The Impact of Legislator Attributes on Interest-Group Campaign Contributions* KEVIN B. GRIER George Mason University, Fairfax, VA 22030 MICHAEL C. MUNGER University of Texas, Austin, TX 78712 Legislators possess political assets that economic interest groups may find valuable in pursuing their goals. This paper examines the effect these legislative assets have on the campaign contributions made by two large and easily identifiable interest groups: corporations and labor unions. Committee assignment, voting record, and electoral security are significant predictors of both corporate and union contributions to House incumbents, while party affiliation and years in office also influence the behavior of union political action committees.

1.

Introduction

The Federal Election Campaign Act (FECA), enacted in 1971 and amended in 1974and 1976, represents a substantial change in the way that economic interest groups participate in the political process. From 1943 to 1971, both corporations and unions were legally proscribed from making contributions to political candidates. Now, these contributions, though regulated, limited, and a matter of public record, are an important source of funds for politicians. This paper focuses on the activities of corporate and union Political Action Committees (PACs) over the period 1978 to 1982. We present and test a model that predicts interest-group contributions based on observable qualities of legislators, finding that these PACs contribute disproportionately to legislators who have a comparative advantage in producing the services desired. This paper reviews the history of corporate and union campaign contribution laws (Section II); summarizes previous work on this issue (Section III); outlines our model of the importance of legislative assets in obtaining interest-group contributions (Section IV); explains the nature of the data and statistical procedures (Section V); and presents the econometric results (Section VI).

*We thank James Bennett, Mark Crain, Art Denzau, Bob Tollison, and Barry Weingast for comments and suggestions of fundamental importance in the development of this paper. Of course, they bear no responsibility for any shortcomings of the final result. The research in this paper was supported by the Sarah Scaife Foundation and the Washington University Committee on Political Economy. JOURNAL OF LABOR RESEARCH Volume VII, No.4 Fall 1986

350

JOURNAL OF LABOR RESEARCH

II. History of Corporate and Union Contribution Laws The first attempt to regulate corporate political activity, the Tillman Act (1907), grew out of a concern that corporate funds could unduly influence elections. The act, which prohibited monetary contributions, was extended by the Corrupt Practices Act (1925) to disallow contributing anything of value. During this period, labor unions were allowed to make direct contributions, an imbalance that was reduced by the Hatch Act (1940) which placed a $5,000 limit on contributions. The Smith-Connaly Act (1943) and the Taft-Hartley Act (1947) equalized the treatment of unions and corporations by outlawing contributions by either group. This status quo was incorporated in the Federal Election Campaign Act (FECA) of 1971, which regulated contributions by individuals and non-affiliated clubs and proscribed all corporate and union contributions. The Supreme Court ruling in Pipefitters Local #52 vs. United States in 1972 opened the door for direct interest group contributions. The Pipefitters created a campaign fund and were charged with violating Taft-Hartley as amended by FECA. The Court ruled, however, that the fund complied with the statutes because it was strictly separate from other union funds for accounting purposes and contributions were not a condition of union membership. This decision marked the inception of the Political Action Committee (PAC) as a vehicle for interest-group contributions to political candidates. In 1974, FECA was amended to incorporate the Court's decision and to impose a significant set of regulations on business and union PACs. The Federal Election Commission (FEC) was created and criminal enforcement provisions were strengthened by increasing fines for violations and closing loopholes by redefining illegal activity. Contribution limits were also imposed: $1,000 by persons per candidate per election and $5,000 by multicandidate committees (i.e., PACs). Despite further amendment in 1976, including the elimination oflimits on contributions by individuals, this is still basically the regulatory environment for union and corporate contributors. I Both the number of PACs and the amount of PAC money in elections have been increasing. Table 1 reports the number of corporate and union PACs from 1978 to 1982, while Table 2 shows the level of contributions by these groups to House incumbents running for re-election over the same time period. Union contributions increase 78 percent from 1978 to 1982, but the explosive growth in corporate PAC contributions is more notable. Corporate contributions have grown by over 300 percent and have become larger than the union figures, both in total and average contributions. Individual union PACs are still larger than their corporate counterparts, with average total contributions per PAC being $27,400 for union and $11,100 for corporations in 1982. 'The 1976 amendments were designed to incorporate another Supreme Court ruling; Buckley vs. Valeo struck down limits on individual contributions by equating political spending with freedom of speech. The interest group limits were not changed.

KEVIN B. GRIER and MICHAEL C. MUNGER

351

Table 1

The Growth of Political Action Committees, 1978-1982" Union

Corporate

201 224 276

682 1,062 1,102

1978 1980 1982

aOata are the number of PACs registered with the Federal Election Commission in each election cycle. Not all PACs exist in all election cycles.

Table 2

Patterns of PAC Contributions" A: TOTAL PAC CONTRIBUTIONS TO HOUSE INCUMBENTS Union

Corporate

1978

$4,390,386

$ 3,623,686

1980

6,673,205

8,195,495

1982

7,566,439

12,271,746

B: AVERAGE NON-ZERO CONTRIBUTIONS TO HOUSE INCUMBENTS Union

Corporate

1978

$13,180

$10,290

1980

20,040

21,910

1982

22,860

34,660

aAll figures are in 1982dollars. Data are from the FEC, as reported by corporate and labor PACs on FEC form A, schedule A, line 15-17, as required by law.

III.

Other Research

The role of interest-group campaign contributions to political candidates has been considered by numerous authors in a variety of contexts. Theoretical papers by Barro (1973), Welch (1974), Ben-Zion and Eytan (1974), Peltzman (1976), and Becker (1983) treat contributions as one of the many transactions through which markets for policy clear. Barro sets out the agency problems that voters face in controlling elected representatives and points out that these problems increase the possibility that inter-

352

JOURNAL OF LABOR RESEARCH

est-group contributions will affect policy outcomes. Welch seeks to distinguish contributions made for the purpose of affecting election outcomes from those made as payment to politicians for services rendered. Ben-Zion and Eytan model the connection between economic and political power using a vote-maximization objective for candidates and allowing interest groups to make contributions in an effort to influence policy. There has been relatively little empirical testing of these theoretical propositions. Abrams and Settle (1978) examine the implications of restricting access to campaign funds. Crain and Tollison (1976) and Crain, Deaton, and Tollison (1977)attempt to measure the value of a legislative seat by examining factors that determine how much candidates will spend to obtain it. Jacobson (1980) and Welch (1980) examine the effect of campaign spending on election outcomes and policy decisions, respectively. None of these papers addresses the question posed in this paper: What determines the contribution patterns of corporate and union PACs to incumbent legislators? The empirical work most similar to ours is Kau and Rubin (1982), who compare PAC contributions in the 1972and 1978elections and Poole and Romer (1984), who study the pattern of interest-group PAC contributions in the 1980 election.

IV.

Legislative Assets and Campaign Contributions

Our model is based on Denzau and Munger's (1986)theory of the interaction between voters, interest groups, and legislators. They assume that legislators maximize the probability of re-election, where that probability is a function of dollars, or campaign contributions, and voter reaction to legislative performance. We focus on the attributes that make a legislator attractive to an interest group. The cornerstone of this model is the recognition that legislators are not identical. Each is elected from a distinct geographic district with voters whose preferences may vary widely across districts. Each legislator also has different endowments of ability and experience and a different position in the legislative committee system. Thus, the cost of providing services to interest groups will vary across legislators, and interest groups can be expected to employ the legislator or group of legislators that are the least-cost suppliers of the desired policy. The contributions received from an interest group depend on the comparative advantage the legislator has in performing the services the group desires. Each legislator will specialize in activities where his effort is the most productive. Attempts to exchange interest-group services for contributions in an area of comparative disadvantage will fail because the interest group can purchase the service elsewhere at lower cost. The following attributes or "legislative assets" will help determine which legislators will be the low-cost suppliers of a specific service and will thus receive disproportionate contributions from a specific interest group.

Committee assignments. Committees are the basic unit of legislative power in the U.S. House of Representatives. Policymaking in the House is divided into

KEVIN B. GRIER and MICHAEL C. MUNGER

353

separate, though sometimes overlapping, jurisdictions: the 22 standing committees. Each committee has effective veto power over legislation within its jurisdiction (Denzau and Mackay, 1983). The committee can always protect the status quo by simply refusing to report out new legislation. It also controls the writing and progress of new legislation in its area. Even after a bill is passed and is being administered or regulated by the bureaucracy, the committee with jurisdiction has substantial power to influence the administrative agency through hearings, agency appointments, and the ability to control future appropriations to the agency (Weingast and Moran, 1983; Weingast, 1984; and Grier, 1985). The committee seat that a legislator' 'owns" giveshim a comparative advantage over politicians not on that committee in influencing policies in the committee's jurisdiction. This implies that legislators on the committee with jurisdiction over an area of policy that concerns an interest group willhave a cost advantage in producing services and that the group will tend to purchase the desired service from these low-cost suppliers through larger campaign contributions. Voter preferences. Denzau and Munger demonstrate that an important determinant of the cost of providing interest-group service is the reaction of the legislator's constituency. While interest-group contributions produce votes indirectly, if voters disapprove of the action the legislator may lose more votes directly than can be gained through use of the campaign funds. Politicians with a pro-union constituency would charge a very high price to produce corporate interest-group services, regardless of productivity, because it will be costly in terms oflost votes.' On the assumption that legislators in some broad sense "vote their constituencies," the legislator's voting record is a good proxy for constituent preferences. Electoral security. The effort that a legislator willoffer for a given contribution should depend on how badly money is needed to finance his campaign. That is, if a legislator expects a close race and needs extra money to defeat his opponent, he will be motivated to solicit contributions by producing interest-group services simply because the money is more valuable to him than to an unopposed or secure incumbent. Tenure. Years in office may serve as a proxy for a variety of legislative assets, such as procedural expertise and collegial respect. Tenure also helps legislators advance in the committee hierarchy, but in the House the link between tenure and committee leadership is weakening.'

'We take voting as a determinant of contributions, rather than the reverse hypothesis that offers of money determine votes, based on the view that politicians are the constrained agents of their constituents in the final analysis and on the evidence that the voting behavior of incumbent politicians is relatively stable. See Glazer and Robbins (1985) and Dougan and Munger (1986). 'For example, Les Aspin and Pat Gray are non-senior chairs of the Armed Services and Budget committees. The possibility of taking the chairmanship without being the senior majority member has existed since 1975, when the House rules were changed to de-emphasize seniority.

354

JOURNAL OF LABOR RESEARCH

Party. A legislator's party affiliation may carry information about his voters' preferences and his access to the leadership that affects legislative productivity.

V. Data and Model Selection Our data consist of three election cycle (1978-1982) cross-sections of PAC campaign contributions. The FEC classifies PACs into five groups: labor, corporate, trade/membership/health, cooperatives and non-connected. Here we analyze the contributions of corporate and union PACs to U.S. representatives running for re-election.' In the corporate PAC contributions data, 46 incumbents did not receive any money in at least one of the three elections. The average per-election contribution by all corporate PACs to incumbents is $21,412, and the maximum is a little over $200,000 (received in 1982by John Rousselot, R-Calif.). In the union data, there are 176incumbents who receive no money in one of the elections; the average contribution is $16,501; and the maximum contribution is $134,060 (in 1982to James Howard, D-N.J.). The type of data available to test our model is necessarily truncated at zero. An incumbent can receive anywhere from zero to an infinite amount of contributions, but explicit negative contributions do not occur. This mass point at zero in the distribution of PAC contributions biases ordinary least squares estimates. Yet, we have much more information than purely qualitative estimation procedures, such as Logit or Probit, use. The best estimation procedure in this case is Tobit. This method estimates a non-linear, maximum-likelihood probability model and then uses the first partial derivatives of the likelihood function to generate linear regression coefficients that can be used to predict levels of contributions. For detailed discussions of the problems of truncated data and the properties of Tobit estimation, see Tobin (1958), Ameniya (1973), and Heckman (1976). Our theory assumes that interest-group contributions are purchases of legislative services and implies that groups will seek out the low-cost suppliers of such services. There are three major elements of the cost of special-interest service that we test. First is the institutional cost of providing the service. Legislative jurisdictions are apportioned across committees; and the major substantive committee for union interests is Education and Labor, while the prime business committee is Energy and Commerce.' Our theory predicts that members of these committees are the most cost-effective in obtaining services for union and business groups respectively, and thus should, ceteris paribus, receive a greater amount of campaign funds from their respective demanders. As a secondary proposition, mem-

·We exclude the FEe category of corporations with no stock from the corporate data set. 'If the same legislators served on both committees, we would have problems separating the effects. In the House, most members have only one substantive committee assignment. and over our 1978-1982 sample there is no membership overlap.

KEVIN B. GRIER and MICHAEL C. MUNGER

355

bers of the business (labor) committee should receive less funding from union (corporate) PACs. We test these propositions by including dummy variables that indicate membership on a substantive committee (EDLAB and ENCOM). Second, there is the cost of disenchanting constituents by servicing special interests. Legislators who have a reputation of being conservative (pro-business) will not surprise constituents by supporting organized business interests. Indeed, this behavior will broadly reflect the economic or ideological interests of the voters that elected him. The same argument applies to liberal legislators and their constituency. Legislators with conservative voting records should get more corporate money; liberal voters, extra union funds. The Chamber of Commerce Vote Score (CHAMBER VOTE), where pure pro-business is 100and anti-business is 0, is used to test this hypothesis. The third cost component is the legislator's marginal disutility of effort. Serving interest groups is hard work, and legislators who have "safe" seats will need fewer campaign funds and be less willing to supply special-interest service at low cost. We use a measure. of the margin of victory in the incumbent's last election (% VOTEE - 1) to test for this effect. To complete the empirical model, we include two other variables that are often considered important and may proxy for cost parameters in our theory: a dummy variable for party affiliation (DEMOCRAT = 1 if legislator is a democrat, otherwise = 9) and a measure of the length of service of each incumbent (TENURE). Tobit regressions of this model on corporate and union data sets are presented in Table 3 and Table 4. To allow for the effects of inflation, the equations are estimated using contribution figures in 1982dollars. We examine the stability of the slope coefficients over time by estimating a separate regression for each cross-section and then testing the hypothesis that the coefficients are equal. The null hypothesis of coefficient equality is rejected only in the case of Chamber of Commerce Vote Scores." The reported regressions constrain all coefficients but Chamber vote and the intercept to be constant across election cycles.

VI.

Corporate and Union Regression Results

Table 3 contains the estimates of our model on the corporate data set. In general, our theory is not in discord with the empirical results. Using the linearized coefficients, the R 2 between actual and predicted corporate contribution levels is .31.

'The particular test involves comparing the maximized values of the unrestricted and restricted likelihood function, called the likelihood ratio test. Specifically, - 2*log (ratio of the likelihood functions) is distributed chi-square with k degrees of freedom, where k is the number of restrictions. We also put union and corporate observations into one large data set (n = 2267) and tested H.: union slope coefficients = the corresponding corporate coefficients. This hypothesis is resoundingly rejected at any significance level chosen.

JOURNAL OF LABOR RESEARCH

356

Party and years in office (DEMOCRA T and TENURE) are not significant predictors of the likelihood of an incumbent receiving money, but all the other regressors are significant. The committee variables perform exactly as expected: a seat on Education and Labor greatly reduces corporate contributions ( - $6,6(0) while Energy and Commerce is worth about an extra $6,700. Reputation is also very important. CHAMBER VOTE is positive and significant in each year, with the significance level rising each election cycle. In 1978, 10 additional Chamber of Commerce Vote points were worth $1,350; in 1980, $4,730; in 1982, $4,070. Taking an extreme pro-business position (a 100percent

Table 3 Predicting Corporate Contributions to House Incumbents, 1978-1982° Tobit Coefficient

Variable

TENURE

0.005

Regression Coefficient

95

(1.12)

%VOTEE _ 1

-0.013 (6.24)

-264

DEMOCRAT

-0.115 (1.41)

-2,384

CHAMBER VOTE-78

0.007 (2.89)

135

CHAMBER VOTE-80

0.023 (4.38)

473

CHAMBER VOTE-82

0.019 (8.30)

407

EDLAB

-0.319 (2.93)

-6,640

ENCOM

0.325 (3.11)

6,743

CONSTANT

1.037 (5.22)

21,526

DUMMY-80

-0.638 (1.79)

-13,247

DUMMY-82

0.449 (2.77)

9,316

Log of Likelihood Fn: -12360.5 Non-Limit Observations: 1084 Limit Observations: 46

Durbin-Watson: 1.83 Pseudo R 2 : .310 p(Y> 0 I Average X's): .84

"Data are in 1982 dollars. DUMMYSO and DUMMYS2 are time-series intercept shifts. Numbers in parentheses are asymptotic T-ratios.

KEVIN B. GRIER and MICHAEL C. MUNGER

357

Table 4 Predicting Union Contributions to House Incumbents, 1978-1982° Tobit Coefficient

Regression Coefficient

TENURE

-0.009 (2.36)

-176

%VOTEE _ ,

-0.030 (13.88)

-545

DEMOCRAT

0.983 (11.46)

17,700

CHAMBER VOTE-78

-0.009 (4.16)

-177

CHAMBER VOTE-80

-0.030 (5.77)

-544

CHAMBER VOTE-82

-0.030 (12.34)

-550

EDLAB

0.446 (4.04)

7,999

ENCOM

0.229 (2.16)

4,126

CONSTANT

2.442 (11.62)

43,824

DUMMY-80

1.940 (5.41)

34,824

DUMMY-82

1.750 (10.49)

31,585

Variable

Log of Likelihood Fn: -10896.9 Non-Limit Observations: 961 Limit Observations: 176

Durbin-Watson: 1.86 Pseudo R': .521 P(Y> 0 I AverageX's): .78

aData are in 1982 dollars. DUMMYSO and DUMMYS2 are time-series intercept shifts. Numbers in parentheses are asymptotic T-ratios.

voting record) generates $40,700 in extra corporate contributions as of 1982. The ease of victory in the previous election is a significant negative influence on corporate PAC contributions to incumbents, reflecting our hypothesis that secure incumbents provide fewer interest-group services because their marginal valuation of additional campaign funds is lower than that of an incumbent in a tight race. Consider now the union PAC contribution equation in Table 4, where all the regressors are statistically significant. Unlike corporations, union PACs treat party affiliation as an important signal in distributing money. Of the 176incumbents who received no union contributions in an election cycle, 131 of them were

JOURNAL OF LABOR RESEARCH

358

Republicans. Also, the coefficient on DEMOCRATin the union equation implies that, ceteris paribus, being a Democrat generates around $17,000 in extra union contributions. The effect of committee assignment on union contributions is also somewhat at variance with the corporate result. A seat on the main jurisdictional committee (EDLAB) does generate extra campaign revenue of around $8,000, but members of the Energy and Commerce committee also receive about half that amount in extra union PAC funds. As predicted, the effect of the incumbent's voting record is the mirror image of the corporate equation. The negative effect of pro-business voting increases in size and significance each election. By 1982, each 10points of Chamber Vote score cost an incumbent $5,500 in union contributions. The incumbent's percentage of the vote in the previous election also significantly reduces union contributions. Also, years in office (TENURE) has a small negative and significant effect on union PAC contributions. The linear R 2 of this equation is .52. 7 We experimented with several other hypotheses in unreported regressions. First, the models were checked for possible interaction effects between party, voting record, and committee assignment and tenure and committee assignment with no significant effects uncovered. Second, we included dummy variables for the chairmen of the Education and Labor and Energy and Commerce committees. These turned out to be negative, but insignificant. Third, dummies for all other

Table 5 Data Summary Variable

Mean

Std. Deviation

Min

Max

Total Union Contributions Total Corporate Contributions Tenure % VoteE-t Chamber of Commerce Vote

16,501 21,412 8.07 68.61 56.0

22,560 24,215 7.35 14.50 23.5

0 0 0 38 0

134,060 209,770 41 99 99

'Our results on the effects of party and % Vote.-t on corporate and union PAC contributions, confirm and extend the Poole and Romer study of the 1980election data. Poole and Romer also use a measure of conservatism they generate with a "multidimensional unfolding" technique that has the same qualitative effect as our CHAMBER VOTE variable. They model seniority non-linearity and find it significant for corporations, but insignificant for unions, which is the opposite of our findings with a linear variable and the elections before and after 1980included in the sample. Poole and Romer do not have a model that predicts the sign or significance of their regressors.

KEVIN B. GRIER and MICHAEL C. MUNGER

359

committee assignments were included and all, with the exception of Foreign Affairs, were insignificant.' A seat on Foreign Affairs has a small negative effect on contributions by both types of PACs. We have no explanation for this effect, but it does not influence the size or significance of our reported results.

VII.

Conclusions and Extensions

Our argument assumes that special interest groups are interested in obtaining cost-effective legislative services. From this we predict that specific legislator characteristics will attract contributions from some, but not necessarily all, interest groups. We test the model by predicting corporate and union PAC contributions to House incumbents as functions of committee assignment, voting record, and electoral security. The available data are explained relatively well and clearly demonstrate that the two interest groups purchase services from different types of legislators. Our results imply that having a seat on Energy and Commerce and a voting record two standard deviations above the mean produce about $25,500 in additional corporate PAC contributions while a seat on Education and Labor and a voting record two standard deviations below the mean garners almost $34,000 in extra union contributions compared to the amount received by an "average" legislator. Specific legislative assets clearly have different values to different interest groups.' An important assumption made in this study is that corporations and unions are monolithic interest groups. That is, we assume that a certain committee assignment has the same value to all unions or corporations. The significance of the results indicate that this assumption may not be too extreme, but it would be valuable to disaggregate PAC contributions by line of business and test for intragroup differences in contributions. Another interesting test of our theory would be to apply the model to predicting when and how much PACs will contribute to challengers in House elections.

'Poole and Romer include all 22 committees as dummy variables and note that some committees are significant with no expectation if this result is good, bad, or even interesting. Our theory predicts only certain committees should be significant, and in the empirical tests, they, and only they, are. 'By average, we mean a Democrat with the average vote percentage and voting record, not on the committees of interest. Evaluating our linearized coefficients at the means of the continuous independent variables, our model predicts corporate contributions to the "average incumbent" to be $33,500, while union contributions to this individual are predicted to be $25,000.

360

JOURNAL OF LABOR RESEARCH

REFERENCES Abrams, Burton A. and Russell Settle. "The Economic Theory of Regulation and Public Financing of Presidential Elections," Journal of Political Economy 86 (April 1978): 245-57. Amemiya, Takeshi. "Regression Analysis When the Dependent Variable Is Truncated Normal," Econometrica 41 (November 1973): 997-1016. Barro, Robert J. "The Control of Politicians: An Economic Model," Public Choice 17(Spring 1973): 19-42. Becker, Gary S.• 'A Theory of Competition Among Pressure Groups for Political Influence," Quarterly Journal of Economics 98 (August 1983): 371-400. Ben-Zion, Uri and Zeev Eytan. "On Money, Votes and Policy in a Democratic Society," Public Choice 21 (Spring 1974): 1-10. Congressional Quarterly Weekly Almanac. Washington, D.C.: Congressional Quarterly Inc., various issues.

Crain, W. Mark, Thomas Deaton, and Robert Tollison. "Legislators as Taxicabs: On the Value of Seats in the U.S. House of Representatives," Economic Inquiry 15 (April 1977): 298-302. Crain, W. Mark and Robert Tollison. "Campaign Expenditures and Political Competition," Journal of Law and Economics 19 (April 1976): 177-88. Denzau, Arthur T. and Robert Mackay. "Gatekeeping and Monopoly Power of Committees: An Analysis of Sincere and Sophisticated Behavior," American Journal of Political Science 27 (November 1983): 740-61. Denzau, Arthur T. and Michael C. Munger. "Legislators and Interest Groups: How Unorganized Interests Get Represented," American Political Science Review (March 1986). Dougan, William and Michael C. Munger. "The Rationality of Ideology," University of Chicago Working Paper, May 1986. Glazer, A. and M. Robbins. "How Elections Matter: A Study of U.S. Senators," Public Choice 46 (Spring 1985): 163-72. Grier, Kevin B. "Congressional Preference and Federal Reserve Policy: A Principal - Agent Approach," Center for the Study of American Business Working Paper Series, October 1985. Heckman, James. "The Common Structure of Statistical Models of Truncation, Sample Selection and Limited Dependent Variables and a Simple Estimator for Such Models," Annals of Economic and Social Measurement 5 (Fall 1976): 475-92. Jacobson, Gary C. Money in Congressional Elections. New Haven: Yale University Press, 1980. Kau, James and Paul Rubin. Congressmen, Constituents and Contributors. Boston: Nijhoff, 1982. Mayhew, David R. Congress: The Electoral Connection. New Haven: Yale University Press, 1974. Peltzman, Sam. "Toward a More General Theory of Regulation," Journal ofLaw and Economics 19 (June 1976): 211-40. Poole, Keith T. and Thomas Romer. "Patterns of Political Action Committee Contributions to the 1980 Campaigns for the U.S. House of Representatives," Public Choice 47 (1985): 63-112. Tobin, James. "Estimation of Relationships for Limited Dependent Variables," Econometrica 26 (January 1958): 24-36.

KEVIN B. GRIER and MICHAEL C. MUNGER

361

Weingast, Barry R. "The Congressional-Bureaucratic System: A Principal-Agent Perspective," Public Choice 44 (1984): 147-91. _ _ _ _ _ _ and Mark Moran. "Bureaucratic Discretion or Congressional Oversight? Regulatory Policy-making by the Federal Trade Commission," Journal ojPolitical Economy 91 (October 1983): 765-800. Welch, W. P. "The Economics of Campaign Funds," Public Choice 20 (Winter 1974): 83-97. _ _ _ _ _ _. "The Allocation of Political Monies: Economic Interest Groups," Public Choice 35 (Spring 1980): 97-120.