Essays on Dynastic Politics and Political Participation
- Author(s): Sonnet, Luke
- Advisor(s): Golden, Miriam A
- et al.
This dissertation contains three essays on electoral politics, with focuses on political dynasties, political participation, and the political context of Pakistan. All three essays deal with the barriers that certain individuals face when participating in politics. The first essay explores how certain institutional changes, but not others, curtail the power of political families and the conditions under which electoral reforms create space for new candidates for legislative office. While the second essay makes a largely methodological contribution, the central result that political dynasties hold advantages even in close elections also has implications for who can enter and win elections. The third essay documents how social norms act as barriers to female political participation in Pakistan and considers the kinds of expectations that may increase female turnout when it lags far behind male turnout.
In the first essay, my co-authors, Ali Cheema and Farooq Naseer, and I estimate the effect of institutions on political dynasties. Previous evidence is mixed on whether institutional change disrupts elite power and policy prescriptions for how to improve the competitiveness of elections are unclear. We contribute to this literature by studying the effects of two specific reforms enacted before the 2002 General Elections in Pakistan---reapportionment and an education minimum for legislative candidates. Introduced by General Musharraf's military regime, these reforms were specifically aimed at powerful political families. However, we provide evidence that only one was effective. Using a pseudo-regression discontinuity design relying on the formulaic assignment of seats to districts following reapportionment, we estimate that increasing the number of seats in the legislature dilutes the prevalence of political dynasties in elected positions. Political families were able to win just as many seats as before reapportionment, but were unable to capture many of the new seats, leaving room for politicians from other families to enter and win new seats. On the other hand, the education minimum was less effective at curtailing incumbent power. Using a differences-in-differences design, we estimate that areas most affected by the disqualifications caused by the education minimum were more likely to have connected family members entering and winning elections for the first time, while ``outsiders'' were even less likely to enter and win the newly vacated seats. We argue that elites may be more willing to respond to disqualifications than an expansion in the number of seats in a legislature due to diminishing marginal returns to the number of elected family members. This implies that increasing legislature sizes may be more effective at curtailing elite power than other institutional reforms that restrict who can run in an election.
The second essay makes a largely methodological contribution by providing evidence that members of political families are more likely to win very close elections. This imbalance in the dynastic status of winners and losers of close elections exists in datasets from Japan, the Philippines, Pakistan, and the United States. On average, the winner of a close elections is about 6 percentage points more likely to be a family member of a previously elected politician than a loser of a close election. This imbalance, while perhaps unsurprising given the well-documented advantage dynastic candidates have in general, has important implications for the validity of close elections regression discontinuity designs. Predicated on the assumption that those who barely win and those who barely lose elections are similar on many characteristics, close elections regression discontinuity designs aim to estimate the causal effect of winning an election. However, if there is a detectable imbalance in the characteristics of winners and losers of close elections, as with dynastic status here, then estimates using close elections regression discontinuity designs may be biased. In fact, I calculate that between 9 and 40 percent of top political science and economics papers that use these designs have their most robust estimate quartered by bias of this magnitude.
The third essay moves away from candidacy and the legislature and focuses on the issue of female political participation in Pakistan and the role that norms play in constraining or motivating behavior. Using data from 37 communities on the beliefs of men and women, their expectations about others in their community, and their social networks, this essay provides evidence that behavior and expectations about behavior are more strongly correlated with self-reported female political participation than normative beliefs and expectations. In other words, women self-report voting more often when they believe women in their social network vote and when women in their social network self-report voting. This relationship between self-reported voting and both empirical expectations---beliefs about what others actually do---and the actual behavior of others with a woman's own behavior is stronger than the relationship between self-reported voting and normative expectations---beliefs about what others think one ought to do. I argue that these results are explained by limited female mobility outside the home in Pakistan. Due to this restriction, female turnout takes the form of a coordination game and thus empirical expectations and behavior are more important than normative expectations in explaining when women will vote. Lastly, I provide evidence that expectations of both kinds serve to license political participation; women who think others support their right to vote will be more likely to turn out only if they themselves support a woman's right to vote. In a context like Pakistan, supportive social norms regarding suffrage only mobilize women when they themselves are supportive of female political participation.