What are the impacts of adopting electronic voting machines in Brazil?

Responsible researcher: Adriano Valladão Pires Ribeiro

Article title: VOTING TECHNOLOGY, POLITICAL RESPONSIVENESS, AND INFANT HEALTH: EVIDENCE FROM BRAZIL

Article author: Thomas Fujiwara

Location of intervention: Brazil

Sample size: 558 municipalities

Major theme: Economic Policy and Governance

Type of Intervention: Effects of adopting electronic ballot boxes on the number of valid votes and public spending

Variable of Main Interest: Valid votes in elections

Evaluation method: Discontinuous regression

Policy Problem

The economic policy choices regarding the size of public spending and the direction in relation to which sectors of society will be prioritized are part of the political system. Greater political participation by the lower economic classes, for example, would lead to greater redistribution in their favor. In Brazil, the adoption of the electronic ballot box increased the participation of less educated and poorer voters, leading to the implementation of policies more targeted at this group.

Assessment Context

In the 1990s, Brazil began to adopt electronic ballot boxes in elections to replace paper ballot boxes. Filling out a voting form, which would be a trivial act for people with some level of education, was a challenge at the time, especially when 23% of adults were unable to read and write. The problem became worse since citizens were required to write the candidate's name to validate their vote and the instructions were in writing. This ended up generating a huge number of invalid and blank votes. On the other hand, the electronic ballot box contains error messages when the vote is counted as invalid or blank, shows the photo of the chosen candidate and has step-by-step guidance on how to vote. In other words, the electronic ballot box reduced the difficulties of voting and led to a reduction in blank and invalid votes.

The analysis of the effects of the introduction of the electronic ballot box takes place on the elections for the state legislative power of 1998. In these elections the electronic ballot box was used for the first time and, due to limited supply, only municipalities with more than 40,500 registered voters were able to use the new technology, that is, municipalities below this cutoff line continued to use paper. Next, it was assessed whether the inclusion of less educated voters changed the direction of public policies. The states with the highest proportion of voters in municipalities with more than 40,500 registered voters would be those most impacted by the introduction of the electronic ballot box. It is then possible to compare the trajectory of public spending between states to verify the effects of the greater participation of the less educated population in the electoral process.

Intervention Details

Information about the elections, such as the number of registered voters, how many actually voted and the results, is made available by the Superior Electoral Court for each municipality. Of the 5281 municipalities, all 307 that were above the cutoff used the electronic voting machine. Furthermore, to avoid confusion between the effect caused by the ballot boxes and the characteristics of the municipalities, we used the municipal data made available by the 1991 IBGE Census for each of them. Finally, regarding the effect of using the electronic ballot box, the ratio between the number of valid votes (not null or blank) over the number of people who participated in the vote was considered.

By using votes for the state legislature to evaluate the impact of electronic voting machines, the focus of changes in the budget is on state government spending, in particular, spending on public health. Poorer families depend more on the service provided by the public health system than richer families, so a change in state spending in favor of the public health system can be seen as a redistribution in favor of the poorest. Furthermore, health spending is able to directly reach legislators as it is a common demand from voters and this is a type of spending that can be changed to quickly increase its share of the budget. Specifically, we analyzed both health expenditures over the 4-year legislative term, the number of visits to the doctor during pregnancy and the weight of the newborn.

Methodology Details

The effect of using electronic voting machines can be captured by comparing the voting of municipalities close to the limit of 40,500 registered voters, both below and above the limit. The idea is that municipalities close to this cutoff would be similar and the difference in the ratio between the number of valid votes and turnout would be caused by the use of the electronic ballot box. Consequently, the treatment group is defined as the municipalities that adopted the ballot box and the control group as the municipalities that used paper during the 1998 elections. It should be noted that the results are given for municipalities close to this limit, as other cities very large or small may present different voting dynamics. Finally, it remains to determine how close to the cutoff the municipalities must be to be compared to each other, using the intervals of 20,000, 10,000 and 5,000 registered voters.

Next, to explore the idea that the electronic voting machine had a greater impact on the uneducated population, the exercise of obtaining the effect of adopting the electronic voting machine was repeated by dividing the municipalities into those above and below the median illiteracy rate. Finally, the effect of adopting electronic voting machines on public health spending was captured. As only a few cities adopted electronic voting machines in the 1998 elections, some states ended up being more affected than others, that is, a larger portion of the electorate in some states had access to electronic voting machines when compared to other states. The phased adoption of the voting machine by some states then allows us to compare whether the evolution of health spending also follows the same pattern, in other words, whether public health spending grew earlier in states with a larger share of voters with access to electronic voting machines. .

Results

Following the methodology above, the impact obtained for the adoption of the electronic ballot box is in the order of 12 percentage points on valid votes, that is, the electronic ballot box actually caused an increase in the number of valid votes where it was adopted. This result is valid for the three intervals used. When considering the illiteracy rate of municipalities, the effect of the electronic voting machine was between 15 and 18 percentage points for municipalities with a high illiteracy rate and only between 9 and 11 percentage points when the rate was below the median, reinforcing the idea that the participation of those with less education benefited more.

When evaluating health expenditures, the use of services by pregnant women and the weight of the newborn, the adoption of electronic voting machines was also positive. The complete transition from voting using paper to voting using the ballot box generated an increase of 3.4 percentage points in the portion of the state budget allocated to health. Adoption of the electronic voting machine led to an increase in the share of pregnant women with little or no education who made more than seven prenatal visits by 7 percentage points and a reduction of 0.5 percentage points in the probability of having a child under of the weight.

Public Policy Lessons

Two lessons can be learned from the study. First, access to the electronic ballot box resulted in a higher rate of valid votes in the elections, the result is even more significant for locations with a higher rate of illiteracy, that is, in the Brazilian context, the adoption of the electronic ballot box resulted in greater representation of the population with less education in the electoral process. Second, the greater participation of the poorest and least educated voters leads government officials to allocate a greater portion of spending aimed at this public, that is, there is a redistribution of spending in favor of the poorest voters.

Reference

Fujiwara, Thomas. “Voting technology, political responsiveness, and infant health: Evidence from Brazil.” Econometrica, vol. 83, no. 2, p. 423-464, 2015.