Recovery and Reinvestment Act Spending at the State Level: Keynesian Stimulus or Distributive Politics?

Transcription

1 Recovery and Reinvestment Act Spending at the State Level: Keynesian Stimulus or Distributive Politics? Andrew T. Young Department of Economics West Virginia University Morgantown, WV (304) Russell S. Sobel Department of Economics West Virginia University Morgantown, WV (304) September 2010 JEL Codes: D7, H5, E6, Keywords: fiscal stimulus, fiscal policy, political economy, public choice, congressional dominance model, American recovery and reinvestment act

2 Recovery and Reinvestment Act Spending at the State Level: Keynesian Stimulus or Distributive Politics? Abstract: We examine the US state-level pattern of American Recovery and Reinvestment Act (ARRA) spending. We relate spending to (1) Keynesian determinants of countercyclical policy, (2) congressional power and dominance, and (3) presidential electoral vote importance. We find that the ARRA is, in practice, poorly-designed countercyclical stimulus. After controlling for political variables, coefficients on Keynesian variables are often statistically insignificant. When they are statistically significant they are often the incorrect sign. On the other hand, statistically significant effects associated with political variables are almost always of the sign predicted by public choice theory. One striking result is that the elasticity of ARRA spending with respect to the pre-arra levels of federal grants and payments to state and local governments is between and States previously capturing large amounts of federal funds continue to do so under the ARRA stimulus. JEL Codes: D7, H5, E6, Keywords: fiscal stimulus, fiscal policy, political economy, public choice, congressional dominance model, American recovery and reinvestment act 1

3 1. Introduction On February 13 th, 2009 the American Recovery and Reinvestment Act (ARRA) was passed by the US. Congress. Four days later, President Barack Obama signed the bill into law. The ARRA provided for some $787 billion in combined spending increases and tax cuts aimed at stimulating the national economy. 1 At the aggregate level, this would appear to be textbook Keynesian countercyclical policy. However, especially when it comes to how to geographically allocate government spending, macroeconomic policy is never as simple as increasing G in a model. The ARRA s spending component is $500 billion and projects must be defined; regions to locate the projects must be determined; individuals and firms to carry out the projects must be hired. These choices are made through the political process, and there is a large literature in public choice theory devoted to understanding this complex process. Ideally, to maximize the effectiveness of the stimulus spending from a Keynesian perspective, funds should be targeted at the areas or industries with the most unemployed resources, and/or the areas where the marginal propensity to consume (MPC) is high. Figure 1 illustrates this graphically in the context of a short-run aggregate supply (SRAS) and aggregate demand (AD) model. In the upper panel, a given stimulus (in the form of an increase in AD) yields a larger increase in output/expenditures if there are slack resources (moving along the flatter portion of SRAS). Alternatively, starting from a higher initial output, a given stimulus shows up to a larger extent in nominal price increases (moving along the steeper portion of SRAS). In the bottom panel, an increase in aggregate demand is illustrated in terms of initial government stimulus spending (autonomous expenditures) and then multiplier effects (induced expenditures). A higher MPC will be associated with a higher elasticity of private expenditures 1 Basic facts, including expenditure and tax numbers, can be found at recovery.gov. 2

4 with respect to each stimulus dollar. There will, all else equal, be larger multiplier effects and a greater induced shift in AD. In evaluating the ARRA, neither the Obama administration s Council of Economic Advisors (CEA) nor the Congress Budget Office (CBO) consider, at least explicitly, the statelevel spending patterns (e.g., CEA (2010) and CBO (2010)). As the most explicit example, the CEA (pp ) does try to assess state-level ARRA effects on employment. Their approach is to take their aggregate employment creation number and calculate states shares of that aggregate according to (1) states shares of aggregate non-farm employment, (2) state-level shares of ARRA outlays, or (3) industry employment shares in each state. None of these approaches considers that multipliers may be different from state to state. If state-level ARRA spending patterns are not determined with these differences in mind then policy-makers may be getting less bang than they could for their stimulus buck. Have the Keynesian countercyclical policy criteria been important determinants of ARRA allocations across states? Figures 2 and 3 present the pattern of per capita ARRA funds across US states in relation to per capita income levels and unemployment rates, respectively. We consider three measures of ARRA funds as reported by federal agencies: (1) announced, (2) made available and (3) paid out. 2 The prima facie evidence for the importance of optimal countercyclical policy criteria is mixed. While there is some indication that states with higher unemployment rates have been receiving more ARRA funds, there is also some indication that lower income levels have correlated with less funding. If countercyclical policy criteria are not the whole story, an alternative is that political criteria have important determinants of the ARRA allocations. Given a spending package such as 2 There are also recipient-reported measures of funding. While these are not considered in figures 2 or 3 they are analyzed in the formal empirical analysis below. 3

5 the ARRA, it is possible that political incentives confronting both President Obama and members of the US Congress played an important role in the allocation of funds across states. States that are politically important in the next presidential reelection, or states with tenured and powerful congressional representation may have been able to secure a larger proportion of funding than would have been warranted based on Keynesian criteria. To the extent that his is true, the effectiveness of the countercyclical policy may be compromised by both distributive politics and rent-seeking. The purpose of this paper is to examine the pattern of ARRA spending across states and ask what factors played a significant role in determining state level funding. We relate state-level ARRA spending to (1) Keynesian determinants of countercyclical policy, (2) congressional power and dominance, and (3) presidential electoral vote importance. In terms of congressional power and dominance our data cover both houses of Congress and include average legislator tenure and appropriations committee membership. We also estimate the effects of appropriations subcommittee and authorization committee membership by considering separately the state level funding for the four largest cabinet departments (Health and Human Services, Education, Energy, and Transportation). In terms of electoral vote maximization our data include an electoral importance measure that is increasing in both a state s electoral votes and the closeness of the previous presidential election popular vote. We find that the ARRA is, in practice, poorly designed countercyclical stimulus. After controlling for political variables, coefficients on Keynesian variables are often statistically insignificant. When they are statistically significant they are often in the opposite direction as would be predicted by Keynesian theory. On the other hand, statistically significant effects 4

6 associated with political variables are almost always of the sign predicted by public choice theory. Of the political variables, one striking result is that the elasticity of ARRA spending with respect to the pre-arra levels of federal grants and payments to state and local governments is between and We interpret this variable as a proxy for a state s accumulated ability to extract federal funds (extraction capital). States previously demonstrating the ability to capture large amounts of federal funds continued to demonstrate that ability in regards to the ARRA stimulus package. Thus, the ARRA in practice mostly gave more money to the same states that were the most politically powerful in receiving it pre-stimulus. This paper is organized as follows. Section 2 discusses previous literature relevant to this paper, especially works that inform our selection of political determinants. The data included in our econometric analysis are discussed in section 3 while the results of the analysis are presented in section 4. Section 5 concludes. 2. Existing Theory and Evidence Our evaluation of the ARRA as countercyclical policy is informed largely by the standard, textbook fare of Keynesian macroeconomics. On the other hand, while politics matter may be an uncontroversial statement, choosing variables belong for an analysis of distributive politics requires some care. What kind of political power is important; how do we measure it? In considering the determination of a federal spending package, both the legislative and executive branches are potential sources of influence. Concerning the legislative branch, economists have elaborated a congressional dominance model (Weingast and Moran (1983), Weingast (1984), Weingast and Marshall (1988), and Moe (1987 & 1997)). In this model, 5

7 congressional committee members hold monopolies of authorization and/or appropriation on legislation falling under their committee s jurisdiction. The seniority system guarantees a committee member property rights over his or her seat (Holcombe and Parker (1991), Weingast and Marshall, 1988, p. 132). The seniority system also grants leadership positions based on tenure. Legislation is the result of political competition between self-interested representatives and senators where competitive advantages are functions of committee membership and seniority. Concerning the executive branch, the president wields influence by broadly setting agendas and, more directly, exercising (or threatening to exercise) the veto option to maximize electoral votes (e.g., Wright (1974)). 3 Adding to the political competition, the executive branch favors legislation benefiting politically important states in the Electoral College. The underlying goal is reassuring reelection for the current president or a successor from the current president s party. We can refer to the above as the electoral vote maximization model. In this paper we analyze not only the empirical relevance of Keynesian versus political determinants of the ARRA, but also, within the political determinants, the relevance of the congressional dominance model relative to the electoral vote maximization model. As such, this paper is related to several previous studies presenting evidence of political rent-seeking in various contexts. 4 Most analogous to our study are the works by Wright (1974), Anderson and Tollison (1991), and Couch and Shughart (1997) linking New Deal spending across states to 3 Such an approximation of behavior is probably of increased relevance in the case of a first-term president. 4 E.g., Faith et al. (1982) find that Federal Trade Commission (FTC) rulings tend to be more favorable in districts with congressional oversight committee membership. Young et al. (2001) present evidence that state IRS audit rates are sensitive to the political importance states in the upcoming presidential election. Kosnik (2005) reports that congressional representation effects he Federal Energy Regulatory Commission s (FERC s) relicensing of hydroelectric dams. As a final example, Coats et al. (2006) document the fact the Homeland Security grants following 9/11 were designed in a way consistent with electoral vote maximization. 6

8 measures of both congressional power and the importance of states electoral votes in the upcoming presidential election. The ARRA was response to the worst recession since the Great Depression. During the Great Depression, the New Deal was an analogous response. Anderson and Tollison (1991, p. 164) state: Traditional arguments imply that state-by-state allocation of [New Deal] spending was a direct function of the need [ ] of the residents of various states for federal relief from economic distress. Their evidence, as well as that presented by Reading (1973) and Wright (1974), calls these traditional arguments into question. Anderson and Tollison (1991, p. 175) reach the pointed conclusion that the New Deal was not big government s Garden of Eden, but rather the more familiar stomping ground of Homo economicus. Our analysis is not designed to include the need for federal relief per se. However, our countercyclical determinants (e.g., per capita income levels; unemployment rates) surely control for such. This being the case, our analysis can be interpreted as addressing, with respect to ARRA, the questions considered by the authors above with respect to the New Deal. This paper is also related to series of reports by Brito and de Rugy (2010) de Rugy (2010) examining recipient-reported data at the congressional district level. Whereas much of their focus (in terms of political determinants) is on party affiliation, our state-level analysis is more natural for the seniority and committee membership variables suggested by the congressional dominance model. We also analyze authorization and subcommittee membership, as well as the previous level of federal grants and payments, which their reports do not. 7

9 3. Data Our data on ARRA funding is obtained from the federal government s recovery.gov website. We collect both agency-reported and recipient-reported funds measures for all 50 US states. Agency-reported data comes from federal agency final reports, Federal Procurement Data System, and USASpending. 5 Recipient-reported data is collected directly from contract, grant, and loan recipients. 6 Agency-reported measures are funds announced, funds made available, and funds paid out. Recipient-reported measures are funds awarded and funds received. We consider all five of these ARRA funds measures in our regressions. The ARRA stimulus remains a work in progress as of this writing. We collected data during a period when recovery.gov stated its last update as April 7 th, 2010 (almost 14 months subsequent to the act s passing). As of that date, aggregate agency-reported funds announced were $324 billion (or just about 65 percent of the total legislated spending component). Table 1 reports the aggregate numbers for both of the recipient-reported categories and the three agencyreported categories. (As one would expect funds awarded is greater than funds received; funds announced is greater than those made available, which is greater than those already paid out.) Table 1 also reports separately funds administered by the Department of Health and Human Services, Department of Education, Department of Transportation, and Department of Energy. As the last row of table 1 reports, these four departments together account for a substantial portion of total ARRA funds over 80 percent in terms of the recipient reported data; between 53 and 65 percent in terms of the agency-reported data. Table 2 reports the correlations across the five total ARRA funds measures. Most correlations fall into the or higher range. Notably, agency-reported funds announced have

10 positive but relatively low correlations with other funds measures (including the other two agency-reported measures). However, even these correlations are all above Despite the large differences in levels (table 1) the cross-state variation in funds is quite similar across funds measures (agency-reported funds announced being, again, somewhat of an exception). Summary statistics for the state-level ARRA funds measures are reported in table 3. The standard deviations are all considerable relative to the means. Standard deviations taken as percents of the means (coefficients of variation) are between and The distributions are noticeably skewed, the mean values being typically closer to the minimum values than the maximum values. We will be discussing regression results in terms of elasticities so the last row of table 3 reports the standard deviations of the natural log of each variable. (Table 3 also includes the abbreviated terminology employed in subsequent tables, e.g., RRFUNDAW for recipient-reported funds awarded.) The ARRA funds measures serve as our dependent variables in our empirical analysis. We relate ARRA funds to variables associated with (i) Keynesian countercyclical policy, (ii) the congressional dominance model, (iii) the electoral vote maximization model. We also include (iv) various demographic controls. All of these variables, their definitions, and their sources, are reported in table 4. Summary statistics for the Keynesian (i) variables and political variables (ii) and (iii) (in the forms in which they enter regressions) are reported in tables 5 and 6, respectively. i. Keynesian Variables Countercyclical policy that aims at priming the pump in an economy would best be targeted at depressed states where economic activity is below potential and labor is idle. As such 9

11 we consider the levels of per-capita income and unemployment, as well as their changes, as potential determinants of ARRA funding. The state unemployment level we include is from January of 2009, the last level observable to policy-makers prior to the legislation s passing. Also, acknowledging that policymakers focus more on the change in economic conditions leading up to the stimulus, we include the change in a state s unemployment rate from January 2008 through January State-level GDP is only available as an annual measure, so we include the 2008 level of real per capita state GDP and the growth rate of real state GDP from 2007 through While less plausible as an actual consideration by policy-makers, we also consider that funds may be targeted to states where the marginal propensity to spend is high. This would be sensible if policy-makers wanted to maximize multiplier effects and get the largest economic response from the stimulus. We include the state-level estimates of marginal propensities to consume (MPCs) from Luengo-Prado and Sørensen (2008). While these MPC estimates are based on data from 1964 through 1998, we at least attempt to control for some relative tendencies across state populations that, perhaps, change only slowly over time. Another factor that we consider at potential relevant for designing countercyclical stimulus is a state government s budgetary situation. The recent recession has been characterized by several state government budget crises where public spending and payments to public workers have been threatened. Since such decreases in state government expenditures may essentially offset federal stimulus, policy-makers may target funds where those decreases appear likely. We include the growth in a state s tax revenue from 2007 through 2008 to control for this possibility. 10

12 ii. Congressional Dominance Variables The congressional dominance model suggests the spending composition of a stimulus package such as the ARRA is the result of political competition between self-interested legislators. Advantages enjoyed by a specific congressperson in that competition are a function of committee membership and seniority. We account for seniority by including the average tenure of a state s senators and representatives, separately, at the time of the ARRA s passing. In both the House and the Senate, appropriations committees deal with determining the specific funds allocated to federal agencies, departments, and organization. The appropriations committees deal with drafting legislation specifying such funds allocations. We include the number of legislators for each state who are members of the House appropriations committee as variables; likewise for the Senate appropriations committee. When we turn to analyzing a panel of four department-based ARRA measures, we extend the congressional dominance variables to include memberships in appropriations subcommittees and authorization committees that are associated with those specific departments. A description of those variables is reserved for section 4. (Also see table 10 below.) Finally, we include 2008 federal grants and payments to state and local governments in a given state. This is a variable that we include in all of our regressions. We believe that it is that it is sensible to evaluate ARRA allocations by state relative to initial federal government allocations. However, we also believe that this variable can be interpreted in terms of the congressional dominance model (broadly-conceived) as a state s stock of extraction capital, i.e., accumulated political apparatus and skills aimed at capturing federal funds. In other words, a good measure of a state s political power in securing federal funds is how much they have been 11

13 able to capture in the recent past. There may also be institutional stickiness associated with grant and payment flows and current committee appointments may be partly influenced by a state s previous legislators. If this is true, then the federal grants and payments variable proxies for lagged values of congressional dominance variables. iii. Electoral Vote Maximization Variables Unlike the model s name might naively suggest, we do not simply include the number electoral votes that a state commands. Rather, we follow Garrett and Sobel (2003) in constructing an electoral importance measure. First, we record the percent of a state s popular vote won by Barack Obama in 2008 (call it X). We then calculate Y = 1-4 (X 0.5) 2. Our electoral importance measure is then a state s electoral votes multiplied by Y. The weight (Y) placed on the electoral votes (X) is at a maximum of one for a popular vote share of 50 percent. The weight decreases symmetrically towards zero as the vote share goes to 0 or to 100 percent. Intuitively, the electoral importance of a state increases in (1) the number of electoral votes the state has and (2) the expected closeness of the state in the next election. For example, despite commanding 31 electoral votes, a democratic president may be unlikely to allocate many resources to New York because a democratic landslide is already highly probable. (Barack Obama won 63 percent of the popular vote in 2008.) On the other hand, North Carolina might be deemed worthy of a large amount of resources because, despite commanding only 15 electoral votes, it was a battleground state in the previous election and may be again in the next election. (Barack Obama won the state by just over 50 percent of the popular vote in 2008.) In addition to the electoral importance measure, we also include a dummy variable that takes the value of one if Barack Obama carried the state in the 2008 presidential election; zero 12

14 otherwise. Including this variable is intended to control for the fact that, especially in the case of a first-term president, resources may be allocated to reward states for their past votes (perhaps in the hopes of convincing voters that it behooves them to cast their votes likewise in the next election). iii. Demographic Controls The most basic demographic control that we include is a state s population in In our regressions, dependent variables (i.e., ARRA funds measures) are always logged and logged population is always included on the right-hand-side. This means that our estimated coefficients on other regressors can be interpreted as elasticities of per capita ARRA funds. We also include the 2008 percent of a state s population that is over 65 years of age; the percent that is African American and the percent that is Latino. As with all other variables, definitions and sources are reported in table Regression Results Since we have a large number of regressors (17) relative to state observations (50) we first consider whether some of the Keynesian variables are irrelevant or redundant. Redundancy is a particular concern with the Keynesian variables since we consider real per capita state GDP and unemployment in terms of both levels and changes. i. Keynesian Variables Table 7 reports the results of an ordinary least squares (OLS) regression of the five (logged) ARRA funds measures on (1) a constant, (2) the logs of population and federal grants 13

15 and payments, and (3) the Keynesian variables. Included in the Keynesian variables, Real GDP per capita also enters regressions in logged form. For exposition purposes, in discussing results from table 7 and all subsequent regression results we will refer to recipient-reported funds awarded and received, and agency-reported funds announced, made available, and paid out, respectively, by the shorthand introduced in table 3: RRFUNDAW and RRFUNDRE; ARFUNDAN, ARFUNDAV, and ARFUNDPO The estimated elasticity of ARRA funds with respect to a state s population is always significant at the one percent level and large. (Point estimates are between and ) This is to be expected: larger (smaller) populations are associated with larger (smaller) allocations of ARRA funds. We include the log of population in all of our regressions reported below and the estimated elasticities are always statistically significant and quantitatively similar to those in table 7. Turning to the Keynesian variables, the unemployment level is statistically significant at the one percent level in two regressions (ARFUNDAV and ARFUNDPO). The point estimates are and respectively, so a one percent higher unemployment rate is associated with a very small fraction of a percentage more ARRA funding. 7 A standard deviation increase in the unemployment rate (1.85 percent) is not even associated with a one percent increase in ARRA funding. The change in the unemployment rate is also statistically significant in two regressions (RRFUNDAW and ARFUNDPO). However, it is only significant at the 10 percent level; more importantly, the sign is positive or negative depending on the regression. The per capita GDP level is statistically significant at the one percent level in two cases (ARFUNDAV and ARFUNDPO). In these cases the point estimate elasticities are and 7 The unemployment rate enters the regression as a percentage rate so if, for example, the unemployment rate for a state is 4 percent the corresponding observation is 4 rather than

16 These elasticities are economically very large given that the standard deviation of GDP is about 17.5 percent, which is associated with an increase in ARRA funds between 5.8 and 9.3 percent. (The standard deviations of logged ARFUNDAV and ARFUNDPO are, respectively, 1.0 and 1.1 percent.) However, the effects are of the opposite sign as would be predicted by Keynesian theory: all else equal, more ARRA funds are allocated to less depressed states. There is, on the other hand, one case (RRFUNDRE) where the estimated effect on GDP is negative and significant, though only at the 10 percent level. If the positive effect of GDP on ARRA funds is to believed, it could possibly be unrelated to cyclical conditions, i.e., perhaps it is simply richer (rather than less-depressed) states that are able to extract a large slice of the ARRA pie. (Recall, however, that we are controlling for states previous level of federal aid.) Perhaps the recent GDP growth is more indicative of how the recent recession affected a state. In the case of real GDP growth, the estimated effects are never statistically significant. Also, in five out of the five regressions reported in table 7, the coefficient is again of the wrong sign from a Keynesian perspective: higher recent GDP growth is associated with more ARRA funds. Likewise, the estimated effects on the Luengo-Prado and Sørensen (2008) MPC estimates are never statistically significant. In a majority of the regressions (three out of five) higher marginal propensities to consume are associated with lower ARRA funds. In terms of countercyclical policy, this would imply that funds have been targeted where the potential for multiplier effects is lower. Finally, state tax revenue growth on ARRA funds is significant at the one percent level in two cases (RRFUNDAW and RRFUNDRE) and the estimated elasticity is well above unity (1.881 or 1.774, respectively). However, the sign of the effects in both significant cases implies 15

17 that states with improving budgets are allocated more ARRA funds. We are inclined to dismiss these estimated effects as spurious. While the sign on GDP is wrong in two cases from the perspective of countercyclical policy, one can imagine reasons why wealthier states (with, e.g., more educated populations and politicians) are able to extract a large part of a stimulus. However, in the case of state tax revenue growth, no reasonable, direct and positive relationship between tax revenue growth and ARRA funds seems plausible. Table 7 reports the results of Breusch-Pagan (1979) tests for heteroscedasticity. There are only two regressions (with dependent variables ARFUNDAV and ARFUNDPO) where the null of no heteroscedasticity is rejected, and only at the 10 percent level. We calculated White (1980) heteroscedasticity-consistent errors. The only resulting change in the ARFUNAV results is that state tax revenue growth becomes significant at the five percent level. However, it is positive as in the other significant revenue growth cases. In the ARFUNDPO regression the change in unemployment becomes significant at the five rather than only 10 percent level. Otherwise the results are unchanged meaningfully. Notable, in two regressions (RRFUNDAW and (ARFUNDAV) the logged level of previous federal grants and payments has positive coefficient estimates that are significant at the one percent level. The point estimates are also economically not negligible (0.259 and respectively). The standard deviation of log of FEDAID is about 0.9 percent and is associated with somewhere between 23 and 27 percent of a standard deviation in funds. The only reasonable evidence of ARRA funds being allocated according to countercyclical criteria (in a way that would suggest improved policy) appears in the unemployment level coefficients. However, as remarked above the estimated effects are quite small. On the other hand, when the other coefficients are statistically significant they are most 16

18 often the wrong sign when viewed from a Keynesian perspective and, perversely, economically large. In the case of real per capita GDP levels, the results may suggest that wealthier states are receiving a disproportionately large share of the stimulus. The remainder of the paper explores whether political determinants are better accounted for by political variables suggested by the congressional dominance and electoral vote maximization models. In the regressions that follow, we continue to include the per capita GDP and unemployment levels as controls. ii. Political Variables Table 8 reports the results from OLS regressions of funds measures on the political variables while controlling for demographic (BLACK, LATINO, and SENIORS) factors and the real per capita GDP and unemployment rate levels. As in all of our regressions, population and the previous level of federal aid to state and local governments are controlled for. As before, population is always positive and statistically significant at the one percent level. Average tenure of a state s representatives has a statistically significant (at the one percent level estimated effect in two cases (RRFUNAW and ARFUNDAV). The signs are positive as the congressional dominance model would predict. However, the coefficients are economically small (0.021 and respectively). A standard deviation increase in the average tenure of a state s representatives is associated with only about a tenth of a percent increase in funds. Average tenure of a state s senators is also statistically significant in two regressions (ARFUNDAV and ARFUNDPO). Again, however, thought the coefficients are the right signs (now when assessed relative to the predictions of public choice theory) they are economically 17

19 small: both are relative to the standard deviation of the average Senate tenure variable. OBAMA08 is statistically significant at the five percent level in two regressions (ARFUNDAV and ARFUNDPO) but, like the congressional dominance variables, the estimated effects are economically small. The point estimates are and 0.136, respectively, while the standard deviation of OBAMA08 is only The only political variable that is estimated to have both statistically and economically significant effects is the extraction capital measure. The coefficient on the log of FEDAID is statistically significant in three regressions (RRFUNDAW, ARFUNDAV and ARFUNDPO). Furthermore, the point estimates are all at least as large as those reported on FEDAID in table 7; they range from to Recall that the standard deviation of the log of federal grants and payments to state and local governments is Neither Keynesian variables nor political variables directly predicted by the congressional dominance and electoral vote maximization models appear to be important. What does seem to be important is evidence of states previous skill at capturing federal funds. 8 iii. Department Panel Results; Authorization and Subcommittees In evaluating the congressional dominance model, our evaluation of committee representation effects was limited above to appropriations committee membership. However, it is possible that important effects are associated instead with either appropriation at the subcommittee level or authorization committees. To explore this possibility we explore the data on ARRA funds associated with four different federal Departments: Health and Human Services 8 There is one regression where LATINO and SENIORS both have statistically significant estimated effects. However, we do not dwell on those here because they are not are focus and, importantly, the effects are economically negligible. (The standard deviations of LATINO and SENIORS are, respectively, and ) 18

20 (HHS), Education (ED), Transportation (TR) and Energy (EN). Table 9 reports summary statistics on ARRA funds measures for each of these departments at the state-level. We collect data on each state s membership on not only House and Senate (1) appropriations committees (HAC and SAC), but also (3) appropriations subcommittees (HAPSUB and SAPSUB), (3) authorization committees (HAUTH and SAUTH) and (4) authorization subcommittees (HAUTHS) associated with each federal department. (Note that there are not Senate authorization subcommittees.) Table 10 reports these various committees and subcommittees by department. Tables 11A and 11B report on panel regressions that include all variables from the table 8 regressions plus HAPSUB, SAPSUB, HAUTH, SAUTH, and HAUTHS. There is no time dimension, as one would typically associate with a panel of data, but rather the dimensions are departments by states: 4 50 = 200 observations. Estimation is by OLS with fixed effects associated with each department. The logged previous level of federal grants and payments is again statistically significant at either the one or five percent level in three regressions (RRFUNDAW, ARFUNDAN, and ARFUNDAV). Indeed, the significant point estimates (0.284, 0.356, and 0.310) are slightly larger on average than those from table 8. A state s previously evidenced ability to extract federal funds accounts for ARRA allocations in an economically meaningful way. The political variables included from our previous regression are rarely statistically significant and in all of those case economically small. In the ARFUNDAN regression, average Senate tenure has a statistically significant (at the five percent level) negative estimated effect. This is the only case reported in this paper where a political variable is associated with a 19

21 statistically significant effect of the wrong sign. (As well, the point estimate in this is very small: ; the standard deviation of AVSENATET is ) Of the newly included political variables, the coefficient estimate on SAPSUB is statistically significant and positive in two regressions (RRFUNDRE and ARFUNDPO). In these two cases, respectively, the point estimates are (five percent level) and (10 percent level). Senate appropriations subcommittee membership is de facto binary: a state either has one member or none. The average value of SAPSUB is 0.29 so about one third of states had a senator on a relevant appropriations subcommittee. Since the standard deviations of the logged department funds are approximately one for all departments (save for RRFUNRE where the standard deviation is 1.481), having a relevant appropriations subcommittee member in the Senate is associated with between 17 and 29 percent of a standard deviation increase in ARRA funds from that department. Authorization committee membership in the House of Representatives is statistically significant (at the one percent level) in one regression (ARFUNDAV) and authorization subcommittee membership is significant (at the five percent level) in one regression (ARFUNDAN). The HAUTH point estimate is very small though (0.039). The point estimate is associated with HAUTHS is so getting a representative onto a relevant authorization subcommittee is associated with about 15 percent of a standard deviation more in ARRA funds from that department. This is not economically negligible, but not very substantial either. Table 12 reports the results of re-estimating the panel regression, dropping all variables that are never statistically significant in tables 11A and 11B. There is little change in the results. SAPSUB is no longer significant in the ARFUNPO regression, but it was only significant at the 20

22 10 percent level previously and the p-value in the table 12 regression is The significant point estimates on the log of FEDAID rise noticeably to 0.349, 0.361, and Concluding Discussion The American Recovery and Reinvestment Act (ARRA) is a $787 billion dollar federal stimulus package, $500 billion of which is new federal government spending. According to recovery.gov the ARRA has three immediate goals : Create new jobs and save existing ones Spur economic activity and invest in long-term growth Foster unprecedented levels of accountability and transparency in government spending[.] The government s own wording implies that criteria in addition to those narrowly related to designing Keynesian countercyclical policy underlie ARRA fund allocations. Quite explicitly, while many of Recovery Act projects are focused more immediately on jumpstarting the economy, others, especially those involving infrastructure improvements, are expected to contribute to economic growth for many years. 9 Given this, it is perhaps not surprising that our Keynesian variables fail to account predominantly for spending patterns across US states. However, what is troubling is that such variables are either outright economically irrelevant in the determination of ARRA funding across states (unemployment rates) or they perversely positively correlated with ARRA funds (real per capita income and growth in state government revenues). Aside from the fact that

23 ARRA is in the aggregate a massive spending package (and this fact should not be discounted), the design of state-level funding patterns must be judged poorly-designed countercyclical policy. Political variables are often statistically significant determinants of state-level ARRA funds and the estimated effects are almost always positive. However, few of the specific effects appear to be economically large. There is some evidence that having a member on a relevant Senate authorization committee can lead to a meaningfully larger amount of ARRA funding for a state; there is similar evidence for membership on House authorization subcommittees. There is basically no evidence that the electoral vote maximization model accounts for ARRA spending patterns in an economically meaningful way. Are most robust finding is that a state s previous level of federal grants and payments to their state and local governments predicts a state s current share of ARRA funding. The elasticity of current ARRA funding to previous levels of federal aid is about one third. We interpret the previously level of federal aid a proxy for a state s extraction capital: accumulated political apparatus and skills aimed at capturing federal funds. To put this result as a punch-line: if a state was getting a lot of federal funding it is likely to be getting a larger share of the ARRA. This effect holds while controlling for population, per capita incomes and unemployment rates in a state. Our results suggest that the aggregate economic impact of the stimulus may fall short of the Keynesian ideal. Without the ARRA spending being targeted at areas with the most unemployed resources and/or the highest marginal propensities to consume, its impact will be significantly lessened. We find no evidence that either the administration s Council of Economic Advisors or the Congressional Budget Office consider the potential, systematic variation in multipliers across states. This may mean that policymakers are leaving money on the table in 22

24 terms of generating economic activity beyond the initial government expenditures. How much money was left on the table in the case of ARRA? Our results do not provide a direct answer. We believe that this is an avenue for fruitful future research. 23

25 References Anderson, G. M., and Tollison, R. D Congressional Influence and Patterns of New Deal Spending. Journal of Law & Economics 34 (1): Breusch, T. S. and Pagan, A. R Simple Tests for Heteroscedasticity and Random Coefficient Variation. Econometrica 47 (5): Brito, J. and de Rugy, V Stimulus Facts. Mercatus Center Working Paper, No (mercatus.org/sites/default/files/publication/wp0946.pdf). CBO Estimated Impact of the American Recovery and Reinvestment Act on Employment and Economic Output from January 2010 through March (May 2010) ( CEA The Economic Impact of the American Recovery and Reinvestment Act of 2009: Fourth Quarterly Report. (July 14, 2010) ( Coats, R. M., G. Karahan, and Tollison, R. D Terrorism and Pork-Barrel Spending. Public Choice 128 (1-2): de Rugy, V Stimulus Facts Period 2. Mercatus Center Working Paper, No (mercatus.org/sites/default/files/publication/wp1015_stimulus%20facts%202.pdf). Faith, R. L., D. R. Leavens, and R. D. Tollison Antitrust Pork Barrel. Journal of Law & Economics 25 (2): Garrett, T. A., and Sobel, R. S The Political Economy of FEMA Disaster Payments. Economic Inquiry 41 (3): Holcombe, Randall G. and Glenn R. Parker Committees in Legislatures: A Property Rights Perspective. Public Choice 70(1):

26 Kosnik, L. D Sources of Bureaucratic Delay: A Case Study of FERC Damn Relicensing. Journal of Law, Economics, & Organization 22 (1): Luengo-Prado, M. J. and Sørensen, B. E What Can Explain Excess Smoothness and Sensitivity of State-Level Consumption? Review of Economics and Statistics 90 (1): Moe, T. M An Assessment of the Positive Theory of Congressional Dominance. Legislative Studies Quarterly 12 (##): Moe, T. M The Positive Theory of Public Bureaucracy. in (D. C. Mueller, ed.) Perspectives on Public Choice: A Handbook. New York, NY: Cambridge University Press. Reading, D. C New Deal Activity and the States, Journal of Economic History 33 (##): US Bureau of Economic Analysis Economic Slowdown Widespread Among States in News Release (June) ( Weingast, B. R The Congressional-Bureaucratic System: A Principal Agent Perspective (with Applications to the SEC). Public Choice 44 (1); Weingast, B. R. and M. J. Moran Bureaucratic Discretion or Congressional Control? Regulatory Policy-Making by the Federal Trade Commission. Journal of Political Economy 91 (5): Weingast, B. R. and W. J. Marshall The Industrial Organization of Congress; or, Why Legislatures, Like Firms, Are Not Organized as Markets. Journal of Political Economy 96 (1):

27 White, H A Heteroscedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroscedasticity. Econometrica 48 (##): Wright, G The Economic Determinants of New Deal Spending. Review of Economics And Statistics 26 (##): Young, M., M. Reksulak, and W. F. Shughart II The Political Economy of the IRS. Economics and Politics 13 (2):

28 Figure 1 Stimulus effectiveness in the SRAS; AD model P SRAS stimulus stimulus AD AD * AD AD * Y Y * Y Y * Y P autonomous expenditures induced expenditures SRAS low MPC high MPC AD Y Y * Y ** AD AD * Y 27

29 Figure 2 Agency-reported Recovery and Reinvestment Act (log of) funds per capita (a) announced, (b) made available and (c) paid out versus (log of) 2008 real per capita GDP by state. (a) (b) (c) Note: funds are on the horizontal axis. 28

30 Figure 3 Agency-reported Recovery and Reinvestment Act (log of) funds per capita (a) announced, (b) made available and (c) paid out versus the January 2009 unemployment rate by state. (a) (b) (c) Note: funds are on the horizontal axis. 29

31 Table 1 Recovery and Reinvestment funds measures. Recipient-reported (Bil $) Agency-Reported (Bil $) Funds Awarded Funds Received Funds Announced Funds Made Available Funds Paid Out Total $177 $53 $324 $304 $178 Health and Human Services $15 $3 $56 $71 $56 Education $65 $30 $82 $75 $36 Transportation $37 $11 $34 $34 $10 Energy $28 $3 $20 $18 $10 Four Dept. % of Total Source: Data was collected when website stated its last update was April 7,

32 Table 2 Correlations of state-level Recovery and Reinvestment total funds measures. Recipient-Reported Agency-Reported Recipient-Reported Funds Awarded Funds Received Funds Announced Funds Made Available Funds Paid Out Funds Awarded Funds Received Agency-Reported Funds Announced Funds Made Available Funds Paid Out Source: Data was collected when website stated its last update was April 7,

33 Table 3 Summary statistics for state-level Recovery and Reinvestment total funds measures. Recipient-reported Agency-Reported (Bil $) (Bil $) Funds Awarded (RRFUNDAW) Funds Received (RRFUNDRE) Funds Announced (ARFUNDAN) Funds Made Available (ARFUNDAV) Funds Paid Out (ARFUNDPO) Mean $3.537 $1.072 $6.487 $6.075 $3.551 Maximum $ $7.610 $ $ $ Minimum $0.512 $0.121 $0.659 $0.617 $0.265 Standard Deviation $3.555 $1.148 $ $6.818 $4.270 Coefficients of Variation Standard Deviation (natural log) Source: Data was collected when website stated its last update was April 7, Coefficients of Variation are computed as standard deviations divided by means. 32

34 Table 4 Control variables: descriptions and sources. Variable Description Sources Keynesian Variables UNEMP January 2009, Seasonally-Adjusted State Unemployment Rate BLS ΔUNEMP Change in State Unemployment Rate from January 2008 to January 2009 BLS ΔREV State Tax Revenue Growth from 2007 to 2008 Census GDP Level of Real State Per Capita GDP in 2008 BEA ΔGDP Growth in Real State GDP from 2007 to 2008 BEA MPC Marginal Propensity to Consume in the State Luengo-Prado and Sørensen (2008) Demographic Controls POP State Population in 2008 Census SENIORS Percent of a State s Population over 65 Years of Age in 2008 Census BLACK Percent of a State s Population that was African American in 2008 Census LATINO Percent of a State s Population that was Latino in 2008 Census Political Variables FEDAID Federal Grants & Payments to State & Local Governments by State in 2008 Census AVHOUSET Average Tenure of State s US Representatives various AVSENATET Average Tenure of State s US Senators various HAC Number of Representatives on the US House Appropriations Committee various SAC Number of Senators on the US Senate Appropriations Committee various EIM Electoral Importance Measure (Garret and Sobel (2003)) electoral-vote.com OBAMA08 Dummy Variable (1 if Obama Won State in 2008; 0 otherwise) electoral-vote.com 33

35 Table 5 Summary statistics for Keynesian variables. UNEMP ΔUNEMP ΔREV ln(gdp) ΔGDP MPC Mean Maximum Minimum Standard Deviation Sources: see table 4. Table 6 Summary statistics for political variables. ln(fedaid) AVHOUSET AVSENATET HAC SAC OBAMA08 EIM Mean Maximum Minimum Standard Deviation Sources: see table 4. 34