The Effects of Discretionary Federal Spending on Parliamentary Election Results
Evans, Thomas A., Economic Inquiry
Three compelling reasons to expect that discretionary federal spending will have a different impact on incumbent majority and opposition candidates in the Canadian parliamentary system of government are: only the prime minister and his appointed Cabinet may propose spending or taxing initiatives; the electorate only votes directly for their local Member of Parliament (MP); and, if the prime minister declares a vote-of-confidence all majority party MPs must vote with the prime minister or leave the party. These three characteristics, combined with the fact that the Opposition does not have enough votes to successfully prevent passage of legislation, make it very difficult for opposition MPs to influence discretionary spending. If the electorate is aware of this situation, majority party candidates, regardless of incumbency, stand to reap the electoral benefits of additional federal discretionary spending. Furthermore, opposition incumbents will lose votes as a consequence; if majority candidates receive additional votes, to some extent this must be at the expense of votes for opposition incumbents. The purpose of this paper is to empirically test these two hypothesis.
The idea that discretionary spending can be used to influence election outcomes verges on the trite. Mass media discussions are rife with accusations of pork barrel politics, this idea is implicitly contained the entire interest group theory of political decision making, and it is formally modeled in papers such as Weingast, Shepsle and Johnsen (1981). Despite the general agreement on the electoral motivations of discretionary spending, empirical findings have been relatively unsuccessful in documenting either important or significant effects of discretionary spending, or various other district level explanatory variables. Typical of this literature, Jacobson (1978) finds a negative and small effect of incumbent campaign spending, and Johannes and McAdams (1981) find no effect of case work on election outcomes.
As explained by Fiorina (1981), Rivers and Fiorina (1989) and Stein and Bickers (1994), the presence of a simultaneous equation system has biased the empirical estimates. Since politicians are apt to shift resources towards those electoral districts where the election is expected to be closest, ordinary least squares will underestimate the true effect. Stein and Bickers demonstrate a negative correlation between the closeness of elections and the amount of pork barrel spending in Congressional races; in a more recent article, Case (2001) finds that federal block grants are negatively influenced by the expected closeness of the election in Albania. Despite the recognition of a simultaneity bias, attempts to overcome the problem have generally been unsuccessful. A notable exception is that of Levitt and Snyder (1997) who uses spending outside of a district but inside a state as an instrument for spending in the district. Levitt and Snyder find a positive and significant impact of discretionary spending on the share of the vote for incumbents in U.S. House elections.
This paper contributes to the discretionary spending and election literature in two ways. Using a newly created data set of Canadian federal spending at the electoral district level, the analysis of discretionary spending can be extended beyond the U.S. border and presidential system of government, to a parliamentary system. Additionally, the results contained herein provide a robustness check of Levitt and Snyder's methodology.
Using instruments similar to those proposed by Nagler and Leighley (1992) and Levitt and Snyder (1997), consistent estimates indicate that increases in federal government spending and national gross domestic product (GDP) per capita have positive and significant effects on the share of the vote received by majority incumbents, yet insignificant effects on opposition incumbents (The sign of the point estimates are negative). …