That women are under-represented in politics is one of the truisms of our age. Quite why this is so is the matter at issue obviously enough. Are women not chosen for winnable seats? Is it that not enough women actually want to go into politics? Or, as this paper suggests, is it that the voters don’t want all that many female politicians? If it’s this last, well, what is to be done? The point of democracy being that the voters get what the voters want good and hard. And if what the voters want is not women then why shouldn’t the voters not have women politicians?
Voter bias and women in politics: Evidence from French parliamentary elections
Thomas Le Barbanchon, Julien Sauvagnat 08 December 2018
Despite many efforts to close the gender gap, women remain underrepresented in politics. This column shows that in the case of France, voters’ preferences towards gender shapes political selection and ultimately the gender composition of elected politicians. This suggests that gender parity in policymaking relies on improving the slow-changing attitudes of voters towards male and female political candidates.
Despite significant progress in recent years, women are still largely under-represented among elected politicians, accounting for around 25% of members of parliaments across the world. While recent evidence suggest that the gender composition of politicians has important implications for policymaking (e.g. Chattopadhyay and Duflo 2004), there is no consensus on the key factors that drive the under-representation of women in politics. In this column we show that voters’ preferences towards gender shapes political selection and, ultimately, the gender composition of elected politicians.
The low fraction of women in politics is somewhat surprising in a world in which there are no legal restrictions on women’s ability to vote or run for office. A number of recent studies have explored whether political parties favour male candidates (e.g. Esteve-Volart and Bagues 2012, Casas-Arce and Saiz 2015), and have found mixed empirical evidence on the importance of party bias. While party bias has been extensively studied, less is known about the role of voter bias in the gender composition of elected politicians.
Evidence from survey data on voters’ attitudes towards women
In recent research, we link voters’ attitudes toward gender with gender gaps in both the composition of candidates running for elections, and in electoral outcomes (Le Barbanchon and Sauvagnat 2018). For this, we build on previous work showing that attitudes towards gender permeate across markets (e.g. Alesina et al.2013), and use both administrative data on local gender earnings gaps (the difference in labour earnings between men and women) and survey data on gender roles in politics (the fraction of respondents who agree with the statement “men are better political leaders than women”). We hypothesise that both measures are increasing with the degree of unfavourable attitudes towards women.
We find that voters’ attitudes towards gender have a strong impact on the gender distribution of candidates and elected politicians in both French parliamentary elections and across countries. As shown In Figure 1, a 5% increase in the gender earnings gaps is associated with a 2.5% decrease in the probability that a female candidate runs for office in that district, a 10% drop from the sample mean. Similarly, we find that a 10% increase in the share of respondents who agree with the statement “Men are better political leaders than women” is associated with a 2.5% decrease in the share of female candidate politicians.
Figure 1 Share of female candidates in French parliamentary elections vs gaps in labour earnings
We further exploit the granularity of local gender earnings’ gaps in order to estimate the effect of voters’ attitudes on gender gaps in vote shares for the same female and male candidates, across municipalities of the same electoral districts in French parliamentary elections. We find a positive and strong correlation between gender earnings gaps and electoral gaps across municipalities – a 10% increase in gender earnings gap leads to an increase by 1.5% in vote shares between male and female candidates (a difference that is large enough to change the outcome of around 15% of local elections).
These results indicate that voters’ attitudes towards women matter quantitatively for understanding gender differences in both selection into politics and electoral outcomes.
The effect of gender quotas and electoral competition
We use these stylised facts as a motivation for building a model of electoral competition (in majoritarian single-member constituencies elections) in which political parties select candidates across districts taking into account that voters care about the gender of candidates in their districts. The model allows us to derive new insights on how gender quotas on candidates affect political selection, and ultimately the gender composition of elected politicians. We first show theoretically that in the absence of gender quotas, parties always select the best candidate in each district – i.e. the one that maximises the probability of winning the election whatever the degree of political contestability. However, electoral competition shapes the selection of male versus female candidates in the presence of gender quotas on candidates. In that case, we show that parties strategically select men in contestable districts (that is, in close electoral races) and women in non-contestable districts when voters are biased in favour of male candidates.
We take this prediction to the data and exploit the introduction in 2000 of gender quotas in French parliamentary elections, in which parties face fines when they deviate from a 50% national gender parity rule on candidates. We find strong empirical support for the existence of a voter bias in favour of male candidates in French elections – while electoral competition has no effect on the gender allocation of candidates before 2000, we find that parties are more likely to select male candidates in contestable districts after the introduction of gender quotas.
Finally, we use the model to quantify the importance of competition in restricting women representation in politics. For this, we calibrate our model to match the gender distribution of candidates both before and after the introduction of gender quotas in French Parliamentary elections, and conduct counterfactual simulations in which we vary the degree of electoral competition between the two main coalitions running for elections. The effect is quantitatively large. We find that an increase of 10% in the share of contestable districts reduces the increase in the fraction of elected women due to the introduction of gender quotas by around 25%.
Overall, the findings that voters’ attitudes toward gender affect gender gaps in politics – both for differences in electoral scores and in the gender composition of candidates – have important consequences, and suggest that slow-moving voters’ attitudes might be an important factor that limits convergence towards a gender parity among politicians over the world.