Distributive Politics and Voter Turnout

Abstract

Do distributive policies affect voter turnout? Drawing upon previous research on distributive politics and policy feedback, we hypothesize that more distributive federal grants increase the rate of voting. Our analysis using district-level data from 1993 to 2000 reveals that the larger amount and number of distributive federal grants allocated to congressional districts are associated with the higher percentages of voter turnout in the subsequent congressional elections. We address estimation problems that result from the reciprocal relationship between federal spending and voter turnout and of omitted variable bias by using an instrumental variable. Our findings imply that distributive policies advantage incumbent legislators partly because they are effective for mobilizing potential supporters.

Notes

Supplementary materials are available at http://www.tandf.co.uk/journals/fbep. These include findings referred to but not presented in the text that follows.

Campbell (2003) also argued that public policies affect how likely people are to be mobilized by interest groups.

We combined Arnold's data (1994) on the volume of news coverage for a large sample of legislators and newspapers in 1993 and 1994 and the data of federal grants allocated in 1993 and 1994. Our estimates indicate that the larger amount and number of total grants allocated to the district increases the number of coverage for the incumbent legislators. The estimation results are available among the supplementary materials on the publisher's website at http://www.tandf.co.uk/journals/fbep.

Voting-eligible population is defined as those who are citizens and over 18 years old.

We estimated the size of voting eligible population for 1994, 1996, and 1998 using the Census data in 1990 and 2000.

FAADS data are publicly available from the web site of the Census Bureau.

Unfortunately, the FAADS database includes no information on military expenditures and wages.

A large number of transactions in the FAADS database do not specify a congressional district number to which federal grants are awarded. These transactions are typically entitlement programs awarded to counties that are divided into several congressional districts. Following Stein and Bickers (1995), we apportion the amount and number of federal grants to multiple districts by the fraction of the total county population that lived in a particular congressional district. The data on the fraction of the total population are available from MABLE/Geocorr90 Geographic Correspondence Engine at http://mcdc2.missouri.edu/websas/geocorr90.shtml.

About 60% of programs are new grants between 1993 and 2000. The main results reported in the subsequent section hold even when we included the continuing grants in the analysis.

Formula grants include block grants, while direct payments include financial assistance for specified and unrestricted use.

To assess the possibilities that these outliers have an impact on estimation results, we take a natural log of the grant variables. We also exclude districts if the amount of grants is larger than $2,000 or the number of grants above six. The results are almost identical to those reported below.

Note that all grants allocated to the state are omitted from our dataset.

Note that at-large districts are dropped from our analysis because R̄ _sdt cannot be identified for them.

The demographic variables are estimated using the 1990 and 2000 Census data.

We cannot examine our hypothesis only for the open-seat districts because of lack of temporal variation within districts.