Remuneration vs. reelection: a senatorial balancing act.

Couch, Jim F. ; King, Brett A. ; Stevenson, Taylor P. 等

Unique events often give rise to other unique events. Such was the case of Hurricane Katrina--the costliest and most destructive natural disaster in U.S. history. The resulting casualties and staggering damage estimates from the savage storm in late August 2005, led the U.S. Senate to vote overwhelmingly (92 to 6) against a Senate pay raise. Most senators apparently felt that voters would not look kindly on a Senate pay raise given the devastation caused by Katrina.

The Senate vote was largely symbolic because the House of Representatives was not going to take a vote on rescinding a pay raise for members of Congress. Thus, senators voting against the raise knew their votes would resonate well with voters but have no impact on the annual congressional pay raise sanctioned by law. Of the six senators who did vote for the pay raise, five--James Jeffords (I-Vt.), Daniel Inouye (D-Hawaii), Jeff Bingaman (D-N.M.), Richard Lugar (R-Ind.), and Kit Bond (R-Mo.)--were long-time incumbents whose seats were not threatened, and one--Paul Sarbanes (D-Md.)--was about to retire.

Typically, votes cast by members of Congress are relatively easy to predict. The empirical analysis of congressional voting patterns suggests that measures of ideology and party affiliation play a decisive role in explaining overall voting behavior. A vote on congressional pay, however, is not typical. It creates a dilemma for lawmakers by pitting two margins of self-interest against each other: pecuniary gains and reelection. Senators are clearly made better off by not opposing annual pay raises--salaries have increased from $98,400 in 1990 to $165,200 in 2006. Yet, those increases are likely to irritate voters and could harm a senator's chances for reelection. In this article, we examine those conflicts of interest and how they affect voting behavior in the case of senatorial pay raises.

It is often reported that cooperation among politicians, or bipartisanship, is a thing of the past. However, our results suggest that at least in the case of the Senate, when it comes to an issue as controversial as a pay increase, senators are fully capable of cooperating. More vulnerable senators (in terms of their probability of reelection) are allowed to vote against a pay raise, knowing that those in a more secure position can vote for the increase. The more vulnerable senators likely compensate senators who take the unpopular position of voting in favor of the pay increase with favorable votes on future legislation.

The Framers of the Constitution, after much debate, made two very important decisions about the salaries of those elected to serve in the U.S. Senate. First, the Framers decided that senators' salaries would be paid out of the U.S. Treasury and not by the states that elect them. The idea was that state governments could punish senators by withholding their pay, thereby influencing national decisions. Second, the newly elected Congress would determine the method and amount members would receive

We would like to focus on the second decision. Congress has the responsibility of determining its own pay. To be sure, members of Congress are in a precarious position when they vote to increase their own salaries. They must consider whether or not the taxpaying voter finds them deserving. More important, to benefit from the salary increase, they must continue to hold their seats in Congress.

The discussion of congressional voting has centered on the question of whether the members' votes are determined by the interests of the constituency they represent, an expression of their political ideology, or some combination of the two. Research has shown that the two are interrelated, but it is ideology that is the best predictor of how a senator will vote on a particular issue.

The question we address is whether ideology or some other factor is a good predictor of how senators vote when it comes to the issue of increasing their salaries. We hypothesize those factors other than ideology will ultimately determine voting behavior on this issue. We predict that senators who are less likely to be voted out of office will favor a salary increase, while those who are more vulnerable will oppose a pay raise.

A Brief History of Congressional Pay

In 1789, Congress set its pay at six dollars per day. The pay rate did not change until 1816 when Congress voted to change the structure of congressional pay from a per diem rate to an annual salary of $1,500 per year. The change raised the salary significantly from the $900 per year that most members received under the per diem rate.

Many members of the House either lost their seats or did not run for reelection later that year. Congress responded promptly by changing the pay structure back to a per diem rate in its next session. It was not until 1855 that Congress again established an annual salary for its members. The annual salary for both houses was set at $3,000.

Congressional salaries continued to grow over the years. (1) In 1989, Congress passed the Ethics Reform Act that restricted honoraria that House members could accept but boosted their pay. In 1991, the Senate also placed a ban on honoraria but increased salaries. Every member of Congress would now be entitled to an annual cost-of-living increase unless Congress voted against it.

Constituency Interests versus Ideology

Research in this area began with the question of whether constituency interests or political ideology was more important in determining a representative's vote on a particular piece of legislation. Kau and Rubin (1979, 1982) and Kalt and Zupan (1984) found measures of ideology, such as Americans for Democratic Action scores and the presidential vote by the representative's district, to be good predictors of voting behavior. Peltzman (1984, 1985) got similar results when investigating whether economic incentives of constituents were the most important factor that influenced how a representative would vote.

Another point of view suggests that members of Congress are elected because their ideology reflects the ideology of their constituents (see Rubin 2001). Poole and Rosenthal's (1997) treatment of roll call votes indicates that ideology is the best predictor of voting behavior. They found that most votes in Congress fall into one ideological dimension in which members may be lined up on a scale from "liberal" to "conservative." Therefore, the outcome of most roll call votes is determined by the placement of the "critical point" that is determined by the issue in question. Furthermore, they have found that their measure of ideology--known as "NOMINATE"--can explain around 80 percent of roll call votes. Poole and Rosenthal note that logrolling behavior is not likely to cause a problem in NOMINATE's ability to predict voting behavior because of the tendency of coalitions to be maintained over time.

Empirical Model

In Peltzman's (1976) model of political equilibrium, the politician acts as a political support maximizer in making wealth-transfer decisions. We expect that members of Congress are able to behave in a manner consistent with the self-interest argument when the threat of losing their seat is lower. The same framework can be applied to analyze voting behavior on congressional pay raises: How much can I increase my salary before I lose my seat?

Prior models of voting behavior included two categories of variables: constituency preferences and ideology. The model we develop contains no measures of constituency interest because of the zero-sum nature of salary increases. Anything gained by members of Congress in the form of a pay raise is lost by the taxpaying voter in the form of higher taxes or less funding for current programs (McCormick and Turner 2001: 254). We believe that constituency interests will not play a role in a senator's decision because constituents are likely to view the pay raise as a direct transfer from the taxpayers to senators.

With respect to political ideology, prior studies have shown this variable to be the best estimator of how a senator will vote on any given roll call vote. Furthermore, Poole and Rosenthal's NOMINATE variable is the most comprehensive measure of ideology and is probably the best predictor of a senator's vote on any particular issue. If NOMINATE is significant, then we can conclude that ideology plays a role in Senate voting behavior on legislative pay. (2)

We hypothesize that a senator who enjoys greater job security is more likely to vote in favor of a salary increase. Thus, the senator's tenure in office, the length of time until reelection, and the percentage of the popular vote received in the last election are entered into the model as explanatory variables. The general model is therefore specified as follows:

(1) Yes or No vote = f (Tenure, Year Term Expires, Percent of Vote Received, NOMINATE).

The data are compiled from three Senate roll call votes on legislative pay: one roll call vote each from years 2001, 2002, and 2003.

The senator's vote will be classified as a dichotomous variable. That is, the senator either voted for higher pay (vote = 1) or for lower pay (vote = 0). This type of analysis is best conducted by employing a logistic regression. The predictor (independent) variables are Tenure, Year Term Expires, Percent of Vote Received in Last Election, and NOMINATE.

Empirical Results

The coefficients that were estimated with the logit regression are shown in Table 1. The magnitude of the coefficients is of little value in predicting the likelihood that a senator will vote for a pay raise. Only the sign and level of significance is relevant for our purposes. A positive and statistically significant coefficient indicates that the explanatory variable contributes to the likelihood of a vote for a pay raise.

The results indicate that the variable NOMINATE was not significant. Ideological representations that align the senators from liberal to conservative, such as NOMINATE, divide them almost perfectly by their party designation. However, when the vote is a question of salary, the votes come equally from both sides of the aisle. Of particular interest is that all the job security measures were both positive and statistically significant. The longer the senator had served, the more time left in the senator's current term, and the larger the senator's percentage of the popular vote in the last election, the more likely the senator was to go on record in favor of a pay raise.

The overall model performs well in predicting the salary votes. The model correctly predicted 74.4 percent of the senators' votes in this sample. Also, the entire model proved to be significant when tested with Wald and likelihood ratio tests.

Conclusion

In the majority of Senate roll call votes, a senator's vote can be predicted by a measure of ideology. Likewise, the senator's party affiliation is a good predictor of voting behavior. Such is not the case when the question involves a change in salary. When conflicting margins of self-interest are pitted against each other, the Senate as a whole behaves in a rational manner. Those politicians at greater risk of defeat in the next election allow the more secure members of the body to pass the pay raise. It is likely that logrolling behavior is present in this process. Empirical evidence points toward other characteristics being determinants in those decisions.

It is uncommon for members of Congress to deviate from their party in roll call votes. But votes that determine congressional salary levels seem to be one glaring exception to that rule. While members of the Senate desire higher levels of pay, our results suggest that senators behave in a manner that is influenced by political variables that include self-preservation.

References

Kalt, J. P., and Zupan, M. A. (1984) "Capture and Ideology in the Economic Theory of Politics." American Economic Review 74: 279-300.

Kau, J. B., and Rubin, P. H. (1979) Self-Interest, Ideology and Logrolling in Congressional Voting. Journal of Law and Economics 22: 365-84.

-- (1982) Congressmen, Constituents and Contributors. Boston: Martinas Nijhoff.

McCormick, B. E., and Turner, C. S. (2001) "On Legislatures and Legislative Efficiency Wages." In W. F. Shughart II and L. Razzolini (eds.) The Elgar Companion to Public Choice, 240-57. Northampton, Mass.: Edward Elgar.

Peltzman, S. (1976) "Toward a More General Theory of Regulation." Journal of Law and Economics 19: 211-40.

-- (1984) "Constituent Interest and Congressional Voting." Journal of Law and Economics 27: 181-210.

-- (1985) "An Economic Interpretation of Congressional Voting in the Twentieth Century." American Economic Review 75: 656-75.

Poole, K. T., and Rosenthal, H. (1997) Congress: A Political-Economic History of Roll Call Voting. New York: Oxford University Press.

-- 107th Senate Rank Ordering. Accessed 22 November 2004 (www.voteview.com/SEN107.htm).

-- 108th Senate Rank Ordering. Accessed 22 November 2004 (www.voteview.com/sen108.htm).

Rubin, P. H. (2001) "Ideology." In W. F. Shughart II and L. Razzolini (eds.) The Elgar Companion to Public Choice, 328-36. Northampton, Mass.: Edward Elgar.

(1) For a history of Senate salaries, see www.senate.gov/artandhistory/history/common/briefing/senate_salaries.htm.

(2) NOMINATE is highly correlated with other measures of political ideology. For example, when NOMINATE is regressed with the National Taxpayers' Union grade as an explanatory variable, the adjusted R-square is 0.796. Members of Congress who are fiscal conservatives are also social conservatives, so NOMINATE is a sufficient measure of ideology.

Jim Couch is Professor of Economies at the University of North Alabama; Brett King is the State Farm Professor of Risk Management at the University of North Alabama; and Taylor Stevenson is Assistant Professor of Economies at Austin Peay State University. They thank Jennifer Huddleston for research assistance.

TABLE 1
LOGIT REGRESSION RESULTSDEPENDENT
VARIABLE: SENATORS VOTE ON PAY RAISE

Independent Chi-
Variables Coefficients Error Square Pr > ChiSq

Intercept -686.9 168.7 16.5776 <0.0001
NOMINATE 0.0054 0.0048 1.2558 0.2624
Percent Vote
 Received 0.0139 0.0056 6.2317 0.0125
Year Term
 Expires 0.3419 0.0840 16.5488 <0.0001
Tenure 0.0840 0.0168 25.0137 <0.0001