Transportation Investments in Response to Economic Downturns (2014)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

2 Fiscal Policy in Recessions and the Role of Transportation Spending This chapter describes the economic rationale for undertaking a stimu- lus program in response to a recession and the factors that should be taken into consideration in deciding the role of transportation spend- ing in such a program. The first section below reviews research on the effectiveness of stimulus spending to aid recovery from recessions. The American Recovery and Reinvestment Act of 2009 (ARRA) was the latest such federal program and the most ambitious since the Depression. The second section examines the suitability of federally supported spending for public transportation as stimulus. FISCAL POLICY IN RESPONSE TO RECESSIONS Fiscal policy is the collection of decisions of a government about its bud- get and taxes. The term usually refers to decision making that takes into consideration the macroeconomic impacts of government spending and taxes. A fiscal stimulus program, as defined in Chapter 1, is a package of extraordinary federal government expenditures or tax concessions funded by borrowing and undertaken to reduce unemployment and speed recov- ery from a recession. Stimulus spending historically has taken various forms, including purchases by the federal government directly, grants to state and local governments for specific purposes (e.g., for transportation projects), grants or credit assistance to businesses or individuals for spe- cific purposes (e.g., to subsidize energy efficiency), and unrestricted pay- ments to individuals. Tax changes may be temporary deductions, credits, or rate reductions. Chapter 1 explained that the study charge did not call on the commit- tee to assess the overall usefulness of stimulus spending or of ARRA as 33 

34 Transportation Investments in Response to Economic Downturns a whole. The committee examined these questions because conclusions about the value of transportation stimulus spending (e.g., the transpor- tation programs funded by ARRA) depend on assumptions about the general effectiveness of fiscal policy. Two examples of this dependence are the following: • Infrastructure investments, provided that projects are well chosen, will yield a long-term economic benefit from the services produced by the facilities constructed, regardless of the magnitude of the short- term stimulus impact of the government spending. Forms of stim- ulus spending that do not directly support productive investment do not provide this long-term benefit. Therefore, in the design of a stimulus program, greater uncertainty about the short-term stimu- lus benefit of government spending may justify allocating a larger share of the total program funds to infrastructure investment, which will at least yield the long-term benefits. Other categories of stimulus spending also may yield benefits that are distinct from their stimu- lus effect; for example, extension of unemployment compensation alleviates the distress of unemployed persons, a benefit regardless of the magnitude of the stimulus the spending may provide. Both kinds of benefit (stimulus effect and benefits beyond stimulus) should be taken into account for all spending categories in designing the stimu- lus package. • In the ARRA transportation programs, some established practices with regard to project selection, spending priorities, and funding sources were superseded in an effort to boost the immediate employ- ment impact of the federal funding. Consequently, as Chapter 3 describes, some projects with greater long-term transportation ben- efits probably were passed over in favor of projects that were expected to yield greater short-term stimulus (because they could employ workers quickly or were located in distressed areas). If policy makers are confident that the spending will provide a substantial short-term stimulus benefit, they should be willing to give up some long-term transportation benefit to attain the stimulus benefit. However, if the short-term benefit is uncertain or expected to be minor, the case for altering priorities is weakened. 

Fiscal Policy in Recessions and the Role of Transportation Spending 35 The first subsection below reviews estimates from economic research of the success of stimulus spending as an aid to recovery from recessions, and the second reviews estimates of the effects of ARRA. The third sub- section describes particular features of a stimulus program that research indicates will affect its outcome. The final subsection is a summary. Assessments of the Effects of Stimulus Spending Assessments of ARRA and other fiscal policy actions have measured impact in terms of changes in employment and changes in gross domes- tic product (GDP) compared with the levels that would have prevailed in the absence of the government action. In effect, GDP and employment in the absence of the intervention are estimated by means of a model that simulates the path of the economy as a function of government policy and external factors. The impact measure reported may refer to points in time (e.g., the change in employment or GDP in a particular quarter or for each quarter over a period of years) or a cumulative effect (e.g., total person months of employment attributable to the action or the present value of future changes in GDP over a specified period attributable to the action). An impact reported as an increase in jobs or GDP without reference to the duration of the effect does not provide a basis for evaluation. The impact often is expressed as a multiplier, for example, the change in GDP per dollar of added government spending during a specified period. Most public debate over stimulus spending places the greatest emphasis on job creation, especially on immediate job creation during the recession. Enactment of ARRA aroused political and scholarly debate over the effectiveness of fiscal policy. Reviews and syntheses of the research include Hall 2009; Hall 2010; Auerbach et al. 2010; Boskin 2011; Ramey 2011; and Congressional Budget Office (CBO) 2012. Nearly all published research at the time of these reviews was based on data from before the 2007–2009 recession. Conclusions of the reviews concerning the magnitude of mul- tipliers during recessions were as follows: • Hall (2009) and Hall (2010) review the alternative approaches used to estimate multipliers and their results. The conclusion is that for 

36 Transportation Investments in Response to Economic Downturns government purchases, in times of extreme recession such as in 2007–2009 and when monetary policy is holding interest rates near zero, “a value for the . . . multiplier of around two is representative of recent research” (Hall 2010, 91). A multiplier of 0.8 for government benefit payments to individuals applies under the same conditions (Hall 2010, 93). These estimates describe immediate effects of a stim- ulus program (i.e., within the first year); negative effects (reductions in output) are possible in later years (Hall 2010, 86). However, “the range of estimates for the zero-interest-rate multiplier is wide, even within the same study, because it is sensitive to the timing of the stimulus and the duration of the period when the interest rate will remain pinned at zero” (Hall 2010, 91). • Auerbach et al. (2010) review the history of use of fiscal policy to combat recessions in the United States through ARRA, methods that have been used to evaluate the effects of fiscal policy, and results of such assessments. They emphasize the uncertainty of the estimates. Estimates cited for multipliers range from 0.4 to 4 (Auerbach et al. 2010, 148–151), and for the increase in GDP as a result of ARRA, in the quarter of peak impact, from 0.5 to 4 percent (Auerbach et al. 2010, 156–157), implying a similarly wide range in multipliers. No assess- ment of the most likely range of multiplier values is proposed. • Boskin (2011) compares estimates of the effects of fiscal stimulus that were derived by using diverse estimation methods. The review takes care to include results supporting a skeptical view of fiscal pol- icy effectiveness. It interprets the estimates as follows: at full employ- ment the short-run government purchases multiplier is zero. With high unemployment but with monetary policy reacting to changes in inflation and GDP by adjusting the short-term interest rate, empiri- cal estimates of the multiplier are mostly between 0.5 and 0.7 (Boskin 2011, 10). When monetary policy is holding the interest rate to zero (as the Federal Reserve may choose to do during a recession to provide a monetary stimulus), some models predict short-term government purchases multipliers above 1 (Boskin 2011, 13). Debt-financed gov- ernment spending has long-term costs in the form of depressed future output, particularly when the ratio of government debt to 

Fiscal Policy in Recessions and the Role of Transportation Spending 37 GDP is high, which most studies of short-term stimulus impact ignore (Boskin 2011, 21). • Ramey (2011) summarizes a representative group of 12 studies that estimated multipliers for government purchases for the U.S. national economy and 10 studies that used cross-sectional state-level data to estimate the effect of federal stimulus spending within a state on state income or employment. The estimates in the studies are mostly of short-term effects (i.e., within 2 years of the spending). The review notes several studies that found that the size of the multiplier depends on whether the economy is in recession or expansion. The author concludes that “the U.S. aggregate multiplier for a temporary, deficit- financed increase in government purchases . . . is probably between 0.8 and 1.5” (Ramey 2011, 673). Moreover, “if the increase is under- taken during a severe recession, the estimates are likely to be at the upper bound of this range. It should be understood, however, that there is significant uncertainty involved in these estimates. Reason- able people could argue that the multiplier is 0.5 or 2.0 without being contradicted by the data” (Ramey 2011, 681). • For a series of quarterly reports on the effects of ARRA, CBO (2012) reviewed the economic research literature to select a range of mul- tipliers specific to each of the forms of stimulus provided by the act. For example, the range of multiplier estimates is 0.4 to 2.2 for fed- eral transfers to state and local governments for infrastructure. The CBO multipliers are described in the section below on the impact of ARRA. CBO explains that “the multipliers indicate the cumulative impact of [ARRA funded actions] . . . on GDP over several quarters” (CBO 2012, 6). In summary, these reviews are consistent in concluding that there is evidence that the short-term government spending multiplier is greater than zero, despite varying points of view on modeling methods and on the overall value of stimulus spending or of ARRA. In addition, they cite support for the conclusion that the multiplier depends on eco- nomic circumstances. In particular, most conclude that the multiplier is greater than one when the economy is in a serious recession (although few studies estimate the effect of the state of the economy on the size of 

38 Transportation Investments in Response to Economic Downturns the multiplier)1 and when monetary policy is holding interest rates near zero, and that the multiplier is smaller when the economy is expanding. The reviews cite research findings contrary to these generalizations and emphasize the uncertainty of the multiplier estimates and the conse- quent difficulty in providing policy advice to a government planning a stimulus program. In models that predict a small response to stimulus spending, an increase in government purchases crowds out a substantial amount of consump- tion and investment that would have occurred without the intervention, because government spending bids up interest rates, wages, or prices or because expectations of future higher taxes or slower growth reduce consumption and investment. An estimated government purchases multiplier that is positive and less than one means that GDP increases in response to an increase in government purchases, but the ultimate increase is less than the value of the initial government purchases. In times of full employment, crowding out is important because nearly all resources are already employed; therefore, a multiplier much less than one would be expected. Conversely, measurement of a multiplier greater than one is expected to be more likely during a recession. The stimulus impact of federal spending also will be diminished if the spending is in the form of grants to state and local governments and recipients increase their spending by less than the amount of the grants (substituting the federal funds for state and local funds). Uncertainty in estimates arises from the dependence of multipliers on economic circumstances and on the characteristics of the policy responses, from data limitations, and from uncertainty about the validity 1 Most of the reviews cite a working paper preceding Auerbach and Gorodnichenko (2012), which estimates a model that allows the size of multipliers for categories of government spending in recessions to differ from their values in expansions and finds larger multipliers in recessions, especially when effects over several years are considered. Ramey (2011) also cites two aggregate analyses (one for the United States and one for the Euro area) that measure the dependence of the multiplier on the state of the economy and find little effect (Ramey 2011, 679) but notes that three of the studies using data across the U.S. states find that the multiplier is higher during recessions (Ramey 2011, 683). A more recent estimate of the dependence of multipliers on the state of the economy in seven countries concludes that “multipliers, and especially spending multipliers, are significantly larger in downturns than in expansions. Spending multipliers in the United States are found to be significantly above one during downturns” (Baum at al. 2012). 

Fiscal Policy in Recessions and the Role of Transportation Spending 39 of economic models and assumptions. Data providing direct observa- tion of the effect of stimulus spending in response to a recession are scarce because recessions and stimulus programs both are infrequent events (Parker 2011, 711). The structure and assumptions of the model selected can strongly influence the magnitude of multiplier estimates (Leeper et al. 2011, 20; Auerbach et al. 2010, 157). The structure of some models does not allow the multiplier to depend on the state of the econ- omy; such models are of limited use for evaluation of stimulus spending as an aid to recovery from a recession (Parker 2011, 716). Most of the studies cited in the reviews summarized above estimate short-term effects of stimulus spending (although some studies report present-value multipliers representing the cumulative effect of the stim- ulus over a period of years or quarterly multipliers for a period of years after the application of the stimulus). One of the reviews (Boskin 2011) emphasizes the concern that increasing government debt through stim- ulus spending may depress growth in the long term. Debt may suppress growth because higher debt raises interest rates and depresses saving and investment or because it leads to higher taxes. CBO has estimated the effects of alternative federal budgetary paths on the economy for the decade from 2014 to 2023 on the basis of macro‑ economic model simulations. The analysis concluded that a $2 trillion increase in cumulative primary deficits (i.e., the increase in deficits before counting the resulting increase in interest costs) over the decade, compared with a baseline representing deficits under current law, would increase gross national product (GNP) in the short term (2014 to 2016) but decrease GNP in all later years in the decade, compared with pro- jected GNP with baseline government spending. The projected peak increase is 0.4 percent in 2015 and the decrease reaches 0.9 percent in 2023 (CBO 2013b, 1–7). Impact of ARRA ARRA required CBO to issue quarterly comments on the reports of grant recipients on numbers of jobs supported by ARRA funds (CBO 2012; CBO 2013a). CBO’s reports explain that the recipients’ jobs data do not provide a measure of the full jobs impact and therefore include estimates based on econometric models of the act’s impact on employment and 

40 Transportation Investments in Response to Economic Downturns GDP. CBO estimated these impacts by grouping ARRA provisions into nine categories (federal purchases, payments to state and local govern- ments for infrastructure, payments to state and local governments for other purposes, transfer payments to individuals, one-time payments to retirees, and four categories of tax provisions) and selecting high and low estimates of the multiplier for each category on the basis of model outputs and data. The range of the high and low multiplier values illustrates the uncer- tainty that CBO believes to exist in such estimates. For example, for federal purchases, CBO’s high and low estimates are 2.5 and 0.5; for pay- ments to state and local governments for infrastructure, 2.2 and 0.4; for payments to states and local governments for other purposes, 1.8 and 0.4; and for transfer payments to individuals, 2.1 and 0.4 (Table 2-1). In its later assessments of ARRA, CBO broadened the ranges of uncertain- ties of its multiplier estimates (CBO 2012, 6–7) compared with its ear- lier assessments (CBO 2011, 6–7), allowing for a greater likelihood that multipliers may be less than 1.0. The multiplier estimates incorporate CBO’s assumptions about the fractions of federal grants to individuals TABLE 2-1  CBO Estimates of Output Multipliers for Provisions of ARRA Estimated Output Multipliers Type of Activity Under ARRA Low High Purchases of goods and services by the federal government 0.5 2.5 Transfer payments to state and local governments for infrastructure 0.4 2.2 Transfer payments to state and local governments for other purposes 0.4 1.8 Transfer payments to individuals 0.4 2.1 One-time payments to retirees 0.2 1.0 Two-year tax cuts for lower- and middle-income people 0.3 1.5 One-year tax cut for higher-income people 0.1 0.6 Extension of first-time home buyer credit 0.2 0.8 Corporate tax provisions primarily affecting cash flow 0 0.4 Note: The output multiplier is the increase in dollars of GDP, cumulatively over several quarters, per dollar of federal spending (or tax reduction) under each type of ARRA provision. The estimates assume that the Federal Reserve is holding short-term interest rates low. Source: CBO 2012, 6–7. 

Fiscal Policy in Recessions and the Role of Transportation Spending 41 and to state and local governments that are immediately spent by their recipients. (For example, state and local governments may respond to receipt of grants by reducing spending of their own funds.) CBO’s estimate of the peak GDP impact of ARRA is an increase of between 0.8 and 4.6 percent in the 2010 second quarter. CBO derives employment impact from the estimated GDP impact. The percentage increase in employment attributed to ARRA is somewhat smaller than the percentage increase in GDP, with employment increases lagging GDP increases. The estimated peak impact on the number of people employed was between 1.4 million and 3.6 million in the 2010 third quarter. The estimated increase in employment measured in full-time-equivalents is greater (1.0 million to 5.2 million in the 2010 third quarter), presum- ably because some already employed persons worked more hours (CBO 2011, 3). Research using other models to simulate the effects of the ARRA stim- ulus has produced different estimates. Model estimates other than those of CBO for the increase in GDP as a result of ARRA cited by Auerbach et al. (2010, 157), in the quarter of peak impact or for the year 2010, are in the range of 0.5 to 3.4 percent. Estimates of private-sector fore- casters and investment banks tended to fall in this range, toward the higher end.2 A variety of academic models have generated differing results depend- ing on assumptions made. A simulation by Drautzburg and Uhlig (2011) generates a lower multiplier of roughly 0.5 in the short run (which can turn negative in the long run because of future taxes) (Drautzburg and Uhlig 2011, 13–34).3 The estimated short-term multiplier for the gov- ernment investment component of ARRA is smaller than for the govern- ment consumption component but larger in the long term (Drautzburg and Uhlig 2011, 13, 29). The estimates are sensitive to assumptions about 2 See CEA (2011, 12–13) for examples of private-sector estimates. 3 The sources for the CBO multipliers in Table 2-1 include traditional large-scale empirical macro­ economic models, which incorporate assumptions that tend to lead to large multipliers in reces- sions (Reichling and Whalen 2012, 8–17). The model of Drautzburg and Uhlig (2011) is from a category of models (dynamic general equilibrium models) that take a different theoretical approach and that have tended to produce a wider range of multiplier estimates, depending on model assumptions (Auerbach et al. 2010, 150–151). 

42 Transportation Investments in Response to Economic Downturns expectations with regard to how long the Federal Reserve would hold interest rates at zero, as well as to other assumptions. An alternative sim- ulation by Davig and Leeper (2011, 226) finds much higher multipliers for spending under ARRA. It estimates government spending multipliers of 3 if monetary policy is passive in response. Christiano et al. (2011, 93) similarly conclude that government spending multipliers can be greater than 3 under some assumptions if monetary policy is passive in response to an increase in government spending. The CBO (2012) and Drautzburg and Uhlig (2011) assessments are not derived from observation of the effects of ARRA but rather by apply- ing models whose parameters are estimated on the basis of earlier his- torical data to estimate the effect of the ARRA spending. Research that attempts to measure the actual effects of ARRA is summarized as follows in the paper on stimulus spending commissioned by the committee: Wilson (2012), Feyrer and Sacerdote (2011), and Conley and Dupor (2012) each looked at the employment effects of the stimulus spending from the 2009 American Recovery and Reinvestment Act (ARRA) using state-level data on employment and stimulus received. Though their data and regression specifications differed, each attempted to isolate the causal effect of the stimulus on subsequent employment change using exogenous drivers, or “instruments,” that help determine how much stimulus a given state receives while being uncorrelated with current economic conditions in the state. . . . Both Wilson and Feyrer and Sacerdote find relatively large total and private-sector employment effects of ARRA spending, while Conley and Dupor find positive and significant total employment effects but are unable to reject a zero effect for the private sector (though their confidence intervals are fairly large). (Leduc and Wilson 2012c, 17–18) The last-cited paper (Conley and Dupor 2012), which found a weak employment impact of ARRA, observes that the finding of an employ- ment increase in the public sector but no statistically significant impact on private-sector employment may be a crowding-out effect: because government workers are relatively highly educated on average and the labor market for such workers was relatively less affected by the reces- sion, government hiring or retention of these workers as a result of ARRA grants preempted some hires in the private sector (Conley and Dupor 2012, 22). 

Fiscal Policy in Recessions and the Role of Transportation Spending 43 An additional study directly measuring the effects of ARRA (Chodorow- Reich et al. 2012) evaluated a provision to distribute $88 billion to the states by increasing the rate of federal reimbursement for state Medicaid expenditures, freeing state funds for other purposes. Although this aid was not for infrastructure purposes, the evaluation provides an insight into the effect of federal ARRA grants to states on total state government spending and saving. The study concluded that each $100,000 in addi- tional ARRA Medicaid aid received by a state increased state employ- ment by 3.8 person years through June 2010, with most of the increase occurring outside the government and health sectors (Chodorow-Reich et al. 2012, 137). The authors observe that “for this result to be economi- cally plausible, states must have used the funds to avoid spending cuts or tax increases. Hence we also provide evidence that the transfers do not appear to have increased the states’ end of year balances” (Chodorow- Reich et al. 2012, 121). As the authors of the cross-state studies acknowledge, measurements of the impact of federal stimulus spending within a state on income and employment within the state will miss out-of-state impacts of the spend- ing (for example, when a construction contractor buys materials from a factory in another state). The variation in the in-state effects of stimulus spending occurs against a background of common national effects of federal government actions influencing the economy. While the United States was implementing fiscal stimulus through ARRA and other measures, fiscal policy in some European countries placed more emphasis on measures to restrain the growth of national debt (measures referred to as fiscal consolidation). Experiences of European countries have been a source of direct observations of the effects of fiscal policy actions under the conditions of financial crisis and recession that have occurred globally since 2008. A study comparing forecast with actual growth in 2010 and 2011 in 26 European economies that undertook fis- cal consolidation found that the greater the magnitude of planned fiscal consolidation, the greater the forecast error in GDP growth, indicating that actual fiscal multipliers were consistently larger than forecasters were assuming (Blanchard and Leigh 2013, 4). Reviewing forecast assumptions and recent research, the authors conclude that actual multipliers were around 0.5 at the start of the crisis and “substantially above 1” in the early stage of the crisis (Blanchard and Leigh 2013, 19). 

44 Transportation Investments in Response to Economic Downturns Features of a Stimulus Program That Influence Its Impact Critics of ARRA questioned the act’s effectiveness on two grounds. The first criticism emphasized the low-end estimates of multipliers and argued that the short-term gains from stimulus spending are too small to justify likely long-term costs of adding to the federal debt (Boskin 2011, 19). The second criticism was that, while stimulus spending should be effective in principle, ARRA was poorly designed—the spending package was too small or was not sufficiently timely or well targeted, reducing its jobs and GDP impact (Feldstein 2011; Boskin 2011, 20; Auerbach et al. 2010, 156; Hall 2010, 84–86). Taylor (2011) argued that federal grants to state and local governments, a major component of ARRA, are ineffective as stimulus because recipients will not promptly spend the added funds. On the basis of a simulation model estimated with historical data from the National Income and Product Accounts for the first quarter of 2009 through the first quarter of 2011, the study projected that ARRA grants to state and local governments would reduce state and local govern- ment net borrowing but leave state and local government spending at the level that would have occurred during the recession in the absence of the grants (Taylor 2011, 694–695). Chapter 3 examines the trend in total state and local transportation spending during ARRA. The criticisms of stimulus package design are related to the com- mittee’s task. The major design features are (a) the size of the stimulus; (b) its timing; (c) the distribution of the total spending among fed- eral purchases, grants to states, transfers to individuals, and tax reduc- tions, as well as allocations within each of these four categories and provisions for targeting spending to particular regions or industries; and (d) grant rules and administration (including matching require- ments and maintenance-of-effort requirements for grants to state and local governments and allocation formulas and other rules for selecting recipients). Chapter 3 examines grant design questions in the case of the ARRA transportation grants. Research findings relevant to the commit- tee’s task with regard to the distribution of stimulus among categories of spending and the timing of spending are summarized below. Relative Stimulus Effects of Categories of Government Spending Certain kinds of government purchases appear to have larger multipliers than others during recessions. An understanding of these differences 

Fiscal Policy in Recessions and the Role of Transportation Spending 45 would be valuable in designing a stimulus package; however, estimates of the relative magnitude of sector-specific multipliers are uncertain and the magnitudes probably depend on the circumstances of the recession. Estimates of multipliers, or of relative stimulus effectiveness, for catego- ries of government spending include the following: • The multiplier estimates (CBO 2012, described above and shown in Table 2-1) that CBO used for its assessments of ARRA indicate mostly small differences among the multipliers for direct federal purchases, payments to state and local governments for infrastructure and for other purposes, and most categories of payments to individuals. CBO ranked the multipliers for the ARRA tax cuts lower than those for the spending provisions. • Earlier CBO reports [using the same methodology and multipliers similar to those of CBO (2012)] defined and ranked fiscal policy options for increasing employment within 1 or 2 years as a response to the recession (CBO 2008; CBO 2010; Elmendorf 2010). In terms of person years of employment per million dollars of federal budgetary cost, infrastructure investment ranks high among the options in these estimates, although not at the top (Table 2-2). CBO estimated the effect of additional federal funding of infrastructure to be smaller than the effect of employer payroll tax deductions and aid to the unemployed and slightly larger than the effect of federal aid to states for purposes other than infrastructure. • A study of multipliers for components of government expenditures in recessions and during expansions (Auerbach and Gorodnichenko 2012, Auerbach and Gorodnichenko 2013) estimated that the GDP multiplier is 3.7 for defense spending and 1.3 for nondefense spend- ing during recessions and that the multiplier for government con- sumption purchases during recessions is 1.4 but the multiplier for government investment (i.e., nonconsumption) purchases is 4.3. Multipliers during expansions were estimated to be smaller (Auer- bach and Gorodnichenko 2013, 321). The research does not reveal the particular features of defense spending and investment expenditures that account for their relatively high estimated multipliers. • Feyrer and Sacerdote (2011), one of the studies cited above that used cross-state data to measure responses to ARRA, estimated that fed- eral grants for infrastructure and income support for low-income 

46 Transportation Investments in Response to Economic Downturns TABLE 2-2  CBO Estimates of Effects of Fiscal Policy Options on Employment Cumulative Effects on Employment, 2010–2015 (years of full-time-equivalent per million dollars budgetary cost) Policy Option Low High Increasing aid to the unemployed 6 15 Reducing employers’ payroll taxes 4 11 Reducing employers’ payroll taxes for firms that 7 16 increase their payroll Reducing employees’ payroll taxes 2 7 Providing an additional one-time Social Security 2 8 payment Allowing full or partial expensing of investment 1 8 costs Investing in infrastructure 4 10 Providing aid to states for purposes other than 3 9 infrastructure Providing additional refundable tax credits for 3 7 lower- and middle-income households in 2011 Extending higher exemption amounts for the 1 4 Alternative Minimum Tax Reducing income taxes in 2011 1 4 Note: See source for definitions of policies. Source: CBO 2010. households in ARRA were highly stimulative but that grants to states for education were not (Feyrer and Sacerdote 2011, 16). The authors argue that most federal education aid supported jobs that would have been funded from other sources in the absence of the federal grants, whereas the infrastructure projects mostly would not have been built without the ARRA grants (Feyrer and Sacerdote 2011, 14, 20). • A study of the multiplier for federal highway grants to states (Leduc and Wilson 2012b; Leduc and Wilson 2012a), using state-level data for 1990 to 2010, estimated that the short-term multiplier for state- level income for an addition to federal highway grants was between 1.5 and 3 on average but that this multiplier was much larger during 

Fiscal Policy in Recessions and the Role of Transportation Spending 47 the 2007–2009 recession (Leduc and Wilson 2012a, 3–4). The study does not compare highway spending with other forms of stimulus spending, but the results are evidence that highway spending may produce substantial stimulus. The particular circumstances of past recessions may have influenced the relative stimulus effectiveness of alternative government purchases. In a recession in which the largest initial job losses are in manufactur- ing, purchases of manufactured goods like defense material might be a relatively highly effective stimulus, while in a recession like that of 2007–2009, with particularly severe job losses in construction, govern- ment infrastructure purchases might be effective, because such targeted spending would be more likely to employ directly resources that would otherwise be idle and less likely to crowd out alternative uses of the resources. However, none of the research reviewed evaluated whether such targeted spending could improve stimulus effectiveness, and the impact of the 2007–2009 recession became broadly spread throughout the economy. Proponents of transportation proposals often present economic impact estimates that use an input–output model to project in detail the employment and income that the construction will generate. Similar estimates have been presented to support proposals for spending priori- ties in a stimulus program and to evaluate the components of ARRA. Impact estimates that are derived according to the standard method of such studies are of limited value in judging the effectiveness of stimu- lus spending or for comparing transportation spending with alternative forms of stimulus spending. Box 2-1 gives examples of these studies and describes their limitations. Importance of Timeliness As Chapter 1 described, the failure of past stimulus programs to spend the funds provided when the economy was still declining or in a trough has been one of the main grounds for skepticism about the effectiveness of fis- cal stimulus. Timeliness has been a particular concern for infrastructure spending as stimulus because planning and commencing a construction project typically require months, and carrying it out requires months to years. Chapter 1 also noted that, in past fiscal stimulus programs, the time 

48 Transportation Investments in Response to Economic Downturns BOX 2-1 Estimates of Stimulus Spending Effects from Economic Impact Studies Economists’ estimates of multipliers, which are based on present- day macroeconomic models and econometric methods, vary over a wide range, and the researchers acknowledge a high degree of uncertainty. The estimates do not attempt to make fine distinctions between the effects of spending in one industry versus another. Yet economic impact evaluations of infrastructure projects and programs frequently are presented that offer pre- cise estimates of jobs impacts and compare impacts of narrowly defined alternative expenditures. The following are examples: • The Maryland state government estimated direct employment and total in-state quarterly jobs created by state spending of ARRA funds in each of 13 programs during the period from 2009 to 2012 (State of Maryland n.d.). In the estimates, trans- portation spending created 1.9 total person quarters of employ- ment per direct person quarter of employment on funded projects and 1 person quarter of total employment per $78,000 in spending, while spending in all programs created 2.6 jobs per direct job at $56,000 per person quarter of total employ- ment (excluding one quarter with anomalous estimates). • The Washington State Department of Transportation esti- mated total (direct, indirect, and induced) person years of employment for each ARRA-funded highway project, with employment impact depending on the nature of the project. It reported about 1 person year per $100,000 (Washington State Department of Transportation 2010). Examples of assessments arguing the superiority of specific catego- ries of stimulus spending include the following: • Antonopoulos et al. (2010) compared the effects of hypo- thetical federal stimulus programs dispensing equal sums for 

Fiscal Policy in Recessions and the Role of Transportation Spending 49 BOX 2-1 (continued) Estimates of Stimulus Spending Effects from Economic Impact Studies grants alternatively for social services (early-childhood ser- vices and home-based healthcare) or for infrastructure. They concluded that social service grants generate 24 total jobs per million dollars of federal grants versus 11 jobs per million dollars for infrastructure grants. • The Surface Transportation Policy Project (2004) compared employment effects of alternative categories of transpor- tation spending. The analysis indicated that $1 million of spending on federal-aid supported projects for constructing new highway capacity would generate 35 jobs; for highway maintenance, 38 jobs; and for public transit construction and operations, 41 jobs. The study concluded that “investing in a balanced transportation system rather than a new roads-only system will help get more Americans get back to work.” Each study derived its employment estimates from an input– output model. This model is a system of linear equations that specify the quantities of inputs required from each of the sectors making up an economy (industries and the household sector) to produce one unit of output from each sector. The model can be used to trace the increases in the outputs of all industries in an economy required to produce one unit of additional final product in an industry (UN 1999, 3–6). For example, producing an additional ton of asphalt paving mix requires increased out- put of asphalt from the refining industry and increased truck- ing services, which in turn requires a further increase in refining industry output, and so forth. Input–output model predictions show a multiplier effect, because an initial increase in produc- tion in one industry leads to increases in employment in indus- tries throughout the economy. The input–output model ignores the interactions that are the sources of uncertainty and debate in the recent economic research (continued on next page) 

50 Transportation Investments in Response to Economic Downturns BOX 2-1 (continued) Estimates of Stimulus Spending Effects from Economic Impact Studies on multipliers: the effects of stimulus spending on wages, labor supply, prices, interest rates, and consumers’ and investors’ expec- tations. In the simplest applications of the model, capacity con- straints are ignored and the impacts predicted do not depend on whether the economy is in recession or expansion. The possibil- ity of crowding-out (increased government spending depressing spending that otherwise would have occurred) is not allowed for. Because of these substantial model limitations, input–output assessments of the impact of stimulus spending do not help to narrow the range of uncertainty in the recent economic research concerning the size of fiscal stimulus multipliers. An input–output model may shed light on relative multiplier effects of alternative categories of government purchases, but for this purpose, a model based on current, accurate, and detailed industry data would be necessary, and the variability of market conditions that can change quickly (e.g., prices or import content of goods) would limit the utility of such estimates. The model is most appropriate for revealing the initial distribution of effects of a program by industry or region (Grady and Muller 1988, 60). required to enact legislation appears to have been more important than delays in spending once funds were available. Chapter 3 will describe the rate of spending of ARRA transportation construction funds. The study of differences in multipliers in recessions and expansions cited above found evidence that the size of the multiplier tends to decline quickly as soon as the economy begins to grow following a trough (Auerbach and Gorodnichenko 2012, 11). Therefore, if the stimulus is delayed, there is a risk not only that it will miss the period of greatest need but also that the amount of improvement in employment and income per dollar of spending will be diminished. 

Fiscal Policy in Recessions and the Role of Transportation Spending 51 Government spending has an immediate stimulus effect as soon as it is announced (e.g., as soon as Congress appropriates funds for a pub- lic works program). For example, as soon as it becomes known that a state transportation agency will be letting new contracts, highway con- struction contractors may begin to recruit workers in order to have the capacity to bid on the projects. An analysis of this effect concluded that “the model disputes the common view that the long ramp-up in pur- chases will delay the effects of the stimulus until long after they would be most beneficial. Rather, announcing future purchases delivers immedi- ate stimulus” (Hall 2009, 226–227). If this immediate stimulus occurs, the extended time period for federal disbursements in an infrastructure program (compared with stimulus in other forms, such as payments to individuals) will be a less significant drawback. However, lengthening the delay in undertaking announced construction of productive pub- lic capital may dampen the stimulus, in part because private-sector investors will delay their spending in anticipation of the productivity improvements provided by the public facilities (Leeper et al. 2010, 3–4). At the outset of the 2007–2009 recession, some observers predicted accurately that recovery was likely to be protracted on account of the sever- ity of the financial crisis that accompanied the recession [e.g., Summers (2007), in a newspaper opinion article]. When a slow recovery from a reces- sion can be predicted, stimulus spending is less likely to fail for lack of timeliness, and spending programs (such as infrastructure construction) that extend over multiple quarters may be suitable. In contrast, the infra- structure spending stimulus programs of the 1970s and 1980s described in Chapter 1 were too late to aid the more rapid recoveries from the recessions of those periods, which were already under way when the pro- grams were enacted. Summary The committee’s task was to assess the value of transportation spending as fiscal stimulus. Research results concerning the effects of fiscal pol- icy during recessions that are relevant to this task are those that reduce uncertainty about the effectiveness of fiscal stimulus during a recession, indicate the kind of fiscal stimulus that is most effective, and indicate 

52 Transportation Investments in Response to Economic Downturns the external circumstances that affect a fiscal stimulus program. Results concerning these questions include the following: • Although estimates of the magnitude of effects vary over a wide range, empirical studies support the conclusion that federal stimulus spend- ing during a recession or period of high unemployment, when mon- etary policy is suppressing interest rates, leads to increases in GDP and employment, at least in the short run (within 1 or 2 years after the spending). • Studies that directly measured employment and income responses to ARRA have concluded that ARRA raised employment and income, compared with the levels that would have prevailed if the extra federal spending had not occurred. • The following features of a stimulus program influence its success: – Timeliness: The income multiplier is larger during the recession and diminishes once recovery is under way, so a stimulus that is enacted late (as were U.S. stimulus programs in earlier decades) misses the period when the need and the impact of spending are greatest. Stimu- lus impact begins as soon as definite plans for increased government spending become known. – Sectors of the economy in which federal purchases are made: For example, studies based on historical experience have found that fed- eral defense purchases, federal spending and grants for infrastructure, and payments to low-income individuals have been more effective stimulus than other forms of spending. The relative impacts of differ- ent forms of spending probably depend on the circumstances of each recession. – Form of the federal spending: CBO’s estimates of the impact of ARRA assume that direct federal purchases are more effective stimu- lus than grants to the states earmarked for similar purchases, presum- ably because some substitution of federal funds for state funds by grant recipients is believed to be unavoidable. Research studies rel- evant to judging the extent of this fiscal substitution have not yielded consistent results. • Factors external to the stimulus package also influence success: – If the recession and recovery are expected to be protracted, the risk that a stimulus program will be too late is reduced. 

Fiscal Policy in Recessions and the Role of Transportation Spending 53 – If the initiation of the recession is concentrated in certain sectors or regions, stimulus spending targeting them may be more effective (although empirical evidence of the effectiveness of such targeting is not available). – If the Federal Reserve is enforcing low interest rates, the multiplier will be larger than if interest rates can rise in response to the stimulus spending. TRANSPORTATION SPENDING AS FISCAL STIMULUS An evaluation to decide the role of transportation spending in a fiscal stimulus program, once the budget of the total program had been deter- mined, ideally would include the following steps: 1. Rate the alternative forms of stimulus actions available according to their relative stimulus impact (e.g., in terms of a multiplier). The rating would take into account circumstances that may affect multipliers, for example, the duration of the recession and sectors most affected, mon- etary policy, and differences between direct federal spending and grants. 2. Identify practical constraints on the use of any of the stimulus options, for example, in the case of transportation spending, constraints on institutional capacity to spend new funds promptly and the availabil- ity of worthwhile projects. 3. Estimate long-term effects and benefits other than stimulus of the stimulus options. These may include macroeconomic effects (e.g., consequences of adding to the federal deficit), the value of the services provided by infrastructure constructed, returns from other kinds of capital spending options, and the social benefit that derives from redistribution to relieve distress. Options would be defined in terms of the category of spending (e.g., government purchases, payments to individuals) and the amount of spending of each category. Benefits of added increments of spending in a category might decrease as the amount of spending in the category was increased. 4. Select the package of stimulus expenditures, which may include trans- portation spending, that maximizes the benefit of the stimulus pro- gram, considering the expected short- and long-term benefits and costs of each option, the uncertainty of the estimates, and the practi- cal constraints on the use of each option. 

54 Transportation Investments in Response to Economic Downturns The ARRA experience highlighted characteristics of transportation spending, described below, that would be relevant in each of the four evaluation steps—rating short-term stimulus impact, identifying con- straints, estimating long-term effects, and selecting the makeup of the overall stimulus package. Relative Short-Term Stimulus Impact Studies that ranked alternative stimulus expenditures by size of multi- plier were described above. Few estimates of multipliers for specific cate- gories of government spending are available. However, research indicates that multipliers can vary substantially among spending categories and that infrastructure spending can be at least as effective as other categories of stimulus spending. Transportation spending may have some special advantages that con- tribute to its stimulus impact. As noted in the preceding section, in reces- sions such as that of 2007–2009, in which the construction industry is strongly affected, infrastructure spending may be well-targeted as stimu- lus. Construction spending also may be more useful when a protracted recovery period can be reliably predicted, because the start-up lag in construction is not then a fatal drawback (Summers 2008). Finally, trans- portation infrastructure expansion, by adding to the productive capital stock, may raise consumers’ and investors’ expectations for economic growth, providing an immediate stimulus effect beyond that produced by equal expenditures for nonproductive purposes. Chapter 1 described the objections that have been raised in the past to the use of transportation or infrastructure spending as stimulus. They include the problem of timeliness, the propensity of state and local gov- ernments to substitute federal funds for their own funds (since most federal transportation spending is in the form of grants to state and local governments), and the risk of poor project selection when spending is accelerated. These objections must be assessed alongside the possible advantages of transportation spending as stimulus. Chapter 3 describes the experience of the ARRA transportation grant program with regard to the timing of spending, impact on total state transportation spending, and project selection. 

Fiscal Policy in Recessions and the Role of Transportation Spending 55 Practical Constraints Practical limits on the amount of funding for transportation that could be used effectively in any future stimulus program are imposed by the capac- ity of the transportation construction industry, state and local transpor- tation agencies’ capacities to manage an unanticipated surge in funding, federal administrative agencies’ capacities to manage and oversee new grant programs or expansions of existing programs, and the opportuni- ties available to invest the stimulus funds in worthwhile transportation projects. The experience of ARRA, described in Chapter 3, illustrates these constraints. Other forms of government purchases in a stimulus package would be subject to analogous constraints. Highway and transit aid, although a small part of ARRA, was substan- tial compared with the normal scale of federal highway and transit aid and state and local highway and transit expenditures. The $35.9 billion in highway and transit aid provided by ARRA equals 77 percent of the amount of regular federal highway and transit aid distributed in 2008 and 32 percent of all state and local government highway and transit capi- tal spending in 2008 (FHWA 2010; APTA 2012). The ARRA aid was to be obligated in 18 months; therefore, it represented about a 50 percent increase in the agencies’ federal grant processing workload and a 20 percent increase in their normal capital spending rate. Because the funding increment was moderate and the administrative procedures (of the regular federal trans- portation aid programs) were in place, the ARRA stimulus spending may not have been seriously constrained by the agencies’ administrative capaci- ties. Nonetheless, as described in Chapter 3, the Government Account- ability Office (GAO) found that processing ARRA grants slowed the states’ processing of regular federal-aid highway grants (GAO 2010, 69–73). Total government spending on construction and operation of high- ways and transit in the United States was $249 billion in 2010 (FHWA 2012b, Table HF-10; APTA 2012, Table 56). ARRA provided $831 billion in spending and tax relief (to be spent mostly over 2 years). Therefore, an increase in annual highway and transit spending as large as 50 percent (which would require a buildup of capacity in the highway construction industry and public agencies over a period of years) would constitute only a minority of spending in a comprehensive stimulus program the size of ARRA. 

56 Transportation Investments in Response to Economic Downturns Long-Term Effects Potential long-term benefits of transportation investment are central to the case for including transportation in a stimulus spending package. They include the following: • If projects are well selected, the benefits of the transportation services provided will offset the initial cost. Including transportation invest- ment in a stimulus provides a hedge against the possibility that the government expenditure multiplier is smaller than hoped for because the benefits of the facilities constructed may justify the expenditure even if little short-term stimulus benefit is produced. • Transportation investments, unlike consumption expenditures, may tend to restrain price increases and thus reduce the inflationary effects of stimulus spending that would result from increases in demand, because they increase the productive capacity of the economy (Freedman et al. 2009, 20). This effect on the price level will make it less likely that the Federal Reserve will need to hold back a recovery by increasing inter- est rates to combat inflation. As noted above, an expenditure that increases future productive capacity may have an immediate stimulus effect because expectations of future growth will encourage current spending. • Stimulus spending that consists of accelerating planned expenditures (for example, planned road construction projects and routine reha- bilitation) adds less to public debt than expenditures that would not have been made in the absence of the need for stimulus (Feldstein 2008; Feldstein 2009). Limiting the long-term addition to the public debt was one of CBO’s criteria for judging stimulus options, on the grounds that higher debt eventually slows growth (CBO 2010, 13). • Construction costs are likely to be lower during a recession. Materials prices may have moderated, and there will be less competition for contractors’ services and for use of construction equipment. For exam- ple, the Federal Highway Administration’s National Highway Construc- tion Cost Index declined 24 percent from the second half of 2006 to the second half of 2010 (FHWA 2012a). (See Figure 3-9 in Chapter 3.) With lower prices, transportation agencies can buy more with the funds available, and some projects that were not attractive during the peak 

Fiscal Policy in Recessions and the Role of Transportation Spending 57 period may become feasible. From the perspective of social costs, the opportunity cost of employing workers who otherwise would be idle is lower than the cost to the contracting agency. To gain these advantages, evaluation of the transportation benefits of candidate projects is necessary. GAO, in its reviews of ARRA trans- portation spending, found that the states and the U.S. Department of Transportation (U.S. DOT) could document immediate job creation but had not evaluated the long-term benefits of the spending. GAO rec- ommended that the U.S. DOT determine a method for such evaluation (GAO 2011, 15–16). This need exists not only for stimulus spending but for all public transportation spending. Comprehensive economic evaluation does not play a decisive role in most government transpor- tation investment decision making. Rather, the objective inputs to pro- ject selection typically are comparisons of physical measures of system performance and infrastructure condition with defined goals for these measures. Proponents of infrastructure spending as stimulus have sometimes suggested that the long-term benefits of infrastructure investment are evident (e.g., Summers 2013). However, empirical studies of histori- cal transportation system expansion have not consistently found high marginal returns to expansion. [For example, typical findings indicate positive but modest returns to present-day incremental expansions of the highway system (TRB 2006, 70–73; Jiwattanakulpaisarn et al. 2012).] Ensuring positive returns will require designing infrastructure spending programs so as to minimize the risk of poor project selection. Chapter 3 cites evidence that state and local government total capi- tal spending and state and local noncapital transportation spending are procyclical; that is, they tend to accelerate when the economy is grow- ing strongly and to decelerate during downturns. This pattern amplifies business cycles, may delay project completion, and prevents transporta- tion agencies from taking advantage of lower prices during recessions. Therefore, managing transportation spending so as to aid in smoothing out of business cycles would benefit the economy as a whole and trans- portation programs. Adjusting the timing of transportation spending in this way would incur the cost of deviating from the otherwise optimal schedule of expenditures. 

58 Transportation Investments in Response to Economic Downturns This potential benefit of transportation spending as a countercyclical instrument need not be a major factor in determining how much should be invested in transportation in the long term (i.e., over several business cycles) or what form the investments should take. The primary long- term contribution of transportation investment to growth of income is derived from the transportation benefits it produces for the users of the facilities constructed, which can be evaluated through benefit–cost analysis of individual transportation projects. Reliance on transporta- tion spending as a countercyclical instrument might lead to an increase in total spending because some projects that cannot be justified on the basis of their transportation benefits during normal conditions become attractive during a recession when construction prices and interest costs are lower. Also, if stimulus benefits are substantial, some spending during recessions might be justified for transportation projects that could never be justified solely on the basis of their transportation benefits. However, trans- portation spending could be a useful part of an antirecessionary stimulus program, and the benefits of making spending countercyclical could be obtained, without increasing long-term total spending on transportation. Any short-term stimulus effect from adjusting the timing of transpor- tation spending to be countercyclical may be a worthwhile secondary benefit but should not dictate the level of spending or project selection in the long run. Selecting the Stimulus Package Decisions on the actions to be included in the stimulus package and the share of total expenditure to allocate to each action depend on trade-offs between the stimulus benefits (immediate job creation) of the options and their long-term benefits and costs. The important long-term consider- ations are impact on the public debt [because some actions may add more to debt than others (e.g., advancing spending that would have occurred eventually adds less to debt than spending that never would occur if not for the stimulus program)] and the future benefits produced by spending that adds to the capital stock. Both these considerations are relevant in deciding the transportation share of the package and in the selection of projects to be built with the funds allocated to transportation. For a given 

Fiscal Policy in Recessions and the Role of Transportation Spending 59 budget, the portfolio of transportation projects that produces the greatest immediate employment is not likely to be the same as the portfolio that maximizes transportation benefits. No objective method for evaluating these short-term versus long-term trade-offs was available to guide planning or management of ARRA. Benefit–cost evaluations of transportation investments use a consumer surplus welfare measure, while short-term stimulus benefits are mea- sured in terms of a change in GDP or in employment (neither of which is a measure of the full welfare implications of the stimulus). To com- pare alternative expenditures that are intended to produce benefits in the form of fiscal stimulus and benefits from improved transportation, a common unit of measure for the two kinds of benefits is required.4 Uncertainty in impact estimates also should be a consideration in design of the stimulus package. Once the decision has been made to undertake a stimulus program, a diversified package is a reasonable strategy because the relative sizes of the multipliers for different forms of stimulus are not well known (Auerbach et al. 2010, 158). Including infrastructure in the package would contribute to diversification. REFERENCES Abbreviations APTA American Public Transportation Association CBO Congressional Budget Office CEA Council of Economic Advisers FHWA Federal Highway Administration GAO Government Accountability Office TRB Transportation Research Board UN United Nations 4 Drautzburg and Uhlig (2011) present a calculation of the welfare effects of ARRA that could be applied to assess this short-term versus long-term trade-off in the infrastructure component of a stimulus program. The welfare calculation depends critically on the discount rate for credit-constrained households (households that cannot borrow to tide themselves over during a downturn); that is, how much future consumption such households would be willing to give up to have ARRA in place (Drautzburg and Uhlig 2011, 31–34). 

60 Transportation Investments in Response to Economic Downturns Antonopoulos, R., K. Kim, T. Masterson, and A. Zacharias. 2010. Investing in Care: A Strategy for Effective and Equitable Job Creation. Working paper. Levy Economics Institute of Bard College, Aug. APTA. 2012. 2012 Public Transportation Fact Book: Appendix A: Historical Tables. March. Auerbach, A., W. G. Gale, and B. H. Harris. 2010. Activist Fiscal Policy. Journal of Eco- nomic Perspectives, Vol. 24, No. 4, pp. 141–164. Auerbach, A., and Y. Gorodnichenko. 2012. Measuring the Output Responses to Fiscal Policy. American Economic Journal: Economic Policy, Vol. 4, No. 2, May, pp. 1–27. Auerbach, A., and Y. Gorodnichenko. 2013. Corrigendum: Measuring the Output Responses to Fiscal Policy. American Economic Journal: Economic Policy, Vol. 5, No. 3, August, pp. 320–322. Baum, A., M. Poplawski-Ribeiro, and A. Weber. 2012. Fiscal Multipliers and the State of the Economy. Working paper. International Monetary Fund, Dec. Blanchard, O., and D. Leigh. 2013. Growth Forecast Errors and Fiscal Multipliers. Working paper. International Monetary Fund, Jan. Boskin, M. 2011. The Fiscal Policy Debate. Presented at Conference on Chinese Policy Reform, Beijing, March 21–22. CBO. 2008. Options for Responding to Short-Term Economic Weakness. Jan. CBO. 2010. Policies for Increasing Economic Growth and Employment in 2010 and 2011. Jan. CBO. 2011. Estimated Impact of the American Recovery and Reinvestment Act on Employ- ment and Economic Output from April 2011 Through June 2011. Aug. CBO. 2012. Estimated Impact of the American Recovery and Reinvestment Act on Employment and Economic Output from October 2011 Through December 2011. Feb. CBO. 2013a. Estimated Impact of the American Recovery and Reinvestment Act on Employ- ment and Economic Output from October 2012 Through December 2012. Feb. CBO. 2013b. Macroeconomic Effects of Alternative Budgetary Paths. Feb. CEA. 2011. The Economic Impact of the American Recovery and Reinvestment Act of 2009: Eighth Quarterly Report. Dec. 9. Chodorow-Reich, G., L. Feiveson, Z. Liscow, and W. Woolston. 2012. Does State Fiscal Relief During Recessions Increase Employment? Evidence from the American Recov- ery and Reinvestment Act. American Economic Journal: Economic Policy, Vol. 4, No. 3, Aug., pp. 118–145. Christiano, L., M. Eichenbaum, and S. Rebelo. 2011. When Is the Government Spending Multiplier Large? Journal of Political Economy, Vol. 119, No. 1, Feb., pp. 78–121. Conley, T., and W. Dupor. 2012. The American Recovery and Reinvestment Act. Presented at 80th Meeting of the Carnegie–Rochester Conference on Public Policy, Nov. 9. http:// www.carnegie-rochester.rochester.edu/crNovember12.htm. 

Fiscal Policy in Recessions and the Role of Transportation Spending 61 Davig, T., and E. Leeper. 2011. Monetary–Fiscal Policy Interactions and Fiscal Stimulus. European Economic Review, Vol. 55, No. 2, Feb., pp. 211–227. Drautzburg, T., and H. Uhlig. 2011. Fiscal Stimulus and Distortionary Taxation. Working paper. National Bureau of Economic Research, June. Elmendorf, D. 2010. Testimony: Statement of Douglas W. Elmendorf, Director. Policies for Increasing Economic Growth and Employment in the Short Term. Prepared for the Joint Economic Committee, U.S. Congress, Congressional Budget Office, Feb. 23. Feldstein, M. 2008. Defense Spending Would Be Great Stimulus. Wall Street Journal, Dec. 24. Feldstein, M. 2009. Rethinking the Role of Fiscal Policy. American Economic Review, Vol. 99, No. 2, pp. 556–559. Feldstein, M. 2011. The Economy Is Worse Than You Think. Wall Street Journal, June 8, p. A15. Feyrer, J., and B. Sacerdote. 2011. Did the Stimulus Stimulate? Real Time Estimates of the Effects of the American Recovery and Reinvestment Act. Working paper. National Bureau of Economic Research, Feb. FHWA. 2010. Highway Statistics 2008. FHWA. 2012a. Construction Cost Trends for Highways. Oct. FHWA. 2012b. Highway Statistics 2010. Freedman, C., M. Kumhof, D. Laxton, D. Muir, and S. Mursulal. 2009. Economic Stimulus to the Rescue? Short-Run Benefits and Potential Long-Run Costs of Fiscal Deficits. Work- ing paper. International Monetary Fund, Nov. GAO. 2010. Recovery Act: Opportunities to Improve Management and Strengthen Account- ability over States’ and Localities’ Uses of Funds. Sept. GAO. 2011. Recovery Act: Funding Used for Transportation Infrastructure Projects, but Some Requirements Proved Challenging. June. Grady, P., and R. Muller. 1988. On the Use and Misuse of Input–Output Based Impact Analysis in Evaluation. Canadian Journal of Program Evaluation, Vol. 3, No. 2, pp. 49–61. Hall, R. 2009. By How Much Does GDP Rise If the Government Buys More Output? Brookings Papers on Economic Activity, Fall, pp. 183–249. Hall, R. 2010. Fiscal Stimulus. Daedalus, Fall, pp. 83–94. Jiwattanakulpaisarn, P., R. Noland, and D. Graham. 2012. Marginal Productivity of Expanding Highway Capacity. Journal of Transport Economics and Policy, Vol. 46, No. 3, Sept., pp. 333–347. Leduc, S., and D. Wilson. 2012a. Highway Grants: Roads to Prosperity? Economic Letter, Federal Reserve Bank of San Francisco, Nov. 26. Leduc, S., and D. Wilson. 2012b. Roads to Prosperity or Bridges to Nowhere? Theory and Evidence on the Impact of Public Infrastructure Investment. Prepared for 2012 National Bureau of Economic Research Macroeconomics Annual Conference. 

62 Transportation Investments in Response to Economic Downturns Leduc, S., and D. Wilson. 2012c. Should Transportation Spending Be Included in a Stim- ulus Program? A Review of the Literature. Paper commissioned by the committee, draft, Sept. 24. Leeper, E., N. Traum, and T. Walker. 2011. Clearing Up the Fiscal Multiplier Morass. Work- ing paper, Sept. 11. Leeper, E., T. Walker, and S. Yang. 2010. Government Investment and Fiscal Stimulus. Working paper. International Monetary Fund, Oct. Parker, J. A. 2011. On Measuring the Effects of Fiscal Policy in Recessions. Journal of Economic Literature, Vol. 49, No. 3, Sept., pp. 703–718. Ramey, V. A. 2011. Can Government Purchases Stimulate the Economy? Journal of Eco- nomic Literature, Vol. 49, No. 3, Sept., pp. 673–685. Reichling, F., and C. Whalen. 2012. Assessing the Short-Term Effects on Output of Changes in Federal Fiscal Policies. Congressional Budget Office, May. State of Maryland. n.d. StateStat: Recovery Jobs. Accessed April 5, 2013. http://www. statestat.maryland.gov/recoveryjobs.asp. Summers, L. 2007. The State of the U.S. Economy. Conference remarks. Brookings Insti- tution, Washington, D.C., Dec. 19. Summers, L. 2008. Investing in America’s Infrastructure: From Bridges to Broadband. Conference remarks. Brookings Institution, Washington, D.C., July 25. Summers, L. 2013. America’s Deficits: The Problem Is More Than Fiscal. Washington Post, Jan. 21. Surface Transportation Policy Project. 2004. Setting the Record Straight: Transit, Fixing Roads and Bridges Offer Greatest Job Gains. Jan. 28. Taylor, J. 2011. An Empirical Analysis of the Revival of Fiscal Activism in the 2000s. Jour- nal of Economic Literature, Vol. 49, No. 3, Sept., pp. 686–702. TRB. 2006. Special Report 285: The Fuel Tax and Alternatives for Transportation Funding. National Academies, Washington, D.C. UN. 1999. Handbook of Input–Output Table Compilation and Analysis. Department of Economic and Social Affairs, New York. Washington State Department of Transportation. 2010. Highway Construction Job Estimates for ARRA Projects. Revised May 25. http://www.wsdot.wa.gov/Funding/ stimulus/jobs.htm. Wilson, D. 2012. Fiscal Spending Jobs Multipliers: Evidence from the 2009 American Recovery and Reinvestment Act. American Economic Journal: Economic Policy, Vol. 4, No. 3, Aug., pp. 251–282.