Question 1: Policy Analysis Field Examination Questions Spring 2014 Answer four of the following six questions As the economic analyst for APEC City, you need to calculate the benefits to city residents from a potential city swimming pool. To understand willingness to pay for a city pool, you gather data from 25 other cities that have pools. You collect information on the number of visits to city swimming pools each year for each city. Also reported is the fee charged per visit and each city s average income and population. With that information, you ran the following regression equation describing demand or willingness to pay for swimming in a city pool. VISITS = 140547 14638*FEE 0.001127*INCOME + 0.6031*POP You are pleased to see that this clearly resembles a demand curve because the quantity demanded (VISITS) is inversely related to price (FEE). As an analyst, you are interested in applying this information to your own city, which has average income of $31,500 and a population of 70,200. a) Rewrite the equation above using the specific information on income and population for your city so that the demand curve for visits is a function of a constant term and fees only. b) Assume your city is considering a fee of $1 per visit. Graph the demand curve you obtained in part (a) and calculate the size of consumer surplus when the fee is $1. c) Assuming the consumer surplus will remain the same for each year and the initial cost of the pool is $12 million, calculate the present value of the benefit -cost ratio under the following assumptions: the discount rate is 3%, the pool only has costs in the initial year and will last forever, the pool will require no maintenance, the benefits will exist every year in perpetuity. As a hint, remember that the present value of an amount X to be received annually for an infinite number of years discounted at rate r can be written as X/r. Show your work. d) According to your regression analysis, is the demand for a city swimming pool a normal or inferior good? Discuss. e) The regression analysis comes from 25 cities with pools that are largely located in white, middle-income neighborhoods. APEC city, which is approximately 20 percent African American, has the highest black drowning rate in the Page 1 of 11
nation. Nearly 70 percent of African Americans in APEC City claim they do not know how to swim. Some of the pool s local proponents argue that another benefit of building a pool is that it will provide greater access to learn-to-swim programs for inner-city and minority residents and thereby reduce racial disparities in drowning. Explain how or whether the computations of the net social benefits of building the pool are affected by this new information about black drowning rates. Page 2 of 11
Question 2: The financial problems facing state government are forcing a re-evaluation of the financial relationships between federal, state, and local government in the United States. Demonstrate your knowledge of the economics behind multi-level governmental finance by answering the following questions. a) Under what conditions will public sector provision of certain services increase economic efficiency in a perfectly competitive market economy? Provide examples of three different types of market failures which would require government intervention. Explain the nature of the market failure and how public sector action can overcome those failures. b) Explain how the public s preferences are reflected in the level of spending and taxation chosen in a community. Begin by assuming a single community containing residents with non-identical preferences and a single elected official who sets the level of spending and taxation. In the long run, whose preferences will the community s spending and tax collections represent? Will this lead to a Pareto optimal result? Explain why or why not. c) Another way of determining how much of a public good should be provided is by popular vote or referenda. Assume that the question posed to the electorate is whether level A or level B should be provided. Level A is the current level and is clearly less than optimal. Level B is greater than level A. How do voters determine whether to vote for level A or level B? Could voters ever vote to approve a service level that exceeds the optimal level of services? Explain why or why not. d) Expand your conceptual model to include the existence of many units of government, no barriers to mobility, and no economies or diseconomies of size in the production of the service. Does this increase economic efficiency? Will it lead to a Pareto optimal result? Explain why or why not. e) Show how a system of aid from state aid to local governments can help to overcome some market failures. What key assumptions are necessary for state aid to successfully change the allocation of local services? Explain why this will or will not guarantee that local government services will be provided at optimum levels. How is the allocation of services provided by the state (other than the aid to local government) affected? f) Some argue that growth in local government spending over time should reflect only increases in population served and prices, that is, that real, per capita local government spending should remain constant. Evaluate that statement, what are the assumptions necessary for the statement to be true? Page 3 of 11
Question 3: a) What is implicit discrimination and how does it differ from the two main economic models of discrimination (i.e. Becker s taste for discrimination and statistical discrimination)? (20%) b) What is spontaneous discrimination and how does it differ from the two main economic models of discrimination (i.e. Becker s taste for discrimination and statistical discrimination)? (20%) c) An advocacy group, the Minority Business Enterprise Legal Defense and Education Fund (MBELDEF), believes that Becker-type discrimination explains why disadvantaged business enterprises (DBEs) are more likely than non-dbes to be denied contract awards. MBELDEF claims that its analysis of data on prequalification for bidding on construction projects funded by the Washington Suburban Sanitation Commission (WSSC) demonstrates that the agency responsible for prequalifying firms discriminates (in the Becker sense) against DBEs. The discrimination against DBEs in the prequalification process, according to MBELDEF, then produces a form of selection bias, explaining why it often the case that there are statistically insignificant differences in contract award rates. Table 1 reports the results concerning prequalification. The dependent variable is the probability that the firm is pre-qualified to bid on WSSC construction contracts. The independent variables are: whether the firm is a DBE; whether the firm is an in-state firm; and whether the firm s main business is Building Construction, Heavy Construction vs. all other types. Explain whether or not these results support the MBELDEF claim that a) there is Becker-type discrimination and b) that this discrimination produces a form of selection bias. (30%) Table 1: Logistic Regression Analysis of Pre-Qualification in Construction * Coefficient estimates are significant at 95% significance level. Source: DOT Pre-Qualification of Construction File, 1999 Page 4 of 11 Coefficient Estimate Chi-square p-value Mean Intercept -0.6736* 6.669 0.0098 1.0000 DBE Status 0.6493* 6.280 0.0122 0.1077 Instate 0.9889* 16.546 0.0001 0.8098 Building Construction -0.3717 2.843 0.0918 0.3706 Heavy Construction -1.5555* 46.119 0.0001 0.4448 Mean of Dependent 0.3678 Max-rescaled R-Square 0.1661 p-value 0.0001 Number of Observations 715
d) Write out an alternative model specification and estimation scheme designed to test directly for whether selection bias in prequalification contributes to any observed disparity in contract award rates between DBEs and non-dbes. (30%) Page 5 of 11
Question 4: a) A corporation that is subject to the United States corporate income tax purchases a piece of equipment. Show how you would measure the present value of the tax savings the purchase generates for the corporation. Show the present value per dollar of the purchase price. Define the variables in your formula. Assume there is no investment tax credit. You may use either discrete or continuous time. Assume the corporation pays all profits as dividends to individual shareholders. b) Define (in words) the user cost of capital from the Hall Jorgenson model. According to Hall and Jorgenson, how does the user cost of capital affect a corporation s investment decisions? c) State and explain the formula for the user cost of capital for a U.S. corporation. Again, define the variables in your formula, assume there is no investment tax credit, use either discrete or continuous time, and assume the corporation pays all profits as dividends to individual shareholders. How does the formula change if there is an investment tax credit? d) The 2003 U.S. tax bill included a provision that allowed businesses to expense (deduct immediately) 50% of the purchase cost of some capital assets. The remainder of the cost must be depreciated. This provision is called bonus depreciation. Assuming the equipment purchased in (a) qualifies for this treatment, explain, using the formula in (c), how the cost of capital is affected by the tax law change. What would the Hall-Jorgenson model predict about the effect of bonus depreciation on corporations investment in equipment? Explain your answer. e) Recently, chairman of the U.S. House Ways and Means Committee, Dave Camp, proposed eliminating all accelerated depreciation including bonus depreciation to help finance a cut in the corporate tax rate from 35 percent to 25 percent. Camp proposed replacing accelerated depreciation with depreciation rules that would match more closely the true economic useful life of assets. 1 Evaluate Camp s proposal. Would it make the corporate tax system more or less efficient? 1 http://taxvox.taxpolicycenter.org/2014/05/29/dave-camps-great-bonus-depreciation-flipflop/#sthash.lzefy7c6.dpuf Page 6 of 11
Question 5: A social service agency offers job training services to a group of local disadvantaged workers. These workers typically have low educational attainment and little formal job experience. a) Assuming some information is collected before and after the intervention on participants only, write out a simple linear regression model that you might estimate to determine whether participation in the job training program has an effect on earnings or hours worked. Describe the outcome and treatment variables along with any other variables you might include in your model. b) What assumption or assumptions are needed to identify the causal effect of the job training program in your simple model described in (1)? Discuss. c) Now assume you have information on outcomes and other relevant variables for nonparticipants and participants both before and after the intervention is offered. Rewrite your regression equation to allow for the existence of treated and nontreated individuals. How does having information during the same time period for a control group of nonparticipants help improve the estimation method you described in (1)? You can describe the advantages of having information on these nonparticipants in words rather than equations. d) What estimation problem is solved by using random assignment to determine which of the disadvantaged workers will be offered job training? e) Unlike ordinary regression estimation methods, propensity score methods require a common support region. What is meant by common support? Provide some hypothetical examples of why or how the lack of common support could be a problem given the participants and nonparticipants in this example. Page 7 of 11
Question 6: In Kerwin Charles and Jonathan Guryan s Prejudice and Wages: An Assessment of Becker s Economics of Discrimination, Journal of Political Economy, 2008, the authors claim to test for the predictions of the Becker Model of Discrimination by constructing an index of prejudice using the General Social Survey. They argue that their aggregate index of prejudice closely mirrors the results of specific responses that approximate racial prejudice, e.g., unwillingness to vote for a black for president; support for a law against interracial marriage; and belief that whites have a right to segregate communities. Table 1 below reports the OLS estimates of coefficients on age, education and gender in different measures of prejudice. a) State wherein you agree or disagree with the Charles-Guryan claim that their aggregate prejudice measure mirrors other measures. (10%) b) The authors then construct from the individual-level measures of prejudice, state measures. They compute for each state the 10 th percentile, the median, and the 90 th percentile of the prejudice measures for 45 states. They estimate the residual black-white wage gap in each state in the following manner: They obtain OLS estimates of coefficients in a log wage equation, controlling for education, a quadratic in potential experience, race-specific year dummies, and a black dummy variable for each state. The estimated effects on each of these black dummy variables become the dependent variable in a second equation, where the independent variables are the average, median, 10th, and 90th percentiles of the white prejudice distribution in the state; the fraction of the workforce that is black; and an approximation to the marginal level of prejudice given by the pth percentile of the prejudice distribution, where p is the percentage of the state workforce that is black. These last results are reported in Table 3 below. Define what is meant by the marginal discriminator (10%) c) Explain how or why the results in Table 3 provide evidence of Becker-type discrimination. (15%) d) Charles and Guryan claim that the Becker Model predicts larger black-white wage gaps in states with larger shares of black workers. Explain (15%) e) Provide an interpretation of the coefficients on the fraction black variable. Do the results of the estimated coefficients on the fraction black variable in Table 3 confirm or refute the claim in #4? Explain (15%) Page 8 of 11
f) Provide an alternative explanation for the estimated coefficients on the fraction black variable. (15%) g) Discuss an alternative model specification and estimation to test for statistical discrimination. Compare and contrast your model with the Altonji and Pierret analysis in Employer Learning and Statistical Discrimination, Quarterly Journal of Economics, 2001. (20%) Page 9 of 11
Page 10 of 11 Student ID Number:
Page 11 of 11 Student ID Number: