Inter- and Intra-Chamber Differences and the Distribution of Policy Benefits

Inter- and Intra-Chamber Differences and the Distribution of Policy Benefits Thomas M. Carsey Department of Political Science Florida State University Tallahassee, FL 32306 tcarsey@garnet.acns.fsu.edu and Barry S. Rundquist Department of Political Science (M/C 276) University of Illinois at Chicago Chicago, IL 60607 barryr@uic.edu A paper prepared for delivery at the Annual Meeting of the Midwest Political Science Association, Chicago, IL, April 25, 2002

Inter- and Intra-Chamber Differences and the Distribution of Policy Benefits Introduction Distributive theory offers an explanation for the geographic distribution of governmental expenditures based on the re-election incentives of legislators. 1 Referred to by Krehbiel (1991) as the dominant theoretical approach to congressional politics, distributive theory focuses how the desire for re-election combines with the institutional rules of decision making within a legislature to produce policy outcomes. Tests of distributive theory have focused primarily on the committee structure within congress, though more recent efforts have incorporated political parties and ideology into the collection of institutional forces at work (see Rundquist and Carsey 2002 for a review). Most recent treatments view these internal institutions as at least partially endogenous to the preferences of legislators. In other words, re-election oriented legislators not only work through these institutions to produce their desired distribution of benefits they also work to shape the institutions themselves so as to allow them to pursue their re-election efforts more efficiently. What this work has in common, however, is a focus on institutional structures that operate primarily within a particular legislative chamber. 2 A recent book by Lee and Oppenheimer (1999) offers a different focus. Lee and Oppenheimer (subsequently abbreviated L&O) argue that the difference in how seats are apportioned in the U.S. House and the U.S. Senate creates a small-state advantage, at least on a 1 This research was supported in part by a grant from the National Science Foundation (SES#9809370). There are many people to acknowledge, but Lisa Schmit deserves special mention for her effort in compiling the data set used for this analysis. 2 In fact, most empirical treatments of distributive theory within the context of the United States only analyze a single chamber most typically the House of Representatives. 1

per capita basis, in the ultimate distribution of national policy benefits. This stems specifically from the equality in seats awarded to both big and small states in the U.S. Senate. Put in more general terms, L&O argue that an institutional factor that operates between chambers rather than within them plays a fundamental role in determining policy outcomes, including the geographic distribution of federal benefits. Furthermore, L&O suggest that the inter-chamber difference in apportionment may affect some kinds of policy, such as formula grants, more than others. In this paper, we take a first cut at examining the relationship between within (intra-) chamber distributive politics and the between (inter-) chamber. We begin with a discussion of how traditional distributive theory could be extended to incorporate L&O s argument. We then provide several empirical tests of the small-state advantage thesis. We first examine the direct affect of the small state advantage in the Senate. We then examine whether the intra-chamber factor thought by distributive politics scholars to at least potentially influence the distribution of policy expenditures specifically, committee representation is conditioned by the small-state advantage L&O posit. We conduct our analysis using annual county-level national expenditures across five policy areas from 1983 through 1996. The policy areas are: agriculture, crime, defense, health care, and transportation. We further break down our analysis of agriculture and health care spending into types of spending (grants, loans, direct payments, etc.) to evaluate L&O s claim that the small-state advantage varies as a function of how the benefits are awarded. We conclude with a general discussion of our findings with an eye toward future research. Small versus Large States in Distributive Politics Recently, Lee and Oppenheimer (1999) have argued that states with smaller populations 2

have a distinct advantage over larger states in garnering policy benefits from the national government. L&O argue that this advantage stems from how representation operates in the U.S. House of Representatives as compared to the U.S. Senate. Put quite simply, because representation is the House is based on population, we should expect House members to prefer formulas or other decision rules governing the distribution of federal policy benefits that are population-based. In contrast, representation in the Senate is based on states. As a result, Senators should prefer formulas or other decision rules that allocate funds based on states. In other words, as each member of Congress seeks his/her constituency s fair share of federal expenditures, House members define fair using population-based criteria while each Senator defines fair more in terms of equality across states. Assuming all members of Congress pursue such policy benefits, L&O imply that small states should be no better or worse off in per capita terms as a result of the representational structure of the House. However, L&O explicitly claim that small states should benefit, at least in terms of per capita benefits, over large states due to the representational structure of the Senate. Thus, smaller states should receive higher levels of federal expenditures per capita after controlling for internal Congressional factors like partisanship and committee effects, as well as external factors like characteristics of the constituency. L&O argue that, while such pressures should exist for all policy making, formula-based expenditures are most likely to evidence the small-state advantage. They argue that formulabased programs are less subject to short-term pressures resulting from changing committee control in Congress. They are also less subject to manipulation or influence by government agencies or others during the implementation stage because there is less bureaucratic discretion 3

involved with formula-based expenditures. Thus, formula-based expenditures should embody the long-term structural features of representation in Congress, and the most prominent of these is the state-based representation in the Senate compared to the population-based representation in the House. Policy expenditures that allow for more discretion among bureaucrats and those policies subject to annual (or frequent) re-authorization should be more responsive to the shortterm forces of party control and committee representation. Thus, L&O conclude that the kind of spending programs that have been the subject of most studies of distributive politics are exactly the kinds of programs that should evidence either no or a significantly attenuated small-state bias. Intra- and Inter-chamber Institutions and Distributive Politics The argument put forth by L&O is compelling. However, we suggest that, in its current form, L&O s presentation draws an unnecessary line between distributive (discretionary) and non-distributive (formula-based) spending programs. The result is a somewhat false dichotomy between a focus on short-term institutional forces within each chamber in Congress (such as committee representation) and long-term institutional forces that structure representation between chambers (state-based representation in the Senate). The implication is that some policies (i.e. non-distributive policies) are structured by this basic inter-chamber difference in representation and are not subject to intra-chamber institutional features such as committee representation while, for other policies, the situation is reversed. Furthermore, the assumption is that the nature of intra-chamber representation (i.e. committee representation) is not affected by this large interchamber institutional difference. Certainly the difference between population-based and statebased representation embodied in the two chambers of Congress should be considered. Much of 4

the work on congressional policy making has not only ignored this difference, it has often ignored the Senate entirely. However, the state-based versus population-based nature of representation in these two chambers is just another feature of the institutional structure of policy making that re-election oriented members of Congress must consider when making policy. If the focus of scholars moves away from the nominal type of policy as a means of classification toward a focus on the process of policy making, a general theory of the process can be developed that removes the line drawn by L&O. Specifically, we suggest that a theory of distributive politics can (and should) be extended to incorporate the inter-chamber institutional differences that structure representation and, thus, influence policy making. Elsewhere, we define distributive politics as a process by which the institutional structure of Congress interacts with the re-election incentives of members of Congress to structure the geographic distribution of government benefits (Rundquist and Carsey 2002). Rather than serving as a theory for specific types of policy (a la Lowi 1964; Wilson 1973), ours is a theory of the policy making process. Hence the use of the term distributive politics as opposed to distributive policy. Admittedly, while our previous work has considered both the House and Senate, we have not previously taken into account the role of population-based versus state-based representation. But we view L&O s observation as highlighting another important institutional feature of distributive politics that should be incorporated into our theory and models. In this paper, we first consider the fundamental proposition raised by L&O: that smaller states should be advantaged in the per capita distribution of federal expenditures. We do so by examining the impact of state population on the geographic distribution of per capita federal expenditures in five substantive policy areas: agriculture, crime, defense, health, and 5

transportation. Next, we break agriculture and health spending down into five separate categories based on the mechanisms through which expenditures are made (i.e. direct payments, grants, loans, etc.). This analysis is performed by adding state population to a model of distributive politics that controls for committee representation, partisanship, the MC ideology, and various measures of the constituency. This is the first step at incorporating the hypothesized small-state advantage into a larger model of distributive politics. The analysis just described will allow us to determine whether there is a direct effect of state size on policy benefits. We take the analysis one step further, however, by examining whether the differential nature of representation in the two chambers as it relates to population structures the influence that committee representation has the geographic distribution of expenditures. If small-state senators are, in general, in a position to structure policy outcomes so as to benefit their states disproportionately in terms of per capita expenditures, we suspect that it might be the case that their ability to do so would be further enhanced by serving on those committees in the Senate with jurisdiction over the policy area under consideration. In other words, the small-state advantage that L&O find in many substantive policy areas may be further exaggerated by relevant committee representation. Similarly, the importance of population-based representation in the House may combine with representation on a relevant House committee to the benefit of more populous locations. Thus, the small-state advantage that emerges in many policy areas may stem in part or in whole from the representation of small states on the relevant Senate committees. In other words, the competing institutional structures of inter-chamber representation and intra-chamber committee representation that L&O keep separate may instead interact with each other to shape the geographic distribution of policy benefits. 6

Model and Research Design. L&O s analysis of programmatic spending was primarily concerned with the effect of the mal-apportioned Senate on the distribution of per capita expenditures. To extend their analysis to consideration of intra-chamber structures requires a model that incorporates both intra- and interchamber politics. Here we employ a statistical model that represents a modification of a portion of a model of distributive politics that we have employed in several studies elsewhere (see Rundquist and Carsey 2002 for the most complete treatment). Specifically, we model the level of funding received in a particular county as a function of: previous levels of spending, whether the county was represented on a House and/or Senate committee that substantive jurisdiction over the policy area in question, the partisan make-up of the House and Senate delegations that represent the county in Congress, and the ideological orientations of those same delegations. We also control for constituency characteristics related to the policy problem under consideration (Carsey and Rundquist, 1999; Adler and Lapinski 1997). This model further separates committee representation into representation by either fo the two parties. Because the model includes a lagged value of the dependent variable, the other variables in the model are really predicting change in the level of spending received in a county from one year to the next. To this we first add a measure of state population to capture the small-state effect predicted by L&O. If there is a small-state advantage, we would expect the coefficient estimate operating on the state population variable to be negative and statistically significant. This would indicate that as state population increased, counties in those states would see smaller increases in expenditures or even declines in expenditures per capita. 7

In subsequent analyses, we produce a series of multiplicative interaction terms between state population and the various committee representation variables described above. This allows us to examine whether the affects of committee representation (split by party) on the distribution of policy benefits operates differently for smaller versus larger states. This part of the analysis is somewhat more exploratory in nature as our expectations regarding what we should find are somewhat mixed. The initial expectation is that smaller states should be particularly advantaged by representation on the relevant committees in the Senate, leading us to expect negative and significant coefficients on the Senate representation state population interaction terms. This would suggest that committee representation becomes steadily less valuable to Senators from larger states. This expectation stems from assuming that small states Senators could capitalized on their already aggregate advantage reported by L&O even further if they held the relevant institutional positions within the Senate. Our expectations regarding the House are less clear. On the one hand, the preference for population based decision rules in the House that L&O posit might suggest that representation on a House committee would be increasingly valuable for representatives from populous states. This implies statistically significant positive coefficient estimates on the House representation state population interaction terms. However, L&O never claim that more populous states are disproportionately advantaged by the nature of House representation; only that smaller states are disproportionately advantaged in the Senate. In other words, one might just as readily expect no particular additional payoff from committee representation for more populous states, which implies statistically insignificant coefficient estimates for the House representation state population interaction terms. 8

These expectations are further clouded by the difference between the two chambers in terms of the relative importance of committees in structuring policy choices. Briefly speaking, the absence of a Rules committee, the greater openness of floor debate, the greater informality, and smaller size in the Senate combine to make committees relatively less important in shaping the final outcome of policy decisions in the Senate as compared to the House. Thus, small states have the incentive, along with the structural advantage in representation, in the Senate to pursue their interests, but it may be that committee membership is less central to Senators in their pursuit of their goals as compared to House members. Finally, it is worth noting that we are modeling expenditures to come out of Congress and not floor votes or budget proposals as adopted by each chamber. Thus, the final product is more than just the result of the intrachamber processes taking place within the House and Senate it is also the result both chambers either adjusting to what they anticipate the other doing or the result of conference committee activity (Rundquist and Strom 1977). As mentioned above, the unit of analysis in this study is the U.S. county. Heitshusen (1991), Anton et al. (1980), Crump and Archer (1993), and Rundquist, Carsey, Schmit, and Rhee (1997) have demonstrated the utility of county-level analysis, with Anton et al. (1980) noting that, at the county level, Federal program outlays are closely associated with need in programs designed to address those needs (p. 78). Counties are useful because they are components of both most congressional districts and all states, and because they allow for more narrow specifications of local need and socioeconomic characteristics than do either congressional districts or states. County borders also remain stable, meaning that all of the problems introduced in district-level analyses by the decennial redistricting can be avoided. 9

Data on programmatic expenditures are from the Census Bureau s Consolodated Federal Funds Report (CFFR). Data on the various measures of constituency characteristics included as substantive control variables come from a variety of sources. Those for agriculture, defense, and transportation are taken from the Regional Economic Information Systems (REIS) data. Data for crime were taken from the U.S. Department of Justice Federal Bureau of Investigation s Uniform Crime Reporting Program Data, County Level. Data on health care came from the Area Resource File: 1940-94" produced by the U.S. Department of Health and Human Services Health Resources and Services Administration, Bureau of Health Professions. Member ideology is measured using annual Conservative Coalition scores. The specific constituency characteristic measures are reported in the note to Table 1. Representation on a relevant committee is measured as a simple dummy variable for most counties, coded 1 if the county is represented and 0 otherwise. However, there are several large urban counties that encompass multiple congressional districts. In these instances, the committee representation variables were computed as proportions. So, if a county included 10 House districts, and five of those representatives served on a relevant committee, the committee representation variable was coded as.5. The specific committees in question are listed in the note to Table 1. State population is measured in millions of persons. Each dependent variable is measured in dollars per capita, adjusted for inflation. Findings Table 1 presents the results of our initial models for each of the five substantive policy areas we consider. It shows some support for House committee representation effects on the 10

distribution of per capita benefits in Agriculture, Defense, and Health for counties represented by House committee Democrats and for Agriculture, Crime, and Defense for counties represented by House committee Republicans. Looking at Agriculture, for example, being represented on the House Agriculture committee by either a Democrat or a Republican translates into an increase in county-level per capita agricultural expenditures of between $66.30 to $72.50. In contrast, House committee representation effects are not evident for other policy areas. Looking at Senate committee representation, we see roughly the same pattern of partial support for an affect in some policy areas, but not others. Again, the affect is strong and statistically significant for committee representation by either a Democrat or a Republican on the Senate Agriculture committee, though here we see the affect is more than three times larger for Democrats than it is for Republicans. Oddly, we find statistically significant negative effects for Senate committee representation by a Republican for health care expenditures. Turning to the effect of state population, Table 1 reveals that counties located within relatively smaller states received significantly larger relative changes in per capita expenditures in four out of five of our policy areas than did counties located in more populous states. The one policy area where this finding is not supported crime spending. For example, the coefficient in Table 1 operation on state population in the agriculture spending model can be interpreted as an increase in state population of one million leading to a decrease in change in per capita agriculture spending in a county of $2.99 per capita. These findings support L&O s claim of a small state advantage. Next, we expand the baseline models presented in Table 1 to include the set of committee representation state population interaction terms as described above. The results of this 11

analysis are reported in Table 2. 3 Table 2 shows nine of the twelve committee representation state population interaction terms produce positive and significant (p <.1, two-tailed) coefficient estimates. This means that counties located in more populous states receive a greater benefit from House committee representation than to counties located in less populous states. Furthermore, comparing the results from Table 1 with those of Table 2, we find that the positive House committee representation direct effects uncovered in Table 1 for agriculture, defense, and crime are all substantially reduced in Table 2. This suggests that the results uncovered in Table 1 were driven in large part by House committee representation among larger states. The findings reported in Table 2 for the Senate committee representation state population interaction terms are more varied. In four instances, all within agriculture and transportation policy, we find the expected negative coefficient operating on these interaction terms. In these instances, it appears that small states are benefitting disproportionately from representation on the relevant Senate committees. However, in eight instances, the coefficients operating on these interaction terms in the Senate are positive and statistically significant. In these cases, it appears that representation on a relevant Senate committee pays off more for more populous states. So, while smaller states appear to be using committee representation to their increased advantage in Agriculture and Transportation spending, committee representation 3 Any time a model includes multiplicative interaction terms, the potential for problems stemming from multi-colinearity arise. Of course, multi-colinearity represents an efficiency problem, not a bias problem, so any statistically significant estimates we uncover for these variables has had to overcome a higher hurdle. Never-the-less, as an initial check, we examined the correlations between all possible bi-variate combinations of state population and these interaction terms. The largest we found across all the models was.62, and most were below.2. This approach does not constitute an exhaustive search for multi-colinearity, but it its results and our large sample sizes, we are comfortable concluding that mulit-colinearity does not present a substantial problem here. 12

appears to be countering the small-state advantage in the Senate in the areas of Crime, Defense, and Health care. L&O argue that the small state advantage is most likely to show up in spending programs driven by stable formulas rather than among discretionary funds. More generally, the implication of this argument is that the method of delivering policy benefits may condition how sensitive those benefits are to the small-state advantage in the Senate. To begin to explore this hypothesis, we conduct separate analyses of agriculture and health care spending, this time dividing spending into the various mechanisms by which funds are distributed. We begin by repeating the baseline model presented in Table 1 for the various categories of agriculture spending, the results of which we present in Table 3. The first column of Table 3 simply repeats the analysis of total agriculture spending first reported in table 1. The remaining columns divide total agriculture spending into spending in the form of direct loans, direct payments, grants, and insurance. Without examining each coefficient in detail, several things can be gleaned from Table 3. First, the positive effect of House committee representation by either a Democrat or a Republican that was uncovered in total agriculture spending is found in three of the four separate expenditure types the one exception is in the area of grants. The same can be said for the pattern of results related to Senate committee representation. Our findings also follow a similar pattern regarding the direct effect of state population size on different types of agriculture spending. The negative coefficient for total spending and for all of the areas of spending except grants indicates that counties located in small states enjoy a significant advantage in receiving per capita spending increases. Next, we extend our analysis of agriculture spending by including the full set of 13

committee representation state population interaction terms for each category of spending. The results are presented in Table 4. The direct effects of House committee representation uncovered in Table 3 are all diminished, some to insignificance, by the inclusion of these interaction terms. The interaction terms for both Democratic and Republican committee representation are positive and significant only for expenditures made for insurance. Only the republican interaction term is significant and positive for expenditures in the form of direct loans, while it is the Democratic interaction term that is significant for direct payments. Neither is significant for grants. The effects of Senate representation and of the interaction terms reported in Table 4 follow a pattern similar to those in the House. Finally, the direct effect of state population on changes in agriculture benefits in each of these areas is reduced compared to the results shown in Table 3. However, they remain statistically significant in Table 4 in every instance in which they were so in Table 3, and the effect actually achieves significance in grant spending in Table 4 where had not in Table 3. Finally, in Tables 5 and 6, we report on a similar analysis for various types of health care spending. Looking first at Table 5, we see that the only positive and significant coefficient estimate for representation on a relevant House committee emerges for spending on SSDI, where representation by a Democrat on House Ways and Means results in about a $17 per capita increase in the benefits received in a county. Regarding Senate representation, the results in Table 5 suggest a strong pattern of a negative affect resulting from representation by a Republican on the Senate Labor committee (for Direct Loans, Medicare, and SSDI) or the Senate Finance committee (for all areas except Guaranteed Loans). There is no corresponding affect for Senate committee representation by a 14

Democrat. Finally, the small-state advantage emerges in every area of health care except Guaranteed loans. Once again we find strong and consistent support for the L&O hypothesis that small states benefit directly in terms of per capita benefits. Table 6 reports the models for types of health care spending that include the committee representation state population interaction terms. Twelve of the 20 interaction terms for House committee representation are positive and significant, suggesting that if any representatives are better able to capitalize on committee representation in the House, it appears to be those from larger states. For the Senate, eleven of the twenty interaction terms produce significant coefficient estimates however, they are all positive. In the area of health care, it appears that the small-state advantage is offset somewhat by larger states being represented on the relevant Senate committees. This runs counter to our initial expectations. The direct effects estimated for state population remain negative and significant for all areas except Guaranteed Loans. The absolute values of the coefficients operating directly on state population all increased from Table 5 to Table 6, which is to be expected given that all of the positive effects for state population when larger states are represented on the relevant committees in the House or the Senate were all masked in the analysis presented in Table 5. Conclusion and Discussion We introduced this paper as one that seeks to extend distributive theory beyond the factors internal to the House and Senate to incorporate important institutional differences between the two chambers. As a first, and admittedly rough, step, we focused our attention on 15

the small-state advantage resulting from state-based rather than population-based representation in the Senate. Our argument is that this is just one of the many institutional structures that members of congress must work with and/or around in pursuit of re-election. To this we added a focus on the traditional distributive theory hypothesis about committee representation significantly influence the geographic distribution of policy benefits. Our findings show most clearly that the L&O hypothesis receives consistent and substantial support. In nearly every instance in our analysis, we uncover a statistically significant advantage for smaller states relative to larger states in changes in per capita spending. In one regard, this is strong support for the small-state hypothesis. On the other hand, we seem to find broader support for it than do L&O themselves. They make an argument that it will be most evident in formula-based programs, but we find evidence for it almost across the board. In terms of how the small-state hypothesis interacts with committee representation, we found the most consistent evidence that counties with representatives on the relevant House committees from larger states are often able to outperform their small-state colleagues. Our initial expectation regarding these findings was one of either no effect or a positive one favoring committee members from larger states. Population-based representation in the House appears to be further reinforced by the relative advantage in the geographic distribution of national government policy benefits enjoyed by counties located in larger states stemming from committee representation. Our results for the interactive models in the Senate were less clear. We initially felt more strongly about the hypothesis that the small-state advantage stemming from state representation in the Senate would manifest itself in an even larger small-state advantage for those places with 16

representatives sitting on the relevant committees in the Senate. This hypothesis was not widely confirmed in fact, we were about equally likely to see a significant advantage stemming from Senate committee representation for larger states as we were to uncover such an advantage for smaller states. Of course mixed findings often suggest mixed interpretations. Here are a few speculations. First, the small state advantage may simply manifest itself differently from one substantive policy area to another, based in some way on the substantive nature of the policy. For example, we see the small-state advantage being further enhanced by Senate committee representation in the area of agriculture spending (see Table 2). In contrast, larger states tend to benefit more from Senate committee representation in the area of Crime policy. This may be because agriculture policy is of central concern to many less populous states while crime policy is primarily of concern to those from large urban centers, which obviously are located in more populous states. Thus, in areas of concern to less or more populous states, we see exactly what a theory of re-election oriented Senators would predict rural areas using representation on the Senate agriculture committee to their advantage while urban (populous) areas use the Senate committee representation process to their advantage. This argument based on the substantive type of policy may be part of the issue, but we are less that fully satisfied. First, this would require an identification of the importance of various policies to more urban and more rural areas. Crime and agriculture present relatively easy cases to explain, but what about defense and health care? Second, though limited, we do attempt to control for constituency need in these analyses. These controls are meant to capture the need or demand for policy expenditures in an area that stem from constituency characteristics. For example, the need or demand for 17

agricultural spending, measured as per capita earnings from agriculture in a county, should already capture the substantive differences in policy types. Another hypothesis is that Senators are aware of the built-in small state advantage and Senators from larger states must pick and choose where they will focus their efforts at minimizing this advantage. In particular, large-state Senators may implicitly or explicitly cooperate with their counter-parts in the House when spending in a policy area is relatively large. Large states come out ahead if they sacrifice agricultural spending to smaller states but are able to fight and win on big-ticket items like health care spending and military procurement. In other words, the mixed findings here may stem from the strategic calculations of small- and large-state senators to decide when to fight and when not to fight. Regardless of how the interaction terms behaved, the direct effect of state population remained negative and significant in nearly every instance. This tells us that the small state advantage is not simply a spurious finding driven entirely by how the committee representation process works in the House compared to the Senate. Though population size is a crude proxy for capturing a key institutional difference between the two chambers, it appears to be an institutional difference that goes beyond just influencing the internal workings of the committee system within each chamber. So were we successful in integrating the small-state benefit hypothesis into a general distributive theory of legislative policy making? Ultimately, that judgment is left to the reader. However, we feel we have at least made progress. We feel relatively confident that the statistical model we developed for previous studies of distributive theory has been shown here to be capable of incorporating both the simple direct effect hypothesis as well as our hypotheses about 18

the conditional affects of state population and committee representation. Where we have made less progress is on the predictions we would expect from this analysis and the theoretical implications of our findings. We suggest several post hoc ideas, but the need for further theoretical development may be the clearest conclusion that we reach. 19

References Adler, Scott E. and John Lapinski. 1997. Demand-side Theory and Congressional Committee Composition: A Constituency Characteristics Approach. American Journal of Political Science. 41:895-918. Anton, Thomas J., Jerry P. Hawley, and Kevin L. Kramer. 1980. Moving Money: An Empirical Analysis of Federal Expenditure Patterns. Cambridge, MA: Oelgeschlager, Gunn, and Hain. Carsey, Thomas M. and Barry S. Rundquist. 1999. Targeting Distributive Benefits: Comparing Health, Agriculture, Transportation, Crime, and Defense Spending. Paper presented at the annual meeting of the American Political Science Association, Atlanta, September 2-5. Crump, Jeffrey and Clark Archer. 1993. "Spatial and Temporal Variability in the Geography of American Defense Outlays". Political Geography. Vol. 12, No.1, pp 38-63. Heitshusen, Valerie. 1991. Do Committee Members Get a Bigger Piece of the Pie? The Distribution of Program Expenditures to Members of Congress in the 1980s. A paper presented at the Annual Meeting of the American Political Science Association, Chicago, IL. Krehbiel, Keith. 1991. Information and Legislative Organization. Ann Arbor: University of Michigan Press. Lee, Frances E. and Bruce I. Oppenheimer. 1999. Sizing Up the Senate: The Unequal Consequences of Equal Representation. Chicago: The University of Chicago Press. Lowi, Theodore. 1964. American Business, Public Policy, Case Studies, and Political Theory, World Politics, XVI (July), 677-715. Rundquist, Barry S. and Thomas M. Carsey. 2002. Congress and Defense Spending: The Distributive Politics of Military Procurement. Norman, OK: University of Oklahoma Press. Rundquist, Barry, Thomas M. Carsey, Lisa Schmit, and Jungho Rhee. 1997. Congressional Committees and Interest Representation in Defense Contracting. a paper presented at the annual meeting of the Midwest Political Science Association. Chicago, IL. Strom, Gerald S. and Barry S. Rundquist. 1977. A Revised Theory of Winning House-Senate Conferences. American Political Science Review. 71(2): 448-53. Wilson, James Q. 1973. Political Organizations. New York: Basic Books. 20

Table 1: Factors that influence county-level per capita expenditures in five policy areas Agriculture a Crime b Defense c Health d Transportation e Benefits t-1.78 (.001) 1.13 (.001).932 (.01) 1.01 (.001).553 (.001) HR-Com-Rep-Dem t-1 72.5 (.001) -.84 (.346) 58.6 (.063) -9.2 (.916) 5.9 (.503) HR-Com-Rep-GOP t-1 66.3 (.001) 2.93 (.003) 70.2 (.061) 95.3 (.393) 2.5 (.783) HR-Com-Rep-Dem t-1 174.2 (.102) HR-Com-Rep-GOP t-1-149.9 (.261) SEN-Com-Rep-Dem t-1 99.1 (.001).49 (.255) -15.4 (.352) 18.9 (.799) 21.0 (.122) SEN-Com-Rep-GOP t-1 30.8 (.001).03 (.949) 42.4 (.012-129.1 (.026) 11.4 (.303) SEN-Com-Rep-Dem t-1 50.5 (.307) 55.8(.001) SEN-Com-Rep-GOP t-1-175.5 (.002).17 (.978) Dem HR delegation t-1-54.4 (.001).005 (.990).79 (.973) -18.7 (.018) -7.6 (.153) HR delegation ideology t-1 -.34 (.027) -.013 (.076) -1.12 (.014) -4.5 (.001) -.48 (.001) Dem Senate delegation t-1-15.6 (.004) -.033 (.891) 48.4 (.003) -45.2 (.242) -4.45 (.195) Senate delegation ideology t-1-1.1 (.001) -.011 (.152).61 (.195) -1.76 (.139) -.19 (.086) State Pop t (millions) -2.99 (.001) -.0106 (.68) -5.87 (.001) -11.7 (.002) -2.25 (.001) Constituency factor t-1 209.5 (.001) -22.98 (.001) -14.5 (.203) 284,639 (.001) 514.6 (.001) Constituency factor t-1 126.7 (.001) 37.2 (.003) -8054 (.016) N 40,334 40,328 34,973 40,251 40,345 Adjusted R 2.77.57.52.51.25 Note: Cell entries are unstandardized regression coefficients, two-tailed significance levels in parentheses. Models also include year dummy variables. The relevant committees and constituency characteristic variables are, in order: a House Agriculture Committee, Senate Agriculture Committee, per capita earning from agriculture. b House Judiciary Committee, Senate Judicial Committee, per capita offenses, per capita police employment. c House Armed Services Committee, Senate Defense Committee, economic capacity in Gun Belt states, per capita income. d House Commerce Committee, House Ways and Means Committee, Senate Labor Committee, Senate Finance Committee, doctors per capita, hospital beds per capita. e House Public Works Committee, Senate Banking Committee, Senate Public Works Committee, per capita income from highway construction. 21

Table 2: Interactive ffects of state population and committee representation on county-level per capita expenditures in five policy areas Agriculture Crime Defense Health Transportation HR-Com-Rep-Dem t-1 26.8 (.028) -1.4 (.344) 1.57 (.976) -512.8 (.001) -50.3 (.001) State Pop Int t-1 8.18 (.01).139 (.501) 8.05 (.151) 87.5 (.001) 8.33 (.001) HR-Com-Rep-GOP t-1 21.6 (.110) -3.42 (.081) -82.2 (.189) -141.2 (.420) -34.8 (.023) State Pop Int t-1 8.91 (.001).633 (.001) 23.1 (.003) 33.2 (.141) 6.27 (.003) HR-Com-Rep-Dem t-1-464.8 (.001) State Pop Int t-1 128.2 (.001) HR-Com-Rep-GOP t-1-346.5 (.063) State Pop Int t-1 36.7 (.097) SEN-Com-Rep-Dem t-1 129.3 (.001) -3.46 (.001) -60.8 (.015) -598.8 (.001) 35.3 (.101) State Pop Int t-1-5.35 (.040).54 (.01) 12.6 (.003) 82.3 (.001) -3.80 (.082) SEN-Com-Rep-GOP t-1 70.4 (.001) -.94 (.135) -9.3 (.689) -313.6 (.003) 13.1 (.400) State Pop Int t-1-6.46 (.001).229 (.039) 8.61 (.001) 48.8 (.032) -1.21 (.520) SEN-Com-Rep-Dem t-1-569.3 (.001) -7.7 (.391) State Pop Int t-1 87.4 (.001) 7.29 (.001) SEN-Com-Rep-GOP t-1-274.7 (.002) 275 (.015) State Pop Int t-1 21.8 (.198) -8.03 (.001) State Population t -2.81 (.001) -.0851 (.001) -13.6 (.001) -60.8 (.001) -4.33 (.001) N 40,334 40,328 34,973 40,251 40,345 Adjusted R 2.77.57.52.51.26 Note: Cell entries are unstandardized regression coefficients, two-tailed significance levels in parentheses. Models also include all additional independent variables included in the models in Table 1. 22

Table 3: Factors that influence county-level per capita expenditures in different types of Agricultural spending Total Direct Loans Direct Payments Grants Insurance Benefits t-1.78 (.001).69 (.001).85 (.001).90 (.001).61 (.001) HR-Com-Rep-Dem t-1 72.5 (.001) 37.5 (001) 10.6 (.011) -.60 (.009) 12.5 (.001) HR-Com-Rep-GOP t-1 66.3 (.001) 12.1 (.015) 23.4 (.001) -.23 (.392) 10.1 (.001) SEN-Com-Rep-Dem t-1 99.1 (.001) 40.7(.001) 13.0 (.001) -.17 (.414) 7.4 (.001) SEN-Com-Rep-GOP t-1 30.8 (.001) 17.1 (.001) 20.3 (.001) -.20 (.288) 3.8 (.001) Dem HR delegation t-1-54.4 (.001) -32.3 (.001) -18.8 (.001).05 (.801) -3.9 (.001) HR delegation ideology t- 1 -.345 (.027) -.44 (.001) -.29 (.001) -.009 (.024) -.017 (.279) Dem Senate delegation t-1-15.6 (.004) -1.18 (.661).80 (.755).09 (.518) 1.48 (.008) Senate delegation ideology t-1-1.1 (.001) -.18 (.025) -.07 (.387).002 (.577).025 (.135) State Population t -2.99 (.001) -.910 (.001) -1.55 (.001) -.0218 (.143) -.445 (.001) Constituency factor t-1 209.5 (.001) 16.2 (.256) 34.4 (.011) -1.64 (.032) 2.95 (.315) N 40,334 40,335 40,347 40,366 40,352 Adjusted R 2.77.61.73.58.52 Note: Cell entries are unstandardized regression coefficients, two-tailed significance levels in parentheses. Models also include year dummy variables. 23

Table 4: Interactive effects of state population and committee representation on county-level per capita expenditures in different types of agriculture spending Total Direct Loans Direct Payments Grants Insurance HR-Com-Rep-Dem t-1 26.8 (.028) 35.8 (.001).62 (.914) -.66 (.037) 10.2 (.001) State Pop Int t-1 8.18 (.001).4 (.529) 1.98 (.003).01 (.779).413 (.004) HR-Com-Rep-GOP t-1 21.6 (.110) -12.5 (.059) 26.5 (.001) -.29 (.406) 6.2 (.001) State Pop Int t-1 8.91 (.001) 5.21 (.001) -.5 (.522).02 (.738).743 (.001) SEN-Com-Rep-Dem t-1 129.3 (.001) 35.6 (.001) 20.1 (.001) -.46 (.149) 14.7 (.001) State Pop Int t-1-5.35 (.040) 2.58 (.042) -.7 (.545).07 (.269) -1.74 (.001) SEN-Com-Rep-GOP t-1 70.4 (.001) 30.3 (.001) 41.3 (.001) -.44 (.112) 9.6 (.001) State Pop Int t-1-6.46 (.001) -2.24 (.001) -3.36 (.001).04 (.253) -.926 (.001) State Population t -2.81 (.001) -.861 (.022) -.669 (.061) -.04 (.041) -.195 (.012) N 40,334 40,335 40,347 40,366 40,352 Adjusted R 2.77.61.73.58.52 Note: Cell entries are unstandardized regression coefficients, two-tailed significance levels in parentheses. Models also include all additional independent variables included in the models in Table 3. 24

Table 5: Factors that influence county-level per capita expenditures in different types of health policy spending Total Direct Loans Medicare SSDI Guaranteed Loans Grants Benefits t-1 1.01 (.001).04 (.001).98 (.001) 1.03 (.001) 2.11 (.001) 1.05 (.001) HR-Com-Rep-Dem t-1-9.2 (.916) -.004 (.649) -4.5 (.912) -1.4 (.835) -.023 (.816) -3.7 (.929) HR-Com-Rep-GOP t-1 95.3 (.393).002 (.833) 52.0 (.323) 11.2 (.190) -.033 (.791) 31.4 (.553) HR-Com-Rep-Dem t-1 174.2 (.102) HR-Com-Rep-GOP t-1-149.9 (.261).00001 (.999) 72.8 (.147) 17.1 (.035) -.006 (.961) 87.3 (.084) -.012 (.330) -59.2 (.345) -19.3 (.058) -.059 (.693) -72.0 (.254) SEN-Com-Rep-Dem t-1 18.9 (.779).004 (.478) 28.2 (.373) -1.1 (.829).104 (.170) -5.6 (.860) SEN-Com-Rep-GOP t-1-129 (.026) -.012 (.020) -78.5 (.004) -10.7 (.016) -.017 (.794) -38.0 (.167) SEN-Com-Rep-Dem t-1 50.5 (.307).006 (.156) 22.7 (.328) 2.96 (.434) -.009 (.874) 28.1 (.230) SEN-Com-Rep-GOP t-1-175.5 (.002) -.013 (.011) -102.3 (.001) -16.6 (.001) -.023 (.722) -53.3 (.052) Dem HR delegation t-1-118.7 (.018) -.007 (.132) -63.5 (.007) -12.3 (.001) -.037 (.511) -44.7 (.059) HR delegation ideology t-1-4.5 (.001) -.0003 (.006) -2.3 (.001) -.41 (.001) -.001 (.397) -1.76 (.001) Dem Senate delegation t-1-45.2 (.242) Senate delegation -1.76 ideology t-1 (.139) State Population t -11.7 (.002) -.004 (.211) -29.7 (.103) -2.46 (.405).013 (.768) -14.2 (.436) -.00006 (.554) -1.05 (.062) -.16 (.076.005 (.683) -.57 (.311) -.0007 (.028) -6.05 (.001) -1.15 (.001) -.002 (.656) -4.48 (.012) Constituency factor t-1 284,639 (.001) 18.8 (.001) 132,018 (.001) 27,596 (.001) 72.1 (.002) 127,480 (.001) Constituency factor t-1-8054 (.016) -.16 (.312) -2193 (.176) -899 (.001) -1.33 (.485) -7037 (.001) N 40,251 40,375 40,334 40,251 40,375 40,334 Adjusted R 2.51.01.50.54.84.52 Note: Cell entries are unstandardized regression coefficients, two-tailed significance levels in parentheses. Models also include year dummy variables. 25

Table 6: Interactive effects of state population and committee representation on county-level per capita expenditures in different types of health policy spending Total Direct Loans Medicare SSDI Guaranteed Loans Grants HR-Com-Rep-Dem t-1-512.8 (.001) -.02 (.101) -246 (.01) -44.4 (.001) -.027 (.862) -222 (.001) State Pop Int t-1 87.5 (.001).003 (.094) 42.1 (.001) 7.45 (.001).001 (.959) 38.1 (.001) HR-Com-Rep-GOP t-1-141 (.420) -.006 (.694) -75 (.367) -17 (.205) -.003 (.989) -51.5 (.535) State Pop Int t-1 33.2 (.141).001 (.604) 17.5 (.099 4.1 (.017) -.004 (.888) 11.6 (.277) HR-Com-Rep-Dem t-1-464.8 (.001) -.017 (.198) -174 (.010) -42 (.001) -.007 (.968) -240 (.001) State Pop Int t-1 128.2 (.001).004 (.073) 47.6 (.001) 12.0 (.001) -.001 (.963) 67.5 (.001) HR-Com-Rep-GOP t-1-347 (.063) -.009 (.584) -160 (.070) -40 (.005) -.07 (.756) -147 (.096) State Pop Int t-1 36.7 (.097) -.00008 (.968) 19.1 (.067) 3.77 (.025) -.00057 (.963) 13.8 (.189) SEN-Com-Rep-Dem t-1-599 (.001) -.025 (.017) -308 (.001) -57 (.001).208 (.116) -234 (.001) State Pop Int t-1 82.3 (.001).004 (.001) 46.0 (.001) 7.36 (.001) -.02 (.329) 29.3 (.001) SEN-Com-Rep-GOP t-1-314 (.003) -.016 (.089) -180 (.001) -26 (.001) -.008 (.944) -107 (.030) State Pop Int t-1 48.8 (.032).001 (.641) 26.9 (.012) 4.07 (.019) -.002 (.953) 18.2 (.091) SEN-Com-Rep-Dem t-1-569 (.001).0006 (.934) -274 (.001) -53.5 (.001) -.05 (.559).239 (.001) State Pop Int t-1 87.4 (.001).001 (.144) 42.6 (.001) 7.94 (.001).004 (.609) 37.0 (.001) SEN-Com-Rep-GOP t-1-275 (.002) -.016 (.041) -160 (.001) -24.6 (.001) -.04 (.681) -86.3 (.043) State Pop Int t-1 21.8 (.198).001 (.490) 14.1 (.077) 1.72 (.182).003 (.896) 5.79 (.471) State Population t -60.8 (.001) -.002 (.001) -30.2 (.000) -5.6 (.001) -.003 (.659) -25.1 (.001) N 40,251 40,375 40,334 40,251 40,375 40,334 Adjusted R 2.51.01.50.54.84.53 Note: Cell entries are unstandardized regression coefficients, two-tailed significance levels in parentheses. Models also include all additional independent variables included in the models in Table 1. 26