Looking back on Election 2016 one thing managed to stand out, to me, among unusual political banter. Our federal government did very little to protect the voting rights of our nation’s marginalized communities. President Obama stood idly by when state GOP affiliates blocked voters access in North Carolina1, Wisconsin2, and Arizona3. This begs what should be a lingering question on the American conscious for years to come: what if we let them vote?
While I will not engage in arguing mere hypotheticals, I am open to discussing how economists have attacked this issue lately by briefly outlining two related articles. Grant Miller of Stanford4 and Thomas Fujiwara of Princeton5 critically assess this question in distinct contexts.
Fujiwara 2015 uses regression discontinuity to evaluate how the distribution of electronic voting machines lead to a dramatic shift of voter enfranchisement among Brazil’s poorest and its effect on state-level politics. Miller 2008 studies the effect of women’s suffrage on public policy agendas and public health outcomes within the United States from 1900-1930 by implementing a difference-in-differences approach. Let’s flesh out these papers to understand their results fully.
Before 1998, Brazilians were completely reliant on paper ballots to elect officials. While this may not seem like a big deal to those who can read this post, those who cannot will struggle to fill out the names of their preferred candidate (for those living in São Paolo, that would be out of 1265 candidates for state legislature, 661 for lower chamber of congress, 10 for state governor and 12 for president). These paper ballot votes are typically error-ridden, are counted as residual votes, and are discarded. In a democracy where 23% of the population is illiterate, the induction of electronic voting machines in 1998 proved a boon to their democratic participation. Lucky for us social scientists, it also created a natural experiment worth investigating.
Due to a shortage of these electronic voting machines6, officials decided to only distribute machines to municipalities with more than 40,500 registered voters. Those with 40,500 or less were expected to use paper ballots.
Fujiwara begins his contribution here. First, he compares municipalities just above and below the population cutoff to ensure no other covariates are driving his results. He then merely7 compares the number of valid votes counted in the Brazilian 1998 election in machine municipalities with valid votes in paper ballot municipalities. The estimates are encouraging as machine municipalities see a 12% boost in valid vote-turnout ratio. All while paper ballot technology leads to a quarter of all votes being counted as residual.
In this case, massive voter enfranchisement led to left-leaning parties disproportionately benefiting with robust public health results. On average, over the following eight years, switching to machine voting led to 32.3% growth in state-level healthcare spending, 19% growth in prenatal visits and low-weight births falling by 6.8% among the poorest Brazilians.
The derivative of their growth is worth a brief discussion. Prior to the full rollout of machines, there was a large change in voter participation among Brazil’s least educated, leading to a larger change in healthcare spending8 than was observed in 2002 (the following election where all Brazilian municipalities received electronic voting machines). We can attribute this relationship to the government’s responsiveness to the slowing rise of uneducated Brazilians’ influence in state-level politics between 1998 and 2002. This pattern is important as Fujiwara states that for an endogenous variable to detract from his estimates, it must follow this pattern.
The next paper evaluates women’s suffrage within the United States. As many may know, before the ratification of the 19th amendment, women had the right to vote in many frontier states. While by no means random, we can still compare outcomes within various states that voluntarily enacted women’s suffrage, by comparing public health statistics before and after the passage of women’s suffrage. This difference-in-differences approach controls for unobservable heterogeneity between states to ensure that other covariates are not driving results.
As states begin to expand women’s right to vote, we see a near instantaneous shift in the Senate as senators from women’s suffrage states begin to support the progressive agenda (with negligible outcomes in the House). This is unlikely due to a change of heart among America’s political elites. Before women’s suffrage, senators were constantly lobbied by women’s groups. When these women’s groups could vote, senators started to listen.
Women’s suffrage resulted in a 23% increase in progressive voting in the Senate with various outcomes at the municipal level. Women’s suffrage in America’s cities with a population of 30,000 or more led to an 8% increase in spending, a 6% increase in sanitation spending and a 36% increase in charity, hospital and corrections spending.
Were these dollars well spent? Miller goes a step further to try and evaluate the effect of women’s suffrage on child mortality. In short, giving women the power to vote led to 20,000 fewer child deaths as mortality plummeted, in death registration states, by 72% for children aged 1 to 4 and 59% for children 5 to 9.
It’s clear from the evidence that allowing marginalized groups to vote leads to improved healthcare outcomes for children and their mothers. The bacterial revolution was in full swing, but it appears as if without women’s suffrage, it’s not clear whether it would have caught on at this magnitude.
Ideally, we want to exist in a society whereby ideology alone we can enforce laws associated with the most sacrosanct freedom given to us by our constitution. However, by being able to measure the improvement to society when given a political voice, we can say definitively that enfranchising marginalized populations in the ballot box leads to positive spillovers from which the entire electorate will benefit.
- Miller, Grant. “Women’s suffrage, political responsiveness, and child survival in American history.” The Quarterly Journal of Economics 123.3 (2008): 1287-1327.
- Fujiwara, Thomas. “Voting technology, political responsiveness, and infant health: evidence from Brazil.” Econometrica 83.2 (2015): 423-464.
- The electronic voting machine relied on numbers rather than letters as voters inputted a number, they were then given a picture of the candidate who corresponded to that number, then either submitted their ballot or reset the process to input their desired number.
- Due to large sample size, Fujiwara can engage in non-parametric estimation and chooses the Imbens and Kalyanaraman optimal bandwidth choice. He ends up estimating a relatively simple OLS regression which is legitimized by a battery of robustness checks controlling for state effects. He also checks linear combinations of his parameters to ensure that the sign flip in the derivative does not cancel out the benefit of the initial enfranchisement of voters.
- Why healthcare spending? A nationally representative survey found that 49% of all Brazilians list healthcare as a government priority, serving as a good proxy for determining why Brazilian voters vote.