ABSTRACT 
 
Title of dissertation:  STATE FISCAL INSTITUTIONS: AN EVOLUTION 
 
Maria Isabel Rodriguez Tejedo, Doctor of Philosophy, 2007 
 
Dissertation directed by: Professor Wallace E. Oates 
    Department of Economics 
 
 
This dissertation considers two aspects of the study of fiscal institutions: the 
importance of understanding their evolution and the need for analyzing their specific 
form.  Chapter 1 addresses the tension between commitment and flexibility in the design 
of institutions and stresses the importance of using a comprehensive approach when 
studying them.  
Chapter 2 explores the history of constitutional debt restrictions, using a new dataset 
that allows us to follow their evolution and construct a proper classification. In this 
chapter, we find that the importance of limits on debt has been overstated. Procedures 
regulating debt issuance are the most important aspect of the provisions.  In addition, 
modification procedures and debt itemization play an important role in the actual limiting 
 
of debt at the state level. Second, the analysis shows a long process of ?recursive 
institutional change,? culminating in a move towards greater flexibility in the 1960s and 
1970s.  This new perspective provides a framework for understanding the appearance of 
new agents in debt issuance, the rise of many different forms of financing, and the 
creation of tools and fiscal institutions that today are basic elements in the budgeting 
practices of the states.  
Chapter 3 investigates the determinants of the configuration of one of the latest of 
such institutions. Using multinomial discrete techniques and introducing the ordered 
nature of the requirements in the analysis, this chapter investigates the reasons that have 
led to the adoption of weak or more stringent budget stabilization funds (BSFs). We find 
that economic factors (such as tax effort, volatility of spending and tax revenue) and 
political factors (like the size of the senate and the fractionalization of the lower house) 
have had a significant impact on the way states have chosen to configure their BSFs (or 
"rainy day funds" as they are often called).   Nevertheless, the specific form of these 
funds cannot be properly explained if we ignore the set of other institutions that exist in 
the states.  We find that balanced budget requirements, the scope of tax and expenditure 
limitations, and the appointment method of the state's supreme court influence the design 
of BSFs. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
STATE FISCAL INSTITUTIONS: AN EVOLUTION 
 
 
 
By 
 
 
Maria Isabel Rodriguez Tejedo 
 
 
 
 
 
Dissertation submitted to the Faculty of the Graduate School of the  
University of Maryland, College Park in partial fulfillment 
of the requirements for the degree of 
Doctor of Philosophy 
2007 
 
 
 
 
 
 
 
 
 
 
Advisory Committee: 
Professor Wallace Oates (Chair) 
Professor Robert Schwab 
Professor Mahlon Straszheim 
Professor John Wallis 
Professor David Sicilia 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
? Copyright by 
Maria Isabel Rodriguez Tejedo 
2007 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
  
  
  
 
 ii
Dedication 
 
A mi familia.  
No podr?a haber elegido mejor a los once que me tocaron,  
y los dos que escog? han sido un regalo. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 iii
Acknowledgements 
 
I am deeply thankful to my advisor, Dr. Wallace Oates, for his immense support and 
guidance. His knowledge, passion and commitment helped shape this research from a 
very early stage, and he has devoted much of his own time to see it come to a good end. 
He has been a continuous source of inspiration and encouragement during my experience 
in graduate school, both professionally and personally.  
I would also like to thank Dr. John Wallis for his many suggestions and constant help. 
He has been extremely generous with his time and ideas, providing much counsel and 
guidance. Drs. Robert Schwab and Mahlon Straszheim took the time to carefully read the 
different parts of the dissertation and provided valuable comments that helped to keep 
this thesis focused. Dr. David Sicilia agreed to join the project when it was nearing an 
end, and provided a very useful fresh look. I would like to thank many others in 
Maryland, in particular Drs. Roger Betancourt and Peter Murrell, for their advice.  
I especially want to acknowledge the Fundaci?n Rafael del Pino for its generous 
support, and several professors at the Universidad de Valladolid for providing the early 
inspiration that led me down this path. 
I also want to express my gratitude to a very special group of people. Marilyn, 
Roberto, Bob and Alex read different parts of this thesis and provided detailed comments. 
Many of the hours I spent in grad school would have been gloomier without Lisa?s cheer 
and spirit, and my good friends Helena, Gunjan, Andr?s, and Melissa were always there 
for me. I cannot imagine my life during the last years (or in the ones to come) without 
Yyann?. She has been classmate, roommate, friend, sister, and my home away from 
 iv
home. I will always be indebted to her for everything she has done for (and with) me. 
Joaquin changed his plans around to accommodate mine. I hope he still has some 
patience left after all these years, and that we have enough time for me to give back some 
of the happiness he has brought to my life. Lastly, I want to thank my family for their 
unending love and their faith in me. Without Manuela, F?lix, Chelo, Antonio, Maribel, 
Juan Alberto, Conchita, Dioni, Carlos and ?lvaro nothing would have ever been possible. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 v
TABLE OF CONTENTS 
 
Dedication  ??????????????????????  i 
Acknowledgements  ??????????????????????  ii 
Table of contents ??????????????????????  iv 
List of Tables  ?????????????????????? vi 
List of Figures  ?????????????????????? viii  
Chapter 1. Institutions as disciplining mechanisms ??????????  1 
 1.1. Introduction  ??????????????????..  1 
 1.2. Tying the government?s hands: commitment v flexibility ?  6 
 1.3.  The importance of a comprehensive approach??????..  9 
Chapter 2. Debt Restrictions: Classification and Evolution???????? 16 
 2.1. Introduction ??????????????????.. 16 
 2.2. A framework for debt restrictions ?????????.. 26 
  2.2.1. Debt limits and procedural restrictions ..????.. 28 
  2.2.2. Issuers of debt ?????????????? 34 
  2.2.3. Types of debt ?????????????? 41 
 2.3. Evolution of debt restrictions ??????????. 45 
 2.4. Conclusions and further work ??????????. 67 
Chapter 3. The Determinants of the Structure of Rainy Day Funds?????. 73 
 3.1. Introduction ??????????????????.. 73 
 3.2. Characterization of budget stabilization funds????.... 77 
  3.2.1. Deposit requirements ????????????.. 80 
  3.2.2. Withdrawal requirements ??????????? 82 
 vi
 3.3. Factors influencing the choice of BSF configuration??? 83 
 3.4. Empirical strategy and results????????????    91 
 3.5. Conclusions ??????...????????????  104 
Chapter 4. Conclusions and Research Prospects. ???????????..   108 
Appendix A. Appendixes to chapter 2 ?..????.????..?????.  119 
Appendix B. Appendixes to chapter 3???.????.????????..  132 
Bibliography ???????????????????????  153 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 vii
LIST OF TABLES 
 
Table 2.1. Nature of debt restrictions ?.. p. 27 
Table 3.1. List of control variables ?.. p. 90 
Table 3.2. Ordered logits. Dependent variable: deposit requirement ?.. p. 100 
Table 3.3. Ordered logits. Dependent variable: withdrawal requirement ?.. p. 101 
Table 3.4. Multinomail logits. Dependent variable: deposit requirement ?.. p. 102 
Table 3.5. Multinomial logits.Dependent variable: withdrawal requirement ?.. p. 103 
Table A.1. Government debt by level of government ?.. p. 119 
Table A.2. Trends in outstanding municipal bonded debt -- by type ?.. p. 120 
Table A.3. Classification criteria of special districts ?.. p. 121 
Table A.4. State debt limits and associated modification procedures ?.. p. 121 
Table A.5. Changes to constitutional regulations on debt related to   
 approval method at state level 
?.. p. 124 
Table A.6. Changes in the limits to the amount of debt states are allowed to issue 
 ?.. p. 125 
Table A.7- Constitutional rules on budget balance, by type and date ?.. p. 126 
Table A.8. Debt provisions & levels of government, ranked by  
 number of units per capita 
?.. p. 127 
Table A.9. Rainy day funds and debt provisions ?.. p. 128 
Table A.10- Constitutional rules on budget balance and debt provisions ?.. p. 129 
 viii
Table B.1.1. Dates of adoption of states? Budget Stabilization Funds ?.. p. 132 
Table B.1.2. Deposit and withdrawal requirements in the BSFs ?.. p. 133 
Table B.1.3. Deposit and withdrawal requirements in BSFs ?.. p. 133 
Table B.3.1. States with gubernatorial term limits ?.. p. 138 
Table B.3.2. Method of appointment of the supreme court ?.. p. 141 
Table B.3.3: Tax burden, 1953-75 ?.. p. 148 
Table B.3.4. Strictness of TELs ?.. p. 150 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 ix
LIST OF FIGURES 
 
Diagram 2.1. Dimensionality of debt restrictions   ?.. p. 27  
Graph 2.1. Introduction of debt limits, by type   ?.. p. 31 
Graph 2.2. Absolute debt limits: provisions for modification   ?.. p. 33 
Graph 2.3. Relative debt limits: provisions for modification   ?.. p. 33 
Graph 2.4. Number of governments   ?.. p. 39 
Graph 2.5.A. Constitutional debt provisions, by type   ?.. p. 46 
Graph 2.5.B. Modification procedures and procedural restrictions qualifying 
states? debt limits (whole period, by type, not repeated by state) 
?.. p. 47 
Graph 2.6. 1839-79: Distribution of new constitutional provisions on debt   ?.. p. 51 
Graph 2.7. 1880-1929: Distribution of new constitutional provisions on debt?.. p. 53 
Graph 2.8. 1960-today: Modifications of debt provisions   ?.. p. 64 
Graph A.1: Trends in new municipal bond issuances, 1975-1999   ?.. p. 120 
Graph B.3.1- Average percentage earning distribution   ?.. p. 143 
Graph B.3.2 - Average percentage composition of spending   ?.. p. 144 
Graph B.3.3 - Average percentage composition of tax income   ?.. p. 146 
 
 
 
 
 1
Chapter 1. Institutions as Disciplining Mechanisms 
 
1.1. Introduction 
State budget deliberations are usually conducted within the framework of 
requirements and limits that seek to provide a consistent background for decision-
making. Within these limits, public decisions are transformed into policy. Since fiscal 
results are not independent of the budgetary process, fiscal institutions play an important 
role in the economic outcomes of states. Moreover, beyond their real effects on fiscal 
outcomes, tools that aim at improving the fiscal health of the state ?even at the cost of 
restricting the freedom of officials- may send a signal to the market of the long-run 
intentions of officials and policy objectives.1  
Fiscal institutions are the consequence of the economic circumstances and overall 
institutional framework in which they were created, and are meant to establish guidelines 
for making choices in a situation that is expected to recur. They are created to address 
problems that policy makers expect to face again (potentially several times) in the future, 
and they are concerned that their response to these problems may not be optimal.  As a 
result, many fiscal institutions represent constraints on the ability of future policy makers 
to choose policies.  
Frequently, when using fiscal institutions in analysis, researchers take a narrow view. 
This ?narrow view? has two characteristics. First, institutions may be considered only on 
the basis of whether they exist in a particular state or not, and often we see institutions 
represented by a dummy variable that takes the value of one if the institution exists and 
                                                 
1 Lowry and Alt (2001), for example, find that strict balanced budget rules help investors extract 
information about officials? attitudes towards deficits. 
 
 2
zero if it does not. This approach ignores the various forms that an institution can take; it 
may cause the grouping under the same name of very different rules, only because their 
object of interest is the same.  
The second aspect of this narrow view is considering institutions exogenous and 
immutable. The complications associated with modifying complex rules, which often 
include the amendment of statutes or constitutions, result in a certain bias towards the 
status quo. However, the fact that institutions are difficult to modify does not mean that 
they are immutable. Understanding how institutions evolve is critical to evaluating how 
they affect economic outcomes, directly or in conjunction with other rules.  
This static vision constrains our view of the effect of regulations and has two main 
consequences. First, it is in part responsible for a certain feeling that some of the 
institutions ?don?t work.? As Wallis and Weingast (2006) put it ?scholars regularly find 
that balanced budget provisions do not produce balanced budgets; debt restrictions do not 
restrict debt issue; tax and expenditure limitations limit neither taxes nor expenditures; 
and budget stabilization funds fail to provide budget stabilization.?2 In addition, it can 
contribute to the blurring of the interactions of institutions, since ignoring the way they 
change can lead us to overlook the impact these changes have on the form of other 
institutions and economic outcomes.   
This dissertation concerns itself with these two aspects of what we have called the 
?narrow view of fiscal institutions.? Three common themes are present throughout the 
discussion. First is the tension between commitment and flexibility that gives rise to the 
existence and configuration of institutions. Second, several types of rules and limitations 
may exist under a similar name because their object of attention is the same, but 
                                                 
2 Wallis and Weingast (2006), page 1. 
 3
considering an institution only from the point of view of whether it exists or not may not 
be very informative when trying to understand its effects in the larger context. Lastly, 
institutions do not appear or evolve independently of the economic and political history 
of the state, and without this historical perspective we are unlikely to understand fully 
their objectives and effects. Institutions interact with each other beyond their relationship 
through their stated object of interest: the effects of a rule may create incentives for the 
enactment of a new institution, it may serve to overcome the constraints imposed by 
another, etc. 
To recognize the importance of the form of fiscal institutions, it is necessary to 
describe and understand the variety in fiscal structures. There are different types of fiscal 
institutions. Some establish procedures that govern the process by which a state adopts a 
budget, including legislative procedures (such as a budget calendar, or the process for its 
approval), the assignment of responsibilities and authorities of the agents involved (such 
as the governor?s item veto), or set up direct requirements (like that of a balanced 
budget). Other fiscal institutions impose constraints directly on fiscal outcomes. 
Examples of these are the debt, tax, and expenditure limitations, and budget stabilization 
funds (also known as rainy day funds).  Although the principles discussed in this 
dissertation are applicable to the larger set of state fiscal institutions, in this dissertation 
we will focus on restrictions on debt and rainy day funds. 
Fiscal institutions operate in two main dimensions.  First, the rules can limit specific 
items. For example, restrictions on debt may establish a debt ceiling on total debt 
outstanding. Tax and expenditure limitations often work in similar ways. Limits can 
either be expressed as absolute dollar number (absolute limits) or as a percentage of a 
 4
relative fiscal magnitude such as assessed property valuation, revenues, or previous years 
expenditures (relative limits). 
Second, rules may govern procedures. These procedures govern the process by which 
the state government makes specific decisions.  Procedural debt restrictions require state 
governments to follow certain procedures, like holding bond referendums, before debt 
can be issued.  Balanced budget amendments require the ex ante construction of a budget 
that is in balance, while stipulating how the budgetary process must proceed.  Procedures 
often include voter approval or legislative supermajorities. . 
The English language is not always helpful in separating the two dimensions.  A debt 
limitation of $1,000,000 is a very different kind of constraint on a state government than 
a debt restriction that requires a state to obtain the approval of voters before bonds can be 
issued. This difference is not trivial, since it has real implications for the potential effects 
of the constraint; nevertheless, it is frequently overlooked in the public finance literature. 
Institutions can establish limits, procedural restrictions, or both. Because the objective 
of the restrictions may be very similar (for example, to impose controls on debt), 
procedural restrictions and the limits are often classified under the same name (for 
example ?debt limits?), ignoring the particulars of the constraint. This becomes a problem 
when using the institution in analysis; if the only parameter taken into account is its 
existence, we may overlook the features of the measure that have a real impact on 
economic outcomes. In fact, for some institutions such as rainy day funds or debt 
limitations, it may be the form rather than the existence that determines if any real effects 
exist at all, or in what magnitude. 
 5
Similarly, the second aspect of the narrow view of fiscal institutions can cause 
problems. Understanding how institutions evolve is critical to evaluating how they affect 
economic outcomes, directly or in conjunction with other rules. Fiscal institutions may 
change in response to many of the changing fiscal outcomes.  If, for example, 
constitutional provisions with respect to state government borrowing change as a result of 
state debt accumulation, then explaining levels of state borrowing as a function of the 
presence or absence of a debt restriction is problematic.  As this thesis shows, state level 
fiscal institutions have not been static over the course of American history. Changes in 
fiscal institutions show definite patterns of change connected with the changing fiscal 
circumstances of state governments.   
We discuss the tension between commitment and flexibility in more detail in section 
1. 2, and, in subsequent chapters, use case studies of two actual state fiscal institutions to 
show the importance of recognizing the endogeneity of fiscal institutions and considering 
their form.  Although the basic issues to consider are the same, the chapters use different 
approaches for their study. In chapter 2 we consider state restrictions on debt, using a 
new dataset that comprises the whole constitutional history of debt restrictions in the 
states. Considering the institution from a historical perspective, we are able to distinguish 
between the different types of restrictions, which sheds light on the issue of why many 
observers have noted that debt limits are consistently exceeded. The analysis of the 
evolution of the institution provides evidence supporting the concept that institutions are 
endogenous and need to be considered jointly. 
In chapter 3 we consider another state fiscal institution, rainy day funds. In contrast to 
chapter 2, the analysis in this chapter consists of an econometric analysis of the factors 
 6
that have affected the choice of deposit and withdrawal requirements embedded in the 
funds. We find here some evidence of the endogeneity of fiscal institutions.  
 
1.2. Tying the government?s hands: commitment v flexibility 
Fiscal institutions are aimed, in large part, at constraining the behavior of policy 
makers. In a sense, they intend to limit the capacity to design and implement policies by 
establishing a set of rules, or by imposing a set of procedural constraints, that must be 
followed before a proposal can become policy.  
In principle, self-regulation may seem a paradox since unconstrained maximization 
yields results at least as good as its constrained counterpart. In any one period, the 
existence of the rule limits the set of available choices and potentially might even 
preclude the optimal solution (Mas-Colell et al (1995), de la Fuente (2000)). This loss of 
flexibility is a non-trivial cost of the establishment of rules. 
However, because agents might be time inconsistent or myopic, decisions taken at a 
given point may not be optimal when considered in the longer run.3 This problem stems 
from the time dimension of the problem and the non-continuity of the agents in charge of 
making policy decisions. Limitations on the power of officials to make decisions are 
                                                 
3 For example, the existence of different types of groups with asymmetric influence in the decision to elect 
public officials (such as current vs. future taxpayers, members of a special interest group vs. the general 
public, etc) can result in non-optimal policies. Because of the costs associated with raising taxes and/or 
restricting expenditure, officials may be inclined to postpone adjustments during downturns or overspend 
during booms. These asymmetries also influence the choice of financing mechanisms: for example, using 
debt instead of taxes for the funding of a project shifts part of the costs to future generations, while the 
associated benefits may be enjoyed by the present generation, who elects current officials. Even if present-
day constituencies do not seek to benefit by shifting the cost of projects or adjustments onto future 
generations, ?debt illusion? could cause excessive borrowing if the voters do not correctly perceive the 
future tax implications of today?s borrowing.   
 7
meant as solutions to these problems, establishing general principles that are aimed at 
providing an overall benefit, even if their constraints become burdensome at times. 
The disadvantage of pre-commitment comes from loss of flexibility. Even a rule 
designed to operate differently under different economic circumstances (such as some of 
the rainy day funds) cannot provide for all the potential economic conditions.4 Flexibility 
is desirable because economic circumstances are mutable and, to some extent, 
unpredictable. If legislators could perfectly foresee all potential future events, they might 
(technical possibilities allowing) prepare a response for each contingent situation and 
include them in the institution. However, because it is impossible to describe every 
possible circumstance in sufficient detail, the rules have to operate in a wide range of 
situations with only a certain number of degrees of freedom. Given the dynamic nature of 
the processes fiscal institutions deal with, this is an important fact to take into account. 
On the other hand, excessive flexibility may be used by officials to enact policies 
directed towards re-election, rather than desired economic outcomes. If the rules are too 
pliable, they lose their usefulness as a method to commit.  
If policy makers feel the need to constrain their future decisions, they are faced with 
the problem of how to commit, and how much. In choosing how to commit, policy 
makers have often chosen to make institutions into laws, rather than using promises or 
                                                 
4 A notable example of this reasoning exists in the famous statement signed by many economists, among 
them several Nobel Laureates, claiming that ?to keep the [federal] budget balanced every year would 
aggravate recessions?. This statement was made in reference to a proposal for a balanced budget 
amendment at the federal level, even though the proposal had built-in mechanisms that would allow the 
budget to become unbalanced when certain requirements were satisfied. The concern, however, was that 
these escape valves would not provide enough flexibility in times of need. Another example can be found 
in the replenishment requirements that some states have embedded in their budget stabilization funds. 
These rules specify a period within which the fund?s resources must be replenished, but because these 
periods bear no relation with the economic cycle, a state may find itself in the obligation to deposit money 
in the fund while still in a recession. 
 
 8
commitments because of the differences that set laws (statutory and especially 
constitutional) apart from compromises. First, it would be difficult to achieve the 
formality of a law through compromise. Second ?and more importantly- deviations from 
the law are more difficult: the breaking of a law typically has higher associated and 
recognizable costs, and its modification is more complicated. Constitutional provisions 
have the added benefit of increased credibility. The inclusion of an item in the 
constitution transmits the perception that the issue is fundamental; and it lends additional 
stability to the institution, since constitutions are more difficult to amend than regular 
laws. Buchanan and Brennan (1985) consider the nature of the compromise to be 
fundamental, and they state ?governments can be induced to take the long view only if 
they are appropriately constrained by constitutional rules?.5 
At any point in time, the degree of commitment may seem a given fact rather than a 
choice. However, over time, policy makers can make decisions regarding the existence 
and level of stringency of their fiscal institutions. An examination of these choices can 
provide information on the nature and degree of the commitments. Furthermore, to 
appraise correctly the degree of commitment derived from the institutional structure of a 
state, it may also be necessary to take into account the extent to which court decisions 
have relaxed or tightened the effects of the rule for policy-making, as well as whether 
other rules or institutions are complementing or undermining its effectiveness.  
A possible way for governments to commit is through the creation of institutions that 
formalize their wish to commit. Although theoretical constructs are possible and even 
desirable to better understand the mechanics of commitment, the studying of real-world 
                                                 
5 Buchanan and Brennan (1985). The word ?constitutional? as used by the authors does not necessarily 
refer to a rule embedded in a Constitution, but rather to a norm that is meant to exist and operate during 
several periods. 
 9
institutions complements this approach by showing how the problems are addressed in 
practice rather than considering ? ?institutions? that not only do not exist, but also cannot 
easily be mapped into real-world institutions.?6 With this aim, we turn now to the 
description of some of the tools that have been used in the states to enact commitment in 
the budget.  
 
1.3. The importance of a comprehensive approach 
Because of data limitations or the tractability of using simplified models, empirical 
and theoretical analyses are sometimes restricted to the study of one or a few institutions. 
The omission is unlikely to be trivial. Primo (2006) presents a theoretical and empirical 
model relating the effectiveness in curtailing expenditure to the interaction of spending 
limits and executive veto. He notes the importance of considering the overall institutional 
environment when evaluating the effects of any one tool, since  ?the unintended 
consequences so often observed after reform may reflect little more than a failure to 
understand the interplay of institutions.?7 
Institutions may reinforce or undermine each other?s effectiveness. A historical 
perspective that incorporates the interaction of various fiscal institutions is useful, since 
the creation and modification of institutions is related not only to the economic and 
political circumstances of the time, but also to the set of constraints in place. The 
interactions between institutions can create unexpected results, but may have also been 
planned. For example, rainy day funds can help states achieve the level of equilibrium 
imposed by balanced budget restrictions, but they may also be a way to avoid the 
                                                 
6 Drazen (2000), page 164. 
7 Primo (2006), page 2. 
 10
restrictions imposed by TELs. While the first type of interaction would be publicized and 
widely discussed when the new institution is proposed for enactment, the second set of 
results is likely to be less obvious to the public. However, if one institution is established 
as a way to avoid the limitations imposed by another institution, failure to recognize the 
interaction causes a double problem. On the one hand, the new institution may ?not 
work:? it may not cause the effects officials stated it would achieve. In fact, if one of its 
(unstated) objectives was to achieve more flexibility for policy makers, it would indeed 
be achieving its real goal, only not its publicly stated one. In addition, the perceived 
effects of the original institution are distorted, since the new tool has successfully reduced 
its effective stringency. Casual analysis of the effectiveness of the original institution 
may result in a misguided impression of lack of success.  
There may be as much risk in overlooking perverse interactions as in overstating them. 
When a new tool is established, it is difficult to foresee all its possible effects under all 
circumstances. When economic conditions change, a rule may prove to be too 
constraining, or unsuitable, to meet budgetary exigencies. New developments need not be 
political maneuverings around a rule, but simply an adjustment to the circumstances. 
These can include more than the enactment of new institutions, including the assignment 
of responsibilities to new agents or even the creation of new agents altogether, changes in 
how certain tasks are carried out (such as privatization of certain services, new taxing 
mechanisms?), etc. Moreover, because the results cannot be forecast in advance, the 
process involves some trial-and-error, with room for mistakes and improvements.  
A set of interacting institutions may create difficulties. Because of its complexity, the 
system may not be very transparent, creating opportunities for manipulation, and 
 11
impeding direct control from the citizenry. Indeed, the matter of state finances has 
become an increasingly professionalized field. However, at least part of this difficulty 
derives from the long evolution of the institutions involved. This is a complex issue for 
which there is no simple solution. 
On the positive side, the states? freedom to react to new problems allows them the 
flexibility to respond creatively to their fiscal problems. In a sort of laboratory, states can 
create or adopt tools that fit their own circumstances, and, in this process, can learn from 
(and teach) their neighbors.  
Institutions are not exogenous. By considering their history jointly, we realize that 
they have indeed co-evolved, rather than developed independently. Their uses and forms 
cannot be properly understood when considered separately, and important insights into 
their functioning arise when taking into account their historical evolution. The difference 
between limits and restrictions also becomes clearer when considering the evolution of 
the institutions, as does its importance for empirical and theoretical research. 
Rainy day funds (RDFs) provide us with a good example. RDFs are tools meant to 
provide a fund for the deposit of resources during good economic times so that they are 
available during depressed times. However, if states were free to run unbalanced budgets 
or borrow unchecked, the need that RDFs are created to serve would become less 
pressing. In addition, other institutions, directly part of the design of the RDF, can 
potentially create incentives that affect the configuration of these funds. Without a 
comprehensive view of the institutional background in which RDFs are supposed to 
operate, it becomes difficult to understand why and how they exist and function. A static 
view of the institutional background would be incomplete as well. Debt restrictions first 
 12
appeared in the 1840s as a response to taxless finance, not to forbid debt but to ensure 
that taxes would be raised. As restrictions on the issuance of debt spread across the states 
and were modified in response to economic and political changes, other tools have been 
created. Balanced budget rules, some of which are sometimes confused with debt 
restrictions, have added an additional level of complication. With the pressures to balance 
the budget or to finance projects with restricted issuance of debt, concerns were born over 
the size of government ?in particular about the amount of taxes. As tax and expenditure 
limitations joined the set of rules under which state finances operate, they created an 
incentive for the existence of separate funds ?outside the scope of the limitations. 
Rainy day funds are, then, a step in a long process rather than an isolated tool. The 
history of the evolution of fiscal tools ?both in the creation of new instruments and the 
modifications of existing ones- provides us with the needed background to place each 
individual institution in the larger framework. Without this perspective, we may risk 
omitting important information on the choices that led to the creation and configuration 
of present-day institutions and, consequently, their effects on fiscal outcomes. 
In the second chapter of this dissertation, we explore this evolution for a particular 
class of fiscal constraints. When economic benefits failed to materialize after the heavy 
investment in railroads and canals in the 1840s, states found themselves facing serious 
problems in financing their debt. To restrict future uses of taxless finance, states adopted 
different types of rules on the issuance of debt.  
We construct a new dataset, gathered directly from state constitutional provisions on 
debt. This previously unavailable information describes the creation and modification of 
the original restrictions on debt, as well as related provisions that affect the ability of the 
 13
state to issue debt, even if they do not limit it directly. Unlike previously used data, this 
dataset allows us to follow the evolution of the constitutional provisions that affect the 
ability of the state to issue debt, rather than providing a snapshot of the institution, or 
including only some of the relevant provisions. Our first finding has to do with the proper 
classification of debt restrictions. By ignoring some of the provisions regulating debt 
issuance, previous studies overstated the importance of limits. We show that, in fact, 
procedures regulating the issuance of debt are the most important aspect of provisions 
regulating debt. In addition, modification procedures and debt itemization ?so far largely 
ignored- play an important role in the actual limiting of debt at the state level.  
Secondly, our analysis shows that a long process of recursive institutional change 
ensued, and these debt restrictions underwent changes and modifications. The process 
culminated in the nineteen sixties and seventies, with a move towards greater flexibility 
in the issuance of debt at the state level. This (previously unaccounted for) change 
provides a new perspective for understanding the recent introduction of other fiscal 
institutions for fiscal restraint and stability. In short, the evolution of these norms has 
contributed to the appearance of new agents in debt issuance, the rise of many different 
forms of financing, and the creation of a number of tools that today are basic elements in 
the budgeting practices of the states.  
In the third chapter we investigate the determinants of the configuration of one of the 
latest of such tools. Using econometric techniques, this chapter investigates the reasons 
that have led to the adoption of weak or stringent BSFs. The investigation of the factors 
that have had a significant impact on the adoption choices of RDFs suggests that some 
economic and political characteristics are relevant. In particular, among the economic 
 14
characteristics, we find that states engaged more heavily in highly volatile spending are 
more likely to choose weak RDFs. In contrast, a high proportion of tax revenues coming 
from volatile sources decreases the probabilities that a state adopts a weak RDF. In 
addition, states that exert high levels of tax effort are more likely to establish strict RDFs.  
Among the political variables, we find that states with larger senates are more likely to 
adopt weak RDFs, and increases in the level of fractionalization of the lower house 
reduce the probability that states choose strict requirements. Nevertheless, the 
configuration of these tools cannot be properly explained if we ignore the set of other 
institutions that exist in the states. States with stricter balanced budget requirements are 
less likely to adopt strict deposit requirements, and states where the scope of tax and 
expenditure limitations is wider are more likely to adopt weak deposit requirements. 
Also, states that have appointed supreme courts are more likely to adopt strict 
requirements. In chapter 3, we shall explain the rationale for these findings. 
Lastly, chapter four concludes and proposes some future venues of research derived 
from this dissertation. First, the method developed here can be extended to other areas. In 
particular, the framework used to investigate debt restrictions could be applied to other 
fiscal institutions, where existence rather than form may have been the focus until now. 
The analysis of local debt restrictions has suffered from the same limitations as its state 
counterpart, and constitutes an open area for future work. A second major venue for 
research would consider applying the new dataset on debt restrictions to gain a better 
understanding of the effects of debt restrictions on fiscal outcomes, the proliferation of 
public authorities and special districts and the use of revenue bonds. Rainy day funds and 
debt restrictions play an important role in the fiscal activity of the states. Considering 
 15
their evolution and paying due attention to their form, we can study the role of learning 
and diffusion in the changes of state institutions, in the fashion of a ?laboratory of fiscal 
federalism.? 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 16
Chapter 2. Debt Restrictions: Classification and Evolution    
 
2.1. Introduction 
Budgetary institutions are often assumed to be exogenous. However, what Richard 
Briffault calls ?the byzantine structure of state finances?8 is the result of a long historical 
process involving constitutional amendments, politics, court decisions, tradition, and a 
myriad of other intermingling elements. In what might be considered a process of trial 
and error or learning-by-doing, legislators draft policies and create institutions to solve 
the financing issues that states face. Often, their choices have unexpected consequences. 
Those consequences may later force new rounds of changes, and these new rules and 
instruments are created on the base of the pre-existing institutions. These changes happen 
in a coherent and almost systematic way, which allows us to draw conclusions about the 
way institutions evolve. To understand this process of what we call ?recursive 
institutional change,? it is useful to separate the history of institutions into stages. It then 
becomes clearer how a certain stage creates the conditions for the next one.    
The rules written into the state constitutions regarding debt restrictions are a good 
example of this involved process. Debt restrictions first appeared in the constitutions as a 
response to a very concrete economic and political problem in the 1840s, resulting from 
the crisis created by failed investments in infrastructure and banking. As time passed and 
needs evolved, state and local legislatures amended, avoided, and legislated around the 
limits. The introduction of debt restrictions addressed a real problem, but unexpected 
developments created other issues and gave rise to additional rules, institutions, and 
                                                 
8 Balancing Acts: The Reality Behind Balanced Budget Requirements, R. Briffault 1996.  
 17
practices.  There was another round of constitutional changes from 1877 to 1920, a third 
round during the Great Depression and World War II, and a fourth round of changes in 
the 1960s that affect the way in which governments conduct their business today. For the 
most part, the various types of restrictions on debt are not aimed exclusively, or even 
largely, at preventing borrowing. Restrictions may establish a limit to the issuance of 
debt, but frequently the limit is accompanied by a provision that allows the state to 
borrow without being constrained by the limit if certain procedures (usually voter or 
legislative approval) are satisfied. Concerns about the functioning of the political system, 
the possibility of corruption, the selection and financing of worthy infrastructure or other 
projects with the potential to benefit the community beyond their cost were also 
considerations in the drafting of the regulations. 
States are not, however, the only governments involved in this process. Lower levels 
of government and other governments, such as public authorities, have become active in 
debt issuance at the subnational level. Their relative importance has not remained 
constant, and the creation and increased involvement of these sub-state institutions in 
borrowing play an important role in determining the effects of constitutional limits to 
state debt. In section 2.2 we present an overview of the main sets of governments 
involved in this process, as well as the different types of debt used. 
Because restrictions on debt come in several forms and this variation results in very 
different effective levels of constrains on issuing debt, it is important to understand the 
structure of debt restrictions. These restrictions are often misinterpreted when observers 
fail to account for the channels that allow states to issue debt beyond what appears to be 
the limit. These other options are not ?tricks? to avoid the restriction, but actual 
 18
provisions in the constitutions designed to allow debt issue. Ignoring them results in a 
distorted view of debt restrictions. In fact, after considering their nature from the 
perspective of their form, rather than simply the existence of any kind of provision 
dealing with debt, it becomes obvious that the substance of the restrictions concerns 
procedures, rather than limits. We discuss these differences and present a system to 
properly classify state constitutional debt restrictions in section 2.1. In addition, careful 
classification and consideration of the restrictions is useful in disentangling their 
relationships with other institutions. By analyzing in detail debt provisions, we can 
observe how their close link to balanced budget rules has led to confusion between the 
two. 
Unlike the federal government, states deal with financial matters, such as taxing, 
spending, and borrowing in their constitutions. In particular, almost every state has a 
constitutional provision on balanced budgets9 and restrictions on debt issue. Many state 
constitutions include a whole article on ?state finance,? often spelling out clear 
limitations on the amount and type of debt that can be issued, how debt creation must be 
authorized by the various levels of governments, and specific procedures and timelines to 
do so. These rules are accompanied by statutory provisions, court interpretations and 
guidelines, and policies at state and local levels, but their explicit inclusion in the 
constitutions reflects the importance legislators and the public place on these provisions. 
                                                 
9 According to the Advisory Commission of Intergovernmental Relations, the only exception is Vermont. 
However, as presented in table B.7, once the provisions that refer to debt or that do not deal directly with 
budget processes are classified appropriately, the picture changes considerably. Still, this does not mean 
that states do not aim to balance their budgets: the ?sense? that the budget needs to be balanced may have 
already surpassed the written provision and be part of the policy making process on its own. For more on 
balanced budget requirements see Briffault (1996) or Rodriguez-Tejedo (2006). 
 19
Their constitutional nature may respond to more than their perceived importance: 
constitutional rules result from the realization that systematic features of the incentive 
structure or decision making of the legislature are likely to bring about problems. If 
voters were convinced that a problem was caused by the decision taken by an elected 
official, they would likely resort to the replacement of the official in the next election.  In 
contrast, if voters feel that the problem stems from the structure of the government 
process, then the solution is to enact a constitutional provision that changes the 
procedures of government decision making. Constitutional provisions are a consequence 
of the perception of a systematic problem in government process that cannot be properly 
addressed through regular unrestricted legislative action.  As this chapter shows, the 
substantive part of most debt restrictions is not to limit the amount of debt that can be 
issued but to stipulate the procedures that state and local governments must follow to 
authorize borrowing. 
Constitutions are difficult, but not impossible, to modify. Provisions regarding debt 
limits have been altered in response to changing circumstances or in pursuit of a better 
rule. Often, however, researchers take debt provisions as immutable and consider only 
their present state when analyzing them or their effects on other economic variables. This 
chapter examines how state debt restrictions change over time and places those changes 
in the context of other fiscal institutions adopted by states.  
The main data sources are the constitutions themselves. From 1776 to the present, 
constitutions have been amended and some have been superseded by new ones (for 
example, the state of Louisiana has had twelve active constitutions). An appendix 
containing the actual wording of constitutional limitations on state and local debt and 
 20
their evolution is available from the author by request. The appendix provides a 
comprehensive recount of these provisions, including amendments, based on information 
provided by the NBER/MD Constitution Project website.10 From this information we 
have constructed a dataset that spans the whole constitutional history of debt restrictions 
at the state level. Unlike previous data, the dataset considers not only the original and/or 
present restrictions on debt; it provides us with a panel that includes the date and nature 
of any modifications to the restrictions since their creation. In addition, it includes 
information that had previously been overlooked because only provisions that directly 
included the terms ?debt? or ?debt limit? were included. To take into account provisions 
that affect the issuance of debt, even if they are not contained in the article that directly 
imposes a limit, we reviewed the sections of the constitutions that dealt with state 
finances. This dataset, then, addresses two limitations of the previous information. First, 
it covers the complete constitutional history of debt restrictions over time, providing a 
record of the evolving rules over time and across states, rather than a series of congealed 
pictures corresponding only to the rules originally written in the constitutions or a 
snapshot of regulations at a point in time. Second, it expands coverage to the full set of 
the restrictions that affect debt issuance. 
Originally, limits on the actual amount of casual debt were the dominant form of 
restriction.  These limitations were usually accompanied by procedural requirements to 
obtain voter approval to issue debt beyond the casual limit amount. Casual debt is debt 
used to cover deficiencies or shortages in revenue (as opposed to debt for the financing of 
a project over a long period of time). To be clear, we define debt limits as restrictions on 
the absolute dollar amount of debt allowed or a dollar amount determined by a relative 
                                                 
10 Online at < http://www.stateconstitutions.umd.edu/ >  [Last accessed on June 25th 2007.] 
 21
measure (e.g, 2 percent of assessed property value).  We define procedural restrictions as 
the stipulations constitutions place on the process used to authorize debt.   
Over time states tended to move away from absolute limits into relative ones 
(determined either by a percentage of income or property values).  They also tended to 
alleviate the effect of limits by periodically raising the limit, or by itemizing debt issues 
for specific functions, like highways or water systems, that were not subject to the limit 
on general debt.  The raising of limits, and the widening of debt itemization as a way to 
issue debt beyond the limit, have combined with the procedural restrictions which usually 
allow any amount of debt to be issued as long as it is authorized by the right procedure.  
This led states in the nineteen-sixties and seventies to significantly ease debt restrictions 
and replace them with simpler and more flexible rules like legislative super-majorities 
without voter approval.  The easing of debt restrictions in the nineteen sixties and 
seventies has not been recognized previously in the literature on balanced budget 
amendments, tax and expenditure limitations, and rainy day funds.  Our investigation of 
the evolution of debt restrictions provides new insights for the analysis of other 
institutions created in the last part of the twentieth century. 
Debt at subnational levels is not, by any means, insignificant. In fact, the subnational 
debt market in the United States is the largest in the world.11 According to figures 
published by the Census of Governments, states? overall total debt outstanding per capita 
at the end of the fiscal year grew by almost 30% in real terms between the years 1977 and 
2000. Brecher, Richwerger and Van Wagner (2003) show the increasing importance of 
debt in state finances, noting that the gross amount of outstanding debt tripled from $72 
billion in 1975 to $212 billion in 1985, and reached more than double that amount in 
                                                 
11 Liu and Wallis (2007) 
 22
2000. Table A.1 provides an overview of the evolution of debt by level of government, 
both in amount and as percentage of the total. Although there is a gap in the table 
between 1841 and 1870 due to data unavailability, we can see a pattern in the distribution 
of government debt. In the early years, states issued over six times as much debt as the 
federal and local governments combined. Later on, states were replaced by local 
governments, whose share had risen to over half of the overall debt at the turn of the 
twentieth century and 70% of total government debt in 1913.  Another shift occurs when 
local governments are displaced by the federal government as main issuer of debt, mostly 
because of war-related debt. 
Restrictions on the issuance of debt are put in place because the public or policy 
makers consider that, if unchecked, too much debt would be issued. However, the 
question remains as to what kind of impact these institutions have, and whether formal 
restrictions have real effects.12 One can consider several potential outcome effects, such 
as the amount and kinds of debt issued, the issuers involved in the process, the interest 
rate associated with borrowing and the impact of debt restrictions on the overall set of 
institutions of states. 
The most immediate effect is, of course, on the actual amount of debt issued. There 
is no consensus on whether these restrictions really curtail debt, and some studies find 
that the amount of state long-term debt is not significantly affected by the constitutional 
                                                 
12 Poterba (1995) reports that studies on the impact of fiscal institutions on budget outcomes find that strict 
anti-deficit rules induce lower deficits and faster adjustments when these rules are coupled with anti-
borrowing regulations. Brecher, Richwerger and Van Wagner (2003), on the other hand, draw attention to 
the fact that this effect disappears once other economic, political and service demand factors are taken into 
account ?leaving the shift towards different forms of debt as the main consequence of debt limits. Fat?s and 
Mihov (2006) find that limits make fiscal policy more pro-cyclical, but they argue that fiscal constraints 
have their benefits in terms of reduced budget deficits, sustainable budgetary plans and decreases in 
discretionary changes in fiscal policy, suggesting that the benefits derived from fiscal rules outweigh their 
costs. 
 23
restrictions that aim at limiting it,13 while others claim that some debt limits may have the 
potential to reduce debt.14  
This kind of effect is complicated further by the fact that there is not a definitive list 
of debt instruments. Even if the use of a particular type of debt declines because of the 
introduction of the rules, it is possible that another form may become more widespread. 
In particular, states with limits on general obligation debt may turn to the issuance of 
non-guaranteed debt for their financing needs,15 but some studies suggest that the 
importance of this effect may be overstated.16 Table A.2 depicts the evolution of general 
obligation and revenue debt in the twentieth century. Because of data availability, there is 
a gap between 1937 and 1966, but we can see a large increase between these dates in the 
share of revenue bonds relative to that of general obligations bonds. The share of revenue 
bonds in the total remained at about a third until the early nineteen seventies, when it 
started increasing again. Graph A.1 depicts the relative shares of revenue bonds and 
general obligation bonds issued by municipal governments. The figure shows a 
continuous increase in the share of revenue bondsuntil the mid-nineteen eighties when 
                                                 
13 Mitchell (1967), Pogue (1970), McEachern (1978) and Clingermeryer and Wood (1995) find that the 
amount of state long-term debt is not significantly affected by constitutional restrictions on debt. 
14 Kiewiet and Szalaky (1996) find that states that prohibit debt or require voter approval for its issuance 
have less guaranteed debt than those who are allowed to issue debt based on legislative approval or that 
have revenue-based limitations 
15 In fact, Regens and Lauth (1992) report an increase of almost sixty percentage points in the amount of 
non-guaranteed debt issued by the states in the period 1950-1989, and Hackbart and Leigland (1990) 
describe how during the 1977-1987 period the importance of full faith and credit debt changed from 
representing roughly the same amount as non-guaranteed debt to being only a third of the latter in 1987.  
Bradford and Constantine (2005), Bunch (1991), Nice (1991), and Von Hagen (1991) find evidence of the 
substitution between these types of debt and consider that debt restrictions do not reduce the overall amount 
of debt. However, Bahl and Duncombe (1993) find that constitutional debt limits reduce both general 
obligation and non-guaranteed debt, although they find evidence suggesting that these two types of debt 
may be substitutes 
16 Kiewiet and Szalaky (1996) do not find evidence supporting the theory that debt limits encourage the 
issuance of non-guaranteed debt. At the local level, Farnham (1985) finds that state-imposed limits on local 
borrowing reduce the amount of both guaranteed and non-guaranteed debt, even if there is some 
substitution away from full faith and credit debt to non-guaranteed debt 
 24
new issues of these bonds started decreasing in response to the enactment of the Tax 
Reform Act (TRA) of 1986; this Act limited the number and types of bonds exempted 
from federal taxation. A concern usually associated with this shift is the higher cost of 
non-guaranteed debt, although some studies have focused on the impact of these 
limitations on the price of general obligation debt.17  
Another effect has to do with shifts in issuance among levels of government. We 
have already discussed the shift away from state debt to its local counterpart, as reported 
by table A.1, but other shifts have also taken place. Because the restrictions are written to 
control state and (later on) local debt, many courts have considered debt issued by public 
authorities (PAUs) and special districts to be outside the scope of these provisions. This 
is the reason why some authors link the proliferation of PAUs and special districts to debt 
restrictions,18 although others suggest alternative explanations.19 Two main views on the 
implications of the growing number of PUAs and special districts have been hotly 
debated. On the one hand, some argue that this proliferation may result in a decreased 
                                                 
17 Bayoumi et al (1995) find that (at average levels of debt) the existence of fiscal controls lowers interest 
costs by 50 basis points, a result similar in spirit to that of Poterba and Rueben (1997), who find that states 
with stricter limits on the legislature?s ability to issue debt have lower bond interest rates.   
18 Bunch (1991) uses cross-sectional data from the eighties to test the impact of constitutional debt limits on 
the number of public authorities, the scope of their activities, the existence of a public building authority, 
and the state?s reliance on PAUs to issue debt to build infrastructure. She finds that states with a 
constitutional debt limit that encompasses both general obligation and revenue bonds have a higher number 
of PAUs and more activities fall under the power of PAUs, although such results do not exist if the limit 
applies only to general obligation debt. States with debt limits of either kind are also more likely to have a 
public building authority and to finance their public infrastructure debt through PAUs. Hackbart and 
Leigland (1990) report an increase in the number of state-level entities that issue revenue-backed debt of 
almost sixty percentage points, and Leigland (1994) finds support for the use of PAUs as a way around debt 
limits in his finding that a high amount of PAUs is usually associated with low levels of full faith and credit 
debt. However, Trautman (1995) remarks that limits on general obligation debt alone don?t have a 
significant impact on the number of public authorities; the combination of limits on state general obligation 
and state revenue-backed debt increase the number of public authorities by almost a third. 
19 For example, Bourdeaux (2005) finds that it is not only financial concerns that bring about the creation of 
a PAU; politically competitive environments are also a relevant factor. In addition, when Frant (1997) fails 
to find a relationship between debt restrictions and either the number of PAUs or their issued debt, he 
hypothesizes that they may have other features that explain their existence. Kimball (1976) suggested that 
the increase in the number of PAUs may be partially due to a ?laboratory? effect, where states learn from 
their own successes and those of neighboring states. 
 25
transparency and loss of control by citizens and even by the state legislators themselves. 
An alternative (or complementary) view holds that part of this fragmentation process may 
correspond in some instances to the improved match between those who pay for the debt 
and those who enjoy the derived benefits.20 
Section 2.2 discusses the concepts necessary to provide a clear picture of the nature 
and evolution of restrictions on debt. We discuss the different types of restrictions, 
emphasizing the difference between limits to casual deficit, absolute and relative limits to 
debt, and procedural restrictions. We also call attention to the often overlooked need for 
considering these restrictions jointly with other provisions that allow states to issue debt 
beyond the limit, and the ways in which constitutions themselves can be modified. The 
rest of section 2.2 discusses the issuers of debt as well as the types of debt. Section 2.3 
presents an overview of the evolution of these limits since the time of their creation: here 
we find that debt restrictions have indeed evolved and are not ?as often assumed- 
exogenous. On the contrary, they may have been an important factor in the creation of 
other institutions towards the end of the twentieth century. Section 2.4 summarizes the 
findings of the chapter and discusses future venues for research. 
 
                                                 
20 Wallis and Weingast (2006) explore the possibility that some PAUs may be a natural result of the limits 
to local borrowing paired with an evolution of financing needs and a better alignment of the payers and 
beneficiaries of projects. A piece of evidence that supports this claim can be found in the evolution of 
special districts. As Pagano and Perry (2006) point out, the emphasis on these districts has increasingly 
been on financing infrastructure through the actual users, rather than the entire community itself, and 
creations such as Downtown Development Districts, and Industrial Development Parks have been formed 
within cities, going in numbers from about 6000 to around 14000 between 1962 to 1992. 
 26
2.2. A framework for understanding debt restrictions 
Debt creation and management is a multi-faceted process. The process for 
authorizing debt issues depends on the level(s) of government involved, the type of debt 
instruments used, and the fiscal institutions in place.  
The elements that can configure a debt restriction are sometimes the source of 
confusion. At its most basic, a debt restriction can be constituted by limits (absolute or 
relative, depending on whether the limit is given as an actual amount or as a percentage 
over some variable, such as property value) or procedural restrictions. We assign the 
general name of ?debt restriction? to refer to any and all of these forms, and reserve the 
term ?debt limit? for those provisions that establish an actual maximum amount of debt 
that can be issued. ?Procedural restrictions,? on the other hand, require the state to follow 
certain procedures to issue debt, but establish no maximum on the total amount of debt, 
as long as the procedures are followed. The line between these debt limits and procedural 
restrictions can become blurry, because many debt limits have an associated procedural 
restriction. In addition, constitutional debt restrictions can be modified by amending the 
constitution through upward revision of the limit, debt itemization or by the direct 
modification of the constitutional provision. We refer to these as ?modification 
procedures?. 
But beyond the actual configuration of the provision, debt restrictions operate on a 
variety of levels and they are often misunderstood when a limit on one dimension is 
accompanied by a method for modifying the limit on another dimension.  Diagram 2.1 
and table 2.1 present some of the dimensions that need to be taken into account.  We 
discuss them in more detail in sections 2.2.1, 2.2.2, and 2.2.3. 
 27
 
 
 
 
 
 
 
 
Table 2.1. Nature of debt restriction 
Limits  Procedural restrictions 
Scope Modification procedures  Methods 
Absolute    Upward revisions  Voter approval 
Casual debt Debt items  Legislative majority 
Long term debt    
Relative (% of)    
Some measure of income    
Property value    
 
Reading diagram 2.1 from right to left, debt can be issued by state or local 
governments, and within local governments by county, municipal, or other governments.  
Special districts and public authorities are often created specifically to use debt finance to 
provide a public service or interjurisdictional public goods.  Special districts and public 
authorities can be created by state or local governments.   
The type of debt issued also varies.  General obligation debt obligates the issuing 
government to service and repay debt from all of the revenue sources available.  Revenue 
bonds are debts payable from specific revenue sources, and only from those sources.  
General obligation bonds and revenue bonds are not exclusive categories; some bonds 
mix features of both general obligation and revenue bonds.  
Applicability 
 
 
 
 
 
 
 
Nature of the debt restrictions  
(see table 2.1) 
 
Level 
 
 
 
 
 
 
Types of debt 
State 
 
Local 
 
Other 
Special districts 
 
Public authorities 
General obligation 
 
Limited obligation 
Diagram 2.1. Dimensionality of debt restrictions 
 28
Both the level of government and the type of debt that can be issued are regulated by 
provisions in state constitutions, and we briefly review them later in this section.  
However, the aspect that is most often misunderstood is the nature of the debt restrictions 
themselves. We turn now to their description and present a classification that tries to take 
into account the different kinds of restrictions and provides us with a framework to 
explain why debt limits are seemingly exceeded so frequently among states. 
 
2.2.1. Debt limits and procedural restrictions 
In general, these provisions are called debt limitations, but their structure is more 
complicated.  Debt limits establish an actual limit on the amount of debt that can be 
issued.  The limit can be an absolute limit stating a specific dollar amount, imposing a 
cap on total debt21, fixing the maximum level of annual debt service payments22 or even 
banning debt altogether23. Limits based on absolute maxima of total outstanding debt or 
on the amount of debt service are sometimes very small in real terms, and states that have 
these kinds of limits and have not updated them may end up facing a de facto near 
prohibition of debt.  Alternatively, we also observe relative limits that limit debt issues to 
a percentage of revenues or assessed property value.24  
Procedural restrictions, in contrast, do not limit the amount of debt that can be issued, 
but require the government to follow certain procedures for authorizing debt issues, such 
                                                 
21 Arizona and Nebraska are examples of states that have this kind of ?maximum dollar amount? limit. 
22 Although Louisiana?s ability to issue large amounts of debt is dependent on referenda, its constitution 
allows the legislature to contract debt without voter approval as long as total debt service is kept under 6% 
of the total estimated general fund and dedicated funds 
23 Such as Arkansas and West Virginia. 
24 Georgia and Hawaii, for example, have limits related to the amount of revenue, while Nevada and Utah 
provide examples of states where the maximum admissible borrowing is constrained by the value of taxable 
property. 
 29
as requiring the approval of a majority of voters25 or the inclusion of super-majority 
provisions for debt approval in state or local legislatures.26 An example of such a 
restriction on debt can be found in appendix A.2.2.  
Most constitutional debt provisions really have two parts.  The first part is a limit that 
creates a maximum allowable amount of debt that the legislature can create freely; and 
the second part allows the state to issue debt beyond the amount established by the limit 
if certain conditions are met. The two parts usually exist in the same article.  
For example, the text of the first complete debt restriction was in Article 4, section 6 
of New Jersey?s constitution of 1844; it states that:  
 
?The legislature shall not, in any manner, create any debt or debts, liability or liabilities, of the State 
which shall, singly or in the aggregate with any previous debts or liabilities, at any time exceed one 
hundred thousand dollars, except for purposes of war, or to repel invasion, or to suppress 
insurrection, unless the same shall be authorized by a law for some single object or work, to be 
distinctly specified therein; which law shall provide the ways and means, exclusive of loans, to pay 
the interest of such debt or liability as it falls due, and also to pay and discharge the principal of such 
debt or liability within thirty five years from the time of the contracting thereof, and shall be 
irrepealable until such debt or liability, and the interest thereon, are fully paid and discharged; and 
no such law shall take effect until it shall, at a general election, have been submitted to the people, 
and have received the sanction of a majority of all the votes cast for and against it, at such election; 
and all money to be raised by the authority of such law shall be applied only to the specific object 
stated therein, and to the payment of the debt thereby created. This section shall not be construed to 
refer to any money, that has been, or may be, deposited with this State by the government of the 
United States.? 
 
                                                 
25 Such as Florida and Rhode Island. 
26 This kind of provision exists, for example, in Pennsylvania and Massachusetts.  
 30
The opening clause states the limit: ?The legislature shall not, in any manner, create 
any debt or debts, liability or liabilities, of the State which shall, singly or in the 
aggregate with any previous debts or liabilities, at any time exceed one hundred thousand 
dollars,? and then follows it with the procedural restriction ?unless the same shall be 
authorized by a law for some single object or work?.? The language of the New Jersey 
constitution thus creates an absolute limit on debt of one hundred thousand dollars.  
However, the provision continues to allow debt beyond this limit if voters give their 
consent to a law that allows debt for a single, distinctive object, and provides ways and 
means for the repayment of principal and interest.  
Limits and procedural restrictions are the main types of debt provisions, but both types 
are affected by other modifying provisions. A key element has to do with the purpose and 
the time period of the debt issued. Because some debt is issued to cover deficiencies or 
shortages in revenue, rather than to finance of a project over a long period of time, many 
states allow the legislature to issue ?casual debt? up to a fixed amount.  Casual debt is 
typically authorized to meet ?temporary deficiencies in revenue? (or similar language, an 
example of such a restriction can be found in appendix A.2.3).  If a limit on debt is 
coupled with a procedural restriction that allows an unlimited amount of debt to be issued 
if it is financed by higher taxes approved by the voters, then the debt limit is only binding 
in the sense that the legislature must go to the voters to borrow.  
Graph 2.1 depicts the time patterns in the adoption of the different types of debt 
limits. The height of the bars represents the number of constitutional provisions dealing 
with debt limits (exclusively) that take place in each decade. It is important to note that 
this graph does not include any procedural restrictions associated with the limits; rather it 
 31
is meant to provide an overview of debt limits, both to provide an image of the patterns 
and to illustrate what constitutional restrictions on state debt would look like to an 
observer who reads only the absolute limits without considering procedural restrictions 
and modification provisions.  
 
G raph 2.1 - Introduction of debt lim its, by type
0
2
4
6
8
10
X axes: decade 
     (starting year) 
Y axis: incidence 
     (number)
Absolute amount 0 6 3 2 0 0 0 1 3 0 0 0 2 0 0
Casual deficit 1 4 6 0 3 5 1 1 2 1 0 0 1 1 1
no debt 0 0 0 0 1 0 1 0 0 0 0 0 0 0 0
1830 1840 1850 1860 1870 1880 1890 1900 1910 1920 1930 1940 1950 1960 1970
 
 
Other types of debt are also allowed as exceptions to restrictions.  Most states allow 
debts to be issued to ?repel invasion, suppress insurrections, etc.?  Debt issued for this 
purpose is not subject to limits.  In a similar way, states may authorize debt for specific 
purposes, like highways, education, or public buildings.  We call these ?debt items?; such 
debt items may be subject to a limit or they may not be. 
Finally, all of the limits and procedures are subject to modification through the process 
of constitutional amendment.  Absolute and relative debt limits can be raised, new debt 
items can be authorized, and new procedures put in place.  As we discuss later in the 
chapter, almost all of the states that started with simple absolute limits on debt have 
modified their constitutions in one way or another to make their limits more flexible. 
 32
Of the 24 instances27 in which a state constitution includes an absolute limit (not tied 
to casual deficits, but rather for overall debt), in only three cases is there no constitutional 
record of some type of modification, either explicit,28 and/or implicit.29 It is important to 
note that sometimes these modifying provisions are present in the same constitution (for 
example, in all six cases where the limit was revised upwards other types of procedural 
restrictions existed). A similar result can be observed when examining the relative limits 
to the issuance of debt.30 In short, the real constraint to debt appears to be the procedural 
restrictions, rather than the actual debt limit.  
Graphs 2.2 and 2.3 present the distribution of modification provisions for the absolute 
(applying to long term debt, rather than debt for casual deficits) and relative limits that 
have been included in the states? constitutions during their whole period of existence31. 
                                                 
27 These may include the same state with two separate constitutional provisions at different points in time. 
Also, because of data restrictions, no information is available for Vermont, Nebraska, Missouri, Georgia 
and Tennessee.  
28 Six states allow modification if approved by voters, three if approved by the legislature, two more accept 
either method, and in three instances both approvals are required. 
29 These implicit modification procedures are the inclusion in the constitution of itemized debt (which 
happens in nine of the twenty-four cases) or the upward revision of the limit (which we find in five cases). 
30 A table with a summary of the limits and the modification procedures can be found in the appendix (table 
B.3). 
31 It is worth noting that relative restrictions are often created with reference to concepts such as taxable 
property, whose definition may be in control of the legislature. This ?built-in? flexibility may be the reason 
for the slightly higher number of ?no modification provisions? among relative restrictions. 
 33
Graph 2.2: 
Absolute debt limits: 
Provisions for modification
23%
11%
9%31%
17%
9%
Voter approval
Legislative supermajority
Voter and legislative approval
Debt itemization
Upward adjustment
None      
Graph 2.3: 
Relative debt limits: 
Provisions for modification
23%
12%
6%
23%
18%
18%
Voter approval
Legislative supermajority
Voter or legislative approval
Debt itemization
Upward adjustment
None
 
 
The opportunity to modify debt restrictions by constitutional amendment may or may 
not be more difficult than actually authorizing debt through an existing procedural 
restriction. For example, suppose a constitutional amendment requires a majority vote of 
both houses of the legislature and approval by 50% of the voters. Suppose, moreover, that 
a new bond issue requires a majority vote of both houses of the legislature, the imposition 
of a higher tax rate to fund the bonds, and approval of 50% of the voters. Then it isn?t 
clear whether it is politically more costly to amend the constitution to approve an increase 
in the debt limit for a specific function or to get voters to approve a bond issue that raises 
taxes. 
How difficult it is for a state government to issue debt depends on more than the 
procedures and limits provided in the constitution; it also depends on the difficulty of 
amending the constitution itself. As a result, debts limits are more complicated than they 
seem at a first glance.  
 34
2.2.2. Issuers of debt 
In a decentralized system, the structure of government itself becomes an important 
part of debt issue. Regulations at one level may shift the issuance of debt to others, and 
mandates or transfers can affect the necessity to finance projects through future revenue 
and/or the capacity of the government to sustain debt payments. In addition, the various 
levels of government serve different purposes. In some instances, it may be easier to 
reach an agreement on debt issuance at the lower level, where citizens? preferences are 
likely to be more homogeneous. Alternatively, some projects have an impact over a large 
area, and a higher level of government is better fitted to provide for its financing. 
Considering a rule that establishes limits to debt issuance at the state level without 
taking into consideration the activity of other levels can be misleading, since often they 
are deeply interrelated. An example can be found in Oregon?s Irrigation and Drainage 
District, created by a constitutional amendment in 1919. Although article 11, section 7, of 
Oregon?s constitution establishes that the state?s liabilities are not to exceed 50,000 
dollars ?allowing for an additional issuance of debt equivalent to a certain percentage of 
the value of the property in the state- the amendment allows the state to pay the interest 
on debt issued by the district issuing general obligation bonds under the name of ?district 
interest bonds.? 
The number and identity of government units that issue debt has not remained 
constant over time, and neither has the distribution of debt among the different levels (see 
table A.1).  The type of activities that each government performs and how they are 
related to citizens and other levels of government has also undergone a continuous 
process of change. In fact, not even the types of governments with the legal authority to 
 35
borrow has remained constant. Their geographical distribution is not straightforward 
either: some of these governments exist and act within political boundaries, but others 
cross or supersede them, encompassing several states or restricting their field of action to 
a particular piece of infrastructure, such as a bridge or a toll-road. However, they differ 
more than just in size: their forms of government and the need for financing vary, and so 
do their legal ability and economic capacity to issue debt vary (and hence the amount and 
price at which they are able to borrow).  
Besides the federal government, which we do not consider here, we distinguish 
between states, local governments, special districts and public authorities.32   
 
2.2.2.1. States  
States are intrinsically different from the other governments because states are 
sovereign entities. For debt purposes, sovereignty translates into the ability to create debt 
and to disallow being sued by individuals if the state defaults on its debts, since states are 
immune to suit without their consent.33 They also tend to be more complicated in their 
organization since, as Mikowski (1993) points out, states have legislative, executive and 
judicial branches, all involved in the creation and administration of debt. 
                                                 
32 This classification is not universal. The Census, for example, considers five types of local governments: 
counties, municipal governments, townships, special districts and school districts. School districts are out 
of the scope of this study; and counties, townships and municipal governments ?which all share the feature 
of providing general government services- are lumped together under the general name of ?local 
government? or ?municipal government? 
33 Amendment XI of the United States constitution states that ?The judicial power of the United States shall 
not be construed to extend to any suit in law or equity, commenced or prosecuted against one of the United 
States by citizens of another state, or by citizens or subjects of any foreign state.? There is some argument 
over whether states have waived their immunity regarding bankruptcy laws (in Central Virginia 
Community College v. Katz (2006) the Supreme Court rejected the claim in favor of the state?s sovereignty 
immunity). This topic is, however, out of the scope of this study and is not pursued further here. 
 36
On the other hand, states have wider tax bases and more diverse sources of revenue, 
which make them less vulnerable to specific shocks, and hence more attractive to risk-
averse investors. In addition, they usually have at their disposal a wider set of technical 
and human means to design and manage their debt policies.  
States are also unique in their relationships with the other levels of government: 
Through their relations with the lower and higher tiers, they have obligations and receive 
revenues. They can receive mandates that result in required expenses from the federal 
government, and may also impose similar demands on local governments.  
 
2.2.2.2. Local governments   
Commonly understood to include counties, cities, towns, townships and villages, this 
group includes a wide variety of governments, including very disparate issuers of debt, 
such as cities, school districts, American Indian tribes, territories of the United States, etc. 
Special districts and public authorities are sometimes also considered a type of local 
government, even though they can cover several states or cross state lines, but we 
consider them separately here because of their special nature regarding debt issuance.  
Not every state has the same local government organization, and within states local 
governments vary enormously in size, population, attributes, and powers.34 Here we refer 
to local governments or municipalities interchangeably.  
For the purposes of debt issuance, municipalities often find their overall capacity to 
issue debt constrained by rules at the state level, which impose maxima or demand 
                                                 
34 Local governments are, in a way, creatures of the state and often the rules for their creation and 
management are embedded in state constitutions. According to the Tenth Amendment of the constitution, 
local governments are a state matter rather than a federal one. However, local governments typically have 
considerable discretion in the organization of their activities, with home rule being a staple of their 
independence. 
 37
referenda. However, this is not done in a uniform fashion: some states limit debt at the 
local level in very specific ways, while others do so in general terms. The reality of 
issuance of debt at the local level is complicated and goes beyond the set of borrowing 
constraints, including a wide range of purposes and instruments.  
 
2.2.2.3. Special districts and public authorities 
Because constitutional restrictions on debt issuance are usually restricted to the state 
(and local) governments, debt restrictions may or may not apply to new governments that 
state and local governments create. It is because of this that the proliferation of special 
districts and public authorities is often attributed to the creation of state and local debt 
restrictions. 
Special districts have a long history in the United States, with the first example 
appearing in Rhode Island in 1797.35 The distinctions between special districts and other 
governments are often blurry, and they are frequently confused with school districts 
(which are, in a way, special districts, but are usually treated as a separate category 
because of their homogeneous nature) and public authorities. There is, indeed, no clear 
definition of what constitutes a special district. They serve a wide range of purposes, and 
can vary significantly in size. Florida, for example, lists a total of 1529 special districts, 
divided across sixty-two functions,36 while the U.S. Census reports only 14 for Alaska 
and 45 for Louisiana.37 
                                                 
35 Haas (1996, page 441) 
36 Information from Special District Information Program, part of the Florida Department of Community 
Affairs, online at < http://www.floridaspecialdistricts.org/ > [Last accessed May 21st 2007] 
37 The comparable number for California is 2830. The difference between the two numbers in California 
could be due both to the difference in the reporting year and the fact that no clear definition exists for 
special districts. 
 38
As we will see, there is no exact definition of what constitutes a special district, and it 
is equally difficult to categorize these units following a uniform criterion, since several 
classifications are possible and none identifies uniquely all types. Table A.3 presents 
some of the possible classifications, based on their relationships to the government that 
created them, their geographical area of influence, their purpose, and their mode of 
financing. 
Different authors have concentrated on various defining characteristics, having to do 
with the act of creation of special districts, their purpose, and their organization ?both in 
terms of their operation and management. Special districts are very much like local 
governments, except that where municipalities provide general public services, special 
districts are narrower in their competences, making the purpose of the entity its 
differentiating feature. Based on the definition provided by the Florida Department of 
Community Affairs, special districts are characterized by operating within a specific 
geographic area, having a governing board with the ability to make policy decisions, and 
being created by a legal act. 
When it comes to telling special districts and public authorities apart, the defining 
characteristics have more to do with organization than with purpose. According to the 
Handbook of Local Government Administration, the key characteristic that sets special 
districts apart from public authorities is that the latter have no elected governing board, 
and often lack taxing power. Leigland (1993) adds to the list of differences the fact that 
special districts are usually smaller and do not use money markets as often. 
The US Census states that special districts ?are established to provide only one or a 
limited number of designated functions and having sufficient administrative and fiscal 
 39
autonomy to qualify as independent governments.?38 According to the 2002 Census, total 
expenditure by special districts (as defined by the Bureau) totaled over one hundred 
twenty-two billion dollars, with the largest portion being devoted to utility expenditure 
(almost fifty billion dollars)39. However, it is possible that this number misrepresents the 
actual size of the activity of special districts in the states, since the Census classifies as 
subordinate governments those that identify themselves as districts (Leigland 1993, page 
393). 
Under any classification, it remains a fact that while the number of state, county and 
sub-county governments has stayed roughly the same in the second half of the twentieth 
century, the number of special districts has risen. Graph 2.4 presents the evolution of 
these types of governments over the last half of the century according to the classification 
of the US Census. 
 
Graph 2.4: Number of Governments
0
20000
40000
60000
80000
1952 1962 1972 1982 1992 1997 2002
Source: US Census
Counties Municipal governments Towns
Special districts School districts
 
 
                                                 
38 US Census, Chapter 3 (Framework of Census Statistics on Governments) of the ?Federal, State, and 
Local Governments Government Finance and Employment Classification Manual? available online at < 
http://www.census.gov/govs/www/class_ch3.html > [Last accessed May 12th 2007] 
39 Census? information on expenditure of Special District Governments by Function, Character and Object, 
2001-02. 
 40
The distinction between special districts and public authorities (PAUs) is not clear, 
and it is not even commonly agreed that they should indeed be treated differently. 
Leigland (1994) provides an overview of the different views taken on the issue. In 
general, corporate status is considered the defining major characteristic of public 
authorities. It is sometimes added that PAUs have appointed boards, possess the ability to 
raise money (though not through taxation),40 and ought to be self-sufficient.41 The reader 
is referred to Leigland (1993) for an overview of the different types of PAUs, their 
purposes and roles in financing public projects, their relationships with the various levels 
of government, and their methods of financing. 
By virtue of their corporate status, PAUs ?separate? themselves from the state or 
local government that creates them and can raise revenues or issue debt independently. 
This act has several consequences, among the most notable the disjointing of the 
government that created the PAU from its liabilities, the increased flexibility in the 
conduct of businesses that comes from disassociating the PAU?s operation from the 
procedures of a state agency, and its exclusion from certain limitations imposed by law.  
However, the bonds issued by these authorities are often tax-exempt, sharing the 
benefit of those bonds issued directly by municipalities. This has not always been 
uncontested. As early as 1941, the Treasury Department tried to collect taxes on interest 
from debt issued by the New York Port Authority, claiming that the authority was not a 
political subdivision and hence its debt was taxable, an argument later rejected by the Tax 
Court (Pryde, 1993). 
 
                                                 
40 This is not a defining characteristic, since often public authorities are conferred taxing power.  
41 Again, this is not a clearly defining characteristic, since public authorities display various degrees of 
dependence from their respective creating government. 
 41
2.2.3. Types of debt 
States and municipalities borrow both for the short and the long term. Since many 
revenues accrue at specific times, and not in a continuous fashion, governments borrow 
to manage their cash flows. This type of borrowing, such as that embodied in notes for 
tax anticipation, revenue anticipation, bond anticipation or tax and revenue anticipation, 
is not however meant to finance long-ranging spending or to supply financing for large 
projects, such as infrastructure. Limits to debt in its many forms are meant to regulate the 
issuance of debt for the longer run, and we do not consider instruments for short-term 
borrowing any further. 
Borrowing can be classified in light of who issues debt: states are sovereign to create 
debt, limited primarily only by their own regulations and policies. At lower levels, 
governments and agencies may be constrained by the rules imposed at the state level but, 
as discussed previously, these constrainys are not uniform. Municipal bonds (sometimes 
known as ?munis?) are issued by governments at any sub-state level. Not only cities, but 
also the debt issued by special districts and public authorities is usually considered to fall 
in this category, which can include both general obligation and revenue bonds and may or 
may not be tax exempt at the federal and state level. Often, the funds raised with these 
bonds are restricted by law to be used for a certain purpose (frequently a capital project), 
but sometimes other uses are allowed. 
An alternative classification looks at bonds from the vintage point of the security 
level they provide, based on the sources committed to the service of the debt and the 
extent of the associated guarantee. Debt limits apply often to some ?but not all- of these 
 42
bonds, and the difference is often created by the type of security they carry, making it an 
important characteristic for our purposes. 
  General obligation bonds (GOs) carry the full faith, credit and taxing power of the 
issuer, who commits its available resources for repayment of principal and interest. The 
projects financed with GO debt are usually meant to benefit the community as a whole, 
and many do not generate revenue. Often, the issuance of these bonds is accompanied by 
the levying of a tax to service the debt (at the state level these taxes are typically sales 
and income taxes, while the local governments use property taxes); in many instances 
voter approval is required before issuance, and frequently the debt has a maximum 
maturity period established by law, although none of these are consistently present across 
all GOs.42  
Revenue bonds, also known as limited obligation bonds, are supported from specified 
revenue sources, usually ?although not exclusively- those deriving from a particular 
project with which the bond is associated. They may also be financed from grants, 
excises or other non-ad-valorem taxes, etc. Henceforth, the general credit of the 
government who issues the bond cannot be called upon to repay the principal and/or 
interest. Unlike GO bonds, revenue bonds rarely require voter approval prior to issuance. 
While other types of debt issuance tend to be generic in the ways in which they are 
presented and paid, revenue bonds are usually very specific in the way they are financed. 
Complicated contracts are sometimes built around them, in part because of the greater 
                                                 
42 Two subcategories of general obligation bonds exist: Unlimited tax GOs are not subject to any (usually 
constitutional) tax millage limit, while limited tax GOs can only pledge the taxing authority up to a limit. 
Sometimes, a separation is made between GOs and full faith and credit bonds, although the two are very 
similar and in many instances are not considered as separate categories. Full faith and credit bonds are 
backed by all funds that are legally available to the issuing authority and are paid out of the general fund, 
although they are not necessarily backed by ad-valorem taxes. In our analysis, we use the terminology 
general obligation bonds and full faith and credit bonds interchangeably. 
 43
risk to lenders since they are tied to a particular stream of revenue. If that revenue proves 
insufficient to service the debt, the debt holder cannot make a claim against all the assets 
of the state.43 
The risk associated with a particular revenue bond depends on the committed revenue 
source (net of costs, if the revenue is available after operating costs are paid); some may 
be very robust, such as those tied to essential services, while others can be riskier. In 
general terms, issuers of revenue bonds usually have to pay higher interest costs than if 
the debt were financed through general obligations. According to Ambler et al (1993) 
only 0.6% of revenue bonds received Aaa or Aa1 ratings (while 2.2% received ratings 
below Baa) in Moody?s Investor Services ratings,44 compared to the 3% of Aaa and Aa1 
(and 0.8% below Baa) of general obligation bonds.45 
 On occasion, the distinction between GO bonds and revenue bonds is so fine as to be 
invisible.  Individual bond issues combine elements of general obligation and revenue 
bonds along a continuum.   Some bonds do not pledge the full faith and credit of the state, 
but go beyond using a stream of revenue as their guarantee.46  
                                                 
43 Several types of revenue bonds exist, depending on the guarantee provided or their purpose. In the first 
group we find, for example, limited tax bonds and special tax bonds -secured by the pledge of a specific tax 
or category of taxes; and special assessment bonds, which are payable from revenues derived from a 
special assessment. 
44 Moody?s rating system assigns grades Aaa to bonds considered to carry minimal risk; followed by those 
assigned Aa1, Aa2 and Aa3. A ratings (either 1, 2 or 3) are consider to be of upper-medium quality; and 
together with the previous two and the Baa (1, 2, and 3) complete the group known as ?investment grade?. 
The speculative grade is comprised of the ratings Ba, B and Caa (in their three scales, corresponding to 
substantial, high and very high credit risk respectively), and Ca and C ?which carry the presumption of 
potential default.  
45 However, these differences may in fact be rather small, since about the same proportion of both types of 
bonds (approximately 9%) were reported to have a rating of Aa. The authors point out that, on occasion, 
revenue bonds actually fare better than full faith and credit bonds, and point to their insulation from 
political pressures, and the fact that the committed revenue streams may be more readily available than 
taxes in the short term due to stronger payment enforcement.  
46 One such example can be found in moral obligation bonds, which are usually used to finance a revenue-
producing facility. The principal source of payment for the bonds is the revenue derived from the 
exploitation of the facility, but the bonds are secured by an agreement that states that if the need arises, the 
proposed budget will include a recommendation to cover the shortfall. However, the legislature is under no 
 44
In terms of the purpose they serve, bonds can be issued for a variety of ends, and a 
complete inventory is beyond the scope of this study.47 There is not a simple 
correspondence between the purpose and the type of associated guarantee, as debt for the 
same purpose may be backed by the full faith and credit of the government, or by a 
committed stream of revenues (for example, debt for affordable housing is issued under 
both types of guarantees).  
This is not an exhaustive list of the types of bonds used by governments, special 
districts and public authorities. It is also important to note that not all types of bonds were 
available at any given point in time. As needs and restrictions evolved, so did the types of 
debt used. A very simplified examination of their evolution over time suggests a 
reduction in the importance of full faith and credit in the overall composition of debt (see 
table A.2 and graph A.1), and an expansion in the types of sources used to fund revenue 
bonds. Combined with the increased importance of non-traditional issuers of debt, the 
matter of debt issuance has become a complicated one indeed. 
                                                                                                                                                 
obligation to honor the recommendation and so the appropriation may not be enacted. Double-barreled 
bonds are guaranteed by the revenues generated from a certain project, but if these are not enough the debt 
is serviced by the general tax revenue of the issuing government. Because tax revenue can be called upon 
for payment of these bonds, they are considered GO bonds. 
47 Other types of bonds are closely tied to private activity or involve the direct engagement of the 
government with the private sector. Some examples of these are lease rental bonds, industrial development 
bonds (often known as industrial revenue bonds) and mortgage revenue bonds. Lease rental bonds are 
issued to build a facility that is leased to the bond issuer, and the amount of the lease used to service the 
debt. Industrial development bonds (IDBs) are public bonds used to help private enterprises to establish or 
update their equipment. Originally fully tax exempt, they were heavily used in the mid-twentieth century, 
although public voices claimed that the benefit of foregone taxation fell into private hands. Pryde (1993) 
goes over the numerous attempts at curtailing these and other types of tax-exempt municipal bonds. As she 
explains, the Revenue and Expenditure Control Act 1968 restricted issuance of public debt for private 
purposes, making municipal bonds taxable if more than a quarter of the funds were to be used for private 
benefit (this provision would be substituted by state caps in the Tax Reform Act of 1986). Debt for some 
purposes was exempt from this regulation, but for others tax-exempt debt was banned (and this list would 
be expanded over the years). Mortgage revenue bonds were used to finance assistance to first-time low-
income homebuyers (these were restricted by The Mortgage Subsidy Act of 1980) and multifamily bonds 
are used to finance rehabilitation of buildings under the condition that certain percentage of the units are 
reserved for rent by low-income families. The primary sources for debt service of these bonds are the 
mortgage and rental payments derived from the housing units, but some of the bonds issued for multi-
family mortgage funding are ultimately backed by the federal government. 
 45
2.3. Evolution of debt restrictions 
Restrictions on the state officials? capacity to issue debt are frequently presented as 
fixed, and often the argument is made that they have existed in the original constitutions 
of the states ?unchanged- since the time these states entered the Union, or that restrictions 
to debt appeared rather suddenly and homogeneously across states to limit directly the 
amount of debt a state could issue or have outstanding at any given point in time. These 
ideas suggest a disassociation of debt restrictions from other fiscal institutions that 
developed later. Balanced budget requirements, tax and expenditure limitations and rainy 
day funds all acquire a new perspective when put in the context of the different types of 
debt restrictions. The commonly used definition of balanced budget rules includes 
provisions that, when investigated more closely, refer to debt rather than balanced 
budgets themselves. Because these institutions are often considered exogenous and 
independent of debt restrictions, this relationship is often overlooked. 
However, as we can see in graph 2.5.A, the history of constitutional debt provisions 
does not reflect this static view. The graph presents the introduction of debt restrictions in 
state constitutions over time. To avoid overcrowding of the graph, every point in the 
horizontal axis represents a decade (noted in the graph by its starting year). Debt 
restrictions are grouped in five categories. From left to right, the white bars represent the 
?absolute? debt limits (independently of whether there is an associated procedure or 
modification provision), and include cases where the limit is set at zero (no debt is 
allowed).  The second bar, with a checkered pattern, counts the cases in which limits on 
debt for casual deficits were introduced (without considering associated procedures or 
modification provisions). The third category (represented with a bar crossed by horizontal 
 46
lines) includes the cases in which a ?relative? debt limit was established (again, ignoring 
for the time being, whether there is an associated procedure or modification provision). 
The fourth bar (crossed by sloping lines) counts provisions that consist of a procedural 
restriction (whether the procedure requires voter and/or legislative approval). The last 
group is represented with a black bar, and includes the number of states that started 
allowing itemized debt in that decade.  Less than five bars appear if there were no 
introductions of a certain type of restriction in the corresponding decade. 
 
Graph 2.5.A- Constitutional debt provision, by type
0
2
4
6
8
10
12
18
30
18
40
18
50
18
60
18
70
18
80
18
90
19
00
19
10
19
20
19
30
19
40
19
50
19
60
19
70
19
80
19
90
In
cid
en
ce
absolute relative procedure debt item
 
 
.  Even in the reduced form of the graph, we can see that debt provisions were not 
enacted simultaneously, or even homogeneously, through history. However, graph 2.5.A 
does not reflect all the subtleties that come into play when we consider that limits may 
(and most often are) accompanied by procedural restrictions. In graph 2.5.A, a state is 
classified as introducing a procedure only if it does not have a limit, but many states have 
limits and procedures. Graph 2.5.B completes the picture presented in graph 2.5.A by 
providing an overview of the distribution of procedural restrictions and modification 
 47
provisions in limits. Every state is counted at most once in every bar, for example, a state 
that has 3 provisions for absolute limits at different points in time is included as one 
observation in the corresponding bar. If the state had also a relative limit at some point in 
time it appears once in the ?absolute? bar and the ?relative? bar. As we can see from 
graph 2.5.B, very few states have only limits (either absolute or relative).  
 
Graph 2.5.B. Modification procedures and procedural 
restrictions qualifying states' debt limits 
(whole period, by type, not repeats by state)
0%
50%
100%
None 1 2 0
Both 7 4 6
Modification 5 8 5
Procedural 3 8 5
Absolute (casual deficit) Absolute  Relative
 
 
In what follows, we present an evolution of debt restrictions that illustrates several 
principles. First, today?s restrictions on debt are the result of an evolution, rather than a 
single creation act. The first restrictions on debt were created in response to a specific 
problem, but were not meant to forbid debt or even to place a fixed maximum on the 
allowable amount. As we will see, restrictions on debt extended to other states and, as the 
circumstances changed, took different forms and evolved into the present-day limitations. 
Secondly, the creation of other fiscal institutions (as well as a change in the composition 
of the nature of debt and its issuers) is linked to debt limits, both at the state and local 
 48
level. The evolution of debt restrictions is a coherent process rather than a series of 
random changes. This fact does not become obvious through simple description, but 
requires careful definition of the elements involved and a systematic approach to the 
investigation of the changes. More than a story, the evolution of debt restrictions is better 
understood as a series of changes brought about by the pre-existing conditions. For this 
purpose, separating the history of debt restrictions into a set of distinct stages is useful. 
By doing so, we can see that changes in debt restrictions happened, in part, as a response 
to the existing limits, but also as part of an evolution of their own, where problems and 
opportunities alike surfaced with the new changes.  
 
First regulations on debt at the state level: 1840s to 1879 
Debt restrictions first appeared in state constitutions after states defaulted on 
obligations contracted mainly for investing in financial and transportation infrastructure 
(Wallis, 2005). Because private investors were unable to underwrite the large costs of 
constructing railroads and canals that were considered to be fundamental for economic 
development, state legislatures took positive action to provide such infrastructure. For 
this purpose, states borrowed directly,48 and subsidized and guaranteed loans to private 
enterprises. When these projects failed, states found themselves in a difficult situation, 
aggravated by the banking crisis of 1839, and many of them defaulted on their 
obligations.49 Foreign investment in the United States slowed down, borrowing became 
                                                 
48 By 1836, all but eight states had contracted debt for this purpose (Ratchford, 1941; Wallis, 2005, Wallis, 
Sylla and Grinath, 2004; Sterk and Goldman, 1991). 
49 Maryland, Illinois, Indiana, Michigan, Mississippi, Louisiana, Arkansas, Florida and Pennsylvania 
defaulted on their bonds, either by repudiating or adjusting their debt.  
 49
more difficult in the states that had defaulted, and in some instances excise and property 
taxes had to be instituted (Kiewiet and Szakaly, 1996).  
As a result of this experience, many states considered it necessary to establish rules 
that would prevent such problems from recurring, while not necessarily forbidding state 
borrowing.50 In fact, no constitutional debt restrictions existed before 1839, but twelve 
states had adopted such rules by the end of the 1850s, and eight more would do so before 
1860. Wallis and Weingast (2006) explain this wave of adoptions as a movement away 
from taxless finance.51 Under the new provisions, ?voters now had to approve each new 
debt issue and tax increase prior to new debt issues.?52 It was precisely states with higher 
levels of debt that adopted constitutional limits to control borrowing. 
With these financial concerns in mind, many states adopted other type of restrictions 
on their budgets and introduced rules aimed at providing a certain sense of balance in the 
budget.  
The Advisory Commission of Intergovernmental Relations (ACIR), whose data are 
widely used in empirical analysis, used these constitutional provisions to calculate their 
measure of stringency of balanced budget requirements. However, these were not 
balanced budget rules in the precise modern meaning of the term. Because ACIR lumped 
together what are virtually restrictions on debt and balanced budget requirements, its 
indexes do not measure correctly the true rules that require balancing of the budget. Also, 
this misuse provides an incorrect sense of when these instruments were created. Balanced 
                                                 
50 The exception is Arkansas, where prohibition on debt is established in 1874 in article 16 section 1 of the 
constitution. 
51 Very simply put, this mechanism tries to pay for projects without raising taxes by issuing debt to be 
repaid from the revenue generated by the project. However, if the project does not generate enough 
revenues to service the debt, states have to pay for the debt. 
52 Wallis and Weingast (2006), pages 25-26. 
 50
budget rules are often taken to date from this period, while in reality, ?true? balanced 
budget requirements date from a much later time. Table A.7 provides a sense of the 
difference between rules that are aimed at balancing the budget and rules that deal with 
debt, classifying these provisions in two main types: true rule and other rules.  ?Other? 
rules includes restrictions that dealt directly with an aspect of the budgetary process, by 
establishing a procedure that had to be followed or requiring equality between 
appropriations and expenditures: ?er? in the table. Other states regulated items that had 
consequences for the final status of the budget, but did not directly regulate the budget 
process, such as the requirement that expenditures be kept within the limits of revenues, 
or debt restrictions: ?d? in the table. In contrast, ?true? balanced budget rules actually 
required the state to implement a real budgetary process.   One type of true rule required 
that appropriations not be greater than expected revenues: ?ai? in the table.  Modern 
balanced budget rules specify distinct procedures through which the state writes and 
considers its budget: ?p? in the table.  Most of the restrictions adopted during this period 
correspond to the ?other? kind of balanced budget provision. As shown in table A.7, it is 
only towards the end of this period that we see a few rules aimed directly at providing 
budgetary balance, but the majority (and all minus one of the ?balanced budget? 
constitutional provisions established before 1968) have to do either with debt or a rule 
linking expenditures to revenues.  
As shown in graph 2.6, the first wave of debt restriction adoptions consists mostly of 
absolute limits to debt issuance, either to casual deficit or overall debt, combined with 
procedural restrictions. Of the states that adopted debt provisions at this time, several 
 51
included procedural provisions to issue debt beyond their stated limit, and four also 
started itemizing debt during this period.  
 
Graph 2.6. 1839-79: Distribution of new 
constitutional provisions on debt
43%
31%
6%
9%
11% Limit, without procedural restriction
Limit, with procedural restriction
Voter approval
Legislative approval
Debt item
 
 
A complete list of state debt limits, with their corresponding procedural restrictions 
and modification provisions can be found in table A.4. If a state has a limit and a 
procedural restriction, the procedural restriction is typically part of the same 
constitutional provision that creates the limit, usually requiring voter approval or some 
type of legislative supermajority to issue debt beyond the limit they specify.53 On the 
other hand, modification procedures are usually to be found in different articles of the 
constitution, sometimes years after the original limit was put in place. These include the 
upward revision of the limit and the approval of debt items not subject to the original 
limit.  
The growth in state debt did indeed slow down after the 1840s and, as we can see in 
table A.1,by 1880 state debt as a share of total public debt had fallen to less than an 
                                                 
53 Although not included in table B.4, often these modification procedures include provisions for maturity 
limits of the debt and the requirement to raise taxes or otherwise provide for the payment of debt. Full 
classification of these additional ?operational? requirements is currently underway.  
Total number: 35 
 52
eighth of its 1838 value. After the Civil War, state debt increased, in part due to costs 
associated with destroyed infrastructure, damaged capital, and the high human costs that 
affected the labor force. Recovery was a difficult task, especially among the confederate 
states. Ratchford (1941) provides a description of the financial difficulties of Southern 
states at this time, which resulted in large amounts of debt that states had great difficulty 
repaying.  In the 1870s several southern states defaulted on their debts, and in the 1870s 
and into the 1880s, southern states began imposing debt restrictions. 
In this period, the first restrictions on debt are established as a response to a crisis that 
resulted in defaults in several states. However, their main objective was not to forbid 
debt, but to avoid what were perceived as systematic problems in the way institutions 
worked. Many of the new constitutional restrictions on debt called for taxes to be raised 
for funding debt, and required the voters to approve issuance of new debt. The share of 
state debt in the total fell sharply; by the end of this period, local government was 
growing in importance in public finances, a tendency that became stronger in the next 
period.  
 
1880 to 1929:Modifications of debt restrictions, extension to local governments, debt 
restrictions extend to the South, and their scope is redefined. Decline of state debt and 
rise of local debt. 
The constraints imposed during the Congressional reconstruction turned out to be 
insufficient or ignored in practice. When Southern states changed their constitutions in 
the late 1870s and into the 1880s, many included provisions to constrain debt.54 Also, 
                                                 
54 All the former confederate states (save Arkansas) included provisions for limiting debt in their 
constitutions (Wallis and Weingast, 2006, pages 32-33) 
 53
several states significantly modified the nature of their debt restrictions55 or altered their 
rules without changing their nature.56  
However, states were still in need of financing and, as they became more constrained 
in their ability to become indebted through full faith and credit debt, they looked for other 
ways to meet their financing needs. In this period, most of the constitutional activity 
dealing with debt provisions was to include debt items, or to establish limitations with 
associated procedures for debt issuance. As we can see in graph 2.7, these two categories 
together make for more than 70% of all the provisions in this period.  
 
Graph 2.7. 1880-1929: Distribution of new 
constitutional provisions on debt
18%
29%
7%4%
42%
Limit, without procedural restriction
Limit, with procedural restriction
Voter approval
Legislative approval
Debt item
 
 
But not only the form of the constitutional provisions on debt was changing. Faced 
with constraints on debt, legislatures soon started reconsidering and reshaping what 
                                                 
55 Louisiana substituted its absolute debt limit (which had a provision that allowed for more debt if the 
legislature voters approved the debt issuance) for a prohibition on debt, Alabama eliminated the restriction 
that debt was used to cover casual deficits and increased the overall cap, Virginia adopted legislative 
supermajority as a procedure to allow debt for casual deficit beyond the established maximum, and North 
Carolina incorporated the admissibility of debt without voter approval up to a relative limit that could still 
be surpassed through the voters? consent. Within this period North Dakota substituted its 1889 limit on 
casual deficit for a relative limit in 1918, while Idaho switched form an relative restriction to an absolute 
one in 1912, although in both cases the real limit is given by the possibility to issue debt for specific items 
if a majority of the voters authorizes it; and Utah expanded its 1895 casual deficit limit to allow for debt up 
to 1.5% of the taxable property of the state 
56 Michigan changed its limit for casual deficit, and Rhode Island did the same for its general absolute limit. 
Total number: 28 
 54
?debt? (as limited in their constitutions) meant, often limiting the application of the 
restriction to general obligation bonds only. These reviews of the term and its scope were 
often amended by state courts, which in many instances ruled that bonds payable from 
resources other than the general fund were not debt in the strict sense, and hence were not 
subject to the restrictions specified in the constitutions.57  
States also turned to local and municipal governments to provide the necessary means 
for the financing of projects. At the same time, increased urbanization and 
industrialization translated into higher needs for services and infrastructure at the local 
level, and the combination of these factors led to an increase in sub-state borrowing.58 If 
states had been responsible for the provision of most of the public infrastructure 
previously, local governments became a much more significant actor in public finances in 
this period. This pattern can be seen very clearly in table A.1, which shows large 
increases in the share of local debt during this time. It had taken roughly 40 years for the 
share of local government debt to double, from about 12% in 1838 to roughly 25% in 
1880. Within the next 20 years, it would more than double again, with the share of local 
government in total debt reaching almost 60% in 1902, and rising still further to over 
70% in 1913. By this time, the share of local debt is about 10 times its state counterpart, a 
tremendous change compared to 1838, when the share of state debt was 7 times as large 
as that of the local governments. 
                                                 
57 Since limits applied to debt backed by the full faith and credit of the state, borrowing that was not 
supported in such a fashion was not to be taken under consideration in calculating the amount of debt 
subject to limit a state had at any given point in time. Ratchford (1942), however, claimed that in some 
instances the definition was rather stretched, and payment for these ?debentures? or ?certificates of 
indebtness? was to be provided by the revenues of special, separated funds (Ratchford, 1942, page 460), 
giving rise to the Special Fund Doctrine. 
58 Wallis and Weingast (2006) find that debt limits at state and/or local government reduce the amount of 
borrowing in their corresponding jurisdiction, and local borrowing seems to be higher where debt limits 
exist at the state level, although it is not possible to assert the causal relationship unequivocally.  
 55
In time, many municipalities made some of the same mistakes that had led to financial 
difficulties at the state level, and restrictions on local debt were imposed between 1865 
and 1880,59 often tying local debt to the value of taxable property within its jurisdiction 
and requiring voter approval for debt issuance. As a result, local governments turned to 
non-general obligation debts not covered by the limits, such as revenue bonds and special 
assessment debts. An additional response to the constraints imposed by the constitutional 
rules was the diversification of types of local governments. With the creation of 
specialized units, a better match between the payers and beneficiaries of a project was 
possible. This allowed this new type government unit to issue debt when agreement by 
the relevant voters was required. As we discussed in section 2.2.2.3, these ?agencies? are 
able to separate themselves from the state or local government that creates them and raise 
revenue or issue debt independently. This has been seen as a way for governments to ?get 
around? the constraints imposed by the state constitutions, especially in the legal 
literature. While this motive may in part explain the rise of these new issuers of debt, the 
authors that claim this is the only (or even the main) motive, may be missing part of the 
picture. The key element to explain the missing part is the difference between debt limits 
and debt restrictions. The constitutional provisions that are being ?circumvented? did not, 
for the most part, impose absolute limits on debt, but rather required the appropriate 
authority generate sufficient electoral support among its constituents to issue debt backed 
by full faith and credit.  
These two trends, the usage of revenue bonds and the new types of governments, can 
be understood partly as reactions to the restrictions in place, but they also constitute 
another step in the process of change for the betterment of the system. They both create 
                                                 
59 According to Wallis and Weingast (2006), 36 states had adopted some type of local restriction by 1890. 
 56
problems (such as loss of transparency) but also new opportunities (for example, the 
financing of projects that would otherwise not be undertaken). It is in the next period, 
however, when both the usage of revenue bonds and the expansion of new types of 
governments fully develop. 
 
The 1930s, 1940s and 1950s: crisis, revenue bonds and involvement of the federal 
government 
With tightening economic conditions, many municipalities started defaulting on their 
special assessment bonds after 1926. The Great Depression dramatically aggravated the 
economic problems, and approximately 12% of municipalities with population above 
30,000 were in default by 1934 (Lehman (1936)). Ratchford (1936) argued that these 
defaults ?paralyzed the services of local units and (threw) a staggering load on state and 
federal government,?60 and added that the bad reputation of defaulting governments 
affected the market for bonds for the rest of municipalities. The associated costs were 
even higher due to the legal requirement that an agreement by all creditors be reached 
before any proposed refunding plan could be carried out, which implied that any small 
investor could effectively block an otherwise agreed-on plan. The Federal Municipal 
Bankruptcy Act of 1934 tried to facilitate the readjustment by compelling minority 
bondholders to accept agreements between a majority of the creditors and the debtors 
under certain conditions, but the Supreme Court declared it unconstitutional61, 
considering it undue interference of the federal government in state sovereignty. This 
                                                 
60 Ratchford (1936), page 71. The tense of the verb in parenthesis has been altered. 
61 Ashton v. Cameron Company Water Improvement District N.1 (1936). After Congress modified the Act, 
the Court deemed it constitutional in United States v Bekins (1938). 
 57
problem would be addressed in 1937 with the introduction of a federal law that included 
provisions to deal with municipal bankruptcy.62  
Something about these problems merits particular attention. It is the very different way 
in which they are addressed, compared to the approaches taken before. Rather than being 
constrained by a new wave of restrictions, subnational governments find that the federal 
government becomes more involved in state and local finances, financing assistance to 
poor households, farmers and even actively buying subnational government bonds. States 
were encouraged to borrow to be able to take advantage of matching grants offered by the 
national government, and revenue bonds were a convenient way to raise money. Often 
the need to borrow rapidly made the issue of voter-approved debt (a requirement 
common in many constitutionally imposed rules for full faith and credit borrowing) 
impractical, and revenue bonds provided an expedient alternative, accepted for loans both 
by the Reconstruction Finance Corporation (RFC) and the Public Works Administration 
(PWA) (Ratchford, 1941). 
Local governments kept on issuing debt, increasingly relying on revenue bonds63 
which did not pledge their full faith and credit, and in time special districts and public 
authorities started issuing debt that was not considered to be under the limit of either state 
or local limitations. Much has been argued about the suitability of allowing these local 
entities to borrow. Already Secrist (1914) presented his doubts about the appropriateness 
of allowing unlimited borrowing power to a corporation only because it was coined as 
independent, when the purposes of its debt could be easily squared with a political district 
                                                 
62 Chapter 9 (Adjustments of Debts of a Municipality) of the United States Code, available online at  
< http://uscode.house.gov/download/pls/11C9.txt > [Last accessed on May 5th 2007] 
63 According to Pagano and Perry (2006), page 8, about 5% of the tax-exempt debt in the US was in the 
form of revenue bonds in the early 1930s. By 1960 near 40% of total local and state debt was non-
guaranteed, a percentage that would double to 76% by the mid-nineties.  
 58
that would otherwise be issuing debt under the appropriate regulations. He did not, 
however, argue for the complete prohibition of these entities but rather that borrowing 
authority ?when it is conferred it should be adequate to the purposes at hand? (Secrist 
1914, page 378). 
In 1921 the New York Port Authority was created, and more such authorities followed 
during the Great Depression (for water, sewer, roads, bridges, housing, etc), encouraged 
by the favorable treatment granted by the federal government (Pagano and Perry, 2006). 
Often these public authorities charged for their services and so could issue revenue-
backed bonds, or if the charging of fees for services was not possible (such as debt for 
constructions of jails and office buildings, for example), the public authority could enter 
into a lease agreement in virtue of which the rent paid by the government to the PAU 
could be considered a revenue and used to issue revenue-backed bonds.  
New types of bonds were also being created during this period, such as the Industrial 
Development Bonds64, but this was not a period of high activity in terms of constitutional 
debt provisions at the state level. Two states modified their debt provisions following the 
Great Depression: Arkansas (the only state to have defaulted during the Great Depression 
(Cohen, 1993)) allowed debt issuance with voter approval; and North Carolina required 
two thirds of the previous year?s debt be repaid before issuing additional debt. In 
addition, New Jersey changed its absolute debt limit for a relative one based on revenues 
in 1947 ?a change that took place in several other states in the following decades. When 
                                                 
64 Industrial Development Bonds (IDBs) were first tried out in 1929 in Columbia, Mississippi, where 
citizens signed promissory notes to guarantee the money necessary to build a factory for Reliance 
Manufacturing Company, which would in exchange set a plant in the town. The initiative was extended to 
the whole state under the Balance Agriculture with Industry program in 1936. However, these types of 
bonds were not to become popular until mid-century, and their rapid increase in popularity would bring 
about restrictions by the end of the 1960s. 
 59
Alaska and Hawaii entered the Union in 1959, both included debt limits in their 
constitutions. Furthermore, two new states (North Dakota and Ohio) included debt 
itemization in their constitutions. 
As we can see in table A.1, the level of state indebtness leveled off during World War 
II, due to the concentration of resources on the war effort, but it started growing again 
after war?s end due in part to spending on highways and more state involvement in the 
financing of education. By the middle of the century, state debt had kept growing, more 
municipal revenue bonds were issued, and new special districts and public authorities 
were being created (Pagano and Perry, 2006).  
Even though there was only modest constitutional activity regarding debt restrictions 
in this period, there took place a consolidation of two important trends that will have an 
impact in the decisions that will be taken in the next stage. First, the use of revenue bonds 
increased significantly: although table A.2 has a gap in information during the nineteen 
forties and fifties, but we can see that between 1937 and 1966 the proportion of revenue 
bonds of total debt expanded sevenfold. Related to this expansion in the usage of revenue 
bonds, we can observe a sharp increase in the number of special districts and public 
authorities65.  
Debt issued by special districts and public authorities, or in the form of revenue bonds, 
is often outside the scope of the constitutional debt restrictions established for state and 
local governments. As these forms of debt grow in importance, a smaller share of total 
borrowing falls under the scope of these restrictions. Even if little constitutional activity 
                                                 
65 From graph 2.4 we can see that, by 1957, only the number of special districts and municipal governments 
had increased since they started being counted by the US Census, and the increase in the number of special 
districts dwarfs that of municipal governments.  
 60
happens during this third period, the developments that take place set the stage for the 
changes that will happen in the next period. 
 
 Later part of the twentieth century: increased flexibility, new institutions 
In this last period we can observe a distinct shift in the nature of debt restrictions at 
the state level towards greater flexibility. But this is followed by new measures aimed at 
restricting their flexibility in budgetary issues. These changes are discussed in more detail 
at the end of this subsection. 
With the economy?s continued growth over the fifties and sixties, federal aid on the 
rise, and increased tax revenues derived from expanded tax bases or even the introduction 
of new taxes, states had relatively little trouble keeping up with the requirements to 
balance their budgets. But as revenue growth began to slow down in the seventies and 
early eighties, states experienced fiscal problems. At the same time state involvement in a 
variety of redistribution programs -such as Medicaid and AFDC (Aid to Families with 
Dependent Children, which had been created in 1935) - demanded funding, while aid 
from the federal government remained stagnant. As we can see from tables A.1 and A.2, 
although its share in overall debt remained roughly constant, the amount of state debt 
grew by almost 200% between 1972 and 1982. This increase was not homogeneous 
across debt categories, since non-guaranteed debt took up the larger part of the total share 
of states? obligations.66  
                                                 
66 A new development in local financing come in the form of tax increment financing (TIF), which 
appeared for the first time in California in 1952 and spread to the rest of the states (they became widespread 
in the late seventies, and currently all states except for Arizona have adopted TIF) as funds for 
infrastructure from the federal government diminished and local governments saw their issuance of tax-
exempt bonds reduced by the Tax Reform Act (TRA) of 1986 while their responsibilities in these areas 
increased. As some types of debt, TIFs tap into the future gains of a project to finance it. When the project 
is completed, it will generate an increase in existing property values and/or new investment, increasing tax 
 61
In the mid to late 1980s, a number of factors caused fiscal difficulties for state and 
local governments. The market for tax-exempt bonds became increasingly complicated 
with the addition of numerous regulations and standardizing principles (Mikowski, 1993), 
while the Tax Reform Act (TRA) of 1986 restricted the use of tax-exempt bonds by 
limiting the number and types of bonds exempted from federal taxation; this increased 
the administrative and effective cost of borrowing. Several Supreme Court rulings upheld 
the right of Congress to pass legislation in areas that had previously been considered to 
be the competence of subnational governments, effectively reducing the fiscal freedom of 
the states and municipalities (MacManus, 1990).67 In addition, devolution to states and 
municipalities of various fiscal responsibilities brought about increased fiscal pressures in 
the form of unfunded mandates for expenditures, interest rates increases, and federal 
government reductions in the financing of infrastructure.  
In this increasingly difficult setting, local governments resorted to revenue bonds; in 
addition, tax-exempt debt for private purposes and non-guaranteed bonds became 
important instruments to attract industry to jurisdictions and fund public endeavors (Bahl 
and Duncombe, 1993). As we can see from graph A.1, in 1991 about 65% of all 
municipal debt issued was in the form of revenue bonds. However, the fact that states and 
local governments have created new forms of debt does not necessarily mean a lack of 
concern for debt-related matters. In fact, debt management policies are not unusual; many 
                                                                                                                                                 
revenues. Based on this premise, bonds are issued for the construction of this project, to be repaid out of the 
tax gains generated. Although TIF was in its origin directed towards promoting the growth of 
underdeveloped areas its use was later extended to the financing other types of projects since, according to 
a 2000 study by the National Association of Counties (as quoted by Amt, 2000), more competition for 
investment has led to the financing of an increasing range of projects through TIF. 
67 McManus cites rulings attributing Congress the authority to regulate the terms of employment 
subnational governments? employees, conferring it the power to regulate tax-exempt bonds, and limiting 
the power of these units to contract. (McManus, 1990, page 23) 
 62
states have appointed supervisors and boards to oversee debt issuance and insure the debt 
load is appropriate and sustainable. 
Local borrowing kept growing after the crisis of 1991, and both the amount of full 
faith and non-guaranteed bonds peaked in 1998 before decreasing in 1999 and 2000. 
However, as pointed out by Cohen (1993), when analyzed in terms of capacity the 
increase in state debt is not so significant: in the early nineties the ratio of debt to 
personal income was about the same as in the sixties, and even below that of the 
seventies. 
Decisions on infrastructure involve many governments; local governments have the 
primary responsibility for the design, construction and maintenance of many 
infrastructure projects (such as transportation, water, garbage and sewage facilities and 
systems), but the federal and state governments play important roles in funding and 
coordination. As we have seen, some of this complexity has been addressed by shifting 
provision to private hands and by compartmentalizing users to match payment with use, 
as well as with the use of new debt instruments.68  
The search for ways to avoid having to raise taxes to pay for all or a major part of the 
cost of projects has not ceased. For example, since stadiums were taken off the list of 
purposes admissible for issuing tax-exempt bonds under the form of private-activity 
                                                 
68 A somewhat new trend in debt issuance has to do with the use of expected cash from future federal grants 
to back revenue bonds, expanding the trend to issue debt backed not by the full faith and credit of the 
government, but by a committed stream of revenues. In this, they are not different from other forms of 
financing that are in use by subnational governmental units. As with the other revenue-based forms of debt 
financing, their success depends on whether the expected revenues will realize themselves in a timely 
manner. Lease-purchase agreements allow municipalities to commit to make appropriations every year 
from the stream of revenues deriving from taxes for the payment of a lease on a piece of equipment or 
infrastructure. Certificates of participation are issued, and the lease amounts are used as guarantee. Once 
the leasing period is over and the certificates are redeemed, the property reverts to the government. With 
few limitations on this type of arrangement, municipalities can use it freely to acquire buildings and other 
equipment without increasing the amount of debt capped by their legal limits.  
 63
bonds, it had become difficult to finance these projects. In 2006, the IRS ruled that in the 
case of a stadium in New York its associated revenues were Payment In Lieu Of Taxes69 
(PILOT or PILT): taxes, rather than revenues from private activity, which makes debt 
issued for this purpose governmental bonds (and hence tax exempt) rather than private 
activity bonds. A similar arrangement allows the Port Authority of New York to make 
payments in lieu of taxes for its ownership of the World Trade Center site. 
In terms of constitutional provisions on debt, in the late nineteen sixties and early 
seventies, we can observe a movement to alter what may have been considered obsolete 
or excessively constraining constitutional restrictions on debt. During this period, Florida 
(1968, 1970), Illinois (1970), Louisiana (1974), Mississippi (1960), Montana (1973), 
Pennsylvania (1969), Virginia (1971), Washington (1972), and Wisconsin (1969) 
modified their constitutions to allow for more flexibility in the issuance of debt by 
shifting to less stringent restrictions, including supermajority approval as a way to issue 
guaranteed debt, or explicitly excluding revenue bonds from the debt limit. This pattern is 
captured in tables A.5 and A.6. We may note that an important characteristic of this shift 
among several states is the relaxation of the requirement to attain voter approval for debt 
issuance: public officials are recovering part of the flexibility they had given up with past 
restrictions. In the three cases that require voter approval before issuing debt, this is done 
as a way to create a procedural restriction where only a limit existed previously. Graph 
2.8 shows the distribution of the changes in constitutional provisions regarding debt, 
including the appearance of debt itemization in some states.  
                                                 
69 PILOTs were original tax compensations from the federal government to local government because of 
the use of lands (and hence loss of taxes). 
 64
The trend towards greater flexibility is important: After over one hundred years of 
establishing different types of restrictions on debt, states have moved away from rigid 
rules and have decided to embed more flexibility in the ways in which they allow 
themselves to issue debt. Moreover, this has taken place in a relatively short period of 
time.   
 
Graph 2.8: 1960-today: Modifications of debt provisions 
35%
25%
15%
5%
5%
15%
Change from absolute to relative
limit
Remove voter approval or
subsitute for legislative approval
Create modification procedure 
Change from procedural restriction
to limit
Eliminate limit and retain
procedural restriction
Debt item
 
 
Shortly after these changes, we observe a time of intense institutional activity focused 
on fiscal matters at the state level. While there was a movement towards relaxation of the 
provisions that restricted the issuance of debt, other fiscal institutions were created to 
discipline state finances.  
Table A.7 shows the adoption since the mid nineteen-sixties of several constitutional 
rules that established requisites for some sort of budgetary balance. What makes this 
wave of adoptions interesting is that, rather than referring to debt or to an equilibrium 
between expenses and revenues, most of the provisions directly address budgetary 
Total number: 20 
 65
balance. The rules regulate directly a relationship between appropriations and 
expenditures or establish detailed budgetary procedures aimed at balancing the budget. 
Of the nine states for which we have data indicating the adoption of a ?true? balanced 
budget requirement (that is, states that include in their constitutions a defined procedure 
for their budgets or that demand an equilibrium between appropriations and expenditures) 
in this period, six of them also modify their constitutions to introduce more flexibility in 
their debt restrictions (see table A.10.)  
Tax and expenditure limitations (TELs) were established in this period of economic 
uncertainty under the premise that they would help constrain the size of the government. 
The tax revolt that followed the passage of California?s Proposition 13 in 197870 
translated into the adoption of limits to the taxing and spending powers in several states. 
These limits, paired with the constitutionally imposed restraints on debt and the 
requirements to present some sort of balanced budget, increased the difficulties of the 
states in raising enough revenue to meet their spending needs. This trend was further 
strengthened by the popularization in the nineteen-nineties of supermajority requirements 
for increases in certain types of taxes.  
Starting in the mid-eighties, rainy day funds (RDFs) -also known as budget 
stabilization funds- became a popular instrument for budget stabilization purposes in the 
states, although the first rainy day fund had been in place for a number of years.71 These 
funds interact with the existing institutions in a variety of ways: the use of accumulated 
reserves during economic downturns reduces the necessity of increasing taxes (which 
                                                 
70 California?s Proposition 13 is not often considered a TEL because it only restricted a specific type of 
taxation, rather than overall tax revenue. See Rodriguez-Tejedo (2006) for more information on TELs. 
71 New York adopted the first rainy day fund in 1945, but the second RDF adoption did not come in almost 
15 years (1959, Florida). The rest of the existing RDFs were all adopted during this period, furthermore the 
bulk of them were enacted in the mid-eighties. 
 66
may be problematic in the face of TELs) or otherwise raising revenue to meet the 
requirements of a balanced budget. States that had built demanding RDFs were able to 
deal better with the fiscal stress associated with the crisis of 1991 (Sobel and Holcombe, 
1996) and even though they ?are unlikely to bail out a state in serious fiscal difficulty ? 
they can serve as a stop-gap measure to buy time.?72 Table A.9 relates the dates of 
adoption and main characteristics of RDFs73 to the type of state constitutional debt 
provisions. From table A.9 we can see that, within the group of states that modified their 
constitutions to introduce greater flexibility in their debt restrictions during this period, a 
higher percentage of states adopted RDFs with strict withdrawal requirements (about 
40% compared to 24% in the group where not such changes took place.) 
The debt restrictions that were established in the 1840s were quite different from 
those present in states today. In the fashion of what Wallis refers to as ?recursive 
institutional change?, the evolution of debt restrictions has caused, and been caused by, 
changes in the levels of government involved and the types of debt.  Frequently, casual 
observers see only one direction in this process: restrictions on debt have caused the 
proliferation of PAUs and special districts, and have favored the creation of a large 
variety of types of non-guaranteed debt. It is rarely mentioned that, because of these 
changes and the economic and political circumstances of the states, debt restrictions 
themselves have evolved. In the second half of the twentieth century, the proliferation of 
issuers of debt and types of bonds is so large that some argue that debt restrictions do 
nothing to limit debt.  
                                                 
72 Fundamentals of Sound State Budgeting Practices, NCSL 1995, page 13. 
73 The structure of RDFs and their characteristics are described in more detail in chapter 3 of this 
dissertation. 
 67
And yet, rather than hardening debt restrictions, we can see state constitutions 
increasing the flexibility of their constitutional provisions regarding debt in this stage. 
This is not, however, the only institutional activity happening at the end of the 20th 
century. Other institutions aimed at restricting the fiscal activities of the state (tax and 
expenditure limitations, and balanced budget requirements that directly deal with 
particular aspects of budgeting) have been put in place. This is the most recent step in a 
process that requires both the clear definition of the types of debt restrictions and a 
relatively long perspective on their evolution. The changes we observe are responses to 
the set of circumstances created by the economic and political make-up of the states, and 
by the conditions brought about by previous institutional choices.  
 
2.4. Conclusions and further work 
Restrictions on debt are not exogenous or immutable; they did not appear at the same 
time or in a homogeneous form in all states; nor were they in general created to ban or 
even limit debt with absolute rigidity. The constitutional provisions that create them are 
sometimes misinterpreted by concentrating only on the part that mentions a limit and 
leaving out procedural restrictions or by ignoring other parts of the constitution that 
enable states to issue debt outside the scope of the limitation. 
We have tried to provide a comprehensive treatment that takes all the aspects that 
constitute a debt restriction into account to facilitate the description and understanding of 
these rules.74 Besides the distinction between debt for casual deficits and debt for the 
                                                 
74 Arguably, two additional aspects are sometimes present in the qualification of debt limits: the maturity 
limit of the debt and whether a requirement exists to provide for funds to finance the debt. The larger 
 68
longer run, we make a separation between limits (that create a maximum allowable 
amount of debt, defined either in absolute or relative terms) and procedural restrictions 
(that establish a set of requirements that need to be satisfied to issue debt but do not 
impose a maximum on the amount of debt that can be issued). In addition, we have 
examined modification procedures, such as those that increase the limit, and the 
authorization of debt for certain items that allow a state to issue more debt than originally 
permitted by the limit. Ultimately, debt restrictions ?like any other constitutional 
provision- can be altered by modifying the constitution. Changes have indeed taken 
place, and some of them would not be evident if we restricted the definition of ?debt 
restriction? to the dichotomy of whether a provision of any kind exists regarding debt 
issuance in the state or ignored the modification procedures associated with debt limits. 
The creation of the first constitutional provisions restricting debt is concentrated in 
time around a very specific economic occurrence in the 1840s. Towards the end of the 
19th century, states start including debt itemization in their constitutions. The restrictions 
spread to other states, reaching many of the Southern states in latter part of the 19th 
century. States continue to borrow, but the late 19th century saw a marked increase in 
local borrowing.  States responded by extending debt restrictions to local governments, 
beginning in the 1870s.  Local governments responded to the new restrictions by 
developing the revenue bond and more extensive use of special districts and public 
authorities. 
At the beginning of the 20th century, states began modifying debt limits by explicitly 
allowing debt for certain purposes.  Debt itemization, the use of revenue bonds, and the 
                                                                                                                                                 
dataset includes these variables, but their coherent incorporation into a system to interpret debt restrictions 
is still underway. 
 69
creation of special districts gave state governments a great deal of latitude in borrowing, 
even if the state constitution contained what appeared to be a strong limit on debt 
creation. In response, several states in the nineteen sixties and seventies modified their 
constitutions to allow legislatures a considerably greater degree of flexibility, either on 
the limit itself or in the modification procedures.  
The evolution of state debt limitations plays a role in other areas as well. We have 
presented a brief discussion of the proliferation of revenue-backed bonds, and include 
some discussion of the increased role of sub-state governments in the issuing of debt. But 
this remains an area for further study. A detailed investigation of debt provisions at the 
local level, using the history of the relevant constitutional provisions, is underway to 
complement the study of the evolution of state debt limits. 
An additional conclusion from this study is that debt restrictions cannot be taken to be 
independent from the rest of the institutions and the circumstances of the state. This 
becomes clear when a historical perspective is applied to the problem. In fact, as Cohen 
(1993) states, ?these [debt] restrictions generally reflect a reaction to some adverse event 
in the past. The limit, while designated to address the past failure, will likely become 
obsolete, representing an obstacle to be surmounted by creative interpretation.?75 This 
often-creative restructuring of debt restrictions is driven by the set of economic 
circumstances present (such as the possibility to hand down responsibilities to local 
governments, increase tax bases, raise tax rates or the economic need for financing) as 
well as the general political situation (spending pressures from the federal government, 
enactment of laws, interpretations by the Courts, etc) and even restricted by the available 
?technology? (for example, the ability to charge fees to the users of a project and commit 
                                                 
75 Cohen in ?The Handbook of Municipal Bonds and Public Finance?, page 131. Word in brackets added. 
 70
the expected stream of revenues to the issuance of revenue bonds, or to identify similar 
governments who may benefit from the debt issued by a public authority or special 
district and bond them to a payment scheme).  
During good economic times, states have been able to operate within the restrictions 
imposed by balanced budget requirements and debt restrictions, but when expectations of 
future revenue are not met or the economic circumstances deteriorate, governments find 
themselves in complicated situations and their reactions are constrained by their past 
institutional choices. These reactions are not illegal or even necessarily undesirable, and 
although they sometimes contribute to the diffusion of responsibilities and make the 
financing of public endeavors less transparent, on other occasions they have created new 
fiscal institutions and institutions that allow the market to function more efficiently. For 
example, bonds that are not backed by the full faith and credit of the state may be issued 
as a way to incur debt outside the limits mandated by the constitution or statutes, and this 
may have its benefits76. Moreover, as the pressure brought about by restrictions 
increased, other ways of financing long-ranging projects developed,77 and it has often 
been up to the Courts to decide the extent of the constitutional rule.78 
                                                 
76 Several problems could arise because of the increase use of revenue bonds. First, it could diminish the 
clarity and control over the amount of debt, conditions and purposes of debt. In addition, it could result in 
larger debt service for an equal amount of borrowing derived from higher interest costs, which could 
increase the strain on government finances. Lastly, revenue bonds can be considered a violation of the spirit 
that drives the limit or the procedures embedded in the state?s constitution. On the other hand, revenue 
bonds may allow for the provision of development-promoting infrastructure and services, which would 
otherwise be unavailable to states and municipalities. They would then be means to finance a project from 
service charges or increased tax collection. Similarly, states have contracted the construction of projects 
and agreed to enter in a long-term lease with the promoter, so instead of debt payments the rental fees are 
considered current expenditure of each year. 
77 States have created special funds to repay the costs of projects, and Courts have often interpreted that the 
debt to be repaid out of these funds does not count towards the state?s limit ?since the payments are to be 
derived from the revenues of the project, and not from the general funds of the state 
78 Sterk and Goldman (1991) provide an overview of how Courts have addressed issues regarding debt 
limits in several states. For example, they discuss how Ohio?s Courts have upheld a tight interpretation of 
the state?s debt limit, while the loose interpretation of New Jersey?s rules has watered down the meaning of 
 71
Another development in state financing that is often considered a reaction to state debt 
limits is the proliferation of public authorities and special districts.79 Table A.8 lists the 
states that have more special districts and sub-county general-purpose governments than 
the national average for the period in which such data are available, and provides an 
overview of the constitutional restrictions to debt at the state level in those states.  
The relationship of debt restrictions with other institutions has received less attention. 
Balanced budget rules are closely tied to debt restrictions, and this becomes particularly 
clear when systematically classifying debt restrictions. Indeed, we often observe 
confusion between the two, since constitutional provisions that are often taken as 
balanced budget rules are in fact restrictions on debt.80  
Similarly, RDFs can be construed as a way to evade TELs, but they may also serve as 
fiscal protection against economic downturns in a setting where constitutions and statutes 
restrict debt issuance and require some degree of budget balance, especially if unpopular 
tax increases may be subject to voter approval. All except two of the RDFs were adopted 
in the last period of modifications to debt restrictions, coinciding with a time when 
legislators were changing their constitutions to incorporate more flexibility into their debt 
restrictions.  
The search for an institutional framework that would combine flexibility with stability 
has resulted in transformations in debt restrictions that?far from being exogenous- have 
                                                                                                                                                 
its limit. Briffault (2003) argues that ?there is an enormous gap between the written provisions of state 
constitutions and actual practice? (page 3). 
79 These might be a subterfuge for politicians who wish to enjoy the political benefits of higher spending 
without paying the associated cost of financing the projects, but they may also be a response to the 
fragmentalization of interests and the possibility of better identifying benefits with costs. In a set-up where 
lower levels of government -such as municipalities- are more than administrative extensions of the central 
authority, borrowing at the local level has the potential of realizing local needs that would not be provided 
for adequately at higher levels. 
80 Table B.7 presents the states? balanced budget requirements, as well as a list of the states whose 
provisions are not aimed directly at regulating the budget. 
 72
been both the cause and result of economic changes. Furthermore, they have contributed 
to the creation and shaping of other budgetary institutions at the state and local levels. A 
proper understanding of the real nature of debt restrictions is the first step in 
incorporating these rules into the theoretical and empirical analyses of state finances. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 73
Chapter 3. The Determinants of the Structure of Rainy Day Funds 
 
 
3.1. Introduction 
State legislatures have constrained themselves through history to prevent the 
functioning of a democratic system from negatively affecting finances in the presence of 
an ever changing and, to some extent, unpredictable environment.81 Examples of these 
self-imposed constraints can be found in the limits and rules imposed on state budgets, 
such as balanced budget provisions, tax and expenditure limitations, debt restrictions, etc. 
These rules are aimed at preventing politicians from starting projects and incurring 
excessive expenditures ?whose consequences would become evident later, when the 
officials may no longer be in office. However, they also diminish the state?s ability to 
deal with crises,82 a problem that is aggravated due to the fact that state finances tend to 
be pro-cyclical: in times of prosperity, states receive moneys from expanded tax bases, 
and the number of people who qualify for state assistance diminishes. Conversely, when 
the economy is in recession, revenues fall while spending needs increase.  
Budget stabilization funds (BSFs henceforth), also known as rainy day funds (RDFs), 
are a relatively new addition to the set of tools states have at their disposal to face the 
fiscal pressures brought about by business cycles. BSFs can help states smooth their 
consumption by serving as receptacles for savings to be used in times of economic 
distress. As Poterba (1995) points out in his study of the real effects of capital budgets in 
                                                 
81 Wallis (2005) presents the case of states constraining themselves not to undertake large projects without a 
sound plan for repayment, as a consequence of the experience with canal and railroad construction, where 
failed projects imposed a heavy burden on states. 
82 Lowry and Alt (2001) note that there are costs associated with adjusting a budget and it is impossible to 
guarantee that the benefits from fiscal discipline will compensate for the loss of flexibility in state policy 
that comes with reducing the state?s ability to run deficits. 
 74
the states ?since?fiscal institutions have important effects on policy outcomes, it is 
important to understand the factors that lead to changes in these institutions.?83  This is 
true of RDFs, since their structure, in terms of the rules that control the deposit and 
withdrawal of funds from the fund, has important consequences for their effectiveness.84 
The choices states make regarding the configuration of their BSFs are, therefore, not 
innocuous. Despite its relevance, the literature has so far overlooked the factors that 
determine the adoption of a particular set of regulations for state BSFs. We use a 
categorization based on the stringency of the rules that dictate how funds in the RDF 
enter and leave the fund to analyze why some states adopt very demanding RDFs in 
terms of the rules for deposit and withdrawal, while others prefer more relaxed 
regulations.  
Using multinomial and ordered econometric methods, we analyze the effect of various 
economic, political and institutional factors in the RDF configuration decisions of the 
states. With this method, we find that states with larger Senates are more likely to adopt 
weak rules, and lower levels of fragmentation in the lower House reduce the odds of 
adopting strict requirements. Also, states with appointed supreme courts are more likely 
to establish demanding rules, as are states whose RDFs are constitutional in nature. 
Finally, we find evidence that states with higher volatility in their tax revenues or that 
exert more tax effort are more likely to establish strict RDFs. 
RDFs are just one of the tools states have at their disposal to reduce the negative 
effects of economic downturns, but their importance becomes apparent once we examine 
the alternatives closely. R. Holcombe and R. Sobel (1997) conclude that cyclical 
                                                 
83 pp 185. 
84 Sobel and Holcombe (1996), Douglas and Gaddie (2002), Zahradnik and Ribeiro (2003). 
 75
variability of state revenues (as opposed to expenses) takes the larger share in the 
responsibility for state government financial crises. Their analysis indicates that there is 
no simple recipe to reduce revenue variability, and they propose that states concentrate 
instead on smoothing their resources over the business cycle. The set of tools available 
for state consumption smoothing is, however, more limited than its household 
counterpart. The same institutions that are meant to stimulate responsible fiscal behavior 
restrict the usage of debt for business cycle smoothing,85 leaving four main options open 
to state officials: increasing taxes to match spending needs, reducing spending in 
accordance with the decrease in means, using fiscal gimmicks, and depleting previously 
stored resources.86 The use of moneys from BSFs falls in this last category. Before 
describing these funds in detail, we briefly review the reasons that make the other policy 
venues unattractive or unsatisfactory to fully solve the financial problems states face 
during crises. 
Holtz-Eakin, Rosen and Tuller (1994), using data on aggregate local and state 
expenditure, find that nearly all expenditure is driven by current resources. In this spirit, 
the first of the four solutions mentioned above calls for increases in taxes to meet 
spending demands. However, as the results in Sobel (1998) indicate, raising tax rates to 
face increasing spending needs is not an option welcomed by voters: politicians who use 
this kind of policy are significantly more likely to be voted out of office, making this 
solution unattractive to election-bound officials. In addition, some states have enacted tax 
                                                 
85 States do borrow (see Brecher et al (2003)), but it is rare to find states that use long-term rate debt to 
finance current expenses (Snell (2004)). For a more in-depth review of debt limits and their impact on state 
economies the reader is referred to Rodriguez-Tejedo (2006). 
86 McGranahan (2002) in her analysis of the 2001 crises analyzes ?the combination of cyclical revenues 
with acyclical or even counter cyclical obligations and institutions that are not permitted to use financial 
markets to deal with this disjoint?, pp 20. 
 76
limitations, which reduce the potential of this venue in times of crises.87 On the other 
hand, spending cannot be easily downsized to match decreased revenues: states have 
become increasingly responsible for the provision of care for needy citizens, and these 
obligations only grow during economic hardships; moreover, reduction of state spending 
in such times can also impede the recovery of the economy.88  
Fiscal gimmicks and one-time cash solutions can temporarily correct budget problems, 
but they do not address the problems behind the deficits. An early compilation of these 
strategies can be found in Kirkland (1983) who argues that they likely indicate the state 
officers? belief that the adjustments are meant to weather recessions while keeping an 
otherwise well planned and well functioning budget. Poterba (1995) and Briffault (1996) 
also describe some of these strategies and draw attention to their worrisome long-term 
consequences.  Furthermore, these tricks become scarcer as time goes on and are always 
cosmetic operations, not fit as long-term solutions.89 
Since the restrictions governments face in terms of balanced budget rules are stock in 
nature, states are not required to maintain spending and revenues at the same level at all 
times, leaving savings (such as those stored in BSFs) as a viable alternative (or 
complement) to smooth out consumption over the business cycle. States can save in other 
funds aside from the BSF, and in the next section we discuss the factors that set apart 
these funds from the general fund.90 As we will see, the differences between the general 
                                                 
87 With regard to these limits, Poulson (2005) raises the issue of the tradeoff between controlling the 
government?s size versus diminishing the capacity to smooth out fiscal activity over the cycle. 
88 Lav and Berube (1999) describe in some detail the dynamics of the crisis in the early 1990s, providing 
examples of these issues. 
89 More on accounting gimmicks and how they interact with other budget policies (such as balanced budget 
requirements) can be found in Briffault (1996) and Rodriguez-Tejedo (2006). 
90 Hou (2001) finds that BSFs have taken the lead in counter-cyclical effects of savings, while the general 
fund surplus has been relegated to a second place. Hou (2005) further suggests general fund surpluses may 
have ceased to be used for expenditure smoothing after BSFs were adopted. 
 77
fund surplus and a BSF (and across BSFs) are relevant for their effectiveness in helping 
states cope with recessions. These differences can turn BSFs into relatively ineffective 
policy tools or significantly increase the state?s capacity to weather adverse economic 
conditions. 
Section 3.2 describes the characteristics of the funds; Section 3.3 presents a series of 
potential determinants for the choice of configuration of the BSFs; Section 3.4 discusses 
the empirical strategy and the results and Section 3.5 concludes.   
 
3.2. Characterization of Budget Stabilization Funds 
BSFs are simply a separate account for savings where funds can be stored during good 
times to withdraw them in times of need. However, what constitutes a BSF may not be 
unambiguously clear, as the disagreement over the nature of some funds demonstrates.91 
The definition we use in this paper runs parallel to that most commonly used in the 
literature:92 in rough terms, BSFs are institutionalized budgetary tools that allow for the 
accumulation of funds during expansions for use during recessions.  According to this 
definition, there are currently five states without an RDF: Alabama,93 Arkansas, 
Colorado,94 Montana and Oregon. 
                                                 
91 Two clear examples are Alabama?s Education Proration Prevention Fund (noted as a rainy day fund by 
the National Association of State Budget Officers, but not by most of the literature due to its restrictive 
scope) and Colorado?s Required Reserve (considered as a rainy day fund by both NASBO and several 
authors in the literature, but not by policy makers in Colorado, who repeatedly initiate petitions to amend 
the state?s Constitution to provide for a rainy day fund). 
92 The point where we deviate from the literature is excluding Colorado from the list of states with BSFs. 
The reasons for this elimination are the exchanges with officials and policy analysts in Colorado who 
consider the state as lacking such funds, and the careful study of state documents regarding the Required 
Reserve.  
93 Alabama set up a reserve fund, but its resources can only be used for education so it is not considered a 
budget stabilization fund by most of the literature. 
94 Colorado has only a small emergency fund that cannot be accessed to meet economic downturns since it 
is reserved mainly for natural disasters. State Treasurer Coffman and Dr Poulson, among others, are 
 78
As shown in table B.1.1 in appendix B, BSFs did not become commonplace until after 
the mid-1980s, although dates of adoption vary substantially. Earlier studies of BSFs 
placed much emphasis on the ?lesson effect? of the crisis of the 1980s, often thought to 
be the cause of the cascade of BSF adoption. However, more recent research (Wagner 
and Sobel (2006)) suggests that this explanation may be too simplistic and overlook other 
factors, such as the changes in the set of restrictions and fiscal tools available to states 
that occurred during that period.   
By establishing and funding a BSF, states may increase the amount of assets at their 
disposal during a crisis,95 providing a cushion that can be used as an alternative or 
complement to other fiscal strategies. However, this basic explanation misses the 
complexity of the process of saving and withdrawing embodied in the their everyday 
operation: Not all funds are born equal and, in fact, BSFs have diverse characteristics that 
introduce widely varying elements that make their operation intrinsically different from 
the general fund surplus.96  
There is strong evidence of the importance of the configuration of BSFs. Navin and 
Navin (1994), through the study of BSF characteristics in the context of economic 
indicators, concluded that BSFs acted as countercyclical tools in only three of the 
Midwestern states. Sobel and Holcombe (1996) and Douglas and Gaddie (2002) consider 
the ability of a BSF to reduce fiscal stress during crises, and conclude that the structure of 
                                                                                                                                                 
making strong calls for a significant BSF in Colorado that would fit the state?s special framework, ruled by 
the presence of Colorado?s Taxpayer?s Bill of Rights (TABOR). See A Rainy Day Fund for Colorado, 
Treasure E-notes, January 2003. 
95 Knight and Levinson (1999) find that states with BSFs have more savings than those without funds and, 
furthermore, they save more after the adoption of these funds than they did previously. 
96?Budget Stabilization Funds should not be combined with general fund ending balances because these 
funds serve two different purposes and they generally are not interchangeable? Nevertheless, both serve a 
similar purpose and should be reported as resources available to a state? Fiscal Survey of the States 
(NASBO, July 1985), pp 18. 
 79
the BSF is crucial for its effectiveness ?while the mere existence of a BSF has no real 
effects.97 McGranahan (2002) and Zahradnik and Ribeiro (2003) find that the existence 
of BSFs helps states weather recessions, but remark that an appropriate configuration 
could significantly improve their effectiveness. 
Studies also exist regarding more particular venues through which states may benefit 
from stringently-configured funds: Gonzalez and Paqueo (2003) conclude that funds 
ruled by stringent requirements accumulate higher balances and reduce social sector 
expenditure volatility, and Knight and Levinson (1998) and Wagner (2003) find evidence 
suggesting that states with funds that operate under strict rules save more and receive 
better bond ratings, which makes future borrowing less costly for the state.  
In sum, the configuration of BSFs is a very important choice that significantly impacts 
the fund?s effectiveness. This is not surprising, since there is wide evidence that the 
structure of fiscal tools (such as balanced budget requirements, tax and expenditure 
limitations etc) is a significant factor determining their impact on state fiscal outcomes. 
There are four elements in a budget stabilization fund: deposit mechanisms, 
withdrawal rules, caps, and replenishment requirements. Each of them regulates the 
operation of a BSF in a different way, although they do not have the same importance in 
terms of determining their fund?s effectiveness. Because of their prevalence and their 
particular importance for the operation of the BSF, the deposit and withdrawal 
requirements are the most important characteristics embodied in the fund. Being the 
?gates? of the resources as they move in and out of the fund, they are key in determining 
its success as a stabilizing tool and we examine them in more detail now. A description of 
                                                 
97 Hobel and Solcombe (1996) find that BSFs with strict deposit requirements reduced fiscal stress, while 
the effectiveness of BSFs was not affected by the nature of its withdrawal requirement. 
 80
the other two rules (caps and replenishment requirements) can be found in appendix B.2, 
and tables B.1.2 and B.1.3 contain information on the deposit and withdrawal 
requirements of the funds in each state. 
 
3.2.1. Deposit requirements 
Rules to control the incoming flow of money into the fund are often written in the 
constitutional or statutory rule that establishes the RDF as part of the budgetary structure 
of the state, although in some cases no specific provision is made and deposit is left to the 
discretion of the policy maker.   
These rules vary in the degree of freedom given to policy makers in their deposit 
decision. For example, very weak deposit rules do not require contributions to be made to 
the fund and leave the time and amount of deposits up to discretion of the policy maker. 
In contrast, other RDFs explicitly present regulations that specify the circumstances in 
which deposits ought to be made to the fund as well as the specific amount to be 
contributed. For example, Arizona?s Statute regarding its Budget Stabilization Fund 
states  ?In a calendar year in which the annual growth rate exceeds the trend growth rate, 
the excess growth when multiplied by total general fund revenue of the fiscal year ending 
in the calendar year determines the amount to be appropriated by the legislature to the 
budget stabilization fund in the fiscal year in which the calendar year ends.? 
Wagner (2004) classified deposit requirements according to the strictness of the rule 
from one to four, with higher numbers depicting stricter requirements, as follows: 
 81
1) Deposits made through appropriation, at the discretion of the policy 
maker. Under this configuration, BSFs look a lot like the general fund and many 
elements of substitutability between the funds are introduced. 
2) Deposits happen if there is a surplus in the budget.  In practice, this option 
may be very similar to the previous one, since the existence of surplus in the 
budget is a decision largely in the hands of budget crafters. 
3) Fixed deposit, based on formulae tied to different parts of the budget (the 
most popular are linked to percentages of revenues or spending). 
4) Deposits based on rules tied to economic growth (usually regarding the 
portion of the excess in the general fund to be deposited). 
 
Deposit rules based on formulas guarantee that the fund will actually receive revenues, 
forcing officials to plan on savings while drafting the budget if the conditions established 
by the formula are met. However, this does not imply that the state will necessarily save 
more than it would in absence of the BSF, because it could simply decide to put into the 
budget stabilization fund what would have otherwise been deposited in other funds. 
However, this is only true if the legislature had planned on saving at least as much as the 
formula requires; if this were not the case, the rainy day fund is effectively increasing the 
amount of savings made by the state.  
 
 82
3.2.2. Withdrawal requirements 
Withdrawing resources from the fund can also be done in a variety of ways, ranging 
from discretionary appropriation by the legislature to restrictive formulae that will only 
allow withdrawals if the economic circumstances are severe.98  
As with deposit requirements, withdrawal rules are indexed from least to most 
stringent, again following Wagner: 
1) Withdrawals are possible through appropriation, at the discretion of the 
policy maker. A BSF where legislatures can access funds freely is as open to 
political raid as the general fund, and in this respect constitutes only a formal 
distinction between the two.  
2) Withdrawals are permitted in the event of a revenue shortfall. Although 
more restrictive than the previous requirement, this rule permits access to funds 
whether or not there is serious fiscal stress since revenue shortfalls can be 
triggered in a variety of ways, including cuts in taxes.   
3) A supermajority approval is required for withdrawal. 
4) Withdrawal is conditional on formulas tied to economic decline. 
 
Arizona?s Statute, for example, provides detailed information regarding the time and 
amount of the withdrawal appropriation stating that ?In a calendar year in which the 
annual growth rate is both less than two per cent and less than the trend growth rate, the 
difference between the annual growth rate and the trend growth rate when multiplied by 
                                                 
98 Strict withdrawal requirements have real effects on fiscal outcomes. The reason follows the logic 
presented by Manuel Amador (2003) in the context of political economy models of government savings: 
?illiquidity is a useful characteristic because it reduces the temptation of current governments from 
overconsuming.? 
 83
the total general fund revenue of the fiscal year ending in the calendar year determines 
the amount to be transferred by the legislature from the budget stabilization fund to the 
state general fund at the end of the fiscal year in which the calendar year ends. The 
transfer calculated pursuant to this subsection shall not exceed the available balance in 
the fund, nor shall the legislature transfer an amount which exceeds the amount sufficient 
to balance the general fund budget.? It is therefore classified as having a withdrawal 
requirement of type 4. On the other hand, Rhode Island?s State budget reserve and cash 
stabilization account is classified as having a type 2 requirement, since it allows for 
withdrawals ?upon notification by the budget officer that it is indicated that total 
resources which are defined to be the aggregate of estimated general revenue, general 
revenue receivables, other financing sources and available free surplus in the general 
fund will be less than the original estimates upon which current appropriations were 
based.? 
It is worth noting the disagreement in the literature over whether certain funds, 
reserved for certain purposes (such as education or natural emergencies), can be 
considered real BSFs.99 In general (and for our purposes as well) funds that are not 
available for the reduction of fiscal stress during crises are not considered to be BSFs. 
 
3.3. Factors influencing the choice of BSF configuration 
This section presents several factors that may be relevant in the state?s choice of the 
structure of a BSF and how they approximate or interact with what seem to be the three 
                                                 
99 For example, Hou (2001), repeatedly advocates a much more restrictive definition of what constitutes a 
BSF than the one usually admitted by the literature. These discussions serve to illustrate the wide range of 
strictness in the withdrawal requirements.  
 84
most important factors in the decision to configure a BSF: the uncertainty derived from 
the difficulty in obtaining good forecasts of future economic circumstances, the 
embedded uncertainty of election-driven state politics, and the existence of regulations on 
the budget that limit the policy maker?s ability to control state finances. We briefly 
discuss each of them and present the associated indicators we have considered in the 
regressions. A more detailed description of the variables can be found in Appendix B.3. 
The regression equations can be presented in compact form as Y= f(X1i, X2i, X3i), 
where Y is the discrete variable that represents the deposit (or withdrawal) requirements, 
the vector X1 includes the economic variables, X2 includes the political variables, and X3 
includes the variables regarding institutional structure. 
Uncertainty about the future of the economy is at the core of the decision to establish a 
BSF: if perfect forecasting of cycles were possible, state officials could plan accordingly 
and smooth out consumption by saving in good times and running their reserves down 
during perilous times. This would not be politically taxing because it would be easy to 
justify both behaviors to the public under the light of the predictable nature of the state?s 
economic cycle. However, even with state budget officers devoting much effort and 
resources to getting good forecasts of revenues and expenditures, these are at best good 
approximations that tend to get worse as the time horizon is extended and usually fail to 
foresee sharp downturns in state finances. In the specific economic factors we discuss 
below, increased income volatility, uncertainty or need may raise the optimal level of 
savings for the state, everything else constant, making it more desirable to establish a 
strict BSF. 
 85
The second source of uncertainty comes from the political process. Even politicians 
who are not purely self-interested need first to be (re)elected to ensure that their preferred 
policies will be enacted.100 The desire to remain in office, paired with the fact that state 
budgets often finance targeted public policies, translate into an effort to please voters at 
the cost of shortsighted policies, or the conscious effort to set up an unfavorable 
environment for the successor if he happens to be of the opposite party. In either case, 
these non-economic objectives have the potential to create incentives for suboptimal 
fiscal choices.101  
For the choice of BSF configuration, then, economic uncertainty calls for increased 
savings in the spirit of life-cycle models, which ?in the presence of incentives for 
overspending- may make institutionalized forms of savings attractive. In addition, 
political uncertainty creates incentives for policy makers to consume resources while in 
power in a common pool problem fashion. If those who draft BSF-like funds want to 
reduce the effects of political uncertainty, strict rules are an attractive feature.102 On the 
other hand, weak RDFs may be used as a means to accommodate political needs, making 
weak funds more enticing.  
Other factors, such as the socio-economic configuration of the state and the existing 
set of institutions can strengthen or weaken either motive.103 For example, BSFs could be 
used as means to avoid the budget rigidities imposed by other restrictions, such as 
balanced budget requirements or tax and expenditure limitations. It is important to 
                                                 
100 Poterba (1994) finds that in gubernatorial election years states enact less tax increases and expenditure 
reductions. 
101Velasco (2000) present a model in which government resources are viewed as common property and find 
that fiscal deficits and excessive debt emerge.    
102 Mody and Fabrizio (2006) find, when studying countries in the EU, that budgets are often an expression 
of political rather than economic priorities, but also that budget institutions and rules have significant value 
in alleviating these problems. 
103 See Rodriguez-Tejedo (2006) for an overview of the interactions between several institutions and BSFs. 
 86
consider the effects other institutions may have had in the decision to establish a certain 
type of BSF, since ?the various institutions interact with one another in complex 
ways,?104 and an analysis that omits these interactions is likely to provide an incomplete, 
or even misleading, picture.  
The choice of BSF structure can be understood in terms of a standard random utility 
model. Under every possible fund configuration, we consider the legislature?s utility, 
which would be a (not necessarily linear) function of the funds? characteristics, as well as 
the particularities of the state. A state will choose a particular configuration if its 
associated utility surpasses that of all the other possible configurations (and is also 
greater than the utility associated with not establishing a BSF at all).  
To analyze this decision empirically, we use a panel dataset with information for all 
states that adopted such funds in the period 1951-2000 (the last year in which an adoption 
occurred). Since our primary focus is to investigate the determinants of the configuration 
choice, we include only states that adopt a BSF during our sample period. After a state 
adopts a fund and establishes its preferred configuration, no further observations from the 
state are included in the sample.105 The results are conditional on the state adopting a BSF 
for the first time; the model seeks to explain the factors leading to adoption (rather than 
adoption and retention) of the chosen structure. Seven states are excluded from our 
sample: Alabama, Arkansas, Colorado, Montana and Oregon (because they do not 
currently have a BSF), Alaska (due to the very particular nature of its BSF) and New 
York (which adopted its RDF before 1951). 
                                                 
104 Knight and Levinson (1998), pp 3. 
105 This simplifies the empirical analysis, since it prevents the potential simultaneity bias that would occur 
if we were to include after-BSF years, when some of the regressors may be affected by the existence of the 
fund. The assumption does not stray far from reality, since only Ohio has changed the requirements of its 
BSF, and the procedures to change the configuration of a BSF can be quite cumbersome. 
 87
The data used to approximate the elements that we postulate may have had an effect 
on the process of adoption of these BSFs are listed in table 3.1 at the end of this section, 
grouped in three main categories: political, socio-economic, and institutional factors. 
Appendix B.3 contains detailed descriptions of the construction of the data,106 as well as 
some alternatives and the reasons for the choice of the variables included in the analysis.  
The political science literature suggests several variables that may be of importance 
among the first set mentioned above. In the legislative branch, the finding that larger 
upper houses spend more could translate into a desire for weakly configured BSFs, so 
funds are easily accessible. On the other hand, there is no clear result that links partisan 
composition to spending, leaving the relationship between the composition of the houses 
and the nature of BSFs as a matter open for empirical investigation. There is, however, 
evidence suggesting that the political affiliation of the governor (independently and 
jointly with the legislature?s) and the existence of term limits for governors have real 
fiscal effects. Lastly, appointed State Supreme Courts are thought to be more lenient, 
because appointed judges may be more amenable to deviations from the rule, which 
would make strict BSFs rules less demanding in real terms and hence less politically 
constraining. 
Among the socio-economic variables, we use the yearly deviation from the national 
mean of per-capita personal income as a measure of the state?s general economic 
condition. To investigate the effects of the sector composition of the state, we introduce 
the proportion of total earnings in construction, farming, manufacturing, mining and 
                                                 
106 We discuss, among others, the correction of the data for the balanced budget requirement in Tennessee, 
the consideration of an additional measure for savings that takes into account the Census? warnings 
regarding the construction of series on savings and different measures for volatility. The expected effect of 
the variables in this context is also discussed further in this appendix. 
 88
services. The effects of state?s population density are unclear: a state that has to cover the 
public expenditure demands of a larger population may find BSFs more attractive, an 
effect reinforced by the public-good component of savings in the RDF. However, larger 
states have been found to have less volatile business cycles, so they may find strict BSFs 
less appealing. Beyond their income and population, we expect states engaged in volatile 
spending to be in greater need for easily accessible savings, a fact that may be reflected in 
the type of BSF they adopt. We consider each spending type?s mean standard deviation 
and classify the six types of expenditure in three categories (high, medium and low 
volatility107).  On the other hand, we might expect states with volatile tax revenues to be 
more inclined to establish strict funds. As with spending, we include tax collection by 
grouping the different types of taxes according to their levels of volatility.108  
Tax collections are the most important source of income for states (although their 
share has decreased), followed by intergovernmental revenue (IG). IG revenues include 
local and federal transfers (with the latter making up about 95% of the total109) and are 
mostly outside of state control. IG revenues are likely to decrease during periods of 
crises, when states need resources the most. IG finances are included in the analysis by 
calculating the deviation from the national mean of the per capita net IG transfers 
(revenues minus expenses). 
                                                 
107 After applying the GDP deflator and calculating the overall average standard deviations, we can see that 
the magnitudes of the standard deviations are similar within groups and considerably different across 
groups, so the choice of three groups with two components seems reasonable. Education and welfare 
spending are the most volatile group, while expenditure in highways and health and hospitals fall in the 
middle category, and unemployment compensation and spending in natural resources are the relatively least 
volatile expenses. Although it may seem counterintuitive that education belongs in the most volatile group, 
we must note that capital spending in education is included in this category, which explains its variability. 
108 The percentages of tax income that come from severance and property taxes are grouped in the ?most 
volatile? category, while the percentages received from sale and individual income taxes form the ?least 
volatile? category. The percentage of tax revenues derived from corporate income taxes corresponds to the 
?middle volatility? group, which is used as baseline.  
109 As opposed to IG expenses, where local IG spending makes for most of the total expenditure. 
 89
Beyond the effects of current sources of income, states can use savings in face of a 
recession. It is likely that states that maintain easily liquefiable resources will consider 
the need to establish a stringent fund as less pressing. On the other hand, it may be 
possible that states that decide to have more savings in the form of cash and securities 
have a preference for sound savings, and would be more inclined to establish strict funds. 
Since it is difficult to establish, intuitively or theoretically, a predicted sign for the 
relationship between other savings and the nature of BSFs, it remains a question best 
answered empirically. Aside from using reserves, states can increase the resources they 
raise from taxation. However, states that exert higher levels of tax effort will have less 
room for tax increases, making meaningful BSFs more attractive. On the other side of the 
spectrum, our a priori expectation is that states with higher levels of debt will be, all else 
constant, more inclined to establish demanding BSFs, since it would be relatively more 
costly for them to go further into debt. However, high levels of per capita debt may be 
correlated with a higher tolerance for debt in the state, which could offset the 
aforementioned effect. The final effect of indebtness on BSF rules is then left to 
empirical investigation. 
Among the institutional constraints, tax and expenditure limitations (TELs) restrict the 
state?s ability to cope with recessions through direct action, which may make meaningful 
BSFs more attractive. Alternatively, RDFs may be seen as a way to put funds outside of 
the scope of the TELs, allowing for wider discretion in spending decisions, a proposition 
for which Wagner and Sobel (2006) find supporting evidence. We have explored 
different alternative measures of TELs, using dummies for the existence of each of these 
limitations as well as Poulson?s (2005) indexes of TEL strictness. Another important 
 90
institutional constraint is given by the existence of balanced budget requirements (BBR). 
States with demanding BBRs enact more restrictive spending policies (Poterba (1994)), 
fare better in deficit control (Alesina and Bayoumi (1996)), are more likely to enact tax 
increases and spending cuts during recessions (Alt and Lowry (1994)) and tend to save 
more (Bohn and Inman (1996)). But strict BBRs also introduce rigidities in fiscal policy 
(Alt and Lowry (2001)) and may exacerbate business cycle volatility (Levinson (1998)). 
Demanding BBRs make meaningful BSFs more appealing, since intertemporal 
smoothing becomes more difficult. The last institutional factor is embedded in the BSFs 
themselves. Their legal nature (statutory or constitutional) can also play a role on the 
configuration of deposit and withdrawal requirements. Constitutionally established 
budgetary tools allow decision makers less freedom when establishing the particulars of 
the law and have been shown to have stronger effects on fiscal policy than their statutory 
counterparts.   
Table 3.1. List of control variables 
Variable Source 
X1 ? Political variables 
Number of seats in upper House 
Number of seats in lower House 
% seat gap between main parties (Upper House) 
% seat gap between main parties (Lower House) 
Democratic Governor (dummy) 
ICPSR study #0016, Statistical Abstracts of 
the US, Minnesota Legislative Reference 
Library. 
Appointed Supreme Court (dummy) Bohn and Inman (1996) 
Limit for governor's tenancy (dummy) Council of State Governments 
Citizens? ideology  
Governments? ideology 
Berry, Ringquist, Fording, and Hanson 
(1999) 
X2 ? Socio-economic variables 
Deviation from average per capita personal income Bureau of Economic Analysis 
Percentage of earnings ? by sector Bureau of Economic Analysis 
Deviation from average per capita savings US Census Bureau 
Percentage of tax revenue ? by degree of volatility 
Percentage of expenditure - by degree of volatility 
Deviation from average per capita net intergovernmental 
revenue 
Census of Governments and the Historical 
Statistics of the United States. 
Tax effort ACIR and Tannenwald (2000) 
Deviation from average per capita debt US Census Bureau 
Population density Statistical Abstracts of the United States 
 91
X3 ? Institutional variables  
Constitutional BSF (dummy) Wagner (2001) 
Expenditure limitation (dummy) 
Rueben (1995), Waisanen (2005) and 
Poulson (2005) 
Stringency of the balanced budget rule 
ACIR, NASBO, personal communication 
with Tennessee?s Department of Finance 
and Administration 
 
3.4. Empirical strategy and results 
Since the dependent variables are not continuous, the estimation of the coefficients 
with an OLS regression would not be legitimate.110 Since the actual ?distance? between 
two consecutive numbers is unknown and is likely to be different, more appropriate 
estimation methods are considered below. Our characterization of the dependent 
variables (rules for deposit and withdrawal) allows us to use multinomial and ordered 
techniques that provide new insights into the rationale for the configuration of BSFs. The 
final choice between the explanatory variables is conditioned by the constraints posed by 
data restrictions, which limit the number of parameters that can be estimated.  
An important characteristic of the deposit and withdrawal requirements is that they 
can be classified according to their level of stringency. That is, we can classify funds 
according to how easy or difficult it is to control the moneys that go in or out of the fund 
and assign numbers to each characteristic to reflect this fact. For that purpose, we assign 
numbers one to four for the deposit and withdrawal requirements embodied in each 
BSF.111 The actual values are irrelevant beyond reflecting that outcomes with larger 
values correspond to ?stricter? requirements. We can take advantage of the ordinal 
multinomial nature of the data by estimating ordinal logistic regressions. The structural 
                                                 
110 OLS equivalents of the models were calculated for comparison purposes, and are available upon request. 
As expected, all the models considered performed better than their OLS counterparts. 
111 Meaning that requirements of type ?four? are stricter than those of type ?three?, ?two? and ?one?, 
requirements of type ?three? are stricter than those with values ?two? or ?one? and so on. 
 92
model for an ordered logit (or proportional odds model) is given by y*it = xit ? + ?it , 
where i indexes the state, t the year and ? is a disturbance with the logistic distribution. In 
the most general case we consider (with four possible categories), the model can be 
expressed in terms of probabilities as: 
  x'1 )(plogit 1
1
1
1 ?? +=?= p
p , 
  x'1 )p(plogit 2
21
21
21 ?? +=??
+=+
pp
pp    and 
  
 
x'1 )pp(plogit 3
321
321
321 ?? +=???
++=++
ppp
ppp
 , 
with p1+p2+p3+p4 = 1 and ?1 < ?2 < ?3 < ?4. 
The ordered logit (OL) assumes that all the coefficients on the independent variables 
are equal for every category of the dependent variable, so the slopes of the estimated 
equations are identical. This is known as the parallel equation assumption (or 
proportional odds assumption, because the odds ratio of Y?j is the same for all 
categories). The assumption can be tested using a Brant?s test112 (Brant (1990)) or a 
likelihood ratio test, which provide evidence suggesting that the parallel regression 
assumption is violated. This is not a rare occurrence, since the proportional odds 
assumption is often violated (Long and Freese (2006)) even with large samples and no a 
priori reason that would justify the violation.113 It is in the spirit of this literature that we 
                                                 
112 Formally, the Brant test?s null hypothesis is that the coefficients remain the same across categories. A p-
value lower than 0.005 indicates that the impact of the independent variables is different across categories, 
and the violation of the proportional odds assumption. 
113 Williams (2006) cites Sarah Mustillo saying ?neither of us [referring to herself and a colleague] has ever 
run an ologit model that DID NOT violate the proportional odds assumption. My models always fail the 
Brant test?. 
 93
report the OL results even when the proportional odds assumption seems to be violated, 
but keeping in mind that the estimates may be misleading.  
The results of some relevant models appear in tables 3.2 and 3.3.114 Because 
regression outcomes of ordered logits are difficult to interpret and do not correspond to 
the numbers most frequently used for the interpretation of regression results, the tables 
report the marginal effects of each independent variable (holding the others at their 
means) rather than estimated coefficients.115 Tables 3.2 and 3.3 then report the partial 
derivatives with respect to the explanatory variables of the probabilities of choosing each 
outcome. Following the same logic as the results of a binary logit, they indicate the 
estimated change for each outcome individually. The results indicate the changes in 
probability of a state adopting a weak (leftmost column in each specification) or a strict 
requirement (rightmost column) given a marginal change in the independent variable, and 
keeping every other variable at its mean. From these tables we can see that an increase in 
the number of seats in the upper House significantly increases the probability of adopting 
a weak or strict deposit requirement, but the increase in probability of adopting a weak 
deposit outweighs the increase in probability of adopting a strict deposit. Similarly, 
increases in the percentage gap in the number of seats held by the two main parties in the 
Lower House reduces the likelihood of establishing demanding requirements. These 
effects persist even when controlling for various other political circumstances. Our results 
                                                 
114 Following Long and Freese (2006) we report the McKelvey and Zavonia?s R2, which has been shown by 
Hagle and Mitchell (1992) and Windmeijer (1995) to be closest to the R2 of a linear model estimated using 
the underlying latent variable. 
115 It is important to note that the marginal changes expressed in the tables cannot be directly used to 
consider the effects on the left hand side variable of an arbitrary increase in any of the independent 
variables. For example, because the probabilities are not linear, we would need to calculate directly what 
the effect on the probability of adopting a strict deposit requirement of a 10% increase in the percentage of 
tax revenue that is derived from the least volatile category, rather than simply multiply the reported 
marginal effect by 10.  
 94
run parallel to Matsusaka?s (1995), who concludes that larger upper houses (but not lower 
houses) are significantly associated with higher spending but does not find such effects 
for the existence of divided governments. Other results (not shown) also fail to find any 
significant relationship between the affiliation of either the legislative or executive 
branches and the configuration of a BSF. In addition, there seems to be some weak 
indication (regressions not shown) that more liberal governments are more likely to adopt 
weak withdrawal requirements. Among the group of economic characteristics, states that 
spend comparatively more on high-volatility spending appear to be more likely to 
establish weak rules, while states with higher levels of debt are less likely to establish 
weak withdrawal requirements. Within the institutional data, our results suggest that 
constitutionally configured BSFs are more likely to include strict operating rules. 
However, it is plausible that we are indeed facing a true violation of the parallel 
equation assumption. Aside from modifying the model, other solutions exist. For 
example, we can ignore the ordinal nature of the dependent variable and restrict the 
analysis to multinomial logits, or we can use generalized ordered logits, that allow the 
effects of the explanatory variables to vary with the point at which the categories of the 
dependent variable are dichotomized. 
Ignoring the ordinal nature of the dependent variable does not bias the coefficients, 
although it may lead to loss of efficiency. Two logit models are commonly used: the 
multinomial logit (or generalized logit model) and the conditional logit.116 For the more 
disaggregated case, the probabilities of adoption in the multinomial case can be expressed 
as:  P(y=i) = exp(?i*x)/ ? j ? i  exp(?j*x) for i=1,2,3,4. As usual, for the system to be 
                                                 
116 We choose the multinomial logit as opposed to the conditional logit model, because the former is used 
when the independent variables refer to characteristics of the units, while the second one is usually 
employed when the independent variables are characteristics of the choices. 
 95
identified, we need to set one of the coefficients equal to zero and compare the results to 
the baseline group. The coefficients on the other (non-reference) groups can then be 
interpreted as log odds of being in a particular group as compared to being in the 
reference group. 
A potential for bias in the estimation of the multinomial logit (MNL) exists, brought 
about by the independence of irrelevant alternatives (IIA) assumption (or the 
independence in competing risks assumption). In our case, the IIA translates into a risk 
for bias if we include BSF configuration alternatives that are not available to legislatures, 
or if we are presenting as different choices configurations that are in reality very close 
substitutes. The latter could be a problem if in fact some of our four categories are close 
substitutes. Reducing our classification from four to two and three categories will provide 
some rough idea on whether this is actually a serious problem.117 
As with the ordered logits, the partial derivative of the probability of a given choice 
does not correspond to the associated regression coefficient, so caution must be used 
when interpreting the results. As with any multi-output regression, there are several 
equations and potential comparisons.  Again we report the effect of a unitary change in 
the independent variable on the probability of adopting a weak (or strict) requirement, 
keeping the rest of the independent variables at their means. Examination of tables 3.4 
                                                 
117 The Hausman-McFadden (1984) test for IIA in the four-category case suggests that independence may 
actually exist. However, there are known problems with this test that make its validity questionable. A 
more reliable test (the Small-Hsiao (1985) test) produces mixed results that suggest that the IIA assumption 
may be violated. When we restrict our characterization of the deposit and withdrawal requirements to two 
categories (?strict? vs ?lax?) both the Hausman-McFadden and the Small-Hsiao test indicate that the IIA 
assumption holds. With the three categories split, again we find the same discordance between the 
Hausman-McFadden and the Small-Hsiao tests as we did in the four-category case. 
Although Wald tests for the possibility of amalgamation of the categories suggest that none is possible, the 
associated chi square values for the test for the reduction of categories 1 and 2 into a single group is much 
smaller than the rest, suggesting that categories 1 and 2 may be much more similar than the others (as we 
expected them to be). In addition, the likelihood ratio test suggests that categories 1 and 2 may be 
indistinguishable. 
 96
and 3.5 reveals similar results to those of the ordered regressions in terms of the political 
variables: states with bigger senates are more likely to establish weak rules118; less 
fragmentation in the lower House reduces the probability of adopting strict 
requirements;119 and states with more liberal institutions seem to be more likely to 
establish weak requirements. The effect of the size of the lower house is barely 
significant and very small in quantitative terms. 
The MNL results suggest that additional factors may be of relevance: states with 
appointed supreme courts are more likely to establish strict deposit rules, providing some 
support for the possibility that deviations from the rules may be easier under this type of 
Supreme Court, reducing the cost of adopting stringent requirements. As before, the 
results indicate that constitutional RDFs are more likely to have more demanding rules. 
In particular, the results from table 3.4 suggest that although the increase in probability of 
adopting a ?strict? deposit requirement from having a constitutional RDF is small in 
magnitude, it is highly significant. 
Among the economic variables, we still find a significant increase in the odds of 
adopting weak rules for states with relatively large shares of highly volatile spending or 
whose earnings are comparatively more dependent on agriculture. If the proportion of 
state expenditure that falls in the most volatile category were to increase by just one unit 
(from the average, ceteris paribus), the results predict an increase in probability of the 
                                                 
118 An alternative way of interpreting the results is by introducing odds ratios (calculated, not shown). If we 
compare the effect of the size of the upper house on the odds ratio of adopting a strict or lax deposit 
requirement, we find that a unit increase in the size of the upper house increases the odds of adopting a 
weak deposit requirement by almost twenty percent. 
119 Our results are consistent with Wallis? assertion that ?states where politics were the most competitive, 
where both parties were most responsive to voters? concerns, were the states more likely to adopt new 
constitutional provision.? (Wallis (2005), pp 29). 
 97
state adopting a weak deposit requirement of between seven and nine percentage points, 
depending on the specification of the model.  
In addition, the MNL results suggest that states with higher levels of tax effort may be 
more prone to establishing strict deposit rules. The results indicate that a one percentage 
point increase from the average value of tax revenues coming from this category 
decreases the estimated probability of adopting a weak deposit requirement by almost 
two percentage points. In the tax structure, higher reliance on volatile tax sources seems 
to increase the odds that a state will choose a strict deposit requirement.  
Other alternatives, besides the usage of MNL, to solve the problem of violation of the 
parallel equations assumption exist. The proportional odds assumption is violated if some 
(or all) of the coefficients in the J-1 regressions are found to be statistically different, so 
that the estimated lines are not parallel. The generalized ordered logit (GOL) model 
solves this problem by allowing the coefficients to differ across categories. Although 
promising, this method is problematic in our case due to the high number of parameters 
that need to be estimated, which may render the estimation of the generalized ordered 
logit model difficult.120 An intermediate solution is presented in Bercedis and Harrell 
(1990) and Williams (2006): they propose the estimation of a generalized ordered logit 
where some of the coefficients are fixed across equations while others are allowed to 
vary. We still need to explore this method; given the nature of our data, it might prove to 
be useful. The stereotype ordered model is a compromise between the MNL and the 
ordered logit, which can be used in a case like this when there is some indication that 
                                                 
120 
 Indeed, many of our generalized ordered logit regressions have difficulty converging.  
 98
some alternatives are very similar.121 Other non-ordered alternatives to multinomial logits 
yet to be explored are the multinomial probit (MNP) model and nested logit model (NL). 
MNPs, while being similar to MNLs, have the advantage of assuming an arbitrary 
covariance structure for its multinomial normal distribution, allowing arbitrary 
correlation between the utilities of each choice. However, the large number of parameters 
that have to be estimated may make this method unattractive in our case. NLs create a 
hierarchical structure, grouping choices into categories within which the IIA is assumed 
to hold. However, two problems exist with this approach: the number of parameters to 
estimate may be too high for our problem, and it is not clear how to group the four levels 
of stringency in such a way that the IIA will hold (the best option, which is the one we 
present here, may be to collapse categories 1 and 2 in a single group, and 3 and 4 in 
another). Alternatively, the NL model is a potential solution to bring into the analysis 
states without BSFs, making the choice of establishing a BSF into the first decision 
nodule and considering the decision about the type of fund a second-level choice. 
It is not a priori clear what the preferred model for our problem may be. In theory, 
given the ordered multi-dimensional nature of the deposit and withdrawal requirements 
and the evidence suggesting that the parallel equation assumption may not be satisfied, a 
generalized ordered logit may be the best fit. However, we run into the problem of having 
to estimate too many parameters. The partial generalized ordered logit could be a solution 
to this problem and needs to be considered. Alternatively, multinomial logits ignore the 
ordered nature of the deposit and withdrawal requirements, but capture the 
multidimensionality of the requirements.  
                                                 
121 Lunt (2001) presents the stereotype ordinal regression model as a method to impose ?ordering 
constraints? in a MNL.  
 99
According to the measures of fit, there seems to be some indication that MNL may 
provide a better fit than the OL for the problem at hand. More support for this claim can 
be found in the plots of the predicted probabilities of the OL and MNL models: there is a 
sudden truncation of the ordered logit model?s distribution that seems unrealistic, 
suggesting that the multinomial logit may be a better model for the data.122 However, 
preliminary work with intermediate techniques suggests these may provide better fits for 
the model while taking into account the ordered nature of the data. 
Beyond the search for the best fit, there is an inherent usefulness in the comparison of 
various plausible models. It allows us to see the data under different alternatives, with 
various specifications and a diverse set of assumptions and provides, in its own way, 
checks for robustness of any estimated coefficient.  
 
 
 
 
                                                 
122 The correlations between the sets of predictions for ordered logits and multinomial logits are not very 
high (about 0.6 [0.4] for lax [strict] deposit requirements and 06 [0.7] for lax [strict] withdrawal 
requirements). 
 100
 
Table 3.2. Ordered logits. Dependent variable: deposit requirement 
Description of independent variables  weak  strict  weak  strict  weak  strict  weak  strict  
Number of seats in upper House  8.55E-05 *** 2.51E-05 ***             
Number of seats in lower House  -5.56E-06  -1.63E-06              
% seat gap between main parties (Upper House)     2.62E-05  8.07E-06  2.28E-05  7.02E-06  3.44E-05  1.06E-05  
% seat gap between main parties (Lower House)     -1.29E-04  -3.98E-05  -1.30E-04  -3.98E-05  -1.24E-04  -3.80E-05  
Democratic Governor              -1.26E-03  -3.88E-04  
Appointed Supreme Court          -1.74E-03  -5.33E-04      
Limit for governor's tenancy              -3.02E-03  -9.28E-04  
Deviation from average per capita personal income  -1.06E-06  -3.10E-07  -1.88E-06  -5.80E-07  -1.90E-06  -5.83E-07  -1.93E-06  -5.94E-07  
Percentage of earnings - farming  -6.56E-02 ** -1.92E-02 ** -5.55E-02 * -1.71E-02 * -5.67E-02 ** -1.74E-02 ** -5.93E-02 ** -1.82E-02 ** 
Percentage of earnings - construction  -1.09E-01 * -3.19E-02 * -9.70E-02 * -2.99E-02 * -9.00E-02 * -2.77E-02 * -1.04E-01 ** -3.20E-02 ** 
Percentage of earnings - manufacturing  -3.73E-02 * -1.09E-02 * -3.25E-02 * -1.00E-02 * -3.19E-02 * -9.81E-03 * -3.63E-02 ** -1.12E-02 ** 
Percentage of earnings - mining  -4.44E-02  -1.30E-02  -3.53E-02  -1.09E-02  -3.83E-02  -1.18E-02  -3.77E-02  -1.16E-02  
Percentage of tax revenue - most volatile  -1.64E-02  -4.81E-03  -5.53E-03  -1.70E-03  -7.57E-03  -2.33E-03  -7.44E-03  -2.28E-03  
Percentage of tax revenue - least volatile  -2.09E-02  -6.13E-03  -1.64E-02  -5.04E-03  -1.69E-02  -5.20E-03  -1.51E-02  -4.62E-03  
Percentage of expenditure - most volatile  8.14E-02 *** 2.39E-02 *** 8.13E-02 *** 2.50E-02 *** 7.80E-02 *** 2.40E-02 *** 7.83E-02 *** 2.40E-02 *** 
Percentage of expenditure - least volatile  6.58E-03  1.93E-03  1.07E-02  3.30E-03  7.49E-03  2.30E-03  1.43E-02  4.40E-03  
Deviation from average per capita savings  1.28E+00  3.76E-01  8.40E-01  2.59E-01  8.95E-01  2.75E-01  6.78E-01  2.08E-01  
Tax effort  2.30E-05  6.75E-06  9.24E-06  2.85E-06  2.09E-05  6.43E-06  -4.96E-05  -1.53E-05  
Expenditure limitation  5.00E-03  1.47E-03  5.49E-03  1.70E-03  6.02E-03  1.86E-03  4.19E-03  1.29E-03  
BBR stringency  -5.30E-05  -1.55E-05  2.34E-05  7.21E-06  7.41E-05  2.28E-05  1.96E-04  6.02E-05  
Deviation from average per capita debt  -7.54E-06  -2.21E-06  -4.22E-06  -1.30E-06  -4.61E-06  -1.42E-06  -3.55E-06  -1.09E-06  
Population density  1.15E-06  3.39E-07  5.08E-06  1.57E-06  6.00E-06  1.84E-06  5.92E-06  1.82E-06  
Constitutional BSF  6.15E-03  1.82E-03  6.77E-03  2.10E-03  7.70E-03  2.39E-03  7.43E-03  2.30E-03  
Deviation from average per capita net IG revenue   1.32E-02  3.86E-03  7.18E-03  2.21E-03  8.88E-03  2.73E-03  7.26E-03  2.23E-03  
Log Likelihood  -182.09951  -179.69465  -179.55244  -178.95421  
*** Significant at 1%, ** Significant at 5%, * Significant at 10% 
The dependent variable equals zero in years prior to BSF adoption, in the year of adoption it equals 1 if the adopted deposit requirement is lax and 2 if it is strict 
 
 101
 
Table 3.3. Ordered logits. Dependent variable: withdrawal requirement 
Description of independent variables  weak  strict  weak  strict  weak  strict  weak  strict  
Number of seats in upper House  8.63E-05 *** 2.53E-05 ***             
Number of seats in lower House  -5.19E-06  -1.52E-06              
% seat gap between main parties (Upper House)      1.88E-05  5.74E-06  1.43E-05  4.36E-06  2.88E-05  8.80E-06  
% seat gap between main parties (Lower House)      -1.36E-04  -4.16E-05  -1.36E-04  -4.15E-05  -1.30E-04  -3.98E-05  
Democratic Governor              -1.40E-03  -4.28E-04  
Appointed Supreme Court          -2.16E-03  -6.57E-04      
Limit for governor's tenancy              -2.91E-03  -8.87E-04  
Deviation from average per capita personal income  -1.16E-06  -3.41E-07  -2.09E-06  -6.38E-07  -2.11E-06  -6.42E-07  -2.11E-06  -6.44E-07  
Percentage of earnings - farming  -6.53E-02 * -1.91E-02 * -5.41E-02 * -1.66E-02 * -5.57E-02 * -1.70E-02 * -5.77E-02 * -1.76E-02 * 
Percentage of earnings - construction  -1.13E-01 * -3.30E-02 * -1.00E-01 * -3.07E-02 * -9.16E-02 * -2.79E-02 * -1.06E-01 * -3.23E-02 * 
Percentage of earnings - manufacturing  -3.80E-02 * -1.11E-02 * -3.32E-02 * -1.01E-02 * -3.25E-02 * -9.91E-03 * -3.70E-02 * -1.13E-02 * 
Percentage of earnings - mining  -4.62E-02  -1.35E-02  -3.69E-02 * -1.13E-02 * -4.11E-02  -1.25E-02  -3.95E-02  -1.20E-02  
Percentage of tax revenue - most volatile  -1.48E-02  -4.32E-03  -2.70E-03  -8.24E-04  -4.99E-03  -1.52E-03  -4.82E-03  -1.47E-03  
Percentage of tax revenue - least volatile  -1.48E-02  -4.32E-03  -1.63E-02  -5.00E-03  -1.70E-02  -5.18E-03  -1.51E-02  -4.62E-03  
Percentage of expenditure - most volatile  8.20E-02 *** 2.40E-02 *** 8.08E-02 *** 2.47E-02 *** 7.67E-02 *** 2.34E-02 *** 7.81E-02 *** 2.38E-02 *** 
Percentage of expenditure - least volatile  5.34E-03  1.56E-03  9.14E-03  2.79E-03  5.36E-03  1.63E-03  1.30E-02  3.98E-03  
Deviation from average per capita savings  1.22E+00  3.57E-01  6.81E-01  2.08E-01  7.42E-01  2.26E-01  5.38E-01  1.64E-01  
Tax effort  2.26E-05  6.62E-06  6.46E-06  1.98E-06  2.07E-05  6.31E-06  -4.92E-05  -1.50E-05  
Expenditure limitation  5.41E-03  1.59E-03  6.26E-03  1.93E-03  6.97E-03  2.14E-03  4.77E-03  1.46E-03  
BBR stringency  -5.60E-05  -1.64E-05  2.48E-05  7.60E-06  8.78E-05  2.67E-05  1.94E-04  5.92E-05  
Deviation from average per capita debt  -7.18E-06 * -2.10E-06 * -3.56E-06 * -1.09E-06 * -4.07E-06 * -1.24E-06 * -2.94E-06  -8.96E-07  
Population density  1.24E-06  3.62E-07  5.68E-06  1.74E-06  6.81E-06  2.08E-06  6.47E-06  1.97E-06  
Constitutional BSF  6.27E-03  1.85E-03  7.41E-03  2.28E-03  6.81E-06  2.08E-06  7.88E-03  2.43E-03  
Deviation from average per capita net IG revenue   1.38E-02  4.03E-03  8.26E-03  2.53E-03  1.04E-02  3.17E-03  8.18E-03  2.49E-03  
Log Likelihood  -182.0834  -179.35376  -179.13407  -178.61558  
*** Significant at 1%, ** Significant at 5%, * Significant at 10% 
The dependent variable equals zero in years prior to BSF adoption, in the year of adoption it equals 1 if the adopted withdrawal requirement is lax and 2 if it is strict 
 
 102
 
Table 3.4. Multinomial logits. Dependent variable: deposit requirement 
Description of independent variables  weak  strict  weak  strict  weak  strict  weak  strict  
Number of seats in upper House  3.79E-04 ** -8.41E-11              
Number of seats in lower House  3.87E-06  -3.09E-11 *             
% seat gap between main parties (Upper House)      -1.20E-06  6.94E-11  -5.17E-06  9.09E-12  9.47E-06  3.46E-11  
% seat gap between main parties (Lower House)      -1.23E-04  -9.74E-11  -1.26E-04  -9.77E-12  -1.10E-04  -4.98E-11 * 
Democratic Governor              -2.34E-03  -2.33E-10  
Appointed Supreme Court          -2.98E-03  1.29E-09 *     
Limit for governor's tenancy              -3.22E-03  4.63E-10  
Deviation from average per capita personal income  -7.53E-07  -1.40E-13  -1.67E-06  -1.16E-12  -1.62E-06  -1.15E-13  -1.73E-06  -5.16E-13  
Percentage of earnings - farming  -3.73E-02 * -5.73E-08  -2.17E-02  -1.39E-07 *** -2.44E-02  -1.26E-08 *** -2.58E-02  -6.59E-08 *** 
Percentage of earnings - construction  -4.39E-02  -2.62E-08  -5.00E-02  -8.40E-08 * -3.71E-02  -6.75E-09  -5.46E-02  -4.12E-08 * 
Percentage of earnings - manufacturing  -2.78E-02  2.86E-09  -1.64E-02  6.91E-10  -1.50E-02  1.08E-09  -2.14E-02  2.50E-10  
Percentage of earnings - mining  -2.63E-03  -6.12E-08 ** 3.39E-03  -9.80E-08 * -2.72E-04  -6.01E-09  -4.59E-04  -4.36E-08  
Percentage of tax revenue - most volatile  -2.51E-02  5.75E-09  -8.82E-03  1.33E-08  -1.20E-02  2.59E-09  -1.23E-02  3.33E-09  
Percentage of tax revenue - least volatile  -2.10E-02  1.21E-08  -1.74E-02 * 2.72E-08  -1.82E-02 * 2.88E-09  -1.57E-02  1.26E-08  
Percentage of expenditure - most volatile  7.35E-02 ** 1.60E-08  9.11E-02 *** 2.22E-08  8.47E-02 *** 4.31E-09  8.58E-02 *** 9.81E-09  
Percentage of expenditure - least volatile  2.29E-02  1.20E-08  2.36E-02  1.12E-08  1.73E-02  3.03E-09  3.16E-02  4.12E-09  
Deviation from average per capita savings  1.09E+00  -9.08E-07  6.24E-01  -7.52E-07  6.82E-01  -1.94E-07  3.97E-01  -3.60E-07  
Tax effort  -1.77E-04  3.94E-11  -1.35E-04 ** 1.66E-10 ** -1.15E-04 * 1.29E-11 ** -1.84E-04 ** 8.81E-11 * 
Expenditure limitation  9.01E-03  -5.71E-11  7.17E-03  1.89E-09  8.12E-03  1.20E-10  5.41E-03  9.10E-10  
BBR stringency  1.49E-04  -8.32E-10 *** 2.15E-04  -2.34E-09 *** 2.91E-04  -2.83E-10 *** 3.81E-04  -1.11E-09 *** 
Deviation from average per capita debt  -3.17E-06  -6.35E-12 *** -4.00E-07  -1.48E-11 *** -9.88E-07  -1.28E-12 *** 4.63E-07  -6.83E-12 *** 
Population density  4.31E-06  -1.21E-11 *** 8.85E-06  -2.92E-11 *** 1.04E-05  -2.50E-12 *** 9.04E-06  -1.46E-11 *** 
Constitutional BSF  -3.59E-03  1.19E-05 *** -3.15E-03  8.39E-05 *** -2.60E-03  2.87E-06 *** -2.33E-03  6.38E-05 *** 
Deviation from average per capita net IG revenue   1.48E-02  5.48E-09  8.87E-03  -6.48E-09  1.16E-02  -1.52E-09  8.01E-03  -3.23E-09  
Log Likelihood  -153.05429  -155.28692  -153.23644  -154.11043  
*** Significant at 1%, ** Significant at 5%, * Significant at 10% 
The dependent variable equals zero in years prior to BSF adoption, in the year of adoption it equals 1 if the adopted deposit requirement is lax and 2 if it is strict 
 
 103
Table 3.5. Multinomial logits. Dependent variable: withdrawal requirement 
Description of independent variables  weak  strict  weak  strict  weak  strict  weak  strict  
Number of seats in upper House  2.77E-04 * 4.23E-06              
Number of seats in lower House  4.22E-06  -4.04E-06              
% seat gap between main parties (Upper House)      1.10E-04  -1.35E-06  1.06E-04  -1.20E-06  1.14E-04  -4.51E-07  
% seat gap between main parties (Lower House)      -1.01E-04  -4.22E-06 *** -1.02E-04  -3.62E-06 *** -9.26E-05  -1.69E-06 *** 
Democratic Governor              -1.19E-03  -3.52E-05  
Appointed Supreme Court          -1.45E-03  -5.38E-05      
Limit for governor's tenancy              -5.09E-03  -1.02E-05  
Deviation from average per capita personal income  -1.13E-06  -7.62E-08  -9.52E-07  -5.31E-08 * -9.60E-07  -4.91E-08 * -1.02E-06  -2.29E-08 * 
Percentage of earnings - farming  -4.49E-02 * -1.28E-03  -4.65E-02 * -1.61E-04  -4.69E-02 * -2.52E-04  -5.24E-02  -6.40E-05  
Percentage of earnings - construction  -3.39E-02  -1.22E-02 *** -4.78E-02  -3.65E-03 *** -4.35E-02  -3.09E-03 *** -6.47E-02  -1.34E-03  
Percentage of earnings - manufacturing  -3.68E-02 * -8.53E-04  -3.00E-02 * -3.55E-04  -2.97E-02 * -3.17E-04  -3.70E-02 ** -1.52E-04  
Percentage of earnings - mining  -3.90E-03  -6.70E-03  -4.44E-03  -1.66E-03 *** -5.84E-03  -1.77E-03 ** -1.01E-02  -6.38E-04 *** 
Percentage of tax revenue - most volatile  -4.27E-02 * 1.42E-03 * -3.73E-02 * 5.95E-04 ** -3.88E-02 * 4.75E-04 * -3.95E-02 ** 2.45E-04 * 
Percentage of tax revenue - least volatile  -2.11E-02  -4.86E-04  -2.42E-02 * -1.22E-04  -2.47E-02 * -1.29E-04  -2.05E-02  -5.41E-05  
Percentage of expenditure - most volatile  6.49E-02 ** 5.12E-03  8.42E-02 *** 1.64E-03 ** 8.18E-02  1.37E-03 * 7.55E-02 *** 6.92E-04 ** 
Percentage of expenditure - least volatile  -2.51E-03  2.60E-03  3.37E-03  5.07E-04  3.88E-04  4.65E-04  6.10E-03  2.30E-04  
Deviation from average per capita savings  1.12E+00  -1.40E-02  1.21E+00  -1.39E-02  1.25E+00  -9.67E-03  8.61E-01  -5.33E-03  
Tax effort  -4.08E-05  -8.43E-06  1.79E-05  -1.98E-06  3.05E-05  -1.87E-06  -8.33E-05  -7.42E-07  
Expenditure limitation  6.65E-03  4.92E-04  4.00E-03  2.88E-04 * 4.54E-03  2.77E-04  2.10E-03  9.75E-05 * 
BBR stringency  -3.06E-04  -4.14E-05  -1.52E-04  -5.08E-06  -1.30E-04  -1.32E-06  1.39E-04  -1.09E-06  
Deviation from average per capita debt  -7.03E-06  -3.23E-07 * -6.63E-06  5.32E-09  -7.02E-06  -8.91E-09  -5.53E-06  1.86E-09  
Population density  4.72E-06  -5.77E-09  3.16E-06  1.72E-07  4.09E-06  1.66E-07  4.50E-06  8.93E-08 * 
Constitutional BSF  -1.45E-03  5.63E-03 ** -1.25E-03  3.46E-03 *** -8.34E-04  4.56E-03 ** -9.49E-04  3.18E-03 ** 
Deviation from average per capita net IG revenue   1.09E-02  1.49E-03 ** 5.31E-03  2.78E-04  7.01E-03  2.96E-04  6.17E-03  1.20E-04  
Log Likelihood  -169.88079  -164.80986  -164.37042  -162.30124  
*** Significant at 1%, ** Significant at 5%, * Significant at 10% 
The dependent variable equals zero in years prior to BSF adoption, in the year of adoption it equals 1 if the adopted withdrawal requirement is lax and 2 if it is strict 
 
 104
3.5. Conclusions 
 
BSFs have become popular among states as tools to help them weather recessions and 
other adverse conditions. However, they are very disparate in nature, and the differences 
in terms of deposit and withdrawal requirements have a significant impact on their 
effectiveness. This chapter has investigated the factors that determined the choice of 
BSFs? configuration, using data from the second half of the twentieth century and 
extending or modifying the set of indicators used in the previous literature. In particular, 
we corrected the figure for the stringency of Tennessee?s BBR, considered a new 
measure of the resources easily available to the states (proposing an alternative to the 
measure of savings that had been previously used, which is advised against by the 
Census), introduced additional indicators of the political, economic and institutional 
particulars of the state, and proposed new methods that incorporate the ranked nature of 
the two requirements that have been proved to significantly affect the effectiveness of 
these funds: deposit and withdrawal rules. 
Our results provide several insights: one suggests that the two most lax categories of 
deposit and withdrawal requirements may be indistinguishable, so the possibility of 
collapsing them when analyzing BSFs should be considered. A second result has to do 
with the methodology used: intuitively, ordered techniques should be employed when 
analyzing these funds, but the violation of the proportional odds assumption makes the 
OL model unreliable. Tentative work with generalized ordered logit suggests this option 
may be superior to MNL, and further investigation is under way to apply intermediate 
techniques that would incorporate a sense or ordering into the analysis without imposing 
 105
excessively restrictive assumptions. A lesson to be drawn from this is that we must 
consider carefully the ordered nature of the requirements and the assumption  of 
proportional odds.  
Turning to the investigation of the factors that determined the decision to configure 
these funds,123 we find indications that bigger Senates are conducive to laxer deposit 
requirements and that more fragmented lower houses (which generally have high levels 
of control over the budget and more members) may be correlated with less stringent 
BSFs.124  
Among the economic variables, we find some evidence suggesting that states with 
higher levels of debt are more prone to establish weak deposit requirements but stricter 
withdrawal rules, and that the state earning?s composition may be a factor to take into 
account.125 In addition, we find evidence that states which receive higher percentages of 
their total tax revenue from relatively more volatile sources show some inclination to 
establish funds with stricter deposit requirements, as do states with higher levels of tax 
effort. Both effects provide some indication in favor of the hypothesis that states adopt 
these funds to accumulate resources in order to weather recessions. However, states that 
spend a higher proportion of their budgets on volatile spending categories are more likely 
to establish weak funds.  
                                                 
123 We do not have enough data to allow us for clustering by year. We have, however, run our regressions 
with clustering by economic cycle using the business cycles data reported by NBER. The resulting 
estimates are smaller in magnitude but none of the significant coefficients switches signs. 
124 Besley and Case?s (2001) finding that fiscal cycles exists in states where limits to governor tenure are 
binding may help explain why we find weak and scattered indications linking the existence of these rules 
with increases in the likelihood that a state will institute strict requirements for its BSF. 
125 To take into account the possibility of regional effects, we run our regressions clustering using the BEA-
defined regions and included regional dummies. When including regional dummies, the variables 
representing the New England states (Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, 
Vermont) and the South Atlantic states (Florida, Georgia, North Carolina, South Carolina, Virginia) were 
significantly more likely to adopt weak funds. 
 106
Other state institutions are relevant in the configuration decision, in line with Poterba?s 
(1994) suggestion that fiscal tools should not be studied individually. Firstly, states with 
stricter balanced budget requirements seem to be less likely to establish demanding 
deposit requirements (although no such result appears regarding withdrawal 
requirements). Also, although the mere existence of tax and expenditure limitations is not 
a significant factor in the configuration choices of states? RDFs, the part of the budget to 
which they apply is. In particular, the existence of more comprehensive TELs increases 
the likelihood that states will adopt weak deposit requirements, which suggests RDFs 
may be an attempt to avoid the restrictions imposed by these limits. Additional support 
for this idea comes from the results suggesting that BSFs that were not established by the 
legislature, but rather by voters, and that those embedded into the state constitution are 
more likely to have a strict configuration.  
Further work includes the investigation of the simultaneous choice of deposit and 
withdrawal requirements, with measures of the overall level of stringency of the fund and 
simultaneous estimation of deposit and withdrawal choices. Also, additional work on the 
importance of the ordered nature of these rules is needed. Alternative methods were 
briefly discussed that take into account the fact that the trade-off between the number of 
parameters to estimate and potential explanatory variables is of great importance and 
needs further consideration.  
The results of this study provide a stepping-stone to the discussion of this rather recent 
fiscal tool. BSFs have been found to have the potential to significantly reduce fiscal 
stress, but only if they are properly configured. Their impact on budget stabilization takes 
many forms: adequately designed BSFs improve the state?s credit rating, reduce the need 
 107
for hurried solutions to cash shortages (such as unplanned tax increases or cuts in 
spending) and significantly reduce the volatility of expenditure ?in particular, social 
spending. 
Our results suggest that fiscal characteristics, such as the levels of tax effort or 
volatility of state spending, are important factors for the choice of the form of these 
funds. However, we have gathered some evidence that indicates that factors other than 
budget stabilization may help explain the weak ?and less effective- configuration of 
many funds. Political factors, as well as other institutional constraints, also provide 
incentives that explain the configuration of the funds.  
Given the importance of these rules, states reconsidering the nature of their funds may 
benefit from rethinking the reasons that led to the actual configuration and include them 
in their discussions about the possibility of reform. These lessons may also be valuable 
for other countries, where increased subnational government fiscal responsibilities could 
make instruments for budget stabilization at these levels an attractive option. As with the 
U.S. experience, the institutional details of these funds are likely to be of major 
importance. Others who may consider establishing funds like these could benefit from the 
awareness of considerations other than the purely economic reasons that have impacted 
the choices embedded in BSFs. 
 
 
 
 
 
 
 108
Chapter 4. Conclusions and Research Prospects 
 
 
States have created a wide array of institutions to address fiscal matters and the ways 
in which these institutions affect policy outcomes are not easy to disentangle. Often, 
fiscal institutions are considered in a relatively short time frame, either because data 
becomes scarcer and less reliable as we go back in time or because changes in institutions 
may take place over a long period of time. This gives the appearance that fiscal 
institutions are immobile. As a consequence, fiscal institutions ?in particular those that 
have been in existence for a long period of time- are frequently considered as exogenous 
in economic research. A historical perspective helps us understand the origin and 
evolution of fiscal institutions and clarifies their effects on and relationships with other 
fiscal institutions.  
The specific details of the configurations of fiscal institutions are another feature that 
is sometimes ignored, despite its relevance. Indicator variables, that concentrate only on 
the absence of presence of a particular institution, rather than focusing on the form of the 
institution can produce misleading interpretations of the evolution of an institution. A 
clear classification of the different types of rules that can coexist under the same name is 
necessary before we can proceed to study their evolution and effects.  
This dissertation illustrates the need for a comprehensive treatment of fiscal 
institutions.  The creation, configuration and evolution of fiscal institutions are affected 
by the economic and political circumstances in which they operate, but also by other 
institutions. In fact, the discussion suggests that fiscal institutions, far from being 
exogenous and independent of each other, co-evolve.  
 109
Chapter one considered the issue of the creation of institutions under the light of the 
government?s decision to credibly pre-commit. In particular, the trade-off between the 
flexibility to address future problems and the advantages derived from tying the hands of 
future policy makers is at the core of the decision to establish institutions to deal with 
fiscal affairs. The uncertainty of economic conditions, paired with common pool 
problems and the incentives associated with the fact that public officials are subject to 
election, provide incentives to institutionalize the procedures by which  certain decisions 
are made. 
The choice of how to commit results in fiscal institutions that are used in all sorts of 
economic and political conditions. The effectiveness of one rule depends on the 
restrictions imposed by other rules. This is the reason why a general perspective of some 
of the states? fiscal institutions is necessary before analyzing any other one in detail. The 
evolution of debt restrictions provides us with a good example of an institution whose 
changing nature has been overlooked in favor of a more static view. Careful 
consideration of the history of the rules, however, provides evidence of what may be 
called ?recursive institutional change.? After institutions are created and used in the 
economic and political circumstances of the time, new problems arise. Sometimes it is an 
economic disturbance, or the unforecasted effects of the interaction of the institution with 
an existing one, or new developments that cause the revision of the original institution. In 
turn, this change creates a new set of responses, and even the introduction of new rules. 
Institutional change becomes then a process of recursive change, rather than a series of 
unconnected exogenous events. 
 110
Chapter two analyzes debt restrictions at the state level. Subnational debt is an 
important matter in the United States, where the size of the debt market is sizeable, and 
regulation of state debt issue is constitutional. We gathered data on the full history of the 
constitutional provisions that regulate debt in the states using the information available in 
the NBER/MD Constitution Project.126 This compilation provides previously unavailable 
data on debt restrictions. It contains information on the original rules created in the 
constitutions and their subsequent modifications,127 spanning over a hundred and fifty 
years and, frequently, several constitutions. The compilation includes provisions dealing 
with limits to state debt, as well as restrictions and regulations on municipal borrowing, 
itemized provisions allowing for debt issuance for specific purposes, and regulation of 
non-guaranteed debt. 
This dataset provides us with two main advantages. First, since the data covers the 
entire history of constitutional debt restrictions, it allows us to consider their evolution 
rather than a snapshot of the current situation. Secondly, it provides with coherent and 
comprehensive information on debt restrictions that goes beyond the dichotomy of 
existence/non-existence of a provision on debt issue.  The data set allows us to draw a 
more complete picture of what debt restrictions really look like. The dataset improves on 
previously used indicators of debt restrictions because its information is not restricted to 
one or a handful of provisions related to the limiting of debt, but includes a 
                                                 
126 The actual wording of these provisions is available from the author upon request. The assemblage of a 
full recount of state constitutions is a difficult task. Because this compilation would be severely flawed if 
only the original and/or last version of the constitutions were recorded it becomes necessary to include the 
successive amendments that have altered the constitutional texts without causing the creation of a ?new? 
constitution 
127 These texts are often not readily available and making sense of the string of modifications can be a 
difficult task. As such, the work on the database of historical state constitutions is still underway and 
because of insufficient information, data are currently unavailable for Vermont, Nebraska, Missouri, 
Georgia, and Tennessee. 
 111
comprehensive review of a more complete set of the constitutional provisions that play a 
role in the actual regulation of debt issuance. Frequently, the narrow view of debt limits 
is caused by two factors: considering only the part of the restriction that prescribes a 
certain amount as limit to debt, without paying attention to the modification procedures 
that transform the limit into a threshold for a different type of restriction on debt; and the 
oversight of the fact that regulations of debt may exist over more than one article. 
A careful definition of what constitutes a debt restriction involves some terminology. 
The first distinction separates debt limits from procedural restrictions on debt. Limits 
establish a maximum for debt (either in absolute or relative terms, or for overall debt or 
casual deficit). Procedural restrictions do not limit the amount of debt that can be issued, 
but require public officials to follow certain procedures for authorizing the debt issue.  
In addition, we need to consider the existence of multiple provisions that allow for the 
modification of the restriction and that can, in fact, significantly alter its nature. Most 
constitutional debt provisions explicitly include modification provisions that allow states 
to issue more debt than the maximum established by the limit if certain conditions (such 
as voter or legislative approval) are met. Previous research has often concentrated on the 
existence of debt limits, ignoring procedural restrictions. Furthermore, the implicit 
possibility of modifying the debt limit exists, though upward revision of the maximum or 
the inclusion in the constitution of separate debt items that are explicitly excluded from 
the list. Amending the constitution is difficult, but it may be easier than complying with 
the procedure required to issue debt above the limit.  
Chapter two discusses the evolution of state debt restrictions in context with the 
economic changes of the period, the shifting involvement of other levels of government, 
 112
and the creation of new types of debt. Overall, debt issues have shifted away from 
general obligation bonds to revenue-backed bonds, and a pattern of increased 
involvement of sub-state governments has been widely recognized. Part of these 
developments are attributable to debt restrictions at the state level, but other arguments 
can be made involving a better alignment of users with payers of services and the 
evolution of systems that allow for the identification of revenue sources for the service of 
debt. The literature, however, has ignored the shift towards flexibility in debt restrictions 
that occurred in the second half of the twentieth century and its potential implications for 
our vision of fiscal institutions in the states. This change becomes clear only when debt 
restrictions are put in perspective and their evolution over the course of their existence is 
considered.  
Debt restrictions appeared in the 1840s as a response to the states defaults in 1841 and 
1842. The objective of the new fiscal institutions was not as much to forbid debt but to 
include mechanisms that addressed the problems of taxless finance, concerns about the 
functioning of the political system and corruption. The restrictions spread to other states, 
reaching many of the Southern states in latter part of the XIX century. States continue to 
borrow, but the late XIX century saw a marked increase in local borrowing.  States 
responded by extending debt restrictions to local governments, beginning in the 1870s.  
Local governments responded to the new restrictions by developing the revenue bond and 
more extensive use of special districts an public authorities. 
At the beginning of the XX century, states began modifying debt limits by explicitly 
allowing debts for certain purposes.  Debt itemization, the use of revenue bonds, and the 
creation of special districts enabled state governments a great deal of latitude in 
 113
borrowing, even if the state constitution contained what appeared to be a strong limit on 
debt creation.  In response, several states in the nineteen sixties and seventies modified 
their constitutions to allow legislatures a considerably greater degree of flexibility, either 
on the limit itself or in the modification procedures.  
Immediately following the easing of debt restrictions, states began experimenting with 
new types of fiscal institutions. Several states incorporated balanced budget rules directed 
towards establishing procedures to regulate the budget process, and others laid down 
rules regulating appropriations. Tax and expenditure limitations (TELs) start appearing in 
the late seventies and were joined by large-scale adoption of budget stabilization funds in 
the mid-eighties.  
A major contribution of chapter two is to show how the accumulating body of debt 
restrictions in state constitutions, was not in fact producing a more restrictive 
environment for the creation of state or local government debt. Of the 24 cases where 
states wrote down absolute limits on the dollar amount of debt the state could incur, only 
3 instances actually produced an actual operating limit.  In the other 21 cases, states 
allowed themselves other avenues for authorizing debt issue. This previously 
unrecognized movement towards flexibility is paired with the creation of institutions 
aimed at increasing fiscal control, such as TELs and balanced budget rules, and is then 
followed shortly after by the enactment of rainy day funds, aimed at providing a 
mechanism to help states cope with some of the effects of these fiscal constrains.  
Rainy day funds are the object of study of chapter 3. As one of the latest institutions 
used in state budgeting, these funds have received a considerable amount of attention and 
are often regarded as potentially powerful institutions for budget stabilization, if they are 
 114
properly configured  (that is, if they include stringent rules, especially for deposit 
requirements). The particular structure of rainy day funds is a crucial characteristic in 
determining their effectiveness, and the literature has consistently shown that simply 
considering their existence does not provide accurate measures of their efficiency. Yet, 
no research has considered the reasons that lead states to adopt rainy day funds in such 
ways that compromise their ability to reduce fiscal stress. Chapter three takes a step in 
this direction, considering the choice of the structure of rainy day funds from an 
empirical perspective.  
In particular, the rules that regulate the ways resources are deposited into and drawn 
from the fund play an important role in determining the effect of rainy day funds. Both 
deposit and withdrawal requirements can be classified according to their stringency, 
which is inversely related to the degree of flexibility policy makers have in making 
decisions about the funds? finances. Chapter three describes in greater detail the deposit 
and withdrawal requirements, and how the various options within each category are 
ranked, concentrating on their ordered nature. It then goes on to discuss the potential 
empirical strategies to identify the reasons that lead states to select ?lax? or ?stringent? 
rules for their funds.   
The notion of an ?order? in these rules has existed for some time in the literature on 
rainy day funds, but little or no empirical work existed previously on the reasons for the 
specific configuration choices. We fill this niche in the literature and include the ordered 
nature of the requirements as an element to be explained. For that, we first consider the 
potential for amalgamation among the rules, as some of them have the potential of being 
close substitutes. Indeed, our results suggest that when considering the most 
 115
disaggregated classification we may take the two laxest levels of both deposit and 
withdrawal requirements to be very similar, opening the door to a possible collapse of the 
categories for empirical work.  
This discussion adds to the literature of ordered analytical techniques in the 
econometric analysis. However, care has to be exerted in the application of the simplest 
ordered methods in this case, since the satisfaction of the proportional odds assumption is 
not guaranteed. Because of the thinness of the data, more elaborated ordered techniques 
are difficult to implement, but there is some indication that methods that take into 
account the ordered nature of the rules may be superior to un-ordered multinomial 
methods. In general, the use of econometric techniques for research on these types of 
institutional issues shares the same type of problem, in particular if the object of research 
is the different configurations of the institution, rather than its existence, which requires 
an especially careful selection of the econometric technique. Because econometric 
analysis may be constrained by data availability, a comprehensive vision of the set of 
institutions and their interactions becomes particularly useful, as they can guide the 
research, and complement and inform the results provided by the regression analysis. 
Last, the empirical model is estimated using the states that adopted a rainy day fund 
during the sample period (1951-2000) to examine the role of political, socio-economic 
and institutional factors on the decision to select a particular configuration for the state?s 
rainy day fund.  We find evidence suggesting that states with larger Senates are more 
likely to adopt RDFs with laxer deposit requirements and that more fragmented lower 
houses increase the probability that the state adopts a less strict RDF. Among the 
economic variables, high levels of debt may make states more likely to adopt weak 
 116
deposit requirements, but stricter withdrawal rules. States with more volatile tax sources, 
or that exert a higher level of tax effort adopt stricter RDFs, providing support for the 
hypothesis that states adopted RDFs as a stabilization mechanism. Volatility of spending 
is also a relevant factor, favoring the adoption of weak RDFs. 
Economic motivations were not the only important determinant.  We find support for 
the idea that RDFs need to be considered as part of a bigger set of instruments, rather 
than individually. States with more stringent balanced budget requirements, as measured 
by the Advisory Commission of Intergovernmental Relations, are less likely to establish 
strict deposit requirements. There is also some suggestion in the results that the mere 
existence of institutions does not provide us with a complete picture of their effect on 
each other. The presence of tax and expenditure limits does not significantly affect the 
configuration decisions regarding RDFs, but we find that states with more comprehensive 
TELs are more likely to adopt weak deposit requirements. Paired with the result of the 
effect of balanced budget requirements, this provides some indication that RDFs may 
have been designed to increase the power of decision of legislatures, constrained by other 
rules. Additional support for this idea comes from the method of enactment of RDFs 
themselves, since RDFs that were created through voters? approval are more likely to 
have strict requirements.  
Although this dissertation provides some interesting insights and results, there is still 
much room for future research. A first avenue for future research includes the 
incorporation of other types of institutions. The institutions considered fall in the 
legislative arena, but executive control is an important part of the budgetary process. 
Including the institutions used to regulate this aspect of the process in the analysis would 
 117
provide us with a more complete picture of the ways in which institutions interact. In 
addition, the process has created other, smaller, institutions and instruments such as TIFs 
(tax increment financing) and PILOTS (payment in lieu of taxes).  
A second area for research should concentrate on empirical analysis. A first option is 
empirical investigation of the kind conducted on rainy day funds for other institutions, 
such as tax and expenditure limitation or the last wave of balanced budget rules. This 
could help us further our understanding of the processes that have led to the present 
nature of the institutions state use for fiscal matters. A second area of empirical work 
would take advantage of the dataset constructed on debt restrictions to investigate the 
effects of different types of debt restrictions on fiscal outcomes, and the effects of debt 
restrictions on other aspects, such as the proliferation of special districts and public 
authorities and the use of revenue bonds. Unlike much of the previous research, this 
investigation would be able to distinguish between the different types of debt restrictions 
and separate the effects of each, and use a longer time period for analysis.  
Within the area of debt restrictions, we have yet to analyze the data on debt 
restrictions at the local level. Because of the absence of a compilation of the 
constitutional restrictions in all states, most of the research has been centered on the study 
of particular cases.  Given the importance of local borrowing, the effects of local debt 
restrictions both on debt and on other fiscal outcomes from a larger perspective pose an 
interesting question. This research could benefit from the distinction between the 
different types of restrictions in existence at the state level, although an additional layer 
regarding the type of local government that is subject to the rule would need to be added. 
 118
Further work should incorporate statutory restrictions, to complement the constitutional 
provisions that restrict debt issuance both at the state and local levels. 
Last, because of their importance and potential effect on fiscal outcomes, rainy day 
funds remain an interesting venue of research. The existing research has frequently 
considered the rest of the other institutions at the state level, but frequently the measures 
used had to do with their existence and not their configuration. In addition, with a more 
complete description of debt restrictions, possibilities for better description of both the 
reasons that led to the adoption of rainy day funds (as well as the time and specific form 
of their enactment) and their effects on fiscal outcomes exist.  
Empirical research and careful consideration of the creation and evolution of fiscal 
institutions are complementary approaches, and we have much to gain from their 
integration. While the evolution may happen over extended periods of time, these 
institutions do change, and their modifications follow patterns that can (and should) 
inform our empirical questions. 
 
 
 
 
 
 
 
 
 
 119
Appendices 
 
Appendix A ? Appendix to chapter 2 
 
Appendix A.1. Tables. 
 
 
Table A.1. Government Debt by Level of Government  
Amount (in millions of dollars) Share of total Year 
Local State National Local State National 
1838 25 172 3 12.5% 86.0% 1.5% 
1841 25 193 5 11.4% 86.4% 2.3% 
       
1870 516 352 2436 15.6% 10.7% 73.7% 
1880 826 297 2090 25.7% 9.2% 65.0% 
1890 905 228 1122 40.1% 10.1% 49.8% 
1902 1877 230 1178 57.1% 7.0% 35.9% 
1913 4035 379 1193 72.0% 6.8% 21.3% 
1922 8978 1131 22963 27.1% 3.4% 69.4% 
1932 16373 2832 19487 42.3% 7.3% 50.4% 
1942 16080 3257 67753 18.5% 3.7% 77.8% 
1952 23226 6874 214758 9.5% 2.8% 87.7% 
1962 58779 22023 248010 17.9% 6.7% 75.4% 
1972 129110 59375 322377 25.3% 11.6% 63.1% 
1982 257109 147470 919238 19.4% 11.1% 69.4% 
1992 603920 372319 2998639 15.2% 9.4% 75.4% 
2002 1043900 642200 3540427 20% 12.3% 67.7% 
Sources: 1838-1992: Wallis (2000), 2002: US Census 
Notes: National debt is net debt held by public 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 120
 
 
 
Table A.2. Trends in outstanding municipal bonded debt -- by type  
Amount (in millions of dollars) Share of total 
Year Revenue 
Bonds 
General 
Obligation 
Bonds 
Revenue 
Bonds 
General 
Obligation 
Bonds 
1909 -- 3000 1% 100% 
1925 63 11638 1% 99% 
1931 325 16975 2% 98% 
1934 650 17250 4% 96% 
1937 1000 17800 5% 95% 
     
1966 3838 7241 35% 65% 
1967 4983 9423 35% 65% 
1968 6525 9795 40% 60% 
1969 3573 8129 31% 69% 
1970 6101 11981 34% 66% 
1971 8710 16219 35% 65% 
1972 9404 14289 40% 60% 
1973 10624 13197 45% 55% 
1974 9999 13561 42% 58% 
1975 14736 15964 48% 52% 
1976 17272 18143 49% 51% 
1977 28645 18061 61% 39% 
1978 30641 17711 63% 37% 
1979 30836 12467 71% 29% 
Source: Liu, Wallis (2007) 
 
 
Graph A.1: Trends in New Municipal Bond Issuances, 1975-1999 
0
10
20
30
40
50
60
70
80
90
Ye
ar 
19
75
19
76
19
77
19
78
19
79
19
80
19
81
19
82
19
83
19
84
19
85
19
86
19
87
19
88
19
89
19
90
19
91
19
92
19
93
19
94
19
95
19
96
19
97
19
98
Source: Liu, Wallis (2007)
Revenue Bonds (% of Total)
General Obligation Bonds (% of Total)
 
 121
 
 
Table A.3. Classification criteria of special districts   
Inter-governmental 
dependence 
Geographical 
scope 
Functional 
specialization 
Primary 
Revenue 
Debt 
Financing 
Highly dependent regional single purpose tax-based general obligation bonds 
Highly independent metropolitan multi-purpose non taxed-based revenue bonds 
 urban fringe    
 rural corporation    
Based on Leigland (1993) 
 
 
 
Table A.4. State debt limits and associated modification procedures 
Limit 
Procedural restrictions and  
modification provisions 
associated with the limit State 
Limit Year Limit  Procedure restriction Year PR Modification provision Year MP 
Alabama Absolute [cd] 1875     
 Absolute 1901   DI 1922 
Arizona Absolute [cd] 1912 V 1912   
California Absolute 1847 V 1847   
  Absolute 1879 V 1879   
      DI 1919 
    SL 1960   
Colorado Relative 1876   DI 1920 
      U 1993 
Delaware Absolute [cd] 1897 SL 1897   
Florida Absolute 1868   DI 1875 
Hawaii Absolute 1959 SL 1959   
 Relative 1968     
Idaho Relative 1890 V 1890   
 Absolute 1912 V 1912   
Iowa Absolute 1847 V 1847   
 Absolute [cd] 1857 V 1857   
Illinois Absolute [cd] 1848 V 1848 U 1870 
  Relative 1970 V or SL 1870   
Indiana* Absolute [cd] 1851     
Kansas Absolute 1859     
Kentucky Absolute [cd] 1850 SL 1850   
Louisiana Absolute 1845 V and SL 1845   
  Absolute 1852 SL 1852 U 1852 
  no debt 1910   DI 1818 
  Relative 1993 SL 1993   
 122
Table A.4 Cont?d. State debt limits and associated modification procedures 
Limit 
Procedural restrictions and  
modification provisions 
associated with the limit State 
Limit Year Limit  Procedure restriction Year  PR Modification provision Year MP 
Maine Absolute 1847   DI 1912 
      U 1919 
    V and SL 1951   
Maryland Absolute [cd] 1851   DI 1867 
Michigan Absolute [cd] 1850   U 1909 
      DI 1917 
    V and SL 1954   
Minnesota Absolute 1857   DI 1872 
Mississippi Relative 1960     
Montana Absolute 1889 V 1889   
    SL 1973   
Nevada Absolute 1864   DI 1924 
  Relative 1916   U 1989 
New Jersey Absolute 1844 V 1844   
  Relative 1947 V 1947   
New Mexico Absolute [cd] 1911 V 1911   
      DI 1921 
New York Absolute [cd] 1846 V and SL 1846   
  Absolute [cd] 1894 V and SL 1894 DI 1894 
  Absolute  1938 V and SL 1938 U 1938 
North Carolina Relative 1924 V and SL 1868   
  Absolute 1936 V 1936   
    V and SL 1971   
North Dakota Absolute [cd] 1889   U 1918 
  Absolute 1918   DI 1948 
Oklahoma Absolute [cd] 1975     
Ohio Absolute [cd] 1851   DI 1947 
  Relative 1999 SL 1999   
Oregon Absolute 1857   DI 1912 
      U 1920 
Pennsylvania Absolute [cd] 1839   DI 1918 
      U 1874 
  Relative 1969 V 1969   
Rhode Island Absolute 1843 V 1843   
South Dakota Absolute 1889     
Texas Absolute 1845   U 1876 
 Absolute [cd]* 1876   DI 1917 
  Relative 1997     
Utah Absolute 1895   DI 1997 
  Relative 1911     
       
 123
Table A.4 Cont?d. State debt limits and associated modification procedures 
Limit 
Procedural restrictions and  
modification provisions 
associated with the limit State 
Limit Year Limit  Procedure restriction Year  PR Modification provision Year MP 
Virginia Absolute [cd]* 1870 SL 1902   
 Relative 1928   DI 1920 
  Relative 1971     
Washington Absolute [cd] 1889 V 1889 DI 1972 
  Relative 1972 V 1972   
West Virginia Absolute [cd]* 1872   DI 1920 
Wisconsin Absolute 1848     
  Relative 1969 V 1969   
Wyoming Relative 1889 V    
Notes: This table includes the debt limits, absolute or relative, and the associated modification 
procedures. Procedural restrictions (that is, voter and legislature approval requirements that are not 
associated with a limit but are rather self-standing) are not included. 
* Indiana, Texas, Virginia, West Virginia establish debt can only be issued to meet casual deficits, but 
does not establish a limit 
Limit: Absolute [cd] (absolute limit for debt for casual deficit) Absolute (absolute limit debt).   
Explicit procedure: V (voter approval), SL (legislature approval) 
Implicit procedure: DI (debt itemization) U (upwards revision of the limit) 
Note: Vermont, Nebraska, Missouri, Georgia, Tennessee NA 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 124
Table A.5. Changes to constitutional regulations on debt related to approval method at state level 
State Subdivision Year of change 
Super- 
majority 
and voter 
approval 
Super- 
majority 
or voter 
approval 
Super- 
majority 
Voter 
approval 
Louisiana West South Central 1852   x  
Virginia South Atlantic 1902 x    
New Mexico Mountain 1911    x 
Nevada Mountain 1916     
North Carolina South Atlantic 1924     
Arkansas West South Central 1934    x 
New Jersey Middle Atlantic 1947    x 
Maine New England 1955 x    
California Pacific 1960 x  x  
Michigan East North Central 1965 x    
Florida South Atlantic 1968    x 
Pennsylvania Middle Atlantic 1969    x 
Wisconsin East North Central 1969     
Illinois East North Central 1970  x x  
Washington Pacific 1972   x  
Montana Mountain 1973   x  
Louisiana West South Central 1974   x  
South Carolina South Atlantic 1977  x   
Texas West South Central 1977     
Colorado Mountain 1993    x 
Idaho Mountain 1998    x 
North Dakota West North Central 1918/1981     
Note: Vermont, Nebraska, Missouri, Georgia, Tennessee NA 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 125
Table A.6. Changes in the limits to the amount of debt states are allowed to issue 
  Situation after change 
  No debt Absolute limit Casual deficit Anticipation of revenues Percentage of revenue 
No restriction 
on amount 
of debt ? 
No debt      Louisiana (1974) 
Louisiana 
(1910)    
Nevada 
(1916) 
Maine         
(1955) 
    New Jersey (1947) Michigan (1964) Absolute limit 
    Washington (1972)  
   Pennsylvania* (1969) New Mexico* (1911) Colorado (1993) 
    Texas (1977)  Casual deficit 
    North Dakota (1918)  
Anticipation 
of revenues       
Percentage 
of revenue  
Idaho 
(1912) 
North Dakota 
(1981) 
 North Carolina* (1936)   
Arkansas 
(1934)   
Maine* 
(1967) 
Florida 
(1968)  
Or
igi
na
l s
itu
ati
on
 
No restriction on 
amount of debt?     Wisconsin* 
(1969)  
* Denotes this new category was added and did not substitute the previous one 
? This does not include procedural restrictions for the issuance of debt (supermajority, voter approval, etc) 
Note: Vermont, Nebraska, Missouri, Georgia, Tennessee NA 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 126
Table A.7- Constitutional rules on budget balance, by type and date 
State Type Year  State Type Year 
Massachusetts p 1780  Maryland p 1916 
Iowa d 1846  West Virginia d 1918 
Florida er 1848  Maine d 1919 
Wisconsin er 1848  Alabama ai 1932 
Kentucky er 1850  Arkansas d 1934 
Indiana d 1851  New York p 1938 
Ohio er 1851  New Jersey ai 1947 
Kansas er 1859  Alaska d 1959 
Nevada er 1864  Michigan ai 1964 
Georgia ai 1868  California p 1966 
Illinois ai / p 1870 / 1970  Pennsylvania p 1969 
Colorado ai 1876  North Carolina p 1971 
Texas p 1876  Louisiana ai 1974 
Montana ai 1889  Tennessee d 1974 
North Dakota d 1889  Hawaii ai 1978 
South Dakota d 1889  Delaware ai 1980 
Washington d 1889  Virginia er 1984 
Wyoming d 1889  Rhode Island ai 1986 
Idaho ai 1890  Connecticut er 1992 
Utah ai 1895  Minnesota -- -- 
South Carolina d 1896  Mississippi  --  -- 
Oklahoma p 1907  New Hampshire  --  -- 
New Mexico d 1911  Vermont   --   -- 
Arizona er 1912  Missouri p NA 
Oregon er 1912   Nebraska d NA 
Notes: "True" balanced budget rules: (p) certain procedure must be followed  (ai) 
appropriations shall not exceed revenues | "Other" rules (er) expenditures shall not  
exceed revenues  (d) limit on debt 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 127
Table A.8- Debt provisions & Levels of government, ranked by number of units per capita 
States with more total sub-county units per capita than national average 
State Limit Procedural restriction Modification method 
North Dakota Abs (1918)  DI, U 
South Dakota Abs (1889)   
Kansas Abs (1859)   
Minnesota Abs (1857)  DI 
Nebraska NA   
Vermont NA   
Maine Abs (1847) V and SL (1951) DI, U 
Wisconsin Abs (1848)/Rel (1969) V (1969)  
Iowa Abs [cd] (1857) V (1857)  
Indiana Abs [cd] (1851)    
    
States with more special districts per capita than national average 
State Limit Procedural restriction Modification method 
North Dakota Abs (1918)  DI, U 
Wyoming Rel (1889) V (1889)  
Idaho Abs (1912) V (1912)  
Nebraska NA   
Kansas Abs (1859)   
Montana Abs (1889) V (1889)/SL (1973)  
Colorado Rel (1876)  DI, U 
Oregon Abs (857)  DI 
South Dakota Abs (1889)   
Washington Abs [cd] (1889)/Rel(1972) V (1889)/V(1972) DI 
Missouri NA   
Illinois Abs (1848)/ Rel (1970) V (1840) / V or SL (1870) U  
Delaware Abs [cd] (1879) SL (1879)  
Arkansas ND(1874)/VA(1934)  DI 
Vermont NA     
Notes: ND: no debt, VA: voter approval, Abs: absolute limit, Abs [cd]: Absolute limit on casual 
debt, Rel: relative limit, V: voter approval, SL: legislative approval, DI: debt itemization, U: upward 
revision. If several provisions existed for the same state, only the relevant ones to the period 1952-
2002 are included in the graph  
 
 
 
 
 
 
 
 
 
 
 
 
 128
Table A.9. RDF requirements and increased flexibility of debt restrictions 
State deposit withdrawal 
Florida 0 0 
Hawaii 0 1 
Idaho* 0 1 
Illinois 0 0 
Louisiana 0 0 
Mississippi 0 0 
North Carolina 0 0 
Ohio 0 0 
Oklahoma 0 1 
Pennsylvania 0 1 
South Carolina 1 0 
Texas 0 0 
Virginia 1 1 
Washington 0 1 St
ate
s t
ha
t c
ha
ng
ed
 th
eir
 de
bt 
pro
vis
ion
s t
o 
inc
rea
se 
the
 fle
xib
ilit
y d
uri
ng
 19
60
-20
00
 
Wisconsin 1 0 
 # of states with strict requirements 3 6 
 % of states with strict requirements 20% 40% 
State deposit withdrawal 
Alaska 0 0 
Arizona 1 1 
California 0 0 
Connecticut 0 1 
Delaware 0 1 
Indiana 1 1 
Iowa 0 0 
Kansas 1 0 
Kentucky 0 0 
Maine 0 0 
Maryland 1 0 
Massachusetts 0 0 
Michigan 1 1 
Minnesota 0 0 
Nevada 1 0 
New Hampshire 0 0 
New Jersey 0 0 
New Mexico 0 0 
New York 1 0 
North Dakota 0 1 
Rhode Island 0 0 
South Dakota 0 0 
Utah 0 0 
West Virginia 0 0 
Sta
tes
 th
at 
did
 no
t c
ha
ng
e t
he
ir d
eb
t p
rov
isi
on
s t
o  
inc
rea
se 
the
 fle
xib
ilit
y d
uri
ng
 19
60
-20
00
 
Wyoming 0 0 
 # of states with strict requirements 7 6 
  % of states with strict requirements 28% 24% 
Note: includes only states with RDFs.  Vermont, Nebraska, Missouri, Georgia, 
Tennessee NA. Deposit and withdrawal requirements: 1 if strict, 0 if weak 
 
 129
 
Table A.10- Constitutional rules on budget balance and debt provisions 
State 
Type 
BBR Year 60-00  State 
Type 
BBR Year 60-00 
Massachusetts  p 1780   West Virginia  d 1918  
Iowa  d 1846   Maine  d 1919  
Florida  er 1848 x  Alabama  ai 1932  
Wisconsin  er 1848 x  Arkansas  d 1934  
Kentucky  er 1850   New York  p 1938  
Indiana  d 1851   New Jersey  ai 1947  
Ohio  er 1851 x  Alaska  d 1959  
Kansas  er 1859   Michigan  ai 1964  
Nevada  er 1864   California  p 1966  
Georgia  ai 1868   Pennsylvania  p 1969 x 
Colorado  ai 1876   North Carolina  p 1971 x 
Texas  p 1876 x  Louisiana  ai 1974 x 
Montana  ai 1889 x  Tennessee  d 1974  
North Dakota  d 1889   Hawaii  ai 1978 x 
South Dakota  d 1889   Delaware  ai 1980  
Washington  d 1889 x  Virginia  er 1984 x 
Wyoming  d 1889   Rhode Island  ai 1986  
Idaho  ai 1890 x  Connecticut  er 1992  
Utah  ai 1895   Minnesota  -- --  
South Carolina  d 1896 x  Mississippi   --  -- x 
Oklahoma  p 1907 x  New Hampshire   --  --  
New Mexico  d 1911   Vermont    --   --  
Arizona  er 1912   Illinois  ai / p 1870 / 1970 x 
Oregon  er 1912   Missouri  p NA  
Maryland  p 1916     Nebraska  d NA   
Notes: "True" balanced budget rules: (p) certain procedure must be followed  (ai) appropriations shall 
not exceed revenues | "Other" rules (er) expenditures shall not exceed revenues  (d) limit on debt        
Debt provisions: 60-00: X if the state modified its debt provision during the last considered period (1960-
2000) Type: type of debt provision (excluding modification procedures, both explicit and implicit), -- 
?missing? 
 
 
 
 
 
 
 
 
 
 
 130
Appendix A.2. Samples of constitutional provisions on debt restrictions 
 
A.2.1. Example of constitutional absolute limit on debt qualified with explicit 
modification procedure. 
Constitution of New Jersey (1844), Article 4, Section 6, Part 4:  
?The legislature shall not, in any manner, create any debt or debts, liability or 
liabilities, of the State which shall, singly or in the aggregate with any previous debts or 
liabilities, at any time exceed one hundred thousand dollars, except for purposes of war, 
or to repel invasion, or to suppress insurrection, unless the same shall be authorized by a 
law for some single object or work, to be distinctly specified therein; which law shall 
provide the ways and means, exclusive of loans, to pay the interest of such debt or 
liability as it falls due, and also to pay and discharge the principal of such debt or liability 
within thirty five years from the time of the contracting thereof, and shall be irrepealable 
until such debt or liability, and the interest thereon, are fully paid and discharged; and no 
such law shall take effect until it shall, at a general election, have been submitted to the 
people, and have received the sanction of a majority of all the votes cast for and against 
it, at such election; and all money to be raised by the authority of such law shall be 
applied only to the specific object stated therein, and to the payment of the debt thereby 
created. This section shall not be construed to refer to any money, that has been, or may 
be, deposited with this State by the government of the United States.? 
 
 
 
 131
 
A.2.2 Example of constitutional procedural restriction 
Constitution of North Carolina (1868), Article 2, Section 16: 
?No law shall be passed to raise money on the credit of the State, or to pledge the faith 
of the State directly or indirectly for the payment of any debt, or to impose any tax upon 
the people of the State, or to allow the counties, cities, or towns to do so, unless the bill 
for the purpose shall have been read three several times in each house of the general 
assembly, and passed three several readings, which readings shall have been on three 
different days, and agreed to by each house respectively, and unless the yeas and nays on 
the second and third readings of the bill shall have been entered on the journal.?  
 
A.2.3. Example of constitutional absolute limit on casual deficit debt qualified with 
explicit modification procedure. 
Constitution of Illinois (1870), Article 4, Section 18:  
[?] ?The State may, to meet casual deficits or failures in revenues, contract debts, 
never to exceed in the aggregate two hundred and fifty thousand dollars; and moneys thus 
borrowed shall be applied to the purpose for which they were obtained, or to pay the debt 
thus created, and to no other purpose; and no other debt, except for the purpose of 
repelling invasion, suppressing insurrection, or defending the State in war (for payment 
of which the faith of the State shall be pledged), shall be contracted, unless the law 
authorizing the same shall, at a general election, have been submitted to the people and 
have received a majority of the votes cast for members of the general assembly at such 
election.?  
 132
Appendix B ? Appendix to chapter 3 
 
Appendix B.1 - Tables 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Table B.1.1. Dates of adoption of states? Budget Stabilization Funds 
State Year of adoption State Year of adoption State Year of adoption 
Alabama .   Louisiana 1990   Ohio 1981 
Alaska 1986   Maine 1985   Oklahoma 1985 
Arizona 1990   Maryland 1986   Oregon . 
Arkansas .   Massachusetts 1986   Pennsylvania 1985 
California 1985   Michigan 1977   Rhode Island 1985 
Colorado .   Minnesota 1981   South Carolina 1978 
Connecticut 1979   Mississippi 1982   South Dakota 1991 
Delaware 1977   Missouri 1992   Tennessee 1972 
Florida 1959   Montana .   Texas 1987 
Georgia 1976   Nebraska 1983   Utah 1986 
Hawaii 2000   Nevada 1994   Vermont 1988 
Idaho 1984   New Hampshire 1987   Virginia 1992 
Illinois 2000   New Jersey 1990   Washington 1981 
Indiana 1982   New Mexico 1978   West Virginia 1994 
Iowa 1992   New York 1945   Wisconsin 1981 
Kansas 1993   North Carolina 1991   Wyoming 1982 
Kentucky 1983   North Dakota 1987     
Notes: ?.? indicates the state does not have a BSF. 
Source: Wagner (2004) and documents for the state of Colorado 
 133
Table B.1.2- Deposit and withdrawal requirements in the BSFs 
Deposit requirements Withdrawal requirements 
Alabama . Montana .  Alabama . Montana . 
Alaska 1 Nebraska 2  Alaska 1 Nebraska 2 
Arizona 4 Nevada 4  Arizona 4 Nevada 2 
Arkansas . New Hampshire 2  Arkansas . New Hampshire 2 
California 2 New Jersey 2  California 2 New Jersey 2 
Colorado . New Mexico 2  Colorado . New Mexico 1 
Connecticut 2 New York 4  Connecticut 3 New York 2 
Delaware 2 North Carolina 2  Delaware 3 North Carolina 1 
Florida 2 North Dakota 2  Florida 2 North Dakota 4 
Georgia 2 Ohio 2  Georgia 1 Ohio 1 
Hawaii 1 Oklahoma 2  Hawaii 3 Oklahoma 3 
Idaho* 1 Oregon .  Idaho* 3 Oregon . 
Illinois 2 Pennsylvania 2  Illinois 1 Pennsylvania 3 
Indiana 4 Rhode Island 1  Indiana 4 Rhode Island 2 
Iowa 1 South Carolina 3  Iowa 1 South Carolina 2 
Kansas 3 South Dakota 2  Kansas 1 South Dakota 2 
Kentucky 2 Tennessee 3  Kentucky 1 Tennessee 2 
Louisiana 2 Texas 2  Louisiana 1 Texas 2 
Maine 2 Utah 2  Maine 1 Utah 2 
Maryland 3 Vermont 2  Maryland 1 Vermont 2 
Massachusetts 1 Virginia 4  Massachusetts 1 Virginia 4 
Michigan 4 Washington 2  Michigan 4 Washington 3 
Minnesota 1 West Virginia 2  Minnesota 1 West Virginia 2 
Mississippi 1 Wisconsin 3  Mississippi 1 Wisconsin 2 
Missouri 1 Wyoming 1  Missouri 1 Wyoming 1 
Deposit requirements: (1) appropriation (2) genera fund surplus (3) required appropriation (4) formula  
Withdrawal requirements: (1) appropriation (2) revenue shortfall (3) supermajority required (4) formula  
* Idaho modified its BSF in 1999, making it stricter. Here we record the original requirements as they were established 
when the BSF was adopted in 1981. 
Source: Wagner (2004) and documents of the state of Colorado. 
 
 
Table B.1.3. Deposit and 
withdrawal requirements in BSFs 
  deposit 
  1 2 3 4 
1 7 8 2 0 
2 1 10 3 2 
3 2 5 0 0 with
dra
wa
l 
4 0 1 0 4 
 
 
 
 
 134
Appendix B.2. Other rainy day fund rules 
 
Caps 
5% (usually of expenditure, although other bases exist) is a common number for rainy 
day fund caps128 and one that is widely accepted to have spread from a comment by a 
rating agency executive129, which is why we are not considering it in our analysis. 
However, there is wide evidence that for most states 5% would not suffice in the face of 
an economic downturn130 (see, for example, Lav and Berube (1999), Joyce (2001), Kriz 
(2003), and Wagner and Elder (2003)). The question then turns to what this limitation 
could do to the operation of the fund if it was to be binding and what it signals with 
respect to the general philosophy that governs the fund. Establishing caps for the RDF 
balance has the potential to reduce its stabilization ability if the limit was to be binding 
and the recession deep enough to make desirable the presence of higher balances.131 
 
 
 
 
                                                 
128 Caps are most frequently set at 5%, but 7% and 10% are also common figures. The overall average cap 
is slightly above 6%. 
129 The 5% number was inspired by declarations from Robert H. Mueller who, while being vice assistant to 
the Standard and Poor Corporation and later vice president of the Morgan Guarantee Trust, referred to it as 
a ?key financial number? and a ?good solid number for a state surplus.?  
130 A 1999 report by the Center on Budget and Policy Priorities set the overall desirable level for a BSF at a 
low of 15 percent of a year?s worth of expenditures. The Government Finances Officers Association 
recommends maintaining a minimum of between 5 and 15% of regular general fund operating revenues or 
no less than one-two months worth of general fund operating expenditures as unreserved fund balance in 
the state?s general fund, although notes are made to clarify that significantly higher balances may be needed 
for some governments in special circumstances, and that these figures should be placed in a long-term 
context to avoid putting excessive emphasis on transitory situations. 
131 It seems that the caps have indeed been binding: The National Association of State Budget Officers 
reported in 2004 that the average total balance for rainy day funds in the period 1979-2003 was 5.2 per cent 
(Budgeting Amid Fiscal Uncertainty, NASBO 2004). 
 135
 Replenishment requirements 
Replenishment requirements call for a refilling of fund coffers within a certain period 
of time. They have not received as much attention from the literature as the other three 
rules, perhaps because they are a rare feature among the funds. However, replenishment 
requirements are often cited in the political and academic circles132 as being a restriction 
that can potentially render the fund useless, since the state could decide not to use it in a 
year of recession under the fear of the obligation of having to replenish the fund in the 
near future when the situation (far from getting better) may be getting worse. We do not 
pursue their analysis further due to their relatively low implementation and the fact that 
states are taking steps towards modifying or entirely eliminating these restrictions. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
                                                 
132 See Lazere (2003) and recommendations of the Office of the State Budget Director of Kentucky for 
examples of arguments against the existence of replenishment requirements. 
 136
Appendix B.3 ? Description of the data. 
 
Among the political variables, we include measures of House composition and 
fragmentation,133 as well as indicators of divided control, affiliation of the executive, 
appointment method of the Supreme Court and ideology measures.  
We expect the size of the Houses to matter in light of the findings that states with 
larger Houses spend more. This effect can be explained using Tullock?s (1959) theory of 
the tragedy of the commons: government spending typically benefits a small fraction of 
the population, while the taxes used to fund it are spread among all taxpayers. Weingast 
et al (1989) formally expressed this issue in the ?law of 1/n?, where n represents the 
number of districts. They show that constituents only pay one nth of any public spending 
they receive, becoming obvious that spending increases with the legislature size. Gilligan 
and Matsusaka (1999) provide a model that links the ability to alter fiscal policy by 
gerrymandering with the number of seats. These theories appear to be supported by the 
data: Gilligan and Matsusaka (2001) find that larger upper houses are associated with 
higher spending (independently of the composition of the legislature) through the 20th 
century,134 this does not, however, apply to the Lower House.135  
The composition of the legislature has also been shown to be relevant in fiscal 
outcomes.136 Poterba (1994) finds that having a divided government slows a state?s 
                                                 
133 Given its unicameral nature, wherever bicameral measures are computed, Nebraska is excluded from the 
regressions. Although the choice between bi and unicameralism is potentially relevant in fiscal terms, 
Heller (1997) shows that bicameralism may lead to higher deficits) the fact that only one state has opted to 
operate with only one House makes it impractical to consider its implications for BSF choices. 
134 Due to data constraints, the years corresponding to World War II are excluded from their analysis. 
135 Gilligan and Matsusaka (2001) did several robustness checks to investigate the theoretically puzzling 
effect of the disparity of effects across Houses, the result always remained the same. 
136 Gilligan and Matsusaka (1997) and McCarty (1999) provide brief reviews of theories and empirical 
applications for the interested reader. 
 137
reaction to a fiscal crisis; Crain and Muris (1995) conclude that divided governments 
spend less, while Gilligan and Matsusaka (1995) fail to find significant effects. Alt and 
Lowry (1994) provide a theoretical framework to explain the importance of partisan 
composition for state fiscal policy and, for the period 1968-87, find empirical evidence 
suggesting that democrats tend to tax and spend more and that divided governments have 
a reduced capacity to respond when facing revenue shocks. To investigate the extent of 
the importance of these two factors for the choice of BSF configuration, we consider the 
number of seats as well as the percentage gap in the numbers of seats held by each of the 
two main parties.  
The affiliation of the governor is also relevant for state budgeting:137 governors submit 
budgets, can veto bills, and have in general a wide array of executive powers that can 
affect state fiscal policy. This effect can be particularly noticeable if the legislative and 
executive powers are of opposite political parties.138 To investigate these possibilities, 
specifications with a dummy taking the value of 1 if the elected governor belongs to the 
Democratic Party and/or interactions with the polarization measures explained above 
were included. Also, 36 states currently have a limit on how long governors can serve, 
since term limits have been receiving considerable attention in the literature as a potential 
way to limit government and encourage fiscal responsibility (Basham (2001) and New 
                                                 
137 Lowry et al (1998) find that gubernatorial electoral effects of an increase in the size of the state budget 
vary by party: Republican candidates loose votes, while Democrats may be rewarded. 
138 Krause (2000) finds that, during the period 1948-95, ideological divergences across the different 
branches of the federal government were related with fiscal deficit. At the state level, Alt and Lowry (1994) 
find that Democrats tax and spend more, and divided governments are less able to react to revenue shocks. 
Additionally, Alt and Lowry (2000) explain how a governor?s budget proposal may be rejected by a 
legislature of the opposite sign, the budget may then remain in the status quo if the governor decides to veto 
the legislature?s preferred proposal and the opposing party in the legislature is not able to override the veto 
in both chambers.  
 138
(2001)) a dummy variable that takes the value of one every year after a term limit is 
adopted is also included. Table B.3.1 presents information on gubernatorial term limits. 
I have drawn on a variety of sources to construct a long-ranging series of the states? 
political make-up. The main source for both the numbers of seats held by each party and 
the affiliation of the governor comes from the Inter-university Consortium for Political 
and Social Research (ICPSR) study #0016. The data for the period 1838-1868 were 
compiled by Dean Burnham and later data were added by ICPSR resulting in a unified 
dataset under the name ?Partisan Division of American State Governments, 1834-1985? 
From this study I have extracted the data for the period 1924-1984.  
 
Table B.3.1. States with Gubernatorial term limits 
State year approved state year approved state year approved 
Alabama 1968 Kentucky 1992 North Carolina 1977 
Alaska 1959 Louisiana 1812 Ohio 1992 
Arizona 1992 Maine 1993 Oklahoma 1966 
Arkansas 1992 Maryland 1947 Pennsylvania 1874 
California 1990 Michigan 1992 Rhode Island 1992 
Colorado 1990 Mississippi 1890 South Carolina 1980 
Delaware 1787 Missouri 1821 South Dakota 1972 
Florida 1992 Montana 1992 Tennessee 1978 
Georgia 1976 Nebraska 1966 Utah 1994 
Hawaii 1978 Nevada 1970 Virginia 1851 
Indiana 1851 New Jersey 1844 West Virginia 1872 
Kansas 1972 New Mexico 1986 Wyoming 1992 
Source: U.S. Term Limits, 2003 
 
In the period 1924-74, the ICPSR dataset reports as missing values the inter-election 
periods. I have taken the division of seats to remain unaltered between elections, so that ?
for example- if democrats were reported to hold 34 seats in the upper House in 1968, this 
number will remain unchanged until new data are reported. This procedure should not 
introduce serious distortions, since deaths, resignations and other events that may change 
 139
the composition are not likely to cause a major change in the division of power within the 
houses. 
For the period 1985-2000, the data come from the elections section of the Statistical 
Abstract of the United States (various editions), as compiled by the U.S. Census Bureau. 
However, neither source had information for Minnesota before 1974. We used the data 
compiled by the Minnesota Legislative Reference Library for party control of both the 
House and the Senate (the caucus strength data for this period was collected using 
unofficial legislative directories of the Minnesota Railroads Association) and party 
affiliation of governor from 1951 onwards, so the series for affiliation of governor and 
party control in the Senate and the House of Representatives in Minnesota were extended 
using this information for the period 1951-1973. For most of the years in this period, the 
Legislative Reference Library provides only one set of seat distribution. However, for the 
House of Representatives, several figures are presented for the years 1971 (caused by the 
death of a representative who was substituted via election by a representative from the 
same party a month later, causing two different possibilities for the seat split in that year) 
and 1973 (due to the resignation of two representatives and subsequent elections of two 
additional ones later in the year, which results in five different potential data points). To 
choose one of them, I take advantage of the overlapping of this series with that provided 
by ICPSR. In particular, we want the numbers to be as comparable as possible, so ideally 
we would like to choose the combination that more closely resembles that of the ICPSR 
dataset. In two out of the three comparable cases the ICPSR reports the earliest 
information, so this will be the one I will use for 1971 and 1973. A similar problem came 
up with the data for governor?s affiliation in Nebraska for the period 1938-1969 and 
 140
1985-2000, which was reported as missing in the ICPSR dataset. The information for the 
period 1938-2000 comes then from the Nebraska Library Commission, which provides 
data on name, party affiliation, years of rule and some other details (such as reason for 
early termination) for Nebraska?s governors starting in 1854. 
An alternative measure of the political structure is presented in Berry et al?s (1999) 
measures of citizen and government ideology. From their data we can see that, on 
average, the governments? ideology leans more towards liberalism than the citizens? 
during most of the period,139 and it seems to be more volatile - with a standard deviation 
50% higher than that of the citizens? index. 
Lastly, the appointment method of the State Supreme Court (elected vs appointed) is 
included due to the fact that judges often enforce fiscal rules. Bohn and Inman (1996) 
report that elected Supreme Courts are associated with larger surpluses in state budgets, 
as compared to surpluses in states where courts are appointed. Using their data on the 
nature of the courts, a dummy variable that takes the value of one if the Supreme Court is 
elected and zero otherwise is included in the regressions. The cost of establishing a strict 
BSF (in terms of loss of flexibility for policy making) is likely to be lower if the Court is 
appointed, because appointed judges may be more amiable to deviations from the rule. 
Hence, we expect states with appointed courts to be more prone to establishing 
demanding BSFs. Table B.3.2 presents the method of appointment in the fifty states. 
 
 
 
 
                                                 
139 The only two exceptions are the periods 1967-73 and 1996-02. 
 141
Table B.3.2. Method of appointment of the Supreme Court  
State Method State Method   State Method  
Alabama E   Louisiana E   Ohio E 
Alaska A   Maine A   Oklahoma E 
Arizona A   Maryland A   Oregon E 
Arkansas E   Massachusetts A   Pennsylvania E 
California E   Michigan E   Rhode Island A 
Colorado A   Minnesota E   South Carolina A 
Connecticut A   Mississippi E   South Dakota E 
Delaware A   Missouri A   Tennessee E 
Florida A   Montana E   Texas E 
Georgia E   Nebraska A   Utah A 
Hawaii A   Nevada E   Vermont A 
Idaho E   New Hampshire A   Virginia A 
Illinois E   New Jersey A   Washington E 
Indiana A   New Mexico E   West Virginia E 
Iowa A   New York A   Wisconsin E 
Kansas A   North Carolina E   Wyoming A 
Kentucky E   North Dakota E   
E: Elected    A: appointed 
Source: Bohn and Inman (1996) 
  
Among the socio-economic variables, we chose to use the yearly deviation from the 
national mean of per capita personal income140 as a measure of the state?s general 
economic condition141 (and, to some extent, the needs of its population) rather than gross 
state product.142 The effects of state?s population density are unclear: a state that has to 
                                                 
140 Calculated using the per capita personal income data from the Bureau of Economic Analysis. An in-
detail description of the methodology used for the construction of this measure can be found in their 
methodology section. Alternatively, we also calculated different moving averages of per capita personal 
income, but settled for using the standard deviation. 
141 The Advisory Commission of Intergovernmental Relations (ACIR, 1987) pointed out that personal 
income is a powerful determinant of state fiscal behavior, surpassing in importance the set of budgetary 
constraints.  
142 We also performed the analysis using yearly deviations from the national GSP mean instead of per 
capita personal income for the years in which the data is available, with similar results, which is consistent 
with the BEA?s assertion that ?GSP and state personal income share the following elements of personal 
income: Compensation, proprietors? income, and rental income of persons. The measure of compensation 
used in GSP is based on wage and salary accruals while personal income is based on wage and salary 
disbursements The difference between accruals and disbursements is typically very small? BEA, Local 
Area Personal Income and Employment Methodology - Introduction, 2004.    
 142
cover the public expenditure demands of a larger population143 may find BSFs more 
attractive,144 an effect reinforced by the public good component of savings in the RDF. 
However, larger states have been found to have less volatile business cycles, so they may 
find strict BSFs less appealing. In sum, the effects of population on budgetary outcomes 
are unclear, as is its impact on BSFs configuration.  
Beyond their income and population, we expect states engaged in volatile spending to 
be in greater need for savings, a fact that may be reflected in the type of BSF they adopt. 
To investigate the effects of the sector composition of the state, we introduce the 
proportion of total earnings in construction, farming, manufacturing, mining and services, 
using the data on earnings provided by the Bureau of Economic Analysis (BEA).145 In 
graph B.3.1 we can observe the evolution over the period of the average percentages of 
each of the categories. Both services and manufacturing have grown in absolute terms 
but, as we can see from the percentage distribution, the increase in services has occurred 
mostly at the expense of a decline in the relative importance of manufacturing in the 
average state economy. 
                                                 
143 Sawicky (2003) has proposed a measure for adjusted population that reflects the actual pressure on state 
finances better than raw numbers on population (adjusting for population in poverty, unemployment rates 
and the state?s fiscal capacity). The data to construct indices of this nature is however not available for our 
sample period. 
144 Oates (1988) finds evidence in local governments for what he calls the ?zoo effect?, that is, larger 
localities may be able to provide a wider range of services. 
145 I would like to thank Dr. Jeff Werling, from the Interindustry Forecasting center (Inforum) at the 
University of Maryland, for his assistance in acquiring the BEA data. 
 143
Graph B.3.1- Average percentage earning distribution
0
0.1
0.2
0.3
0.4
0.5
0.6
19
51
19
56
19
61
19
66
19
71
19
76
19
81
19
86
19
91
19
96
construction
farming
manufacturing
mining
services
 
 
The different types of state revenues and expenses are also included in the regressions 
as percentages rather than actual levels, grouped according to their level of volatility, 
which avoids the scale and trend issues that would occur if we included raw dollar 
amounts. Also, this modification addresses the issue of high multicollinearity between 
our dependent variables.146 Furthermore, including these percentages allows us to 
investigate the effect of spending and taxing compositions on the decision of BSF 
configuration.  
In particular, total spending is not very helpful for our purposes, since not all types of 
government expenditures behave in the same way during recessions. Some types of 
expenditure are less under the control of the policymakers, or are more visible to the 
public (so they may trigger stronger responses if cuts are needed during recessions), 
providing extra incentives to establish solid BSFs. Since we are interested in the impact 
of expenditure volatility and to avoid issues of high collinearity between the different 
                                                 
146 Since the Variance Inflation Factor suggested high levels of collinearity, principal component 
techniques were used to reduce the dimensionality of the dataset. However, the decision for maintaining a 
certain number of eigen-vectors was often not clear-cut, and since a conservative use of the Kaiser and 
Scree criteria led to relatively small reductions in dimensionality we opted for transformations of the 
independent variables to avoid obscuring their relationship with the dependent variable. 
 144
types of expenditure, we go beyond the functional nature of the different types of 
expenditure and group them instead with respect to their degree of volatility. We consider 
each spending type?s mean standard deviation, following an approach similar in spirit to 
the calculations of Lane (2003)147 and classify the six types of expenditure in three 
categories according to their volatility148 (high, medium and low) including in our 
regression the state?s percentages in the most and least volatile categories and leaving the 
middle group as baseline.149 
 
Graph B.3.2 - Average percentage 
composition of spending
0
0.1
0.2
0.3
0.4
0.5
19
51
19
54
19
57
19
60
19
63
19
66
19
69
19
72
19
75
19
78
19
81
19
84
19
87
19
90
19
93
19
96
19
99
education health highways
natural resources welfare unemployment
 
 
                                                 
147 Lane (2003) calculates output volatility measured as the standard deviation of output growth.  
148 After applying the GDP deflator and calculating the overall average standard deviations, we can see that 
the magnitudes of the standard deviations are similar within groups and considerably different across 
groups, so the choice of three groups with two components seems reasonable. Education and welfare 
spending are the most volatile group, while expenditure in highways and health and hospitals fall in the 
middle category, and unemployment compensation and spending in natural resources are the relatively least 
volatile expenses. Although it may seem counterintuitive that education belongs in the most volatile group, 
we must note that capital spending in education is included in this category, which explains its variability. 
149 Alternatively, we followed the method Holcombe and Sobel (1997) developed to generate long-run and 
short-run variability of state income and calculate each state?s yearly deviation from the national mean. The 
basic equation is given by: ln ?(Expenditure type) = ? + ? ln ?(per capita personal income) + ?. Taking the 
growth rate does not make the series stationary (and ? is a measure of the long-run variability), but the 
series become stationary after detrending it with the augmented Hodrick-Prescott filter making ? an 
indicator of the short-run variability over the sample period. 
 145
Although the overall level of tax collection would provide some indication on the 
state?s ability to raise revenue through the cycle (which is likely to be inversely correlated 
with the necessity of maintaining savings and hence a potential determinant of BSF 
structure) the intended use of this variable here, however, is not to provide us with a 
measure of the wealth or fiscal capacity of the state. Firstly, tax collections would provide 
a flow ?rather than stock- indicator and would be in that sense inadequate for our 
purposes. In addition, revenue collected through taxes yearly is likely to be highly 
correlated with other explanatory variables in our analysis (such as personal income and 
tax effort). As with spending, we include tax collection by grouping the different types of 
taxes (individual and corporate income tax, property, sales and severance taxes) 
according to their levels of volatility150. Graph B.3.3 depicts the evolution of each tax?s 
share over the period.  
Tax collections has been a decreasingly important source of income for states,151 but 
still is the most important: According to data from the US Census of Governments, tax 
revenue and intergovernmental (IG) revenue together accounted for about 70% of all 
state resources in 2001. IG revenues include local and federal transfers (with the latter 
making about 95% of the total152) and are mostly outside of state control. They have 
grown in importance during our period of observation, although not steadily: In 
particular, and more importantly for our purposes, IG revenues are likely to decrease 
during periods of crises, when states need resources the most. IG finances are included in 
                                                 
150 The percentages of tax income that come from severance and property taxes are grouped in the ?most 
volatile? category, while the percentages received from sale and individual income taxes form the ?least 
volatile? category. The percentage of tax revenues derived from corporate income taxes corresponds to the 
?middle volatility? group, which is used as baseline.  
151 Dropping from about 70% in 1950 to around 60% in 2000. 
152 As opposed to IG expenses, where local IG spending makes for most of the total expenditure. 
 146
the analysis by calculating the deviation from the national mean of the per capita net IG 
transfers (revenues minus expenses). 
All data on state tax collection, expenditure and IG finances come from the Census of 
Governments and the Historical Statistics of the United States.153 
 
Graph B.3.3 - Avg percentage composition of tax income
0
0.1
0.2
0.3
0.4
0.5
0.6
19
51
19
55
19
59
19
63
19
67
19
71
19
75
19
79
19
83
19
87
19
91
19
95
19
99
property sales individual income
corporate income severance  
 
 
Although funds in BSFs are subject to rules that do not apply to other forms of 
savings, it is likely that states that maintain easily liquefiable resources (such as cash, 
short-term deposits and securities154) in larger amounts will see the need to establish a 
stringent fund as less pressing. On the other hand, it may be possible that states that 
decide to have more savings in the form of cash and securities have a preference for 
sound savings, and would be more inclined to establish strict funds. Since it is difficult to 
intuitively or theoretically establish a predicted sign for the relationship between other 
                                                 
153 The data series for Alaska and Hawaii start in 1957 and 1955, respectively, due to their late 
incorporation to the Union. 
154 Examples of items that are (or are not) included in this item can be found in chapter 10 (Cash and 
Security Holdings) of the Census? Federal, State, and Local Governments Government Finance and 
Employment Classification Manual 
 147
savings and the nature of BSFs, it remains a question best answered empirically. The 
Census only provides data for cash and securities (our measure of other savings) since 
1951, imposing the lower time bound for the analysis. Other potential measures of state 
savings exist in the literature. For example, savings haven been measured using data on 
general fund balances from the National Association of State Budget Officers (NASBO), 
but consistent data are not available until rather recently (Hou (2004)). Wagner (2003), 
among others, calculates savings as current revenues minus current expenditures using 
Census Data, an approach we have also explored155 ? keeping in mind that the Census 
discourages this use.156 However, not-spent revenue may not be available for budget 
stabilization and, as Gold (1995) indicates, fungible resources available to the state are 
perhaps the best indicator of the means a state can count on to face a recession, 
suggesting that cash holdings may be a better measure for our purposes.  
Aside from using reserves, states can increase the resources they derive from taxation. 
However, states that exert higher levels of tax effort will have less room for tax increases, 
making meaningful BSFs more attractive. We include information on average tax effort 
(measured as the ratio of actual tax revenue to the tax revenue that would be collected 
under a hypothetical, uniform tax system) using data provided by the Advisory 
Commission of Intergovernmental Relations (ACIR).157  ACIR?s staff compiled data on 
                                                 
155 In particular, I calculated the deviation from the national average per capita savings. Using this 
indicator, we find some weak evidence that states with more savings are less likely to establish strict BSFs. 
156 ?Although the original sources of data for these finance statistics are the accounting records of 
governments, the data derived from them are purely statistical in nature and cannot be used as financial 
statements or to measure a government's fiscal condition. For instance, the difference between a 
government's total revenue and expenditure cannot be construed to be a "surplus" or "deficit."? Census? 
Government Finance and Employment Classification Manual, Chapter 6 (Overview of Government 
Finance Statistics.)  
157 ACIR only reports data for years 1967, 1975, 1977, 1979 and 1980 ? 1988. Robert Tannenwald from the 
Federal Reserve of Boston has calculated tax effort for additional years using methodology based on 
ACIR?s. Since inclusion of his data did not significantly affect the results and to avoid introducing 
 148
tax burden for the period 1953-64, but unfortunately such information was not published 
and, to my knowledge, it is not available. The only information available for those years 
is the overall average and its corresponding standard deviation, presented in table iii. The 
index was calculated for 1967, 1975, 1977, 1979 and yearly from 1980 to 1988 (except 
for year 1987). Based on this information, ACIR staff indicated a movement towards 
equalization of tax burdens across states in the period 1953-64, and a reversal of the 
tendency since 1964. Tannenwald (1997, 1999, 2002) has calculated tax effort for 
additional years using methodology based on ACIR?s158. Since inclusion of his data did 
not significantly affect the results and to avoid introducing differences in the tax effort 
series due to changes in methodology we restrict ourselves to the data provided by ACIR, 
which is the measure commonly used in the BSF literature. Additionally, the Tax 
Foundation calculates measures of state and local tax effort since 1970, however, it is 
impossible to separate the state component from their numbers so their data is not used in 
this application. 
 
Table B.3.3: Tax burden, 1953-75 
 Tax burden Growth 
 1953 1964 1975 1953-64 1964-75 
standard deviation 1.45 1.28 1.46 1.06 1.02 
coefficient of variation 0.179 0.128 0.132 0.487 0.987 
Source: ACIR (1975) 
 
                                                                                                                                                 
differences in the tax effort series due to changes in methodology we restrict ourselves to the data provided 
by ACIR, which is the measure commonly used in the BSF literature. Additionally, the Tax Foundation 
calculates measures of state and local tax effort since 1970, however, it is impossible to separate the state 
component from their numbers so their data is not used in this application. 
158 I would like to thank Dr Robert Tannenwald for his kind assistance. 
 149
An alternative measure exists in the form of the Total Taxable Resources (TTR) 
index159. However, to my knowledge, there is no data in TTR or its components dating 
far back enough in time to be applied in this study. 
On the other side of the spectrum, debt could potentially help weather recessions by 
smoothing out state consumption. Nevertheless, many states face restrictions in their 
capability to issue debt or require voter approval to issue guaranteed debt. Furthermore, 
debt is costly for states because it increases the future financial burden and may trigger 
increases in the future costs of borrowing if rating agencies are not satisfied with the 
state?s amount of savings. There are several potential ways to include the effect of debt 
on the choice of BSF configuration, such as the amount of interest paid on debt from the 
Census data (to approximate the effective weight of accumulated debt in the budget), the 
ratio of total debt to personal income or a set of indicators for the limitations on the 
emission of debt. We settle for using annual deviations from the national mean of total 
debt per capita160 because they are more likely to reflect the real situation of the state in 
terms of debt than the institutional constraints, which can be avoided in a variety of 
ways.161  Our a priori expectation is that states with higher levels of debt will be, ceteris 
paribus, more inclined to establish demanding BSFs, since they it would be relatively 
more costly for them to go further into debt. However, high levels of per capita debt may 
                                                 
159 The measure was designed in the U.S. State Treasury Department  (Sawicky, 1985) as gross state 
product plus the income the residents earn out of the state, federal transfers, and accrued capital gains 
minus federal taxes paid and depreciation. The TTR index, as explained by the Treasury Department, can 
be then understood as the ?unduplicated sum of the income flows produced within a state and the income 
flows received by its residents which a state can potentially tax.?  (Treasury Methodology for Estimating 
Total Taxable Resources, page 2)  
160 We also tried the deviation from the national mean of the state?s ratio of debt to personal income and the 
results and find some scattered indications that states with higher deviations may be more likely to establish 
stricter deposit (although not withdrawal) rules. 
161 A wide literature exists regarding the potential effects of debt restrictions on debt emissions. An 
interesting suggestion from this literature is that debt limits may have had one of its more important effects 
on the way states emit debt rather than on how much total debt is actually issued. 
 150
be correlated with higher tolerance for debt in the state, which could overcome the 
aforementioned effect. The final effect of indebtness on BSF rules is then left to 
empirical investigation. 
Among the institutional constraints, tax and expenditure limitations (TELs) restrict the 
state?s ability to cope with recessions through direct action,162 which may make 
meaningful BSFs more attractive. On the other hand, RDFs may be seen as a way to put 
funds outside of the scope of the TELs, allowing for higher discretion in the spending 
decisions, a proposition for which Wagner and Sobel (2006) find supporting evidence. 
We have explored different alternative measures of TELs, using dummies for the 
existence of each of these limitations as well as Poulson?s (2005) indexes of TEL 
strictness.163 Table B.3.4 presents a list and overview of these indexes.  
 
 
Table B.3.4. Strictness of TELs 
 
Method  
of  
approval 
Extent 
of the 
limit 
Size 
Treatment  
of  
surpluses 
Provisions 
for  
increases 
and waivers 
TOTAL 
SCORE 
mean 2.04 1.96 1.6 1.14 1.56 8.3 
std. dev. 1.94 2.14 1.9 1.69 1.72 7.77 
Source: Poulson (2005) 
 
 
                                                 
162 Different views exist on the issue: Elder (1992) finds that TELs are associated with a significant decline 
in state tax revenues, while Poterba (1996) concludes that TEL-states deal more quickly with deficits by 
raising taxes.  
163 Poulson (2005) creates indexes that consider the overall strictness of the TEL, whether voter approval is 
required for certain actions, what part of the budget is covered by the limits, the method of approval of the 
limitation and the treatment of surpluses. Of these, only the indicator regarding what part of the budget is 
covered by the TEL seems to have significant effects on the configuration of the BSF. In particular, there 
seems to be some evidence that the more demanding this feature of the TEL, the more likely states are to 
establish weak deposit requirements, although no effect is found on the determination of withdrawal rules. 
 151
Another important institutional constraint is given by the existence of balanced budget 
requirements (BBR). States with demanding BBRs enact more restrictive spending 
policies (Poterba (1994)), fare better in deficit control (Alesina and Bayoumi (1996)), are 
more likely to enact tax increases and spending cuts during recessions (Alt and Lowry 
(1994)) and tend to save more (Bohn and Inman (1996)), but strict BBRs also introduce 
rigidities in fiscal policy (Alt and Lowry (2001)) and may exacerbate business cycle 
volatility (Levinson (1998)). Demanding BBRs make meaningful BSFs more appealing, 
since intertemporal smoothing becomes more difficult. All states (except Vermont) have 
a BBR, although the rules differ in their degree of stringency. We consider the same 
measure the literature has used: the index constructed by ACIR164 (1987) and the 
dummies NASBO derived from it; and, additionally, we follow Poterba (1994) in 
constructing a binary indicator to classify BBRs as ?lax? or ?strict.?165  
In most cases the balanced budget provisions were approved before 1970, often when 
the state entered the Union, and have been in place since. The only exception is 
Tennessee, where a constitutional amendment was proposed on November 30th of 1977 
to establish a balanced budget restriction. Tennessee?s citizens approved it on March 7th 
1978 and it was finally proclaimed on March 31st, 1978.166 Before the 1978 amendment, 
                                                 
164 The index ranges from zero to ten, where higher numbers indicate stricter BBRs. It consideres the 
following order of (increasing) stringency: (1) rule requires governor to submit a balanced budget, (2) the 
legislature must pass a balanced budget, (3) carry-over of deficit is allowed to the next year but it must 
resolved within the following year, (4) deficit carry-over to the next biennium is not allowed, and (5) deficit 
carry-over to the next year is not allowed. Additionally, constitutional rules receive additional points. 
165 The advantage of using this binary indicator as opposed to the measure provided by ACIR is that it 
avoids giving the same relevance to unitary changes at all points in the scale while maintaining the 
difference between lax and strict rules for budget balancing. To investigate possible spurious effects 
brought about by the ?border? cases, we run the regressions two more times, including all border cases 
(states that scored 5 or 6 in the ACIR index) first in the lower category and then in the upper category.  
166 Before the amendment, Article II, Section 24 of the Constitution of Tennessee read: "No money shall be 
drawn from the treasury but in consequence of appropriations made by law; and an accurate statement of 
the receipts and expenditures of the public money shall be attached to and published with the laws at the 
rise of each stated session of the general assembly." 
 152
there was no balanced budget requirement, although according to state officials 
Tennessee had been enacting balanced budgets for decades before the requirement was 
adopted.167 The existing literature on budget has ignored the fact that Tennessee?s 
regulations regarding balanced budgets -unlike the other states- changed in the recent past 
and used the measure reported by ACIR in 1987. This becomes especially important for 
studies like this one, which concern themselves with the set of circumstances that led to 
the adoption of BSFs, because Tennessee adopted its fund in 1972 ?which means that 
using the ACIR index of 10 to measure overstates the strictness of the balanced budget 
rule in Tennessee in the years prior to the adoption of the fund. 
The last institutional factor is embedded in the BSFs themselves. Their legal nature 
(statutory or constitutional) can also play a role on the configuration of deposit and 
withdrawal requirements. Constitutionally established BBRs and TELs are regarded as 
stricter budgetary tools, because they allow decision makers less freedom when 
establishing the particulars of the law and have been shown to have stronger effects on 
fiscal policy than their statutory counterparts. To investigate whether constitutional BSFs 
are more likely to be endowed with stricter requirements, we include a dummy that takes 
the value of one if the BSF is embedded in the state?s Constitution. 
 
 
 
 
 
                                                 
167 I would like to thank Mr Bill Bradley, from the Budget Office of the State of Tennessee, for his 
assistance with the balanced budget requirements of Tennessee.  
 153
Bibliography 
 
Abney, G. and Lauth, T. ?The Line-Item Veto in the States: An Instrument for Fiscal 
Restraint or an Instrument for Partisanship?? Public Administration Review, Vol. 
45, N. 3, May June, 1985 
Abney, G. and Lauth, T. ?The Item Veto and Fiscal Responsibility? The Journal of 
Politics, Vol. 59, N. 3, August 1997. 
Advisory Commission on Intergovernmental Relations, State Constitutional and Statutory 
Restrictions on Local Government Debt, ACIR, September 1961. 
Advisory Commission on Intergovernmental Relations, State Constitutional and Statutory 
Restrictions Upon the Structural, Functional, and Personnel Powers of Local 
Governments, ACIR, October 1962. 
Advisory Commission on Intergovernmental Relations ?Measuring the Fiscal Capacity 
and Effort of State and Local Areas? Washington, D C. ACIR, 1972. 
Advisory Commission on Intergovernmental Relations ?Measuring the Fiscal ?Blood 
Pressure? of the States ? 1964-1975? Washington, D C. ACIR, 1977. 
Advisory Commission on Intergovernmental Relations, ?Intergovernmental Service 
Arrangements for Delivering Local Public Services: Update 1983?. ACIR, October 
1985. 
Advisory Commission on Intergovernmental Relations. ?Fiscal Discipline in the Federal 
System: National Reform and the Experience of the States.? Washington, D C. ACIR, 
1987. 
 154
Advisory Commission on Intergovernmental Relations ?State Fiscal Capacity and 
Effort? Washington, D C. ACIR, 1990. 
Alesina, A. and Bayoumi, T. ?The Costs and Benefits of Fiscal Rules: Evidence from 
U.S. States.? NBER Working Paper 5614, 1996. 
Alt, J. and Lowry, R. ?Divided Government, Fiscal Institutions, and Budget Deficits: 
Evidence from the States? American Political Science Review, Vol. 88, N. 4, 
December 1994. 
Alt, J. and Lowry, R. ?A Dynamic Model of State Budget Outcomes under Divided 
Partisan Government? The Journal of Politics, Vol. 62, No. 4, 2000. 
Amador, M. ?Savings under Political Compromise? MIT Mimeo, 2003. 
Ambler, D., Burr, J., McManus, K., Mischel, H. and Roswick, D. ?Revenue Bond Credit 
Analysis? The Handbook of Municipal Bonds and Public Finance Ed. Lamb, R., 
Leigland, J. and Rappaport, S. New York Institute of Finance, 1993, pp 143-165. 
Amt, W. ?Tax Increment Financing Boosts Local Tax Base? Economic Development 
Digest, Vol. 11, N. 4, September 2000. 
Bahl, R. and Duncombe, W. ?State and Local Debt Burdens in the 1980s: A Study in 
Contrasts? Public Administration Review, Vol. 53, N. 1, 1993. 
Bails, D. and Tieslau, M. ?The Impact of Fiscal constitutions on State and Local 
Expenditures? The Cato Journal, Vol. 20, N. 2, Fall 2000. 
Basham, P. ?Assessing the Term Limits Experiment of California and Beyond? Cato 
Institute's Center for Representative Government, No. 143, 2001. 
 155
Bayoumi, T., Goldstein, M. and Woglom, G. ?Do Credit Markets Discipline Sovereign 
Borrowers? Evidence from the U.S. States? Journal of Money, Credit, and Banking, 
Vol. 27, N. 4, 1995. 
Beckman, A. "The Item Veto Power of the Executive" Temple Law Quarterly, Vol. 31, 
1957. 
Berry, W. Ringquist, E., Fording, and Hanson, R. ?The Measurement and Stability of 
State Citizen Ideology? Forthcoming in State Politics and Policy Quarterly, 2006. 
Besley, T. and Case, A. ?Political Institutions and Policy Choices: Empirical Evidence 
from the United States? Mimeo, 2001. 
Bohn, H. and Inman, R. ?Balanced Budget Rules and Public Deficits: Evidence from the 
U.S. States.? National Bureau of Economic Research Working Paper, No. 5533 
Cambridge, MA. 
Bourdeaux, C. ?A Question of Genesis: An Analysis of the Determinants of Public 
Authorities,? Journal of Public Administration Research & Theory, Vol. 15, N. 3, 
July 2005. 
Bradford, E. and Constantine, R. ?The Debt Ceiling and Executive Latitude? Harvard 
Law School, Federal Budget Policy Seminar, Briefing Paper 11, May 2005. 
Brecher, C. and Lynam, E. ?New York?s Endangered Future: Debt Beyond Our Means? 
Citizens? Budget Commission, September 2005. 
Brecher, C., Richwerger, K. and Van Wagner, M. ?An Approach to Measuring the 
Affordability of State Debt? Public Budgeting and Finance, 2003. 
Brennan, G. and Buchanan, J. The Reason of Rules: constitutional Political Economy 
Cambridge University Press, Cambridge, 1985. 
 156
Briffault, R. "The Item Veto in State Courts? Temple Law Review, Vol. 66, 1993. 
Briffault, R. Balancing Acts: the Reality Behind State Balanced Budget Requirements. 
The Twentieth Century Fund Press, New York, 1996. 
Briffault, R. ?The Disfavored constitution: State Fiscal Limits and State constitutional 
Law? Rutgers Law Journal, Vol. 34, N. 4, Summer 2003. 
Buchanan, J. and Tullock, G. The Calculus of Consent University of Michigan Press,  
Ann Arbor, 1962. 
Buchanan, J. Public Finance in Democratic Process: Fiscal Institutions and Individual 
Choice University of North Carolina Press, Chapel Hill, 1967. 
Bunch, B. ?The effect of constitutional Debt Limits on State Governments? Use of Public 
Authorities? Public Choice 68, January 1991. 
Calgary City ?The U.S. Experience with Tax Increment Financing (TIF): A Survey of 
Selected U.S. Cities? [online] [Accessed July 3, 2007]   
< http://www.calgary.ca/docgallery/bu/corporateproperties/final_report_tif.pdf >  
Clarke, W. and Eger, R. ?Special Districts, Authorities, Corporations, and the Bond 
Market? Forthcoming in the Municipal Finance Journal. 
Clingermayer, J. and Wood, D. ?Disentangling Patterns of State Debt Financing? 
American Political Science Review, Vol. 89, N. 1, March 1995. 
Coffman, M. ?A Rainy Day Fund for Colorado.? Treasur-E-Notes, Vol. 3, Issue 40, 
2002. 
Cohen, C. ?Analyzing Governmental Credit? The Handbook of Municipal Bonds and 
Public Finance Ed. Lamb, R., Leigland, J. and Rappaport, S. New York Institute of 
Finance, 1993, pp 127-142. 
 157
Congressional Budget Office ?Balancing the Federal Budget and Limiting Federal 
Spending: constitutional and Statutory Approaches? CBO, September 1982.  
Congressional Budget Office ?The Mortgage Subsidy Bond Tax Act of 1980: Experience 
Under the Permanent Rules? CBO Staff Working Paper, March 1982, Washington 
DC. 
Crain, W. and Muris, T. "Legislative Organization of Fiscal Policy? Journal of Law and 
Economics, Vol. 38, No. 2, 1995. 
De la Fuente, A. Mathematical Methods and Models for Economists Cambridge 
University Press, Cambridge, 2000. 
Denison, D., Hackbart, M. and Moody, M. ?Evolving Role of Debt Limit Policies? 
University of Kentucky, Mimeo, 2005. 
Denison, D., Hackbart, M. and Moody, M. ?State Debt Limits: How Many Are Enough?? 
Public Budgeting and Finance, Winter 2006. 
Dickson, S. ?Civil Wars, Railroads and Road Bonds: Bond Repudiations in the Days of 
Yore? The Handbook of Municipal Bonds and Public Finance Ed. Lamb, R., 
Leigland, J. and Rappaport, S. New York Institute of Finance, 1993, pp 166-173. 
Douglas, J. and Gaddie, R. ?State Rainy Day Funds and Fiscal Crises: Rainy Day Funds 
and the 1990-1991 Recession Revisited.? Public Budgeting and Finance, Spring 
2002. 
Drazen, A. Political Economy in Macroeconomics. Princeton University Press, Princeton, 
New Jersey, 2000. 
Drazen. A. Political Economy in Macroeconomics Princeton University Press, Princeton, 
New Jersey, 2000. 
 158
Duke Law Journal Editorial Board ?State Administrative Supervision of Local 
Government Debt: The North Carolina Model Notes? Duke Law Journal, Vol. 1972, 
No. 2, June 1972. 
Eger, R. ?Casting Light on Shadow Government: An Exploratory Analysis of Public 
Authorities in the Southern States.? Paper presented at the Association for Budgeting 
and Financial Management at Washington, DC, January 17-19, 2002. 
Elder, H. ?Exploring the Tax Revolt: An Analysis of the Effect of State Tax and 
Expenditure Limitation Laws,? Public Finance Quarterly, vol. 20, 1992. 
Endersby, J. and Towle, M. ?Effects of Constitutional and Political Controls on State 
Expenditures? Publius, Vol. 27, N. 1, Winter 1997. 
Farnham, P ?Re-examining Local Debt Limits: A Disaggregated Analysis? Southern 
Economics Journal, Vol. 51, N. 4, April 1985. 
Fat?s, A. and Mihov, I. ?The Macroeconomic Effects of Fiscal Rules in the US States? 
Journal of Public Economics, Vol. 1-2, 2006. 
Frant, H. ?Reconsidering the Determinants of Public Authority Use? Journal of Public 
Administration Research and Theory, Vol. 7, No. 4, October, 1997. 
Gillete, C. ?Direct Democracy and Debt? NYU Center for Law and Business Working 
Paper Series 03-010, 2003. 
Gilligan, T. and Matsusaka, J. ?Deviations from Constituent Interests: The Role of 
Legislative Structure and Political Parties in the States.? Economic Inquiry, Vol. 33, 
No. 3, 1995. 
 159
Gilligan, T. and Matsusaka, J. ?Fiscal Policy, Legislature Size, and Political Parties: 
Evidence from State and Local Governments in the First Half of the 20th Century? 
National Tax Journal, Vol. 54. No. 1, 2001. 
Gilligan, T. and Matsusaka, J. ?Structural Constraints on Partisan Bias under the 
Efficient Gerrymander.? Vol. 100, No. 1, 1999. 
Gold, S. The Fiscal Crises of the States: Lessons for the Future. Washington, DC: 
Georgetown University Press, 1995. 
Gonzalez, C. and Paqueo, V. ?Social Sector Expenditures and Rainy-Day Funds.? World 
Bank Policy Research Working Paper 3131, 2003. 
Goodrich, C. ?The Revulsion Against Internal Improvements? The Journal of Economic 
History, Vol. 10, No. 2, Nov. 1950. 
Government Finance Officers Association (Committee on Accounting, Auditing and 
Financial Reporting and the Committee on Governmental Budgeting and 
Management.) ?Appropriate level of unreserved fund balance in the general fund.? 
January 30th 2002. 
Government Finance Officers Association ?Recommended Budget Practices: Tax 
Increment Financing as a Fiscal Tool? GFOA, 2006. 
Government Finance Officers Association. ?Debt Management Policy? GFOA 1995 and 
2003. 
Government Finance Officers Association. ?Recommended Budget Practices: A 
Framework for Improved State and Local Government Budgeting? GFOA 1998. 
Government Finance Officers Association. ?Recommended Budget Practices: A 
Framework for Improved State and Local Government Budgeting? GFOA 1998. 
 160
Government Finance Officers Association. Recommended Budget Practices: A 
Framework for Improved State and Local Government Budgeting, GFOA 1998. 
Gricar, J. ?Municipal Corporations: Circumventing Municipal Debt Limitations?  
Michigan Law Review, Vol. 48, No. 7, May, 1950. 
Haas, P. ?Managing Single Function Governments: The promise and challenge of special 
districts? in Handbook of Local Government Administration, Public Administration 
and Public Policy, Vol. 62, Ed. Gargan, J., Marcel Dekker, New York, 1996. 
Hackbart, M. and Leigland, J. ?State Debt Management Policy: A National Survey? 
Public Budgeting and Finance, Spring 1990. 
Hagle, T. and Mitchell, G. ?Goodness of Fit Measures for Probit and Logit? American 
Journal of Political Science, Vol. 36, 1992. 
Hausman, J. and McFadden, D. ?Specification Tests for the Multinomial Logit Model? 
Econometrica, Vol. 52, No. 5, 1984. 
Heins, A. constitutional Restrictions Against State Debt, University of Wisconsin Press, 
Madison, 1963. 
Heller, W. ?Political Denials: The Policy Effect of Intercameral Partisan Differences in 
Bicameral Parliamentary Systems? Journal of Law, Economics, and Organization 
Vol. 17, 2001. 
Hildreth, W. ?State and Local Governments as Borrowers: Strategic Choices and the 
Capital Market? Public Administration Review, Vol. 53, No. 1, Jan-Feb. 1993. 
Hildreth, W. and Zorn, C. ?The Evolution of the State and Local Government Municipal 
Debt Market over the Past Quarter Century? Public Budgeting & Finance, Vol. 25, 
N. 4s, December 2005.  
 161
Holcombe, R. and Sobel, R. Growth and Variability in State Tax Revenue: An Anatomy 
of State Fiscal Crises. Greenwood Press, Westport, 1997. 
Holcombe, R. and Sobel, R. Growth and Variability in State Tax Revenue: An Anatomy 
of State Fiscal Crises Greenwood Press, Westport, 1997. 
Holtz-Eakin, D., Rosen H. and Tuller, S. ?Intertemporal Analysis of State and Local 
Government Spending: Theory and Tests? Journal of Urban Economics 35, 1994. 
Hou, Y. ?Fiscal Reserves and Budgetary Reactions to Revenue Shocks;? University of 
Georgia, Mimeo, 2004. 
Hou, Y. ?Testing the Effects of Sub-National Counter-Cyclical Fiscal Policies: Budget 
Stabilization Funds, General Fund Surpluses, and State Total Own-Source 
Expenditures? Syracuse University, Mimeo, 2001. 
Hou. Y. ?Fiscal Reserves and State Own-Source Expenditure in Downturn Years.? 
Public Finance Review, Vol. 33, No. 1, 2005. 
Jeweler, R. ?Municipal Reorganization: Chapter 9 of the U.S. Bankruptcy Code? CRS 
Report for Congress, March 2007. 
Johnson, C. and Kriz. K. ?Fiscal Institutions, Credit Ratings, and Borrowing Costs? 
Public Budgeting & Finance, Vol. 25, N. 1, Spring 2005. 
Joyce, P. ?What?s So Magical about Five Percent? A Nationwide Look at Factors that 
Influence the Optimal Size of State Rainy Day Funds.? Public Budgeting and Finance, 
Summer 2001. 
Kiewiet, R. "Constitutional Limitations on Indebtedness: The Case of California.'' In 
constitutional Reform in California Ed. Cain,B. and Noll, R. Berkeley, California: 
Institute of Governmental Studies, 1995. 
 162
Kiewiet, R. and Szalaky, K. ?Constitutional Limitations on Borrowing: An Analysis of 
State Bond Indebtness? The Journal of Law, Economics and Organization, Vol. 12, 
N. 1, 1996. 
Kirkland, K. ?Creative Accounting and Short-Term Debt: State Responses to the Deficit 
Threat? National Tax Journal, Vol.  36, No. 3, 1983. 
Kirkland, K. ?Creative Accounting and Short-Term Debt: State Responses to the Deficit 
Threat? National Tax Journal, Vol. 36, N. 3, September 1983. 
Knight, B. and Levinson, A. ?Fiscal Institutions in U.S. States? University of Wisconsin, 
Mimeo, 1998. 
Knight, B. and Levinson, A. ?Rainy Day Funds and State Government Savings.? 
National Tax Journal, Vol. 52, No. 3, 1999. 
Krause, G. ?Partisan and Ideological Sources of Fiscal Deficits in the United States? 
American Journal of Political Science, Vol. 44, No. 3, 2000. 
Kreps, M. ?Ups and Downs of Municipal Bonds? Volume and Yields in the Past Century? 
The Handbook of Municipal Bonds and Public Finance Ed. Lamb, R., Leigland, J. 
and Rappaport, S. New York Institute of Finance, 1993, pp 104-114. 
Krishnakumar, A. ?In Defense of the Debt Limit Statute? Harvard Journal on Legislation, 
Vol. 42, N. 1, Winter 2005. 
Kriz, K. ?The Optimal Level of Local Government Fund Balances: A Simulation 
Approach? State Tax Notes, Vol. 27, No. 10, 2003. 
Lane, P. ?The Cyclical Behaviour of Fiscal Policy: Evidence from the OECD? Journal of 
Public Economics, Vol. 87, No. 12, 2003. 
 163
Lav, I. and Berube, A. ?When It Rains It Pours: A look at the adequacy of State Rainy 
Day Funds and Budget Reserves.? Center on Budget and Policy Priorities, 1999. 
Lazere, E. ?Fixing D.C.?s Rainy Day Fund? D. C. Fiscal Policy Institute, 2003. 
Lee, R. "State Item-Veto Legal Issues in the 1990s," Public Budgeting and Finance, Vol. 
20, N. 2, Summer 2000. 
Lehmann, H. ?The Federal Municipal Bankruptcy Act? The Journal of Finance, Vol. 5, 
No. 3, September 1950. 
Leigland, J. ?Overview of Public Authorities and Special Districts? The Handbook of 
Municipal Bonds and Public Finance Ed. Lamb, R., Leigland, J. and Rappaport, S. 
New York Institute of Finance, 1993, pp. 375-396. 
Leigland, J. ?Public Authorities and Determinants of their Use by State and Local 
Governments? Journal of Public Administration Research and Theory, Vol. 4, N. 4, 
1994. 
Levinson, A. ?Balanced Budgets and Business Cycles: Evidence from the States.? 
National Tax Journal Vol. 51 no. 4, 1998. 
Lewis, W. ?Budgetary Balance: The Norm, Concept and Practice in Large US Cities? 
Public Administration Review, Vol. 54, N. 6, December 1994.  
Liu, L. and Wallis, J. ?Infrastructure Financing, Debt Restrictions and Subnational Debt 
Market. Lessons from the United States? Mimeo, April 2007. 
Long, J. and Freese, J. Regression Models for Categorical Dependent Variables Using 
Stata, Stata Press, College Station, 2006. 
Lowry, R. and Alt, J. "A Visible Hand? Bond Markets, Political Parties, Balanced 
Budget Laws, and State Government Debt.? Economics & Politics 13, 2001. 
 164
Lowry, R. and Alt, J. "A Visible Hand? Bond Markets, Political Parties, Balanced 
Budget Laws, and State Government Debt? Economics & Politics, Vol. 13, N. 1, 
March 2001. 
Lowry, R. and Alt, J. ?A Visible Hand? Bond Markets, Political Parties, Balanced 
Budget Laws, and State Government Debt? Economics and Politics, Vol. 13, N. 1, 
March 2001. 
Lowry, R. and Alt, J. ?Divided Governments, Fiscal Institutions, and Budget Deficits: 
Evidence from the States? American Political Science Review, Vol. 88, No. 4, Dec 
1994. 
Lowry, R., Alt, J. and Ferree, K. ?Fiscal Policy Outcomes and Electoral Accountability 
in American States? American Political Science Review, Vol. 92, No. 4, 1998. 
MacManus, S. ?Financing Federal, State, and Local Governments in the 1990s? Annals 
of the American Academy of Political Science, May 1990. 
Maco, P. ?Building a Strong Subnational Debt Market: A Regulator?s Perspective? 
Richmond Journal of Global Law & Business, Vol. 2, N. 1, Winter/Spring 2001. 
Maddala, G.  Limited-dependent and qualitative variables in econometrics Cambridge, 
MA: Cambridge University Press, 1983. 
Man, J. ?Fiscal Pressure, Tax Competition and the Adoption of Tax Increment 
Financing? Urban Studies, Vol. 26, N. 7, 1999. 
Mas-Colell, A., Winston, M. and Green, J. Microeconomic Theory. Oxford University 
Press, New York, 1995. 
McCarty, N. ?The Policy Consequences of Partisan Polarization in the United States? in 
the 2004 Seminar of Institutions and Positive Political Theory, Berkeley. 
 165
McEachern, W. ?Collective Decision Rules and Local Debt Choice: A Test of the Median 
Voter Hypothesis? National Tax Journal, Vol. 31, N. 2, June 1978. 
McGranahan, L. ?Unprepared for Boom or Bust: Understanding the Current State Fiscal 
Crisis? Economic Perspectives, Federal Bank of Chicago, 2002. 
McGrane, R. Foreign Bondholders and American State Debts. The Macmillan 
Company, New York, 1935. 
Mikowski, C. ?Organizational Dynamics of State Debt Issuance? The Handbook of 
Municipal Bonds and Public Finance Ed. Lamb, R., Leigland, J. and Rappaport, S. 
New York Institute of Finance, 1993, pp 360-375. 
Mitchell, W. The Effectiveness of Debt Limits on State and Local Government 
Borrowing New York Institute of Finance, September 1969. 
Mody, A. and Fabrizio, S. ?Can Budget Institutions Counteract Political Indiscipline?? 
IMF Working Papers, Washington DC, 2006. 
National Association of State Budget Officers. ?Budget Stability: A Policy Framework 
for the States.? NASBO, 1995. 
National Association of State Budget Officers. ?Budgeting Amid Fiscal Uncertainty. 
Ensuring Budget Stability by Focusing on the Long Term.? NASBO, 2004. 
National Conference of State Legislators ?Fundamentals of Sound State Budgeting 
Practices? NCSL, Washington DC, June 1995. 
National Conference of State Legislators ?Fundamentals of Sound State Budgeting 
Practices? NCSL, Washington DC, June 1995. 
Navin, J. and Navin, L. ?An Evaluation of Budget Stabilization Funds among Midwestern 
States? Growth and Change, Vol. 25, No. 4, 1994. 
 166
New, M. ?Limiting Government Through Direct Democracy. The case of state tax and 
expenditure limitations.? Policy Analysis, N. 420, Cato Institute, 2001.  
New. M. ?Where Term Limits lead to Tax Cuts?, Cato Institute, 2001. 
Nice, D. ?The Impact of State Policies to Limit Debt Financing? Publius, Vol. 21, No. 1, 
Winter 1991. 
Oates, W. ?An Essay on Fiscal Federalism? Journal of Economic Literature, 1999. 
Oates, W. ?On the Measurement of Congestion in the Provision of Local Public Goods? 
Journal of Urban Economics, Vol. 24, 1988. 
Oates, W. ?Towards a Second Generation Theory of Fiscal Federalism? International 
Tax and Public Finance, Vol. 12, 2005. 
Pagano, M. and Perry, D. ?Financing the 21st Century City: How Did I Get Stuck 
Holding the Bag? Great Cities Institute, May 2006.  
Peters, S., Smith Lovette, S. and Choe, K. T. ?Emerging Borrowing Priorities ? 
Financing Infrastructure? The Handbook of Municipal Bonds and Public Finance 
Ed. Lamb, R., Leigland, J. and Rappaport, S. New York Institute of Finance, 1993, 
pp. 788-806. 
Peterson, B. and Harrell, F. ?Partial Proportional Odds Models for Ordinal Response 
Variables? Applied Statistics, 1990. 
Pogue, T. ?The Effects of Debt Limits: Some New Evidence? National Tax Journal, Vol. 
23, N. 1, March 1970. 
Pollock, R. and Suyderhoud, A. ?The Role of Rainy Day Funds in Achieving Fiscal 
Stability.? National Tax Journal 39(4), 1986. 
 167
Poterba, J. ?Balanced Budget Rules and Fiscal Policy: Evidence from the States? 
National Tax Journal, Vol. 48, 1995. 
Poterba, J. ?Budget Institutions and Fiscal Policy in the U.S. States.? American 
Economic Review, Vol. 86, No. 2, 1996. 
Poterba, J. ?Capital Budgets, Borrowing Rules and State Capital Spending? Journal of 
Public Economics 56, 1995. 
Poterba, J. ?State Responses to Fiscal Crises: The Effects of Budgetary Institutions and 
Politics.? Journal of Political Economy 102(4), 1994. 
Poterba, J. and Rueben, K. "State Fiscal Institutions and the U.S. Municipal Bond 
Market" Fiscal Institutions and Fiscal Performance, Ed. Poterba, J. and von Hagen, J. 
University of Chicago Press, Chicago, 1999, pp 181-209. 
Poulson, B. ?Creating a Budget Stabilization Fund for Colorado.? Information for 
Colorado?s Advisory Group Members, 2003. 
Poulson, B. ?Creating a Budget Stabilization Fund for Colorado.? Information for 
Colorado?s Advisory Group Members, 2003. 
Poulson, B. ?Grading the States? Tax and Expenditure Limits.? Americans for Prosperity 
Foundation, June 2005. 
Poulson, B. ?Grading the States? Tax and Expenditure Limits.? Americans for Prosperity 
Foundation, 2005. 
Primo, D. ?Stop Us Before We Spend Again: Institutional Constraints on Government 
Spending? Economics & Politics, Vol. 18, N. 3, November 2006. 
 168
Pryde, J. ?The Ongoing Battle: Almost 70 Years of Assaults on Tax-Exempt Municipals? 
The Handbook of Municipal Bonds and Public Finance Ed. Lamb, R., Leigland, J. 
and Rappaport, S. New York Institute of Finance, 1993, pp. 836-850. 
Raher, S. and Donner, C. ?Debt by Any Other Name is Still Debt? Independence 
Institute, Colorado, December 2003.  
Ratchford, B. ?A Formula for Limiting State and Local Debts? Quarterly Journal of 
Economics, Vol. 51, N. 1, November 1936. 
Ratchford, B. ?New Forms of State Debts? Southern Economic Journal, Vol. 8, N. 4, 
April 1942. 
Ratchford, B. ?Public Debts in the South? Southern Economic Journal, Vol. 2, N. 3, 
January 1936. 
Ratchford, B. American State Debts, Duke University Press, Durham, NC, 1941. 
Regens, J., and Lauth, T. ?Buy Now, Pay Later: Trends in State Indebtness, 1950-1989? 
Public Administration Review, Vol. 52, No. 2, March-April, 1992. 
Rodriguez-Tejedo, I. ?Requirements and Limits on State Budgets? University of 
Maryland, Mimeo, 2006. 
Sawicky, M. ?Altered States: How the Federal Government can Ease the States? Fiscal 
Crisis? Economic Policy Institute Issue Brief, No. 187, 2003. 
Saye, A. ?American Government and Politics: Georgia?s Proposed New constitution? 
The American Political Science Review, Vol. 39, No. 3, June 1945. 
Sbragia, A. Debt Wish: Entrepreneurial Cities, U.S. Federalism, and Economic 
Development University of Pittsburgh Press, Pittsburgh PAU, 1996. 
 169
Secrist, H. ?constitutional Restrictions on Municipal Debt? The Journal of Political 
Economy, Vol. 22, N. 4, April 1914. 
Simonsen, B., Robbins, M. and Kittredge, B. ?Do Debt Policies Make a Difference in 
Finance Officers? Perceptions of the Importance of Debt Management Factors?? 
Public Budgeting and Finance, Spring 2001. 
Small, K. and Hsiao, C. ?Multinomial Logit Especification Tests? International 
Economic Review, Vol. 26, No. 3, 1985. 
Snell, R. ?State Balanced Budget Requirements: Provisions and Practice.? National 
Conference of State Legislatures. Article posted in April 1996, updated in March 
2004. 
Snell, R. ?State Balanced Budget Requirements: Provisions and Practice.? National 
Conference of State Legislatures. Article posted in April 1996, updated in March 
2004. 
Sobel, R. ?The Political Costs of Tax Increases and Expenditure Reductions: Evidence 
from State Legislative Turnover? Public Choice, Vol. 96, 1998. 
Sobel, R. and Holcombe, R. ?The Impact of State Rainy Day Funds in Easing State 
Fiscal Crises During the 1990-1991 recession.? Public Budgeting and Finance, Fall 
1996.  
Sobel, R. and Holcombe, R. ?The Impact of State Rainy Day Funds in Easing State 
Fiscal Crises During the 1990-1991 Recession.? Public Budgeting & Finance, Fall 
1996. 
Sterk, S. and Goldman, E. ?Controlling Legislative Shortsightedness: The Effectiveness 
of constitutional Debt Limitations? Wisconsin Law Review, 1991.  
 170
Sutch, R. and Carter, S. (Editors) Historical Statistics of the United States: Millennial 
Edition, 2007 Cambridge University Press. 
Tannenwald, R. ?Fiscal Disparity Among the States Revisited? State Tax Notes, Vol. 17, 
No. 15, 1999. 
Tannenwald, R. ?Methodology for Estimating Total Taxable Resources? U.S. 
Department of Treasury, 2002. 
Tannenwald, R. and Cowan, J. ?Fiscal Capacity, Fiscal Need, and Fiscal Comfort among 
U.S. States? The Journal of Federalism, Vol. 27, No. 3, 1997. 
Temple. J. ?The Debt/Tax Choice in the Finance of State and Local Capital 
Expenditures? Journal of Regional Science, Vol. 34, N. 4, 1994. 
Trautman, R. ?The Impact of State Debt Management on Debt Activity? Public Budgeting 
& Finance, Vol.15, N.2, Summer 1995. 
Tullock, G. ?Problems of Majority Voting.? Journal of Political Economy, Vol.67, No. 6, 
1959. 
Vanberg, V. and Buchanan, J. ?Organization Theory and Fiscal Economics: Society, 
State, and Public Debt? Journal of Law, Economics & Organization, Vol.2, N. 2, 
Autumn, 1986. 
Velasco, A. "Debts And Deficits With Fragmented Fiscal Policymaking," Journal of 
Public Economics, Vol. 76, 2000. 
Von Hagen, J. ?A note on the Empirical Evidence of Formal Fiscal Restrains? Journal of 
Public Economics 4(2), 1991. 
 171
Wagner, G. ?Are State Budget Stabilization Funds Only the Illusion of Savings? 
Evidence from Stationary Panel Data.? Quarterly Review of Economics and Finance, 
Vol. 43, No. 2, 2003. 
Wagner, G. ?The Bond Market and Fiscal Institutions: Have Budget Stabilization Funds 
Reduced State Borrowing Costs?? National Tax Journal, Vol. 57, No. 4, 2004. 
Wagner, G. and Elder, E. ?The Role of Budget Stabilization Funds in Smoothing 
Government Expenditures over the Business Cycle.?  Public Finance Review 33(4), 
2005. 
Wagner, G. and Sobel, R. ?State Budget Stabilization Fund Adoption:  
Preparing for the Next Recession or Circumventing Fiscal Constraints?" Public 
Choice, Vol. 126, No. 1, 2006. 
 Wallis, J. ?American Government Finance in the Long Run: 1790 to 1990? The Journal 
of Economic Perspectives, Vol. 14, N. 1, Winter, 2000. 
Wallis, J. ?Constitutions, Corporations and Corruption: American States and 
Constitutional Change: 1842-52.? The Journal of Economic History, Vol. 65, No. 1, 
2005. 
Wallis, J. and Weingast, B. ?Dysfunctional or Optimal Institutions?: State Debt 
Limitations, the Structure of State and Local Governments, and the Finance of 
American Infrastructure? Mimeo, 2006. 
Wallis, J. NBER/University of Maryland State constitution Project, [online]  
< www.stateconstitutions.umd.edu > [Accessed July 3, 2007] 
Wallis, J., Sylla, R. and Grinath, A. ?Sovereign Debt and Repudiation: The Emerging-
Market Debt Crisis in the U.S. States? NBER Working Paper No. W10753. 
 172
Weingast, B., Shepsle, K. and Johnsen, C. ?The political Economy of Benefits and Costs: 
A Neoclassical Approach to Distributive Politics? Journal of Political Economy, Vol. 
3, No. 4, 1981. 
Wells, R. "The Item Veto and State Budget Reform" The American Political Science 
Review, Vol. 18, N. 4, November 1924. 
Williams, R. ?Gologit2: Generalized Ordered Logit/ Partial Proportional Odds Models 
for Ordinal Dependent Variables.? Stata Journal, Vol. 6, No. 1, 2006. 
Windmeijer, F. ?Goodness of Fit Measures in Binary Choice Models? Econometrics 
Review, Vol.14, No. 1, 1995. 
Zahradnik, B. and Ribeiro, R. ?Heavy Weather: Are State Rainy Day Funds Working?? 
Center on Policy and Budget Priorities, May 2003. 
Zahradnik, B. ?Rainy Day Funds: Opportunities for Reform.? Center on Budget and 
Policy Priorities, March 2005. 
Zimmerman, D. ?Tax?exempt Bonds, Professional Sports Stadiums, and Economic 
Policy? American Tax Policy Institute, March 2007.