Printer Friendly

Campaign spending in congressional elections.



Empirical studies of the effects of campaign spending on the outcome of Congressional elections have found that campaign spending by incumbents is not as effective as that of their challengers. Indeed, Kau and Rubin [1982! and Jacobson [1985 1987! find that incumbent spending reduces the number of votes they receive and Netter [1983! that it increases the challenger's votes. Incumbents seeking reelection, however, are very successful, winning nine times out of ten in the postwar era. If their campaign spending is so inefficient, why do incumbents continue to outspend their challengers in a majority of elections?

We argue that these researchers counterintuitive findings are the result of using models which do not adequately correct for the simultaneous determination of the election outcome and campaign contributions, and which do not include an electoral check on incumbent behavior. (1) In their studies Jacobson and Netter, as well as Lott [1986! and Laband and Lentz [1985!, assume the incumbent's ability to repel challengers is independent of his or her behavior while in office. Few would agree that the electoral check is so ineffective. We develop a simultaneous model to correct these deficiencies. It captures the effects of campaign expenditures, tenure, the consistency of the incumbent's votes with his constituents' preferences, and the size of the voter turnout on the votes received by incumbents and challengers in four congressional elections between 1978 and 1984. The model is very robust over these four elections. The use of primary election data insures model



We assume that the platforms of any two candidates for a political office can be measured along a single left-right spectrum. Voters know their preferred position on the dimension but cannot observe the positions of the two candidates with certainty because of rational ignorance, a theory accepted since Downs [1957!. Following Congleton [1986!, we assume voters are risk-averse and receive more utility from candidates with positions closer to the voters' preferred position, but lose utility as the candidate's position becomes more uncertain. Voters compare the candidates and vote for the one that provides them greater expected utility.

Voters can observe the incumbent's performance and are thus less uncertain of his views and credibility. Uncertainty should decrease with an incumbent's tenure in office. Tenure provides voters with knowledge of the incumbent's position and the incumbent with political capital. This accumulation of political capital may make it more difficult for challengers to mount a successful campaign. In this sense, tenure acts as a barrier to entry. (2)

If all incumbents were at the same point on the spectrum as the median voter of their districts, all incumbents would win reelection. If all potential challengers were aware of this fact, they would not challenge since campaigns are costly. But one observes many challengers and some defeated incumbents. So we proceed from the hypothesis that challengers enter the race when they perceive a gap between the position of the incumbent and that of the median voter and assume this position as their platform. A large difference between the incumbent's position and the median voter's position increases the likelihood of a successful challenge. If voters were certain of the positions of both candidates, and the challenger's assessment was correct, the challenger would win.

But challengers are usually less well known than incumbents, and this additional uncertainty gives incumbents an advantage. Thus the challenger must convince the voters that his or her position is closest to the median voter's position, and advertising is the primary means to that end. (3) Repetition of the message reduces the voters' uncertainty over a challenger's position. The closer the incumbent is to the median voter, the less is the difference in expected voter utility from the two candidates, and the larger will be the amount of advertising necessary to generate a given number of challenger votes.

Strong challengers induce increased incumbent spending to solidify a position on the spectrum. (4) Past incumbent actions and advertising have already reduced voter uncertainty additional expenditures are not likely to be as effective as previous expenditures, as found in Jacobson and Kernell [1982!. Advertising is funded largely by campaign contributions, and Grier and Munger [1986!, Kau and Rubin [1982!, and Poole, Romer and Rosenthal [1987! find that these contributions are not independent of the number of votes the candidate is likely to receive.

Political advertising potentially benefits candidates through two channels. Advertising can inform voters of a candidate's position or can increase uncertainty about an opponent's, reducing votes for the opponent. At the same time, contributors' expectations about the strength of the challenge are formed from past experiences with the candidates, from polls taken during the campaign, and performances in primary elections. The closeness of the race and the probability of a successful challenge can be estimated. This implies that the votes and contributions received by the candidates are determined simultaneously as expectations of the election outcome change during the campaign. (5)

A simultaneous model designed to capture the relationship between the votes and expenditures of candidates for political office is presented in equations (1) and (2)

(1) Candidate votes = a + b(own spending) + c(opponent's spending) + d(tenure) + e(democrat tenure) + f(consistency) + g(democrat consistency) + f(total votes [l/times - 1 if republican!) + g(own primary votes) + error.

(2) Candidate Spending = a + b(challenger votes) + c(tenure) + d([tenure.sup.2!) + e(democrat tenure) + f(democrat [tenure.sup.2!) + g(consistency) + h(democrat consistency) + i(dummy: dem=1 rep=0) + j(percent of own primary votes) + error.

Votes for each candidate are expected to be positively related to their own campaign spending and negatively related to the opponent's spending. (6) Tenure is likely to prove beneficial for the incumbent because of its importance in securing committee assignments and because it indicates "brand loyalty" by the constituents. (7) This will not draw votes away from challengers, but rather enhances the vote total gained by incumbents. Consequently, tenure measures are not included in the challenger vote equations.

Incumbents who vote the preferences of their constituency would be voting the median voter's position. An incumbent's consistency with voter preferences can be readily estimated. If a congressional district voted more strongly for the President than did the nation as a whole, the representative will be closer to the median position of the district if he or she votes in support of the President more often than the average representative. Our vote equations include a measure designed to capture the effects of the consistency of the incumbent's voting pattern. The deviation in the incumbent's presidential support (PS) rating from the average presidential support rating in all congressional districts, times the deviation in the percent of votes for the president in the incumbent's district (VS) from its average is our measure of consistency. (8)

Consistency of incumbent i = ([PS.sub.i! - mean PS) * ([VS.sub.i.! - mean VS)

Large positive coefficients for consistency in the incumbent vote equations identify an incumbent whose position is close to the median negative values indicate that challengers can establish positions closer to the median voter. High positive values of consistency should be correlated with higher incumbent vote shares. The sign on this variable should be positive in the incumbent vote equation and negative in the challenger equation.

Campbell [1960! found that Republicans are more likely to vote than are Democrats, ceteris paribus. Republicans therefore benefits from smaller turnouts. To capture the effect of turnout on the election outcome, total votes cast, multiplied by -1 if the incumbent is Republican, is included in the vote equations. A positive sign on this measure would support the proposition that Republicans benefit from low turnouts.

The total number of primary votes a candidate receives is included in the general election vote equations. A primary identifies a group of voters relatively committed to a candidate. They constitute a political base for a candidate from which defections seem relatively unlikely. This variable is included to identify the general election vote equations. (9)

The unique conditions of each election may affect candidates of opposing parties differently. Consequently, we include the length of tenure of Democratic incumbents and the consistency of Democratic incumbents with their constituents' preferences in the incumbent vote equation. The a priori signs on the variables are indeterminate. Squared values of the expenditure and tenure measures are included to test for diminishing returns.

Incumbent and challenger expenditures are modelled as a function of the votes received by incumbents and challengers, the tenure of the incumbent, the position of the incumbent relative to the median voter, the party of the incumbent, and the percentage of votes received by the candidate in question in the party primaries. Both canditates and contributors use all available information to form rational expectations of the challenger's strength, and funds will tend to flow to close races. We posit this to be the information set: Incumbent tenure is a measure of voter certainty over the incumbent's position. Longer tenure should induce higher contributions. Facing an incumbent whose positions on issues are farther from the median voter should improve the challenger's fund-raising efforts. Moreover, when the gap between the position established by the incumbent and that of the median voter is large, we anticipate that incumbents will compensate with greater spending. The percentage of votes received by the candidates in the primary election is included in the expenditure equations to (1) identify the equations and (2) to capture the effects of the popularity of the candidate within the party on campaign spending. The closer the primary challenge, the greater the spending. If one party provides more support to its candidates than the other, we can expect higher expenditures for all the party's candidates, ceteris paribus. We include a dummy variable that is equal to one if the candidate is a Democrat in the expenditure equations. Interaction terms for party with tenure and incumbent position are also entered as explanatory variables. Jacobson and Kernell [1982! discuss the steering of funds by party committees towards particular candidates. Republicans are noted for disbursing campaign funds toward cadidates in close campaigns, against freshmen Democratic incumbents, and to protect an incumbent who may not be voting consistently with constituent preferences. We anticipate that the interaction terms will account for these selective interventions.


Estimates of the second stage of the vote equations for the congressional election years 1978, 1980, 1982, and 1984 are reported for incumbents and challengers in Table I. (10) The estimates of the effects of campaign spending on the votes received are consistent with the predictions of the model. Incumbent expenditures have the effect of increasing the number of votes incumbents receive at a decreasing rate. For the same years, challenger expenditures reduce the votes received by the incumbent at a decreasing rate, but the negative effect is less than the positive effect of incumbent spending. Diminishing returns to incumbent and challenger expenditures on incumbent votes are not found over the range of 1978 reported expenditures.

Challenger votes are affected to a greater extent by their own spending than by incumbent spending. Although incumbent spending is negatively related to challenger votes in every year but 1980, the coefficients are not always significant (they were not significant in 1980) at the usual acceptance levels. Diminishing returns to challenger spending on their own votes is observed only during the 1978 election. The best estimates indicate diminishing returns to incumbent spending on challenger votes in every year except 1980.

We computed the marginal products of incumbent and challenger spending on incumbent vote pluralities for each election year by subtracting the challenger vote equations from the incumbent vote equations for each year. Partial derivatives of the vote difference with respect to incumbent and challenger spending were then taken. Diminishing returns to incumbent and challenger spending on pluralities are evident in all elections. Moreover, the returns to expenditures on incumbent vote pluralities become negative at expenditure levels well below those reached by challengers in many races. This is not expected in the long run however, elections are held in the short run, where overspending is much more likely to occur.

Table II reports the spending levels at which negative returns to advertising appear and the percent of candidates exceeding the level. The percent of candidates "overspending" in 1982 is particularly high because zero-return levels of spending in that year are low. Redistricting between 1980 and 1982 led to a significant change in the composition of some districts, and these were removed from the sample. Districts in which only minor changes were made were not eliminated from the sample, and their inclusion may affect the results. (11) On the other hand, Republican freshmen may have spent overzealously in 1982 in their effort to hold the seats they won in the 1980 Reagan landslide.

The marginal products of campaign expenditures on the difference between incumbent and challenger votes are presented for mean and median levels of spending in Table III. Except in 1980, the marginal productivity of incumbent spending is greater than the marginal productivity of challenger spending. (12) This result, which is contrary to Lott [1986!, suggests that the productivity of incumbent spending is enhanced by the electorate's experience with the incumbent or by the value of past incumbent advertising.

Tenure increases the votes received by incumbents. Incumbents are more likely to deliver promised benefits to their constituents, and hence more desirable than challengers who promise the same benefits. There are diminishing returns to tenure, which for Repulicans becomes negative sometime during their eighth term (in 1982, zero returns occur in the sixth term).

The separate effects of tenure on a Democratic incumbent are determined by summing the coefficients on tenure and Democrat tenure (and their squares). In all cases Democration incumbents benefit from tenure, albeit less so than Republicans. This may reflect the use of party labels by voters to determine the positions held by incumbents and to punish those less likely to have positions consistent with the median voter's positions. The longer Democratic incumbents stay in office, however, the less harmful is their party affiliation (the squared interaction term has a positive sign).

Incumbents are rewarded at the polls for casting votes consistent with the preferences of the median voter. Tenure allows incumbents to vote their preferences, but at some cost. (13) Closeness of representative voting records with constituent preferences does not reduce challenger votes. The shift term, Democrat consistency, is always negative, and is significant in reducing challenger votes in three of four elections (1984 is a borderline case). A negative sign shows that Republican challengers

fared relatively poorly against Democrats whose votes were consistent with their district's preferences. If voters are ill-informed, as theories of rational ignorance predict, they may use party identification to deduce a candidate's view on policy issues and the likelihood that specific platforms will be implemented.

Other variables in the model are either of secondary interest or were included for specification purposes. The interaction of party and turnout was included in the vote equations to test whether high or low turnouts systematically help Republicans. The finding that Republicans fare better in low turnout districts, while Democrats do better in high turnout districts, is consistent with the previous research of Campbell [1960!. Primary election data were included to identify the model. Higher vote totals for incumbents in primaries are correlated with lower votes in the general election. This is reasonable: an incumbent does not get large numbers of primary votes unless there is a challenge. A primary challenge indicates a weak incumbent. The correlation between the incumbent consistency measure and the number of votes an incumbent received in the primary election is significant at the 10 percent level in 1978 and 1980, and at the 5 percent level in 1982 and 1984. Further study of this effect is beyond the scope of this paper, but remains a viable avenue for future research.


The model we use to estimate the vote-expenditure relationship in Congressional elections succeeds in solving the simultaneity problem that has plagued research in this area. Contrary to previous results, we find the expected relationship between the candidates' campaign spending and their opponents' votes.

The inclusion of the consistency variable eliminates the counterintuitive findings Kau and Rubin [1982! and Jacobson [1984 1987! that higher incumbent expenditure reduces incumbent votes. We find that an incumbent is rewarded for voting the wishes of the median voter and is punished for failing to do so. The scope for deviation is constrained by the ability of the challenger to advertise that fact.

Both challengers and incumbents can advertise effectively. The marginal product of advertising by incumbents may be higher because of experience, but the difference is usually small and not persistent. The cost of acquiring an additional vote through advertising also varies. In 1978, the cost of the additional vote for the incumbent was $1.79, while for the challenger it was $2.43, a difference of 36 percent. It was never that large again: the challenger had a 15 percent advantage over the incumbent in 1980, while the incumbent's advantage was 7 percent and 15 percent in 1982 and 1984. The error bands are too large to suggest that any of these differences are statistically significant. Incumbent and challenger spending may produce votes at the same rate. While incumbents possess an absolute advantage, this need not imply that the marginal benefit of their expenditures is higher than that for challengers.

Our results do not support the view that advertising is a barrier to entry into politics. The continued success of incumbents results from their serving the wishes of their constituents and from tenure. These are the "barriers" confronting challengers. Given a risk-averse voter, neither is undesirable.

(*1) Associate Professor and Professor of Economics, St. Cloud State University. We thank John Lott, John McArthur, Masoud Moghaddam, Michael White, Thomas Willett and two anonymous referees for useful suggestions to earlier drafts.

(1.) Welch [1981! models votes and expenditures simultaneously for the 1972 election only and excludes other important explanatory variables.

(2.) Lott [1986! and Jacobson and Kernell [1982! identify name recognition and tenure and Laband and Lentz [1985! heredity as barriers to entry into the political industry. Lott notes that past campaign spending and tenure contribute to the incumbent's name recognition. Each dollar spent by a challenger, therefore, will likely have a lower return the greater the level of past incumbent spending and the longer the tenure. Lott presents evidence which tends to support his position. But the same argument can be made for challengers who have held another office or who are sons or daughters of successful politicians, as Laband and Lentz [1985! find. for this reason, we do not consider the effects of past campaign expenditures. The effects of past advertising will be proxied by tenure.

(3.) Dougan and Munger [1989! and Lott and Reid [1989! emphasize the importance of ideology in binding a candidate to support certain policies. But the distinction of whether candidates are search or experience goods is not relevant to our empirical estimates: the analysis would be the same for either viewpoint.

(4.) Glazer and Robbins [1984! show that changing one's position on the scale increases voter uncertainty and is therefore counterproductive.

(5.) Jacobson and Kernell [1982! consider two types of campaign contributors. One group gives to the candidate that represents their viewpoint, regardless of the candidate's electoral chances. The contribution is an act of consumption, like contributing to a favorite charity. The other contributes to influence legislation. Contributions flow to candidates expected to win since only winners can supply legislation.

(6.) We use vote totals rather than vote shares to account for abstaining votes. A vote gained by one candidate from spending may not simultaneously reduce the vote total of the other. Higher spending by both candidates should decrease abstinence in this model.

(7.) Tenure can increase brand loyalty if more senior representatives are able to deliver more government goods to their districts. The effects of tenure are beneficial whether politicians are search or experience goods.

(8.) Kau and Rubin [1984! find that both the ideology and the economic interests of constituencies influence their voting choices. Glazer and Robbins [1984! show that senior senators moved towards the position of the newly elected senator if the newcomer unseated an incumbent. Peltzman [1984! concludes that the ideological interests of one's constituents are determined by the economic interests of the district. Voting against the "ideological interest" of one's constituency would appear to have a negative effect on re-election success.

(9.) We use total expenditures rather than expenditures only for the general election because the expenditures for the two campaigns cannot be reliably separated. The National Election Commission does not separate primary and general election campaign spending because federal law allows for contributions subsequent to the expenditures, i.e., a candidate can borrow and spend, then obtain funds from supporters subsequent to the primary campaign. This information was obtained in a telephone conversation with an employee of the Federal Election Commission in spring 1987.

(10.) The squared values for incumbent and challenger expenditures in the vote equations are the squares of the predicted values for these variables from the first-stage estimates. Estimates of the expenditure equations, which are not reported because of space limitations, are available upon request. The data used to estimate the model are from Barone and Ujifusa [1983 1985! and Ornstein et. al. [various issues!.

(11.) It seems likely that the excluded districts have a higher zero return level, as many constituents are less familiar with the "incumbent". This would bias our estimates downward and explain the higher level of overspending in 1982. But we cannot correct for this problem, as vote share for president from the realigned districts are not available. We are indebted to John McArthur for illuminating this point.

(12.) The marginal products of incumbent and challenger spending are greater in presidential election years. This may reflect the greater public interest and participation in the electoral process during these years. The computations in Tables II and III use the often insignificant coefficients on the incumbent expenditure variable in the challenger vote equations. This has the effect of overstating the negative impact of incumbent spending on challenger votes and the marginal productivity of incumbent expenditures. Forcing these coefficients to zero has little effect on the results.

(13.) They may also vote against the wishes of the district when doing so gains the incumbent campaign funds from special interests, as Grier and Munger [1986! find.


Barone, Michael and Grant Ujifusa. The Almanac of American

Politics. Washington, D.C.: National Journal, 1983 1985.

Campbell, Angus. "Surge and Decline: A Study of Electoral Change." Public Opinion Quarterly, Fall 1960, 397-418.

Congleton, Roger D. "Rent-Seeking Aspects of Political Advertising." Public Choice 49(3), 1986, 249-63.

Dougan, William R. and Michael C. Munger. "The Rationality of Ideology." Journal of Law and Economics, April 1989, 119-42.

Downs, Anthony. An Economic Theory of Democracy. New York: Harper and Row, 1957.

Glazer, A. and M. Robbins. "How Elections Matter: A Study of U.S. Senators." Public Choice 46(2), 1984, 163-72.

Grier, Kevin B. and Michael C. Munger. "The Impact of Legislator Attributes on Interest-Group Campaign Contributions." Journal of Labor Research, Fall 1986, 349-61.

Jacobson, Gary." Money and Votes Reconsidered: Congressional Elections, 1972-1982." Public Choice 47(1), 1985, 7-62.

_____. The Politics of Congressional Elections, 2nd ed. Boston: Little, Brown, 1987.

Jacobson, Gary and samuel Kernell. Strategy and Choice in Congressional Elections. New Haven, Conn.: Yale University Press, 1982.

Kau, James and Paul H. Rubin. "Economic and Ideological Factors on Congressional Voting: The 1980 Election." Public Choice 46(3), 1984, 385-88.

_____. Congressmen, Constituents and Contributors. Boston: Martinus Nijhoff, 1982.

Laband, David N. and Bernard F. Lentz. "Favorite Sons: Intergenerational Wealth Transfers among Politicians." Economic Inquiry, July 1985, 395-414.

Lott, John R., Jr. "Brand Names and Barriers to Entry in Political Markets." Public Choice 51, 1986, 87-92.

Lott, John R., Jr. and W. Robert Reid. "Shirking and Sorting in a Political Market with Finite-Lived Politicians." Public Choice 61, April 1989, 75-96.

Netter, Jeffery M. "Political Competition and Advertising as Barriers to Entry." Southern Economic Journal, October 1983, 510-20.

Ornstein, Norman J., Thomas E. Mann, Michael Malbin, Allen Schick and John F. Bibby. Vital Statistics on Congress. Washington, D.C.: American Enterprise Institute for Public Policy Research, various issues.

Peltzman, Sam. "Constituent Interest and Congressional Voting." Journal of Law and Economics, April 1984, 181-210.

Poole, Keith T., Thomas Romer and Howard Rosenthal. "The Revealed Preference of Political Action Committees." American Economic Review 77, May 1987, 298-302.

Welch, William. "Money and Votes: A Simultaneous Model." Public Choice 36(2), 1981, 209-34.
COPYRIGHT 1991 Western Economic Association International
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1991 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Banaian, King; Luksetich, William A.
Publication:Economic Inquiry
Date:Jan 1, 1991
Previous Article:Determinants of voluntary overtime decisions.
Next Article:Short-run pricing strategies to increase corporate average fuel economy.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters