ABSTRACT Title of Thesis: REGIONAL PUBLIC TRANSPORTATION IMPROVEMENTS FOR DISADVANTAGED USERS: AN INTEGRATED ANALYSIS OF THE SOCIAL AND TRANSPORTATION NEEDS OF LOW INCOME POPULATIONS FOR THE WASHINGTON D.C. METROPOLITAN REGION Lynna Nguyen, Masters of Science, 2013 Thesis directed by: Professor Cinzia Cirillo Department of Civil and Environmental Engineering Public transportation is a crucial part of the economic and social fabric of metropolitan areas. However, transit ridership has been decreasing over the decades, putting preference on the convenience of owning personal vehicles. It is seen that low income individuals are less likely to own a vehicle, thus becoming dependents on the public transportation system. However, there are few studies performed to analyze how e ectively transit connects people and jobs within and across these metropolitan areas. And as a result, few federal and state programs related to transportation use factors like job accessibility via transit to make investment decisions. There are even fewer studies and programs relating to subsidizing vehicle ownership. Analyzing characteristics of low income individuals, understanding travel pat- terns, job availability, accessibility, and trip chaining are the methods used in this analysis to better understand the transportation needs of low income individuals. In addition, understanding the relationship that transit and personal vehicles play on the location of low income individuals and low income employment is crucial in creating and implementing programs that will improve and maintain transit and vehicle ownership options for metropolitan residents. REGIONAL PUBLIC TRANSPORTATION IMPROVEMENTS FOR DISADVANTAGED USERS: AN INTEGRATED ANALYSIS OF THE SOCIAL AND TRANSPORTATION NEEDS OF LOW INCOME POPULATIONS FOR THE WASHINGTON D.C. METROPOLITAN REGION by Lynna Nguyen Thesis 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 Masters of Science 2013 Advisory Committee: Professor Cinzia Cirillo, Advisor Professor Paul Schonfeld Professor Marie Howland ? Copyright by Lynna Nguyen 2013 Table of Contents 1 Transportation Needs of Low Income Individuals 1 1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.2.1 Travel Patterns . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.3 Impact of Transportation on Accessibility . . . . . . . . . . . . . . . . 5 1.4 Current Transportation System . . . . . . . . . . . . . . . . . . . . . 7 1.5 Public Practices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 1.5.1 Subsidized Public Transportation . . . . . . . . . . . . . . . . 10 1.5.2 Paratransit . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 1.5.3 Subsidized Vehicle Ownership . . . . . . . . . . . . . . . . . . 14 1.6 Spatial Mismatch Hypothesis . . . . . . . . . . . . . . . . . . . . . . 17 1.7 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 2 Statistical Analysis of Low Income Households and Individuals 22 2.1 Low Income Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.1.1 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.1.2 Low Income Household Statistics . . . . . . . . . . . . . . . . 24 2.1.2.1 Labor Force Distribution of Low Income Households 25 2.1.2.2 Distribution of Low Income Household Size . . . . . 26 2.1.2.3 Income-to-Rent Ratio . . . . . . . . . . . . . . . . . 27 2.1.3 Low Income Individual Statistics . . . . . . . . . . . . . . . . 28 2.1.3.1 Labor Force Participation of Low Income Individuals 28 2.1.3.2 Age Distribution of Low Income Individuals . . . . . 29 2.1.3.3 Education Level of Low Income Individuals . . . . . 30 2.2 Low Income Households and Individuals Conclusions . . . . . . . . . 30 3 Low Income Employment Analysis 32 3.1 Low Income Employment Analysis . . . . . . . . . . . . . . . . . . . 32 3.1.1 Low Income Employment Data . . . . . . . . . . . . . . . . . 32 3.1.2 Low Income Employment Analysis by Jurisdiction . . . . . . . 33 3.1.3 Low Income Employment Analysis by Occupation and Sector . 34 3.2 Low Income Employment Forecast Analysis . . . . . . . . . . . . . . 36 3.2.1 Low Income Employment Forecast Data . . . . . . . . . . . . 37 3.2.2 Low Income Employment Forecast by Jurisdiction . . . . . . . 38 3.2.2.1 Employment Forecast by Jurisdiction Methodology . 38 3.2.2.2 Low Income Employment Forecast by Jurisdiction Results . . . . . . . . . . . . . . . . . . . . . . . . . 39 3.2.2.3 Low Income Employment Forecast by Sector Method- ology . . . . . . . . . . . . . . . . . . . . . . . . . . 41 3.2.2.4 Low Income Employment Forecast by Sector Results 42 3.2.3 Low Income Employment Forecast Major Trends . . . . . . . 45 3.2.3.1 Median Household Income and Value . . . . . . . . . 46 3.3 Low Income Employment Analysis and Forecast Conclusions . . . . . 48 ii 4 National Household Travel Survey Trip Analysis 50 4.1 Trip Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 4.1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 4.1.2 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 4.1.3 Trip Analysis Results and Discussion . . . . . . . . . . . . . . 51 4.1.3.1 Trip Analysis by Mode . . . . . . . . . . . . . . . . . 52 4.1.3.2 Trip Analysis by Work Day . . . . . . . . . . . . . . 54 4.1.3.3 Trip Analysis by Employment Status . . . . . . . . . 55 4.1.3.4 Trip Analysis by Number of Household Vehicles . . . 56 4.1.3.5 Trip Analysis by Gender . . . . . . . . . . . . . . . . 57 4.1.3.6 Trip Analysis by Origin and Destination Purpose . . 58 4.2 Tour Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 4.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 4.2.2 Data and Methodology . . . . . . . . . . . . . . . . . . . . . . 60 4.2.3 Tour Analysis Results and Discussion . . . . . . . . . . . . . . 61 4.2.3.1 Work Statistics . . . . . . . . . . . . . . . . . . . . . 62 4.3 National Household Travel Survey Trip and Tour Analysis Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 5 Policy Analysis 66 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 5.2 Accessibility Methodology . . . . . . . . . . . . . . . . . . . . . . . . 66 5.2.1 Maryland Statewide Transportation Model . . . . . . . . . . . 69 5.2.1.1 Parameters . . . . . . . . . . . . . . . . . . . . . . . 70 5.2.1.2 Zone Selection . . . . . . . . . . . . . . . . . . . . . 71 5.2.1.3 Trip Generation . . . . . . . . . . . . . . . . . . . . 72 5.2.1.4 Trip Attraction . . . . . . . . . . . . . . . . . . . . . 73 5.2.1.5 Trip Distribution . . . . . . . . . . . . . . . . . . . . 73 5.2.1.6 Mode Choice . . . . . . . . . . . . . . . . . . . . . . 74 5.2.1.7 Trip Assignment . . . . . . . . . . . . . . . . . . . . 76 5.2.2 Log-Sum Measure Using the Maryland Statewide Transportation Model . . . . . . . . . . . . . . . . . . . . . . 76 5.2.2.1 Assumptions . . . . . . . . . . . . . . . . . . . . . . 76 5.2.2.2 Log-Sum Measure by Income Group . . . . . . . . . 77 5.3 Implemented Policy Programs . . . . . . . . . . . . . . . . . . . . . . 78 5.4 Analysis Region . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 5.4.1 Trips Made . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 5.4.2 Income Level . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 5.4.3 Drive Alone . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 5.4.4 Share Ride . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 5.4.5 Bus Transit . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 5.4.6 Rail Transit . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 5.5 Policy Analysis Conclusions . . . . . . . . . . . . . . . . . . . . . . . 95 iii 6 Conclusions and Discussions 96 6.1 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 A Statistical Analysis of Low Income Households and Individuals 98 B Policy Analysis Figures 105 iv Chapter 1 Transportation Needs of Low Income Individuals 1.1 Introduction Public transportation is a crucial part of the economic and social fabric of metropolitan areas. However, transit ridership has been decreasing over the decades, with commuters preferring the convenience of personal vehicles. However, studies show that low income individuals are less likely to own a vehicle, thus becoming dependents on the public transportation system. There are few studies performed to analyze how e ectively transit connects people and jobs within and across these metropolitan areas. And as a result, few federal and state programs related to trans- portation use factors like job accessibility via transit to make investment decisions. There are even fewer studies and programs relating to subsidizing vehicle owner- ship. In addition, understanding the relationship that transit and personal vehicles play in the location of low income individuals and low income employment is cru- cial in creating and implementing programs that improve and maintain transit and vehicle ownership options for metropolitan residents. Analyzing characteristics of low income individuals, understanding travel patterns, job availability, accessibility, and trip chaining are the methods used in this analysis to better understand the transportation needs of low income individuals. 1 1.2 Literature Review 1.2.1 Travel Patterns Numerous studies of the the U.S. Census and National Household Travel Sur- vey (NHTS) were performed to understand the di erences in the travel patterns of low income individuals versus higher income individuals. The studies showed that a secular decline in transit demand that started in the 1930s still continues today. This leaves the public transit industry with two major markets: downtown com- muters and transit dependents. The downtown commuters still persist due to cost and limited parking availability, road congestion, and large concentrations of jobs that make transit access relatively convenient. Transit dependents are those who are unable or unwilling to drive or do not have access to a personal vehicle [Giuliano et al., May, 2001, Giuliano, 2005]. This transit-dependent market is increasingly an inner city, minority market; however, studies show that this market is shrinking as car ownership continues to increase [Giuliano et al., May, 2001]. According to studies of the U.S. Census data, the use of public transit as the primary transportation mode to work has decreased from 12.6% to 4.7% between 1960 and 2000. During that period, automobile use has increased from 84.1% to 87.9%. There has been an increasing trend in single-occupancy-vehicle commuters and a decreasing trend in high-occupancy-vehicle commuters. According studies of the Nationwide Personal Transportation Surveys (NPTS) and the National House- hold Travel Survey (NHTS), all person-trips on transit have decreased from 3.2% to 1.6% between 1969 and 2001 [Pucher and Renne, 2003]. The loss of the middle- 2 and higher-income passengers have been greater than the loss of low income passen- gers, therefore, low income passengers make up an increasing share of transit users [Giuliano, 2005, Waller, Dec., 2005]. These statistics are summarized in Table 1.1. Table 1.1: Summary of Trips. Percent of Trips in Year Percent of Trips in Year Work Commute via Public Transit Year 1960: 12.6% Year 2001: 4.7% All Person Trips via Public Transit Year 1969: 3.2% Year 2001: 1.6% Work Commute via Automobile Year 1960: 84.1% Year 2000: 87.9% Source: Nationwide Personal Transportation Surveys and National Household Travel Survey Analysis of the 2001 NHTS indicates that low income individuals make fewer trips than those of higher income households. Low income individuals make about 20% fewer trips than people at other income levels (1,340 person trips compared to 1,648 person trips) [Giuliano, 2005, Murakami and Young, Oct., 1997, Pucher and Renne, 2003, Waller, Dec., 2005]. From the 1995 National Personal Transportation Survey (NPTS) data, it was found that the average trips per day ranged from 3.4 trips for those in the lowest income category to 4.2 trips for those in the highest income category [Pucher et al., 1998]. Low income individuals also travel nearly 40% fewer miles than higher income individuals, yet low income individuals have longer travel times, especially for trips that require transfers [Giuliano, 2005, Murakami and Young, Oct., 1997, Pucher and Renne, 2003, Waller, Dec., 2005]. The average miles traveled per day ranged from 17.4 miles for those in the lowest income category to 28.6 miles for those in the highest income category [Giuliano et al., May, 2001]. About 60% of low income individual trips are 3 miles or less compared to 50% for other individuals. This is more apparent in low income single parent households where about 66% of trips 3 are 3 miles or less. Because vehicle availability is lower, vehicle miles traveled is also lower by 40% compared to other households (9,060 versus 14,926 person miles) [Murakami and Young, Oct., 1997]. The di erence between trips made and mileage can be explained in car ownership and modal use [Giuliano et al., May, 2001]. Averaged across all households, commuting to work by private vehicle can take half as long as commuting via other modes { 20 minutes compared to 42 minutes [Waller, Dec., 2005]. Personal automobile are especially advantageous and suitable for multiple-stop trip chaining like taking a child to school, grocery shopping, and other errands that require signi cantly more time when relying on public transit as the primary transportation mode [Goldberg, 2001, Lucas and Nicholson, 2003, Waller, Dec., 2005]. Though low income individuals are less likely to own a vehicle, they still make a majority of their trips in private vehicles. These trips are much more likely to be made using a vehicle owned by someone else, like a friend or relative at a rate of 8% compared to 1% for other income groups. This percentage is higher for households with children [Murakami and Young, Oct., 1997]. Data from NHTS 2001 show that car ownership dramatically decreases public transit usage from 19.1% for households with no vehicle to 2.7% for households with one vehicle [Pucher and Renne, 2003]. Instead of taking public transit, a large percentage (43.5%) of individuals use non-motorized (i.e., walking and biking) trips. Individuals with access to a vehicle sacra ce mobility by having to rely on public transit and non-motorized trips, resulting in shorter and fewer trips. For work and work-related trips, low income households reported 5% walk modal share compared 4 to 3% for other income groups [Murakami and Young, Oct., 1997]. Low income individuals are also more likely to take public transit to work with 5% modal share compared to 2% for other income groups [Murakami and Young, Oct., 1997, Waller, Dec., 2005]. Low income individuals are more likely to take bus transit, whereas other income level individuals are more likely to take rail transit. Therefore, the average trip distance for low income households using public transit is 10 miles compared to 13 miles for other income levels. Associated with the shorter trip is a shorter commute time, 36 minutes for low income households compared to 43 minutes for other households [Murakami and Young, Oct., 1997]. These statistics are summarized in Table 1.2. Table 1.2: Summary of Trips Made by Low Income and Other Income levels. Low Income Other Income Levels Vehicle Ownership 73.5% 96.0% Vehicles per Adult 0.7 1.0 Average Age of Vehicle 10 years 7.3 years Person Trips (Per Day) 3.4 trips 4.2 trips Person Trips (Per Year) 1,340 trips 1,680 trips Average Miles Traveled (Per Day) 17.4 miles 28.6 miles Trips 3 Miles or Less 60% 50% Trips Made via Automobiles Owned by Others 8% 1% Trips via Walking 5% 3% Work Commute via Walking 5% 3% Work Commute via Public Transit 5% 2% Average Miles on Public Transit (Per Day) 10 miles 13 miles Average Commute Time on Public Transit 36 minutes 43 minutes Source: National Household Travel Survey, 2001 1.3 Impact of Transportation on Accessibility Studies have shown that increased mobility can positively a ect employment status for low income individuals. Although policy makers tend to want to subsi- 5 dize transit, there is little empirical evidence that supports increasing public transit mobility positively e ects employment status [Sanchez et al., 2004]. Sanchez, Shen, and Peng [2004] examined whether increased transit access is associated with the employment status of Temporary Assistance for Needy Families (TANF) recipients in six metropolitan areas. A limited dependent variable regression analysis was performed and the results indicate that access to xed-route transit and employ- ment locations had virtually no association with employment outcomes of TANF recipients [Sanchez et al., 2004]. Numerous studies attempt to address the relationship between labor force participation and the spatial separation of jobs and houses; however, most analyses concentrate on commuting time or distance as a function automobile accessibility. Few studies consider the relative impacts of employment accessibility resulting from public transportation services. Various studies have recognized that traditional pub- lic transportation services have limited capacity to meet the travel needs of persons with little or no access to automobiles [Sanchez, 2008]. Shen [1998] examined the impact of public transit on connecting urban res- idents with job locations using labor participation rates (average annual weeks worked, as reported by the 1990 U.S. Census). The analysis was performed on Port- land, Oregon and Atlanta, Georgia residents living within walking distance (i.e., quarter of a mile) and not within walking distance of a transit stop. Census block groups were examined and for all block groups, average employment levels decreased as the distance from transit stops increased. Average employment levels were found to increase substantially as vehicle ownership increased. The results partially sup- 6 port policies for improving public transit access to improve urban employment. 1.4 Current Transportation System City residents and lower-income neighborhoods tend to be better served by transit than residents of suburbs and middle/higher-income neighborhoods [Tomer et al., May, 2011]. Therefore, public transportation systems have been the response of upper-income groups (business owners) to the mobility needs (demand) of lower- income groups and the notion that transportation mobility is important for economic opportunity. There is little evidence suggesting that the public transportation sys- tem development resulted from lower-income groups [Sanchez, 2002]. Hence the primary mode of public transportation is xed bus route service where riders must adapt to service availability [Sanchez, 2008]. Although, nearly half of all work commutes still originate from or terminate in central cities, 39% of work commutes are entirely suburban, that is originating and terminating in suburban areas. Older rail transit systems fail to capture most suburban commuting patterns because they were mostly developed for trips into and out of the city. This hub-and-spoke transit pattern provides dense metropolitan cores, but may not serve other metropolitan areas very well. From 2002 to 2007, the amount of developed land in the US increased by 8.4%, nearly twice the rate of population growth (4.5%). A majority of large metropolitan residents live under traditional or exclusionary zoning requirements that separate di erent land uses and/or emphasizes low-density development, thereby making it harder to connect 7 people and jobs using public transportation alone [Tomer et al., May, 2011]. The Brookings Institute performed a large scale accessibility study on the top 100 largest metropolitan areas across the U.S. They found that nearly 70% of U.S. Census de ned block groups have access to at least one transit stop within 3/4 mile of their population weighted centroid. Transit coverage is best in Western metropolitan areas and worst in Southern metropolitan areas. In the Northwest and Midwest, neighborhoods that developed decades before the adoption of automobiles have neighborhoods that are well-served by public transit [Tomer et al., May, 2011]. The service frequency is calculated as the median of typical headways in all block groups with public transit access, weighted by their working-age population. Typical headways are shown in Table 1.3. Good transit service does not necessarily equate to high service frequency. Some metropolitan areas have short headways, but focus on smaller areas, thus the area overall has poor transit coverage. In cities, low income neighborhoods experience shorter headways than high-income neighborhoods. Transit systems provide service to city residents nearly twice as frequently as suburban residents (94% versus 58%). This re ects built up transit and rail systems prior to suburbanization, thus serving cities better than suburbs. Within metropolitan areas, city headways are consistently shorter than suburban headways [Tomer et al., May, 2011]. The Brookings Institute found that typical metropolitan residents can reach about 30% of jobs within their area via public transit in 90 minutes. Metro-wide job access was calculated as the average share of jobs reachable within 90 minutes across all block groups with transit coverage, weighted by block group working-age 8 population. This varies considerably across metropolitan areas re ecting variable transit coverage levels, service frequencies, employment levels, and population de- centralization. The percentage of jobs accessible via transit in 90 minutes by region is listed in Table 1.3. In all regions, city residents have better transit access to jobs than suburban residents. The biggest disparity is seen in the north-east, and in the mid-western and southern suburbs, the accessibility level is below 20%. Low income neighborhoods have higher job accessibility than other income levels (36% for low income level, 28% for middle income level, and 23% for higher income level) [Tomer et al., May, 2011]. However for most individuals, a 90 minute commute on public transit is very costly in terms of time and other social factors such as comfort and convenience that is preferred in private automobiles [Sanchez, 1999]. Table 1.3: Public Transit Service Performance for Di erent Metropolitan Regions. Metropolitan Region Transit Service Headway (During Peak Hours) Job Accessibility (for 90 Minute Public Transit Travel Time) North-East 8 minutes 32% Western 9.2 minutes 33% Mid-West 11 minutes 28% Southern 12 minutes 26% Source: Brookings Institute [Tomer et al., May, 2011] 1.5 Public Practices There are various current practice transportation policies and programs im- plemented to promote job accessibility and accessibility to medical, government, and other necessary services and locations for low income peoples, elderly, disabled, and otherwise disadvantaged population. From an external point-of-view these im- 9 plemented programs have helped thousands of disadvantaged individuals nd and maintain jobs, get necessary access to medical facilities, help and promote a more e cient everyday life. However, there are few empirical studies performed to analyze the e ectiveness of these programs. These programs range from: paratransit services, carpool/vanpool type ser- vices, assistance and subsidization of car ownership, subsidized public transit, etc. The following sections will provide case studies of implemented transportation pol- icy practices and program to increase the understanding of di erent possible policies and programs that may be introduced into the Washington D.C. Metropolitan Re- gion. Potential programs will be recommended for the region based on the urban and suburban e ective regions that require attention. Though the region of interest would not be designated as rural, there are many transportation programs that are recommended which can be modi ed to be more e ective in suburban areas, therefore these initiatives should not be readily dismissed. 1.5.1 Subsidized Public Transportation Subsidizing public transit has two con icting objectives: 1) to provide a basic level of mobility for all persons, but especially to those disadvantaged, and 2) to provide an e ective substitute for personal vehicles in order to reduce automobile travel and its associated consequences, including: tra c congestion, air pollution, and urban sprawl [Giuliano et al., May, 2001, Hodge, 1995]. The latter objective has 10 emphasized the provision of rail transit, which is more attractive to discretionary riders and therefore more e ective in achieving environmental goals. Also, with the smart growth movement, even more emphasis has been placed on rail transit [Giuliano et al., May, 2001]. Transit investments manifest the emphasis on rail transit. Between 1991 and 1998, the total revenue vehicle miles of light rail, commuter rail, and heavy rail service increased 59, 20, and 8 percent respectively. Over the same period, bus service increased by only 6 percent [of Transportation, 1998]. There is little evidence that suggest that these investments are generating the desired increases in transit ridership. And in some cases, new rail service replaces pre-existing bus route service and attracts few new riders from cars. New rail systems appeal to long distance, downtown commuters, who are disproportionally a uent. Often enough, the high cost of building and operating rail systems have led to reducing transit ridership as fares increase and bus service is reduced in response to budge constraints. Low income individuals concentrated in central cities would bene t more from increased bus service frequency, lower bus fares, and fewer bus transfers [Giuliano et al., May, 2001]. Mobility is essential for access to jobs, services, and social activities, therefore it is seen as a public responsibility to supply basic levels of transportation services to those who do not or cannot drive. Public transit agencies have shifted resources from basic local transit services to more costly commuter services designed to attract discretionary riders. Since the local transit services are used more by low income individuals, the bene ts of other commuter services are not seen by these individuals, 11 therefore the public subsidies are inequitably distributed [Hodge, 1995]. The housing near the rail transit is also becoming more expensive, making it di cult for low income households to reside near the rail transit system [Giuliano, 2005, Waller, Dec., 2005]. Despite signi cant investment and public subsidies in public transit network, transit usage as a whole is on the decline. Nationwide travel survey data show that less than 2% of all person trips are made by public transit [Giuliano, 2005, Giuliano et al., May, 2001]. The trend is to cut central city transit investment and put it towards suburban commutes (rail and discretional riders) [Waller, Dec., 2005]. US transit ridership is heavily concentrated in a few of the largest cities. New York City accounts for 40% of US daily transit ridership, when adding Los Angeles, Chicago, Boston, San Francisco, and Washington DC, these cities account for two-thirds of the nations total daily transit ridership [Association, 2000]. In Lincoln/Lancaster, Nebraska [Boesch, May, 2005], a six month pilot project was launched to test the hypothesis that decreasing cost would increase ridership. From October 1, 2004 to March 31, 2005, Human Services and StarTran jointly operated the program in e orts to increase ridership on the Citys busses and Handi- Vans, thus proving more a ordable transportation to low income individuals. Bus and Hand-Van passports were subsidized making bus passports $5 per month (orig- inally $30 per month) and Handi-Van passports $10 per month (originally $60 per month). Over the six-month pilot program, ridership continued to increase from 692 low income passports in October to over 1,200 passports each month. Ridership 12 increased by 25% in youth riders each month and was evenly split between males and females. Ridership increased by 15.2% from the previous year. An estimated additional 1,000 people are using public transportation on a regular basis. Rider surveys indicate that 60% use the bus to get to work and 70% use the bus system as their primary transportation mode. The Handi-Van had a 6.5% total increase in ridership. 1.5.2 Paratransit Within Virginia, Richmond has taken initiative to provide the elderly, disabled, and low income individuals access to transportation. In 2000, the U.S. Census re- ported that approximately 12% of the U.S. population is below poverty and within Virginia, almost 10% of its residents are living below the poverty level. The Rich- mond area has 64,000 people (age 5 years and older) living below poverty, amounting to 9% of the regions population. Over half of these low income peoples live within the City limits. Therefore to address these concerns, transportation programs and services have been implemented to increase the mobility of low income individuals. The Goochland Fellowship and Family Services (GFFS) [Feb., 2006] provides free transportation services to its residents who cannot otherwise provide trans- portation for themselves due to age, disabilities, or poverty. GFFS employs one full time van driver which provides transportation with one handicapped accessible van to medical and dental appointments and for pharmaceutical needs. Reservations are requested to be made 24 hours in advance and the service runs during weekdays. 13 During 2005, GFFS provided transportation to 649 individuals and made 201 trips to medical and dental appointments and to local pharmacies. Another vanpool paratransit transportation program initiated in Richmond, VA is Access Chester eld (in Chester eld County, Virginia). It is coordinated van- pool transportation program for the disabled, elderly (age 60 and over), and those who meet federal income guidelines, where household income is less than 200% the poverty level. The vanpool service has no limit on the trip purpose within the ser- vice area as long as it remains in the service area. Services are made by reservation only from Monday through Saturday. The cost is two vouchers for any trip regard- less of distance, where a pack of 10 vouchers is $25. The average weekly trips have increased from several trips to almost 300 trips per week during the November 2004 to June 2005 observation period. 1.5.3 Subsidized Vehicle Ownership The share of households without vehicles dropped from 21% in 1969 to 9% in 1995. The majority of zero-vehicle households are the elderly and retired persons. Most of the remainder is single individuals without children, and two-thirds of zero- vehicle households have no workers. Low income households are less likely to have a vehicle, largely in part because a greater portion of their income is spent on shelter and food. 8% of all urban households do not have a car. However, for households with an annual income less than 20,000, 26.5% do not have a car compared to 4% of other households [Murakami and Young, Oct., 1997, Pucher and Renne, 2003, 14 Waller, Dec., 2005]. Though low income individuals are less likely to own a vehicle, 84% of work trips in low income households use private vehicles compared to 90% in other house- holds. However, the average age of the car for low income individuals is also higher than other households, 10 years compared to 7.3 years. Within low income house- holds, there is 0.7 vehicles per adult compared to 1 vehicle per adult in other households [Murakami and Young, Oct., 1997, Waller, Dec., 2005]. Low income individuals also tend to pay more when purchasing and nancing a vehicle, possibly attributable to bad, short, or no credit history. For those who do own a vehicle, insurance payments and ongoing maintenance costs, especially for older vehicles), are often very high [Goldberg, 2001, Lucas and Nicholson, 2003, Waller, Dec., 2005]. The total car ownership cost varies from region to region, but is generally between $1,100 to $1,400 per year (excluding purchase price and any major repairs). For a low income individual working earning minimum wage and working less than 40 hours per week, this $1,100 to $1,400 cost can fall between 10 to 14 percent of their income [Goldberg, 2001]. The research by Ong and Blumberg [1998] showed that the labor market does not generate compensating wages for welfare recipients who travel far distances to work, such that wages are negatively associated with distance [Wachs and Taylor, 1998]. Therefore better geographical job access has both direct and indirect e ects on recipients. Out-of-pocket expenses are reduced and the opportunity costs as- sociated with travel to work is reduced when the geographical access is improved. Because the labor market does not produce compensating wages for long distance 15 commutes, improved access to jobs indirectly a ects recipients though increased earnings. The ndings stress the importance of increasing job access through lo- cal economic development and by improving mobility, both housing mobility and transportation mobility, among low income individuals [Ong and Blumberg, 1998]. Vermont is primarily a rural state with over 60% of its population living in rural areas. There are only a few towns in Vermont with a public transportation system. The public transportation systems that are available do not provide the necessary mobility for low income individuals to nd and retain jobs. The Good News Garage was then created in 1996 to facilitate the ownership of functional vehicles by low income individuals, with the focus and objective of helping them nd and retain employment. The Good News Garage (GNG) program acquires vehicles from donors, repairs or refurbishes them, and sells them at minimal cost to individuals with income less than 150% of the poverty level. Priority is given to applicants who need a vehicle to retain their current employment or those who are participating in a job training program. The program calls for a personal contribution of $400 to $800 to cover some of the costs of repairs or refurbishing, vehicle registration, and taxes. Since its start in 1996, the GNG has provided over 1,000 donated cares to low income Vermont residents. An analysis of the e ectiveness of the program found that mean earned income was $220 higher per month after individuals received the vehicle ($361.40/month after receiving the vehicle compared to $141.90/month prior). Lucas and Nicholson [2003] studied the impacts of vehicle acquisition using reduced-form random e ects 16 and censored regression models to account for other factors, such as the simultaneous decision to work and to participate in the welfare program. The impacts of the vehicle acquisition program were still found signi cant in increasing the probability of employment. The researchers found that income increased between $124 (using random e ects model) and $127 (using censor regression model) per month due to car ownership. Individuals were also 19% more likely to have earned income after acquiring a car. The car cost to the individual and the program would be recovered within a few months as earnings replaced welfare cash assistance [Lucas and Nicholson, 2003]. 1.6 Spatial Mismatch Hypothesis For metropolitan regions with highly dense city centers and expanding sub- urbs like the Washington D.C. Metrpolitan region, there is a tendency for a spatial mismatch betwen where low income individuals reside and where low income jobs are expected to grow. Central city residents experience high unemployment even in times when the overall demand for labor is high. Traditional explanations are: inferior quantity and quality of education, poor health, and low motivation of low income individuals [Kalachek, 1968]. A majority of low income households reside in rural areas and central cities. And within the last century, residential and em- ployment patterns have in metropolitan areas have reversed [Murakami and Young, Oct., 1997, Wachs and Taylor, 1998, Waller, Dec., 2005]. Also, basic amenities are increasingly located in the suburbs; therefore access to transportation is limited for 17 low income individuals [Criden, 2008, Murakami and Young, Oct., 1997, Waller, Dec., 2005]. The spatial mismatch hypothesis stresses the spatial separation between the central-city residential locations of low income individuals and expanding job op- portunities in the suburbs and the long commute that is needed to connect them [Blumberg, 2004, Blumberg and Manville, Nov., 2004, Criden, 2008, Giuliano, 2005, Tomer et al., May, 2011, Wachs and Taylor, 1998]. A study by Kasarda [1995] found that 70% of all jobs in manufacturing, retail, and wholesale industries that typically have large number of entry level jobs are located in the suburbs. Much of the job growth within cities requires higher skill levels that many low income workers dont have [Criden, 2008, Kalachek, 1968, Murakami and Young, Oct., 1997, Tomer et al., May, 2011, Waller, Dec., 2005]. There are higher concentrations of high-skill industries located within cities than low- and middle-skill industries. About a quarter of low- and middle-skill industries are accessible via public transit in 90 minutes compared to a third of high-skill industries. 94 of the top 100 metropolitan areas provide greater public transit access to high-skill industries than low- and middle-skill industries [Tomer et al., May, 2011]. As low-wage jobs are emerging further and further away from central cities, many low income workers have di culty accessing jobs, training, and other services such as childcare due to inadequate transportation options [Criden, 2008, Kalachek, 1968, Murakami and Young, Oct., 1997, Waller, Dec., 2005]. While new jobs are located in the suburbs, there is little public transit that connects the central-city low income residents to suburban employment [Organization, Feb., 18 2006, Waller, Dec., 2005]. Therefore the low income workers experience a relative decline in job accessibility, which leads to higher unemployment rates and longer commutes for those employed in the suburbs, which imply lower net wages [Giuliano et al., May, 2001, Giuliano, 2005]. The rapid increase in suburban employment does not mean entry-level jobs no longer exist in central cities. Net job growth and job turnover in existing jobs needs to be distinguished, and there is plenty of evidence that suggests that urban job turnover rate actually exceeds the rate of suburban job growth [Blumberg and Manville, Nov., 2004]. As urban sprawl increases around the country due to the adoption of automobiles and the development of auto-dependent societies, the transportation mode choices share of walking and biking decreases and the reliance on automobiles increase [Criden, 2008]. A study by Ong and Blumberg [1998] showed that welfare recipients who lived in job-rich neighborhoods are more likely to nd work in close proximity to their home than those who lived in job-poor neighborhoods. The job distance model emphasized the importance of creating employment opportunities in neighborhoods where jobs are scarce. During the 2000s, poverty grew ve times faster in the suburbs than in cities. While low income households in the suburbs are not nearly as geographically con- centrated as some central cities low income neighborhoods, there is a trend to- wards suburbanization of poverty. Low income suburban residents also tend to live in less job-rich communities than their higher income counterparts [Tomer et al., May, 2011]. The isolation of inner-city and unemployed persons from suburban em- ployment opportunities was identi ed many years ago and from a transportation 19 perspective, the reverse commute presents a signi cant challenge for low income individuals because many do not own a vehicle and transit service does a poor job of servicing these types of reverse commute trips [Sanchez et al., 2004]. An argument against the spatial mismatch hypothesis include the notion that low income individuals do not have a geographical disadvantage with respect to job opportunities, rather, many of them su er from a spatial disadvantage because they are dependent on relatively slow, in exible, and limited public transit services [Sanchez et al., 2004]. 1.7 Motivation The existing transportation system cannot always bridge the distance between where low income individuals live and where jobs are located. The existing sys- tems were originally established to transport inner-city residents to city locations and bring suburban residents to central city work locations. However, the major- ity of entry-level jobs that low income individuals are likely to ll are located in the suburbs which may have limited or no accessibility through the existing public transportation system. The isolation of inner city and underemployed persons from suburban employment opportunities was identi ed as the poverty transportation problem. The reverse commute experienced by low income individuals who travel from a central city region to a suburban region in the morning and returning in the evening represents a signi cant challenge for individuals who do not own personal automobiles and rely on public transit because public transit services do a poor job 20 in serving these types of trip [Sanchez, 2008]. Therefore this research shines light on who low income individuals are, where they live, where they work, what they do, how they travel, and the e ect of implementing certain policies on accessibility. 21 Chapter 2 Statistical Analysis of Low Income Households and Individuals 2.1 Low Income Statistics A demographic pro le of the Washington D.C. Metropolitan Region was ana- lyzed to identify the main characteristics of the low income population, the composi- tion of low income households, and the factors that hinder the advancement of these households using the American Community Survey 2006-2010, 5-year estimates. 2.1.1 Data The American Community Survey (ACS) 2006-2010 5-year estimates was used to analyze the demographic pro le and the location of low income individuals and households. The data are aggregated at the Public Use Microdata Areas (PUMAs) level. The ACS provides information of 293,492, 785,361, and 895,776 households in Washington, D.C., Maryland, and Virginia, respectively. A general reference of the PUMA locations is show in Table 2.2. It should be noted that PUMAs vary in shape and size; hence encompassing di erent communities. Household and individual weights provided by the ACS were also used to reduce sampling bias and error. The sampling bias includes over and under represented subpopulations. Low-income households are identi ed as households with whose total income is below 1.5 times the 2010 National Poverty Guidelines for the respective household 22 size. These guidelines are updated each year and issued in the Federal Register by the Department of Health and Human Services (HHS), see Table 2.1. Table 2.1: 2010 Poverty Guidelines for the 48 Contiguous States and the District of Columbia. Persons in Family/Household Poverty Guideline Low Income Threshold 1 $10,830 $16,245 2 $14,570 $21,855 3 $18,310 $27,465 4 $22,050 $33,075 5 $25,790 $38,685 6 $29,530 $44,295 For families with more than 6 persons, add $3,740 for each additional person. Additional $5,610 23 Table 2.2: PUMAs within the Washington D.C. Metropolitan Region. State County PUMA Reference Location D.C. 00101 North-West D.C. 00102 Northern D.C. 00103 North-East D.C. 00104 South/South-East D.C. 00105 Downtown D.C. Montgomery 01001 Columbia MD Frederick 00300 Fredrick 01002 Germantown 01003 Gaithersburg 01004 Bethesda, Potomac 01005 Aspen Hill, Rockville 01006 Colsville 01007 Silver Spring Prince George 01101 College Park, Hyattsville 01102 Bowie, Crofton 01103 Landover 01104 Walker Mill, Kittering 01105 Harwood, Davidsonville 01106 La Plata, Huntingtown 01107 Suitland VA Arlington 00100 Arlington Alexandria 00200 Alexandria Fairfax 00301 Fairfax, McLean 00302 Mt. Vernon 00303 Burke 00304 Clifton, Centerville 00305 Reston, Great Falls Prince William 00501 Warrenton, Bristow 00502 Dale City, Montclair Loudon 00600 Gainesville 2.1.2 Low Income Household Statistics Washington, D.C. has the highest percentage of low income household in the area, with approximately 18.9%. Maryland has the second highest number of low income households at 9.4%. Virginia has the lowest percentage of low income house- holds at 7.4%. Tables A.1 to A.3 in Appendix A shows the distribution of the 24 percentage of low income households by PUMA. Within Washington D.C., the central and south-east regions (PUMAs 00105 and 00104) have the overall highest percentage of low income households, 24.1% and 21.9%, respectively (out of all low income households within Washington D.C.). Of all individuals in the south-east region (PUMA 00104), 17.9% of households are low income households. 41.3% of all low income households in Maryland are located in Prince George?s County. Within Prince George?s County, 19% of households in the College Park area (PUMA 01101) are considered low income. Virginia has the least number of low income households, with a more even distribution of low income households. Overall, there is no PUMA that contains more than 9% of low income households within it. Fairfax County has 46.6% of all low income households within Virginia. 2.1.2.1 Labor Force Distribution of Low Income Households Understanding the low income household employment composition can help better understand factors that hinder low income households from achieving in- creased income. ACS de nes the labor force as all individuals above the age of 16 and has no illness or condition that prevents him/her from working. ACS breaks down employment status into: at least one member is in the labor force, a household that is described as a couple where neither are in the labor force, a single female as the head of the household whom is not in the labor force, and a single male as the head of the household whom is not in the labor force. Figure 2.1 shows the 25 breakdown of the employment status of only low income households in Washing- ton D.C., Maryland, and Virginia. Overall, low income households in Washington D.C. have a lower employment rate than those in Maryland and Virginia. However, Washington D.C. has the lowest percentage of households which consist of a couple where neither are in the work force. There is also a larger percentage of single fe- males as head of the household whom are not in the labor force than single males, with the largest overall percentage found in Washington D.C., which in both cases is very concerning. Single female head of households are typically unemployed single mother households. The labor force distribution by PUMA can be seen in Tables A.4 to A.6 in Appendix A. Figure 2.1: Labor Force Distribution of Low Income Households. Source: American Community Survey 2006-2010, 5-year Estimates 2.1.2.2 Distribution of Low Income Household Size Figure 2.2 shows distribution of the size of low income households. Surpris- ingly, more than 50% of households consist of only one person in Washington D.C. and over 40% in Maryland and Virginia. Tables A.7 to A.9 in Appendix A shows 26 the distribution of low income household size by PUMA. Figure 2.2: Distribution of Low Income Household Size. Source: American Community Survey 2006-2010, 5-year Estimates 2.1.2.3 Income-to-Rent Ratio Figure 2.3 shows the distribution of the income-to-rent ratio. It is alarming to nd that approximately 38% of all low income households within the region spend more than 100% of their household income on rent. This can be the result of having numerous low income households that consist of one member, high unemployment rate, and high rental rates in the Washington D.C. Metropolitan area. It should be noted that only rental units were observed which consist of 81% of low income households in Washington D.C. and 61% of low income households in Maryland and Virginia. Tables A.10 to A.12 in Appendix A shows the distribution of low income income-to-rent ratio by PUMA. 27 Figure 2.3: Income-to-Rent Ratio of Low Income Households. Source: American Community Survey 2006-2010, 5-year Estimates 2.1.3 Low Income Individual Statistics The distribution of low income individuals within Washington D.C. is clus- tered around the south-southeast portion of Washington D.C. (34%). Within the Maryland portion of the region, the highest percentages of low income are seen in Frederick and College Park (10% and 14%, respectively) with a more uniform dis- tribution in other areas. The low income population is more uniform and at a lower percentage in the Virginia. 2.1.3.1 Labor Force Participation of Low Income Individuals Figure 2.4 shows the percentage of the low income persons that is economically active. From the ACS de ned labor force, approximately 22% to 30% of the low income persons cannot work. It is apparent that low income individuals in Wash- ington D.C. are worse o than those in Maryland and Virginia. The low income individuals in Washington D.C. have the lowest employment rate, highest unem- 28 ployment rate, and highest rate of individuals not in the labor force. Tables ?? to ?? in Appendix A shows the distribution of labor force participation of low income individuals by PUMA. Figure 2.4: Labor Force Participation of Low Income Individuals. Source: American Community Survey 2006-2010, 5-year Estimates 2.1.3.2 Age Distribution of Low Income Individuals When decomposing Washington D.C.s low income population by age, it is found that 53% of these individuals are between 16 and 25 years old. Larger per- centages can be found in Northwest D.C. and Central D.C (75% and 43%, respec- tively). This can be expected from a young population that hasnt yet progressed into the labor force thus their low income status. Nevertheless, this highlights a problem of a possible future low income middle aged and elderly population if no measures are taken to guarantee minimal professional and personal progress to the young population. In Maryland, nearly half (51%) of low income individuals are between 16 and 25 years old. The middle aged and elderly (i.e., 36 years and older) low income individuals are more uniformly distributed in Virginia. 29 2.1.3.3 Education Level of Low Income Individuals Numerous studies have shown the correlation between education and income level. Data from the U.S. Bureau of Labor Statistics shows that in 2010 the median earnings increase at every level of education. Within the region, around 57% of the low income population attained at the most a high school diploma. Within the east-southeast portion of D.C., nearly 80% of low income individuals completed high school or less. Within Maryland, the percentage range of low income high school graduates or less range from 35% to 75%. Within Virginia, the range is 44% to 75%. 2.2 Low Income Households and Individuals Conclusions Larger clusters of low income individuals and households are found more in Washington D.C. and Maryland than in Virginia, where the low income population are somewhat uniformly scattered across the PUMAs. Around 40% to 50% of low income households consist of one person. Consequently, their living expenses rise, as demonstrated by the fact that approximately 60% of low income households reported spending more than 80% of their household on rent. Another concerning nding was that 35% of all low income households in Washington D.C. have an unemployed single female as head of the household. In terms of work, around 10% of low income population was unemployed and 50-60% was not part of the labor force in the year before the survey. This could be a key reason why these people are not rising over the low income threshold. Educa- tion can also be a big factor; however, a fair percentage of individuals obtained an 30 associate?s degree or higher. This raises the question of what is impeding low income households from advancing. Future studies sould focus on nding the underlying factors (e.g., accessibility to jobs, personal/family commitments, health condition, etc.) that are hindering this demographic. Furthermore, future researchers should consider creating a tailored survey to obtain information that is not usually found in publicly available data sets. 31 Chapter 3 Low Income Employment Analysis 3.1 Low Income Employment Analysis A statistical analysis of the Washington D.C. Metropolitan Region is per- formed to examine the types of jobs held by workers within low income house- holds using the American Community Survey 2006-2010, 5-year estimates along with Round 8a Coooperative Forecast Data from the Metropolitan Washington Council of Governments. 3.1.1 Low Income Employment Data The same American Community Survey (ACS) 2006-2010 5-year estimates were used and low income households are identi ed in the same way as the previous chapter to look at the types jobs that low income individuals hold. In this chapter, individuals who live in Washington D.C., Maryland, or Virginia and work in the Washington D.C. Metropolitan Region were analyzed. The employment analysis was performed on 134,559 individuals living within Washington D.C., Maryland, and Virginia that work within the region from the Public Use Microdata Sample (PUMS). When taking into account weights, there is employment data on 2,772,892 workers in the region. Of those individuals analyzed, 26.4% (731,328) work in Washington D.C., 31.3% (867,043) work in the Maryland 32 portion of the region, and 42.4% (1,174,521) work in the Virginia portion of the region. The data are aggregated at the county level therefore the data are not at the desired disaggregate level however it is useful to see the general employment and job availability of the region. Table 3.1 shows the percent distribution of where individuals work and reside. Workers in Washington D.C. show the largest share of working outside of where they live. Table 3.1: Percent Distribution of Where Individuals Work and Live. Reside In Work In Washington D.C. Maryland Virginia Washington D.C. 28.2% 42.7% 29.1% Maryland 4.3% 89.6% 6.1% Virginia 2.9% 10.2% 86.8% Source: American Community Survey 2006-2010, 5-year Estimates 3.1.2 Low Income Employment Analysis by Jurisdiction Figure 3.1 shows the percent of low income workers out of the total number of all low income workers within the region. 24.8% of all low income workers in the entire Washington D.C. Metropolitan region are found within the District of Columbia. Large percentages of low income workers are also found in: Montgomery, Prince George?s, and Fairfax County. 33 Figure 3.1: Percent of Low Income Workers by Jurisdiction out of Total Number of All Low Income Workers in the Washington D.C. Metropolitan Region. 3.1.3 Low Income Employment Analysis by Occupation and Sector The occupation codes in the data set are identi ed using the 2000 Standard Occupation Classi cation (SOC) System. However, there is a large portion of N.A. (not applicable) responses (21%). There are 26 designations for occupation indus- try within the SOC system. Table 3.2 shows a breakdown of the jobs within the Washington D.C. Metropolitan Region, the percentage of that speci c occupation that are employed by low income workers, the percentage of low income workers for each occupation as a fraction of the total number of low income workers, and the categorization of the occupation into one of the four sectors identi ed by MWCOG. 20.7% of workers put "N.A." as their job occupation, of which 7.1% reside in low income households, but all this accounts for 22.3% of all low income household workers in the region. Managers (11.3%) are the majority of workers in the region, 34 of which less than 2% are in low income households. The top ve largest occupations employing low income household workers are: o ce jobs (10.3%), sales (8.9%), the food industry (8.0%), the cleaning industry (7.5%), and the construction industry (7.4%). Within each occupation industry, the top ve occupations with the largest percentage of workers in low income households are: agriculture (24.7%), the clean- ing industry (19.9%), the food industry (18.8%), personal care and services (16.7%), and healthcare support (14.6%). Figure 3.2 shows the percentage rate breakdown of all low income workers into the four sectors. The lowest percentage of low income workers are o ce workers. A majority of low income workers are found working other and industrial jobs. Figure 3.2: Percent of Low Income Workers by Sector out of Total Number of All Low Income Workers in the Washington D.C. Metropolitan Region. 35 Table 3.2: Percentage of Low Income Workers for the Washington D.C. Metropolitan Region by Occupation. Occupation Percent of Total Workers Percent of Low Income Workers in each Occupation Percent of Total Low Income Workers Round 8a Cooperative Forecast Employment Data Sector Manager 11.34% 1.86% 3.23% O ce Business 3.36% 1.6% 0.82% O ce Finance 2.95% 1.84% 0.83% O ce Communication 5.83% 1.37% 1.22% Retail Engineering 2.00% 0.99% 0.30% Other Science 1.79% 2.21% 0.60% Other Community and Social Services 1.14% 5.58% 0.97% Retail Legal Services 2.66% 2.42% 0.98% O ce Education 4.45% 5.03% 3.42% Other Entertainment Industry 2.50% 3.75% 1.43% Industrial Medicine (Doctors and Nurses) 3.31% 3.51% 1.77% Other Healthcare Support 1.10% 14.61% 2.44% Other Protective Services 2.25% 5.48% 1.88% Retail Food Industry 2.79% 18.77% 7.98% Industrial Cleaning Industry 2.49% 19.86% 7.54% Industrial Personal Care and Service 2.20% 16.66% 5.60% Retail Sales 6.54% 8.88% 8.86% Retail O ce 10.18% 6.60% 10.25% O ce Agriculture 0.07% 24.69% 0.28% Industrial Construction 3.80% 12.69% 7.36% Industrial Extraction 0.01% 7.52% 0.02% Industrial Repair 1.88% 6.76% 1.94% Industrial Production 1.55% 8.79% 2.09% Industrial Transportation 2.55% 14.14% 5.50% Other Military 0.55% 4.88% 0.41% Other Not Applicable 20.69% 7.06% 22.28% Other Total 100% 6.56% 100% Source: American Community Survey 2006-2010, 5-year Estimates 3.2 Low Income Employment Forecast Analysis A statistical and spatial of the Washington D.C. Metropolitan Region is per- formed to examine the low income job location, forecast, and growth using the 36 American Community Survey 2006-2010, 5-year estimates along with Round 8a Coooperative Forecast Data from the Metropolitan Washington Council of Govern- ments. 3.2.1 Low Income Employment Forecast Data The data analysis from job availability is used in conjunction with the Round 8a Cooperative Forecast Employment Data, a long-range economic and demographic forecast. The Round 8a Cooperative Forecast Employment Data are aggregated at the Transportation Analysis Zone (TAZ) level, which amounts to 3,675 zones. This encompasses the Washington D.C. Metropolitan Region of interest (2,950 zones) and then some. Both a regional top-down and local bottom-up approach is employed for the Round 8a Cooperative Forecast Employment Data. A regional model based on national economic and demographic factors is used to forecast. Employment is forecasted with 2005 being the base year and forecasts for the years 2010 to 2040 and every ve years in between for every TAZ within the region. The dataset also includes: land area (in acres and square miles), number of households, population, and group quarter population. Given the employment and land area, employment density can be calculated for each TAZ and GIS maps can be created for a visual depiction of job availability and large concentration of jobs employed by low income peoples. Employment data are separated into four occupational sectors: industrial, re- tail, o ce, and other. The poverty analysis from above is used to forecast and 37 identify locations of low income jobs within the region. The occupation codes des- ignated by the SOC System are divided into the four categories: industrial, retail, o ce, and other as shown in Table 3.2. In the Round 8a Cooperative Forecast Em- ployment Data set, there are 3,780,290 jobs in the Washington D.C. Metropolitan Region for the 2005 base year. Of which, 14% are categorized as industrial jobs, 18% retail, 47% o ce, and 21% other. 3.2.2 Low Income Employment Forecast by Jurisdiction A statistical and spatial of the Washington D.C. Metropolitan Region is per- formed to examine the low income job location, forecast, and growth within each jurisdiction using the American Community Survey 2006-2010, 5-year estimates along with Round 8a Coooperative Forecast Data from the Metropolitan Wash- ington Council of Governments. 3.2.2.1 Employment Forecast by Jurisdiction Methodology Figure 3.3 shows the percent of low income workers within each jurisidction. This means that 6.35% of all employed individuals in the District of Columbia are low income individuals. These percentages of low income workers from the ACS were used to forecast and identify concentrations of low income workers in the region from the Round 8a Cooperative Forecast Employment Data. Because the ACS data are aggregated at the county (jurisdiction) level, it was assumed that the percentage of low income workers within the jurisdiction is uniformly distributed across all TAZs 38 for that jurisdiction. For example, it is assumed for all TAZs within the District of Columbia, 6.35% of workers in all of these TAZs are low income workers. This is a very general assumption for each jurisdiction within the region; therefore, it should be recognized that certain TAZs will have larger concentrations of low income workers, but without additional information on individual TAZs it would not be possible to distinguish between them. Figure 3.3: Percent of Low Income Workers within Each Jurisdiction in the Wash- ington D.C. Metropolitan Region. 3.2.2.2 Low Income Employment Forecast by Jurisdiction Results Table 3.3 is the result of the applied assumption that low income workers are uniformly distributed across all TAZs within each jurisdiction. The growth of low income jobs range from 22% in Arlington (with one of the highest employment densities within the region) to 107% in Loudoun County (with one of the lowest 39 employment densities within the region). Table 3.3: Low Income Employment Forecast by Jurisdiction for the Washington D.C. Metropolitan Region. Jurisdiction 2005 2010 2015 2020 2025 2030 2035 2040 Percent Change Total 171,934 180,011 190,569 203,991 214,736 225,300 235,146 243,848 41.83% District of Columbia 41,508 43,434 45,815 48,561 50,145 52,403 54,677 56,202 35.40% Frederick, MD 7,776 9,058 9,639 10,072 10,377 10,611 10,846 11,097 42.71% Montgomery, MD 29,178 29,808 31,418 33,665 35,836 38,164 39,832 40,909 40.20% Prince George, MD 29,302 30,132 31,065 32,100 33,303 34,766 36,597 38,989 33.06% Arlington, VA 9,430 9,719 9,873 10,505 11,002 11,176 11,434 11,483 21.77% Alexandria, VA 7,128 7,284 7,670 7,950 8,421 8,699 9,156 9,442 32.46% Fairfax, VA 27,564 29,113 30,724 32,973 34,466 35,697 36,687 37,626 36.50% Loudoun, VA 9,077 10,206 11,723 14,130 15,912 17,169 17,980 18,786 106.96% Prince William, VA 10,971 11,257 12,642 14,035 15,274 16,615 17,937 19,314 76.05% Figure 3.4: 2005 (Base Year) and 2040 (Forecasted Year) Low Income Employment Dot Density. The location of this low income employment growth is in Figure 3.4, where the 2005 base year data and the 2040 forecasted year data is plotted. As it can be 40 seen, there are larger clusters of low income employment along the major highway corridors. As it can be seen, the job growth (Frederick County, Loudoun County, and Prince William?s County) does not correspond well with the current location of low income households (District of Columbia and Prince George?s County). 3.2.2.3 Low Income Employment Forecast by Sector Methodology Figure 3.5 shows the percent of low income workers within each sector. This means that 12.43% of all employed individuals in the region are working in the industrial sector. These percentages of low income workers from the ACS along with the Round 8a Cooperative Forecast Employment Data were used to forecast the sector in which low income workers are employed. Because the ACS data are aggregated at the county (jurisdiction) level, it was assumed that the percentage of low income workers within the sector is uniformly distributed across all TAZs. For example, it is assumed for all TAZs within the entire Washington D.C. Metropolitan Region, 12.43% of all industrial workers reside in low income house- holds, 6.76% of all retail workers reside in low income households, 3.46% of all o ce workers reside in low income households, and 6.61% of all other workers reside in low income households. This is a very general assumption for the entire region; therefore this percentage of low income workers by sector is calculated for each jurisdiction. It should be recognized that certain TAZs will have larger concentrations of jobs of a particular sector, but without additional information on individual TAZs it would not be possible to distinguish between them. 41 Figure 3.5: Percent of Low Income Workers within Each Sector in the Washington D.C. Metropolitan Region. 3.2.2.4 Low Income Employment Forecast by Sector Results The growth of low income jobs for each sector for the Washington D.C. Metropolitan Region is also shown in Table 3.4. Low income o ce jobs are expected to have the largest job growth at 53.0% and other low income jobs are expected to have the smallest job growth at 30.5% which corresponds to the increase share of low income o ce jobs (" 2.4%) and the decrease in other low income jobs (# 1.9%). However o ce jobs within the region have the lowest percentage of low income work- ers (ref. Figure 3.5). Therefore a shift is seen to have more low income o ce jobs than other low income jobs. The largest number of low income workers is found in o ce jobs, in which growth is forecasted to be 53%. The low income employment growth is then separated by sectors to visualize the location of low income employment for the industrial, retail, o ce, and other sectors. Figure 3.6(a) shows low income employment for the industrial sector clus- 42 Table 3.4: Low Income Employment Forecast by Sector for the Washington D.C. Metropolitan Region. 2005 2010 2015 2020 2025 2030 2035 2040 Percent Change Employment Forecast 2,925,554 3,075,139 3,280,147 3,546,854 3,762,996 3,948,338 4,119,340 4,275,070 46.13% Industrial 324,086 337,693 352,357 376,800 393,823 418,018 439,913 452,189 39.53% Retail 530,519 564,064 594,505 636,686 668,951 699,765 729,570 758,954 43.06% O ce 1,500,577 1,576,613 1,703,094 1,876,953 2,021,355 2,130,357 2,229,306 2,319,928 54.60% Other 570,372 596,769 630,191 656,415 678,867 700,198 720,551 743,999 30.44% Low Income Employment 171,934 180,011 190,569 203,991 214,736 225,300 235,146 243,848 41.83% Industrial 41,270 42,885 44,701 47,699 49,796 52,815 55,572 57,096 38.35% Retail 38,456 40,747 42,933 45,826 48,051 50,275 52,469 54,738 42.34% O ce 51,952 54,389 58,567 64,223 69,063 72,846 76,297 79,493 53.01% Other 40,256 41,990 44,368 46,243 47,826 49,364 50,808 52,521 30.47% tering around major highway corridors (e.g., I-95, I-270, and I-66). There is a larger cluster of low income industrial employment around the Reston, Virginia area. As seen in Figure 3.6(b), the retail sector employs a lot of low income indi- viduals in a more dispersed manner across the region. This is expected due to the nature of retail positions. There is clustering around major highway corridors (e.g., I-95, I-270, I-66, Rt. 70, etc.) as many retail businesses would operate. Retail businesses are more evenly distributed in the Virginia portion of the region which would be expected for the suburban lifestyles seen in Fairfax, Loudoun, and Prince Williams County, Virginia. The largest percentage of low income jobs are o ce jobs when not taking into account those employed in the "other" sector. This is apparent in Figure 3.6(c). The clustering of low income o ce jobs is also seen around major corridors (i.e., I-95, I-270, Rt. 70) and much more densely packed within the Capital Beltway (i.e., 43 I-495). The distribution of low income jobs in the "other" sector is shown in Figure 3.6(d). The largest percentage of low income jobs can be found in jobs outside the: industrial, retail, and o ce sectors. This includes a wide range of occupations, consequently, trends for these low income jobs can only be compared to the other three sectors. The distribution of low income jobs in the "other" sector is uniform throughout the region except in Montgomery County where a majority of low income jobs are in the industrial, retail, and o ce sectors. (a) Industrial Sector. (b) Retail Sector. 44 (c) O ce sector. (d) Other Sector. Figure 3.6: Low Income Employment by Sector Forecasted for 2040. 3.2.3 Low Income Employment Forecast Major Trends A large percentage of low income workers are found to be currently employed in: the District of Columbia, Montgomery County, Prince George?s County, Fairfax County. From the forecast of low income employment, it can be seen that a major- ity of employment growth is seen in the suburban areas and counties surrounding Washington D.C. These can be seen in: Loudoun County, Prince Williams County, and Frederick County. This employment growth is also spatially clustered around major corridors like Interstate-270 and Interstate-66. Therefore these areas should be places of focus for increasing accessibility to these regions. To ll these new po- sitions, low income individuals will have to spread out even more. These job growth locations do not coincide with where low income workers currently live; therefore, 45 low income individuals may need to travel even further to obtain these jobs, mak- ing this highly problematic. This alludes to the possibility of a spatial mismatch between the two. 3.2.3.1 Median Household Income and Value To better understand if there is a spatial mismatch occuring, the median house- hold inocme and value taken from the ACS was examined at the U.S. Census tract level. Figure 3.7(a) shows the superposition of the low income household and low income employment growth. There are several TAZs where the median household income is identi ed as greater than $250,000. Lower income areas are seen in the eastern region of Wash- ington D.C., portions of Prince Georges County, northern portions of Frederick County, and portions of Northern Virginia (parts of Arlington, Alexandria, and Fairfax County). The more a uent areas are seen in Montgomery, portions of Fair- fax County, and overall Loudoun County. There are also several TAZs where the median household value is identi ed as greater than $1,000,000. As expected, house- hold income and household value are highly correlated such that the more a uent neighborhoods correspond to residing in more expensive housing and lower income area residents residing in less costly homes. In the eastern portion of the region including: eastern Washington D.C. and Prince Georges County there is a large region of low income households with some dispersed low income employment. This is also seen where there is major corridors 46 such as Interstate-270 thorough Montgomery County and Frederick County and along Interstate-95 through Prince Williams County. However, there are large areas of high income households right outside the immediate vicinity of these major inter- states/highways. Therefore these low income workers either live in the immediate vicinity of the corridor or major commuting is required to reach these employment centers. Also the growth of low income employment in Loudoun County is expected to be quite high, however, the overall median household income level and the median household value (ref. Figure 3.7(b)) is quite high. Therefore low income workers in the area would have to perform a reverse commute into the suburbs of Loudoun County. (a) Median Household Income. (b) Median Household Value. Figure 3.7: Spatial Analysis of Median Household Income and Value and Low In- come Employment. 47 3.3 Low Income Employment Analysis and Forecast Conclusions Growth of low income jobs can be seen along major corridors and counties surrounding Washington D.C. With already high employment density, the District of Columbia is expected to have limited job growth. Hence, the most growth is seen within the surrounding counties outside the District of Columbia. Growth is seen along the major interstate/highway corridors, I-270, I-66, and I-95 within Prince George?s, Prince William?s, and Fairfax County. However, there is also a dispersal of jobs seen throughout the county. This is especially apparent in Loudoun County where low income job growth is dispersed throughout the county. Therefore, low income employment are increasingly being located within the surrounding counties, moving further away from the Washington D.C. business district. But looking at the higher median household value of these areas (as compared to the household value in other counties), it may not be feasible to work and reside in other than Prince George?s County and parts of Washington D.C. The overall low income job growth is lower than the total employment growth for the region as a whole. There is a higher distribution of low income jobs within the industrial, retail, and other jobs sector as compared to the employment growth distribution of the entire region. More than 50% of the all jobs within the region are o ce jobs. However, out of the total low income jobs for the entire region, only 33% of them are o ce jobs. Therefore a majority of the forecasted o ce job growth will not be low income o ce jobs. Future studies should attempt to minimize the assumption used to forecast 48 low income job growth with additional TAZ information. This will result in a more realistic forecast because currently, is impossible to distinguish those individual TAZs with higher low income households when assuming a uniform low income employment distribution across a big area such as the PUMA. 49 Chapter 4 National Household Travel Survey Trip Analysis 4.1 Trip Analysis 4.1.1 Introduction As mentioned in Chapter 1, research has shown that the daily travel patterns of low income individuals are quite di erent from higher income level individuals. The travel time and distance is analyzed di erentiated by mode, work day, employment status, number of household vehicles, gender, and destination purpose. Looking at who makes di erent types of trips, on di erent modes with di erent employment status gives a better understanding of the travel patterns of low income individuals. Understanding these travel patterns can give a better sense of the transportation di culties that pretain to low income individuals and households. 4.1.2 Data The 2007/2008 National Household Travel Survey (NHTS) data set (i.e., per- son, household, and trip data les) was used to analyze the trips made by low income individuals within the Washington D.C. Metropolitan Region. The NHTS individuals were ltered to just those living within the Washington D.C. Metrpoli- tan region. Individuals in low income households were identi ed as those households 50 whose household income falls below 1.5 times the poverty level as identi ed in Table 2.1 in Chapter 2. 4.1.3 Trip Analysis Results and Discussion Figure 4.1 shows the average number of trips per low income and non-low income individuals and the average travel time and distance per trip for all trip purposes and modes. Low income individuals spend 25% more time traveling during each trip but travel 21% fewer miles than higher income level individuals. This extra time spent traveling adds up over multiple trips contributing to substantial amounts of total daily travel time. Therefore it is especially important for low income individuals to have fast, convenient, and reliable transportation modes. (a) Average Number of Trips. (b) Average Travel Time per Trip. (c) Average Travel Distance per Trip. Figure 4.1: Trip Analysis for All Trip Purposes and Modes. Source: 2007/2008 National Household Travel Survey 51 4.1.3.1 Trip Analysis by Mode The transportation mode taken by low income individuals can explain the longer travel times (ref. Figure 4.2(a)). Low income individuals use transit twice as much as higher income level individuals. There are also 37% more drivers for trips made by non-low income indiviudals. More trips made by carpooling, walking, and other transportation mode trips for low income individuals as well. The average travel time and distance for these trips (ref. Figure 4.2(b) and 4.2(c)) explains the increased travel time per trip for low income individuals. For al- most every mode, low income individuals spend more time per trip traveling shorter distantances than non-low income individuals. Though the transit travel times are comparable, low income individuals travel nearly 46% less far on transit. This al- ludes to the idea that rail transit attracts more discretionary due to its cost (as mentioned in Chapter 1). Hence, higher income level individuals can a ord the pricier rail transit that have longers travel distances with comparable travel times as other public transit modes. 52 (a) Percent Mode Share. (b) Average Travel Time by Mode per Trip. (c) Average Travel Distance by Mode per Trip. Figure 4.2: Trip Analysis by Mode. Source: 2007/2008 National Household Travel Survey 53 4.1.3.2 Trip Analysis by Work Day It is expected that the travel pattern for work days and non-work days would be di erent (ref. Figure 4.3). However, it was unexpected to see that there are more trips made during non-work days for both low income individuals and non-low income individuals. The overall travel time and distance is more for work day trips. Again it is seen that low income individuals have longer travel times for shorter travel distances. This is especially prominent for non-work day trips. (a) Average Number of Trips by Work Day. (b) Average Travel Time by Work Day. (c) Average Travel Distance by Work Day. Figure 4.3: Trip Analysis by Work Day. Source: 2007/2008 National Household Travel Survey 54 4.1.3.3 Trip Analysis by Employment Status Employment is separated from work day such that Figure 4.4(a) represents trips made by individuals who identi ed as either employed or unemployed. Again it is apparent the di erences in the number of trips, average travel time, and average travel distance between low income and non-low income individuals (ref. Figure 4.4). Not employed individuals travel shorter distances overall compared to employed individuals. (a) Average Number of Trips by Employment Status. (b) Average Travel Time by Employment Status. (c) Average Travel Distance by Employment Status. Figure 4.4: Trip Analysis by Employment Status. Source: 2007/2008 National Household Travel Survey 55 4.1.3.4 Trip Analysis by Number of Household Vehicles The average number of household vehicles and the distribution of the number of household vehicles is shown in Figure 4.5. Low income households are six times more likely to not own a vehicle with an average car ownership rate of less than one. 78% of low income households own either one or no vehicle. 78% of higher income households own either one or two vehicles resulting in a vehicle ownership rate of 1.62. (a) Average Household Vehicle Ownership. (b) Distribution of Number of Household Vehicles. Figure 4.5: Vehicle Ownership. Source: 2007/2008 National Household Travel Survey The average number of trips given the number of household vehicles is shown in Figure 4.6(a). The travel time per trip for low income individuals for all numbers of household vehciles is lower than for higher income individuals. For households with no vehicles, trip travel distance is lower for higher income individuals, therefore they are living closer to work, shopping, school, etc, enabling them to make shorter trips. For households with one or more vehicles, trip travel distance increases for low and 56 non-low income individuals. Therefore with the attainment of at least one vehicle, individuals are willing to travel further to work, shopping, school, etc. However, overall, low income individuals still make less number of trips even for households with equivalent number of vehicles (ref. Figure 4.6(c)). (a) Average Number of Trips by Number of Household Vehicles. (b) Average Travel Time by Number of House- hold Vehicles. (c) Average Travel Distance by Number of Household Vehicles. Figure 4.6: Trip Analysis by Number of Household Vehicles. Source: 2007/2008 National Household Travel Survey 4.1.3.5 Trip Analysis by Gender The average number of trips made by males and females (separated by low income and non-low income) are shown in Figure 4.7(a). There isn?t much di erence in the number of trips made and the average travel time and distance between males 57 and females. However, there is a slight trend for females to make more trips, but take less time traveling shorter distances (ref. Figure 4.7(b) and 4.7(c)). This may be explained with household chores being undertaken by females more so than males. Again it is apparent the di erences in the number of trips, average travel time, and average travel distance between low income and non-low income individuals. (a) Average Number of Trips by Gender. (b) Average Travel Time by Gender. (c) Average Travel Distance by Gender. Figure 4.7: Trip Analysis by Gender. Source: 2007/2008 National Household Travel Survey 4.1.3.6 Trip Analysis by Origin and Destination Purpose The origin-destination purpose is shown in Figure 4.8(a). The travel time and distance per trip for each origin-destination purpose is shown in Figure 4.8(b) and 4.8(c). Note that only select origin-destination purposes are shown. Intercity rail, 58 intercity bus, airport, gas station, and external origin-destination purposes were not included. As expected there are less trips made to and from work for low in- come individuals as compared to higher income individuals. They make comparable number of shopping trips. Low income individuals only make slightly more trips to daycare than non-low income individuals and make nearly twice as many school trips than non-low income individuals. When comparing travel time and distance, low income individuals consistently spend more time traveling to these destinations while traveling shorter distances. (a) Destination Purpose. (b) Average Travel Time by Destination Pur- pose. (c) Average Travel Distance by Destination Purpose. Figure 4.8: Trip Analysis by Origin and Destination Purpose. Source: 2007/2008 National Household Travel Survey 59 4.2 Tour Analysis 4.2.1 Introduction As mentioned in Chapter 1, research has shown that the daily travel patterns of low income individuals are quite di erent from higher income level individuals. Looking at tours gives a better understanding of the travel patterns of low income individuals and how e ciently they chain trips together. Understanding these travel patterns can give a better sense of the transportation di culties that pretain to low income individuals and households. 4.2.2 Data and Methodology Again, the 2007/2008 National Household Travel Survey (NHTS) data set (i.e., person, hosuehold, and trip les) was used to analyze the trips made by low income individuals within the Washington D.C. Metropolitan Region. The NHTS individuals were ltered to just those living within the Washington D.C. Metrpolitan region. Individuals in low income households were identi ed as those households whose household income falls below 1.5 times the poverty level as identi ed in Table 2.1 in Chapter 2. The trips reported in the travel survey were chained to create home-based tours. Therefore for each tour, the origin and destination is home and all trips in between are chained together for that particular tour. 60 4.2.3 Tour Analysis Results and Discussion The number of trips per tour are shown in Figure 4.9(a). As expected, low income individuals make fewer trips in each tour, therefore they are chaining fewer trips together than non-low income individuals. The average travel time and dis- tance during a tour is also shown in Figures 4.9(b) and 4.9(c). With more trips chained together, higher income individuals travel further distances per tour. Higher income individuals travel further per trip and per tour, resulting in an overall larger daily travel distance than low income individuals (ref. Figure 4.9(d)). (a) Average Number of Trips per Tour. (b) Average Travel Time within a Tour. (c) Average Distance Traveled within a Tour. (d) Daily Distance Traveled. Figure 4.9: Tour, Trip Chaining Analysis. 61 4.2.3.1 Work Statistics The percentage of low income and higher income individuals working from zero to four jobs are shown in Figure 4.10. There is a clear distinction in the number of jobs held by low income and higher income individuals. There are nearly 30% more low income individuals with no job as compared to higher income individuals. Also, there is 25% more higher income individuals with just one job as compared to low income individuals. Figure 4.10: Percentage Share for the Number of Jobs. The time of day in which individuals start work is shown in Figure 4.11. For non low income individuals (ref. Figure 4.11(b)), there is a large trend for workers to go to work by 9 or 10 am and come back to work after lunch around 2 pm, with very few individuals arriving at work after 5 pm. For low income individuals (ref. Figure 4.11(a)), there are more workers arriving before 8 am and after 6 pm as compared to higher income individuals. 62 (a) Time of Day in which Low Income Individuals Start Work. (b) Time of Day in which Non-Low Income Individuals Start Work. Figure 4.11: Time of Day in which Individuals Start Work. 63 4.3 National Household Travel Survey Trip and Tour Analysis Conclusions There is a signi cant di erence between the number of trips and tours made by those living in low income households and those living in higher income households. This is especially emphasized when looking at the total daily travel distance traveled. Low income individuals are consistently traveling shorter distances for their trips and tours. However, low income individuals are also consistently traveling for longer and going less far. This can be explained by the higher mode share percentages for slower forms of transportation (i.e., public transit, walking, and biking). The biggest di erence is seen in public transit usage, where low income individuals make a signi cant percentage of trips by public transit, taveling shorter distances, and for longer travel times. This is supportive of the idea introducted in Chapter 1 that higher income individuals are discretionary riders being able to a ord the pricier rail transit services that have longer travel distances with comparable travel times as the slower bus transit system. The average number of household vehicles is less than one for low income individuals. This is apparent when 35% of households have no vehicles while this accounts for only 6.5% of higher income individuals. For both income levels, an increase from zero household vehicles to one household vehicles increases the average number of trips, but plateaus for higher income levels at two vehicles and above. For the selected origin-destintation purposes, it is obvious that higher income individuals go to work more often than low income individuals. Low income individ- 64 uals also travel to school more often. Day care, shopping, other, drop-o /pick-up, and parking point are all similar for both income levels. When comparing travel time and distance, low income individuals consistently spend more time traveling shorter distances to these destinations. Corresponding with the low share of work trips being made, it is seen that low income individuals are not working when examining the number of jobs held. A signi cantly larger proportion of individuals in low income households are simply not employed. Low income workers also have more time constraints as a larger percentage of them arrive to work in the early hours between 5 and 8 am while those in higher income jobs arrive between 9 and 10 am. This can appeal to the exibility of higher income salary jobs held by non-low income individuals. Also, it is also seen that there are larger number of low income workers starting work at 6 pm as well. Therefore, transit dependent low income workers would need a public transportation system that serves these time frames. Higher income individuals chain more trips together, making on average more trips per tour than low income individuals. However, low income individuals are still traveling longer for shorter distances for less number of trips within a tour. This supports the larger public transit mode share usage seen by low income individuals. This supports the notion of the possibility of transit dependent low income riders mentioned in Chapter 1. This dependency on ine cient, slow, and expensive public transit due to lack of an available privately owned vehicle shines light onto one of the main factors hindering job accessibility for low income individuals. 65 Chapter 5 Policy Analysis 5.1 Introduction The typical recommended programs for urban regions with some level of pub- lic transit connectivity is subsidizing public transportation. Though this is widely suggested, low levels of connectivity or ine cient public transportation networks hinder the e ectiveness of such programs. Therefore while this may be an easier to execute, those who qualify may not be using it to its fullest potential. Two policies, subsidizing public transportation and subsidizing vehicle operational costs, were implemented in this research to analyze its e ect on accessibility and to see which program would be more e cient for increasing job accessibility. The rst step is to choose an accessibility measure. The next step is to measure the accessibil- ity of status quo. And nally, the change in accessibility can be measured when implementing these two strategies. 5.2 Accessibility Methodology Transportation analysis typically uses logit choice models to measure consumer surplus. Changes in transportation cost and time are commonly used to evaluate a traveler?s bene t. In this research, a disaggregate log-sum accessibility approach us- 66 ing the Maryland Statewide Transportation Model (MSTM) is used to measure the consumer surplus. This method measures the full accessibility bene ts from land- use and transport policies when discrete choice travel demand models are available that already produce log-sums. It takes into account changes in both generalized transportation costs and destination utility; thus it is capable of providing acces- sibility from changes in distribution of activities due to transportation or land-use policies. Log-sums are advantageous beecause it incorporates a degree of heterogeneity within the population. The log-sum also incorporates various factors that in uence choice into a common framework. These factors include: di erent travel time and travel cost components, varying service quality, and individual and household at- tributes. More extensive introduction can be found in the textbooks on discrete choice models (e.g. [Train, 2003]). The consumer surplus by de nition is the utility that a person recieves from their choice in monetary terms. The utility of di erent alternatives for a decision maker is comprised of an observed and an unobserved (random) component: Unj = Vnj + nj (5.1) where Unj is the utility that decision maker n obtains from alternative j from al- ternatives j(n = 1; :::; N ; j = 1; :::; J); Vnj is the \representative utility"; and nj captures the unobservable factors that a ect utility. In a standard multinomial logit (MNL) model, with nj i.i.d. extreme value with standard variance, the choice 67 probabilities are given as: Pnj = eVnj P j e Vnj (5.2) If the unobserved component of utility is an independently identically dis- tributed (i.i.d.) extreme value and the utility is linear in income, then the expected utility becomes the log of the denominator of a logit choice probability, divided by the marginal utility of income, plus arbitrary constants. This is often called the \log-sum" and it is the log of the denominator of this logit choice probability. It gives the expected utility for an alternative from a set of alternatives. The log-sum can also be used in policy evaluation in an expression for the consumer surplus [de Jong et al., 2007]. Decision makers would then choose the alternative that provides the greatest utility. Provided that the utility is linear in income, the consumer surplus (CSn) can be calculated in monetary terms as: CSn = 1 n Un = 1 n maxj (Unj8j) (5.3) where n is the marginal utility of income and equals dUnj=dYn, if alternative j is chosen. Dividing the consumer surplus by n translates the utility into monetary terms (e.g., dollars). Yn is the income of person n and Un is the overall utility for person n. If the model is MNL and utility is linear in income, then expected consumer surplus becomes: E (CSn) = 1 n ln JX j=1 eVnj ! + C (5.4) where C is an unknown constant that represents the fact that the absolute value of utility can never be measured. Aside from the division and addition of constants, the 68 expected consumer surplus in a standard logit model is simply the log-sum. Under the usual interpretation of the distribution of errors, E(CSn) is the average consumer surplus for the subpopulation of people who have the same representative utilities as person n. The total population consumer surplus can then be calculated as the weighted sum of E(CSn) over a sample of decision makers, whom have weights that re ect the number of people in the population that have the same representative utilites. 5.2.1 Maryland Statewide Transportation Model The log-sum accessibility analysis is performed using the Maryland Statewide Transportation Model (MSTM) developed by the Maryland State Highway Admin- istration (SHA) that allows for quick, consistent, and defensible estimates of how di erent patterns of future development can change key measures of transportation performance. The MSTM is a multi-layer model working at the regional, statewide, and urban level. Key input data to the MSTM includes the population and em- ployment data by income category for each tra c zone. The highway network is based on the networks from Baltimore and Washington metropolitan planning orga- nizations (MPOs), supplemented by the statewide network and the networks from surrounding states. The transit networks are derived from the Baltimore and Wash- ington MPO networks which include: WMATA, the MTA system, MARC trains, and all local transit systems within the Baltimore-Washington D.C. area. The tran- sit networks also include the incomplete Baltimore Red Line and the Montgomery 69 to Prince George?s County Purple Metro Line and the completed Inter County Con- nector in Montgomery County. 5.2.1.1 Parameters Based on the MSTM framework, trips are divided into four time periods: AM peak, Mid-Day, PM peak, and Night shown in Table 5.1. Table 5.1: Time of Day Periods. Time Period Time of Day AM Peak Period 6:30 am to 9:30 am Mid-Day O -Peak Period 9:30 am to 3:30 pm PM Peak Period 3:30 pm to 6:30 pm Night Time O -Peak Period 6:30 pm to 6:30 am (of the next day) The population, households (categorized by size and ve income levels), and employment (categorized into four industries) are used for trip generation. Income categories are based on the 2000 Census data shown in Table 5.2 . Table 5.2: Income Groups (in 1999 Dollars). Income Group Income Range Median Income 1 Lower Quartile < $20,000 $10,720 2 Lower-Middle Quartile $20,000 to $39,999 $29,840 3 Middle Quartile $40,000 to $59,999 $49,240 4 Upper-Middle Quartile $60,000 to $99,999 $76,350 5 Upper Quartile > $100,000 $161,330 The MSTM is a four-step model. The parameters obtained from the model runs are used to calculate accessibility. The procedure is as follows: 70 5.2.1.2 Zone Selection The MSTM uses the Statewide Model Zones (SMZ) system which consists of 1,607 zones covering all of Maryland and selected counties in adjacent states. To an extent, SMZs conform to census geography to best utilize census data products in model development/updates and model calibration and validation. SMZs range from 0.25 to 10 square miles. In areas where MPOs aggregate at the transportation analysis zone (TAZ) level, they are aggregated to the SMZ level. SMZs are nested within counties and confrom to county boundaries. SMZs correspond to MWCOG identi ed TAZs or are an aggregation of MWCOG identi ed TAZs. Thus, the analysis at the SMZs captures geographical levels that are equivalent to aggregated MWCOG TAZs. The Washington D.C. Metropolitan Region as identi ed by MWCOG was extracted from the MSTM data. Figure 5.1 shows the comparison between SMZs and TAZs in the study area with Washington D.C. magni ed. The SMZs boundaries are shown in red, and are placed on top of TAZs boundaries shown in black. For the accessibility analysis, the income categories are strati ed at the SMZs level. Individual and household information from the U.S. Census is aggregated at the Public Use Microdata Area (PUMA) level. SMZs are nested within and conform to these PUMAs, therefore the result from the accessibility analysis can be aggregated to the PUMA level. 71 Figure 5.1: SMZs (red) Comparison to TAZs (black) (Washington D.C. area mag- ni ed). 5.2.1.3 Trip Generation The trip generation model within the MSTM generates trip productions by trip purpose. The trip distributions is based on joint distributions of households for each SMZ and trip production rates, which is cross-classi ed by household category. Work trips is the main focus of this accessibility analysis. The trips generated for work trips are based on trip production rates cross-classi ed by income and number of workers. In MSTM, the trip generation rates by household category were taken directly from the Baltimore Metropolitan Council (BCM)/Maryland Transit Administration (MTA) model but adjusted to the MSTM income categories (quartiles). 72 5.2.1.4 Trip Attraction The trip attraction model within the MSTM calculates trips by SMZ based on a regression applied to SMZ socioeconomic variables for non-home trip ends. Table 5.3 indicates the variables used for home-based and non-home-based trip purposes in the MSTM trip attraction model. Table 5.3: MSTM Home-Based and Non-Home-Based Trip Attraction Variables. Variable De nition Trip Attraction Purposes HB Work HB School HB Shop HB Other NHB JTW NHB JAW NHB OBO Maintained Variables HH Households X X X WORKERS Workers RE Retail X X X X X OFF O ce Employment X IND Industrial Employment OTH Other Employment X X ENROLL School X Derived Variables NRE Non-Retail Employment X X TE Total Employment X CBDEMP Employment in CBD Zones X X CBD 1 if Zone is in CBD, else 0 X 5.2.1.5 Trip Distribution The trip distribution model within the MSTM uses a gravity model formula- tion which employs composite travel time functions by purpose, highway and transit time, as well as roadway tolls and value of time. 73 5.2.1.6 Mode Choice The mode choice model within the MSTM uses a nested logit choice model, with a nesting structure shown in Figure 5.2. The top nest consists of trips made by vehicles, be it driving alone or share riding, or public transit. In the lower nest, public transit can be accessed by driving or walking. The rail alternative includes light rail (LRT) and metro. The commuter rail (CR) alternative includes AMTRAK and MARC commuter rail services. Figure 5.2: MSTM Mode Choice Model Nested Logit Structure. Mode choice is based on generalized utility functions for auto and transit travel. Separate utilities were developed for peak and o -peak travel times. Variables in auto utilities include: driving time and cost, terminal time and parking costs at the attraction end, and tolls. Variables in transit utilities include: walking and drive- access times, initial wait times, in-vehicle travel time, and transfer time. The portion of each zone within walking distances of transit stops and stations were determined using GIS techniques. Table 5.4 list the variables that are included in the utility expression for each alternative. Table 5.5 shows the nested logit coe cients for the nests and Table 5.6 shows the coe cients for the variables found in the utility functions. 74 Table 5.4: Variables Included in Utility Function Based on Travel Mode Accessed by Walking and Driving. Mode Variable wBus weBus wRail wcRail DA/SR dBus deBus dRail dcRail In-Vehicle Time X X X X X X X X X Terminal Time X Auto Operating Costs X Auto Tolls X Auto Parking Cost X Walk Time X X X X X X X X Initial Wait Time (under 7.5 min) X X X X X X X X Initial Wait Time (over 7.5 min) X X X X X X X X Transfer Time X X X X X X X X Number of Transfers X X X X X X X X Transit Fare X X X X X X X X Drive Access Time X X X X Attraction Zone Area Type Bias X X X X X X Table 5.5: MSTM Nesting Coe cients. Nest Value Walk Transit Route (Bus, Rail, MARC) 0.30 Drive Transit Route (Bus, Rail, MARC) 0.30 Transit Access (Walk vs. Drive) 0.65 Share Ride Occupancy (2 vs. 3+) 0.30 Auto Mode (Drive Alone vs. Shared Ride) 0.65 Table 5.6: MSTM Mode Choice Utility Coe cients. Attribute HBW, JTW In-Vehicle Time -0.025 Terminal Time -0.05 Auto Operating Cost -0.0042 Auto Parking Cost and Tolls -0.0084 Walk Time -0.05 Initial Wait Time (under 7.5 min) -0.05 Initial Wait Time (over 7.5 min) -0.025 Transfer Time -0.05 Number of Transfers -0.125 Transit Fare -0.0042 Drive Access Time -0.05 75 5.2.1.7 Trip Assignment Travel demand forecasts from the MSTM was assigned to a network. Factors were applied to the respective daily trip matrices to derive peak and o -peak trip matrices for network assignment. Separate assignments are done for the AM and PM peak periods and the periods for the rest of the day were combined. Tran- sit trips were assigned on a daily basis, work trips were assigned based on peak service characteristics, and all other trips were assigned based on o -peak service characteristics. 5.2.2 Log-Sum Measure Using the Maryland Statewide Transportation Model 5.2.2.1 Assumptions The assumptions used in the accessibility analysis are: 1. only work trip pur- poses are considered (indexed by p), 2. accessibility measures is aggregated at ve income levels (indexed by i), 3. the 11 mode choice alternatives as de ned in Section 5.2.1.6 (mode choice index by j, mode choice group indexed by m) are considered, 4. and utilities are speci ed with a nested logit structure and parameters as de ned in Section 5.2.1.6. 76 5.2.2.2 Log-Sum Measure by Income Group For each origin zone z within the area, the log-sum is computed for each income group i, travel purpose p (only work trips are considered), and transportation alternatives j J to all destinations within the study area: Lpiz = ln JX j=1 e pVpijz ! (5.5) where p is the nesting coe cient (based on the nested logit structure) and Vpijz is the deterministic utility obtained from transportation mode j. The log-sums are aggregated for work trips at ve income levels. Then the log-sum is converted into travel time by time coe cient, p and then converted into travel cost by external values of time, V oTpi. Thus, the monetary value of the accessibility of zone z for a person of income group i can be written as: CSLpiz = V oTpi 1 p Lpiz (5.6) The accessibiility bene t is measured in terms of consumer surplus, which is expressed in monetary terms as cents (?) in 2000 dollars. The consumer surplus of the total population can be calculated as a weighted sum of log-sums over a sample of decision makers, where the weights re ect the number of people in the population who face the same representative utilities as the sample. Thus, the consumer surplus for each income category i from each origin zone z is calculated by multiplying the log-sum by the number of trips (Apiz) by people in that income category: E (CSn) = V oTpi 1 p " Apiz ln JX j=1 e pVpijz !# (5.7) 77 where (V oTpi 1= p) equates to a negative cost coe cient n mentioned in Section 5.2. Therefore, the larger the utility, the larger the log-sum and the larger number of trips results in a larger negative consumer surplus. Hence, the larger the negative consumer surplus, the greater the accessibility. 5.3 Implemented Policy Programs Two policy programs were implemented in this research. The rst is subsi- dizing transit fare by 50% to see the e ect it had on consumer surplus at di erent income levels. The second is subsidizing driving costs by 50%. Subsidizing the capital cost of vehicle ownership would be interesting to examine, however, the ac- cessibility model used in this study does not take into account capital cost. Also there is a lot of issues with studying capital cost due to the fact that it is a one time payment that is di cult to annualize over the vehicle?s life span. And as stated in Chapter 1, low income individuals tend to keep their vehicles for longer which leads to another problem of higher maintenence costs for older vehicles. In the ac- cessibility model, the operation cost of driving is 9.9 cents per mile which includes gas and maintenance costs. Therefore in this research, the operational costs which include fuel and maintenence costs were subsidized by 50% to see the e ect it has on consumer surplus. The change in consumer surplus when implementing these two policies is also plotted for each SMZ at di erent income levels. These plots show the change in con- sumer surplus in monetary terms, in 2000 dollars because the analysis was performed 78 using the MSTM model which also uses 2000 dollars. This change in consumer sur- plus is measured similar to the study done by de Jong, Daly, Pieters, and ver der Hoorn [2007], method 2. Consumer surplus is plotted for each SMZ for the various available trans- portation modes at di erent income levels available in Appendix B. For a given transportation mode in a given SMZ, the scale is the same at every income level unless noted otherwise; therefore they can be compared. However, the scale between di erent transportation modes within a SMZ is not the same; therefore they can not be compared. However, the overall magnitude of the consumer surplus can be compared across all transportation modes and income levels. 5.4 Analysis Region Seven TAZs (represented by the green stars in Figure 5.3) were selected within the Washington D.C. Metropolitan Region. These TAZs were selected based on the household value, the concentration of low income individuals or jobs, and the variety of available transportation modes within the TAZ. The selected TAZ and their average household income and value are listed in Table 5.7. The accessibility measure is calculated using the Maryland Statewide Trans- portation Model (MSTM), therefore the data is aggregated at statewide modeling zones (SMZs), and thus the corresponding SMZ for the analyzed TAZs are also listed in Table 5.7. The analysis also refers to the alternatives depicted in Figure 5.2 and income levels de nied in Table 5.2. 79 Figure 5.3: Selected TAZs (Green Stars) within Washington D.C., Maryland, and Virginia. Table 5.7: Statistics About Selected Transportation Analysis Zones. Location TAZ Location Median HH Income SMZ Washington D.C. 219 Rhode Island Ave. Metro $28,814 1223 362 Anacostia Metro $19,238 1268 Frederick County, MD 2926 Near I-270 & Rt-70 $42,529 956 Prince George?s County, MD 842 Suitland $39,788 813 1006 Landover, Near Rt-50 $39,028 796 Fairfax County, VA 2044 Huntington $41,277 1317 Loudoun County, VA 2270 Leesburg, Near Rt-267 $38,231 1368 5.4.1 Trips Made The trips made from each of the seven selected SMZs are mapped in Figure 5.4. The trips made by individuals at income level 1 and 2 with origin in SMZ 1223 and 1268 in Washington D.C. is shown in Figure 5.4(a) and 5.4(b), respectively. As expected, most trips are into downtown Washington D.C. The trips for 796 80 (Landover), 813 (Suitland), and SMZ 956 (Frederick) in Maryland is shown in Figure 5.4(c), 5.4(d), and 5.4(e), respectively. As expected, the trips made from Prince George?s County (SMZ 813 and 796) are in Prince George?s County or downtown Washington D.C. From Frederic (SMZ 956), most trips are made in Frederick County expanding into Montgomery County along I-270. The trips made for SMZ 1317 and 1368 is shown in Figures 5.4(g) and 5.4(f). Within Fairfax County (SMZ 1317), most trips are made mostly within Fairfax County. And as expected, most trips made from within Loudoun County (SMZ 1368) stay within Loudoun County. 81 (a) SMZ 1223, Rhode Island Metro, D.C. (b) SMZ 1268, Anacostia Metro, D.C. (c) SMZ 796, Landover, MD (d) SMZ 813, Suitland, MD (e) SMZ 956, Frederick, MD (f) SMZ 1368, Leesburg, VA (g) SMZ 1317, Huntington, VA Figure 5.4: Trips Made by Low Income Individuals with Origin in Washington D.C., Maryland, and Virginia. 82 5.4.2 Income Level The status quo consumer surplus for all SMZs by income level 1, 2, and 3 are shown in the Figures 5.5 to 5.7. As expected, the accessibility increases as the income level increases, for each SMZ. However, the consumer surplus range varies greatly between the di erent SMZs with Prince George?s County SMZs (SMZ 956 and 813) having the lowest overall consumer surplus and accessibility. (a) SMZ 1223, Income Level 1. (b) SMZ 1223, Income Level 2. (c) SMZ 1223, Income Level 3. (d) SMZ 1268, Income Level 1. (e) SMZ 1268, Income Level 2. (f) SMZ 1268, Income Level 3. Figure 5.5: Consumer Surplus for Income Levels 1, 2, and 3 for SMZ 1223 (Rhode Island Metro) and 1268 (Anacostia Metro) in Washington D.C. 83 (a) SMZ 796, Income Level 1. (b) SMZ 796, Income Level 2. (c) SMZ 796, Income Level 3. (d) SMZ 813, Income Level 1. (e) SMZ 813, Income Level 2. (f) SMZ 813, Income Level 3. (g) SMZ 956, Income Level 1. (h) SMZ 956, Income Level 2. (i) SMZ 956, Income Level 3. Figure 5.6: Consumer Surplus for SMZ 956 (Frederick), 813 (Suitland), and 796 (Landover) in Maryland. 84 (a) SMZ 1317, Income Level 1. (b) SMZ 1317, Income Level 2. (c) SMZ 1317, Income Level 3. (d) SMZ 1368, Income Level 1. (e) SMZ 1368, Income Level 2. (f) SMZ 1368, Income Level 3. Figure 5.7: Consumer Surplus for SMZ 1317 (Huntington) and 1368 (Leesburg) in Virginia. 5.4.3 Drive Alone Figures B.1 to B.3 in Appendix B show the consumer surplus for driving alone at income levels 1, 2, and 3. For all SMZs, there is a large di erence in driving accessibility between the di erent income levels. Individuals at income level 1 are less likely to own a vehicle therefore it is expected that they experience the lowest 85 accessibility. For SMZ 1223 and 1268 in Washington D.C. at income level 1 and 2, there is higher accessibility for northern Virginia (e.g., Arlington, Alexandria, Fairfax), therefore it is easier to reach northern Virginia by driving than downtown Washing- ton D.C. where most trips are being made. The decreased accessibility within the District can be explained due to the high congestion and other unfavorable driv- ing conditions associated with downtown Washington D.C. For SMZ 813 and 796 in Prince George?s County, MD, the consumer surplus seems to be more spread out be- tween Prince George?s County, D.C., and northern Virginia. Within Frederick, MD (SMZ 956), there is high accessibility for Frederick and Montogomery County, there- fore reaching the District is di cult even when driving. For SMZ 1317 in Fairfaix, VA, there is higher accessibility and southern Fairfax County and Prince William?s County, than within Washington D.C. and Maryalnd. For SMZ 1368 in Leesburg, VA, there is higher accessibility to stay within Loudoun County until income level 3, where the accessible regions greatly increases. The change in consumer surplus when subsidizing operational costs for trans- portation alternative driving alone is shown in Figure ??. For all SMZs, there is a large increase in consumer surplus between income level 1 and 2. This is also re- ective of the fact that individuals in income level 1 are less likely to own a vehicle, therefore subsidizing operational costs will not be helpful to them. Surprisingly the largest SMZ that bene ts from subsidized operational costs is in Huntington, VA. Therefore, these individuals have the propensity to drive even though they have access to public transit. Frederick, MD (SMZ 956) and Leesburg, VA (SMZ 1368) 86 Figure 5.8: Change in Consumer Surplus when Subsidizing Operational Costs for Driving Alone in 2000 Dollars. has the least available public transportation, therefore, most residents will rely on driving or carpooling. Then, it is expected that these areas would have a large increase in consumer surplus as evident in the gure. 5.4.4 Share Ride Figures B.4 to B.6 in Appendix B show the consumer surplus for share ride (carpooling) at income levels 1, 2, and 3. Carpooling accessibility patterns are very similiar to drive alone accessibility pattners. However, the consumer surplus scale is much smaller than for driving alone, as expected. Again, there is a large carpooling accessibility between the di erent income levels. The change in consumer surplus when subsidizing operational costs for trans- portation alternative share ride (i.e., carpooling) is shown in Figure ??. The same trend is seen for the consumer surplus change in alternative drive alone in the pre- 87 Figure 5.9: Change in Consumer Surplus when Subsidizing Operational Costs for Share Ride (Carpooling) by Income Level in 2000 Dollars. vious section. This is expected with the similar accessibility patterns in share ride and drive alone. Also there are signi cantly less carpoolers than drivers, therefore the consumer surplus increase when subsidizing operational costs is expected to be lower than the consumer surplus increase when driving alone. 5.4.5 Bus Transit The destinations reachable by bus transit only (accessed by walking) is shown in Figures B.7 to B.9 in Appendix B. For all SMZs, an increase in consumer sur- plus and accessibility is observed with increasing income level. However, for all SMZs except for those within Washington D.C., the areas reached by bus transit accessed by walking decreases with increasing income level. Therefore higher income individuals whom use bus transit tend to travel in shorter distances with grouped together. The SMZs that can access bus by walking are: SMZ 1223 (Washington 88 D.C.), 1268 (Washington D.C.), 813 (Suitland, MD), 796 (Landover, MD), and 1317 (Huntington, VA). The area reached from SMZ 1223 and 1268 is contained mainly within in the Capital Beltway. The area reached from SMZ 813, Suitland, MD spans Washington D.C. and Prince George?s County. The area reached from SMZ 796, Landover, MD spans the north-east portion of Washington D.C. and Prince George?s County inside the Capital Beltway. The area reached from SMZ 1317, Huntington, VA spans the downtown District and into Fairfax County. The destinations reachable by bus transit only (accessed by driving) is shown in Figures B.10 to B.12 in Appendix B. Due to limited available parking near bus transit stops, SMZs that can access bus by driving are: SMZ 1268 (Washington D.C.), 956 (Frederick, MD), 796 (Landover, MD), and 1317 (Huntington, VA). For all SMZs, an increase in accessibility is observed with increasing income level. This is especially evident because this transportation alternative requires acess by driving, where low income households are less likely to own a vehicle. The area reached from SMZ 1268, Anacostia Metro, D.C. and SMZ 1317, Huntington, VA mainly reaches inside the Capital Beltway with less coverage than when acessing bus transit by walking. Within Frederick, MD (SMZ 956), there is no access to bus transit by walking, however, there is access to bus transit by driving. As expected, those who would drive to take bus transit would travel futher and in this case, they travel along the I-270 corridor. That leaves accessibility within the city of Frederick is achieved by driving alone. This will present di culty for low income households with no vehicles. Those in higher income levels have higher accessibility and travel futher away from I-270 and making some trips into downtown Washington D.C. as well. 89 These individuals may be avoiding traveling along the congested I-270 corridor. The areas reached from SMZ 796 by bus transit accessed by driving is very similar to the those areas reached when accessed by walking. For some areas, transit may be used in favor of the metro rail system due to its cost and coverage. This is especially true when traveling shorter distances or areas that are not covered by the metro rail system. The accessibility already present in some areas like Washington D.C. and northern Virginia will support the new low income employment growth expected those areas. There is large employment growth expected in the two suburbs of Frederick County, MD and Loudoun County, VA. However, the two suburban SMZs, 956 (Frederick MD) and 1368 (Loudoun County, VA) do not have bus transit that can be accessed by walking. This is a major problem for low income households that do not own vehicles. Therefore, they must rely on other transportation modes like carpooling and other available public transit. For public transportation, bus transit has a higher consumer surplus magnitude than rail transit. The change in consumer surplus when subsidizing transit fare for the bus transit alternative is shown in Figure 5.10. For all SMZs except SMZ 956 (Frederick, MD) and 1368 (Leesburg, VA), there is a large increase in accessibility moving from income level 1 to income 2. Also in these cases, income level 2 has the largest increase in accessibility. Large increases in consumer surplus is seen at income level 2 and 3 indicating that those in income level 2 and 3 are the largest users of bus transit, with income level 2 individuals having making up the biggest share of bus transit riders. Higher income individuals use transit less, therefore the increase 90 Figure 5.10: Change in Consumer Surplus when Subsidizing Transit Fare for Bus Transit by Income Level in 2000 Dollars. in accessibility is less for increasing income levels. Frederick, MD (SMZ 956) and Leesburg, VA (SMZ 1368) have already shown to have little to no bus transit, therefore they would not bene t much from subsidizing transit fare until a public transit system is implemented there. 5.4.6 Rail Transit The destinations reachable by rail transit only (accessed by walking) is shown in Figures B.13 to B.15 in Appendix B. For all SMZs, an increase in consumer surplus and accessibility is observed with increasing income level. However, for all SMZs the areas reached by rail transit accessed by walking decreases with increasing income level. The SMZs that can access rail transit by walking are: SMZ 1223 (Washington D.C.), 1268 (Washington D.C.), 813 (Suitland, MD), 796 (Landover, MD), and 1317 (Huntington, VA). The area reached from SMZ 1223 (next to Rhode 91 Island Metro station) and 1268 (next to Anacostia Metro station) is mostly outside of the Capital Beltway. The areas reached from SMZ 813 (approximately 1.5 miles from Suitland Metro station) and 796 (approximately 3 miles from the New Carrollton Metro station) includes areas inside and outside of the Capital Beltway. However, those from Landover, MD (SMZ 796) and Suitland, MD (SMZ 813) do not travel in areas north of Landover and Suitland and further south into Prince George?s County. The area reached from SMZ 1317 (less than 2 miles from the Huntington Metro station) covers the entire inside of the beltway, into Prince George?s and Montgonmery County and along major corridors (I-270 and somewhat, I-66). The destinations reachable by rail transit only (accessed by driving) is shown in Figures B.16 to B.18 in Appendix B. Due to limited available parking near rail transit stops, SMZs that can access rail by driving are: SMZ 1268 (Washington D.C.), 956 (Frederick, MD), 796 (Landover, MD), and 1317 (Huntington, VA). For all SMZs, an increase in accessibility is observed with increasing income level. This can be explained by the fact that rail transit is less accessible to low income individuals due to their cost and the fact that this alternative is accessed by driving to a metro station. Only the further most metro rail stations, almost exclusively outside the Capital Beltway and along major corridors (i.e., I-270 and I-66) are reached by rail transit accessed by driving. The destinations reachable by commuter rail transit only (accssed by walking or driving) is shown in Figures B.19 to B.22 in Appendix B. There are very few SMZs that can access commuter rail by walking or driving. The SMZs that can access commuter rail by walking are: SMZ 1223 (Washington D.C.), 1268 (Washington 92 D.C.), 813 (Suitland, MD), 796 (Landover, MD), and 1317 (Huntington, VA), which can be accessed by walking or driving. This alternative does not include transferring from rail to commuter rail. The destinations reached on commuter rail is expected because the nearest commuter rail line is the MARC commuter rail service. It should be noted that in Leesburg, VA in Loudoun County (SMZ 1368), there is no consumer surplus for any public transit mode, be it: bus, express bus, rail, or commuter rail being accessed by walking or driving. This is expected as the closest metro rail station is over 20 miles away. Loudoun County is expecting a lot of low income employment growth and the consumer surplus and accessibility of Loudoun county is much less than other observed SMZs due to its lack of available public transit services. Therefore, low income individuals that live within Leesburg would need to have a vehicle to access anywhere. If low income households have no vehicles, this is especially problematic in Loudoun County. However, when look- ing at the accessibility of driving and carpooling, it isn?t until income level 3 that the destination region is reached. Therefore, even if low income households own a vehicle, they may not be able to reach their destionation if it is too far outside of Loudoun County. For all SMZs, the destinations reachable by rail transit accessed by walking is the inverse of their respective destinations reachable by bus transit accessed by walking. Therefore, within the downtown District, it is easier to travel by bus transit, and futher distances is more easily achieved by rail transit. Those in higher income levels have a higher consumer surplus for rail transit that follow along major corridors (i.e., I-270, I-95, and I-66). The higher consumer surplus and accessibility 93 for higher income levels corressponds to the idea that rail users are discretional riders Chapter 1. In combination with bus transit, the region where most trips are being made for most SMZs is covered, except for the more outlying suburbs (i.e., Frederick County, MD, Loudoun County, VA, and Prince William?s County, VA). These are also the areas where there is a large low income employment growth is expected. Low income individuals who want to access these areas must then rely on carpooling or driving. This is problematic for low income households with no vehicles. Figure 5.11: Change in Consumer Surplus when Subsidizing Transit Fare for Rail Transit by Income Level in 2000 Dollars. The change in consumer surplus when subsidizing transit fare for the rail transit alternative is shown in Figure 5.11. For all SMZs except SMZ 956 (Frederick, MD) there is a large increase in consumer surplus moving from income level 1 to income 2. Also in these cases, income level 2 has the largest increase in consumer surplus. Large consumer surplus is seen at income level 2 and 3 indicating that 94 those in income level 2 and 3 are the largest users of rail transit, with income level 2 individuals having making up the biggest share of rail transit riders. Higher income individuals use transit less, therefore the increase in accessibility is less for increasing inocme levels. Frederick, MD (SMZ 956) have already shown to have no rail transit, therefore they would not bene t much from subsidizing transit fare until a public rail transit system is implemented there. The overall consumer surplus increase for rail transit is less than the consumer surplus increase seen in bus transit. 5.5 Policy Analysis Conclusions Subsidizing operational costs does not increase the accessibility for individuals at income level 1 nearly as much as for those at income level 2. This is also re ective of the fact that individuals in income level 1 are less likely to own a vehicle, therefore subsidizing operational costs will not be helpful to them. Those living in areas with the least available public transportation (i.e., Frederick, MD and Leesburg, VA) see a large increase in accessibility. Subsidizing transit fare does not increase the accessibility for individuals at income level 1. These individuals are not traveling much in general, therefore they would not bene t from the reduced fare. Individuals at income level 2 sees the largest increase in accessibility. A large increase in accessibility is also seen at income level 3, indicating that individuals in income levels 2 and 3 are the largest users of bus and rail transit. Areas with little to no rail or bus transit (i.e., Frederick, MD and Leesburg, VA) would not bene t from subsidized transit fare. 95 Chapter 6 Conclusions and Discussions A large percentage of low income households are found in: the District of Columbia, Montgomery County, Prince George?s County, and Fairfax County. From the forecast of low income employment, it can be seen that a majority of employment growth is seen in the suburban areas and counties surrounding Washington D.C. These can be seen in: Loudoun County, Prince Williams County, and Frederick County. This employment growth is also spatial clustered around major corridors like Interstate-270 and Interstate-66. These job growth locations do not coincide with where low income workers current live; therefore, low income individuals may need to travel even further to obtain these jobs, making this highly problematic. This alludes to the possibility of a spatial mismatch between where low income individuals live and where they work. Looking at the trip patterns of low income individuals, it is apparent that low income individuals spend more time traveling shorter distances. This is explained by the large larger portion of low income individuals that use slower transportation modes including walking, biking, and especially public transit. A larger portion of transit users are low income transit captives, therefore more low income individuals rely on the existing public transportation system. The typical recommendation for urban areas with some level of connectivity via public transportation is to subsidize 96 public transportation. In this case, subsizing public transit fare and subsidizing operational costs were analyzed. Subsidizing operation costs did not increase accessibility for those of very low income. This is expected as those in this income level are much more unlikely to own a vehicle in the rst place. Subsidizing transit fare also does not increase accessibility for individuals for those of very low income. This can be explained by the fact that individuals in this income level (i.e., income level 1) do not travel very much in the rst place. However, there is large increase in accessibility for those at slightly higher income levels. Areas with little or no available public transit see no bene t for this subsidization. Even with the increased accessibility from subsidizing transit fare, there is still the problem that certain areas are still not being reached. Therefore low income individuals in the District of Columbia, Montgomery County, Prince George?s County, and Fairfax County will have trouble reaching the job growth in Loudoun County, Prince William?s County, and Frederick County. 6.1 Future Work The lack of connectivity of suburban counties limits the job accessibility for low income individuals. Therefore, speci cally focusing on connecting these outer suburban counties creating a connected public transit network would be bene cial to increasing accessibility for the expected job growth. Also, future studies should measure the change in consumer surplus and accessibility when subsidizing the cap- ital cost of owning a vehicle, rather than subsidizing marginal operational costs. 97 Appendix A Statistical Analysis of Low Income Households and Individuals Table A.1: Low Income Household Percentages in Washington D.C. PUMA Overall Distribution within D.C. Distribution within PUMA Not Low Income Low Income Total Not Low Income Low Income Total 00101 16.4% 1.3% 17.7% 92.5% 7.5% 100% 00102 13.9% 2.8% 16.7% 83.0% 17.0% 100% 00103 15.9% 3.7% 19.6% 81.2% 18.8% 100% 00104 15.2% 6.8% 21.9% 69.2% 30.8% 100% 00105 19.8% 4.3% 24.1% 82.2% 17.8% 100% Overall 81.1% 18.9% 100% 81.1% 18.9% 100% Source: American Community Survey 2006-2010, 5-year Estimates Table A.2: Low Income Household Percentages in Maryland. PUMA Overall Distribution within MD. Distribution within PUMA Not Low Income Low Income Total Not Low Income Low Income Total 00300 10.3% 1.0% 11.3% 91.5% 8.5% 100% 01001 5.2% 0.3% 5.5% 95.4% 4.6% 100% 01002 5.5% 0.5% 6.1% 91.2% 8.5% 100% 01003 7.8% 0.7% 8.6% 91.5% 8.5% 100% 01004 9.1% 0.5% 9.7% 94.5% 5.5% 100% 01005 5.6% 0.8% 6.4% 87.3% 12.7% 100% 01006 5.0% 0.6% 5.5% 89.8% 10.2% 100% 01007 5.0% 0.7% 5.7% 87.8% 12.2% 100% 01101 3.7% 0.8% 4.5% 81.2% 17.9% 100% 01102 6.0% 0.5% 6.5% 91.7% 8.3% 100% 01103 4.2% 0.8% 5.0% 84.1% 15.9% 100% 01104 4.3% 0.7% 5.0% 85.5% 14.5% 100% 01105 7.4% 0.3% 7.8% 95.6% 4.4% 100% 01106 6.3% 0.4% 6.7% 93.8% 6.2% 100% 01107 5.1% 0.7% 5.8% 87.9% 12.1% 100% Overall 90.6% 9.4% 100% 90.6% 9.4% 100% Source: American Community Survey 2006-2010, 5-year Estimates 98 Table A.3: Low Income Household Percentages in Virginia. PUMA Overall Distribution within VA. Distribution within PUMA Not Low Income Low Income Total Not Low Income Low Income Total 00100 10.4% 1.0% 11.4% 91.4% 8.6% 100% 00200 7.3% 0.7% 8.0% 91.1% 8.9% 100% 00301 8.7% 0.9% 9.6% 91.0% 9.0% 100% 00302 8.1% 0.7% 8.8% 91.8% 8.2% 100% 00303 7.8% 0.4% 8.2% 94.8% 5.2% 100% 00304 6.7% 0.4% 7.0% 94.4% 5.6% 100% 00305 12.3% 0.7% 13.0% 94.4% 5.6% 100% 00501 7.5% 0.6% 8.2% 92.2% 7.8% 100% 00502 8.1% 0.7% 8.8% 91.9% 8.1% 100% 00600 15.8% 1.2% 17.0% 92.9% 7.1% 100% Overall 92.6% 7.4% 100% 92.6% 7.4% 100% Source: American Community Survey 2006-2010, 5-year Estimates Table A.4: Labor Force Distribution in Washington D.C. PUMA At Least 1 Householder in the Labor Force Couple Neither in the Labor Force Single Female Head of Household Not in the Labor Force Single Male Head of Household Not in the Labor Force Total 00101 30.4% 44.6% 22.5% 2.5% 100% 00102 63.4% 6.0% 24.8% 5.7% 100% 00103 49.2% 10.5% 32.5% 7.8% 100% 00104 50.8% 4.8% 40.5% 3.9% 100% 00105 57.4% 7.8% 30.2% 4.6% 100% Overall 52.9% 7.3% 34.9% 4.9% 100% Source: American Community Survey 2006-2010, 5-year Estimates 99 Table A.5: Labor Force Distribution in Maryland. PUMA At Least 1 Householder in the Labor Force Couple Neither in the Labor Force Single Female Head of Household Not in the Labor Force Single Male Head of Household Not in the Labor Force Total 00300 70.8% 11.0% 17.6% 0.7% 100% 01001 57.2% 20.1% 22.7% 0.0% 100% 01002 81.6% 1.5% 14.6% 2.3% 100% 01003 71.7% 12.0% 13.0% 3.4% 100% 01004 56.4% 30.3% 8.9% 4.4% 100% 01005 76.3% 11.7% 8.6% 3.5% 100% 01006 73.6% 11.8% 12.9% 1.7% 100% 01007 86.4% 5.1% 7.0% 1.5% 100% 01101 84.7% 5.3% 7.6% 2.5% 100% 01102 76.6% 8.1% 13.6% 1.8% 100% 01103 79.9% 7.0% 10.0% 3.1% 100% 01104 59.4% 4.1% 29.6% 6.9% 100% 01105 58.5% 16.1% 19.2% 6.2% 100% 01106 50.7% 20.7% 27.2% 1.3% 100% 01107 69.7% 4.7% 21.5% 4.2% 100% Overall 72.3% 9.7% 15.1% 2.9% 100% Source: American Community Survey 2006-2010, 5-year Estimates Table A.6: Labor Force Distribution in Virginia. PUMA At Least 1 Householder in the Labor Force Couple Neither in the Labor Force Single Female Head of Household Not in the Labor Force Single Male Head of Household Not in the Labor Force Total 00100 76.9% 8.6% 12.6% 1.9% 100% 00200 71.9% 9.0% 17.4% 1.7% 100% 00301 71.7% 15.2% 11.4% 1.7% 100% 00302 75.1% 8.2% 16.1% 0.6% 100% 00303 79.9% 6.7% 11.5% 1.9% 100% 00304 71.8% 13.2% 8.0% 7.0% 100% 00305 67.8% 19.6% 12.6% 0.0% 100% 00501 78.2% 8.9% 12.3% 0.6% 100% 00502 77.7% 7.5% 14.2% 0.6% 100% 00600 69.9% 15.3% 12.6% 2.2% 100% Overall 73.9% 11.4% 13.0% 1.6% 100% Source: American Community Survey 2006-2010, 5-year Estimates 100 Table A.7: Distribution of Low Income Household Size in Washington D.C. PUMA 1 2 3 4 5 6+ Total 00101 66.9% 22.2% 5.1% 4.3% 1.4% 0.0% 100% 00102 55.0% 18.5% 9.2% 9.6% 4.9% 2.8% 100% 00103 53.4% 19.2% 10.6% 6.4% 6.7% 3.8% 100% 00104 38.0% 21.5% 13.6% 14.7% 6.2% 6.1% 100% 00105 65.4% 17.6% 7.6% 5.2% 2.2% 2.0% 100% Overall 51.8% 19.8% 10.4% 9.4% 4.8% 3.8% 100% Source: American Community Survey 2006-2010, 5-year Estimates Table A.8: Distribution of Low Income Household Size in Maryland. PUMA 1 2 3 4 5 6+ Total 00300 44.4% 19.1% 13.2% 10.6% 9.2% 3.4% 100% 01001 28.6% 27.2% 21.9% 12.3% 6.4% 3.5% 100% 01002 24.7% 15.4% 24.1% 12.9% 14.4% 8.5% 100% 01003 39.4% 24.6% 11.5% 12.1% 8.7% 3.7% 100% 01004 1.9% 16.7% 11.9% 4.9% 2.3% 2.3% 100% 01005 36.5% 18.9% 13.2% 14.2% 7.6% 9.6% 100% 01006 35.1% 17.1% 18.0% 14.5% 11.1% 4.2% 100% 01007 48.3% 20.7% 9.6% 10.4% 5.8% 5.1% 100% 01101 27.2% 23.6% 7.2% 15.1% 11.7% 5.0% 100% 01102 44.5% 13.2% 16.1% 9.7% 9.0% 7.4% 100% 01103 42.4% 13.9% 10.7% 10.7% 10.0% 12.3% 100% 01104 38.1% 19.1% 15.5% 11.1% 7.8% 8.4% 100% 01105 47.5% 23.6% 7.7% 12.1% 2.3% 6.8% 100% 01106 53.9% 19.6% 9.9% 7.2% 4.7% 4.6% 100% 01107 43.2% 20.4% 15.2% 9.4% 6.1% 5.8% 100% Overall 40.8% 19.3% 14.2% 11.3% 8.2% 6.2% 100% Source: American Community Survey 2006-2010, 5-year Estimates 101 Table A.9: Distribution of Low Income Household Size in Virginia. PUMA 1 2 3 4 5 6 Total 00100 51.6% 18.6% 12.7% 11.1% 3.1% 2.9% 100% 00200 47.3% 19.5% 13.4% 11.8% 4.4% 3.5% 100% 00301 38.3% 18.4% 15.0% 14.8% 5.9% 7.6% 100% 00302 34.0% 16.6% 18.0% 16.1% 5.8% 9.4% 100% 00303 35.2% 8.4% 17.3% 23.2% 10.2% 5.7% 100% 00304 26.1% 16.8% 17.3% 21.7% 8.2% 10.0% 100% 00305 45.5% 18.7% 12.4% 10.2% 7.8% 5.5% 100% 00501 23.5% 15.3% 18.0% 19.6% 13.9% 9.6% 100% 00502 26.7% 17.0% 12.9% 18.1% 13.3% 11.9% 100% 00600 44.4% 20.9% 13.1% 11.2% 5.6% 4.8% 100% Overall 39.0% 17.7% 14.6% 14.7% 7.3% 6.7% 100% Source: American Community Survey 2006-2010, 5-year Estimates Table A.10: Income-to-Rent Ratio of Low Income Households in Washington D.C. PUMA 0-20% 21-40% 41-60% 61-80% 81-100% >100% Total 00101 0.0% 4.7% 3.5% 6.1% 4.1% 81.6% 100% 00102 3.8% 18.2% 19.2% 14.9% 11.7% 32.2% 100% 00103 4.6% 26.8% 15.8% 14.4% 7.1% 31.4% 100% 00104 0.6% 20.7% 19.4% 13.4% 7.5% 29.5% 100% 00105 5.6% 21.8% 9.7% 7.3% 7.3% 48.4% 100% Overall 6.3% 20.8% 15.5% 11.9% 7.8% 37.8% 100% Source: American Community Survey 2006-2010, 5-year Estimates Table A.11: Income-to-Rent Ratio of Low Income Households in Maryland. PUMA 0-20% 21-40% 41-60% 61-80% 81-100% >100% Total 00300 5.7% 23.0% 22.3% 10.3% 7.4% 31.4% 100% 01001 0.0% 18.5% 15.6% 17.6% 12.0% 36.3% 100% 01002 1.0% 12.1% 12.5% 18.8% 10.3% 45.3% 100% 01003 1.4% 18.1% 15.5% 11.8% 13.7% 39.5% 100% 01004 6.8% 12.6% 5.7% 13.7% 12.2% 49.0% 100% 01005 2.6% 19.1% 22.2% 17.2% 7.3% 31.6% 100% 01006 2.5% 9.0% 11.8% 16.6% 7.1% 53.0% 100% 01007 2.0% 14.5% 17.4% 8.8% 15.3% 42.1% 100% 01101 1.8% 8.5% 18.5% 18.4% 12.6% 40.3% 100% 01102 1.1% 12.3% 15.9% 20.7% 9.0% 40.9% 100% 01103 0.3% 21.0% 14.2% 20.3% 16.7% 27.4% 100% 01104 1.5% 15.7% 12.7% 25.0% 8.8% 36.3% 100% 01105 0.0% 18.6% 0.4% 19.5% 11.8% 49.7% 100% 01106 5.5% 26.9% 10.1% 13.8% 9.7% 34.0% 100% 01107 3.0% 12.7% 21.9% 21.3% 9.7% 31.4% 100% Overall 2.4% 15.4% 16.2% 16.8% 11.1% 38.2% 100% Source: American Community Survey 2006-2010, 5-year Estimates 102 Table A.12: Income-to-Rent Ratio of Low Income Households in Virginia. PUMA 0-20% 21-40% 41-60% 61-80% 81-100% >100% Total 00100 1.6% 16.9% 10.6% 14.3% 11.6% 45.0% 100% 00200 1.2% 15.7% 19.5% 10.7% 12.7% 40.3% 100% 00301 6.7% 17.9% 13.5% 19.1% 15.1% 27.8% 100% 00302 1.0% 16.4% 20.2% 21.9% 9.2% 31.3% 100% 00303 5.0% 9.7% 9.2% 12.7% 10.5% 52.9% 100% 00304 3.5% 10.8% 10.5% 5.2% 2.4% 67.5% 100% 00305 3.6% 19.3% 13.5% 12.3% 14.8% 36.6% 100% 00501 0.8% 10.1% 15.9% 18.9% 6.0% 48.4% 100% 00502 2.6% 15.0% 12.1% 18.8% 18.5% 33% 100% 00600 1.9% 18.0% 18.6% 15.1% 11.2% 35.3% 100% Overall 2.6% 15.8% 14.8% 15.4% 12.0% 39.3% 100% Source: American Community Survey 2006-2010, 5-year Estimates Table A.13: Labor Force Participation of Low Income Individuals in Washington D.C. PUMA Employed Unemployed Not in the Labor Force Total 00101 26.2% 3.6% 70.2% 100% 00102 32.9% 11.9% 55.2% 100% 00103 23.6% 11.2% 65.2% 100% 00104 21.6% 15.7% 62.8% 100% 00105 27.9% 9.8% 62.3% 100% Overall 25.6% 11.2% 63.3% 100% Source: American Community Survey 2006-2010, 5-year Estimates 103 Table A.14: Labor Force Participation of Low Income Individuals in Maryland. PUMA Employed Unemployed Not in the Labor Force Total 00300 33.4% 7.3% 59.3% 100% 01001 25.3% 11.0% 63.7% 100% 01002 50.0% 11.7% 38.2% 100% 01003 32.8% 6.1% 61.2% 100% 01004 28.8% 4.0% 67.3% 100% 01005 35.6% 12.8% 51.6% 100% 01006 35.0% 9.4% 55.6% 100% 01007 46.5% 11.3% 42.2% 100% 01101 39.1% 8.7% 52.2% 100% 01102 39.6% 11.0% 49.5% 100% 01103 33.5% 11.1% 55.3% 100% 01104 31.3% 18.8% 49.9% 100% 01105 23.8% 4.8% 71.4% 100% 01106 22.8% 10.6% 66.6% 100% 01107 39.0% 16.3% 44.7% 100% Overall 35.4% 10.1% 54.5% 100% Source: American Community Survey 2006-2010, 5-year Estimates Table A.15: Labor Force Participation of Low Income Individuals in Virginia. PUMA Employed Unemployed Not in the Labor Force Total 00100 43.9% 5.2% 50.9% 100% 00200 35.3% 9.3% 55.4% 100% 00301 39.4% 7.7% 52.9% 100% 00302 40.3% 10.1% 49.5% 100% 00303 38.5% 7.6% 53.9% 100% 00304 41.1% 8.1% 50.8% 100% 00305 37.6% 7.9% 54.5% 100% 00501 41.3% 8.8% 49.8% 100% 00502 54.8% 7.4% 37.9% 100% 00600 36.1% 9.7% 54.2% 100% Overall 40.9% 8.2% 51.0% 100% Source: American Community Survey 2006-2010, 5-year Estimates 104 Appendix B Policy Analysis Figures Figure B.1: Consumer Surplus for Income Levels 1, 2, and 3 for SMZ 1223 (Rhode Island Metro) and 1268 (Anacostia Metro) in Washington D.C. for Driving Alone. 105 Figure B.2: Consumer Surplus for Income Levels 1, 2, and 3 for SMZ 796 (Landover), 813 (Suitland), and 956 (Frederick) in Maryland for Drive Alone. 106 Figure B.3: Consumer Surplus for Income Levels 1, 2, and 3 for SMZ 1317 (Hunt- ington) and 1368 (Leesburg) in Virginia for Drive Alone. 107 Figure B.4: Consumer Surplus for Income Levels 1, 2, and 3 for SMZ 1223 (Rhode Island Metro) and 1268 (Anacostia Metro) in Washington D.C. for Share Ride (Car- pooling). 108 Figure B.5: Consumer Surplus for Income Levels 1, 2, and 3 for SMZ 796 (Landover), 813 (Suitland), and 956 (Frederick) in Maryland for Share Ride (Carpooling). 109 Figure B.6: Consumer Surplus for Income Levels 1, 2, and 3 for SMZ 1317 (Hunt- ington) and 1368 (Leesburg) in Virginia for Share Ride (Carpooling). 110 Figure B.7: Consumer Surplus for Income Levels 1, 2, and 3 for SMZ 1223 (Rhode Island Metro) and 1268 (Anacostia Metro) in Washington D.C. for Walk to Bus Transit. 111 Figure B.8: Consumer Surplus for Income Levels 1, 2, and 3 for SMZ 796 (Landover) and 813 (Suitland) in Maryland for Walk to Bus Transit. 112 Figure B.9: Consumer Surplus for Income Levels 1, 2, and 3 for SMZ 1317 (Hunt- ington) in Virginia for Walk to Bus. Figure B.10: Consumer Surplus for Income Levels 1, 2, and 3 for SMZ 1223 (Rhode Island Metro) in Washington D.C. for Drive to Bus Transit. 113 Figure B.11: Consumer Surplus for Income Levels 1, 2, and 3 for SMZ 796 (Lan- dover) and 956 (Frederick) in Maryland for Drive to Bus Transit. 114 Figure B.12: Consumer Surplus for Income Levels 1, 2, and 3 for SMZ 1317 (Hunt- ington) in Virginia for Drive to Bus. 115 Figure B.13: Consumer Surplus for Income Levels 1, 2, and 3 for SMZ 1223 (Rhode Island Metro) and 1268 (Anacostia Metro) in Washington D.C. for Walk to Rail Transit. 116 Figure B.14: Consumer Surplus for Income Levels 1, 2, and 3 for SMZ 796 (Lan- dover) and 813 (Suitland) in Maryland for Walk to Rail Transit. 117 Figure B.15: Consumer Surplus for Income Levels 1, 2, and 3 for SMZ 1317 (Hunt- ington) in Virginia for Walk to Rail Transit. Figure B.16: Consumer Surplus for Income Levels 1, 2, and 3 for SMZ 1223 (Rhode Island Metro) in Washington D.C. for Drive to Rail Transit. 118 Figure B.17: Consumer Surplus for Income Levels 1, 2, and 3 for SMZ 796 (Lan- dover) and 956 (Frederick) in Maryland for Drive to Rail Transit. 119 Figure B.18: Consumer Surplus for Income Levels 1, 2, and 3 for SMZ 1317 (Hunt- ington) in Virginia for Drive to Rail Transit. 120 Figure B.19: Consumer Surplus for Income Levels 1, 2, and 3 for SMZ 1223 (Rhode Island Metro) and 1268 (Anacostia Metro) in Washington D.C. for Walk to Com- muter Rail Transit. 121 Figure B.20: Consumer Surplus for Income Levels 1, 2, and 3 for SMZ 796 (Lan- dover) and 813 (Suitland) in Maryland for Walk to Commuter Rail Transit. 122 Figure B.21: Consumer Surplus for Income Levels 1, 2, and 3 for SMZ 1317 (Hunt- ington) in Virginia for Walk to Commuter Rail Transit. Figure B.22: Consumer Surplus for Income Levels 1, 2, and 3 for SMZ 1317 (Hunt- ington) in Virginia for Drive to Commuter Rail Transit. 123 Bibliography American Public Transit Association. Apta 2000 transit fact book, 2000. E. Blumberg. En-genering e ective planning: spatial mismatch, low income women, and transportation policy. Journal of American Planning Association, 70:269{ 281, 2004. E. Blumberg and M. Manville. Beyond the spatial mismatch: welfare recipients and transportation policy. Journal of Planning Literature, 19:182{205, Nov., 2004. K. Boesch. Ride for ve: low-income transportation, pilot project nal report, May, 2005. M. Criden. The stranded poor: recognizing the importance of public transportation for low income housheolds, 2008. G. de Jong, A. J. Daly, M. Pieters, and A. I. J. M. ver der Hoorn. The logsum as an evaluation measure: review of the literature and new results. Transportation Research Part A, 41:874{889, 2007. G. Giuliano. Low income, public transit, and mobility. Transportation Research Board: Journal of the Transportation Research Board, 1927:63{90, 2005. G. Giuliano, H. Hu, and K. Lee. The role of public transit in the mobility of low income households, May, 2001. H. Goldberg. State and county supported car ownership programs can help low- income families secure and keep jobs, 2001. D. C. Hodge. My fair share: equity issues in urban transportation. Geography of Urban Transportation, pages 359{375, 1995. E. Kalachek. In: Ghetto dwellers transportation and employment: Conference on transportation and poverty, 1968. J. Kasarda. Industrial restructuringand the changing locations of jobs. America in the 1990s, Volume I: Economic Trends:215{268, 1995. M. T. Lucas and C. F. Nicholson. Subsidized vehicle aquisition and earned income in the transition from welfare to work. Transportation, 30:483{501, 2003. E. Murakami and J. Young. Daily travle by persons with low income, Oct., 1997. U.S. Department of Transportation. National transit summeries and trends, 1998. P. Ong and E. Blumberg. Job access, commute and travel burden among welfare recipients. Urban Studies, 35:77{93, 1998. Richmond Area Metropolitan Planning Organization. Public transportation for the elderly, disabled, and low income: phase i - needs assessment report, Feb., 2006. 124 J. Pucher and J. L. Renne. Socioeconomics of urban travel: evidence from the 2001 nhts. Transportation Quarterly, 57:49{77, 2003. J. Pucher, T. Evans, and J. Wenger. Socioeconomics of urban travel: evidence from the 1995 npts. Transportation Quarterly, 53:15{33, 1998. T. W. Sanchez. The connection bewteen public transit and employment. Journal of American Planning Association, 65:284{296, 1999. T. W. Sanchez. The impact of public transportation on us metropolitan wage inequality. Urban Studies, 39:423{436, 2002. T. W. Sanchez. Poverty, policy, and public transportation. Transportation Research Board: Transportation Research Part A, 42:883{841, 2008. T. W. Sanchez, Q. Seng, and Z. Peng. Transit mobility, jobs access and low icnome labour participation in us metropolitan areas. Urban Studies, 47:1313{1331, 2004. Q. Shen. Location characteristics of inner-city neighborhoods and employment ac- cessibility of low-wage workers. Environment and Planning B, 25:345{365, 1998. A. Tomer, E. Kneebone, R. Puentes, and A. Berube. Missed opportunities: Transit and jobs in metropolitan america, May, 2011. K. Train. Discrete Choice Models with Simulation. Cambridge University Press, Cambridge, 2003. M. Wachs and B. D. Taylor. Can transportation strategies help meet the welfare challange? Journal of American Planning Association, 64:15{19, 1998. M. Waller. High cost or high opportunity cost? transportation and family economic success, Dec., 2005. 125