Principal investigator: Viviane Pires Ribeiro
Paper Title: Voting technology, political responsiveness, and infant health: Evidence from Brazil
Authors: Thomas Fujiwara
Location of the Intervention: Brazil
Sample Size: 5281 Brazilian municipalities
Main theme: Economic Policy and Governance
Main Variable of Interest: Voting technology
Type of Intervention: Analysis of electronic voting technology in Brazilian elections.
Methodology: Discontinuous regression
Exploring a regression discontinuity design in the attribution of voting technology in the 1998 election, Fujiwara (2015) estimates the effects of electronic voting technology in reducing a “simple,” yet still important, obstacle to political participation: difficulty in operating paper ballots. The results indicate that electronic voting did indeed cause a significant emancipation of less educated voters, leading to the election of more left-leaning state candidates, increased spending on public health, utilization (prenatal visits), and child health (birth weight). The results are consistent with predictions from political economy models and demonstrate that electronic voting can promote the political empowerment of low-income populations and raise their living standards.
Evaluation Context
Filling out a ballot might be a trivial task for educated citizens in developed countries, but the same is not true in Brazil, where 23% of adults over 25 are “unable to read and write a simple message” and 42% have not completed the 4th grade. Furthermore, paper ballots required Brazilian voters to write down the candidate's name or electoral number and provided no visual aids. The combination of these two factors resulted in a significant number of blank and error-ridden votes in the elections, generating a large number of residual votes (those not assigned to a candidate and discarded from the results count).
In the mid-1990s, the Brazilian government developed electronic voting technology to replace paper ballots. Although its introduction aimed to reduce the time and costs of counting votes, other features of the technology, such as the use of candidate photos as a visual aid, the use of error messages for voters about to cast residual votes, and step-by-step guidance of the voting process, facilitated voting and reduced errors.
State elections in Brazil are held every four years, and all states share the same election date. In the 1990 and 1994 elections, only paper ballots were used. In the 1998 elections, the electoral authority decided that only municipalities with more than 40.500 registered voters (as of 1996, when the municipal elections took place) would use the new electronic technology, while municipalities below that limit would use paper ballots. This allocation rule was adopted due to the manufacturer's limited production capacity and economies of scale in distribution across large municipalities. In the following election, electronic voting became the sole method of vote collection.
Intervention Details
Exploring a regression discontinuity design in the attribution of voting technology in the 1998 election, Fujiwara (2015) studies the introduction of electronic voting technology in Brazilian elections, more specifically, the author estimates the effects of electronic voting technology in reducing a “simple”, but still important, obstacle to political participation: difficulty in operating paper ballots.
Information on voter registration, participation, and election results at the municipal level was obtained from the federal electoral authority. The institution also published reports listing the municipalities that used electronic voting, showing near-perfect compliance with the discontinuous allocation rule. All 307 municipalities (out of 5281) above the 40.500 voter threshold used electronic voting in 1998. Additional data on municipal characteristics are from tabulations of the 1991 Brazilian Census.
According to the author, the main result of interest is the number of valid votes (i.e., non-residual votes). A vote is considered valid if, and only if, it can be attributed to a specific candidate or party and then counted in their share of the vote. An invalid vote is defined as a residual vote. Therefore, the vote share is equal to the sum of valid and residual votes. A vote cast on a paper ballot is considered residual if it is left blank or if the name or number written on the ballot does not correspond to a candidate. In electronic voting, a residual vote can be cast by pressing the “blank” button or by typing and confirming a number that does not correspond to any candidate.
Methodology Details
The state-level results highlight the fact that discontinuous allocation in the 1998 elections created specific and unusual differences in exposure time between states. The progressive introduction of the new technology was carried out in three consecutive elections held in 1994, 1998, and 2002.
In 1994, only paper ballots were used. In 1998, there was a discontinuous allocation, with municipalities above 40.500 registered voters using the new electronic technology, while municipalities below that limit used paper ballots. In 2002, only electronic voting was used. This schedule implies that the evolution of electronic voting in a state is entirely determined by a cross-sectional variable that remains constant over time: the proportion of voters living in municipalities above the cutoff point for its use in 1998.
If a state has S% of its voters living above the cutoff threshold, then S% of its voters switched from using paper ballots to using electronic voting technology between the 1994 and 1998 elections. Furthermore, between the 1998 and 2002 elections, the remaining (1-S)% of voters switched to electronic voting. Consequently, states with higher voter turnout above the 40.500 voter threshold experienced the most emancipatory effects of electronic voting before states with lower turnout. Intuitively, the empirical strategy tests whether the outcomes of interest follow this same pattern.
Results
Estimates that exploit a regression discontinuity design embedded in its progressive introduction over time indicate that electronic voting reduced residual votes (those full of errors and uncounted votes) and, in fact, emancipated a large fraction (11%) of the electorate. This greater political participation of voters with lower education (low-income population) generated: (1) an increase in the number of elected state candidates who are themselves less educated; and (2) a diversion of government spending to public health, a policy that is particularly beneficial to the low-income population; leading to (3) better use of health services (prenatal visits) by mothers with lower education and (4) a reduction in the occurrence of births with low birth weight newborns in this group. However, no effects were found on the use of health care by mothers with higher education or on the weight of their newborns.
Estimates indicate that the emancipation of approximately one-tenth of Brazilian voters increased the share of state budgets spent on healthcare by 3,4 percentage points, raising spending by 34% over an eight-year period. It also increased adequate prenatal care for mothers with lower levels of education by 7 percentage points and reduced the prevalence of low birth weight by 0,5 percentage points (respectively, a change of 19% and -6,8% from the sample averages).
Lessons in Public Policy
The study conducted by Fujiwara (2015) quantifies the effects of voting technology in Brazilian state elections, demonstrating that, by facilitating the operation of the ballot boxes, it has indeed emancipated a significant number of voters (mostly voters with lower levels of education). Estimates exploring the unique pattern of gradual introduction of technology in the states over time suggest that, as predicted by political economy models, it has diverted government spending to healthcare, which is particularly beneficial for low-income citizens. Positive effects on both the use of healthcare services (prenatal consultations) and newborn health (low birth weight births) are also found for mothers with lower levels of education, but not for those with higher levels of education.
Fujiwara (2015) then argues that this emancipation of less educated citizens has indeed affected politics in a way consistent with theories of political economy redistribution and democratization. Since discontinuous allocation was observed in state elections, the author focuses on state government spending, particularly in an area that disproportionately affects the less educated: healthcare. That is, low-income Brazilians rely primarily on a publicly funded system for healthcare services, while middle- and high-income voters are more likely to use private services.
The population with lower levels of education therefore has relatively stronger preferences for public health services, and a change in healthcare spending can be interpreted as redistribution to those with lower incomes. Consistent with this interpretation, the author also found evidence that electronic voting increased the number of prenatal consultations by healthcare professionals and reduced the prevalence of low birth weight (below 2.500 g) births among women with lower levels of education.
Finally, the author makes a small caveat: the estimated effects may depend on a particular context. The level of literacy required to operate paper ballots is higher in contexts with more candidates than in contexts with fewer candidates, where the candidates' names and photos can be listed. The other results may also depend on characteristics of the Brazilian political and health systems. However, the study exemplifies how increased political participation of disadvantaged groups can change policymaking and affect outcomes, and, more importantly, provides evidence supporting general mechanisms in redistributive policy.
References
FUJIWARA, Thomas. Voting technology, political responsiveness, and infant health: Evidence from Brazil. Econometrics, v. 83, no. 2, p. 423-464, 2015.
