Abstract
Despite the centrality of voting costs to the paradox of voting, little effort has been made to measure these costs accurately, outside of a few spatially limited case studies. In this paper, we apply Geographic Information Systems (GIS) tools to validated national election survey data from New Zealand. We calculate distance and travel time by road from the place of residence to the nearest polling place and combine our time estimate with imputed wages for all sample members. Using this new measure of the opportunity cost of voting to predict turnout at the individual level, we find that small increases in the opportunity costs of time can have large effects in reducing voter turnout.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
Feddersen (2004) considers a two-candidate election with five million voters and candidate 1’s expected vote share of 50.1%; the benefit to a voter who prefers candidate 2 must be more than eight billion times greater than the cost of voting, in order for expected benefits to outweigh costs. Such a ratio of benefits is inconceivable for any voter, with typical stakes in an election (in terms of the amount of compensation needed for indifference between who wins) of only a few thousand dollars (Tollison and Willett 1973).
Tollison and Willett (1973) point out that predictions about whether higher or lower income individuals have a greater incentive to vote depend on both income and substitution effects, and so will be a priori unclear.
For example, Matsusaka and Palda (1999) use surveys from four national elections in Canada where the rate of self-reported voting in their samples was 16 percentage points higher, on average, than the actual turnout in each election. Similarly, 22% of non-voters in local government elections in Sweden claimed to have voted when they were surveyed (Karp and Brockington 2005).
Even more non-randomness may come from the tendency for false reports of voting to be more likely for some demographic groups than others (Silver et al. 1986).
Even though registration is compulsory, there is no legal compulsion to vote, unlike the case in neighboring Australia.
The exception is that a party that would not be represented in the Parliament if it got less than the threshold 5% of the party vote can still have Members of Parliament to match its vote share, if it wins at least one electorate seat. While voters in that particular electorate could have more than average influence, this situation occurs only rarely. In the empirical analysis below, the results are robust to including electorate fixed effects.
In the 2005 national election studied here, less than 7% of votes from resident voters were cast before Election Day.
Evidence for this assumption also comes from the low share of Special Votes, which are for people voting outside of their home electorate on Election Day. In the 2005 election studied here, these were just 6.7% of total votes cast. Moreover, those who work on Saturdays must be allowed to leave their workplace no later than 3 pm for the purpose of voting (or leave for at least two hours if they have essential work that goes after 3 pm) and their employer cannot make deductions from the employee’s remuneration for the time taken off. See Sect. 162 of the Electoral Act, 1993.
Manhattan distance is defined as: d i =|(x 1−x 2)+(y 1−y 2)| where x, y are the longitude and latitude coordinates for the origin (1) and the destination (2).
A drawback of their study is that the data are from voter files, so are restricted to those citizens actually registered to vote, which creates a potential sample selection bias, and also limits the availability of predictor variables compared to what would be available in surveys.
In the raw data files, there are almost 3,700 observations but with missing values for some of the covariates used in our models we have a sample of n=3,005. The voting rate and the average distance from the polls are the same in the full sample and the estimation sample, supporting the assumption that the observations are missing at random.
In terms of physical area, a typical urban meshblock would, if perfectly square, have dimensions of just over 200 metres (0.13 miles).
These were batch processed using the traveltime command written for Stata by Ozimek and Miles (2011).
Skinner used a regression estimated on Consumer Expenditure Survey data to impute consumption estimates for households in the Panel Study on Income Dynamics.
This survey is run as a supplement to New Zealand’s main labor market survey (the Household Labour Force Survey) and hence is equivalent to the March CPS in the United States.
The following controls are included: age, gender, the personal and household income bracket, employment status, two-digit industry and occupation, highest level of education, ethnicity, urbanity and region. The R-squared from this regression is 0.386, indicating that bracketed income along with other socioeconomic characteristics are strong predictors of wage rates.
Details on this policy are available at: http://www.beehive.govt.nz/release/holidays-act-changes-announced (accessed October 1, 2011).
To allow for workers to be either under- or over-employed (hours less than or exceeding the equilibrium) it is necessary to use survey data that captures workers’ willingness to trade wages for leisure, and then use switching regression approaches. See Lee and Kim (2005) for an example. Such data are not available in New Zealand.
See for example, http://www.yourbalance.com.au/flexible-work-survey-highlights-worker-friendly-countries/ (accessed October 2, 2011).
However, even in the rural sector the maximum distances are not especially large, with the 99th percentile of road distances to the closest polling place being 22 kilometers (14 miles).
These results are also robust to including electorate fixed effects.
Approximately 1,400 polling places in 2005 were in schools but the Ministry of Education lists over 2,500 schools in their directory. Approximately 1,100 polling places in 2005 were in churches or community halls but an enumeration in 2004 found over 3,000 places of worship in New Zealand (Statistics New Zealand, 2007) and well over 1,000 town and community halls.
These data on the distance to schools were kindly provided to us by Jaime Pearce and are described in Pearce et al. (2006). The other facilities studied by Pearce et al. (2006) are health- and exercise-related, and so are less relevant to the sorting hypothesis, and do not include churches or community halls. The results of these additional probit models where distance to the nearest school was either added to or used in place of the distance to the nearest polling place variable are available from the authors.
Tables of predicted values for every integer value of distance, time and cost from zero to two standard deviations above the mean are available from the authors. Closer to the mean there is even less difference between the predictions from the different specifications.
References
Aldrich, J. (1993). Rational choice and turnout. American Journal of Political Science, 37, 246–278.
Brady, H., & McNulty, J. (2011). Turning out to vote: the costs of finding and getting to the polling place. American Political Science Review, 105, 115–134.
Cebula, R., & Toma, M. (2006). Determinants of geographical differentials in the voter participation rate. Atlantic Economic Journal, 34, 33–40.
Dyck, J., & Gimpel, J. (2005). Distance, turnout, and the convenience of voting. Social Science Quarterly, 86, 531–548.
Feddersen, T. (2004). Rational choice theory and the paradox of not voting. Journal of Economic Perspectives, 18, 99–112.
Frey, B. (1971). Why do high income people participate more in politics. Public Choice, 11, 101–105.
Gimpel, J., Dyck, J., & Shaw, D. (2006). Location, knowledge and time pressures in the spatial structure of convenience voting. Electoral Studies 25, 35–58.
Gimpel, J., & Schuknecht, J. (2003). Political participation and the accessibility of the ballot box. Political Geography, 22, 471–488.
Haspel, M., & Knotts, H. (2005). Location, location, location: precinct placement and the costs of voting. The Journal of Politics, 67, 560–573.
Hausman, J., Abrevaya, J., & Scott-Morton, F. (1998). Misclassification of the dependent variable in a discrete response setting. Journal of Econometrics, 87, 239–269.
International Monetary Fund (2008). Climate change and the global economy. In World economic outlook: housing and the business cycle (Chapter 4). Washington: IMF.
Karp, J., & Brockington, D. (2005). Social desirability and response validity: a comparative analysis of over-reporting voter turnout in five countries. The Journal of Politics, 67, 825–840.
Larson, D. (1993). Separability and the shadow value of leisure time. American Journal of Agricultural Economics, 75, 572–577.
Lee, K., & Kim, I.-M. (2005). Estimating the value of leisure time in Korea. Applied Economics Letters, 12, 639–641.
Matsusaka, J., & Palda, F. (1999). Voter turnout: how much can we explain? Public Choice, 98, 431–446.
McNulty, J., Dowling, C., & Ariotti, M. (2009). Driving saints to sin: how increasing the difficulty of voting dissuades even the most motivated voters. Political Analysis, 17, 435–455.
Niemi, R. (1976). Costs of voting and nonvoting. Public Choice, 27, 115–119.
Ozimek, A., & Miles, D. (2011). Stata utilities for geocoding and generating travel time and travel distance information. The Stata Journal, 11, 106–119.
Pearce, J., Witten, K., & Bartie, P. (2006). Neighborhoods and health: a GIS approach to measuring community resource accessibility. Journal of Epidemiology and Community Health, 60, 389–395.
Pesaran, M. (1974). On the general problem of model selection. Review of Economic Studies, 41, 153–171.
Pesaran, M. (1982). Comparison of the local power of alternative tests of non-nested regression models. Econometrica, 50, 1287–1305.
Riker, W., & Ordeshook, P. (1968). A theory of the calculus of voting. American Political Science Review, 62, 25–42.
Silver, B., Anderson, B., & Abramson, P. (1986). Who over-reports voting? American Political Science Review, 80, 613–624.
Skinner, J. (1987). A superior measure of consumption from the panel study of income dynamics. Economics Letters, 23, 213–216.
Statistics New Zealand (2007). Non-profit institutions satellite account: 2004. Wellington: Statistics New Zealand.
Tollison, R., & Willett, T. (1973). Some simple economics of voting and not voting. Public Choice, 16, 59–71.
Vowles, J. (2010). Electoral system change, generations, competitiveness and turnout in New Zealand, 1963–2005. British Journal of Political Science, 40, 875–895.
Acknowledgements
We are grateful for the financial support from Marsden Fund grant 07-MEP-003 and to comments from two anonymous referees and the associate editor. All remaining errors in this paper are those of the authors.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Gibson, J., Kim, B., Stillman, S. et al. Time to vote?. Public Choice 156, 517–536 (2013). https://doi.org/10.1007/s11127-011-9909-5
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11127-011-9909-5