ABSTRACT Title of dissertation: EXCHANGE RATE REGIMES AND RISK PREMIA UNDER ALTERNATIVE WAGE STRUCTURE Hyunsoo Joo, Doctor of Philosophy, 2012 Dissertation directed by: Professor Carlos Vegh Department of Economics This paper analyzes the relationship between risk premium and exchange rate regimes. I conclude that xed exchange regime is preferred to exible regime, and risk premium is lower under xed regime. I analyze this problem with the friction where there are two types of wages; a conventional wage available to the current period consumption and a deferred wage paid at the end of period. When deferred wage increases, the real exchange rate and capital used for the next period produc- tion is higher under the exible exchange regime. Since production in the current period can be de ned as a negative function of real exchange rate, higher increase of real exchange rate leads into lower production when a positive deferred wage shock occurs under exible regime. As a result, xed regime is preferred thanks to lower volatility in consumption. In addition, remaining wealth is further reduced. The reduce of remaining wealth, increase of real exchange rate, and a surge of capital lead into the increase of leverage ratio. Therefore, the risk premium under the ex- ible regime is higher. When I replace a deferred wage shock with technology shock and world interest rate shock, still risk premium under exible regime is higher than under xed regime. The addition of the asset holders with the assumption of exogenous segmented asset market does not change these results. The second chapter utilizes a uniques high-frequency database to measure how exchange rates in nine emerging markets react to macroeconomic news in the U.S. and domestic economies from 2000 to 2006. We nd that major U.S. macroeco- nomic news have a strong impact on the ruturns and volatilities of emerging market exchange rates, but many domestic news do not. Emerging market currencies have become more sensitive to U.S. news in recent years. We also nd that market sen- timent could sway the impact of news on these currencies sustematically, as good (bad) news seems to matter more when optimism (pessimism) prevails. Market un- certainty also interacts with macroeconomic news in a statistically signi cant way, but its role varies across currencies and news. EXCHANGE RATE REGIMES AND RISK PREMIA UNDER ALTERNATIVE WAGE STRUCTURE by Hyunsoo Joo Dissertation submitted to the Faculty of the Graduate School of the University of Maryland, College Park in partial ful llment of the requirements for the degree of Doctor of Philosophy 2012 Advisory Committee: Professor Carlos Vegh, Chair/Advisor Professor Anton Korinek Professor Pablo D?Erasmo Professor John Rust Professor Phillip Swagel c Copyright by Hyunsoo Joo 2012 Acknowledgments I owe my gratitude to all the people who have made this thesis possible and because of whom my graduate experience has been one that I will cherish forever. First and foremost I?d like to thank my main advisor, Professor Carlos Vegh for giving me an invaluable opportunity to work on challenging and extremely inter- esting projects related to a graduate textbook. He has always made himself available for help and advice and there has never been an occasion when I?ve knocked on his door and he hasn?t given me time. It has been a pleasure to work with and learn from such an extraordinary individual. I would also like to thank my advisor, Professor Anton Korinek. Without his extraordinary theoretical ideas, this thesis would have been a distant dream. Thanks are due to Professor Pable D?Erasmo, Professor John Rust and Professor Phillip Swagel for agreeing to serve on my thesis committee and for sparing their invaluable time reviewing the manuscript. My co-authors at the International Moneytary Fund and Ferderal Reserves have enriched my graduate life in many ways and deserve a special mention. Dr. Zhiwei Zhang supported and reviewed my statistical programs closely, and Dr. Fang Cai provided real time exchange rate time series data. In addition, Seung-Jae Lee at the TYIB was kind to update major datasets whenever it is required. I owe my deepest thanks to my family - my wife Hyunah Lee and two kids Seo- Young and Bumjin who sacri ced their own lives for the dissertation, have always stood by me, and have pulled me through against impossible odds at times. Most ii of all, Hyunah had to stand up and played a role of anchor in my family with no foreseeable future. Without her support, this work would never have nished. Words cannot express the gratitude I owe them. It is impossible to remember all, and I apologize to those I?ve inadvertently left out. Lastly, thank you all and thank God! iii Table of Contents List of Tables vi List of Figures vii 1 Exchange Rate Regimes and Risk Premia under Alternative Wage Structure 1 1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 1.2.1 Basic Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 1.2.2 Timing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 1.2.3 Households . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 1.2.4 Capitalists . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 1.2.5 Final Goods Producer . . . . . . . . . . . . . . . . . . . . . . 18 1.2.6 Equilibrium Condition . . . . . . . . . . . . . . . . . . . . . . 18 1.2.7 Steady State . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 1.2.8 Fixed Exchange Regime . . . . . . . . . . . . . . . . . . . . . 21 1.2.9 Flexible Exchange Regime . . . . . . . . . . . . . . . . . . . . 27 1.2.10 Comparison between Exchange Regimes . . . . . . . . . . . . 29 1.3 Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 1.4 Extension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 1.4.1 Trader . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 1.4.2 Government . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 1.4.3 Equilibrium Condition . . . . . . . . . . . . . . . . . . . . . . 39 1.5 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 1.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 2 The Impact of Macroeconomic Announcements on Real Time Foreign Ex- change Rate in Emerging Markets 50 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 2.2 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 2.2.1 Exchange Rate Data . . . . . . . . . . . . . . . . . . . . . . . 54 2.2.2 Actual and Predicted Economic Variables . . . . . . . . . . . 56 2.2.3 Foreign Exchange Forecasts . . . . . . . . . . . . . . . . . . . 57 2.3 Announcements and FX responses . . . . . . . . . . . . . . . . . . . . 58 2.3.1 Contemporaneous E ect from OLS Regression . . . . . . . . . 58 2.3.2 Contemporaneous E ect from Dynamic Regressions with Het- eroskedasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 2.3.3 Announcements and FX Volatility . . . . . . . . . . . . . . . . 64 2.3.4 Testing for Asymmetry . . . . . . . . . . . . . . . . . . . . . . 66 2.4 Market Sentiment, Uncertainty, and Macroeconomic News . . . . . . 67 2.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 iv A Appendix 72 A.1 Data Description and Source . . . . . . . . . . . . . . . . . . . . . . . 72 A.2 Contract between capitalist and investors . . . . . . . . . . . . . . . . 73 A.3 The signs of Major Variables in Linearization . . . . . . . . . . . . . 77 A.4 Convergence under Perfect Foresight . . . . . . . . . . . . . . . . . . 79 A.5 Analytical Solution for Real Exchange Rate . . . . . . . . . . . . . . 82 A.6 Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 A.7 Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 Bibliography 136 v List of Tables 1.1 Relations between log EMBIG+ spread and Exchange regime . . . . 3 1.2 Steady state parameter values for simulation . . . . . . . . . . . . . . 32 A.1 Regression During Tranquil Times . . . . . . . . . . . . . . . . . . . . 85 A.2 Short History of Crises from 1997 to 2010 . . . . . . . . . . . . . . . 86 A.3 U.S. and National News Announcements . . . . . . . . . . . . . . . . 88 A.4 U.S. and Domestic News Response and R squares . . . . . . . . . . . 93 A.5 The Impact of Major News Surprises on FX Returns and FX Volatility 95 A.6 Response of Major News Surprises and Announcement E ects . . . . 96 A.7 F-Test Results with Symmetric Response between Positive and Neg- ative News Surprises . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 A.8 Impact of Major News Surprises with FX Forecasts . . . . . . . . . . 98 A.9 Impact of Major News Surprises with FX Forecasts Dispersion . . . . 99 A.10 Summary Table for FX Time Series . . . . . . . . . . . . . . . . . . . 100 A.11 Exchange Regime Changes from 2000 to 2006 . . . . . . . . . . . . . 101 A.12 Summary Statistics for Market Forecast . . . . . . . . . . . . . . . . 104 A.13 Summary Statistics for Market Forecaset Dispersion . . . . . . . . . . 105 A.14 Return and Volatility News Response Coe cients . . . . . . . . . . . 106 A.15 Return and Volatility Response with Announcement Dummy . . . . . 109 A.16 Regression Results with Expected Appreciation . . . . . . . . . . . . 113 vi List of Figures A.1 IS and BP Curves under Di erent Exchange Regimes . . . . . . . . . 115 A.2 Impulse Response under the Fixed Exchange Regime with a Deferred Wage Shock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 A.3 Impulse Response under the Flexible Exchange Regime with a De- ferred Dage Shock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 A.4 Impulse Response under the Fixed Exchange Regime with a Technol- ogy Shock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118 A.5 Impulse Response under the Flexible Exchange Regime with a Tech- nology Shock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 A.6 Impulse Response under the Fixed Exchange Regime with a World Interest Rate Shock . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 A.7 Impulse Response under the Flexible Exchange Regime with a World Interest Rate Shock . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 A.8 Impulse Response of Consumptions with a positive deferred wage Shock122 A.9 Impulse Response of Composite Price with a Positive Deferred Wage Shock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 A.10 Impulse Response of Risk Premium with a Positive Deferred Wage Shock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124 A.11 Impulse Response of Risk Premium with Technology Shock . . . . . . 125 A.12 Impulse Response of Risk Premium with Interest Rate Shock . . . . . 126 A.13 Impulse Response under the Fixed Exchange Regime with Deferred Wage shock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127 A.14 Impulse Response under the Flexible Exchange Regime with Deferred Wage shock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 A.15 Impulse Response of Composite price with Deferred Wage shock . . . 129 A.16 Impulse Response of Risk Premium with Deferred Wage shock . . . . 130 A.17 Sample Autocorrelation Graphs of 5-minute Returns across Countries 131 A.18 Sample Autocorrelation Graphs of 5-minute Absolute Returns across Countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132 A.19 Evolution of EM Exchange Rates Responses to U.S. News . . . . . . 133 A.20 U.S. News Impact curve . . . . . . . . . . . . . . . . . . . . . . . . . 135 vii Chapter 1 Exchange Rate Regimes and Risk Premia under Alternative Wage Structure 1.1 Introduction As nancial markets have developed, various nancial goods are designed for emerging market countries in order to overcome country speci c risk. One of the examples is the country default swap, which measures relative risk of a country compared with the bond return of a so called riskless country. These nancial goods provide some insurance against a country?s default risk to bond holders by paying a premium to the counterpart while it has an obligation to buy the bond usually at par value when country issuing the bond declares default. However, the concept of measuring the risk factor of a country from the nancial market is not new at all. For instance, J.P. Morgan provided the Emerging Market Bond Index Plus (EMBIG+) spreads for widespread countries for the same purpose in the mid 1980s. This index displays the di erence between a developing country?s bond return and the U.S. Treasury bond adjusting some factors such as maturity and dividends. Although there are some minor di erences1, both indices are widely used to check 1For example, there is a counterpart risk in the credit default swap, the risk that the insurer may go bankrupt when it is forced to buy the defaulted bond. The demise of AIG during recent worldwide crisis illustrates this risk. 1 how some countries face default risks. Since those indices, notably credit default swap are actively traded in the Over-The-Counter (OTC) market, it has become more convenient to obtain high frequency data. Considering the fact that the defaults of a sovereign country are rare, these indices are treated as a good proxy to measure the default risk of emerging markets. Therefore, a lot of work to analyze country risk turned their attention from the default events itself to the factors that a ect these indices. This also contributes to broaden our knowledge by including some other countries who are rarely or never defaulted.2 In addition, this issue is very attractive for both analysts in the nancial markets and economists in the academic world. Analysts have a great incentive to precisely estimate the price of the nancial derivatives to get the arbitrage chances, and economists have a better tool to understand the nature of crises. From regression results, which will be discussed in detail in section 5, I nd out that risk premium under the xed regime is lower than under exible one. For the case of crawling regime, risk premium is lower but not signi cant. When emerging countries went into the free falling, then risk premium increases sharply. Under the free falling regime risk premium is high since the country is experiencing economic devastation. However, there is no tangible answer that explains the di erence of risk premium between xed and exible regimes. This empirical result may be caused 2Before Asian crises in late 1990s, the default events analyzed widely is the cases of Latin American countries such as Mexico, Argentina, and Brazil who frequently defaulted their debt in order to have relevant data for the defaults. Even in those cases, there are always some critiques how those events provides general aspects. 2 by a lot of economic factors, but I focus on the low level of nancial development in emerging countries to explain this phenomenon. Table 1.1: Relations between log EMBIG+ spread and Exchange regime Regime Coe cient Fixed -0.16*** Crawling -0.06 Free Falling 0.55*** There are some factors that should be considered in order to overcome the equivalent response of risk premium under various exchange regimes, as in CCV (2004). In developing countries, the access to the international nancial market is heavily restricted as a tool for savings. This may be caused by various rea- sons: the government may prohibit this access in order to satisfy the economy?s need of capitals. Since developing countries are su ering lack of xed capitals to increase production, capital control is usually one of tools used to avoid capital out ows. Transaction costs are another factor hampering access. Considering that the amounts of savings in developing countries are relatively small, transaction costs that is acceptable in developed countries can be a major obstacle to facilitate holding foreign assets in developing countries. Therefore in developing countries, a relatively small number of people can go to the nancial market for savings. This idea is the basic cornerstone of segmented asset market models. According to Lahiri et al. (2007), in the United States as a developed country \as of 1989, ... 59 percent of 3 U.S. households did not hold any interest bearing assets". Lahiri et al. also com- mented that 25 percent of households do not have checking account. We can easily imagine that the nancial situation of developing countries are well behind that of the U.S. For example, Jeon and Lim (2008) state, according to Korean Retirement and Income Study (KReIS) panel data, only 50.18 percent of households who joined the survey in 2005 have savings for the purpose of retirement. 3 Considering this low ratio of asset market participants in developing countries, it is critical to analyze economic behavior without any saving tools except wage. Using this setup, I illustrate the following properties: rst, the response of risk premia among exchange regimes vary. The response of risk premium under the xed exchange regime is smaller in response to a positive deferred wage shock. That under the exible regime with in ation target is higher. Second, the xed regime is preferred to the exible exchange regime. Next, the order of exchange regimes based on the response of risk premium does not change when other real shocks are applied. When it is assumed that there exist some asset holders in the economy with a segmented asset market model, the results do not change with only smaller magnitude of response of risk premium among exchange regimes. What causes this di erence among exchange regimes? Based on the di erent de nitions of exchange regimes, in which nominal exchange rate is set to be constant 3The reader may be surprised this relatively low percentage of savings in Korea. According to National Statistics O ce in Korea, the share of nancial account holders in households is about 98.6 percent in 2007. However, the share of the households that hold accounts for the investment is surprisingly low, 39.6 percent. 4 under the xed regime while domestic price is assumed to be constant under the exible regime, the slopes of IS and BP curve are steeper under the exible regime. Also, the di erence of de nitions causes higher magnitude of impact from a positive deferred wage shock. Therefore, real exchange rate and capital used for the next period?s production is higher under the exible exchange regime than under the xed regime. Since the production at the current period is a negative function of real exchange rate, depreciation in real exchange rate leads into lower level of production at the current period. After the shock, the production is higher under the exible regime. However, this is not enough to cover initial loss of welfare. Furthermore this means that the volatility of both production of nal goods and consumption is higher under the exible regime. On the other hand, the reduce of output at the current period has a side e ect. Coupled with higher increase of real exchange rate under the exible regime, the lower level of nal goods production causes lower level of remaining wealth that will be used for the next period capital production. This will increase the leverage ratio by borrowing more from the international capital market, which results in higher level of risk premium under exible regime. This paper is related to a lot of previous work. From the empirical point of view, Longsta et al. (2007) states the relation between credit default swap and major economic variables. Longsta et al. (2007) insist that the excess returns from investing in sovereign credit are largely compensation for bearing global risk, and there is little or no country-speci c credit risk premium. Their focus is on how country speci c factors may a ect the credit default swap, so the authors ran 5 the regression on country by country basis. Also, the authors did not consider the possibility of exchange rate regimes as a main factor. In my work, I use panel data to check these relations, and include exchange rate regime to check the di erence of the impact on risk premium. Jahjah and Yue (2007) is more related to the exchange rate regime. They show that spreads depend on exchange rate regime and that bond spread are the highest when the exchange regime is a hard peg. One of the interesting point in Jahjah and Yue (2007) is that exchange regime each country declares is not so relevant, therefore exchange rate regime classi cation is important to investigate actual impact of exchange rate. Talvas et al. (2008) tackle the problem of de facto regime codings in this regard. On theoretical front, Lahiri et al. (2007) and Cespedes et al. (CCV, 2004) should be noted. Based on the \ nancial accelerator" from Bernanke et al. (1999), CCV investigate how exchange rate policies a ect the small open economy under rigid wage. The authors claim that the conventional idea of preferring a exible regime to a xed one survives with nancial imperfection and balance sheet e ects. This is because under a xed regime real devaluation drops real wage as does the production while a exible regime successfully insulate real shocks. Two things are worth mentioning: rst, their model does not provide any distinction of risk premia across exchange rate policies. As mentioned in the working paper version, this is contrary to the recent policy literature. Second, the worker is passive in the sense that the response to the economic shock is restricted only to the demand of con- sumption goods and supply of labor. There is no nancial asset in this model so it is not possible to assess how the nancial market for worker a ects the economy. 6 On the other hand, Lahiri et al. suggest that under a xed regime the volatility of consumption for non-asset holders is lower since they can pool the risk intertem- porarily. As a result, the xed regime is preferred under the general condition where share of asset holders are large. Since they assumed explicit output shock without production, it is not possible to investigate the role of balance sheet e ects. There are other works on exchange rate regime comparison. Aghion et al. (2009) suggest that nancial development level is important in the sense that a xed exchange rate regime is bene cial for an economy with a lower nancial development level. Devereux et al. (2006) insist that the degree of exchange rate pass-through for import goods is critical for the assessment of monetary rules. However, they con- clude exchange rate pass-through degree does not a ect welfare ranking for exchange rate regimes so that exible regime is always preferred. Choi and Cook (2004) have a di erent opinion on the comparison issue. They argue that when the default risk premium depends on domestic banks? balanced sheets due to asymmetric informa- tion, a xed regime stabilizes bank balance sheets and so o ers greater stability than exible regimes. Devereux et al. (2006) suggest that openness of the economy may a ect the implication of exchange rate regime from their empirical studies. Magud (2010) shows that with high level of external debt, small open economies are better o with exible regimes to the extent that they are su ciently open. In the case of relatively closed economies his conclusion is that \ xed regimes are better real shock observers". The structure of this paper is as follows: in section 2, I provide theoreti- cal model that generates di erent response of risk premium to real exchange rate 7 changes under various exchange regimes. It is followed by simulation results and implications in section 3. In section 4, the reader may nd out extended model where the assumption of no foreign bond holdings is loosened such that there are xed proportion of asset holders. Section 5 provides empirical evidence that sup- ports lower level of risk premium under the xed regime. Section 6 concludes. In addition, Source of data and technical issues are summarized in the appendix. 1.2 Model 1.2.1 Basic Model The main objective of this model is to study how the change of two wages, that is conventional wage and deferred wage, a ects the economy according to the exchange regimes. When the share of a deferred wage that can be used for the next period consumption increases, the conventional wage that can be used for the current period consumption is reduced. Furthermore, the de nition of exible regime that make domestic goods price constant a ects the larger magnitude of response from the shock, which is discussed later. Therefore, there is a di erence in response of the real sector in the sense that volatility of the real variables under the exible regime is higher. In order to connect this uctuation of real economy with risk premium I fol- low the model from CCV, where the risk premium is generated by random pro t level of individual capital producer. In this setup, the source of risk premium is that individual capitalist producing capital for the nal goods production may go 8 bankrupt since its pro tability is a random variable and it is realized after nancial transaction. Since foreign investors know the distribution of pro t and possibility of defaults in some of invested money, their required return should be higher than riskless interest rate in order to compensate the loss from defaults. However, the model from CCV does not generate di erent risk premium level as exchange regime varies. The main source of making risk premium di erent among the exchange rate regimes is distinction in real variables behavior when those are faced with a positive deferred wage shock. When the volatility of real variables are higher under the exible regime, then the behaviors of nancial variables such as capital, debt, and remaining wealth are di erent as well. This logic is well known as a balance sheet e ect, but this is not the only source of the higher risk premium. The volatility of real variables is the core that initiates this e ect in the balance sheet. In this model, the core assumption is that households receive two di erent types of earnings, wage and deferred wage. As a compensation of labor supplying, households receive wages at the current period. At the end of the period after the production and all market clears, the producer will provide a deferred wage that is a xed share of rms? revenue. This deferred wage cannot be used for the current period?s consumption, but will be used for the next period. It is assumed that households provide labor and receive conventional wages during the period and deferred wage at the end of the period. In addition, it is also assumed that they do not hold nancial asset (or debt). Finally, their consumptions are restricted by Cash-In-Advance constraint. 9 There are four di erent types of players in this model; households, capitalist, nal goods producer, and government. Households provide labor to nal goods pro- ducer in order to produce nal consumption goods. They receive wages that consists of two parts, wage and deferred wage. Wage is provided to the households at the same period so that they can use this salary for the current period consumption. At the end of the period, nal goods producer provide deferred wagees to the house- holds that can be used for the next period consumption. The deferred wage is a fraction of total revenue of production. Even though the households have the infor- mation of deferred wage, they cannot adjust their current consumption level based on the amount of deferred wagees since they are subject to the Cash-In-Advance constraint. Capitalists produce capital that will be used for the next period production of nal goods. The idea of capitalist adopted in this model is exactly the same as those from Bernanke, Gertler, and Gilchrist (1999, hereinafter BGG) and Cespedes, Chang, and Velasco (2004, hereinafter CCV). Household?s Utility depends on consumption and labor. Since it is assumed that consumption is restricted on the money holdings as in cash-in-advance con- straint, the money holdings are not included in the utility function. Vt = Et " 1X s=t s tu (Cs; Ls) # (1.1) Utility function on each period follows GHH utility function, where > 1 represents the elasticity of labor and measures risk averseness of the households. 10 u (Cs; Ls) = (Cs 1 L s ) 1 1 (1.2) It is assumed that there are two goods for consumption; home goods and foreign goods (imports). Since it is also my interest to understand the behavior of exchange rate, it is indispensible to include two goods so that it is possible to de ne the real exchange rate as the relative price between those two goods. And the composite goods are de ned as follows: Cs = (CH;s) (CF;s) 1 (1 )1 (1.3) The imported good has a xed price, normalized to one, in terms of a foreign currency. It is freely traded internationally and the Law of One Price holds, so that the local price of a unit of imports is equal to the nominal exchange rate, St, per foreign currency. 1.2.2 Timing The timing issue should be clear in this model since there are many participants in this economy. At the start of period t, labor market opens with the knowledge of a shock to the share of deferred wage, where nal goods producers and households join to determine the equilibrium level of labor and wage with predetermined level of capital at the previous period t-1. As a result of labor market transaction, the equilibrium level of labor is used for nal goods production. The households can use the conventional wage at the current period and the deferred wage that is paid 11 at the end of the last period for the consumption in the current period. Then, it turns into the payment time. First of all, nal goods producers pay wages to households, and provide interest rates to capitalists. The capitalists pay back the debt from the last period with an interest rate, consumes foreign goods only to simplify goods market clearing, and leave some of the money for the next period of capital production as a remaining wealth. Based on this remaining wealth, the capitalist decide the level of capital for the next period and borrow money from foreign investors in order to prepare capital production that will be used in the next periods. Households consume nal composite goods with wage and the deferred wage that is given at the end of last period. At the end of the period when all the markets clear, nal goods producer provide households the deferred wage that will be used for the next period consumption. This deferred wage does not provide interest since this is given to the households at the end of the period. 1.2.3 Households Following the assumption households do not hold foreign assets, the wealth for the current period?s consumption is based on money holdings from the previous period and wage earned from current period labor. For the convenience of analysis, it is assumed that the households take the deferred wage, denoted by Mt as given.4 4Even though it looks too restrictive, this assumption does not change the results which is suggested later. For instance, it is more reasonable to consider that the deferred wage is exogenous if it is de ned as a portion of total revenue of nal goods producer. Under this de nition, the risk premium under the xed regime is still less than under the exible regime with a real shock. 12 Cash-in-advance Constraint is provided by the following equation: Mt + (1 vt)wtLt = PtCH;t + StCF;t (1.4) where Pt is the price level of home product, St is the price of foreign product, that is the same as nominal exchange rate. In equation (1.4), vt is the share of the deferred wage from total wage income. So, the total deferred wage that is paid at the end of the period is the share vt times total wage income. Mt+1 = vtwtLt (1.5) Then the price level of composite goods is denoted by Qt such that Qt = (Pt) (St) 1 (1.6) Then, the right handed side of (1.4) can be rearranged into a multiplication of composite price and consumption by simple calculation with rst order conditions of home and foreign goods consumptions. Mt + (1 vt)wtLt = QtCt (1.7) Using utility function suggested above and (1.7), Lagrangian is L = Et " 1X s=t s t (Cs 1 (Ls) )1 1 + 1X s=t s s t(Ms + (1 vs)wsLs QsCs) # (1.8) , and rst order conditions are as follows: Ct 1 (Lt) = tQt (1.9) 13 (Ct 1 (Lt) ) (Lt) 1 = t(1 vt)wt (1.10) By solving utility maximization problem, consumption and labor level will be determined as follows: Ct = Mt + (1 vt)wtLt Qt (1.11) (1 vt)wt = Qt (Lt) 1 (1.12) There are nothing particular but the deferred wage shock in those equations. In equation (1.67), the reader may easily understand that labor supply depends on the portion of real wage, wt=QT , that can be consumed in current period. Since the households take the deferred wage as given and are bound to the cash-in-advance constraint, the consumption in the current period is governed by the real value of deferred wage from last period and a portion of real wage. 1.2.4 Capitalists In this model, capitalists produce physical capital and sell it to nal home good producer. They need home goods and foreign goods as sources for capital production. To nance investment, he can use his own money that was left at the last period or borrow money denominated as foreign currency from abroad. For reference, this capitalist setup is adopted from Cespedes et al. (2004). There are some reasons why capitalists should be included in the model. First of all, it is capitalists who may go bankrupt in the model rather than governments. In the model suggested here, there exist individual capitalists with mass 1 who are 14 identi ed with random pro tability. Based on this random variable, that is prof- itability, each individual may go bankrupt if the realized pro t level is below the level of debt repayments. As a result, it is possible to de ne risk premium in the model without any government default based on budget de cits. Second, it is easier to adopt capitalists in order to add a nature of nancial crisis. One of the core rea- sons that lead into nancial crisis is nancial accelerator in the sense that the value of debt repayment may be higher when devaluation (or depreciation under exible exchange regime) occurs. Without capitalists, banking sector and government deci- sion should be included to take this nature into consideration. Finally, it should be noted that the model suggested here is based on the shock of wage structure. This shock directly a ects on the demand and supply of labor, so that the level of capital will change indirectly. As a result, the behavior of risk premium will be passive if capitalists do not exist in the model. Physical Capital production is de ned as the same fashion as the composite consumption goods: Kt = (XH;t) (XF;t) 1 (1 )1 (1.13) where XH and XF mean home goods and foreign goods, respectively. Also, it is assumed that physical capital is entirely depreciated after the nal home good production. Due to the structure of the production function of physical capital, the cost of a unit of capital is Qt, as presented above in (1.6). Then the budget constraint of the capitalist is 15 PtNt + StDt+1 = QtKt+1 (1.14) where Dt+1 is borrowing from abroad, and Kt+1 means investment in period t+1 capital. In this setup, the borrowing is subject to friction. Following from Bernanke et al. (1999) capitalists can go bankrupt due to the idiosyncratic disturbance at- tached on their ex-post gross return. In this case, it is assumed that foreign lenders will monitor the situation that the capitalists face with some costs, and seize all the remaining. This informational asymmetry is the main reason why there ex- ists risk premium in this model. Considering risk premium, the expected return to investment is de ned by Et [Rt+1Kt+1=St+1] QtKt+1=St = (1 + rt+1)(1 + t+1) (1.15) where t+1 is the risk premium between period t and t+1. Using producer?s rst order condition, this equation can be rearranged as follows, which governs demand of capital: Et [ Pt+1Yt+1=St+1] QtKt+1=St = (1 + rt+1)(1 + t+1) (1.16) Bernanke et al. (1999) show that risk premium is an increasing function of the ratio of the value of investment to net wealth and risk premium. This governs the supply of capital. It should be noted that this equation is the result of the maximization of capitalists. The theorectical analysis of capitalist maximization is 16 provided in the Appendix A.2. 1 + t+1 = F QtKt+1 PtNt (1.17) Following CCV, it is assumed for the calibration later that F (G) = G , where > 0. This assumption has a trade o in the analysis. Most of all, it simpli es the model so that it is possible to solve the model analytically using log linearization. Without this assumption, a group of rst order conditions from the maximization problem for capitalists should be solved simultaneously, which does not provide any further insights for understanding. On the other hand, the assumption of function form will erase the impact of statistical structure of random variable for pro tability. As a result, it is not possible to analyze how the economy responds to the change of random variable of capitalists. Since the model is concentrated on the respose of the economy to the change of wage structure, it is justi ed that the loss from this assumption is minimal. It is assumed that capitalist consume 1 share of the remaining after the debt repayment, and he only consumes imports. Then the level of wealth remaining for producing capital at the next period is PtNt = f PtYt (1 + rt) (1 + t)StDtg (1.18) 17 1.2.5 Final Goods Producer Producer simply uses labor and capital to produce nal goods with given price level of capital. Since the contract with households includes deferred wage as a xed share of revenue as well as wage, the producer should consider this deferred wage when it maximizes its own pro t level. Production function is assumed to have standard Cobb-Douglas function form. Yt = AK t L 1 t (1.19) Considering the return for the capital is decided by the capitalists and inter- national investor, pro t maximization problem will be as follows: t = PtYt RtKt wtLt (1.20) Then, First order conditions for the capital and labor will be suggested. PtYt = RtKt (1.21) (1 )PtYt = wtLt (1.22) 1.2.6 Equilibrium Condition Home goods produced by nal goods producers can be consumed by house- holds, used for the capital production by capitalists, or exported to foreigners. In order to simplify the model, the exports to the foreigner are assumed to be constant across periods. Then market clearing condition for home goods are as follows: PtYt = Qt (Kt+1 + Ct) + StX (1.23) 18 where X denotes exports, which is constant. 1.2.7 Steady State In order to derive steady state solution, the price level of home goods is nor- malized to 1, i.e. P = 1, without any loss of generality. Then the price of home goods can be dropped in the steady state equations. The following equations are the main results for steady state variables, where I drop the time subscript. (1 )Y = Q(L) (1.24) Y QK = (1 + r)(1 + ) (1.25) QC = wL = (1 )Y (1.26) Y = Q (K + C) + SX (1.27) N = [ Y (1 + r)(1 + )SD] (1.28) Q = S1 (1.29) N + SD = QK (1.30) 19 The starting point to derive steady state variables is risk premium. Plugging equation (1.25) and (1.30) into equation (1.28), it is possible to derive the following: [1 (1 + r)(1 + )] (QK SD) = 0 (1.31) If it is assumed that remaining wealth for the capitalist is positive, then the second term in the equation cannot be zero. Therefore, the risk premium is a function of risk-free interest rate and the share of consumption for capitalist or 1 + = 1 (1 + r) . (1.32) The next step is to pin down Y; S at the steady state. Plug Demand of capital (1.25) and budget constraint for households (1.26) above into (1.27), then the rst equation for the (Y,S) space is derived. [1 (1 + )]Y = SX (1.33) For the second equation for (Y,S) space, using (1.24) L = (1 )Y S1 1 (1.34) And from production function, K = Y AL1 1 = Y A 1 (1 )Y S1 1 Then plug this equation into (1.25), Y = (1 + r)(1 + )S1 Y A 1 (1 )Y S1 1 20 A 1 (1 + r)(1 + ) (1 ) 1 = S1 ( 1)+1 Y ( 1)(1 ) (1.35) Therefore, we can derive the steady state level of a pair (Y; S) using equation (1.33) and (1.35). It can be easily shown that (1.33) has a positive slope and (1.35) has a negative slope since is assumed to be greater than 1, therefore these two equations provide unique pair of solutions for (Y; S). The other variables can be easily derived. The composite price level Q is driven by equation (1.29), level of labor from (1.34), and capital for production can be derived from L and Y using production function. 1.2.8 Fixed Exchange Regime In order to track dynamic behavior of main variables, I use log linearization for system of equations. All the lower case letters below with time subscript mean log linearization of the variables except denoted otherwise 0t+1 means the log lin- earization of 1 + t+1. It is assumed that the economy remains in the steady states before the shock in the deferred wage occurs. Since the capital level when the shock occurs is in the steady state level, it is clear that kt = 0. Then, log linearized version of production function can be denoted as yt = (1 )lt. (1.36) Since the real exchange rate in the model can be de ned as Et = St=Pt, the linearized version of this can be denoted as et = st pt. Then, from the de nition 21 of composite price, Qt, the following equation is derived. qt pt = (1 )et (1.37) Since the labor is governed by the demand and supply of labor, the linearized version of labor is denoted as lt = 1 1 et + v (1 v) 1 (1 ) vt. (1.38) From this equation, the reader can easily nd out that there is negative rela- tionship between the real exchange rate and labor under the assumption that > 1. This means that depreciation makes labor lower under any exchange rate regime. In additioin, it should be noted that the term for a deferred wage shock, vt, a ects la- bor negatively since the coe cient is negative. It should be reminded that a positive deferred wage shock means drop of wage in the current period. Since the marginal wealth from additional labor supply reduces while there is no change from marginal disutility of labor supply, the equilibrium level of labor will be less than the steady state level. Since the labor is denoted as a function of real exchange rate, so is the output. yt = (1 )lt = (1 )(1 ) 1 et+ v (1 v) (1 ) (1 ) vt = et+ v (1 v) (1 ) (1 ) vt (1.39) The reader can easily nd out that < 0. Therefore, the depreciation leads into lower nal goods production in period t. 22 Using labor demand, equation (1.38), and the fact that pt = et under the xed regime, the linearized wage can be described as wt;fix = 1 1 et v (1 v) (1 ) vt. (1.40) The behavior of nominal wage per unit of labor depends on both real exchange rate and a deferred wage shock. Keeping in mind that > 1, one can easily understand that nominal wage per labor is a negative function of real exchange rate and a positive function of a deferred wage shock. When we look into the behavior of total income, that is wt + lt, the meaning is clearer. wt;fix + lt = ( 1 + ) et + v (1 v) (1 ) (1 ) vt. (1.41) With some simple rearrangement, it is possible to show that total labor income is a negative function of both real exchange rate and a deferred wage shock under the xed regime. So when a positive deferred wage shock happens, total wage income will be below from the steady state level. In addition, the decrease of wage income is not only from direct e ect of deferred wage shock, but also from indirect e ect with depreciation. Linearized version of equation (1.66) can be used for the response of consump- tion in households. Since it is derived that nominal wage and labor are functions of real exchange rate, the consumption is denoted as a function of real exchange rate and a deferred wage shock. 23 ct = [ + (1 v) ( 1 + )] et + v 1 vt = Afixet + v 1 vt (1.42) With the fact that is negative and some mild restriction on parameters, consumption of household is a negative function of both real exchange rate and deferred wage shock.5 Therefore, with a positive shock and increase of real exchange rate, consumption is lower than steady state level. The next step is to follow nal goods market using linearized equations to have IS curve. In order to track the relations, equation (1.66) is inserted into linearized version of equation (1.81) and can be rearranged into the following (1 2(1 v)) yt = 1(qt + kt+1 pt) 2vvt 2vpt + (1 1 2)et (1.43) , where 1 = QK=PY = and 2 = QC=PY = (1 ) respectively. Using equation (1.39), (1.37), and the fact that pt = et under the xed regime, the equation (1.43) can be presented as a function of the real exchange rate and the capital level at the next period. fixet = 1kt+1 + CISvt (1.44) fix = ( 1 + 2)(1 ) (1 1 2) 2Afix < 0 (1.45) CIS = v (1 v) (1 ) (1 ) ((1 v) 1) < 0 (1.46) 5The restriction and sign of Afix is discussed in appendix. 24 The reader may nd out how the signs of fix and CIS is derived with an assumption on parameters. Following those signs, the IS curve represented by the equation (1.44) has negative slope, and this curve moves upward when positive deferred wage shock occurs. The other function that covers another relationship between et and kt+1 is from following linearized equations. 1(qt + kt+1 pt yt) = Bfixet + CBPvt (1.47) Bfix = ( 1 + ) (1 1 2(1 v)) < 0 (1.48) CBP = (1 1 2(1 v)) v (1 v) (1 ) (1 ) + 2v (1.49) 0t+1 0 t = [(qt + kt+1 pt yt) + [(et Et 1et) (yt Et 1yt)]] (1.50) 0t+1 = kt+1 + yt+1 et+1 + et (1.51) yt+1 et+1 = 0 t+1 (1.52) Equation (1.47) is an rearranged version of equation (1.43), and the reader can easily nd out that Bfix is negative from equation (1.48) as before. Equation (1.50) is derived from equation (1.14), (1.16), (1.17), and (1.18). the next equation (1.51) is a linearized version of equation (1.16). The last equation is from the saddle path stability, which can be found out from Appendix, where it can be seen is 25 greater than zero and less than one. Taken the perfect foresight into consideration and 0t = 0, the equations described above can be summarized into the following equation that governs the relationship between et and kt+1. (1 ) 1 Bfix et = kt+1 + (1 ) 1 CBPvt (1.53) The coe cient of real exchange rate from the left handed side has positive sign because Bfix < 0 and 0 < < 1. Therefore the BP curve has positive slope and moves upward when there is a positive deferred wage shock, vt > 0. Both IS and BP curve moves upward when there is a positive deferred wage shock. Under the general situations of parameter values, it can be seen that IS curve goes higher in response to the shock. Therefore, it is easy to prove that both real exchange rate for the current period and capital for the next period increase as a result. Then, using the depreciation of real exchange rate, we can verify that labor and nal goods production decrease. Considering that real exchange rate increases and product decreases, the level of wealth that will be used for the next period production by the capitalist also is reduced from equation (1.18). We can nd out the behavior of risk premium using equation (1.17). Three factors a ect the risk premium: real exchange rate, capital for the next period, and the remaining wealth. I have already proved that both real exchange rate and capital for the next period increases, and the remaining wealth decreases. Therefore, risk premium for the next period will increase. 26 1.2.9 Flexible Exchange Regime Under the exible exchange regime, the behavior of variables including pro- duction and labor are the same as those under the xed exchange regime. The rst di erence is from nominal wage per a unit of labor. wt;flex = (1 ) 1 et v (1 v) (1 ) vt (1.54) In addition, we need to see the behavior of wage income for the clear picture as before. wt;flex + lt = et + v (1 v) (1 ) (1 ) vt (1.55) It is clear that wage income under the exible regime is a negative function of both real exchange rate and a deferred wage shock. The di erence of wage income is from the assumption that pt = 0 under the exible regime. Comparing wage incomes for both exchange regimes, we can nd out that depreciation makes nominal wage income decrease less under the exible regime. Now using nominal wage income under exible regime, it is possible to derive consumption. ct;flex = ((1 v) (1 )) et + v 1 vt = Aflexet + v 1 vt (1.56) It is clear that the coe cient of real exchange rate in this equation is negative, so the combination of depreciation of real exchange rate and a positive shock leads into lower consumption than steady state level of consumption. Furthermore, com- parison of coe cients in real exchange rate shows that slope under the xed regime 27 is bigger than under the exible regime, 0 > Afix > Aflex . That means the impact of depreciation in real exchange rate causes bigger drop in consumption under the exible regime. Deriving IS curve for the exible regime is almost the same as the case under the xed regime, except that pt = 0 and nominal wage suggested above are used. flexet = 1kt+1 + CISvt (1.57) flex = ( 1 + 2)(1 ) (1 1 2) 2Aflex < 0 (1.58) The only di erence between two IS curves is the coe cient of real exchange rate in the left handed side. It can be veri ed that fix flex = v 2 < 0, which means that the slope of IS curve is negative for both regimes and steeper under the exible regime. In addition, the response from positive deferred wage shock is higher under the exible regime sinced the di erence of the coe cients for real exchange rates. 0 < CIS fix vt < CIS flex vt (1.59) The same equations are used for deriving BP curve for the exible regime with the use of other assumption, that is pt = 0. Due to the di erence of de nition of regimes, we can nd out that there is a little di erence in the BP curve as we can see in the IS curve. (1 ) 1 Bflex et = kt+1 + (1 ) 1 CBPvt (1.60) 28 Bflex = ( 1 + ) (1 1 2(1 v)) + 2v (1.61) As in the case of IS curve, the only di erence is the coe cient of real exchange rate. We can verify that the di erence of the coe cient under xed regime from exible regime is Bfix Bflex = 2v < 0. As a result, the slope of BP curve under the exible regime is positive and steeper than under the xed regime. 1.2.10 Comparison between Exchange Regimes Using IS and BP curves in each exchange regime, it is possible to nd out analytical solutions of et and kt+1 as a function of a shock vt. Figure A.1 presents how real exchange rate and capital changes when a positive deferred wage shock occurs in the economy. Both variables stay at the origin before the shock since they are at the steady state level. Both IS and BP curve moves upward with a shock, but IS curves move higher since it is more responsive to the shock. Furthermore, IS curve under the exible regime moves higher than under the xed regime due to the di erence of coe cients. For BP curves, there are slight di erence between the regimes. Therefore, the capital and real exchange rate changes higher under the exible regime at the period when shock occurs. For the analytical solutions, the reader may nd out in the appendix for the derivation of the inequality of real exchange rates. Finally, it is possible to compare the response of risk premium under di erent exchange regimes. With perfect foresight and the fact that risk premium is at the 29 steady state, i.e. t = 0, equation (1.50) can be rearranged as a function of real exchange rate and deferred wage shock. 0t+1 = (qt + kt+1 pt yt) (1.62) Now using equation (1.47) and the compatible equation for the exible regime, then it is easy to derive the di erence of risk premia between xed and exible regime. 0t+1;fix 0 t+1;f lex = 1 [Bfixet;fix Bflexet;flex] < 0 (1.63) The inequality in the equation is veri ed in the appendix. Therefore, the risk premium under the xed regime is less than under the exible regime. The workhorse in this model that brings the di erence between two exchange regime are combination of de nition and composite price level. Following the as- sumptions of policies under exchange regimes, composite price level can be denoted di erently as qt;fix = pt + (1 )st = et;fix < 0 (1.64) qt;flex = pt + (1 )st = (1 )et;flex > 0. (1.65) According to IS-BP analysis discussed above, both real exchange rates increase in response to the positive deferred wage shock. Since the amount of money that can be used in the current period is being reduced, the demand of nal goods decrease. So, the relative price of domestic goods, which is the inverse of real exchange rate 30 should also decrease. Under the xed regime, the price of domestic goods is the only variable that can be adjusted. Therefore, domestic price should decrease and so should composite price level. However, the price of foreign goods should be adjusted by moving upward under the exible regime since domestic price is set to be constant. In consequence, the composite price level should increase under the exible regime. Taking these results in composite price into consideration as well as the decrease of nominal wage income, the consumption drop must be less under the xed regime. Since the consumption is lower under the exible regime, the resources used for the next period capital production will be higher. This means the capitalist needs to borrow more money from abroad to nance investment. However, due to the drop of the production and depreciation, their remaining wealth is lower under the exible regime. Those to forces leads into higher leverage ratio, and risk premium increases higher under the exible regime. 1.3 Simulation I set several parameters used in the model so that predictions of the model are empirically meaningful. For the parameters in the utility function, the coe cient of relative risk averse coe cient is set to 2 following Mendoza (1991). Also, (1 plus the inverse of the intratemporal elasticity of substitution in labor supply) is set to 2. I set the risk-free interest rate to 0.04 based on a 1-year constant maturity U.S. Treasury bill interest rate, and discount factor is set to the inverse of 1 plus risk-free interest rate. For composite consumption goods, share of home goods is 31 Table 1.2: Steady state parameter values for simulation Parameter Description Value Share of capital to output 0.35 Proportion of income for the investment 0.92 v Share of deferred wage from wage income 0.10 r Risk-free interest rate 0.04 Elasticity of labor 2 Share of home goods in composite goods 0.6 A Technology in production function 1 Discount factor 1/(1+r) Coe cient of risk averseness in utility 2 Elasticity of the risk premium 0.02 set to 0.6 based on CCV. For the production function, technology coe cient is set to 1, which is widely accepted in previous literature. Capital?s share in output in the production function is set to 0.35, in line with standard estimates. I choose other variables based on the nancial vulnerability case in CCV. They choose and to imply 400 basis points of risk premium and the leverage ratio as 1.2. Due to the structural di erence in households, the suggested parameters cause a little bit di erent results in my results with the and . The risk premium at the steady state is set to 450 basis points and the leverage ration as 1.12. The rest of the parameters that is important are the share of deferred wage from total revenue. The parameter is calculated from the enlisted companies in 32 Korean Stock Exchange in 2009 that clearly distinguish deferred wage from regular wage in the annual balance sheet in 2009. From that data, the ratio of deferred wage to total wage income is about 10% of conventional wage. The lower ratio of deferred wage to conventional wage does not change the main implication of the simulation results. Only the di erence of variables such as risk premium between two exchange rate regimes is smaller. The simulation results under a xed regime can be seen from Figure A.2 for a positive deferred wage shock. As expected from dynamic analysis, the capital for the next period and real exchange rate is higher at the period when a positive deferred wage shock occurs. Depreciation of real exchange rate governs the behavior of nominal wage and output level of nal goods. With depreciation and a positive shock for deferred wage share, the consumption is lower than steady state level as well when shock hits the economy. To turn our focus into nancial sector, it should be noted that remaining wealth is lower than steady state level since output is reduced and real exchange rate increases. In addition, debt is above steady state level since capital is higher but remaining wealth is lower. In order to understand the behavior of risk premium, it is crucial to check the behavior of leverage ratio as in eq (1.17). The key variable governing the behavior of leverage ratio is the remaining wealth, N , which is re ected by eq (1.18). Considering the decrease of nal goods production, increase of risk-free interest rate, the decrease of remaining wealth overwhelms relative increase of home goods price. As a result, the risk premium increases as well as leverage ratio. The same logic can be applied to the dynamics under the exible regime. The only di erence is the magitude of behaviors, 33 which is from distinctive response of wages from real exchange rateaccording to the exchange rate regimes. Figure A.8 provide the comparisons in consumptions under each exchange rate regime. It is clear that both consumptions are lower than inital steady state level. However, the consumption under the xed regime shows less deviation than under the exible regime. At the next period, the consumption is higher under the exible regime. This result is consistent since the deferred wage able to be used for the next period consumption is high from the shock. In summary, we can conclude that the volatility of consumption to a deferred wage shock is bigger under the exible regime. In addition, it is expected that the utility will be lower under the exible regime if it is assumed that utility is negatively a ected by the volatility of consumption, as in Lahiri et al. (2007).6 In addition, the main di erence from exchange regimes are from the dynamics of composite prices as discussed above. The reader may nd out that the initial response of composite price to a positive deferred wage shock in Figure A.9. With increasing real exchange rates, the composite price under the xed regime deviates downward from steady state level. However, the dynamics of composite price under the exible regime jumps up in response to the shock. For the comparison of risk premia according to the regime, Figure A.10 presents the di erence according to the real shock. As expected from the result of log lin- earization, the risk premium under the xed regime is lower than under the exible 6It should be noted that the utility function used here does not contain volatility of consumption as a factor. 34 regime. Due to the sharp increase of capital and decrease of net wealth, the risk premium increases at the period when a deferred wage shock occurs. At the next period, the risk premium converges to the original steady state level with a lower speed since capital for the nal goods production goes back to the steady state. Taken this result into consideration, I checked the behaviors of risk premium with other shocks such as technology shock and world interest rate shock. It is as- sumed that there is a 10 % technology development in a single period for technology shock, and there is a 1 % increase in a single period for world interest rate shock. The reader can nd out the results of impulse response for major variables in Figure A.4 to Figure A.7. To focus on the di erence of risk premium under both exchange regime, Figure A.11 and A.12 are helpful. The result that the risk premium is lower under the xed regime does not change at all for the other shocks. In addition, the di erence of risk premium is the biggest when there is a technology shock. The risk premium under the xed regime with technology shock seems to be quite smaller than under the exible regime. 1.4 Extension Considering the assumption that the households do not have foreign assets is quite strict, it is loosened by assuming there are some xed share of households holding foreign assets, which is called trader following the terminology from Lahiri et al. (2007). Under this set-up, I can prove that the result from above does not change even though there are some asset traders in the model. For the households 35 who do not have an access to the asset market, who are called Non-trader, denoted by NT, the maximization problem is the same as the households analyzed in section 1.2.3. Hence, it is used with superscript NT for the equilibrium level of labor supply, deferred wage, and consumption equations to obtain: CNTt = MNTt + (1 vt)wtL NT t Qt (1.66) (1 vt)wt = Qt LNTt 1 (1.67) 1.4.1 Trader Trader goes to the asset market before production begins and adjust money holdings with deferred wage received at the end of the last period, transfer from governments, and foreign asset holdings. M^Tt = M T t + St (1 + rt) ft Stft+1 + Tt (1.68) where ft is foreign riskless bond, and Tt means transfer from government. It is noted that only traders can join the asset market so the transfer from government is applied to only traders. Therefore, the amount of transfer is adjusted by the measure of traders, . Trader is also governed by Cash-In-Advance constraint. M^Tt + (1 vt)wtL T t = PtC T H;t + StC T F;t (1.69) From (1.68) and (1.69), MTt + Tt + (1 vt)wtL T t = PtC T H;t + StC T F;t + Stft+1 St (1 + rt) ft (1.70) 36 At the end of the period, deferred wage as a fraction of total revenue is provided to traders and this is used for the next period consumption. For the next period t+1, MTt+1 = vtwtL T t (1.71) Considering that PtCTH;t + StC T F;t = QtC T t , The lagrangian will be L = Et " 1X s=t s tu CTs ; Ls + 1X s=t s t s MTs + Tt + (1 vt)wtL T s QsC T s + Ssfs+1 Ss (1 + rs) fs # (1.72) First order conditions are as follows: CTt 1 (LTt ) = Tt Qt (1.73) CTt 1 (LTt ) (LTt ) 1 = Tt wt (1.74) Tt St = Etf T t+1St+1 (1 + rt+1)g (1.75) From (1.73) and (1.74), wt = Qt(L T t ) 1 (1.76) This is labor supply function from traders. Since GHH utility function is assumed, the labor supply does not depend on wealth e ect. Therefore, the labor supply of trader has the same functional form as that of non-trader described in (1.67). The euler equation is derived from (1.73) and (1.75). 1 CTt 1 L t = Et 2 4(1 + rt+1) Qt Qt+1 St+1 St 1 CTt+1 1 L t+1 3 5 (1.77) 37 This equation means that marginal utility of current period should be equal to that of next period when adjusted by price changes. 1.4.2 Government Government can use various tools to stabilize economy under the di erent FX regimes. Under the xed exchange regime, St = S, it can use nominal money to balance the economy. Under the exible exchange regime, There are two policy tools for government to follow: government can x nominal money supply Mt = M , or it can x price of home goods Pt = P , which is usually called in ation target policy. Under the constant nominal money supply, the change of the production level due to the exogenous shock will lead into the change of price level, so the amount of bonus that households will receive at the end of the period will be the same. In the case of in ation target policy, the change of output level will directly result in the change of bonus amount since the price never changes. Stht+1 (1 + rt)Stht + Tt = Mt+1 Mt (1.78) where h is foreign asset holdings. For the next step, the behaviors of capitalist and nal goods producers are the same as before, as de ned in sections 1.2.4 and 1.2.5 respectively. 38 1.4.3 Equilibrium Condition Since the trader in households is added in the model, the equilibrium conditions should be adjusted accordingly. Since the share of trader is assumed to be xed at , money market clearing condition should be the sum of money from traders and non-traders: Mt = M T t + (1 )M NT t (1.79) Also, since the labor supplies are from both households, the labor market clearing condition should be de ned. Lt = L T t + (1 )L NT t (1.80) Finally, home goods market clearing condition should be adjusted accordingly: PtYt = Qt Kt+1 + C T t + (1 )C NT t + StX (1.81) From deferred wage payment, it is possible to construct quantity theory equa- tion. Mt+1 = M T t+1 + (1 )M NT t+1 = vtwtLt = vt(1 )PtYt (1.82) From (1.70), (1.78), (1.82), and de ning gt = ht + ft, the ow constraint of the economy can be obtained: gt+1 (1 + rt) gt = ( vt + (1 ))PtYt St Qt St CTt + 1 Mt+1 Mt St (1.83) Using this ow budget constraint and rst order conditions, it is possible to 39 pick up the consumption level of traders. It should be noted that the last term of right handed side in (1.83) is the source of the redistribution of this economy, as suggested in Lahiri et al (2007). Since the deferred wage is the sole money in this economy at the end of the period, any change of money will belong to the trader from the participation of the asset market. As ! 1 meaning households are all asset holders, this term goes zero. This implies that this channel only exists when there is an asset market segmentation. In this extension with exogenously segmented asset market model, the reader may nd out that the same logic applies for the real variable movements. As you can nd out from Figure A.15, the responses of composite price di er in direction between exchange regime, and so does the wealth e ect. As a result, it can be seen that the risk premia under the xed regime is lower than under the exible regime in Figure A.16.7 1.5 Data The main focus is to check if there is any di erence in risk premium according to exchange rate regimes. In order to check if there is any di erence of risk premium under various exchange rate regimes, I include dummy variables for exchange rate regimes as explanatory variables except exible regime. One of the problems that arose when the CCV model was adopted was data availability. Since the CCV model assumes the default possibility of individual capitalist, it is consistent to use 7In order to handle unit root problem in the model, I used endogenous discount factor in simulating the extension model with asset holder. 40 rm level data in the empirical study suggested above. However, data availability problems arise when the rm level behavior is analyzed. First, it is extremely di cult to discover rms in developing countries that have regularly issue U.S. dollar denominated bonds. Second, even though some rms in developing countries issued foreign currency denominated bonds, the issue size of bond issued are so small that the nancial market for rm level bonds are not well developed. On the other hand, the rm should have high (at least investment grade level) grades from credit rating companies. This may cause some bias in pro tability distribution. Finally, the reader may think of Credit Default Swap (CDS) market data to overcome this problem. Unfortunately, the time series of CDS data for rm level have been too short until now8. Considering all the restrictions related to using micro-level data to obtain risk premium, it is still widely accepted to use risk premium from government issues bonds. Here, I provide detailed information for the variables used in the model. In order to check the relation between risk premium of developing countries and ex- change regimes, I use the EMBIG+ index spread from J.P. Morgan for 34 countries on a quarterly basis. The time periods of the data are from 1998 to 2007. Since the risk premium used here is unbalanced panel data set, Perron type unit root test for unbalanced panel data is used to check the possibility of unit root process. The null hypothesis that all the panel data follow unit root is rejected with 95 percent 8For instance, CDS for Samsung electronics and POSCO in Korea are available from November 2004, and LG electronics from May 2007 while CDS for Korea government bonds is available from April 2002. This data was obtained from Bloomberg terminal. 41 of con dence interval. However, this test is not perfect since it does not provide the evidence that the time series for each country does not follow unit root process. Therefore, I use both the level and the lag di erence as a dependent variables. Since our interest in on the relation of impact on risk premium, lag di erence is a better proxy than the level itself. Other than risk premia, the classi cation of the exchange rate regime is the most important variables. To de ne each country?s exchange rate regime, I use the coarse classi cation from Reinhart and Rogo (2004)9. This classi cation is based on actual behavior of exchange rates rather than the declaration of the governments. The choice of regime classi cation is important because the actual behavior of the exchange rate can be di erent from what the governments announce. Even though a country label its exchange regime as \ exible", it can use its power to intervene in the exchange rate market so that the exchange rate does not move exibly as expected. Furthermore, Reinhart distinguishes crises periods by adding a class called \free falling", so that the analysis based on their classi cation can be clear without any potential distortion from crisis. In order to clarify the relations of risk premia and exchange rate regimes, it is essential to study the extent to which economic and nancial variables explain the 9In the case of developing countries, there are relatively small number of countries that do not intervene in the foreign exchange market. Those countries are in general classi ed as \managed oating" by Reinhart and Rogo (2004). Since there are a relatively small number of countries classi ed as \ oating", I combined these two classi cation as \ exible" regimes. This changes little in the regression results. 42 variation of risk premia. Explanatory variables used for the purpose of controlling risk premia can be divided into two groups, domestic and global. In the domestic group, local stock index, government e ciency index, GDP per capita, real GDP growth, external debt, short term debt, and reserves are included. On the other hand, U.S. bond price, regional risk premia, PER of U.S. market, spreads for in- vestment grade bond, high yield bonds? spread, term premium of U.S. bond, and volatility of U.S. stock market are included as global economic variables. Details about these variables are presented in the Appendix. For domestic economic environments, debt related variables are considered indispensable for potential event of defaults, which is clearly described in various previous works such as Kaminsky and Reinhart (1997). In this regard, the ratio of external debt to GNI and short term debt to reserves are included in the explanatory variables. Also, reserves holding should be considered to check the capacity of repayment of foreign debt. Since the capacity of repayment is related to the level of debt, the ratio of reserves to external debt is included as an explanatory variable. From the perspective of capital ows to clear debts, the ratio of trade balance to GDP is adopted. In addition to these debt related variables, there are a lot of nancial and economic forces that may impact the risk premia. To capture the state of economy, I include the local stock index denominated on local currency, GDP per capita, and real GDP growth. Finally there may be some factors which are independent of economic situation and risk premia, but can e ect both. To control this endogeneity issue, government e ciency index is used as explanatory variable. 43 It is equally important to review global economic environments as a factor that can cause developing countries to have trouble repaying the debt. International investors may be reluctant to lend money to developing countries because their commitments are questionable, but it should also be considered that investors will not invest since there may be more pro table (or risk reducing) options in the markets. This logic is denoted as \ ight to quality" in the nancial markets, which can be easily seen when the world economic situation is pessimistic. As a stock market variable, the price-earning ratio of S&P 100 index is in- cluded. It should be noted that multiple stock related indices can cause multi- collinear problem when included at the same time. When excess return of S&P 500 index is adopted to re ect the behavior of the equity market with PER of S&P 100 index, the correlation of those indices are higher than 0.9, and coe cients change drastically as diverse subsamples are applied. Therefore, only the PER of S&P in- dex is chosen solely for the equity market behavior. For the bond market variation, I include the change in the ve year constant maturity Treasury (CMT) bond yield. This index is included because it is one of the best proxies for the U.S. economic growth, and it is highly a ected by the ight to quality issue. To consider di erent behaviors of bond according to the bond grades, the spreads of U.S. investment-grade and high-yield corporate bonds are included as explanatory variables. The core concept of ight to quality is that the portfolio will be concentrated on safe assets when the economic environment is getting worse. In this situation, the spread of investment grade level bonds will be lower compared with high yield bonds. To include these indices is important especially since the 44 bond grade of developing countries are lower. The contagion e ect is another issue to be seriously considered in the empirical work. It is considered one of main factors in emerging market crises, especially in the Asian Crises of the late 1990s and a series of defaults in Latin American countries in early 2000s. To control this issue, I construct spreads of regional EMBIG+ based on geographical locations of Asia, Latin America, Europe, and others. For each country, the regional spreads are calculated as the average of the spreads of other countries in the same region. Following the logic of Longsta et al. (2007), the changes of these regional spreads are regressed on the other explanatory variables and the residual is used as an explanatory variable. It should be noted that empirical work suggested here is di erent from Longsta et al. (2007) since they make a regression based on country by country basis while I explore the regression for the panel that covers all the countries at the same time. Furthermore they included regional and global variables to check the contagion factor by constructing global spreads as the same way. However, when the two variables are included in the panel model, serious collinearity problem arise. When both indices are included in the regression model, the coe cients for regional spreads show opposite signs. So I chose regional spread as the only explanatory variables for representing contagion impact on risk premia. Considering suggested controlling issues, the regression model is as follows: RPi;t = 1 + 21(FX = Fixed) + 31(FX = Crawl) + 41(FX = Free Falling) + 51(FX = Dual Rate) + 0X + i;t 45 In the equation, the dependent variable denoted by RPi;t is risk premium of country i at time t, and there are four dummy variables for exchange rate regime. X in the righted haned side of mean equation means the set of control variables, and i;t is an error term. The reader may nd out from Table 1.1 that the coe cient of risk premium is negative under the xed regime after considering control variables which is explained in detail below. 10 From Table 1.1, it should be noted that the coe cient of risk premium under the xed regime is signi cant on both level and di erence regressions. Second, the coe cient for crawling regime do not show signi cance under the log di erence regression model. Finally, under the unstable regimes such as free falling and dual rates, the regression results do not show consistency on the sign of the coe cients. The reader can nd the regression results in Table ?? for this equation. The most notable point that should be mentioned is that the risk premium is lower under xed regime than exible regime. This also applies to Crawling regimes. However, for other regimes such as free falling and dual rates risk premia are bigger than exible regime. It is also worth mentioning that volatility also impact the risk premium. This result is interesting considering I already include regional risk spread to control the contagion e ect. This can be treated as a market in uence that is well known for the nancial market data. Local variables related to external debts show signi cance and correct signs. Finally, there may exist some factors that a ect on both risk premium and exchange regimes. To control this, the e ciency of government index from the World Bank is included in the model and shows that as 10The full results of the regressions can be found in the appendix. 46 a government is more e cient, risk premium is lower. There are some observances that during the crisis periods the behaviors of economic variables are di erent from tranquil times. For instance, one might expect that risk premium is much volatile during the crisis periods so that this "irregular" movements of risk premium may a ect to the result of this kind of regression. In order to explore this possibility, I run the regression by dropping out some crisis episodes. The result is presented in Table ??. To be short, the coe cients for exchange rate regime dummies, especially for xed regime, does not show big dif- ference. Moreover, the magnitude of coe cient is generally bigger when the crisis episodes are dropped, and the biggest when the banking crisis episodes are excluded. To check the consistency of this result, I change the risk premium from the EMBIG+ index to credit default swap for smaller size of countries and time periods. This data is obtained from the Bloomberg terminal, which provides the CDS data for 22 countries from the years 2000 to 2007. After the change of dependent variables, the regression results for the xed and exible regimes do not change much. From the independent variables, exchange rates are most important in this regression. Therefore, I change the real exchange rate into real e ective exchange rate to verify the consistency, and the results do not show signi cant di erence. In addition, there may be side e ects from crises periods since the risk premia and exchange rate changes drastically. I included a dummy variable for the crisis periods as an explanatory variable, and conclude that there are little changes in the regression results. 47 1.6 Conclusion It is widely accepted that the exible exchange regime is preferred to the xed regime against real shocks because the former insulate shock from real economy by quickly adjusting relative price level. Previous work concentrates on the issue of trade sector, such as slower import price transfer or heavily consuming foreign goods. I analyze this problem with the friction in the wage structure where there are two types of wages; a conventional wage available to the current period of con- sumption and a deferred wage that is paid at the end of the period. When a deferred wage shock occurs such that share of conventional wage decreases and that of de- ferred wage increases, the real exchange rate and capital used for the next period production is higher under the exible exchange regime. Since the production in the current period can be de ned as a negative function of real exchange rate, higher increase of real exchange rate leads into lower production in the period when a pos- itive deferred wage shock occurs under the exible exchange regime. Even though the production at the next period is higher under the exible exchange regime, that does not cover initial loss of welfare at the current period. As a result, the xed regime is preferred to the exible regime thanks to lower volatility in consumption. In addition to facing sharp drop of production at the current period under the ex- ible regime as well as higher level of capital for the next period?s production, the remaining wealth that will be used for the next period of capital production is fur- ther reduced. The reduce of remaining wealth, increase of real exchange rate, and a surge of capital for the next period lead into the increase of leverage ratio, which is 48 de ned by value of money for capital production to own capital. Therefore, the risk premium under the exible regime is higher. When I replace a deferred wage shock with other real shocks, such as technology shock and world interest rate shock, still the risk premium under the exible regime is higher than under the xed regime. The addition of the asset holders do not change these results with the assumption of exogenous segmented asset market. There are some points that should be investigated further. Even though it was possible to distinguish among the exchange rate regimes in terms of the response ratio, the ratio itself is relatively small, compared with empirical data suggested. In addition, the di erence between xed regime and exible regime with in ation target policy is small. Other factors such as openness of market can be a potential candidate for widening this response, which will be the issue of future research. 49 Chapter 2 The Impact of Macroeconomic Announcements on Real Time Foreign Exchange Rate in Emerging Markets 2.1 Introduction Information transmission across foreign exchange markets has become a widely studied topic in the academic literature.1 One strand of this literature focuses on the impact of macroeconomic data announcements on foreign exchange markets. An- dersen et al. (2003) (ABDV (2003) hereafter) nds that news about macroeconomic fundamentals a ect both conditional mean returns and volatilities of exchange rates for major currencies. Some other recent papers in this vein include Andersen et al. (2007), Dominguez and Panthaki (2006), Ehrmann and Fratzscher (2005), Fair (2003), Chaboud et al. (2004), Laakkonen (2004), and Faust et al.(2007). Evans and Lyons (2008) connects the impact of news in the FX market to order ows. Most existing studies have however been limited to major currencies exchange rates. The price discovery process and the information transmission mechanism in emerging economy foreign exchange markets have not yet been well understood. 1This chapter is co-authored by Fang Cai at Federal Reserve Board, Zhiwei Zhang at Nomura International. Fang Cai is at the Division of International Finance of the Federal Reserve Board, Washington DC, 20551, and can be reached at fang.cai@frb.gov or (202) 452-3540. Zhiwei Zhang can be reached at zhiwei.zhang@nomura.com or 852-2536-1111. 50 This paper is the rst to focus on how U.S. and domestic macroeconomic announcements a ect exchange rates in nine emerging markets: Czech Republic, Hungary, Indonesia, Korea, Mexico, Poland, South Africa, Thailand, and Turkey.2 We construct a unique database that covers high frequency exchange rates for the nine emerging market economies from January 2, 2000 to December 31, 2006. The database is complemented by information from Consensus Forecast on market expec- tations for these exchange rates, and data from Bloomberg on market expectations for macroeconomic news and the actual announcement. Although similar databases have been studied for major currencies, this is the rst time such data for emerging markets are utilized for economic research. We try to address the following questions in this paper: (i) what macro news announcements move emerging market exchange rates? (ii) did emerging market currencies become more sensitive to news as government controls of foreign exchange (FX) markets have reportedly weakened in some of these countries? (iii) how does market sentiment a ect the way emerging market currencies respond to news? and (iv) does uncertainty in the FX market a ect how these currencies react to news? We nd that the answer to the rst question depends on whether the news is about the U.S. or the emerging economies and varies across countries. Domestic macro news in emerging markets generally do not have signi cant e ect on exchange rates, with the notable exception for Czech Republic. The set of U.S. macro news that moves major currencies signi cantly turns out to a ect 6 out of 9 emerging 2Galati (2000) examines the relationship between trading volumes, volatility and bid-ask spreads in foreign exchange markets in 7 emerging economies, but does not measure the impact of news. 51 market currencies in the same direction in the sample. For the other three currencies, Mexican Peso also reacts to U.S. news signi cantly but almost always in the opposite direction, while the Thai baht and the Turkish lira rarely respond to U.S. news signi cantly. We nd evidence that exchange rates in emerging markets have become more sensitive to U.S. news in recent years, probably due to loosened government controls of the FX markets in some of these countries. This pattern is clear for most exchange rates in the sample except for Thailand, where the Thai bahts lack of reaction to news is persistent through out the sample. The other two Asian currencies, the Korean won and the Indonesian rupiah, used to be irresponsive foreign and domestic macro news in the early part of the sample, but became more in uenced by U.S. news in recent years. Do macro news have more e ect on emerging market currencies when market sentiment is strong, i.e., investors expect the currencies to move substantially in one direction? The answer is yes. We nd strong evidence across country and macro news that market reaction is reinforced by investors conviction on the direction of the emerging market currencies. The magnitude of this reinforcement e ect is large. For instance, when investors expect the Indonesian rupiah to appreciate by 5 percent, the e ect of news on non-farm payroll in the U.S. on the Indonesian rupiah became twice as much as when investors expect the Korean won to stay unchanged. Does market uncertainty amplify or dampen the impact of news on exchange rates? The answer is ambiguous. While regressions show that market uncertainty dampens more news than it ampli es, the evidence is not overwhelmingly one-sided. 52 In some special cases, the e ect of uncertainty on the same news di ers across countries. Further analysis on this issue is necessary. Our paper complements other studies on the impact of news on asset prices in emerging markets. Wongswan (2006) provides evidence of transmission of in- formation from the U.S. and Japan to Korean and Thai equity markets. Using high-frequency intraday data, he nds a large and signi cant association between emerging-economy equity volatility and trading volume and developed-economy macroeconomic announcements at short time horizons. Andritzky, Bannister, and Tamirisa (2007) examine how emerging market bonds react to macroeconomic an- nouncements and nd that global bond spreads respond to rating actions and changes in U.S. interest rates rather than domestic data and policy announcements. Consistent with their studies, we nd a signi cant impact of major U.S. macroeco- nomic news on emerging market currencies using high-frequency data. Compared with their papers, the innovations of our work are: (a) the longer sample of our data makes it possible to track the evolution of reactions to news in the emerging currency markets, and (b) the reaction of exchange rates to news is linked to market sentiment and uncertainty. The rest of the paper proceeds as follows. Section 2 provides the description of the data. Section 3 presents the econometric speci cations and the estimates of how news surprises a ect exchange rate returns and volatility in the nine emerging markets. Section 4 shows how market forecasts and uncertainty interact with macro news and a ect exchange rates in emerging markets. Section 5 concludes. 53 2.2 Data 2.2.1 Exchange Rate Data The paper uses high-frequency exchange rate data for nine emerging markets, drawn from Olsen Financial Technologies. The data report exchange rates of the nine EM currencies versus U.S. dollar at 5-minute intervals. The full sample period is from January 2, 2000 to December 31, 2006, covering 2,557 days of bid-ask prices for each currency with two exceptions.3 It should be noted that the dataset has quite many missing values, in particular for the earlier years. The number of non- missing values for bid and ask price of each countrys exchange rate is presented in Table A.10.4 Using bid-ask price quotes from the raw exchange rate data, we take the simple arithmetic average to get the middle price quote. Then we calculate 5- minute currency returns by taking log di erences. We multiply the log di erences of currency returns by 100 to obtain log currency returns. Following ABDV (2003), we exclude data on weekends and national holidays, since the quoted prices may have some bias based on low transaction volumes. First, we drop the period from 3For KRW/USD, January 2004 (31 days) data are not included. For TRY/USD, the sample period is from January 2, 2001 to December 31, 2006. The number of observations of high frequency FX data is 2,557 * 288 = 736,416 for Czech Republic, Hungary, Indonesia, Mexico, Poland, South Africa, and Thailand, 727,488 for Korea, and 631,008 for Turkey. 4We also estimate the same model with fully lled data using interpolation for missing values. The estimates with interpolated data show similar but a little bit weaker results compared with the results presented in this paper. 54 Friday 21:05 to Sunday 21:00 (local time) for weekends. Second, we drop national holidays in the U.S. and the nine emerging markets. In order to check how the series of currency returns vary over time, we plot autocorrelations of the currency returns and its absolute value in Figures A.17 and A.18, respectively. The general pattern of the two gures is similar with what pre- vious studies show for major currencies: the autocorrelations of currency returns are statistically signi cant in the short term, and decay fast; the autocorrelations of absolute returns are statistically signi cant in the short term and stay high persis- tently. In addition, it should be taken into account that the exchange rate regimes in some emerging markets (such as Hungary, Poland, and Turkey), might have changed within the sample period (Table A.11). The Hungarian foreign exchange regime changed from a crawling peg to a pegged exchange rate within horizontal band in October 2001, and devalued on June 2003. For the Polish zloty, a crawling peg based on 55% of Euro and 45% of dollar changed into independent oating on April 2000. For the Turkish lira, many changes happened during the sample period due to the nancial crisis in 2001. The regime changed from a crawling peg to independent oating on February 2001, and the New Turkish lira was introduced on 2005 and became a sole legal tender from January 1, 2006 with a conversion rate of YTL 1 = TL 1 million. We convert all previous TL quotes into YTL based on the conversion rate when calculating log returns of its exchange rate. 55 2.2.2 Actual and Predicted Economic Variables We use economic forecast data from Bloomberg on various actual and predicted economic indices in the U.S. and nine emerging market countries.5 Many economists and analysts in the nancial markets who use Bloomberg submit their own forecasts to Bloomberg. However since such forecast data submission is voluntary, the number of the observations varies for each observation of economic index. For instance, 39 people submitted forecasts of the initial jobless claims in the U.S. that is published by the Department of Labor on January 5, 2008. Based on those forecasts, Bloomberg provides mean, median, maximum, and minimum values for each economic index. In some cases (mostly in emerging markets) the forecasts are based on the views from a small number of economists. We drop all forecasts that are based on views from fewer than 5 economists. In Table A.3, we present the number of the observations for each variable used in the empirical analysis. There are 26 indices for the U.S. news, 12 for Hungary, 11 for Mexico and Poland, 9 for Turkey, 6 for South Africa, 5 for Korea and Thailand, and 4 for Indonesia. Since the unit of each economic index is di erent, we standardize the time series of each economic index by calculating the surprise as (actual number - forecasts) divided by its sample standard deviation Skt = Akt Fkt ^k (2.1) 5These forecasts of economic indices are easily obtained from ECO menu in the Bloomberg terminal by the country. 56 where Akt is the actual announced value for economic index k at time t, Fkt is the mean of forecasts, and ^k is sample standard deviation of Akt Fkt. 2.2.3 Foreign Exchange Forecasts To measure market expectations on exchange rates, we use forecast data from Consensus Forecasts, which provides a simple arithmetic average of the forecasts for foreign exchange rates over 90 countries as well as major economic indices on a monthly or bimonthly basis.6 The exact date when the survey is conducted is shown in the published data. We collect information on the survey date, spot rate on the survey date, sample average, maximum, minimum, and standard deviation of each exchange rate for the following 1, 3, 12 and 24 months. A variable is constructed to measure which direction the market anticipates the exchange rates move, FXDi;d;t = CFXi;d;t SFXi;t SFXi;t (2.2) where CFXi;d;t is consensus forecast for country is exchange rate at day t for the next d months, and SFXi;t is the spot exchange rate on day t. If FXDi;d;t is positive, then market participants expect that local currency i will depreciate for next d months, and vice versa. In Appendix 3a, we provide summary statistics for FXD. 6Monthly forecasts for Asian economies are available for the full sample. For Latin American economies, monthly forecasts are available after May 2001, and bimonthly forecasts are available before. For other economies, only bimonthly forecasts are available. 57 2.3 Announcements and FX responses 2.3.1 Contemporaneous E ect from OLS Regression We start by running an OLS regression Ri;t = i;kSi;k;t + i;t (2.3) where Ri;t denotes 5-minute exchange rate returns from time t to time t+1 in country i, Si;k;t is the surprise of macroeconomic news k at time t in country i. The estimates are based on only those observations (Ri;t; Si;k;t) such that an announce- ment was made at time t. This speci cation has the advantage of simplicity. The drawback is that it does not control for the potential dynamic feature of exchange rates and news, and does not correct for heteroskedastic disturbances in the error terms. We will move to a more sophisticated model in the next subsection that addresses these issues. Table A.6 shows the estimates from these regressions. For comparison, we also examine the impact of U.S. news on the euro/dollar exchange rate. Three features stand out. First, exchange rates for South Africa and emerging markets in Europe react to many U.S. news in a similar way as major currencies do (as documented in previous literature), but many of the same news have little e ect on currencies in Asia and Turkey. Second, most domestic macroeconomic news have no impact on EM exchange rates. Finally, the euro exchange rate responds to major U.S. news in a similar way to European emerging market currencies. We elaborate more on 58 these ndings before moving to the more sophisticated speci cation. In the case of U.S. news, positive surprises on consumer con dence, durable goods order, GDP, non-farm payroll, retail sales and trade balance lead to appre- ciation of the U.S. dollar and depreciation of EM currencies in Czech Republic, Hungary, Poland, and South Africa, with a few exceptions. This set of news is also found to be signi cant in ABDV (2003). New home sales turns out to be highly signi cant for emerging markets in Europe, re ecting the importance of the U.S. housing sector for the sample we study. On the other hand, very few U.S. news have signi cant impact on the Mexican, Turkish and Asian currencies. Of the 26 U.S. news we studied, only 3 show up signi cantly for Korea, 6 for Indonesia, 1 for Thailand, 3 for Mexico, and 2 for Turkey. In contrast with the large number of signi cant U.S. news, few domestic news in emerging markets have a signi cant impact on their exchange rates. For Indone- sia, Thailand, and Turkey, no domestic news are signi cant in the regressions. Even for Hungary and Poland where many U.S. news move exchange rates signi cantly, only one domestic news is signi cant in each country. Of the 14 cases where domes- tic news announcements move the exchange rates, 9 cases are related to domestic growth or external balance: the current account in Czech Republic and Poland, GDP in Czech Republic and Mexico, industrial production in Hungary, and the trade balance in Czech Republic, Mexico, and South Africa. Given the long sample of the dataset, we can examine if exchange rates in emerging markets have become more sensitive to news in recent years. We estimate equation (2.3) using a two-year rolling window, and plot the point estimates of 59 i;k over time. Charts in Figure 3 plot the signi cant estimates from such rolling regressions. Two patterns stand out. First, most EM currencies have become more sensitive to news in recent years than before. For instance, in Korea, few U.S. news had signi cant impact on the won before late 2002, while 4 out of 9 news are persistently signi cant in recent years. Thailand is an exception, where Thai baht barely reacts to any U.S. news throughout the whole sample. Second, the fact that some news do not a ect certain currencies cannot be explained by the lack of observations. In the later part of the sample, the numbers of observations for given U.S. news are fairly equal across countries. Yet, some currencies persistently react to news, while others seem to be irresponsive. 2.3.2 Contemporaneous E ect from Dynamic Regressions with Het- eroskedasticity We follow ABDV (2003) in their econometric speci cations to include lag terms of currency returns and news, and control for heteroskedestic errors. First, we estimate a linear regression model based on I lags of 5-minute returns, and J lags for all the news surprises. We choose the lags I = 5 and J = 2 according to the Akaike Information Criteria and Schwartz Criteria.7 The number of news surprises in the model is di erent for each country since that of the domestic news surprises is di erent. 7The exact AIC and BIC show di erent optimal number of lags across the countries. However, 6 of the sample countries showed that 5 lags of FX returns are good enough, whereas the other countries showed relatively small lags for FX returns. 60 Rt = 0 + IX i=1 iRt i + KX k=1 JX j=0 kjSk;t j + t t = 1; : : : ; T: (2.4) j ^tj = c+ ^d(t)p 288 + KX k=1 J 0X j0=0 kj0 jSk;t j0 j+ QX q=1 q cos 2q t 288 + q sin 2q t 288 ! + t (2.5) As in ABDV (2003), the absolute value of the residual from equation (2.4) is modeled as the sum of three terms: daily volatility forecast to measure average volatility level during the day; the absolute value of news surprise including lags to assess the impact from the news; and the Fourier exible form with trigonometric terms for the calendar e ect. Equations (2.4) and (2.5) are estimated by 2-stage WLS. First, we run an OLS regression with equation (2.4). Then we estimate equation (2.5) to get a linear prediction of the absolute value of the residuals in equation (2.4). Finally, using the linear prediction from equation (2.5) as a weight, we perform a weighted least-squares estimation of equation (2.4). It is necessary to be more speci c on the independent variables used in equation (2.5). The daily level of volatility in the second term is based on the residual from the regression of GARCH (1,1) model using daily spot exchange rate returns from January 1, 1993 as described above in the data description. GARCH (1,1) models are generally used to extract predictions in high-frequency nancial data in a wide variety of papers. The third term represents the impact of news surprise on the volatility. In order to enhance tractability, we impose a polynomial speci cation on the response 61 patterns associated with kj0 , as in ABDV (2003). This ensures that the response patterns related to the news surprise are determined by the restriction we provide on the speci cations. Consider the general form of polynomials, p( ) = c0 + c1 + + cp p, for = 0; 1; : : : ; J . The restrictions we apply to this equation are J = 12, p = 3, and p(J) = 0. As a result, we have p( ) = c0 [1 ( =12)3] + c1 [1 ( =12)2] + c2 2 [1 ( =12)]. Using this equation, we estimate three coe cients for each FX returns and each news surprise, and plug the xed value from the estimation into the disturbance equation (2.5). j ^tj = c+ ^d(t)p 288 + KX k=1 J 0X j0=0 k " c^0 1 j0 12 3 ! + c^1j 0 1 j0 12 2 ! + c^2(j 0)2 1 j0 12 # jSk;t j0 j (2.6) + QX q=1 q cos 2q t 288 + q sin 2q t 288 ! + t The fourth term of Fourier series covers calendar e ects in the model. AIC and Schwartz criteria suggest that Q = 4 is appropriate for the model, and it means that the seasonal pattern of intra-day trading quote is relatively smooth. Table A.5 presents the estimates for a selected group of U.S. news. Compared with Table A.4, emerging market exchange rates react to U.S. news more consistently across countries. Currencies in Thailand and Turkey remain rather insensitive to most U.S. news. For the other 7 countries, all of the 9 major U.S. news have signi cant signs in the expected direction, with few exceptions. As in the OLS regressions, the Mexican pesos reaction to U.S. news remain mostly the opposite of those of other currencies. 62 The dynamic structure of this model allows us to estimate the persistence of news e ects on exchange rates. The lagged variables of U.S. news surprises mostly show the same sign as the contemporaneous variables. There are some exceptions for news such as Nonfarm Payroll and Producer Prices, which show mean reversion e ects across the time. However, the size of impact seems to decay as time goes by. A complete table with all U.S. and domestic news is provided in Table A.14. Among domestic news surprises, the consumer price index and current account bal- ance show signi cance for the contemporary FX impact across the countries. The trade balance and producer price also seem to be signi cant when lagged variables are considered. Major domestic macroeconomic news surprises in Eastern Euro- pean countries also have a signi cant impact on their exchange rate returns. For the Czech Republic, the budget de cit, current account, consumer price index, ex- ports, imports, industrial production index, producer price index, retail sales index, and trade balance are all signi cant in the model. The current account, consumer price index, and industrial production show signi cance in Hungary. And in Poland, the signi cant news surprises include current account, GDP, money supply, unem- ployment, and wholesale sales index. Along with European countries, exchange rate returns in South Africa are strongly responsive to domestic news surprises. Among the 6 domestic macroeconomic announcements we collect, the consumer price index, interest rate, money supply, and trade balance are all statistically signi cant. In Asian countries, nevertheless, the impact of domestic news surprises on exchange rate returns are somewhat smaller compared with that of the U.S. news surprises. Only one of the domestic news surprises in Thailand is signi cant in the estimation 63 model. None of the domestic news is signi cant in Indonesia and Korea. 2.3.3 Announcements and FX Volatility In order to assess how the news surprises a ect FX volatility, we compare contemporaneous coe cients with the sum of those across 12 lags (i.e., 60 minutes of time) used in the regression model suggested in equation (2.5). In this case, we concentrate on the 9 news surprises that are statistically signi cant for at least 6 countries in the current terms or more than 13 including additional 2 lags in equation (2.4). It should be noted that we use equation (2.5) for the estimation, so the impact of the news surprise on the volatility should last until the next 60 minutes.8 Results presented in the middle section of Table 3 suggest that several of the coe cients for news surprises in the volatility equation have statistical signi cance, although they tend to be smaller compared with the contemporaneous return response coe cients in the top panel. Only 7 of the coe cients for 7 countries excluding Thailand and Turkey are insigni cant. Comparing the signi cance of coe cients in the conditional mean equation (2.4) with those of volatility equation (2.5), it can be seen that the news surprises provide more impact on volatility than on conditional mean of ex- change rate. To summarize, 87.5% of 9 major economic news surprises in 9 countries which are statistically signi cant have a more prolonged impact on volatility for 60 minutes. The whole set of coe cients including contemporaneous and cumulated 8We can extend the time period for this estimation by assigning a bigger number for the time lag J than 12, however this may introduce other sources of volatility within the period. 64 coe cients are presented in Table A.14. As shown in the bottom panel of Table A.5, the cumulative response of volatil- ity is much larger than the contemporaneous volatility response, which is consistent with ABDV (2003)s nding that volatility adjusts to news surprises gradually. An alternative possibility is that the announcement itself can in uence on FX market rather than the size of the news surprise. To check for this possibility, we include dummy variables that represent the announcement in both equation (2.4) and (2.5) such that the lags should be the same as news surprise. Then the equation model changes as follows: Rt = 0 + IX i=1 iRt i+ KX k=1 JX j=0 kjSk;t j + KX k=1 JX j=0 kjDk;t j + t t = 1; : : : ; T: (2.7) j ^tj = c+ ^d(t)p 288 + KX k=1 J 0X j0=0 kj0 jSk;t j0 j+ KX k=1 JX j=0 kjDk;t j+ QX q=1 q cos 2q t 288 + q sin 2q t 288 ! + t (2.8) As before, we present major 9 economic indices that show signi cant impact on FX markets across the countries in Table A.6. The set of all coe cients can also be found in Table A.6. In Table A.6, many major economic indicators seem to have an announcement e ect on FX changes even after taking into account the news surprise impact. Furthermore, the announcement e ects exist not only for FX returns but also for the volatility. 65 2.3.4 Testing for Asymmetry We test if there is any asymmetry in the impact of the news surprises according to the sign. ABDV (2003) reports asymmetric response of US news in the case of major currencies. The long sample and the large number of currencies in our sample provide a good opportunity to check if such patterns also exist in emerging markets. First, we divide news surprises into two groups based on their signs, and estimating two equations below: Rt = 8 >>< >>: 0kSkt + 1kS2kt + t if St 0 2kSkt + 3kS2kt + t if St > 0 (2.9) With this estimation, we reconstruct the set of graphs that contain the t- ted value on the vertical axis and the standard deviation of the news surprise in horizontal axis in Figure A.21 (using the average impact over all news surprises). There appear to be some di erences between the two subgroups in our sample. To investigate this more formally, we try a modi ed equation to test if there is any asymmetry across the sign of the news surprise. Rt = 0kSkt + 1kS 2 kt +Dkt 2kSkt + 3kS 2 kt + t (2.10) where Dkt denotes a dummy variable which takes the value 1 if the news surprise is positive, and the value of 0 if negative. To test for asymmetry, we de ne the null hypothesis such that FX returns have symmetry ( 2k = 0; and 3k = 0) for major 9 economic indicators. The results of the test are presented below in Table A.7. Only 9 cases suggest that the symmetry hypothesis is rejected at 5% 66 signi cance, while 72 other cases cannot reject the symmetric null hypothesis. This symmetric impact of news surprises on FX returns is in contrast with the ndings in ABDV (2003). To look into the source of this di erence, we repeat the regression above for euro. As it turns out, the euro responds to most U.S. news in a symmetric way as well in our sample, suggesting that the di erent ndings between ours and ABDV (2003) come from the di erent sample periods rather than di erences between emerging market currencies and major currencies. 2.4 Market Sentiment, Uncertainty, and Macroeconomic News In this section, we examine the interaction between market sentiment on emerging market currencies and the exchange rate response to news surprises. For instance, if market participants expect that Korean won will depreciate in a near future as a consensus, then the news surprise that suggests the U.S. economy be stronger than expected may have a greater impact on returns of Korean won by making this currency depreciating more rapidly, and vice versa. Therefore, this case consists of two di erent expectational errors from market participants: a rst error from news expectations, and a second error consisting of an FX forecast error. On the other hand, if we can think that the expectation of future appreciation or depreciation is related to the economic cycle in a country, then this approach may become the alternative way to assess symmetry in the impact described in the above section. We use the median value of 1-month-ahead FX forecasts from Consensus Forecasts as a proxy of market expectation of each currency. 67 We use an ordinary least square regression with some modi cation in equation (2.4), by adding an FX consensus variable multiplied by news surprises. If the hypothesis described above is true, then the coe cients on the interaction variable will be positive. The modi ed equation is as follows: Rt = 0+ IX i=1 iRt i+ KX k=1 JX j=0 kjSk;t j+ KX k=1 JX j=0 kjFXDd;t jSk;t j+ t t = 1; : : : ; T: (2.11) In Table A.8, we focus on 9 U.S. major economic indices discussed earlier. All the coe cients for variables used in this regression are presented in Appendix 6. The rst part of the table presents coe cients for news surprises only, and the second part for FX forecasts (FXD) multiplied by the news surprises. Notably, many of the FX forecast-related coe cients show statistically signi cant and positive values, suggesting that market sentiment plays an important role in how news surprises move EM currencies. It acts as an ampli er when the market is pessimistic (opti- mistic) about the EM currencies and news surprises suggest stronger (weaker) U.S. economy. For instance, if market analysts think that the Czech Republic koruna will depreciate (appreciate) by 10% for next d months and the durable good orders data is 1 standard deviation higher (lower) than expected, then exchange rate re- turns will depreciate (appreciate) 2.2 basis points more than when no exchange rate change is expected for next d months.9 On the other hand, when the EM currencies are under pressure to appreciate, positive sentiment for these currencies works as 9Since we multiply log di erence of FX by 100 to increase the scale of coe cients, we need to divide by 100 again, so that the magnitude of the shock can be measured correctly. 68 a shock absorber against strong U.S. news. This evidence is consistent with the ndings of Mian and Sankaraguruswamy (2008) that the stock market response to good (bad) news is greater during a high (low) sentiment period. One explanation for these results is investor overcon dence as documented in Barberis and Thaler (2003) and Hirshleifer (2001), i.e., investor are more likely to accept news that is in line with their prior beliefs and ignore information that is contradictory to their prior beliefs. The accelerator e ect of market sentiment provides a potential explanation why we nd no evidence for asymmetry in EM currencies reaction to news as in ABDV (2003). EM currencies experience more ups-and-downs than major currency pairs. The long sample of our dataset contains both periods of market optimism and pessimism for each EM currency. Over the market sentiment cycle, this asymmetry might be averaged out. In contrast, ABDV (2003)s sample period covers one side of the business cycle, when market sentiment might be persistently one-sided as well. We further test the e ect of uncertainty on exchange rate response to news. The speci cation is Rt = 0+ IX i=1 iRt i+ KX k=1 JX j=0 kjSk;t j+ KX k=1 kDISPd;tSk;t+ t t = 1; : : : ; T: (2.12) where DISP is a measure of market uncertainty de ned by the dispersion of market forecasts for each EM currency. It is constructed as DISPd;t = j CFXhighd;t CFX low d;t SFXt j, where CFXhighd;t denotes the maximum of FX forecasts at time t, and CFX low d;t denotes the minimum of FX forecasts at time t (the summary statistics for DISP is presented 69 in Table A.13). The role of market uncertainty in these regressions is not conclusive. The estimates are shown in Table A.9. Despite many signi cant estimates, the signs of the parameters for market uncertainty vary across country and across news. The diverse set of parameters leaves the regressions inconclusive. Nonetheless, the fact that market uncertainty shows signi cance in many regressions indicates it does have in uence on how exchange rates react to news, but the channel of such in uence is not yet well understood. 2.5 Conclusion This paper documents some interesting features in the FX market for emerg- ing market currencies. Except for Thailand and Turkey, whose currencies are not sensitive to news, the other 7 currencies show consistent reactions to news. First, U.S. news matters much more than domestic news. Second, currencies have become more sensitive in recent years than before. Third, market sentiment on these cur- rencies plays an important role by swaying the impact of news surprises, i.e., good (bad) news matters more when optimism (pessimism) prevails. These nding are robust across countries and news we studied. The role of uncertainty in FX market is also studied but is not fully ex- plored. The signi cant yet inconclusive estimates indicate that its role could be state-dependent, and we are not yet able to capture what is the missing state vari- able. On the role of market sentiment, although we found signi cant and consistent 70 results for emerging markets, it is not clear if this is a unique phenomenon for emerging market currencies, or it also exists for major currencies and other nancial assets. These are potential topics for future research. 71 Appendix A Appendix A.1 Data Description and Source The list of 34 countries that are included in the regression (for EMBIG+) is as follows: Algeria, Argentina, Brazil, Bulgaria, Chile, China, Colombia, Cot?e di Voire, Croatia, Dominican Republic, Egypt, El Salvador, Greece, Hungary, Ja- maica, Korea, Malaysia, Mexico, Morocco, Pakistan, Panama, Peru, Phillippines, Poland, Russia, Serbia, South Africa, Thailand, Tunisia, Turkey, Uruguay, Urkraine, Venezuela, Vietnam. Foreign Exchange Rate: Monthly, End of period. From IFS. For consistence check, Real E ective Exchange Rate (REER) from OECD and BIS are used. Rating : Moody?s ratings, Long term dollar denominated bond (Government issued). From Bloomberg. Stock index: Local index for each country. From Bloomberg. For some coun- tries (Algeria, Cote di Voire, and El salvador) where local stock market index is not available, the regional MSCI Index is used. Reserves: From IFS US Treasury Bond 5 year maturity: From BEA Regional/Global EMBIG (CDS): In the case of the region, I divide into 4 (Asia, Latin America, Europe, and Others). To construct regional EMBIG, the 72 simple average of countries? EMBIG (CDS) in the region is used except the applied country itself. For Global EMBIG (CDS), all the EMBIG (CDS) available excluding the applied region are used to make a simple mean. CDS from Bloomberg and EMBIG from J.P. Morgan. S&P 100 Index PER: Price to Earning ratio of S&P 100 index. From Bloomberg. Corporate Yield Spread in the U.S.: basis point spread between AAA and BBB- industrial bonds yields for investment grade, between BBB- and BB- for high yield bonds. Data from Bloomberg. Term Premium: Based on Cochrane and Piazessi (2005), expected excess re- turn on US treasury bonds can be estimated from linear function of forward rates with 1 to 5 year maturities. I reconstruct predicted excess return on 5 year maturity US Treasury bond. External Debt, Short Term Debt: From World Bank Economic Policy and External Debt. For Hungary and Korea, Deutsche Bank estimates are used due to data availability. A.2 Contract between capitalist and investors Contracting Problem between Capitalist j and foreign lender j?s net worth: PtN j t dollar interest rate: 1 + r prices in period t are known 73 the period t+1 rental rate on capital in dollars Rt+1=St+1 is known Choice variable investment Kjt+1 dollar loan Djt+1 repayment Bt+1 Kjt+1 yields ! j t+1K j t+1 Rt+1 St+1 next period. !jt+1 : random shock, iid across j and t with Etf! j t+1g = 1, cannot be observed by lenders. Monitoring cost !jt+1K j t+1 Rt+1 St+1 . ! is such that Bt+1 = !K j t+1 Rt+1 St+1 . Payo s for lenders and borrower. if !j !j, 8 >>>< >>>: lender: !jKjt+1 Rt+1 St+1 borrower: !jt+1K j t+1 Rt+1 St+1 !jKjt+1 Rt+1 St+1 if !j < !j, 8 >>>< >>>: lender: (1 )!jt+1K j t+1 Rt+1 St+1 borrower: 0 default Risk neutral lender: expected return should be 1 + r. Kjt+1 Rt+1 St+1 ! (1 H( !)) + (1 ) Z ! 0 !jt+1dH(! j t+1) = (1 + r)Djt+1 = (1 + r) QtK j t+1 PtN j t St (A.1) Capitalist?s Budget constraint PtNt + StDt+1 = QtKt+1 (A.2) 74 Capitalist?s Utility Z 1 ! !jt+1dH(! j t+1) ! (1 H( !)) Rt+1 St+1 Kjt+1Qt (A.3) De ne ratio of the value of investment to net wealth. j;t = QtKt+1 PtN j t (A.4) Also, de ne risk premium. 1 + t+1 = Rt+1St QtSt (1 + r) (A.5) Then, from (A.1), j;t 1 = (1 + t+1) j;t ! (1 H( !)) + (1 ) Z ! 0 !jt+1dH(! j t+1) (A.6) Utility function (A.3) using known variables at period t is Z 1 ! !jt+1dH(! j t+1) ! (1 H( !)) j;t (1 + t+1) (A.7) Now, construct maximization problem using (A.7) and (A.6). max j;t; ! Z 1 ! !jt+1dH(! j t+1) ! (1 H( !)) j;t (1 + t+1) s.t. j;t 1 = (1 + t+1) j;t ! (1 H( !)) + (1 ) Z ! 0 !jt+1dH(! j t+1) De ne the followings for the convenience of calculation. ( !) : expected gross share of pro ts going to the lender ( !) = Z ! 0 !jt+1dH(! j t+1) + ! Z 1 ! dH(!jt+1) 1 ( !) = Z 1 ! !jt+1dH(! j t+1) ! (1 H( !)) 0( !) = 1 H( !) 00( !) = h( !) 75 G( !) : expected monitoring cost G( !) = Z 1 0 !jt+1dH(! j t+1) G0( !) = !h( !) Then, maximization problem can be rewritten. max j;t; ! (1 ( !)) j;t (1 + t+1) s.t. j;t 1 = (1 + t+1) j;t [ ( !) G( !)] First order conditions ( !): 0( !) [ 0( !) G0( !)] = 0 ( ): (1 ( !)) (1 + ) + f[ ( !) G( !)] (1 + ) 1g = 0 ( ): [ ( !) G( !)] (1 + ) + 1 = 0 Then, from Bernanke et al.(1999), it is shown that the ratio of the value of investment to net wealth, , and risk premium is is an increasing function of !. j;t = ( !) 1 + t+1 = ( !) Therefore, we can construct the relation between and risk premium. 1 + t+1 = 1 ( j;t) = 1 QtKt+1 PtNt F QtKt+1 PtNt (A.8) where F ( ) is an increasing function. 76 A.3 The signs of Major Variables in Linearization For Afix: Afix = + (1 v)( 1 + ) = + (1 v) (1 ) 1 = 1 1 [v (1 ) + (1 v ) ] Therefore, Afix < 0 is and only if (1 v ) > v (1 ) . Stronger assumption in this case is 1 v > 0, but the parameters suggested in the simulation satisfy this stronger assumption without any problem. For CIS: CIS = v (1 v) 1 (1 ) ((1 v) 2 (1 )) = v (1 v) (1 ) (1 ) ((1 v) 1) CIS < 0 if and only if ((1 v) 1) > 0. Also, the parameters used in the simulation satisfy this inequality. For fix: 77 fix = ( 1 + 2)(1 ) (1 1 2) 2Afix = Afix + Afix ( 1 + 2)(1 ) (1 1 2) 2Afix = v + (1 v ) ( 1 + 2)(1 ) (1 1 2) + (1 2)Afix = v (1 1 2) v + (1 2)Afix < 0 For flex: flex = ( 1 + 2)(1 ) (1 1 2) 2Aflex = Aflex + Aflex ( 1 + 2)(1 ) (1 1 2) 2Aflex = v + (1 ) ( 1 + 2)(1 ) (1 1 2) + (1 2)Aflex = v (1 1 2) + (1 2)Aflex < 0 For CBP : CBP = [1 1 2 (1 v)] v 1 v (1 ) (1 ) + 2v = [1 1 2 (1 v) + (1 v) (1 )] 2v (1 v) (1 ) = [1 (1 v) ] 2v (1 v) (1 ) > 0 78 The rst two terms in the large bracket is negative according to the assumption used in CIS. Since the denumerator of the second term in the righted handed side is also negative, CBP should be positive. A.4 Convergence under Perfect Foresight To examine convergence to the steady state, assume that there are no stochas- tic shocks a ecting the system and that the situation is under the perfect foresight such that the expectation of the variable is the same as the variable itself. Based on equations (1.16), (1.17), and (1.81), the summarized linear equations are as follows: qt + kt+1 pt = yt+1 0 t+1 et+1 + et (A.9) (1 2(1 v)) yt = 1(qt + kt+1 pt) 2vpt + (1 1 2)et (A.10) 0t+1 0 t = [qt + kt+1 pt yt] (A.11) Now, plugging equation (A.9) into (A.10) and (A.11) leads into 1 0 t+1 = 1(yt+1 et+1) (1 2(1 v))(yt et) (A.12) 0t+1 0 t = 1 1 2(1 v) 1 (yt et). (A.13) 79 By de ning zt = yt et, the dynamic system of three variables are turned into 2-variable dynamic equations, which is quite convenient to check the convergence. These two equations can be rearranged into 1 0 t+1 = 1zt+1 (1 2(1 v))zt (A.14) 0t+1 = 1 1 2(1 v) 1 zt + 0 t. (A.15) Rewriting equations (A.14) and (A.15) into a matrix form leads us into 2 6 6 4 zt+1 0t+1 3 7 7 5 = 2 6 6 4 zt 0t 3 7 7 5 (A.16) where = 2 6 6 4 1 1 f(1 2(1 v)) (1 + ) 1g 1 1 1 2(1 v) 1 1 3 7 7 5 (A.17) From this matrix, saddle path stability requires that one of the eigenvalues of should be located inside the unit circle and the other should be outside the unit circle. Using trace and determinant of the matrix , it is possible to check the sign of eigenvalues. Tr ( ) = 1 1 f(1 2(1 v)) (1 + ) 1g+ 1 > 1 (A.18) Det ( ) = 1 1 (1 2(1 v)) > 0 (A.19) 80 Since the trace and determinant of the matrix is positive, all the eigenvalues should be real and positive. In order to check if there is any eigenvalue lesser than a unit, we should have to derive the analytic solution of eigenvalue, which is denoted by . = 1 2 " 1 1 f(1 2(1 v)) (1 + ) 1g+ 1 s 1 1 f(1 2(1 v)) (1 + ) 1g+ 1 2 4 1 1 (1 2(1 v)) # Since the eigenvalue suggested above is lesser one, it should be less than one in order to satisfy the saddle path condition. That means the following condition must be satis ed: 1 1 f(1 2(1 v)) (1 + ) 1g+ 1 < 2 + s 1 1 f(1 2(1 v)) (1 + ) 1g+ 1 2 4 1 1 (1 2(1 v)) The reader may nd out with ease that it is equivalent to the following in- equality: (1 1 2(1 v)) > 0 (A.20) , which is clear since 1 1 2 > 0. In addition, it can be shown that 1 1 [(1 2(1 v)) (1 + ) 1] 1 > 0. 81 A.5 Analytical Solution for Real Exchange Rate From IS and BP curve, it is possible to derive analytical solution of real ex- change rate and capital as a function of a shock. In this section, I show the that the real exchange rate under the xed regime at the shock is less than under the exible regime. And using this result, it is veri ed that the risk premium under the xed regime is also less than under the exible regime. Using IS and BP curve presented in the main body, the real exchange rate can be presented as follows: et;j = 1(1 ) 1 CBP + CIS j 1 h (1 ) 1Bj ivt j (A.21) where j = fix; flex denotes exchange regimes. Furthermore, is positive consid- ering the fact that CBP > 0 and CIS < 0. In addition, j is negative since j < 0 and Bj < 0. Now, remind that fix flex = 2v < 0 and Bfix Bflex = 2v < 0. Then it is possible to derive the relationship between ?s. fix = fix 1 (1 ) 1 Bfix = flex 2v 1 (1 ) 1 (Bflex 2v) = flex 2v 1 + 1 (1 ) 1 Then, it is easy to nd out that fix < flex < 0. Therefore, the following inequality is satis ed: 82 0 < et;fix = fix < flex = et;flex This inequality proves that real exchange rate under the xed regime is less than under the exible regime. The next step is to prove the inequlity of risk premia. Following equation (1.63), what needs to be veri ed is Bfixet;fix Bflexet;flex < 0. Bfixet;fix Bflexet;flex = fix flex ( Bfix flex + Bflex fix) = fix flex Bfix flex + Bflex flex 2v 1 + 1(1 ) 1 = fix flex flex 2v 2v 1 + 1(1 ) 1 Bflex = fix flex 2v [ flex Bflex 1 ] The last term in big bracket in the right handed side is as follows: flex Bflex 1 = ( 1 + 2)(1 ) (1 1 2) 2Aflex [( 1 + )(1 1 2(1 v)) + 2v] 1 = ( 1 + 2)(1 ) 2Aflex + ( 1 + 2(1 v)) 1 = 1( 1 + ) + 2 ((1 v) (1 ) Aflex) = 1( 1 + ) < 0 Now, it is easier to see the veri cation by summariization using two equations above. 83 0t+1;fix 0 t+1;f lex = 1 [Bfixet;fix Bflexet;flex] = 1 fix flex 2v [ flex Bflex 1 ] = 1 fix flex 2v 1( 1 + ) < 0 Therefore, 0t+1;fix < 0 t+1;f lex. A.6 Tables 84 Table A.1: Regression During Tranquil Times Excluded Periods Based ona VARIABLES Currency Crisis Sovereign Debt Crisis-Domestic Sovereign Debt Crisis-External Banking Crisis PEG -0.22*** -0.22*** -0.20*** -0.45*** (0.00) (0.00) (0.00) (0.00) Crawling -0.08 0.01 0.03 -0.07 (0.15) (0.85) (0.52) (0.25) Free Falling 0.56*** 0.35 (0.00) (0.13) Dual Rates 0.52 0.29* 0.00 0.03 (0.14) (0.08) (.) (0.85) US Tr 5 year 0.57*** 0.30*** 0.43*** 0.40*** (0.00) (0.00) (0.00) (0.00) Regional EMBIG 0.23*** 0.25*** 0.20*** 0.26*** (0.00) (0.00) (0.00) (0.00) S&P 100 PER -0.07 -0.39*** -0.44*** -0.43** (0.67) (0.00) (0.00) (0.01) IG Spread -0.07 0.09 0.14 0.04 (0.58) (0.46) (0.23) (0.74) High Yield Spread 0.44*** 0.30*** 0.29*** 0.32*** (0.00) (0.00) (0.00) (0.00) Govt. e ciency -0.84*** -0.69*** -0.68*** -0.85*** (0.00) (0.00) (0.00) (0.00) TB/GDP 0.01 0.00 0.01 0.00 (0.26) (0.47) (0.10) (0.40) Real GDP growth 1.18** -1.18** -1.93*** -0.63 (0.02) (0.01) (0.00) (0.26) External Debt/GNI 0.35*** 0.02 -0.05 -0.06 (0.00) (0.73) (0.36) (0.39) StDebt/Res 0.02 -0.01 -0.02* -0.01 (0.13) (0.59) (0.05) (0.24) Res/External Debt -0.17*** -0.28*** -0.26*** -0.32*** (0.00) (0.00) (0.00) (0.00) Time trend -0.02*** -0.02*** -0.02*** -0.02*** (0.00) (0.00) (0.00) (0.00) Constant 5.17*** 7.85*** 8.32*** 8.42*** (0.00) (0.00) (0.00) (0.00) Observations 590 750 684 578 R-squared 0.720 0.702 0.697 0.735 *** p<0.01, ** p<0.05, * p<0.1 aThe excluded periods based on classi cations of crises are the periods of crisis 2 years. 85 Table A.2: Short History of Crises from 1997 to 2010 Crisis Country Year Currency Crisisa Turkey 1997,1998,1999,2000,2001,2008 South Africa 1998,2000,2001,2008 Argentina 2002 Brazil 1999,2001,2002,2008 Chile 2008 Colombia 1997,1998,1999,2000,2002 Dominican Republic 2002,2003 Mexico 1998,2008 Peru 1998 Uruguay 1997,2001,2002 Venezuela, Rep. Bol. 2002,2004,2010 Egypt 2001,2003 Korea 1997,2008 Malaysia 1997 Philippines 1997,2000 Thailand 1997,2000 Russia 1998,1999,2008 Hungary 1997,1999 Poland 1997,1999,2008 Sovereign Debt Crisis-Externalb Turkey 2001 Argentina 2001,2002,2003,2004,2005 Brazil 2002 Dominican Republic 2005 Peru 1997 Uruguay 2003 Venezuela, Rep. Bol. 1997,2004,2005 Russia 1997,1998,1999,2000 aAn annual depreciation versus the US dollar (or the relevant anchor currency - historically the UK pound, the French franc, or the German DM and presently the euro) of 15 percent or more. bA sovereign default is de ned as the failure to meet a principal or interest payment on the due date (or within the speci ed grace period). The episodes also include instances where rescheduled debt is ultimately extinguished in terms less favorable than the original obligation. 86 Crisis Country Year Sovereign Debt Crisis-Domestica Turkey 2001 Argentina 2001,2002,2003,2004,2005,2007,2008,2009,2010 Brazil 2002 Dominican Republic 1997,1998,1999,2000,2001 Venezuela, Rep. Bol. 1997,1998 Russia 1998,1999 Banking Crisisb Turkey 2000 Argentina 2001,2002,2003 Brazil 1997 Colombia 1998,1999 Dominican Republic 2003 Mexico 1997,1998,1999,2000 Peru 1999 Uruguay 2002 Korea 1997,1998,1999,2000,2001,2002 Malaysia 1997,1998,1999,2000,2001 Philippines 1997,1998,1999,2000,2001 Thailand 1997,1998,1999,2000,2001 Russia 1998,2008,2009 China 1997,1998,1999 Hungary 2008,2009,2010 aThe de nition given above for external debt applies. In addition, domestic debt crises have involved the freezing of bank deposits and or forcible conversion of such deposits from dollars to local currency. bA banking crisis is marked by two types of events: (1) bank runs that lead to the closure, merging or takeover by the public sector of one or more nancial institutions; and (2) if there are no runs, the closure, merging, takeover, or large-scale government assistance of an important nancial institution (or group of institutions), that marks the start of a string of similar outcomes for other nancial institutions. 87 Table A.3: U.S. and National News Announcements News Announcements Source No of Obs Start Date Final Date Timea United States 1 Business Inventoriesb US treasury 83 14-Jan-00 13-Dec-06 15:00 2 Budget De citc BEA 83 21-Jan-00 12-Dec-06 19:00 3 Current Accountd Federal reserve 27 15-Mar-00 18-Dec-06 13:30 4 Capacity Utilizatione Conference board 70 14-Jan-00 15-Dec-06 14:15 5 Consumer Con dence Federal reserve 84 25-Jan-00 28-Dec-06 15:00 6 Consumer Credit Census 84 7-Jan-00 7-Dec-06 20:00 7 Construction Spending BLS 84 4-Jan-00 1-Dec-06 15:00 8 Consumer Price Indexf Census 82 18-Feb-00 15-Dec-06 13:30 9 Durable Goods Orders Census 84 27-Jan-00 22-Dec-06 13:30 10 Factory Orders BEA 84 5-Jan-00 5-Dec-06 15:00 11 Gross Domestic Product dept of commerce 84 28-Jan-00 21-Dec-06 13:30 12 Housing Startsg BLS 83 19-Jan-00 19-Dec-06 13:30 13 Importsh Federal reserve 80 12-Jan-00 14-Dec-06 13:30 14 Interest rate (FOMC) Federal reserve 56 2-Feb-00 12-Dec-06 19:15 15 Industrial productioni ISM 84 14-Jan-00 15-Dec-06 14:15 16 NAPM Conference board 84 3-Jan-00 1-Dec-06 15:00 17 Leading Indicatorsj Census 83 2-Feb-00 21-Dec-06 15:00 18 New Home Salesk BLS 84 6-Jan-00 27-Dec-06 15:00 19 Nonfarm Payroll Employment BEA 84 7-Jan-00 8-Dec-06 13:30 20 Personal Spending dept of commerce 60 31-Jan-02 22-Dec-06 13:30 21 Personal Income BLS 84 31-Jan-00 22-Dec-06 13:30 22 Producer Price Census 84 13-Jan-00 19-Dec-06 13:30 23 Retail Salesl Census 83 13-Jan-00 13-Dec-06 13:30 24 Trade Balance dept. of Labor 84 20-Jan-00 12-Dec-06 13:30 25 Initial Unemploymentm Census 363 6-Jan-00 28-Dec-06 13:30 26 Wholesales Census 84 11-Jan-00 11-Dec-06 15:00 aThe time presented in this table is based on GMT time. b3/04 is a missing observation. c3/04 is a missing observation. d1st Quarter of 04 is a missing observation. e1/01~11/01, 8/02,3/04, 8/04 are missing observations. f1/00, 8/04 are missing observations. g8/04 are missing observation. h3/00, 4/01, 10/01, and 3/04 are missing observations. i3/04 and 8/04 are missing observations. 9/06 and 12/06 have revisited observations. j8/04 is a missing observation. k1/01 has a revised observation. 1/04 is a missing observation. l3/04 is a missing observation. m8/21/2004 and 3/13/2004 are missing observations. 88 News Announcements Source No. of Obs Start Date Final Date Time Czech Republic 27 Budget De cita MoF 15 2-May-00 1-Apr-05 13:00 28 Current Accountb CNB 42 16-Jun-03 13-Dec-06 9:00 29 Current Account(US Dollar)c CNB 13 5-Jun-01 6-Sep-05 8:00 30 Consumer Price Index CSO 84 10-Jan-00 8-Dec-06 8:00 31 Exportsd CSO 15 23-Jun-03 3-Jun-05 7:00 32 Gross Domestic Product CSO 29 22-Mar-00 8-Dec-06 8:00 33 Importse CSO 31 21-Jan-00 3-Jun-05 7:00 34 Industrial productionf CSO 82 11-Jan-00 12-Dec-06 8:00 35 Money Supply CNB 14 31-Mar-00 30-Apr-01 8:00 36 Producer Priceg CSO 81 13-Jan-00 14-Dec-06 8:00 37 Retail Salesh CSO 83 14-Jan-00 18-Dec-06 8:00 38 Trade Balancei CSO 82 21-Jan-00 6-Dec-06 8:00 39 Initial Unemploymentj MoL 73 10-Jan-00 12-Jul-06 7:00 Hungary 40 Budget De citk HFM 25 4-Aug-03 8-Aug-06 15:00 41 Current Accountl MNB 53 3-Apr-00 29-Sep-06 6:30 42 Consumer Price Indexm HSO 79 14-Jan-00 12-Dec-06 8:00 43 Gross Domestic Productn HSO 28 31-Mar-00 14-Nov-06 8:00 44 Industrial productiono HSO 50 4-Feb-00 13-Oct-06 7:00 45 Producer Pricep HSO 45 1-Mar-00 30-Nov-06 8:00 46 Trade Balanceq HSO 34 10-Oct-02 9-Nov-06 8:00 a6/01~12/01, 1/04~11/04, 1/05~3/05, and 5/05~12/05 are missing observations. b3/04 is a missing observation. c3Q/01, 2Q/03~4Q/03, and 3Q/04 are missing observations. d7/03, 12/03, 1/04, 6/04, 7/04, 9/04, 10/04, and 1/05~3/05 are missing observations. e5/01~12/01, 1/02~12/02, 1/03~5/03, 7/03, 12/03, 1/04, 6/04, 7/04, 9/04, 10/04, and 1/05~3/05 are missing observations. f9/02 and 3/04 are missing observations. g11/02, 3/04, and 9/04 are missing observations. h8/04 is a missing observation. i6/04 and 11/04 are missing observations. j1/06~6/06 are missing observations. k9/03~1/04, 3/04~5/04, 1/05, 2/06, 3/06, and 5/06 are missing observations. lCurrent Account is announced quarterly since 2005. 12/01, 3/02,4/02,8/02,12/02, 3/04, 7/04, 8/04, 10/04, 11/04 and 1Q/06 are missing observations. m7/00, 11/00, 4/01, 3/03, 3/03 are missing observations. n4Q/01, 1Q/03, 2Q/03, and 4Q/06 are missing observations. o1/00, 8/00~11/00, 3/01, 12/01~4/02, 6/02~8/02, 11/02~3/03, 6/03~6/04, 2/05, 2/06 are missing observations. p1/00, 7/00, 9/00, 10/00, 12/00, 2/01, 9/01, 12 /01, 2/02, 3/02, 11/02, 1/03~3/03, 5/03~3~04, 5/04~7/04, 9/04, 12/04, 3/05, 7/05, 9/05~12/05, 2/06, 5/06, 7/06, 10/06 are missing observations. q11/02, 12/02, 1/03~3/03, 6/03, 8/03~1/04, 4/04, 5/04, 7/04, 6/06 are missing observations. 89 News Announcements Source No of Obs Start Date Final Date Time Indonesia 47 Exportsa BPS 75 1-Sep-00 1-Dec-06 7:00 48 Gross Domestic Product BPS 24 15-Nov-00 16-Nov-06 7:00 49 Importsb BPS 72 1-Sep-00 1-Dec-06 7:00 50 Trade Balancec BPS 75 1-Sep-00 1-Dec-06 7:00 Korea 51 Consumer Price Indexd NSO 62 31-Aug-00 29-Dec-06 4:30 52 Exportse MoC 53 2-Feb-01 1-Dec-06 1:00 53 Gross Domestic Productf BOK 22 22-Aug-00 24-Oct-06 23:00 54 Importsg MoC 53 2-Feb-01 1-Dec-06 1:00 55 Industrial productionh NSO 56 31-Jan-01 29-Dec-06 4:30 56 Initial Unemployment NSO 3 18-Apr-05 13-Sep-06 4:30 Mexico 57 Current Account Banco de Mexico 14 27-Aug-03 24-Nov-06 20:30 58 Consumer Con dencei INEGI 38 4-Aug-03 5-Dec-06 20:30 59 Consumer Price Indexj Banco de Mexico 47 7-Jan-00 7-Dec-06 20:30 60 Fixed Invest INEGI 33 7-Apr-04 7-Dec-06 20:30 61 Gross Domestic Productk INEGI 27 16-Feb-00 22-Nov-06 20:30 62 Industrial productionl INEGI 81 11-Jan-00 13-Dec-06 20:30 63 Producer Pricem Banco de Mexico 29 7-Jan-00 7-May-04 19:30 64 Retail Salesn INEGI 78 20-Jan-00 19-Dec-06 20:30 65 Trade Balanceo INEGI 117 24-Jan-00 26-Dec-06 20:30 66 Unemploymentp INEGI 80 19-Jan-00 20-Dec-06 20:30 67 Wholesales INEGI 31 20-Jan-00 22-Jul-02 19:30 a1/02 is a missing observation. b12/01, 6/03, 2/06, 3/06 are missing observations. c12/01 is a missing observation. d10/00, 12/00, 1/01, 3/01, 4/01, 6/01, 8/01, 3/02, 4/02, 9/03, 12/03, 12/04, 2/05, 5/05 are missing observations. e4/01, 1/02, 3/02, 4/02, 6/02, 11/02, 12/02, 1/03, 2/03, 5/03, 6/03, 8/03, 9/03, 12/03, 2/04, 12/04, 1/05, 2/06 are missing observations. f2Q/01, 3Q/02, 2Q/04 are missing observations. g4/01, 1/02, 3/02, 4/02, 6/02, 11/02~2/03, 5/03, 6/03, 8/03, 9/03, 12/03, 2/04, 12/04, 2/05, 2/06 are missing observations. h2/01, 4/01~8/01, 1/02, 3/02, 7/02~10/02, 2/03, 12/03, 12/04, 12/05 are missing observations. i1/04, 8/04, 11/04 are missing observations. j12/01 is a missing observation. k3Q/02, 4Q/02, and 2Q/05 are missing observations. l11/02, 12/02 and 3/04 are missing observations. m12/01, 6/02~4/04 are missing observations. n8/02, 11/02, 12/02, 2/03, 6/03, and 8/04 are missing observations. o12/01, 9/02, 11/02, 12/02, 1/03, 2/03, 4/03, and 5/03 are missing observations. p8/02, 11/02, 12/02, and 2/03 are missing observations. 90 News Announcements Source No of Obs Start Date Final Date Time Poland 68 Budget De cita MoF 33 15-Nov-01 15-Dec-06 13:30 69 Current Accountb NBP 77 3-Apr-00 13-Dec-06 13:00 70 Consumer Price Indexc PSO 79 15-Feb-00 14-Dec-06 13:00 71 Exportsd NBP 75 3-Apr-00 13-Dec-06 13:00 72 Gross Domestic Producte Eurostat 25 21-Jun-00 30-Nov-06 9:00 73 Importsf NBP 76 3-Apr-00 13-Dec-06 13:00 74 Money Supplyg NBP 72 14-Apr-00 14-Dec-06 13:00 75 Producer Priceh PSO 77 18-Apr-00 19-Dec-06 13:00 76 Retail Salesi PSO 48 20-Dec-02 21-Dec-06 9:00 77 Unemploymentj PSO 80 26-Apr-00 21-Dec-06 9:00 78 Wholesalesk PSO 77 18-Apr-00 19-Dec-06 13:00 South Africa 79 Current Account SARB 2 21-Sep-06 8-Dec-06 9:00 80 Consumer Price Indexl SSA 83 18-Jan-00 20-Dec-06 9:30 81 Gross Domestic Productm SSA 26 28-Feb-00 28-Nov-06 9:30 82 Interest raten SARB 24 15-Nov-01 7-Dec-06 13:20 83 Money Supplyo SARB 79 1-Feb-00 29-Dec-06 6:00 84 Producer Pricep SSA 82 26-Jan-00 21-Dec-06 9:30 85 Retail Sales SSA 4 4-Nov-04 6-Dec-06 9:00 86 Trade Balanceq SARB 81 31-Jan-00 28-Dec-06 12:00 a12/01, 1/04, 3/04, 2/06, and 9/06 are missing observations. b8/03, 10/03, 11/03, 2/04, and 3/04 are missing observations. c2/01~4/01, and 3/04 are missing observations. d8/03, 10/03, 11/03, 2/04, 3/04, 11/05, and 12/05 are missing observations. e2Q/00, 3Q/00, and 4Q03 are missing observations. f8/03, 10/03, 11/03, 2/04, 3/04, and 12/05 are missing observations. g10/01, 5/02, 8/02, 3/04, 4/05, 8/05~10/05, 12/05 are missing observations. h2/01~4/01, and 8/04 are missing observations. i2/03 is a missing observation. j10/04 is a missing observation. k2/01~4/01, and 8/04 are missing observations. l9/04 is a missing observation. m4Q/00, and 3Q/02 are missing observations. nBimonthly announcements. 2/01~2/03 are missing observations. o7/02, 12/02, 2/03, 3/03, and 2/05 are missing observations. p5/03 and 8/03 are missing observations. q12/00, 2/05 and 2/06 are missing observations. 91 News Announcements Source No of Obs Start Date Final Date Time Thailand 87 Current Accounta BOT 27 30-Sep-04 29-Dec-06 7:30 88 Consumer Price Indexb Commerce Ministry 31 3-Nov-03 1-Dec-06 3:30 89 Exportsc BOT 7 30-Sep-04 31-May-05 8:00 90 Gross Domestic Productd BOT 22 19-Jun-00 4-Dec-06 2:30 91 Interest rate BOT 9 19-Oct-05 13-Dec-06 7:30 Turkey 92 Current Accounte CBT 27 24-Jun-04 11-Dec-06 14:35 93 Consumer Price Indexf SIS 48 3-Jan-00 4-Dec-06 14:30 94 Exportsg SIS 8 31-Mar-05 29-Jul-06 13:30 95 Gross Domestic Producth SIS 24 31-Aug-00 11-Dec-06 8:00 96 Importsi SIS 8 31-Mar-05 29-Jul-06 13:30 97 Industrial productionj SIS 71 8-Aug-00 8-Dec-06 8:00 98 Producer Price SIS 23 3-Feb-05 4-Dec-06 14:30 99 Trade Balancek SIS 32 24-Jun-02 29-Dec-06 14:30 100 Unemploymentl SIS 10 9-Dec-04 20-Nov-06 8:00 a4/05 is a missing observation. b12/03~6/04 are missing observations. c2/05, and 4/05 are missing observations. d3Q/02~2Q/03, 4Q/04 are missing observations. e8/04, 9/04, 11/04, and 1/06 are missing observations. f2/01~6/03, and 8/03~2/04 are missing observations. g10/05~6/06 are missing observations. h4Q/01, and 3Q02 are missing observations. i10/05~6/06 are missing observations. j12/00, 3/01, 4/01, 6/01, 8/01, and 12/02 are missing observations. k1/04, 2/04, 4/04, 5/04, 8/04 are missing observations. l1/05~4/05, 7/05, 9/05, 12/05, 2/06~5/06, and 7/06~9/06 are missing observations. 92 T abl e A.4 : U.S . an d Domesti c New s Res p ons e an d R square s Eur o Cze ch Republi c Hungar y Indonesi a Kore a Mexic o P olan d Sout h Afric a Thailan d T ur ke y R 2 R 2 R 2 R 2 R 2 R 2 R 2 R 2 R 2 R 2 U.S . Announceme n t Busines s In ve ntorie s 0.0 0 0.0 0 0.0 1 0.0 0 -0.0 1 0.0 0 0.0 1 0.0 3 0.0 0 0.0 0 -0.0 1 0.0 0 -0.0 1 0.0 0 -0.0 3 0.0 3 0.0 0 0.0 0 -0.0 3 0.0 2 Budge t De ci t 0.0 0 0.0 1 -0.0 1 0.0 2 0.0 1 0.0 0 0.0 1 0.0 0 0.17 * 0.1 5 0.0 0 0.0 1 0.0 0 0.0 0 -0.0 1 0.0 1 -0.0 3 0.0 2 0.0 8 0.6 6 Curre nt Accou nt 0.0 1 0.0 6 0.04 * 0.1 5 0. 05 0.1 3 -0.07 * 0.2 1 0. 01 0.0 2 0.0 0 0.0 0 0.0 2 0.1 0 0.0 4 0.0 9 0.0 3 0.1 5 0.0 4 0.0 4 Capita l Utilizatio n 0.01 * 0.1 0 0.0 1 0.0 2 0. 00 0.0 0 -0.0 1 0.0 1 -0.0 3 0.0 4 0.0 0 0.0 1 0.0 0 0.0 0 0.06* * 0.1 3 0.0 1 0.0 2 0. 02 0.0 3 Consume r Con denc e 0.06* * 0.3 8 0.11* * 0.4 8 0.10* * 0.4 0 0.0 1 0.0 2 0. 03 0.0 7 -0.03* * 0.1 3 0.05* * 0.1 7 0.05* * 0.1 8 0.0 0 0.0 0 0.04 * 0.0 9 Consume r Credi t 0.0 0 0.0 1 0.0 0 0.0 0 -0.02 * 0.0 6 -0.0 1 0.0 4 -0.0 1 0.0 7 -0.0 1 0.0 2 0.0 0 0.0 0 -0.0 2 0.0 2 0.0 1 0.0 1 -0.0 5 0.5 9 Constructio n Sp endin g 0.0 0 0.0 0 0.04 * 0.0 8 0. 04 0.0 5 0.0 0 0.0 0 0.0 1 0.0 1 0.0 0 0.0 0 0.0 2 0.0 5 -0. 01 0.0 0 -0.0 2 0. 05 -0.0 2 0.0 1 Consume r Pric e Inde x 0.0 1 0.0 4 0.0 1 0.0 0 0.0 2 0.0 2 -0.05* * 0.1 3 0. 00 0.0 0 0.03 * 0.0 5 0.0 2 0.0 1 0.0 5 0.0 4 0.0 1 0.0 0 0.0 6 0.0 4 Durabl e G oo ds Order s 0.02* * 0.0 9 0.06* * 0.2 1 0.07* * 0.2 2 0.05 * 0.1 3 0. 02 0.0 6 -0.0 1 0.0 2 0.04* * 0.1 4 0.05* * 0.1 1 -0.0 1 0.0 0 -0.0 2 0.0 1 Factor y Order s 0.0 0 0.0 0 0.0 3 0.0 5 0.0 2 0.0 3 0.0 0 0.0 0 -0.0 1 0.0 1 0.0 0 0.0 0 0.0 0 0.0 0 0.03 * 0.0 6 0.0 1 0.0 3 0.05* * 0.1 7 Gros s Domesti c Pr oduc t 0.05* * 0.2 7 0.11* * 0.4 2 0.11* * 0.3 9 0.06* * 0.2 9 0. 02 0.0 6 0.0 0 0.0 0 0.07* * 0.3 7 0.07* * 0.1 6 0.0 0 0.0 0 0. 01 0.0 0 Housin g Start s 0.0 0 0.0 0 0.0 1 0.0 1 0.0 1 0.0 1 0.0 0 0.0 0 0.0 2 0.0 2 0.0 0 0.0 0 0.0 1 0.0 0 0.0 1 0.0 1 -0.0 1 0.0 1 0.0 1 0. 00 Im p ort s -0.0 1 0.0 1 0.03 * 0.0 7 0. 02 0.0 4 0.0 1 0.0 1 0.0 3 0.0 7 0.0 2 0.0 1 0.0 3 0.0 4 0.0 0 0.0 0 -0.0 1 0.0 1 -0.0 2 0.0 1 In teres t rat e 0.0 0 0.0 0 0.0 1 0.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 1 0.0 0 0.0 0 0.0 0 Industria l pr oductio n 0.0 1 0.0 1 0.02 * 0.0 6 0. 02 0.0 4 0.0 0 0.0 0 0.0 0 0.0 0 -0.0 1 0.0 1 0.0 1 0.0 2 0.0 2 0.0 1 0.0 0 0.0 0 0.0 1 0.0 0 NAP M 0.1 7 0.0 2 1.39* * 0.2 1 1.35* * 0.1 6 0.2 0 0.0 1 0. 29 0.0 2 0.0 4 0.0 0 0.82* * 0.1 9 0. 44 0.0 3 -0.44 * 0.0 9 -0.2 8 0.0 1 Leadin g Indicator s 0.0 0 0.0 0 0.0 0 0.0 0 0.0 2 0.0 4 -0.0 1 0.0 2 -0.0 1 0.0 1 0.0 0 0.0 0 0.0 0 0.0 0 -0.0 1 0.0 0 0.0 1 0.0 1 0.0 1 0.0 0 Ne w Hom e Sale s 0.01 * 0.0 5 0.04 * 0.0 7 0.05* * 0.1 3 0.03 * 0.1 4 0. 03 0.0 6 0.0 1 0.0 1 0.03* * 0.1 4 0. 01 0.0 0 -0.0 1 0.0 2 -0.0 1 0.0 0 Nonfar m P ayrol l 0.09* * 0.1 8 0.22* * 0.3 4 0.24* * 0.3 6 0.06 * 0.1 2 0.09 * 0.1 6 0.0 3 0.0 3 0.18* * 0.3 1 0.19* * 0.2 4 0.0 1 0.0 1 0. 02 0.0 0 P ersona l Sp endin g 0.0 0 0.0 0 0.0 1 0.0 0 0.0 2 0.0 2 0.0 1 0.0 3 -0.0 1 0.0 1 0.0 0 0.0 0 0.0 0 0.0 0 0.0 1 0.0 1 -0.0 1 0.0 2 -0.0 1 0.0 0 P ersona l Incom e 0.0 1 0.0 4 0.0 0 0.0 0 0.0 1 0.0 0 -0.0 1 0.0 2 0.0 0 0.0 0 0.0 1 0.0 1 0.0 0 0.0 0 0.0 1 0.0 0 0.0 0 0.0 0 0.0 2 0. 01 Pr oduce r Pric e 0.0 1 0.0 3 0.0 2 0.0 1 0.0 2 0.0 1 0.0 3 0.0 5 0.0 1 0.0 3 0.0 1 0.0 1 0.0 3 0.0 5 0.0 3 0.0 2 0.0 0 0.0 0 0.0 0 0.0 0 Retai l Sale s 0.03* * 0.2 0 0.08* * 0.2 2 0.08* * 0.2 4 0.0 3 0.0 6 0.03 * 0.0 9 -0.05 * 0.0 8 0.07* * 0.2 3 0. 03 0.0 1 0.0 1 0.0 3 0.0 0 0.0 0 T rad e Balanc e 0.04* * 0.2 0 0.12* * 0.3 8 0.11* * 0.3 6 0.0 0 0.0 0 0. 02 0.0 5 -0.0 1 0.0 1 0.10* * 0.3 5 0.08* * 0.1 9 0.0 1 0.0 0 0. 02 0.0 1 Initia l Unempl oyme nt -0.01* * 0.0 4 -0.03* * 0.0 7 -0.03* * 0.0 6 0.0 0 0.0 0 -0.0 1 0.0 1 0.0 0 0.0 0 -0.01 * 0.0 2 -0.0 1 0.0 0 0.0 0 0.0 0 -0.0 1 0.0 0 Wholesale s 0.0 1 0.0 4 0.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 0 -0.0 1 0.0 1 0.0 0 0.0 0 0.0 1 0.0 1 0.0 0 0.0 0 -0.0 1 0.0 1 93 Eur o Cze ch Republi c Hungar y Indonesi a Kore a Mexic o P olan d Sout h Afric a Thailan d T ur ke y R 2 R 2 R 2 R 2 R 2 R 2 R 2 R 2 R 2 R 2 Domesti c Announceme n t Budge t De ci t 0.0 0 0.0 0 0.0 4 0. 16 0.0 0 0. 00 Curre nt Accou nt -0.03* * 0.2 4 -0.0 2 0.0 4 0.0 0 0.0 2 - 0.078* * 0.2 4 0.1 3 0.1 1 0. 37 0.1 1 0.0 3 0.0 5 Curre nt Acc . (US ) -0.0 1 0.0 2 Consume r Con denc e 0.0 0 0.0 0 Consume r Pric e In de x -0.0 2 0.0 3 0.0 2 0.0 3 0.01 * 0.1 6 0.0 1 0.0 6 0.0 1 0.0 2 -0.07* * 0.2 8 0. 00 0.0 0 -0.0 1 0.0 0 Ex p ort s -0.0 3 0.0 3 0.0 0 0.0 0 0.0 1 0.0 4 0.0 1 0.0 0 0.0 0 0.0 2 0.0 4 0.0 4 Fixe d In ves t 0.0 0 0.0 0 GD P -0.03 * 0.1 4 0.0 1 0.0 1 0. 00 0.0 0 0.0 1 0.0 0 -0.03 * 0.2 5 -0.0 4 0.1 8 -0.0 3 0.1 0 -0.0 7 0.0 6 -0.0 2 0.0 1 Im p ort s -0.0 1 0.0 1 -0.0 1 0.0 3 0.0 0 0.0 0 0.0 4 0.0 6 -0.1 4 0.3 0 In teres t rat e -0.26 * 0.2 0 -0.0 1 0.0 3 Industria l pr oductio n -0.0 1 0.0 4 -0.028 * 0.1 0 0.0 0 0.0 0 0.0 1 0.0 1 -0.0 1 0.0 0 Mone y Suppl y 0.0 0 0.0 0 -0.0 1 0.0 3 0.0 4 0.0 7 Pr oduce r Pric e -0.02* * 0.1 0 -0.0 2 0.0 8 0.0 1 0.1 4 0.0 0 0.0 0 0.0 1 0.0 1 0.1 5 0.1 5 Retai l Sale s -0.03* * 0.1 0 -0.0 1 0. 05 -0.0 1 0.0 1 -0 .0 2 0.0 5 T rad e Balanc e -0.07* * 0.3 3 -0.0 1 0.0 0 0.3 6 0.0 0 -0.01* * 0.1 0 -0.13* * 0.3 0 0.0 4 0.0 3 Initia l Unempl oyme nt -0.0 1 0.0 2 0.0 0 0.2 9 -0.0 2 0.0 2 0.0 2 0.0 3 -0.0 1 0.0 1 Wholesale s -0.03 * 0.2 5 -0.0 2 0.0 4 Notes : W e estimat e th e ex chang e rat e conditiona l mea n m ode l (1.3 ) , wher e R t is th e 5-mi nut e retur n fro m p eri od t to p eri od t+1 , an d S k t is th e standardize d new s surpris e as descri b ed in th e text . W e estimat e th e regressio n onl y usin g non-missin g dat a fo r ea ch new s surprise . k an d R 2 ar e re p orte d fo r ea ch regressio n result . A st erisk s denot e statistica l signi canc e (** * at 1- p erce nt le vel , ** at 5- p erce nt le vel , an d * at 10- p erce nt le vel) . 94 Table A.5: The Impact of Major News Surprises on FX Returns and FX Volatility Announcements Czech Republic Hungary Indonesia Korea Mexico Poland South Africa Thailand Turkey Response of Contemporaneous News Surprises on FX Returns Durable Goods Orders 0.05** 0.06** 0.04** 0.03* -0.01* 0.04** 0.05** -0.03** -0.01 Nonfarm Payroll 0.19** 0.21** 0.04** 0.09** 0.02** 0.18** 0.18** 0.00 0.02* Trade Balance 0.12** 0.12** 0.00 0.02** -0.02** 0.12** 0.09** 0.00 0.02 Producer Price 0.02** 0.02** 0.04** 0.01 0.01* 0.02** 0.05** 0.00 -0.01 New Home Sales 0.02** 0.03** 0.03** 0.03** 0.01* 0.03** 0.00 -0.01 -0.02 GDP 0.10** 0.11** 0.07** 0.02** 0.00 0.07** 0.07** -0.01 0.03 Consumer con dence 0.11** 0.10** 0.01 0.02** -0.03** 0.04** 0.05** 0.00 0.05 Retail Sales 0.07** 0.07** 0.03** 0.03** -0.07** 0.07** 0.02 0.01 -0.01 Initial Unemployment -0.03** -0.03** -0.01 -0.01** 0.00 -0.02** -0.02** 0.01 0.01 Impact of Contemporaneous News Surprises on Volatility Durable Goods Orders 0.02** 0.05** 0.01* -0.01** 0.00 0.01** 0.01** 0.02** 0.01 Nonfarm Payroll 0.17** 0.17** -0.01** 0.09** 0.05** 0.16** 0.18** 0.01** 0.10** Trade Balance 0.04** 0.04** 0.04** 0.04** 0.00** 0.06** 0.03** 0.00 0.01 Producer Price 0.03** 0.03** 0.02** -0.01* 0.01** 0.01 0.01** 0.00 0.07** New Home Sales 0.03** 0.03** 0.02** 0.02** 0.00 0.03** 0.00 0.00 -0.01 GDP 0.05** 0.04** 0.04** 0.04** 0.00 0.04** 0.04** 0.01 0.03** Consumer con dence 0.03** 0.04** 0.03** 0.03** 0.02** 0.03** 0.00 0.01** 0.00 Retail Sales 0.03** 0.03** 0.03** 0.05** 0.03** 0.04** 0.00* 0.00 0.00 Initial Unemployment 0.01** 0.00 0.01** 0.01** 0.01** 0.02** 0.02** 0.00 0.01** Cumulated Impact of News Surprises on Volatility Durable Goods Orders 0.17** 0.20** 0.05* 0.05** 0.00 0.07** 0.13** 0.02** 0.03 Nonfarm Payroll 0.26** 0.32** 0.16** 0.39** 0.19** 0.37** 0.38** 0.05** 0.16** Trade Balance 0.05** 0.06** 0.01** 0.25** 0.02** 0.11** 0.11** -0.03 0.02 Producer Price 0.03** 0.06** 0.04** 0.04* 0.01** 0.01 0.10** -0.01 0.08** New Home Sales 0.10** 0.06** 0.05** 0.03** -0.01 0.08** 0.01 0.00 -0.01 GDP 0.06** 0.02** 0.05** 0.04** 0.00 0.11** 0.08** 0.03 -0.02** Consumer con dence 0.03** 0.10** 0.13** 0.07** 0.05** 0.12** 0.01 0.02** -0.04 Retail Sales 0.09** 0.05** 0.06** 0.13** 0.03** 0.16** 0.06* 0.00 0.01 Initial Unemployment 0.03** 0.00 -0.01** 0.01** 0.04** 0.12** 0.07** 0.00 0.01** Notes: We estimate the exchange rate conditional mean model (1.4) Rt = 0 + PI i=1 iRt i + PK k=1 PJ j=0 kjSk;t j + t, and we report estimates of the contemporaneous response of exchange-rate returns to news surprises, k0. We also estimate the disturbance volatility model (1.5), and we report estimates of the contemporaneous response of exchange-rate volatility to news surprise, k0= kpk(0). In addition, we report estimates of the cumulative volatility response, P12 j0=0 kpk(j 0), as described in the text. Asterisks denote statistical signi cance (*** at 1-percent level, ** at 5-percent level, and * at 10-percent level). 95 Table A.6: Response of Major News Surprises and Announcement E ects Announcements Czech Republic Hungary Indonesia Korea Mexico Poland South Africa Thailand Turkey Impact of Major News Surprises on FX Returns Durable Goods Orders k0 0.06*** 0.07*** 0.05*** 0.02*** -0.02*** 0.05*** 0.05*** -0.03*** -0.01 k0 -0.01 -0.02*** 0.00 0.01 0.01** -0.02*** -0.02 0.03*** 0.04* Nonfarm Payroll k0 0.22*** 0.25*** 0.06*** 0.09*** 0.01*** 0.20*** 0.18*** 0.01 0.05*** k0 0.05*** 0.06*** 0.02*** 0.04*** -0.05*** 0.02*** -0.01 0.01 0.10*** Trade Balance k0 0.12*** 0.11*** 0.00 0.02*** -0.02*** 0.11*** 0.10*** 0.00 0.03* k0 0.01 0.00 -0.02*** 0.02*** 0.00 0.01** 0.01 0.01 0.05** Producer Price k0 0.00 0.01 0.03*** 0.02*** 0.01** 0.01** 0.04*** 0.01 0.00 k0 -0.02*** -0.03*** -0.02** -0.01 -0.03*** -0.02*** 0.00 0.01 0.09*** New Home Sales k0 0.02*** 0.03*** 0.02*** 0.03*** 0.01** 0.03*** 0.00 -0.01 -0.03 k0 -0.01** 0.03*** 0.01 -0.02*** -0.01** 0.01 0.01 -0.01 0.03 GDP k0 0.10*** 0.11*** 0.06*** 0.02*** 0.00 0.07*** 0.07*** 0.00 0.03* k0 -0.03*** -0.04*** -0.04*** -0.03*** -0.01** -0.01 -0.02* 0.01 0.07*** Consumer Con dence k0 0.11*** 0.10*** 0.01 0.01 -0.03*** 0.05*** 0.05*** 0.00 0.05 k0 0.00 -0.01 -0.01 0.04*** 0.01** -0.02*** 0.00 0.00 -0.02 Real Sales k0 0.09*** 0.08*** 0.03*** 0.03*** -0.07*** 0.08*** 0.03** 0.01 -0.01 k0 0.00 0.00 0.00 0.00 0.02*** 0.00 -0.06*** 0.01 -0.04** Initial Unemployment k0 -0.04*** -0.03*** -0.01** -0.02*** 0.00 -0.02*** -0.02*** 0.01 0.01 k0 -0.01*** -0.01** 0.00 0.00 0.00 -0.01*** -0.01 -0.01* 0.01 Impact of Major News Surprises on Volatility Durable Goods Orders k0 -0.01 0.00 0.00 -0.01 0.00 0.00 0.00 0.03*** 0.01 k0 0.05*** 0.05*** 0.04*** 0.03*** 0.02*** 0.04*** 0.01 0.00 -0.01 Nonfarm Payroll k0 0.07*** 0.06*** 0.00 0.02* 0.02*** 0.08*** 0.12*** 0.01 0.04** k0 0.17*** 0.16*** 0.08*** 0.08*** 0.06*** 0.12*** 0.08*** 0.00 0.07*** Trade Balance k0 0.02*** 0.02*** 0.01 0.03*** 0.00 0.03*** 0.03*** -0.01** -0.03 k0 0.04*** 0.04*** 0.05*** 0.00 0.02** 0.04*** 0.02 0.03*** 0.04* Producer Price k0 -0.02*** -0.02*** 0.00 0.00 -0.01* 0.00 0.00 0.00 -0.01 k0 0.06*** 0.06*** 0.04*** 0.02 0.03*** 0.03** 0.04*** -0.01 0.08*** New Home Sales k0 -0.02*** -0.02** 0.00 -0.01 0.01 0.01 0.00 0.00 0.01 k0 0.07*** 0.05*** 0.03** 0.06*** 0.01 0.02*** 0.03*** 0.00 -0.03 GDP k0 0.02*** 0.03*** 0.03*** 0.02** 0.00 0.01* 0.03** 0.02*** -0.01 k0 0.04*** 0.03*** 0.01 0.01 0.02*** 0.03*** 0.01 -0.02* 0.06*** Consumer Con dence k0 0.02*** 0.03*** 0.00 0.00 0.01** 0.01 0.00 0.01*** 0.00 k0 0.03*** 0.02*** 0.03** 0.04** 0.00 0.04*** 0.01 -0.01 0.00 Retail Sales k0 0.01* 0.01 0.01 0.02 0.03*** 0.02*** 0.00 0.00 -0.04* k0 0.04*** 0.04*** 0.01 0.02** 0.03*** 0.02** 0.06*** 0.00 0.04 Initial Unemployment k0 0.00 0.00 0.01 0.00 0.00 -0.01 0.00 0.00 0.00 k0 0.02*** 0.02*** 0.01 0.02*** 0.01*** 0.03*** 0.01** 0.00 0.01 Notes: We estimate the exchange rate conditional mean model (1.7), where Dk;t j is dummy variable for the announcement. We report estimates of the contemporaneous response of exchange-rate returns to news surprises, k0. We also estimate the disturbance volatility model (1.8). Asterisks denote statistical signi cance (*** at 1-percent level, ** at 5-percent level, and * at 10-percent level). 96 T abl e A.7 : F- T es t Resul ts wit h Symmetri c Res p on se b et w ee n P ositi ve an d Negati ve New s Su rprise s New s Eur o Cze ch Republi c Hungar y Indonesi a Kore a Mexic o P olan d Sout h Afric a Thailan d T ur ke y Consume r Con denc e F valu e 1.4 0 0.8 3 0.6 4 1.2 2 1.6 7 4.0 3 2.4 7 0.5 6 1.7 9 2.0 2 P valu e 0.2 5 0.4 4 0.5 3 0.3 1 0.2 0 0.0 2 0.0 9 0.5 7 0.1 8 0.1 5 Durabl e G oo ds Order s F valu e 1.2 6 1.6 7 2.5 3 2.4 3 0.7 7 1.2 5 1.3 3 2.6 3 6.8 6 0.3 5 P valu e 0.2 9 0.2 0 0.0 9 0.1 0 0.4 7 0.2 9 0.2 7 0.0 8 0.0 0 0.7 0 Gros s Domesti c Pr oduc t F valu e 0.6 5 3.0 9 4.4 8 0.0 3 0.7 7 0.0 1 0.6 2 0.8 2 0.1 5 2.6 5 P valu e 0.5 2 0.0 5 0.0 1 0.9 7 0.4 7 0.9 9 0.5 4 0.4 5 0.8 6 0.0 8 Ne w Hom e Sale s F valu e 2.3 1 0.2 3 0.0 2 0.2 2 0.4 3 0.5 5 1.0 2 0.2 0 1.7 8 0.8 8 P valu e 0.1 1 0.8 0 0.9 8 0.8 1 0.6 5 0.5 8 0.3 7 0.8 2 0.1 8 0.4 2 Nonfar m P ayrol l F valu e 1.2 5 1.4 7 1.0 5 2.6 3 1.6 4 1.6 2 0.8 5 1.8 0 0.1 8 2.1 1 P valu e 0.2 9 0.2 4 0.3 6 0.0 8 0.2 1 0.2 1 0.4 3 0.1 7 0.8 3 0.1 3 Pr oduce r Pric e F valu e 1.3 0 1.2 7 1.5 1 1.3 7 0.8 9 0.3 1 1.1 4 0.5 8 0.3 8 1.6 7 P valu e 0.2 8 0.2 9 0.2 3 0.2 6 0.4 2 0.7 4 0.3 3 0.5 6 0.6 9 0.2 0 Retai l Sale s F valu e 0.0 3 0.9 3 0.6 6 0.3 4 0.2 9 2.1 4 2.8 1 0.3 4 1.5 6 0.7 0 P valu e 0.9 7 0.4 0 0.5 2 0.7 1 0.7 5 0.1 3 0.0 7 0.7 1 0.2 2 0.5 0 T rad e Balanc e F valu e 0.2 1 1.1 0 0.7 6 1.1 4 0.2 9 3.5 9 3.2 5 0.5 5 0.1 8 2.6 0 P valu e 0.8 1 0.3 4 0.4 7 0.3 3 0.7 5 0.0 3 0.0 5 0.5 8 0.8 4 0.0 8 Initia l Unempl oyme nt F valu e 1.6 0 4.1 0 3.7 0 0.0 6 3.3 2 0.5 4 2.4 1 1.1 0 1.0 0 1.8 3 P valu e 0.2 0 0.0 2 0.0 3 0.9 4 0.0 4 0.5 8 0.0 9 0.3 3 0.3 7 0.1 6 Notes : W e estimat e th e ex chang e rat e conditiona l mea n m ode l (1.9) , wher e D k ;t is dum m y variabl e fo r tha t ha s valu e 1 if th e ne w surpris e is p ositi ve , an d valu e of 0 if ne gati ve . Th e nul l hy p othesi s use d in th e tes t is 2 = 3 = 0. 97 Table A.8: Impact of Major News Surprises with FX Forecasts Announcements Czech Republic Hungary Indonesia Korea Mexico Poland South Africa Thailand Turkey Impact of News Surprises Only Durable Goods Orders 0.05** 0.07** 0.06** 0.02* 0.01 0.06** 0.06** -0.02 0.04 Nonfarm Payroll 0.19** 0.21** 0.07** 0.08** 0.11** 0.23** 0.26** 0.01 0.14** Trade Balance 0.04** 0.06** -0.01 0.01 -0.01 0.11** 0.09** 0.01 0.03 Producer Price -0.01 -0.01 0.02* 0.01 0.01 0.01 0.03 0.00 0.02 New Home Sales -0.02 0.03* 0.04** 0.03** 0.01 0.03** -0.01 -0.01 -0.04 Gross Domestic Product 0.09** 0.09** 0.06** 0.02 0.01 0.07** 0.09** 0.00 0.00 Consumer con dence 0.12** 0.10** 0.02 0.02 0.02 0.07** 0.05** 0.00 0.10 Retail Sales 0.09** 0.07** 0.04** 0.04** 0.02 0.08** 0.03* 0.01 -0.01 Initial Unemployment -0.04** -0.03** -0.01 -0.02** -0.02** -0.02** -0.02** 0.01 0.01 Impact of News Surprises with FX Forecasts Durable Goods Orders 0.22** 0.19** 1.08* 0.39 0.04** 0.10 0.29 -1.04 0.00* Nonfarm Payroll 0.73** 0.65** 1.36** 1.07* 0.10** -0.52* -0.13 -1.80** 0.00** Trade Balance 0.38** 0.31** -0.74 1.12* -0.01 -0.84* -1.00** 0.06 0.00 Producer Price 0.13** 0.18** 0.50 0.16 0.01 0.24 0.11 0.43 0.00* New Home Sales 0.18** -0.01 0.89* -0.75 0.01 0.00 0.34 0.28 0.00 Gross Domestic Product 0.20** 0.24** 0.87* 0.81* 0.03 0.17 -0.51 -0.18 0.00 Consumer con dence -0.11* -0.03 0.52 2.35** 0.05** -1.02** 0.15 0.65 0.00 Retail Sales 0.06 0.13* 1.85** -0.29 0.12** 0.30 0.39 -0.49 0.00 Initial Unemployment -0.01 -0.04 -0.09 -0.16 -0.03** 0.44** 0.14 0.32 0.00 Notes: We estimate the exchange rate conditional mean model (1.10) , where FXDj;t is the index that measures the change between consensus and spot price. Asterisks denote statistical signi cance (*** at 1-percent level, ** at 5-percent level, and * at 10-percent level). 98 Table A.9: Impact of Major News Surprises with FX Forecasts Dispersion Announcements Czech Republic Hungary Indonesia Korea Mexico Poland South Africa Thailand Turkey Impact of News Surprises Only Consumer Con dence 0.08** 0.09** -0.01 0.03 -0.03 0.07** 0.04 -0.01 0.12 Durable Goods Order 0.12** 0.11** 0.13** 0.02 0.01 0.06** 0.03 -0.05 0.12* Gross Domestic Product 0.19** 0.09** 0.11** 0.01 0.01 0.04 0.03 0.04 0.01 New Home Sales 0.00 0.07** -0.02 0.01 -0.02 0.01 -0.05 -0.05 -0.05 Nonfarm Payroll 0.25** 0.17** 0.08** 0.19** 0.10** 0.47** 0.16** 0.00 0.17** Producer Price 0.02 -0.01 0.04* 0.05* -0.01 0.02 -0.01 0.00 0.12** Retail Sales 0.16** 0.06** 0.07** 0.02 0.14** 0.14** 0.03 0.02 -0.03 Trade Balance 0.07** 0.17** -0.03 0.02 -0.04 0.10** 0.02 -0.02 0.03 Initial Jobless Claim -0.05** -0.03** -0.01 0.01 -0.02 -0.02* -0.01 0.01 0.01 Impact of News Surprises with Dispersions Consumer Con dence 0.61* 0.32 0.10 -0.09 0.04 -0.20 0.08 0.16 -0.31 Durable Goods Order -0.76** -0.54 - 0.60** 0.16 -0.47 -0.02 0.15 0.34 -0.82** Gross Domestic Product -1.96** 0.58 -0.51 0.15 -0.31 0.33 0.26 -0.72 0.06 New Home Sales 0.11 -0.75* 0.43 0.28 0.38 0.18 0.28 0.55 0.20 Nonfarm Payroll 0.87** 2.23** -0.21 -1.59** -1.36** -2.02** 0.50* 0.05 -0.76** Producer Price -0.16 0.50* -0.23 -0.57 0.35 -0.02 0.22 0.12 -0.68** Retail Sales -0.98** 0.45 -0.28 0.25 -3.42** -0.45* 0.03 -0.17 0.06 Trade Balance 0.83** -1.10** 0.31 0.03 0.49 0.05 0.34 0.37 -0.10 Initial Jobless Claim 0.17 -0.03 0.03 -0.47* 0.33 0.00 -0.07 -0.04 0.00 Notes: We estimate the exchange rate conditional mean model (1.12) , where DISP is the index that measures the magnitude of dispersions between consensus and spot price, de ned by , means the maximum of FX forecasts at time t, and means the minimum of FX forecasts at time t. Asterisks denote statistical signi cance (*** at 1-percent level, ** at 5-percent level, and * at 10-percent level). 99 Table A.10: Summary Table for FX Time Series Country Number of Observations Number of Nonmissing Observation Period Czech Republic 736,416 479,119 January 2, 2000 ~ December 31, 2006 Hungary 736,416 517,950 January 2, 2000 ~ December 31, 2006 Indonesia 736,416 365,843 January 2, 2000 ~ December 31, 2006 Korea 727,488 341,508 January 2, 2000 ~ December 31, 2006 Mexico 736,416 302,674 January 2, 2000 ~ December 31, 2006 Poland 736,416 409,279 January 2, 2000 ~ December 31, 2006 South Africa 736,416 366,973 January 2, 2000 ~ December 31, 2006 Thailand 736,416 446,514 January 2, 2000 ~ December 31, 2006 Turkey 631,008 175,967 January 2, 2001 ~ December 31, 2006 Source: Olsen Financial Technology (www.olsendata.com) * For Korea, January 2004 data is not included. 100 T abl e A.11 : Ex chan ge Regim e Change s fro m 200 0 to 200 6 Cou ntr y Cu rrenc y P eri od Classi catio n Note s Cze ch Republi c Cze ch korun a ful l sampl e Manage d oatin g wit h no predetermine d pat h fo r th e ex chang e rat e Th e externa l valu e of th e korun a is determine d by suppl y an d deman d in th e foreig n ex chang e mar ket . Th e Cze ch Nationa l Ban k (CNB ) m ay in ter ven e in th e foreig n ex chang e mar ke t in or de r to sm oot h larg e in trad ay volatili ty swing s of th e Euro- korun a rate . Th e CN B publishe s dail y rate s of 29 selecte d currencie s agains t th e korun a fo r custom s an d accou ntin g pur p oses . Commercia l bank s se t thei r ow n ex chang e rat e wit h no limitation . Hungar y Hungaria n fori nt 6/4/200 3 { curre nt P egge d ex chang e rat e withi n horizo nta l band s Th e Hungaria n fo ri nt trad es agains t th e E ur o withi n a ban d of 15 % aroun d th e ce ntra l pari ty , whi ch is xe d to th e Eur o at F t 282.3 6 p er EU R 1. 10/1/200 1 { 6/4/200 4 P egge d ex chang e rat e withi n horizo nta l band s Th e cr awlin g p eg w as ab olishe d an d th e ce ntra l pari ty of th e fori nt w as xe d to th e Eur o at F t 276. 1 p er EU R 1. T hus , th e ex chang e arrangeme nt of th e fori nt w as reclassi e d to th e categor y p egge d ex chang e rat e withi n ho rizo nta l ba nd s fro m th e categor y cr awlin g band . (10/1/2001 ) 5/4/200 1 { 10/1/200 1 Cr awlin g Ban d Th e widt h of th e ban d w ithi n whi ch th e fori nt trade s ag ains t th e Eur o w as widene d to 15 % fro m 2.25 % aroun d th e pari ty . (5/4/2001 ) 4/1/2001 { 5/4/200 1 Cr awlin g Ban d Th e mo nthl y depreciatio n of th e fori nt w as adjuste d to 0.2 % fro m 0.3 % (4/1/2001 ) 4/1/200 0 { 4/1/200 1 Cr awlin g Ban d Th e mo nthl y depreciatio n of th e fori nt w as adjuste d to 0.3 % fro m 0.4 % (4/1/2000 ) 1/1/200 0 { 4/1/200 0 Cr awlin g Ban d Th e preannounce d rat e of cr aw l agains t th e Eur o w as a ected . (1/1/2000 ) Indonesi a Indonesia n rupia h ful l sampl e Manage d oatin g wit h no predetermine d pat h fo r th e ex chang e rat e Th e ex chang e rat e is determine d by suppl y an d deman d condition s in th e foreig n ex chang e mar ket . H ow ev er , th e Ban k Indonesi a (BI ) m ay in ter ven e in th e for eig n ex ch ang e mar ke t to mai ntai n stabili ty of th e ex chang e rate . Kore a Korea n w on ful l sampl e Inde p ende ntl y oatin g Th e ex chang e rat e of th e w on is determine d on th e basi s of suppl y an d deman d in th e foreig n ex chang e mar ket . H ow ev er , th e authoritie s in ter ven e whe n necessar y to cou nte r disorderl y condition s in th e mar ket . 101 Mexic o Mexica n p es o ful l sampl e Inde p ende ntl y oatin g Th e ex chang e rat e of th e p es o is determine d freel y in th e foreig n ex chang e mar ket . Th e Ex chang e Commissio n establishe d a rules-base d me chanis m to red uc e th e rat e of in ter nationa l reser ve s accu m ulation . Th e Ban k of Mexic o (BOM ) sell s dollar s directl y in th e foreig n ex chang e ma rk et ev er y da y accordin g to th e foll owin g pr ocedure : th e BO M announce s ev er y quarte r th e tota l amou nt of dollar s it wil l o e r to th e currenc y mar ke t ea ch da y durin g th e foll owin g fou r quarters . Th e tota l amou nt of dollar s to b e so ld wil l equa l 50 % of th e ne t in ternationa l reser ve s accu m ulate d du rin g th e previou s quarter , wit h one-fourt h of th e establish ed amou nt b ein g auctione d ea ch quarter , no t includin g th e cu m ulati ve amou nt of dollar s sol d throug h th e auctio n me chanis m durin g th e sam e p eri od . Base d on th e tota l amou nt of dollars , th e BO M aucti on s on a dai ly basi s a xe d amou nt of dollar s foll owin g a preestablishe d sc hedul e (th e dail y am ou nt to b e so ld is determine d accordin g to th e nu m b er of w orkin g da ys in th e curre nt quarter) . P olan d P olis h zlo ty 4/12/200 0 { curre nt Inde p ende ntl y oatin g Th e ex chang e rat e of th e zlo ty is determine d on th e basi s of suppl y an d deman d in th e foreig n ex chang e mar ket , an d th e zlo ty is trad ed freel y agains t al l currencies . 3/24/199 9 { 4/12/200 0 Cr awlin g p eg E ecti ve 1/1/1999 , th e cu rrenc y bas ke t w as change d to 55 % Eur o an d 45 % dolla r. E ecti ve 3/24/199 9, th e widt h of th e ban d w as increase d to 15 % aroun d th e ce ntra l pari ty . Sout h Afric a Sout h Africa n ran d ful l sampl e Inde p ende ntl y oatin g Th e ex chang e rat e of th e ran d is determine d by deman d an d suppl y in th e foreig n ex chang e mar ket . Thailan d Tha i ba ht ful l sampl e Manage d oatin g wit h no predetermine d pat h fo r th e ex chang e rat e Th e ex chang e rat e of th e ba ht is determine d in th e foreig n ex chang e mar ket . Th e ba ht-dolla r referenc e ex chang e rat e is announce d dail y, base d on th e av erag e ex ch ang e rat e of th e previou s da y. Th e authoritie s in ter ven e in th e foreig n ex chang e mar ke t as condition s require . 102 T ur ke y Ne w T urkis h lir a (YT L 1 = T L 1 million , 1/1/2006 ) 2/21/200 1 { curre nt Inde p ende ntl y oatin g Th e lir a w as all ow ed to oat . A s a consequence , th e ex chang e rat e arrangeme nt w as reclassi e d to th e categor y inde p ende ntl y oatin g fro m th e cate gor y cr awl in g p eg (2/22/2001) . Th e ex chang e rat e of th e lir a is determine d on th e basi s of supp ly an d deman d in th e foreig n ex chang e mar ket . Th e Ce ntra l Ban k of th e Republi c of T ur ke y (CB R T ) conduct s dail y auctio ns to buil d up reser ve s here by it buy s a xe d amou nt of dollar s an d pr ovide s th e successfu l bidder s wit h th e optio n to pur chas e up to 200 % of th ei r successfu l bi d amou nt at th e av erag e auctio n price . Th e dail y xe d pur chas e amou nt w as raise d to $2 0 millio n in 2006 . Th e dail y foreig n ex chang e pur chas e auction s w er e sus p end ed on M ay 16 , 2006 , in res p ons e to nancia l mar ke t volatili ty . O n Jun e 26 an d Jun e 27 , 2006 , th e CB R T hel d forei gn ex chan ge auction s unde r whi ch it sol d $50 0 millio n on ea ch da y throug h m ultipl e pric e aucti ons . O n N ov em b er 10 , 2006 , th e CRB T resume d it s dail y foreig n ex chang e auctio n program , wit h a dail y xe d pur chas e amou nt of $1 5 million . Th e CB R T reser ve s th e rig ht to in ter ven e in th e foreig n ex chang e mar ke t in cas e of excessi ve volatili ty in th e foreig n ex chang e rates . T urkis h Lir a De c 199 9 { 2/21/200 1 Cr awlin g Ban d In Dece m b er 1999 , th e Ce ntra l Ban k of T ur ke y (CBT ) m odi e d it s ex chang e arrangeme nt by m ovin g to a preannouncem en t of th e ex chang e rat e pat h of th e lir a aga ins t th e curre nt bas ke t comprisin g th e dolla r an d th e Eur o (i n amou nt s equi vale nt to $1 an d EU R 0.77) . 103 Table A.12: Summary Statistics for Market Forecast Country # month ahead Mean Median Max Min Std. Dev. Czech Republic 1 0.004 0.006 0.024 -0.015 0.008 3 0.004 0.005 0.035 -0.021 0.010 12 -0.002 -0.005 0.092 -0.040 0.022 24 -0.004 -0.007 0.053 -0.052 0.026 Hungary 1 0.005 0.006 0.034 -0.037 0.013 3 0.008 0.010 0.045 -0.034 0.017 12 0.016 0.013 0.077 -0.040 0.030 24 0.013 0.016 0.081 -0.070 0.034 Indonesia 1 0.002 -0.001 0.140 -0.043 0.023 3 0.000 -0.001 0.172 -0.048 0.029 12 -0.005 -0.002 0.169 -0.071 0.038 24 -0.008 -0.007 0.133 -0.111 0.040 Korea 1 0.000 0.000 0.032 -0.034 0.013 3 -0.004 -0.004 0.034 -0.047 0.016 12 -0.024 -0.023 0.016 -0.067 0.018 24 -0.030 -0.028 0.019 -0.084 0.023 Mexico 1 0.006 0.006 0.032 -0.028 0.014 3 0.015 0.017 0.054 -0.023 0.019 12 0.046 0.045 0.108 -0.011 0.029 24 0.076 0.073 0.176 0.020 0.035 Poland 1 -0.001 -0.003 0.056 -0.046 0.022 3 -0.005 -0.007 0.048 -0.058 0.026 12 -0.015 -0.017 0.071 -0.097 0.042 24 0.005 0.001 0.115 -0.104 0.052 South Africa 1 0.009 0.009 0.082 -0.051 0.030 3 0.020 0.016 0.119 -0.056 0.041 12 0.060 0.063 0.216 -0.061 0.069 24 0.110 0.111 0.319 -0.043 0.096 Thailand 1 0.000 0.003 0.024 -0.035 0.012 3 -0.003 -0.001 0.022 -0.045 0.016 12 -0.014 -0.013 0.030 -0.056 0.020 24 -0.022 -0.023 0.024 -0.066 0.022 Turkey 1 0.024 0.023 0.097 -0.066 0.033 3 0.057 0.049 0.165 -0.057 0.048 12 0.164 0.146 0.378 -0.040 0.107 24 0.298 0.305 0.652 -0.013 0.186 104 Table A.13: Summary Statistics for Market Forecaset Dispersion Country # month ahead Mean Median Max Min Std. Dev. Czech Republic 1 0.049 0.042 0.140 0.008 0.029 3 0.061 0.055 0.210 0.026 0.027 12 0.122 0.104 0.385 0.043 0.057 24 0.179 0.157 0.395 0.053 0.071 Hungary 1 0.058 0.055 0.132 0.007 0.031 3 0.067 0.061 0.145 0.009 0.030 12 0.120 0.107 0.230 0.047 0.043 24 0.144 0.150 0.224 0.054 0.039 Indonesia 1 0.121 0.096 0.468 0.045 0.085 3 0.158 0.142 0.518 0.053 0.083 12 0.269 0.259 0.540 0.119 0.108 24 0.346 0.344 0.667 0.154 0.121 Mexico 1 0.063 0.063 0.116 0.022 0.020 3 0.080 0.079 0.174 0.038 0.025 12 0.103 0.101 0.168 0.052 0.024 24 0.110 0.111 0.200 0.030 0.041 Poland 1 0.110 0.098 0.342 0.033 0.053 3 0.132 0.136 0.242 0.053 0.036 12 0.202 0.200 0.337 0.101 0.052 24 0.228 0.216 0.416 0.124 0.071 South Africa 1 0.177 0.171 0.506 0.058 0.074 3 0.226 0.210 0.575 0.081 0.089 12 0.315 0.303 0.807 0.135 0.111 24 0.347 0.323 0.664 0.165 0.120 Korea 1 0.071 0.065 0.183 0.032 0.029 3 0.103 0.100 0.220 0.047 0.026 12 0.175 0.178 0.252 0.104 0.035 24 0.209 0.199 0.324 0.133 0.047 Thailand 1 0.063 0.060 0.206 0.023 0.025 3 0.089 0.085 0.205 0.046 0.031 12 0.130 0.127 0.226 0.073 0.031 24 0.150 0.145 0.369 0.072 0.050 Turkey 1 0.178 0.146 0.449 0.071 0.093 3 0.227 0.194 0.512 0.095 0.111 12 0.382 0.334 0.817 0.161 0.170 24 0.605 0.588 1.320 0.187 0.283 105 Table A.14: Return and Volatility News Response Coe cients Announcements Czech Republic Hungary Indonesia Korea Mexico Poland South Africa Thailand Turkey Impact of News Surprises on FX Returns Business Inventories 0.00 -0.01 0.00 -0.01 0.01* -0.02** -0.03* 0.00 -0.04** Budget De cit 0.00 0.01 0.01 0.14** 0.00 0.00 0.05 0.01 0.00 Current Account 0.07** 0.05** -0.08** 0.01 -0.01 0.01 0.05 0.03* 0.06 Capital Utilization 0.00 0.00 -0.02 -0.02** -0.01 0.01 0.13** 0.01 0.04 Consumer Con dence 0.11** 0.10** 0.01 0.02** -0.03** 0.04** 0.05** 0.00 0.05 Consumer Credit -0.01 -0.01 -0.01 -0.01 0.00 0.00 -0.02 0.00 0.00 Construction Spending 0.03** 0.02** 0.02 0.00 0.00 0.01 -0.02 -0.04** 0.00 Consumer Price Index 0.00 0.02** -0.04** 0.00 0.02** 0.00 0.04** 0.00 0.08** Durable Goods Orders 0.05** 0.06** 0.04** 0.03* -0.01* 0.04** 0.05** -0.03** -0.01 Factory Orders 0.02** 0.02** 0.00 -0.01 0.00 0.00 0.03* 0.00 0.08** Gross Domestic Product 0.10** 0.11** 0.07** 0.02** 0.00 0.07** 0.07** -0.01 0.03 Housing Starts 0.01* 0.01 -0.01 0.01* -0.01 0.01 0.02 -0.01 0.03 Imports 0.00 -0.01 0.02* 0.03** 0.05** -0.01* -0.02 -0.01 -0.05** Interest rate 0.01 -0.01 0.00 0.00 0.01 0.00 0.00 0.01 0.00 Industrial production 0.02** 0.02* 0.03** 0.01 0.00 0.00 -0.01 -0.01 -0.02 NAPM 1.36** 1.27** 0.07 0.34* 0.00 0.65** 0.31 -0.27 0.00 Leading Indicators 0.00 0.02* -0.01 -0.01 -0.01 0.00 -0.01 0.01 0.00 New Home Sales 0.02** 0.03** 0.03** 0.03** 0.01* 0.03** 0.00 -0.01 -0.02 Nonfarm Payroll 0.19** 0.21** 0.04** 0.09** 0.02** 0.18** 0.18** 0.00 0.02* Personal Spending 0.00 0.01 0.01 0.00 0.00 -0.01 0.01 -0.01 0.00 Personal Income 0.00 0.00 -0.01 0.00 0.01 0.00 -0.01 0.00 0.02 Producer Price 0.02** 0.02** 0.04** 0.01 0.01* 0.02** 0.05** 0.00 -0.01 Retail Sales 0.07** 0.07** 0.03** 0.03** -0.07** 0.07** 0.02 0.01 -0.01 Trade Balance 0.12** 0.12** 0.00 0.02** -0.02** 0.12** 0.09** 0.00 0.02 Initial Unemployment -0.03** -0.03** -0.01 -0.01** 0.00 -0.02** -0.02** 0.01 0.01 Wholesales 0.00 0.00 0.00 0.01 -0.01 0.00 0.00 0.00 0.00 Budget De cit 0.05** 0.03 0.01 Current Account -0.03** -0.02* -0.01 -0.08** 0.09 0.22** 0.03 Current Account(US) -0.01 Consumer Con dence -0.01 0.00 Consumer Price Index -0.04** 0.01* 0.01 0.01 0.00 -0.10** -0.01 0.01 Exports -0.03** 0.00 0.01 0.02 0.02 0.09 Fixed Invest 0.00 Gross Domestic Product -0.02 0.01 -0.02 -0.04 -0.02* -0.03** -0.02 -0.04 -0.03 Imports -0.03** 0.00 0.00 -0.01 -0.19** Interest rate -0.30** -0.01 Industrial production -0.01* -0.03** 0.00 0.01 -0.03 Money Supply 0.00 -0.01* 0.15** Producer Price -0.02** -0.02 0.01 0.00 0.01 0.14** Retail Sales -0.03** -0.01 -0.01 0.06 Trade Balance -0.07** 0.00 0.52 -0.01 -0.16** 0.00 Initial Unemployment 0.07 -0.01 -0.02** 0.02* -0.01 Wholesales -0.03 -0.03**106 Announcements Czech Republic Hungary Indonesia Korea Mexico Poland South Africa Thailand Turkey Impact of Contemporaneous News Surprises on Volatility Business Inventories 0.00 0.02** 0.00 0.00 0.01 0.00 -0.01** 0.00 0.05** Budget De cit -0.01 0.01 0.01 0.11** 0.00 0.00 0.01 -0.01 0.13 Current Account -0.02** 0.00 0.02** 0.01 0.00 -0.01** 0.01 0.01 0.00 Capital Utilization 0.03** 0.02** -0.01 0.00 0.00 0.00 -0.02** -0.01* -0.02 Consumer Con dence 0.03** 0.04** 0.03** 0.03** 0.02** 0.03** 0.00 0.01** 0.00 Consumer Credit 0.00 0.00 0.01* 0.00 0.00 0.01 0.00 0.00 -0.08 Construction Spending 0.01** 0.01** -0.02** 0.02* 0.01 0.00 0.01 0.00 0.00 Consumer Price Index 0.05** 0.04** 0.00* 0.00 0.03** 0.04** 0.05** 0.00 0.01 Durable Goods Orders 0.02** 0.05** 0.01* -0.01** 0.00 0.01** 0.01** 0.02** 0.01 Factory Orders 0.04** 0.02** 0.01 0.00* 0.00 0.00 0.00** 0.01 -0.01 Gross Domestic Product 0.05** 0.04** 0.04** 0.04** 0.00 0.04** 0.04** 0.01 0.03** Housing Starts 0.01** 0.02** 0.02** 0.01 0.00 0.00 0.00 0.00 0.00 Imports 0.02** 0.01** 0.00 0.02* 0.01* 0.01 0.00 0.00 0.02* Interest rate 0.00 0.01** 0.00 0.00 0.00 0.00 0.03** 0.00 0.00 Industrial production 0.00 0.00** 0.00 0.00 0.00 0.02** 0.00 -0.01 0.00 NAPM 0.05* 0.05 0.42** 0.10 0.05 0.82** 0.06 0.24 0.02 Leading Indicators 0.01 0.00 0.00 0.00 0.00 0.00 0.01 0.00 0.00 New Home Sales 0.03** 0.03** 0.02** 0.02** 0.00 0.03** 0.00 0.00 -0.01 Nonfarm Payroll 0.17** 0.17** -0.01** 0.09** 0.05** 0.16** 0.18** 0.01** 0.10** Personal Spending 0.02** 0.01 0.01 0.01 0.00 -0.01 0.00** -0.01 0.03** Personal Income 0.01** 0.00 0.00 0.00 0.01** 0.05** 0.01 0.00 0.01 Producer Price 0.03** 0.03** 0.02** -0.01* 0.01** 0.01 0.01** 0.00 0.07** Retail Sales 0.03** 0.03** 0.03** 0.05** 0.03** 0.04** 0.00* 0.00 0.00 Trade Balance 0.04** 0.04** 0.04** 0.04** 0.00** 0.06** 0.03** 0.00 0.01 Initial Unemployment 0.01** 0.00 0.01** 0.01** 0.01** 0.02** 0.02** 0.00 0.01** Wholesales 0.01** 0.01 0.00 0.02** 0.00 -0.01** 0.00 0.00 0.00 Budget De cit 0.00 -0.01 0.00 Current Account 0.00 0.01 -0.02* 0.02** 0.06 0.24** 0.00 Current Account(US) 0.03** Consumer Con dence 0.00 0.00 Consumer Price Index 0.04** 0.04** 0.00 0.01* 0.00 0.01 0.00 0.00 Exports 0.02 0.00 0.00 0.00 0.00 0.00 Fixed Invest 0.00 Gross Domestic Product 0.01 -0.01 -0.02 -0.02 0.00 0.02** 0.03* -0.01 0.00 Imports 0.02** 0.00 0.01 0.02** 0.09* Interest rate 0.26** 0.00 Industrial production 0.01** 0.01** 0.00 0.00 0.00 Money Supply 0.00** 0.01* 0.03** Producer Price 0.00 -0.01 0.00* 0.01 0.02* 0.19** Retail Sales 0.00 -0.01** -0.01 -0.04* Trade Balance 0.03** 0.01 0.17 -0.01 0.07** -0.02 Initial Unemployment -0.04** 0.00** 0.02** 0.00 0.00 Wholesales 0.00 0.03** 107 Announcements Czech Republic Hungary Indonesia Korea Mexico Poland South Africa Thailand Turkey Impact of Cumulated news Surprises on volatility Business Inventories 0.01 0.04** -0.01 -0.01 0.01 0.00 0.04** 0.00 0.10** Budget De cit -0.02 0.06 -0.04 0.11** -0.01 -0.01 0.05 -0.01 -0.12 Current Account 0.03** 0.00 -0.05** 0.03 0.00 0.05** 0.00 0.01 -0.02 Capital Utilization 0.11** 0.05** 0.02 0.01 0.01 0.01 0.08** -0.07* -0.02 Consumer Con dence 0.03** 0.10** 0.13** 0.07** 0.05** 0.12** 0.01 0.02** -0.04 Consumer Credit 0.00 0.00 -0.04* 0.00 -0.01 -0.04 0.01 0.00 0.75 Construction Spending 0.00** 0.00** 0.09** 0.01* 0.02 0.00 0.03 0.00 0.00 Consumer Price Index 0.13** 0.08** 0.06* 0.01 0.09** 0.05** 0.12** 0.00 0.08 Durable Goods Orders 0.17** 0.20** 0.05* 0.05** 0.00 0.07** 0.13** 0.02** 0.03 Factory Orders 0.13** 0.06** 0.02 0.02* 0.01 0.02 0.07** -0.02 -0.05 Gross Domestic Product 0.06** 0.02** 0.05** 0.04** 0.00 0.11** 0.08** 0.03 -0.02** Housing Starts 0.02** 0.03** 0.04** 0.00 0.00 0.00 0.01 0.00 0.01 Imports 0.14** 0.06** -0.01 0.04* 0.02* 0.00 -0.01 -0.01 0.01* Interest rate -0.01 0.01** 0.00 0.00 0.01 0.01 -0.09** -0.01 0.00 Industrial production 0.00 0.04** 0.01 -0.01 0.00 0.06** 0.01 -0.03 -0.03 NAPM -0.11* 0.08 2.57** 0.10 -0.11 2.43** 0.46 0.03 0.00 Leading Indicators 0.01 0.00 0.00 0.00 0.00 0.02 0.01 0.01 0.00 New Home Sales 0.10** 0.06** 0.05** 0.03** -0.01 0.08** 0.01 0.00 -0.01 Nonfarm Payroll 0.26** 0.32** 0.16** 0.39** 0.19** 0.37** 0.38** 0.05** 0.16** Personal Spending 0.08** 0.00 0.01 0.02 0.00 -0.01 0.09** -0.02 0.03** Personal Income 0.03** 0.00 0.00 0.00 0.03** 0.14** -0.01 0.00 0.02 Producer Price 0.03** 0.06** 0.04** 0.04* 0.01** 0.01 0.10** -0.01 0.08** Retail Sales 0.09** 0.05** 0.06** 0.13** 0.03** 0.16** 0.06* 0.00 0.01 Trade Balance 0.05** 0.06** 0.01** 0.25** 0.02** 0.11** 0.11** -0.03 0.02 Initial Unemployment 0.03** 0.00 -0.01** 0.01** 0.04** 0.12** 0.07** 0.00 0.01** Wholesales 0.02** 0.00 0.00 0.00** -0.02 0.04** 0.01 -0.01 -0.02 Budget De cit 0.00 -0.01 0.03 Current Account 0.00 -0.01 -0.06* 0.06** 0.09 0.40** 0.01 Current Account(US) 0.11** Consumer Con dence -0.01 0.00 Consumer Price Index 0.14** 0.12** 0.01 -0.01* 0.00 0.00 0.00 0.00 Exports 0.03 0.02 0.00 -0.03 -0.01 0.01 Fixed Invest 0.00 Gross Domestic Product -0.01 0.01 -0.03 -0.04 0.00 0.09** 0.07* 0.01 -0.16 Imports 0.12** 0.04 0.01 0.13** 0.23* Interest rate 0.31** -0.03 Industrial production 0.05** 0.07** 0.00 0.01 0.00 Money Supply 0.00** 0.04* 0.07** Producer Price 0.00 -0.01 0.07* 0.01 0.02* 0.49** Retail Sales 0.00 -0.07** 0.01 0.04* Trade Balance 0.10** 0.01 1.96 -0.02 0.02** -0.02 Initial Unemployment -0.07** 0.19** 0.01** -0.01 0.01 Wholesales 0.04 0.18** 108 Table A.15: Return and Volatility Response with Announcement Dummy Announcements xvar id Czech Republic Hungary Indonesia Korea Mexico Poland South Africa Thailand Turkey U.S. Contemporaneous Announcements in equation (1.7) Business Inventories k0 -0.01 -0.01 0.00 -0.01 0.02*** -0.03*** -0.04*** 0.00 -0.07*** k0 0.00 0.00 0.01 0.01 -0.02*** 0.00 -0.03** 0.01 0.03 Budget De cit k0 0.00 0.02 0.01 0.13*** 0.00 0.00 0.05 0.02 0.00 k0 0.01* 0.00 0.00 0.03*** 0.00 0.00 0.01 0.00 0.00 Current Account k0 0.04*** 0.05*** -0.08*** 0.01 -0.01 0.03** 0.11*** 0.03** 0.07** k0 0.00 -0.02 -0.03*** 0.00 0.01 0.02** 0.09*** 0.00 0.02 Capital Utilization k0 0.00 -0.01 -0.02** -0.03*** 0.00 0.01 0.12*** 0.00 0.05 k0 0.01 -0.02 -0.02 0.03* 0.01 0.02 0.02 -0.01 -0.02 Consumer Con dence k0 0.11*** 0.10*** 0.01 0.01 -0.03*** 0.05*** 0.05*** 0.00 0.05 k0 0.00 -0.01 -0.01 0.04*** 0.01** -0.02*** 0.00 0.00 -0.02 Consumer Credit k0 -0.01 -0.01 -0.01 -0.01 0.00 0.00 -0.02 0.00 0.01 k0 0.00 0.01 0.00 0.00 0.01 0.00 0.00 -0.01 0.00 Construction Spending k0 0.03*** 0.02*** 0.01 0.00 0.00 0.02** -0.01 -0.04*** 0.01 k0 -0.03** -0.01 -0.05** -0.09*** -0.03** -0.01 -0.10*** -0.02 0.00 Consumer Price Index k0 0.00 0.02*** -0.05*** -0.01 0.02*** 0.01** 0.04*** 0.00 0.08*** k0 -0.03*** -0.02*** -0.01 0.01 0.01 -0.01 -0.03** 0.01 0.06*** Durable Goods Orders k0 0.06*** 0.07*** 0.05*** 0.02*** -0.02*** 0.05*** 0.05*** -0.03*** -0.01 k0 -0.01 -0.02*** 0.00 0.01 0.01** -0.02*** -0.02 0.03*** 0.04* Factory Orders k0 0.03*** 0.02*** 0.00 -0.01** 0.00 0.00 0.03*** 0.00 0.07*** k0 0.00 0.00 -0.02** 0.02*** -0.01** 0.00 0.00 0.01 0.03 Gross Domestic Product k0 0.10*** 0.11*** 0.06*** 0.02*** 0.00 0.07*** 0.07*** 0.00 0.03* k0 -0.03*** -0.04*** -0.04*** -0.03*** -0.01** -0.01 -0.02* 0.01 0.07*** Housing Starts k0 0.02*** 0.02*** -0.01 0.02** 0.00 0.01** 0.04*** -0.01 0.03* k0 0.00 -0.02** -0.03*** -0.02*** -0.02*** -0.01* -0.02* 0.00 -0.01 Imports k0 0.01 0.00 0.01 0.02*** 0.06*** 0.00 -0.03*** -0.01 -0.04*** k0 0.03*** 0.02*** -0.02*** 0.02** 0.03*** 0.04*** -0.03*** -0.01 0.10*** Interest rate k0 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 k0 0.00 0.05* 0.00 0.00 0.00 0.00 0.00 0.00 0.00 Industrial production k0 0.02*** 0.02** 0.03** 0.01** 0.00 0.01 -0.01 0.00 -0.02 k0 -0.02** -0.01 0.00 -0.03 -0.01 -0.02 0.04 0.00 0.02 NAPM k0 1.39*** 1.42*** 0.12 0.56*** -0.07 0.87*** 0.17 -0.31** 0.04 k0 0.04*** 0.04*** 0.06** 0.09*** 0.02 0.01 0.12*** 0.05** 0.03 Leading Indicators k0 0.01 0.02*** -0.01 -0.01 -0.01 0.00 0.00 0.01 0.00 k0 0.01 0.00 0.01 0.01 0.01 0.01 -0.01 0.00 -0.03 New Home Sales k0 0.02*** 0.03*** 0.02*** 0.03*** 0.01** 0.03*** 0.00 -0.01 -0.03 k0 -0.01** 0.03*** 0.01 -0.02*** -0.01** 0.01 0.01 -0.01 0.03 Nonfarm Payroll k0 0.22*** 0.25*** 0.06*** 0.09*** 0.01*** 0.20*** 0.18*** 0.01 0.05*** k0 0.05*** 0.06*** 0.02*** 0.04*** -0.05*** 0.02*** -0.01 0.01 0.10*** Personal Spending k0 0.00 0.02* 0.01 0.00 0.00 0.00 0.00 -0.01 0.03 k0 0.00 0.00 0.01 0.03** 0.03*** 0.09*** -0.03 0.01 0.15*** Personal Income k0 0.00 0.01 -0.02 0.00 0.01 0.01 -0.01 0.00 0.02 k0 -0.03*** -0.03*** 0.00 -0.02** -0.05*** -0.09*** 0.04** 0.01 -0.07** Producer Price k0 0.00 0.01 0.03*** 0.02*** 0.01** 0.01** 0.04*** 0.01 0.00 k0 -0.02*** -0.03*** -0.02** -0.01 -0.03*** -0.02*** 0.00 0.01 0.09*** 109 Announcements xvar id Czech Republic Hungary Indonesia Korea Mexico Poland South Africa Thailand Turkey U.S. Contemporaneous Announcements in equation (1.7) Retail Sales k0 0.09*** 0.08*** 0.03*** 0.03*** -0.07*** 0.08*** 0.03** 0.01 -0.01 k0 0.00 0.00 0.00 0.00 0.02*** 0.00 -0.06*** 0.01 -0.04** Trade Balance k0 0.12*** 0.11*** 0.00 0.02*** -0.02*** 0.11*** 0.10*** 0.00 0.03* k0 0.01 0.00 -0.02** 0.02*** 0.00 0.01** 0.01 0.01 0.05** Initial Unemployment k0 -0.04*** -0.03*** -0.01** -0.02*** 0.00 -0.02*** -0.02*** 0.01 0.01 k0 -0.01*** -0.01** 0.00 0.00 0.00 -0.01*** -0.01 -0.01* 0.01 Wholesales k0 0.00 0.00 0.00 0.01 -0.01 0.00 0.01 0.00 0.01 k0 -0.01*** -0.01* -0.01 -0.04*** 0.00 -0.02** -0.01 0.00 -0.02 Domestic Contemporaneous Announcements in equation (1.7) Budget De cit k0 0.01 0.02 0.00 k0 0.11*** 0.00 -0.02** Current Account k0 -0.02** -0.01** 0.00 -0.08*** 0.00 0.18*** 0.02 k0 0.01 0.02** 0.00 -0.01 -0.38** -0.07*** -0.02 Current Account(US) k0 -0.01 k0 0.04*** Consumer Con dence k0 0.00 0.00 k0 0.01 0.00 Consumer Price Index k0 -0.01 0.02*** 0.01 0.01 -0.01 -0.09*** -0.01 0.03 k0 -0.01 -0.04*** -0.01 0.00 -0.02* -0.01 0.00 -0.13* Exports k0 -0.01 -0.05*** 0.00 0.02** 0.09 -0.01 k0 -0.15 0.00 0.00 0.00 0.01 -0.18*** Fixed Invest k0 0.00 k0 0.00 Gross Domestic Product k0 -0.01 0.01 -0.03 -0.05 -0.02** -0.03** -0.03 -0.08*** -0.03 k0 -0.03*** 0.00 0.01 -0.01 -0.01 0.00 -0.08** -0.06*** 0.01 Imports k0 -0.07 0.04** 0.00 -0.01 -0.10* k0 0.12 0.22*** 0.00 0.00 0.00 Interest rate k0 -0.14*** -0.01 k0 0.58*** 0.00 Industrial production k0 -0.01 -0.03*** 0.00 0.01 -0.03 k0 -0.01 0.00 0.00 0.01 -0.02 Money Supply k0 0.02 -0.01 0.09*** k0 -0.15 0.00 0.10*** Producer Price k0 -0.02*** 0.01 0.01 0.00 0.01 0.17*** k0 -0.01** -0.02** -0.01 -0.09*** 0.03* 0.24*** Retail Sales k0 -0.03*** -0.01 -0.01 0.12 k0 -0.01 0.01 -0.03*** 0.00 Trade Balance k0 -0.07*** -0.01 4.97*** -0.01** -0.17*** 0.03 k0 -0.03*** 0.01 -0.21*** 0.00 0.04*** 0.04 Initial Unemployment k0 0.01 -0.01 -0.01*** 0.02** -0.01 k0 0.02*** 0.00 -0.01 0.03** -0.01 Wholesales k0 -0.03* -0.03*** k0 0.00 0.06*** 110 Announcements xvar id Czech Republic Hungary Indonesia Korea Mexico Poland South Africa Thailand Turkey U.S. Contemporaneous Announcements in equation (1.8) Business Inventories k0 0.00 -0.01 0.01 0.01 0.01 0.01 -0.01** 0.00 0.06*** k0 0.03*** 0.04*** 0.02 0.02 -0.01 0.02 0.02* -0.01 -0.04 Budget De cit k0 0.00 0.01 0.01 0.15*** 0.00 0.00 0.01 0.03 0.11 k0 0.00 0.00 -0.01 -0.01 0.00 -0.01 -0.02 -0.01 0.00 Current Account k0 0.01 -0.02 0.03** -0.02 0.00 -0.02** 0.01 0.02 -0.02 k0 -0.01 0.03* 0.04** 0.03* 0.01 0.03* 0.00 0.00 0.01 Capital Utilization k0 -0.01 -0.02 0.01 0.01 -0.01 0.00 -0.02 0.00 -0.01 k0 0.01 0.00 0.01 0.02 0.02 0.01 0.03 0.01 -0.03 Consumer Con dence k0 0.02*** 0.03*** 0.00 0.00 0.01** 0.01 0.00 0.01*** 0.00 k0 0.03*** 0.02*** 0.03** 0.04** 0.00 0.04*** 0.01 -0.01 0.00 Consumer Credit k0 -0.01 -0.01 0.00 -0.01 0.00 0.00 0.00 -0.01 -0.01 k0 0.01 0.00 -0.01 0.00 0.00 0.00 0.01 0.00 -0.07 Construction Spending k0 -0.01* 0.00 0.00 0.01 0.00 0.00 -0.01 0.00 0.00 k0 0.03*** 0.03** 0.00 0.01 0.02* 0.01 0.00 0.00 0.02 Consumer Price Index k0 0.01 0.00 0.00** -0.02** 0.00 -0.03* -0.03 0.00 -0.01* k0 0.04*** 0.05*** 0.03*** 0.03* 0.05*** 0.09*** 0.09*** 0.01 0.10*** Durable Goods Orders k0 -0.01 0.00 0.00 -0.01 0.00 0.00 0.00 0.03*** 0.01 k0 0.05*** 0.05*** 0.04*** 0.03*** 0.02*** 0.04*** 0.01 0.00 -0.01 Factory Orders k0 -0.01 -0.01** -0.03** 0.00 0.00 0.00 0.00 0.00 0.01 k0 0.06*** 0.05*** 0.05*** 0.05*** 0.01 0.03*** 0.02* 0.00 -0.03* Gross Domestic Product k0 0.02*** 0.03*** 0.03*** 0.02** 0.00 0.01* 0.03** 0.02*** -0.01 k0 0.04*** 0.03*** 0.01 0.01 0.02*** 0.03*** 0.01 -0.02* 0.06*** Housing Starts k0 -0.01*** -0.01 0.00 -0.03*** 0.00 0.00 -0.01 0.00 0.00 k0 0.04*** 0.03*** 0.02 0.06*** 0.01 0.02** 0.01 0.00 0.02 Imports k0 -0.01 0.00 0.00 0.01 -0.01* 0.02** 0.00 0.00 0.01 k0 0.01 0.00 0.03** 0.00 0.02** 0.00 0.04*** 0.00 0.03 Interest rate k0 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 k0 -0.03 0.06** 0.00 0.00 -0.03 -0.03 -0.05 -0.04 0.00 Industrial production k0 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 -0.01 k0 0.03*** 0.03*** -0.01 0.02 0.00 -0.01 0.01 -0.02 0.03 NAPM k0 0.08*** -0.03 0.03 -0.46** 0.01 0.06 -0.04 0.17 -0.26 k0 0.03*** 0.04*** 0.04 0.06** -0.01 0.03 0.02 0.00 0.01 Leading Indicators k0 -0.01* 0.00 0.00 0.00 0.00 0.00 -0.01 0.00 0.01 k0 0.04*** 0.02*** 0.01 0.04*** 0.01 0.00 0.03 -0.01 -0.04 New Home Sales k0 -0.02*** -0.02** 0.00 -0.01 0.00 0.01 0.00 0.00 0.01 k0 0.07*** 0.05*** 0.03** 0.06*** 0.01 0.02*** 0.03*** 0.00 -0.03 Nonfarm Payroll k0 0.07*** 0.06*** 0.00 0.02* 0.02*** 0.08*** 0.12*** 0.01 0.04** k0 0.17*** 0.16*** 0.08*** 0.08*** 0.06*** 0.12*** 0.08*** 0.00 0.07*** Personal Spending k0 0.00 0.00 0.00 0.01 -0.01 -0.01 0.00 -0.01 0.00 k0 -0.01 0.01 -0.01 -0.01 -0.01 -0.01 0.00 0.00 0.00 Personal Income k0 0.00 0.00 -0.01 0.00 -0.01 0.01 0.00 0.01 -0.01 k0 0.04*** 0.02** 0.02 0.03** 0.03*** 0.05*** 0.04** -0.01 0.05* Producer Price k0 -0.02*** -0.02*** 0.00 0.00 -0.01* 0.00 0.00 0.00 -0.01 k0 0.06*** 0.06*** 0.04*** 0.02 0.03*** 0.03** 0.04*** -0.01 0.08*** 111 Announcements xvar id Czech Republic Hungary Indonesia Korea Mexico Poland South Africa Thailand Turkey Retail Sales k0 0.01* 0.01 0.01 0.02 0.03*** 0.02*** 0.00 0.00 -0.04* k0 0.04*** 0.04*** 0.01 0.02** 0.03*** 0.02** 0.06*** 0.00 0.04 Trade Balance k0 0.02*** 0.02*** 0.01 0.03*** 0.00 0.03*** 0.03*** -0.01** -0.03 k0 0.04*** 0.04*** 0.05*** 0.00 0.02** 0.04*** 0.02 0.03*** 0.04* Initial Unemployment k0 0.00 0.00 0.01 0.00 0.00 -0.01 0.00 0.00 0.00 k0 0.02*** 0.02*** 0.01 0.02*** 0.01*** 0.03*** 0.01** 0.00 0.01 Wholesales k0 -0.01 -0.01* 0.00 0.00 0.00 0.01 0.00 -0.01 0.00 k0 0.04*** 0.04*** 0.02 0.02 0.02*** 0.02* 0.03*** 0.02 -0.03 Domestic Contemporaneous Announcements in equation (1.8) Budget De cit k0 0.01 -0.02* 0.00 k0 -0.01 0.03** 0.04*** Current Account k0 -0.01* 0.02*** -0.01 -0.01 -0.11 0.37*** 0.00 k0 0.01 0.03*** 0.00 -0.04 0.02 -0.05** 0.01 Current Account(US) k0 0.01 k0 0.01 Consumer Con dence k0 0.00 0.00 k0 -0.01 0.00 Consumer Price Index k0 -0.02* 0.01 0.00 0.00 -0.01 0.01 0.00 -0.01 k0 0.04*** 0.03*** -0.01 0.00 0.02* 0.02 0.00 -0.01 Exports k0 0.01 0.00 0.00 0.00 0.00 -0.07 k0 -0.04 0.07*** 0.02 0.10* 0.00 0.06 Fixed Invest k0 0.00 k0 -0.01 Gross Domestic Product k0 0.01*** 0.00 -0.02 -0.01 0.01 0.02 0.01 -0.08*** 0.00 k0 0.02** 0.01 0.00 -0.01 -0.01 0.01 -0.01 0.16*** 0.01 Imports k0 -0.01 0.00 0.01 0.00 -0.02 k0 0.04 -0.26*** -0.04 0.02 0.00 Interest rate k0 0.00 0.00 k0 0.42*** -0.03 Industrial production k0 0.01** -0.01* 0.00 0.01*** 0.00 k0 0.00 0.03*** -0.01 0.00 -0.01 Money Supply k0 0.00 -0.02 0.04*** k0 0.02 0.01 0.00 Producer Price k0 0.00 -0.01 0.00*** -0.02 0.00 0.16*** k0 0.01 0.02*** 0.01 0.04* 0.01 0.02 Retail Sales k0 0.01 0.00 0.01 -0.08*** k0 0.02*** -0.01 0.02* 0.05 Trade Balance k0 0.01* 0.00 0.17 0.01 0.09*** -0.03 k0 0.02*** 0.01 0.19*** -0.01 0.00 0.05 Initial Unemployment k0 0.01 0.00 0.03*** -0.01 -0.01 k0 0.00 -0.01 -0.01 0.00 -0.02 Wholesales k0 0.00 0.01 k0 0.01 0.00 112 Table A.16: Regression Results with Expected Appreciation Announcements Czech Republic Hungary Indonesia Korea Mexico Poland South Africa Thailand Turkey Impact of Contemporaneous News Surprises on FX Return Business Inventories -0.01 -0.01 0.01 -0.01 0.01 -0.01 -0.04** 0.00 -0.03 Budget De cit -0.01 0.05 0.01 0.02 -0.01 0.00 0.04 0.02 0.00 Current Account 0.10** 0.10** -0.06** 0.01 -0.01 0.03* 0.03 0.03 0.03 Capital Utilization -0.01 -0.02 -0.02 -0.03** 0.01 0.01 0.08** 0.00 0.06* Consumer Con dence 0.12** 0.10** 0.02 0.02 0.02 0.07** 0.05** 0.00 0.10 Consumer Credit -0.01 -0.01 -0.01 -0.01 -0.01 0.00 -0.02 -0.01 0.31 Construction Spending 0.06** 0.02 0.01 0.01 0.00 0.03** -0.02 -0.05** -0.01 Consumer Price Index -0.02 0.02 -0.04** 0.01 0.02* 0.01 0.05** 0.00 0.05 Durable Goods Orders 0.05** 0.07** 0.06** 0.02* 0.01 0.06** 0.06** -0.02 0.04 Factory Orders 0.05** 0.02 0.00 -0.03* 0.01 0.00 0.03* 0.00 0.02 Gross Domestic Product 0.09** 0.09** 0.06** 0.02 0.01 0.07** 0.09** 0.00 0.00 Housing Starts 0.02 0.03* -0.01 0.02* 0.01 0.01 0.02 -0.01 -0.03 Imports 0.01 0.01 0.03* 0.02 0.00 0.00 -0.02 -0.01 -0.05* Interest rate 0.01 0.02 0.00 0.00 0.01 0.00 0.00 0.01 0.00 Industrial production 0.07** 0.04 0.03* 0.02 0.00 0.01 -0.02 0.01 -0.01 NAPM 0.61* 0.83** 0.40 0.72** 0.24 0.78** 0.55 -0.23 0.52 Leading Indicators -0.01 0.02 0.00 -0.01 0.00 0.00 -0.01 0.00 0.00 New Home Sales -0.02 0.03* 0.04** 0.03** 0.01 0.03** -0.01 -0.01 -0.04 Nonfarm Payroll 0.19** 0.21** 0.07** 0.08** 0.11** 0.23** 0.26** 0.01 0.14** Personal Spending -0.01 0.01 0.02 0.00 0.01 0.00 -0.01 -0.01 0.00 Personal Income 0.03 0.02 0.00 0.01 0.01 0.00 0.03* 0.00 0.00 Producer Price -0.01 -0.01 0.02* 0.01 0.01 0.01 0.03 0.00 0.02 Retail Sales 0.09** 0.07** 0.04** 0.04** 0.02 0.08** 0.03* 0.01 -0.01 Trade Balance 0.04** 0.06** -0.01 0.01 -0.01 0.11** 0.09** 0.01 0.03 Initial Unemployment -0.04** -0.03** -0.01 -0.02** -0.02** -0.02** -0.02** 0.01 0.01 Wholesales 0.00 0.01 0.01 0.01 -0.03 0.00 0.00 0.00 -0.01 Budget De cit -0.05 -0.34 0.01 Current Account -0.02 -0.04* -0.07 -0.07** 0.34** 0.03 0.03 Current Account(US) -0.01 Consumer Con dence -0.01 0.00 Consumer Price Index -0.03* 0.05** 0.01 0.01 0.01 -0.04* -0.01 1.74** Exports -0.41** -0.01 0.01 0.00 0.01 -0.20 Fixed Invest 0.00 Gross Domestic Product 0.01 -0.16 -0.01 -0.05 0.00 -0.06** -0.08 -0.16** -0.03 Imports 0.08 0.02 0.00 -0.01 0.00 Interest rate -0.36** -0.01 Industrial production -0.01 -0.02 0.00 -0.02 -0.02 Money Supply 0.00 -0.01 0.00 Producer Price -0.02 -0.02* 0.01 0.00 0.02 0.14** Retail Sales -0.02 -0.01 -0.01 0.01 Trade Balance -0.09** -0.01 2.03 -0.01 -0.14** 0.06 Initial Unemployment 0.35** -0.04 -0.01 0.01 0.00 Wholesales -0.02 -0.03** 113 Impact of Contemporaneous News Surprises Multiplied by Appreciation Expectation Business Inventories -0.05 -0.05 -1.37* -0.29 0.00 -0.77* -0.28 1.05 0.00 Budget De cit -0.06 -0.31* -0.26 7.53** -0.01 -0.34 0.53 -0.42 0.00 Capital Utilization 0.01 0.11 0.04 -1.88* 0.04* -0.03 -1.20** 0.22 0.00 Current Account -0.23* -0.26 1.41 0.09 -0.02 0.38 0.04 -1.53 0.00 Consumer Credit -0.02 -0.08 0.23 -0.18 -0.01 -0.15 0.35 0.65 -6.12 Consumer Con dence -0.11* -0.03 0.52 2.35** 0.05** -1.02** 0.15 0.65 0.00 Construction Spending -0.18* 0.01 -0.06 -1.57* 0.00 0.96* 0.68* 2.34* 0.00 Consumer Price Index 0.18** 0.06 -2.67** 0.23 0.00 -0.46 0.28 0.96 0.00 Durable Goods Orders 0.22** 0.19** 1.08* 0.39 0.04** 0.10 0.29 -1.04 0.00* Factory Orders -0.12** 0.02 0.05 1.68* 0.02 -0.03 0.34 0.14 0.00 Gross Domestic Product 0.20** 0.24** 0.87* 0.81* 0.03 0.17 -0.51 -0.18 0.00 Housing Starts 0.02 -0.08 -0.21 -0.93 0.01 0.46 -0.19 -0.47 0.00* Imports 0.02 -0.04 3.19** 0.39 -0.04** 0.05 -1.26** -0.59 0.00 Interest rate 0.00 0.12 0.00 0.00 0.00 0.00 0.00 0.00 0.00 Industrial production -0.21** -0.09 0.29 1.62 -0.01 0.24 -0.20 -0.21 0.00 NAPM 2.55* 2.20 12.73 17.15 0.45 3.42 -7.16 12.46 0.00* Leading Indicators 0.09 0.02 0.39 0.49 0.02 0.40 0.26 0.38 0.00 New Home Sales 0.18** -0.01 0.89* -0.75 0.01 0.00 0.34 0.28 0.00 Nonfarm Payroll 0.73** 0.65** 1.36** 1.07* 0.10** -0.52* -0.13 -1.80** 0.00** Personal Spending 0.08 0.00 0.26 0.26 0.01 0.17 0.06 -0.19 0.00 Personal Income -0.14* -0.10 0.50 0.93 0.01 -0.43 1.57** -0.14 0.00 Producer Price 0.13** 0.18** 0.50 0.16 0.01 0.24 0.11 0.43 0.00* Retail Sales 0.06 0.13* 1.85** -0.29 0.12** 0.30 0.39 -0.49 0.00 Trade Balance 0.38** 0.31** -0.74 1.12* -0.01 -0.84* -1.00** 0.06 0.00 Initial Unemployment -0.01 -0.04 -0.09 -0.16 -0.03** 0.44** 0.14 0.32 0.00 Wholesales -0.01 -0.06 0.34 -1.28* -0.03 -0.28 0.76** 0.11 0.00 Budget De cit 0.09 1.52 0.53 Current Account -0.03 -0.51** -0.07 1.32** -8.49** 11.61** 0.00 Current Account(US) -0.09 Consumer Con dence -0.01 0.00 Consumer Price Index 0.15* -0.06 0.24 -0.01 -0.40 1.00** -0.26 0.00** Exports 1.42** -0.34 -0.49 -1.07* -1.64 8.92 Fixed Invest -0.02 Gross Domestic Product -0.38** 0.48 -0.02 -1.84 0.04 2.02* 0.30 -5.16 0.00 Imports 1.29 0.77 0.29 1.99** 9.11** Interest rate -7.96** 0.00 Industrial production 0.00 -0.08 0.11 -0.02 0.00 Money Supply -0.10 0.10 -2.15** Producer Price -0.06 0.20 0.00 -0.39 -0.83* -2.74** Retail Sales 0.05 0.02 -0.21 -0.21 Trade Balance 0.08 0.05 51.91 0.01 -0.28 0.00 Initial Unemployment -1.87** -2.18 0.02 0.22 0.00 Wholesales 0.00 -0.10 114 A.7 Figures Figure A.1: IS and BP Curves under Di erent Exchange Regimes 115 Figure A.2: Impulse Response under the Fixed Exchange Regime with a Deferred Wage Shock 116 Figure A.3: Impulse Response under the Flexible Exchange Regime with a Deferred Dage Shock 117 Figure A.4: Impulse Response under the Fixed Exchange Regime with a Technology Shock 118 Figure A.5: Impulse Response under the Flexible Exchange Regime with a Tech- nology Shock 119 Figure A.6: Impulse Response under the Fixed Exchange Regime with a World Interest Rate Shock 120 Figure A.7: Impulse Response under the Flexible Exchange Regime with a World Interest Rate Shock 121 Figure A.8: Impulse Response of Consumptions with a positive deferred wage Shock 122 Figure A.9: Impulse Response of Composite Price with a Positive Deferred Wage Shock 123 Figure A.10: Impulse Response of Risk Premium with a Positive Deferred Wage Shock 124 Figure A.11: Impulse Response of Risk Premium with Technology Shock 125 Figure A.12: Impulse Response of Risk Premium with Interest Rate Shock 126 Figure A.13: Impulse Response under the Fixed Exchange Regime with Deferred Wage shock 127 Figure A.14: Impulse Response under the Flexible Exchange Regime with Deferred Wage shock 128 Figure A.15: Impulse Response of Composite price with Deferred Wage shock 129 Figure A.16: Impulse Response of Risk Premium with Deferred Wage shock 130 Figure A.17: Sample Autocorrelation Graphs of 5-minute Returns across Countries 131 Figure A.18: Sample Autocorrelation Graphs of 5-minute Absolute Returns across Countries 132 Figure A.19: Evolution of EM Exchange Rates Responses to U.S. News 133 134 Figure A.20: U.S. News Impact curve 135 Bibliography [1] Aghion, Philippe, Philippe Bacchetta, Romain Ranciere, and Kenneth Rogo , Exchange Rate Volatility and Productivity Growth: The Role of Financial De- velopment, Journal of Monetary Economics, 56, pp.494-513, 2009. [2] Andersen, Torben G., Tim Bollerslev, Francis X. Diebold, and Clara Vega, 2003, Micro E ects of Macro Announcements: Real-Time Price Discovery in Foreign Exchange, The American Economic Review, 93(1), pp.38-62. [3] Andersen, Torben G., Tim Bollerslev, Francis X. Diebold, and Clara Vega, 2007, Real-Time Price Discovery in Global Stock, Bond and Foreign Exchange Markets, Journal of International Economics, 73(2), pp.251-277. [4] Andritzky, Jochen R, Geo rey J. Bannister, and Natalia T. Tamirisa, 2007, The impact of macroeconomic announcements on emerging market bonds, Emerging Markets Review, 8(1), pp.20-37. [5] Aysun, Uluc, Automatic Stabilizer Feature of Fixed Exchange Rate Regimes in Emerging Markets, Economics Working Papers, University of Connecticut, 2006. [6] Barberis, N. and R. Thaler, 2003, A Survey of Behavioral Finance, Handbook of the Economics of Finance, volume 1, part 2, pp.1053-1128. [7] Bental, Benjamin, and Dominique Demougin, Incentive Contract and Total Factor Productivity, manuscript, October 2004. [8] Bernanke, Ben S., Mark Gertler, and Simon Gilchrist, The Financial Accel- erator in a Quantitative Cusiness Cycle Framework, Handbook of Macroeco- nomics, Vol. 1, Chapter 21, 1999. [9] Bollerslev, Tim, Jun Cai, Frank M. Song, 2000, Intraday Perodicity, Long Mem- ory Volatility, and Macroeconomic Announcement E ects in the US Treasury Bond Market, Journal of Empirical Finance, 7(1), pp.37-55. [10] Calvo, Guillermo A., Carmen Reinhart, Fear of Floating, Quarterly Journal of Economics, May 2002, Vol. CXVII, pp.379 - 408. [11] Cespedes, Luis Felipe, Roberto Chang, and Andres Velasco, Balance Shets and Exchange Rate Policy, American Economic Review, Vol. 94 No. 4, September 2004. 136 [12] Chaboud, A., Chernenko, S., Howorka, E., Iyer, R., Liu, D., Wright, J., 2004, The high-frequency E ects of U.S. Macroeconomic Data Releases on Prices and Trading Activity in the Global Interdealer Foreign Exchange Market, Board of Governors of the Federal Reserve System, IFDP No. 823. [13] Cheung, Yin-Wong, Menzie D. Chinn, Ian W. Marsh, 2004, How Do UK-based Foreign Exchange Dealers Think Their Market Operates?, International Journal of Finance and Economics, 9(4), pp.289-306. [14] Choi, Woon gyu, and David Cook, Liability dollarization and the Bank Balance Sheet Channel, Journal of International Economics, 2004, pp. 247 - 275. [15] Cochrane, John H., Monika Piazzesi, Bond Risk Premia, The American Eco- nomic Review, Mar 2005, Vol. 95, No. 1, pp.138-160. [16] Devereux, Michael B., Lane, and Xu, Exchange Rates and Monetary Policy in Emerging Market Economies, The Economic Journal, 2006, 116 (April), 478 - 506. [17] Dominguez, Kathryn M.E., Freyan Panthaki, 2006, What de nes News in For- eign Exchange Markets?, Journal of International Money and Finance, 25(1), pp. 168-198. [18] Ederington, Louis H., Jae Ha Lee, 1996, The Creation and Resolution of Market Uncertainty: The Impact of Information Releases on Implied Volatility, The Journal of Financial and Quantitative Analysis, 31(4), pp.513-539. [19] Ehrmann, Michael, Marcel Fratzscher, 2005, Exchange Rates and Fundamen- tals: New Evidence from Real-time Data, Journal of International Money and Finance, 24(2), pp.317-341. [20] Evans, Martin D.D., Richard K. Lyons, 2005, Do Currency Markets Absorb News Quickly?, Journal of International Money and Finance, 24(2), pp.197- 217. [21] Evans, Martin D.D., Richard K. Lyons, 2008, How is Macro News Transmitted to Exchange Rates?, Journal of Financial Economics, 88(1), pp. 26-50. [22] Fair, Ray C., 2003, Shock e ects on stocks, bonds, and exchange rates, Journal of International Money and Finance, 22(3), pp.307-341. [23] Faust, Jon, John H. Rogers, Shing-Yi B. Wang, Jonathan H. Wright, 2007, The High-frequency Response of Exchange Rates and Interest Rates to Macroe- conomic Announcements, Journal of Monetary Economics, 54(4), pp.1051-1068. 137 [24] Galati, Gabriele, 2000, Trading Volumes, Volatility and Spreads in Foreign Ex- change Markets: Evidence from Emerging Market Countries, BIS Working Pa- pers, No. 93. [25] Galati, Gabriele, Corrinne Ho, 2003, Macroeconomic News and Euro/Dollar Exchange Rate, Economic Notes by Banca Monte dei Paschi di Siena SpA, 32(3), pp.371 - 398. [26] Garman, Mark B., Michael Klass, On the Estimation of Security Price Volatil- ities from Historical Data, Journal of Business, 1980, vol. 53, no. 1, pp.67 ? 78. [27] Gertler, Mark; Simon Gilchrist, and Fabio Massimo Natalucci, External Con- straints on Monetary Policy and The Financial Accelerator, manuscript, Febru- ary 2001. [28] Hirshleifer, D., 2001, Investor Psychology and Asset Prices, Journal of Finance 56(4), pp.1533-1597. [29] Hull, John, M. Predescu, and A. White, The Relationship Between Credit De- fault Swap Spreads, Bond Yields, and Credit Rating Announcements, Journal of Banking and Finance, 2004 (28), pp. 2789-2811. [30] International Monetary Fund, Annual Report on Exchange Arrangements and Exchange Restrictions, IMF, 2000 2007. [31] Jahjah, Samir, Vivian Zhanwei Yue, Exchange Rate Policy and Sovereign Bond Spreads in Developing Countries, manuscript, MArch 2007. [32] Jeon, Seung-Hoon, and Lim Byung In, The E ects of the National Pension Wealth on Household Behaviors to hold the Private Pension Wealth and Its Policy Implications, Journal of Insurance Studies, November 2008, vol. 19, no. 3. [33] Kim, Gloria M., EMBI Global and EMBI Global Diversi ed: Rules and Method- ology, J.P. Morgan, December 2004. [34] Laakkonen, H., 2004, The impact of macroeconomic news on exchange rate volatility, Bank of Finland Discussion Paper 24. [35] Lahiri, Amartya, Carlos A. Vegh, Living with the Fear of Floating: An Optimal Policy Perspective, NBER Working Paper 8391, July 2001. 138 [36] Lahiri, Amartya, Rajesh Singh, Carlos A. Vegh, Segmented Asset Markets and Optimal Exchange Rate Regimes, NBER Working Paper 13154, June 2007. [37] Longsta , Francis A., Jun Pan, Lesse H. Pedersen, and Kenneth J. Single- ton, How Sovereign Is Sovereign Credit Risk?, NBER Working Paper 13658, December 2007. [38] Lui, Yu-Hon, David Mole, 1998, The Use of Fundamental and Technical Anal- yses by Foreign Exchange Dealers : Hong Kong Evidence, Journal of Interna- tional Money and Finance, 17(3), pp.535-545. [39] Magud, Micolas E., Currency Mismatch, Openness and Exchange Rate REgime Choice, Journal of Macroeconomics, 32(2010), pp.68-89. [40] McKenzie, Michael, 2002, The Economics of Exchange Rate Volatility Asym- metry, International Journal of Finance and Economics, 7(3), pp.247-260. [41] Mendoza, Enrique G., REal Business Cycles in a Small Open Economy, The American Economic Review, Vol. 81, No. 4 (Sep. 1991), pp.797-818. [42] Mian, G. M., S. Sankaraguruswamy, 2008, Investor Sentiment and Stock Market Response to Corporate News, National University of Singapore working paper. [43] Reinhart, Carmen, and Kenneth Rogo , The Modern History of Exchange Rate Arrangements: A Reinterpretation, Quarterly Journal of Economics, February 2004, Vol. CXIX, Issue 1, pp. 1-48. [44] Schmitt-Grohe, Stephanie, Martin Uribe, Closing small open economy models, Journal of International Economics, 61(2003), pp.163-185. [45] Tanner, Glenn, 1997, A Note on Economic News and Intraday Exchange Rates, Journal of Banking and Finance, 21(4), pp.573-585. [46] Tavlas, George, Harris Dellas, and Alan C. Stockman, The classi cation and Performance of Alternative Exchange-rate Systems, European Economic Re- view, 52, pp.941-963, 2008. [47] Uribe, Martin, Vivian Z. Yue, Country Spreads and Emerging countries: Who Drives Whom?, Journal of International Economics, 69, pp.6-36, 2006. [48] Veronesi, Pietro, 1999, Stock Market Overreaction to Bad News in Good Times: A Rational Expectations Equilibrium Model, The Review of Financial Studies, 12(5), pp.975-1007. 139 [49] Wongswan, Jon, 2006, Transmission of Information across International Equity Markets, The Review of Financial Studies, 19(4), pp. 1157-1189. 140