Requests and the Distribution of Pork Eric Lawrence George Washington University
[email protected] May 29, 2008
Paper prepared for presentation at the History of Congress Conference, George Washington University, May 30-31, 2008. I thank Amit Kapadia, Mitch Killian, and Maeve Carey for their research assistance.
Once upon a time two boys found a cake. One of them said: “Splendid! I will eat the cake.” The other one said: “No, that is not fair! We found the cake together, and we should share and share alike; half for you and half for me.” The first boy said, “No, I should have the whole cake!” The second said, “No, we should share and share alike; half for you and half for me.” The first said, “No, I want the whole cake.” The second said, “No let us share it half and half.” Along came an adult who said: “Gentlemen, you shouldn’t fight about this; you should compromise. Give him three-quarters of the cake.” (Smullyan 1980, p. 56)
Introduction/Member Preferences for Pork The study of the distribution of government goods, whether federal spending, committee assignments, or earmarks has long been central to legislative studies. Who gets what, who wins, whose ox is being gored, and so forth animate the most active debates in the studies of the U.S. Congress. Is the production of legislation a function of the location of the median voter, of a partisan cartel, or of little legislatures in the form of standing committees operating independently? The study of political “pork” in particular has long informed and been informed by these debates. Scholars have examined the distribution of federal moneys on a broad scale (Stein and Bickers 1995), as case studies (Balla et al. 2002, Lee 2003), and as elements in coalition formation (Evans 2004). In most studies of the distribution of pork, individual legislators’ demand for pork is unknown. Rather, we typically assume that members prefer more pork to less, and the more the better. At the same time, though, we know that some members are more successful at obtaining pork than others, even given similar demands from their constituencies (compare Sens Byrd and Rockefeller, e.g.). This paper offers a first cut at one aspect of the distribution of pork. What happens if we account for member demands directly rather than proxying for it with other variables? Put differently, if we could tell the difference between the two boys described in the Smullyan anecdote (one a sharer, one greedy), would it give us more explanatory power when modeling why some members obtain more pork than others?
Data The rationale for incorporating member preferences and demands for pork is clear. If member preferences have an independent effect on the distribution of pork and their preferences happen to correlate with other variables in our models, not including them in statistical models will lead to omitted variable bias. Member preferences tend not be included in studies of pork not due to researcher ignorance, however. Rather, the omission stems from the lack of data on member preferences. Member preferences are not typically available in the written or archival record, so proxy measures such as constituency characteristics are used in their place.1 While constituency characteristics have the virtue of availability over long periods of time, they can not capture variation across legislators with similar (or identical, in the case of senators) constituencies. Given the absence of publicly available data on member preferences, studies that do use such data will necessarily be episodic, taking the form of case studies. This paper takes advantage of one such episode from the Sixty-fourth Congress (1915-1917). In the Sixty-fourth Congress, the public building bill drew the ire of various good government groups and progressive reformers. The 1917 omnibus public buildings bill was criticized in terms familiar to our modern ears. To wit, minority party member and future Speaker Frederick Gillett (R-MA) offered the following critique.
We all know this bill is not brought in here to improve the post office efficiency. It is brought in largely to improve Congressional fences. It is not peculiar to this committee; it has always been so. These public building bills are brought in the interest of Members...The very way in which this bill was prepared shows that it was not prepared for the public interest. It was prepared for the private interests of Congressmen. The chairman of the committee last year sent out letters for every Congressman suggesting that he send in an item for the bill. Is that the way the other committees go to work? Is that the way any appropriation bill would be prepared? (“Political Public Buildings” 1917) We know, however, that not every member sent in an item for the bill. We know that because one frustrated group, the American Institute of Architects, obtained the building requests and published them in an attempt to bring sunshine to the legislation. The Institute published a list of the appropriation requests, along with the member or Senator making the request and a few other characteristics of the requests in its flagship journal (“Public-Building Bills Introduced...” 1916).2 1 2
See, e.g., the work of Adler and Lapinski (1997) on committee assignments. Details of how precisely the list was obtained and compiled are not provided in the article.
2
This initial publication was followed by an updated list published in a special issue of the first volume of The Searchlight on Congress(Whitaker 1916). The Searchlight was a publication of the National Voters’ League, a good government group headed by Lynn Haines.3 Both the American Institute of Architects and the National Voters League made plain their hopes that publicizing the bill and its contents would embarrass the House and contribute to the demise of the omnibus bill.4 To assess the effect of member demand on the distribution of pork, controlling for traditional measures, the omnibus bill data is combined with other data sources. From the Searchlight data (Whitaker 1916), it is possible to connect individual appropriation requests, the amount of the requests that are granted by the Public Building and Grounds Committee, and the member or Senator making the request.5 Given the three primary organizing theories of legislative organization, data on party, ideology, and committee membership are also needed. For ideology, I use the first dimension DW-Nominate score of each member (McCarty et al. 1997). Committee roster data for membership on the committee of jurisdiction (Public Buildings and Grounds) as well as membership on Appropriations, Rules, and Ways and Means (a proxy for House leadership)6 come from Canon et al. (1998).
Empirical Results Two sets of results demonstrate the impact, or lack thereof, of knowing member demand for pork and controlling for its effect. In the first set of results, the distribution of public building pork is modeled as a function of traditional independent variables and member demand. In the second set of results, member votes on the Omnibus building bill are modeled as a function of traditional independent variables and member demand. The former results demonstrate a significant effect of knowing member demands; the latter show little effect. Those results are summarized shortly. First, however, Figures 1 through 10 illustrate how the 3
Haines had strongly criticized the operation of Congress, the party system, the committee system, etc. in his book Your Congress(1915) prior to the publication of the Searchlight. His work on pork, in particular, reads a bit like modern day critiques made by groups like Citizens against Government Waste. 4 The effort here was therefore similar to the publicity brought to the “A to Z bill” by the Wall Street Journal in the early 1990s (Krehbiel 1995, Binder et al. 1999). 5 This paper does not use the Senate request and appropriation data; only the House member requests are considered. 6 Measuring leadership for this time period by membership on one of these three committees is standard practice. See Wilson 1986a, 1986b, e.g.
3
grant and request data relate to a series of five independent variables. The relationship of each variable to amounts granted (in thousands of dollars) and requested is shown in turn. Figures 1 and 2 show the relationship between member ideology and amount granted (Figure 1) and amount requested (Figure 2). Exploratory analysis indicates that the relationship here is quadratic, though not dramatically so. This implies that both amount granted and requested decline as members get farther from the middle of the distribution, given the shape of the quadratic function. Figures 3 and 4 show that majority party members (Democrats) are granted more money on average (Figure 3) but request similar amounts to the minority Republicans (Figure 4). A similar pattern holds for Public Building membership (Figures 5 and 6) and House leadership membership (Figures 7 and 8). No relationship between amount requested and seniority holds in the sample (Figures 9 and 10). These bivariate relationships suggest, then, that the amount of pork granted correlates with party status, committee membership, and leadership membership, and ideology, with no correlation to seniority. A second set of figures suggest that part of the effect of member demands is interactive. Figure 11 is a scatterplot with the amount granted as a function of amount requested, with separate linear fits for majority party and non-majority party members. The magnitude of the request effect is much larger for majority Democrats than minority Republicans (with slopes of about .20 and .04, respectively). Similar interactive effects appear for committee membership (Figure 12) and House leadership status (Figure 13), with the former being particularly pronounced. This second set of figures provide evidence that models of pork distribution that omit demand will be misspecified. The estimates in Table 1 bear out the bivariate results. The table has three equations. Equation (1) is a garden variety model, explaining grants as a function of party, ideology, and committee membership. Those estimates conform to expectations and are consistent with the bivariate results above. Equation (2) adds the amount requested to the equation, and adding the variable has dramatic effects on the other parameter estimates. In four out of five cases, the parameters are more than cut in half. The majority party effect, e.g., declines from a mean difference of $126,000 to $42,000. The evidence of omitted variable bias is clear. Equation (3) adds variables implied by the interactions in Figures 11-13, increasing the model fit.7 Taken together, the results from Table 1 demonstrate the potential importance of incorporating 7
A Wald test for the joint non-significance of the interaction terms can be rejected at the .01 level.
4
demand data when it is available. The results of Table 2, however, show that knowing member demand does not always change the inferences drawn regarding variables in standard specifications. In Table 2, the vote (yea/nay) on HR 18994 provides the dependent variable. This vote took place in January 1917 after the 1916 congressional elections. A yea vote supports the omnibus bill. Ideology is dropped here due to correlation between ideology and party membership.8 Again, three equations are estimated, with equation (1) omitting requests, equation (2) adding requests, and equation (3) adding requests and the same interactions as Table 1. One additional variable is added to the equation, a dummy for whether the member has an item granting him money in the omnibus bill. As expected, that variable is significant (increasing the likelihood of voting yea by about %21). Adding requested funds does not have the same impact on the parameter estimates exhibited in Table 1. Taking the two tables together, we can conclude that while including member demands can make a difference on our estimates (Table 1), pronounced effects are not guaranteed (Table 2).
Conclusion/Discussion These preliminary results are suggestive of the utility of accounting for member preferences over pork, but several refinements are necessary to move the results from the category of speculative to persuasive. First, the issue of the endogeneity of member preferences should be considered. It may be the case that members have private information about the likely success of their requests and submit their bids on that basis. If that is the case here, demands are not exogenous and an alternative modeling strategy is required.
8 That choice was arbitrary. Dropping party status and including ideology does not affect the parameter estimates on the other regressors
5
References Adler, E. Scott and John S. Lapinski. 1997. “Demand-Side Theory and Congressional Committee Composition: A Constituency Characteristics Approach,” American Journal of Political Science41(3): 895-918. Balla, Steven J., Eric D. Lawrence, Forrest Maltzman and Lee Sigelman. 2002. “Partisanship, Blame Avoidance, and the Distribution of Legislative Pork,” American Journal of Political Science 46(3): 515-525. Canon, David, Garrison Nelson, and Charles Stewart. 1998. Historical Congressional Standing Committees, 1st to 79th Congresses, 1789-1947: House and Senate, 2/14/1998. Evans, Diana. 2004. Greasing the Wheels: Using Pork Barrel Projects to Build Majority Coalitions. New York: Cambridge University Press. Groseclose, Tim, and Charles Stewart III. 1998. “The Value of Committee Seats in the House, 1947-91,” American Journal of Political Science 42(2): 453-474. Haines, Lynn. 1915. Your Congress. Washington D.C.: National Voters’ League. Haines, Wilder H. 1915. “The Congressional Caucus of Today,” American Political Science Review 9(4): 696-706. Lee, Frances E. 2003. “Geographic Politics in the U.S. House of Representatives: Coalition Building and Distribution of Benefits,” American Journal of Political Science 47(4): 71428. McCarty, Nolan, Keith Poole, and Howard Rosenthal. 1997, Income Redistribution and the Realignment of American Politics. Washington D.C.: AEI Press. Peters, Ronald. 1997. The American Speakership: The Office in Historical Perspective. Baltimore, MD: Johns Hopkins University Press. “Political Public Buildings.” 1917. The Searchlight on Congress II(1): 1-4. “Public-Building Bills Introduced into the 64th Congress of the United States.” 1916. Journal of the American Institute of Architects June: 243-254.
6
Smullyan, Raymond. 1980. This Book Needs No Title: A Budget of Living Paradoxes. New York, NY: Prentice Hall. Stein, Robert M. and Kenneth N. Bickers. 1995. Perpetuating the Pork Barrel: Policy Subsystems and American Democracy. New York: Cambridge University Press. Whitaker, Charles Harris. 1916. “Mostly Pork! The Most Far-Reaching and Indefensible Distribution of ‘Pork’ Takes Place When Congress Passes What is Known as the Omnibus Public Buildings Bill,” Searchlight on Congress 1: (November/December). Wilson, Rick. 1986a. “What Was It Worth to Be on a Committee in the U. S. House, 1889 to 1913?” Legislative Studies Quarterly 11(1): 47-63. Wilson, Rick. 1986b. “An Empirical Test of Preferences for the Political Pork Barrel: District Level Appropriations for River and Harbor Legislation, 1889-1913,” American Journal of Political Science 30(4): 729-754.
7
Majority party dummy, = 1 if Democrat First dimension DW-Nominate, 64th House Squared first dimension DW-nominate MC is on Public Buildings and Grounds committee MC is house leader (Rules, Approp., W & M)
(1) 125.660 (1.96) 139.979 (1.90) -81.209 (-1.14) 185.501 (1.85) 68.015 (1.39)
(2) 41.725 (1.28) 47.626 (1.34) -17.394 (-0.47) 131.105 (1.80) 22.614 (1.31) 0.190 (6.11)
-1.234 (-0.04) 0.081 325
-3.792 (-0.20) 0.513 325
Total amount of funds requested, in 1000s Amount requested X Majority status Amount requested X Public Buildings membership Amount requested X House leader Constant R2 N
(3) 44.220 (1.53) 46.631 (1.50) -24.999 (-0.85) -62.884 (-2.00) -5.261 (-0.41) 0.106 (2.06) -0.010 (-0.12) 0.462 (5.71) 0.116 (1.50) 13.971 (0.74) 0.629 325
Table 1: OLS estimates of building funds awarded to House members. Robust z-statistics in parentheses below estimates. One outlying case, with an award of $4.25 million, is omitted from the analysis.
8
MC has item in omnibus bill Majority party dummy, = 1 if Democrat MC is on Public Buildings and Grounds committee MC is house leader (Rules, Approp., W & M)
(1) 0.958 (5.83) 0.534 (3.38) 0.139 (0.32) -0.597 (-2.62)
(2) 0.904 (5.55) 0.519 (3.28) 0.099 (0.23) -0.608 (-2.65) 0.000 (1.28)
-0.220 (-1.49) .135 325
-0.246 (-1.64) .144 325
Total amount of funds requested, in 1000s Amount requested X majority status Amount requested X Public Buildings membership Amount requested X House leader Constant McFadden R2 N
(3) 0.817 (4.78) 0.735 (3.92) -1.335 (-1.81) -1.126 (-2.97) 0.001 (1.49) -0.001 (-1.57) 0.014 (1.27) 0.006 (1.47) -0.293 (-1.81) .176 325
Table 2: Probit estimates of vote on HR 18994, the Omnibus building bill. Robust z-statistics in parentheses below estimates. One outlying case, with an award of $4.25 million, is omitted from the analysis.
9
1500
1000
500
0 1
.12
First dimension DW Nominate, 64th House Median DW
Nominate score =
.12
Figure 1: Amount granted by ideology, with quadratic fit
10
1
4000
3000
2000
1000
0 1
.12
1
First dimension DW Nominate, 64th House Median DW
Nominate score =
.12
Figure 2: Amount requested by ideology, with quadratic fit
.015
.01
.005
0 0
200
400
600
1000s of dollars Majority party Minority party Four members with allocations greater than $1,000,000 suppressed for clarity
Figure 3: Amount granted by party
11
800
.004
.003
.002
.001
0 0
2000
4000
6000
8000
1000s of dollars Majority party Minority party Four members with allocations greater than $1,000,000 suppressed for clarity
Figure 4: Amount requested by member by party
.015
.01
.005
0 0
200
400
600
800
1000s of dollars Public Buildings and Grounds member non member Four members with allocations greater than $1,000,000 suppressed for clarity
Figure 5: Amount granted by Public Buildings and Grounds membership
12
.005 .004 .003 .002 .001 0 0
2000
4000
6000
8000
1000s of dollars Public Buildings and Grounds member non member Four members with allocations greater than $1,000,000 suppressed for clarity
Figure 6: Amount requested by Public Buildings and Grounds membership
.015
.01
.005
0 0
200
400
600
1000s of dollars Rules, Ways and Means, or Appropriations member non member Four members with allocations greater than $1,000,000 suppressed for clarity
Figure 7: Amount granted by leadership status
13
800
.005 .004 .003 .002 .001 0 0
2000
4000
6000
8000
1000s of dollars Rules, Ways and Means, or Appropriations member non member Four members with allocations greater than $1,000,000 suppressed for clarity
Figure 8: Amount requested by member by leadership membership
1000s of dollars
1500
1000
500
0 0
5
10
15
Number of terms served in House
Figure 9: Amount granted by seniority
14
20
8000
1000s of dollars
6000
4000
2000
0 0
5
10
15
20
Number of terms served in House
Figure 10: Amount requested by seniority, with quadratic fit
Amount granted, in 1000s
2000
1500
1000
500
0 0
2000
4000
6000
Total amount of funds requested, in 1000s minority party member minority fit
majority party member majority fit
One member with allocation greater than $1,500,000 suppressed for clarity
Figure 11: Granted vs. requested, by party status
15
8000
Amount granted, in 1000s
1500
1000
500
0 0
2000
4000
6000
8000
Total amount of funds requested, in 1000s non member non member fit
Public Buildings and Grounds member Public Buildings and Grounds fit
One member with allocation greater than $1,500,000 suppressed for clarity
Figure 12: Granted vs. requested, by committee membership
Amount granted, in 1000s
2000
1500
1000
500
0 0
2000
4000
6000
8000
Total amount of funds requested, in 1000s non member non member fit
Rules, Ways and Means, or Appropriations member Rules, Ways and Means, or Appropriations fit
One member with allocation greater than $1,500,000 suppressed for clarity
Figure 13: Granted vs. requested, by leadership status 16
