Political Dynasties, Term Limits and Female Political Empowerment

NBER WORKING PAPER SERIES

POLITICAL DYNASTIES, TERM LIMITS AND FEMALE POLITICAL EMPOWERMENT:

EVIDENCE FROM THE PHILIPPINES

Julien Labonne

Sahar Parsa

Pablo Querubín

Working Paper 26431

http://www.nber.org/papers/w26431

NATIONAL BUREAU OF ECONOMIC RESEARCH

1050 Massachusetts Avenue

Cambridge, MA 02138

November 2019

We thank Cesi Cruz, Raquel Fernandez, Olle Folke, Johanna Rickne and seminar participants in

Galatina Summer meetings 2018, Hunter College, NEUDC 2015, the Political Dynasties

workshop in Oslo, the 3rd Ottawa Applied Microeconomics Workshop, Petralia Applied

Economics Conference 2019, Queens College CUNY, 2019 Scottish Workshop in Development

Economics, Tufts University, the University of Oxford, and the University of Washington for

constructive feedback. The views expressed herein are those of the authors and do not necessarily

reflect the views of the National Bureau of Economic Research.

NBER working papers are circulated for discussion and comment purposes. They have not been

peer-reviewed or been subject to the review by the NBER Board of Directors that accompanies

official NBER publications.

of text, not to exceed two paragraphs, may be quoted without explicit permission provided that

full credit, including © notice, is given to the source.

Political Dynasties, Term Limits and Female Political Empowerment: Evidence from the Philippines

Julien Labonne, Sahar Parsa, and Pablo Querubín

NBER Working Paper No. 26431

November 2019

JEL No. D72

ABSTRACT

We investigate the effect of term limits on female political representation. Using data from

Philippine municipalities where strict term limits have been in place since 1987, we show that

term limits led to a large increase in the number of women running and winning in mayoral

elections. However, we show that this increase is entirely driven by female relatives of the term-

limited incumbents. We further show that the differential gender impact of this policy is driven

by political dynasties' adaptive strategies to stay in power.

Julien Labonne

University of Oxford

United Kingdom

[email protected]

Sahar Parsa

New York University

Department of Economics

19 W 4th Street

New York, NY 10012

[email protected]

Pablo Querubín

Department of Politics

New York University

19 West Fourth Street, Room 428

New York, NY 10012

and NBER

[email protected]

1 Introduction

The second half of the 20th century has seen women break gender roles and stereotypes in all

spheres of life. Despite such dramatic shifts, they still lag in leadership positions, particularly

in politics, in both developed and developing countries. In 2019, only 24.3 percent of national

legislators around the world are women, and such low representation is even more severe for

elected executive positions, with only 11 female Heads of State.

Scholars and policy-makers are interested in understanding women’s pathways to elected

oﬃces. Recent studies have brought attention to structural characteristics of the political en-

vironment such as electoral systems (Rosen, 2013), political competition (Escobar-Lemmon and

Taylor-Robinson, 2005; Lawless and Pearson, 2008; Folke and Rickne, 2016) and the politicization

of ethnicity (Arriola and Johnson, 2014) as well as cultural norms such as matrilineality (Robinson

and Gottlieb, 2019). In parallel, a growing body of research has focused on how policies such as

gender quotas (Krook, 2009; Pande and Ford, 2012; O’Brien and Rickne, 2016) or political reser-

vations (Chattopadhyay and Duﬂo, 2004; Bhavnani, 2009; Cassan and Vandewalle, 2017) shape

female political representation both substantively and descriptively.

We investigate the eﬀect of a widely used policy on female political representation: term limits.

Although the intended goal of term limits is not explicitly to improve women’s representation, a

plausible side eﬀect could be a rise in female elected oﬃcials. After all, open-seat races – races

where the incumbent is not running – are known to attract outsiders and lesser-known candidates

to the political scene (Cain, Hanley and Kousser, 2006). The search for a context to elucidate these

issues takes us to the rise of female politicians in mayoral positions in the Philippines.

The Philippines ranks among the top countries in the world in terms of female political represen-

tation. Women currently hold 29 percent of the seats in the Senate and the House of Representatives.

However, the Philippines didn’t always have such high share of female politicians. In the after-

math of the fall of Ferdinand Marcos’ autocratic regime in the mid 1980s, only 9 percent of women

were elected to the Senate and the House of Representatives. Other elected oﬃces followed similar

trends. Of relevance to the present study, in 1988 only 9 percent of the municipalities had a female

https://www.unwomen.org/en/what-we-do/leadership-and-political-participation/facts-and-ﬁgures accessed on

September 5, 2019.

mayor, climbing to 21 percent in 2010.

We provide evidence that this increase in female political representation is linked to the term

limits introduced in the 1987 constitution. Our analysis relies on a diﬀerence-in-diﬀerences em-

pirical strategy where we compare forced open-seat races to other races, before and after term

limits bind. Hereafter, we use the term forced open-seat races to refer to open-seat races following

binding term limits, to diﬀerentiate from open-seat races in which the incumbent was eligible to

run but decided not to.

First, we show that forced open-seat races are signiﬁcantly more likely to have (i) a woman

running for oﬃce, and (ii) a woman being elected to the mayoral oﬃce. Our estimates suggest that

term limits can account for about two thirds of the increase in the share of female mayors in the

Philippines. We next show that dynastic candidates are more likely to run in forced open-seat races

and that the increase in female representation following binding term limits is entirely driven by

dynastic women. While non-dynastic women are slightly more likely to run in forced open-seat

races, they are not more likely to win. Thus, in our context, term limits increase the share of a very

speciﬁc group of women in power: dynastic women. This is a striking and unexpected result given

the low share of municipalities that had a female representative to start with.

What explains this dramatic increase in the number of female dynastic politicians? Recent

studies have shown that family ties are an important vehicle for women’s access to politics, in both

developed and developing countries (Dal Bo, Dal Bo and Snyder, 2009; Jalalzai, 2013; Chandra,

2016; Folke, Rickne and Smith, 2016; Smith and Martin, 2017; van Coppenolle, 2017; Jalalzai and

Rincker, 2018). If political dynasties are more likely to ﬁeld female candidates, then the increase in

the share of dynastic candidates in forced open-seat races may explain the rise of female politicians.

However, we show that term limits also aﬀect the gender composition of these dynastic successions.

We ﬁnd that the share of dynastic candidates that are female increases from 15% prior to binding

term limits to 45% in forced open-seat races. A simple simulation exercise suggests that these two

eﬀects can help account for the rise of female politicians, though neither by itself can fully explain

the observed change.

We provide evidence of two mechanisms that help explain why term-limited incumbents are

more likely to nominate female relatives. First, term-limited incumbents who wish to return to oﬃce

after waiting out one term, may be more likely to select female relatives who, given existing social

norms, might be more willing to step aside after one term to restore their male relative’s political

career. Our data shows that female relatives are three times more likely than male relatives to hold

oﬃce for one term and then retire to allow the termed-out incumbent to run again. Following

Coronel et al. (2004), we refer to these politicians as “benchwarmers”. Second, we show that given

their age, term-limited incumbents may be constrained in the number of eligible male relatives. In

particular, if term-limited incumbents are, on average, younger than those who retire voluntarily,

they may not have any sons of eligible age when the term-limit binds and may thus have to rely

on their wives to keep the mayoral position in the family.

Our results are relevant for multiple reasons. Most recent research on female descriptive

representation focuses on institutions such as quotas that mandate candidacy or incumbency and

are often relevant for legislative bodies. Here, we bring attention to term limits, an electoral

institution widespread across the world, and most common for executive oﬃces, for which women

are particularly underrepresented. Unlike quotas, term limits are not introduced with the deliberate

goal of increasing female representation. We ﬁnd that in our context forced open-seat races increase

the probability of women accessing elected oﬃces, but mainly for a small group of women: dynastic

women.

This unintended consequence of term limits is related to a recent study by Cassan and

Vandewalle (2017), where female reservations in India increased the representation of low-caste

groups. In their case, the increase in low-caste representation is related to traditional gender

norms prevalent among high caste people in Indian villages. Thus, while our empirical analysis is

limited to local elections in the Philippines, the patterns we document are likely relevant for other

democracies.

Our ﬁndings are also broadly related to the nascent literature emphasizing the importance of

culture when exploring the eﬀects of policies, see Ashraf et al. (2016) and Corno, Hildebrandt

and Voena (2017). Ashraf et al. (2016) show that female education is more responsive to school

construction programs for ethnic groups practicing bride price. Our paper highlights a similar point

within a political context. Although we do not exploit cultural variation within the Philippines,

our results provide strong evidence that the impact of term limits on political outcomes is tied to

the dynastic nature of the Philippine political institutions and the high share of male term-limited

We thus contribute to existing work that explores the characteristics of women who run for public oﬃce. See for

example, Escobar-Lemmon and Taylor-Robinson (2009) and Schwindt-Bayer (2011)

incumbents. Absent a strong dynastic environment, the impact of term limits will most likely

manifest itself through other channels such as increased competition.

2 Term limits and the Rise of Female Politicians

The Philippines saw a dramatic increase in the share of female mayors in the last 30 years. Using

oﬃcial candidate-level results of all mayoral elections between 1988 and 2010, provided by the

Commission of Elections (COMELEC), Figure 1 reports the share of municipalities with an elected

female mayor in each of the eight elections between 1988-2010.

Starting at nine percent in 1988,

the share of municipalities with a female mayor jumped to 16% in 1998 and kept on increasing to

21% by 2010. The discontinuous jump in 1998 is key. It hints at the importance of term limits in

the rise of female politicians as this is the ﬁrst election year with forced open-seat races following

binding term-limits.

Term limits for all elected oﬃces were introduced in the 1987 constitution. At the municipal

level, a politician can only be elected to the same oﬃce three times consecutively (not counting

elections before 1987). The ﬁrst municipal elections following the fall of Ferdinand Marcos were

organized in 1988. In accordance with provisions of the 1987 constitution, the next municipal

elections were organized in 1992. Thereafter, all municipal oﬃcials – the mayor, vice-mayor and

eight municipal councilors, were elected in ﬁrst-past-the-post elections organized every three years.

Upon reaching a binding term limit, a politician can run immediately for a diﬀerent oﬃce or run

again for the same oﬃce after waiting out an election cycle. Thus, the 1998 election was the ﬁrst

election with potentially term-limited incumbents. In what follows, we formally test the hypothesis

that term limits led to an increase in the number of women elected as mayors.

The dataset includes the full names and votes received for all candidates. The only exception is 1988; we only

have data on winning candidates for that election. We used the ﬁrst name of the candidates in our sample to code their

gender.

Figure 1: Share of municipalities with a female mayor

2.1 Empirical Strategy and Main Results

To test the impact of term limits on the rise of female politicians, we use a diﬀerence-in-diﬀerences

estimation strategy. We compare municipalities with a term-limited incumbent to municipalities

without term-limited incumbents, before and after the forced open-seat race. Our basic estimating

equation is:

= α + βTerm_Limit

+ µ

+ ρ

+ 

(1)

where y

is one of two indicators: a dummy for whether a woman ran in a mayoral election

in municipality m and election year t and a dummy for whether a woman won a mayoral election

in municipality m and election year t. Term_Limit

is a dummy equal to 1 if at time t municipality

m had a forced open-seat race and zero otherwise. Our main coeﬃcient of interest is β. We cluster

standard errors at the province level. We include election-year dummies to account ﬂexibly for

common shocks or trends at the national level and a full set of municipality ﬁxed eﬀects to account

for all time-invariant municipal characteristics. We also estimate a more demanding speciﬁcation

by including interactions between election-year dummies and province ﬁxed eﬀects.

Table 1 presents our main results. The dependent variable is a dummy for whether a woman ran

(Columns 1-2) or won the race (Columns 3-4). Municipalities are 17 percentage points more likely

to have a woman running for the mayoral oﬃce in a forced open-seat race (Column 1). Women

are not only more likely to run, they are also 10.5 percentage points more likely to win in forced

open-seat races (Column 3). These results are robust to using the interactions between election-year

dummies and province ﬁxed eﬀects instead of country-wide year ﬁxed-eﬀects (Columns 2 and 4).

Contrasting our point estimates to the jump in the share of female politicians in 1998 (the ﬁrst year

with forced open-seat races), binding term limits could account for more than two thirds of the 6.09

percentage points increase in the share of female mayors between 1995 (9.91%) and 1998 (16.0%).

Given that 41% of the municipalities had a forced open-seat in that year, our estimates would imply

a 4.3 percentage points increase in the share of municipalities with female mayors.

Table 1: Term Limits and the Rise of Female Politicians

Female Candidate Female Mayor

(1) (2) (3) (4)

Term Limit 0.170 0.165 0.105 0.104

(0.014) (0.014) (0.011) (0.012)

Municipal FE Yes Yes Yes Yes

Election FE Yes No Yes No

Province × Election FE No Yes No Yes

Observations 11,892 11,892 11,835 11,835

0.309 0.354 0.317 0.353

Mean Dep. Var. 0.299 0.299 0.146 0.146

Notes: Results from municipality*elections regressions and include municipal and time ﬁxed eﬀects. The

dependent variable is a dummy for whether a woman ran for mayor (Columns 1-2), or a dummy for

whether a woman was elected mayor (Columns 3-4). The standard errors (in parentheses) account for

potential correlation within provinces.

To sum up, the estimates in Table 1 show that women are more likely to run and win in forced

open-seat races. This results from (both) an increase in the likelihood of women running, and to

a minor extent in the likelihood of women winning conditional on running. Prior to 1998 (before

term limits bind), conditional on at least one woman running, the likelihood that a woman was

Sample sizes are a bit smaller in Columns 3 and 4 since, for a few cases, we don’t have information on who won

the election.

elected mayor was 0.45. In forced open-seat races, that likelihood increases slightly to 0.5.

2.2 Testing for the Parallel-Trends Assumption

We now provide evidence of the key identifying assumption in diﬀerence-in-diﬀerences designs:

namely, that municipalities with and without forced open-seat races were on similar trends before

binding term limits. To do so, we estimate an extended version of equation (1) where we include

leads and lags of the term limit variable. More concretely, we estimate the following equation:

= α +

S=−3

Term_Limit

+ µ

+ ρ

+ 

, (2)

where β

are the estimates on the full set of leads and lags (up to three years). For example,

Term_Limit

is the ﬁrst lead of the term-limit dummy. It is equal to one for all term-limited

municipalities one election before they became term-limited and zero otherwise.

Figure 2 provides strong evidence in support of the parallel trends assumption. The coeﬃcients

on the one and two period leads are small and statistically insigniﬁcant, which suggests that treated

and control groups follow a common trend prior to term limits. This reinforces our conﬁdence

that the results discussed above capture the causal eﬀect of binding term limits on female political

participation.

In addition to documenting parallel trends, Figure 2 highlights an interesting feature. The

eﬀect of term limits becomes much weaker one and two periods after the forced open-seat race.

The coeﬃcients on the one and two-period lags, while positive, are substantially muted compared

to the contemporary term-limit dummy. We return to this ﬁnding below in Section 4.

Figure 2: Estimates and 95% Conﬁdence Intervals for Leads and Lags of Term Limit Variable

3 The Dynastic Nature of the Rise of Female Mayors

What drives the increase in female politicians during forced open-seat races? Are the patterns

consistent with the original goal of the policy which was to promote alternation in oﬃce and curb

the power of incumbent political elites, namely political dynasties? Political dynasties play an

important role in politics at both the national and local levels (McCoy, 2009) and have persisted

across many decades. It is common for relatives to take turns holding the same oﬃce, and

for the same family to control multiple elected oﬃces at the same time (Querubin, 2012). To

understand how term-limits interact with this prevalent role of political dynasties, we next explore

the composition of candidates across diﬀerent races between dynastic and non-dynastic. Contrary

to the original intent of the policy, we provide evidence that forced open-seat races lead to an

increase in the number and share of dynastic politicians, particularly of incumbent relatives and

show that the increase in female representation following binding term limits is driven by the entry

of female relatives of the incumbent into the mayoral races.

A key step to analyzing changes in the composition of mayoral races is to identify dynastic

candidates and the relatives of the incumbents in our sample. Individuals in the Philippines carry

two family names: the mother’s maiden name and father’s family name (men and single women)

or the father’s family name and husband’s family name (married women). We follow Querubin

(2016) and Fafchamps and Labonne (2017) and classify a candidate as dynastic if it shares at least

one family name with any current or previous mayor and as an incumbent relative if it shares at least

one family name with the current mayor.

This approach assumes that a shared family name indicates an actual family tie, which is

valid given the historical way in which family names were allocated in the Philippines. In 1849,

Governor Narciso Claveria y Zaldua assigned a diﬀerent family name to every household in each

municipality. As a consequence very common family names are not as prevalent in the Philippines

as in other countries and thus, sharing a family name is very strongly correlated with an actual

family tie. For example, using biographical data, Querubin (2016) documents that the rate of false

positives (i.e. candidates who share a family name but are not related to each other) amongst

candidates for provincial oﬃces (a larger sub-national level) is around 5%. This rate is likely lower

at the municipal level since unique family names were originally allocated at this administrative

level.

Fafchamps and Labonne (2017) and Cruz, Labonne and Querubin (2017) validate and

discuss this method for tracing relatives in more detail.

Our empirical strategy is identical to regression (1), but we use a diﬀerent set of outcome

variables related to whether candidates or elected mayors are dynastic.

In Panel A of Table

2 we focus on the composition of candidates and mayors, between dynastic and non-dynastic,

independently of their gender. In Column 1, we use as dependent variable the number of candidates

in the mayoral race and ﬁnd, as expected, that forced open-seat races attract a larger number of

candidates. However, once we turn to the composition of the pool of candidates between dynastic

and non-dynastic candidates in Column 2, we ﬁnd that the share of candidates that are dynastic

(relative to all candidates) increases by 6 percentage points in forced open-seat races. This is driven

by the higher likelihood of dynastic candidates (Column 3) and in particular incumbent relatives

(Column 4), running in forced open-seat races. The point estimates show that incumbent relatives

are almost 50 percentage points more likely to run in forced open-seat races relative to other

For example, in principle there should be one Aquino family per municipality, even if the Aquino family name

may exist in several municipalities. Two individuals with the Aquino family name living in the same municipality are

thus much more likely to share an actual family tie than two individuals with the Aquino family name from diﬀerent

municipalities in the same province.

We show the results for our preferred speciﬁcation with municipality and year ﬁxed eﬀects. The point estimates

are very similar when we include province-speciﬁc election-year ﬁxed eﬀects.

races. Finally, the dependent variables in Columns 5 and 6 are dummies for whether dynastic

candidates or incumbent relatives win the election, respectively. The point estimate in Column

6 shows that incumbent relatives are 33 percentage points more likely to get elected following

binding term limits. Contrasting the estimated coeﬃcient to the share of mayors that are relatives

of the incumbent in non-open seat races (4.7%), incumbent relatives are 7 times (33/4.7) more likely

to become mayors in forced open-seat races. In sum, our ﬁndings show that while forced open-seat

races attract more candidates, they do not necessarily lead to an increase of “new blood” in the

system.

Table 2: Term Limits and the Rise of Female Dynastic Politicians

(1) (2) (3) (4) (5) (6)

Panel A All Candidates:

No. Cand. Share Dyn. Dyn. Ran Inc. Rel. Ran Dyn. Won Inc. Rel. Won

termlim 0.595 0.060 0.197 0.464 0.139 0.334

(0.031) (0.006) (0.011) (0.013) (0.011) (0.012)

Observations 11,928 11,928 11,928 11,928 11,928 11,928

R-squared 0.547 0.491 0.495 0.372 0.462 0.327

Mean Dep. Var. 2.654 0.176 0.366 0.188 0.234 0.101

Panel B Female Candidates:

Dyn. Ran Inc. Rel. Ran Non-Dyn Ran Dyn. Won Inc. Rel. Won Non-Dyn Won

Term Limit 0.171 0.243 0.043 0.105 0.161 -0.000

(0.011) (0.011) (0.012) (0.009) (0.009) (0.007)

Observations 11,927 11,927 11,928 11,892 11,892 11,835

R-squared 0.305 0.251 0.295 0.307 0.217 0.336

Mean Dep. Var. 0.111 0.0617 0.209 0.0691 0.0368 0.0764

Notes: Results from municipality*elections regressions and include municipality and election-year ﬁxed

eﬀects. In Panel A, the dependent variable is the number of candidates (Column 1), the share of candidates

who are dynastic relative to the pool of candidates (Column 2), a dummy equal to one if a dynastic

candidate ran (Column 3), a dummy equal to one if an incumbent relative ran (Column 4), a dummy for

whether the elected mayor is dynastic (Column 5) and a dummy for whether the elected mayor is related to

the previous incumbent (Column 6). In Panel B, the dependent variable is a dummy for whether a dynastic

woman ran (Column 1), a dummy for whether a woman related to the incumbent ran (Column 2), a

dummy for whether a non-dynastic woman ran (Column 3), a dummy for whether a dynastic woman won

(Column 4), a dummy for whether a woman related to the incumbent won (Column 5), and a dummy for

whether a non-dynastic woman won (Column 6). The standard errors (in parentheses) account for

potential correlation within provinces.

Turning to the sample of female candidates in Panel B, in Columns 1-3 we use as dependent

variables dummies for whether a dynastic, incumbent relative or non-dynastic woman ran in the

race, respectively. The estimates show that the increased probability of women running for oﬃce

documented in Table 1 is mostly driven by dynastic candidates. Dynastic women are 17 percentage

points more likely to run in forced open-seat races (Column 1). The eﬀect is even larger for female

relatives of the incumbent who are 24 percentage points more likely to run following binding term

limits (Column 2). The eﬀect for non-dynastic women (Column 3), while positive and statistically

signiﬁcant, is substantially smaller than that for women related to previous or current incumbents.

Investigating the composition of winning candidates (female mayors), in Columns 4-6 we use

dummies for whether dynastic, incumbent relatives or non-dynastic women won the election,

respectively. The estimates show that the increase in the number of female politicians following

binding term limits is driven entirely by the entry of dynastic female politicians, in particular, by

women related to the term-limited incumbents (Column 5) who are 16 percentage points more

likely to win in forced open-seat races. In fact, close to 75 percent of the women elected in the ﬁrst

wave of forced open-seat races in 1998 were related to the term-limited incumbent and out of the 342

dynastic women winning an election in open-seat races, only 15 are dynastic but unrelated to the

term-limited incumbent. On the other hand, the coeﬃcient in Column 6 shows that non-dynastic

women are not more likely to win forced open-seat races.

In sum, the descriptive patterns and regressions presented in Figure 1 and Tables 1 and 2

provide strong evidence that forced open-seat races constitute critical junctures in which women

are disproportionately more likely to access political oﬃce. However, not all women beneﬁt from

term-limits: relatives of the term-limited incumbent are substantially more likely to win in forced

open-seat races.

4 The Gender Nature of Dynastic Politics and Term Limits

So far we’ve shown that term limits led to an increase in female political representation in mayoral

positions and that most of these women “succeeded” their term-limited relatives in oﬃce. This

suggests that a key factor behind the rise of female mayors in the Philippines is related to political

successions and the response of term-limited incumbents when faced with the constraint imposed

The fact that voters often vote for and elect female relatives of term-limited incumbents is consistent with the

argument of Dolan (2014) that traditional political forces (such as membership in the incumbent dynasty) are often a

more important determinant of voter behavior than a candidate’s gender.

by term limits on their political careers. In this section, we explore two plausible explanations for

the role of term limits in the increase in the probability of female relatives of the incumbent running

for oﬃce. We focus on the likelihood of running for oﬃce, rather than on the likelihood of winning,

since incumbents only have full control on whether a relative runs for oﬃce (not on whether they

actually win). Moreover, recall from Section 2.1 that the increase in female representation in forced

open-seat races is mostly driven by the higher likelihood of women running for oﬃce, rather than

of women winning conditional on running.

Speciﬁcally, the share of female relatives running for oﬃce is the product of i) the likelihood of

incumbent relatives running for oﬃce and ii) the gender composition of these incumbent relatives.

Let Prob(FemRel) be the probability that a female relative of the incumbent runs for mayor. We can

use the following identity to illustrate the point:

Prob(FemRel) = Prob(Rel) × Prob(Fem|Rel)

where Prob(Rel) is the probability that a relative of the incumbent runs for mayor and

Prob(Fem|Rel) is the probability that, conditional on a relative of the incumbent running, the rel-

ative is female. In what follows, we will explore quantitatively the signiﬁcance of each of these

components in explaining the change in the share of female relatives running in forced open-seat

races.

4.1 Term Limits and the Likelihood of Dynastic Successions

Based on the above expression, a ﬁrst explanation behind the rise of female relatives running for

oﬃce is that, as shown in Panel A of Table 2, term limits are associated with an increase in the

likelihood of incumbent relatives running for oﬃce (an increase in the likelihood of dynastic suc-

cessions). As documented by Querubin (2012), this reﬂects the adaptive strategies of term-limited

incumbents who want to continue their political careers and preserve their political power. In a

context like the Philippines, where incumbents enjoy a very large electoral advantage (Querubin,

2016), it is essential to prevent opponents from gaining access to oﬃce in forced open-seat races.

Given the weakness of political parties in the Philippines (Montinola, 1999) and the importance of

families in politics, term-limited incumbents often attempt to maintain political control by having

a relative run to replace them. Moreover, as mentioned above, previous studies have documented

the importance of political dynasties as a vehicle for women’s access to oﬃce both for executive

positions and for legislatures in other developed and developing countries (Dal Bo, Dal Bo and

Snyder, 2009; Jalalzai, 2013; Chandra, 2016; Folke, Rickne and Smith, 2016; Smith and Martin, 2017;

van Coppenolle, 2017; Jalalzai and Rincker, 2018). In our setting, during 1988-1995, prior to any

binding term-limits, the female share amongst candidates related to the incumbent (Prob(Fem|Rel))

was 0.15. Can the increase in the likelihood of dynastic candidates running for oﬃce (documented

in Panel A of Table 2) explain the increased probability of female relatives running and the ensuing

rise in female political representation? Speciﬁcally, what would be the predicted change in the

likelihood of female relatives running for mayor (Prob(FemRel)) if we held the female share of dy-

nastic successions constant at 0.15, but account for the increased likelihood of incumbent relatives

running for oﬃce?

Recall from Table 2, Panel A, Column 4, that forced open-seat races are associated with an

increase of 46 percentage points (0.46) in the probability of an incumbent relative running for

oﬃce. Thus, ∆Prob(FemRel) = 0.46 × 0.15 = 0.069. Contrasting this number with the estimated

diﬀerence-in-diﬀerence coeﬃcient in Column 2 of Table 2 Panel B, we can see that the change in

the likelihood of dynastic successions can account for 29% of the 24 percentage increase in the

probability of female relatives of the incumbent running for oﬃce (0.069/0.24=0.29). This suggests

that the increase in the number of “dynastic successions” following term-limits, holding everything

else constant, plays an important role, but leaves 71% of the estimated change in the likelihood of

female relatives running for oﬃce unexplained.

4.2 Term Limits and the Gender Composition of Dynastic Successions

This brings us to our second explanation: the change in the gender composition of dynastic

successions. Remember that the share of female relatives running for oﬃce (conditional on a

relative running for oﬃce) is 0.15 during the baseline period (1988-1995), prior to any binding term

limits. This share increases to 0.45 in forced open-seat races. This observed diﬀerence roughly

coincides with the diﬀerence-in-diﬀerence estimate reported in Column 2, Panel B of Table 2. In

other words, the share of female relatives diﬀers substantially in dynastic successions following

binding term limits compared to other dynastic successions (that occur due to diﬀerent causes

such as death or retirement, for example). Why do term limits change the gender composition of

dynastic successions? We investigate two plausible explanations rooted in the design of the term

limits policy in the Philippines.

4.2.1 Age Proﬁle of Term-Limited Incumbents and Availability of Eligible Male Relatives

A ﬁrst explanation is related to the fact that term-limits force incumbents out of oﬃce after three

terms. As a consequence, strong incumbents are forced to exit their positions at a younger age, on

average, compared to strong incumbents in standard (non-forced) succession decisions that do not

follow binding term-limits. We argue that the younger age of the exiting incumbent can generate

large diﬀerences in the gender composition of political successions in line with what we observe in

our setting. This is because the term-limited incumbent is less likely to have children (in particular

sons) of eligible age (which in the Philippines is 21) compared to other successions, and hence he

may resort to his wife to succeed him in oﬃce. This argument is better exempliﬁed by the story

of one prominent political family in the Philippines. Jejomar Binay was elected mayor of Makati

in 1988. He was re-elected in both 1992 and 1995 and thus was term-limited in 1998 at age 55. At

that time, Jejomar did not have any sons of eligible age. His kids were Nancy (25, female), Abby

(23, female), Jun (20, male), Marita (19, female) and Joanna (10, female). Elenita, his wife, ran and

was elected for the position but did not run in 2001. She was replaced by her husband who was

again re-elected in 2004 and 2007. In 2010, when he became term-limited again, his son Jun (now

of eligible age) ran and was elected.

Next, we use a simple quantitative simulation to show that, even in a world with a strong male

preference in dynastic succession decisions, this channel can generate changes in the share of female

candidates similar in magnitude to the ones observed in the data, under plausible assumptions.

To that end, suppose that incumbents prefer a male succession: the exiting incumbent chooses an

eligible male successor as long as there is one available, otherwise he chooses amongst the potential

set of eligibile female successors (consisting of the wife and daughters). Consider ﬁrst a standard

succession in which a male politician voluntarily retires or dies. In this case, such politician is

likely to be of older age and succeeded by one of his children. Fertility data for wealthy Filipino

couples in the 70s suggests an average of 3 children per family, which given a gender ratio of

50%, would imply a share of male relatives in this succession of around 87.5%.

Interestingly, this

The estimates for the number of children per family comes from Table 5 in Hirschman and Guest (1990). In 1975-79,

unconstrained male biased scenario is close to what we observe in non-term-limited successions

where the share of women related to the incumbent is around 15%.

As mentioned above, dynastic successions following term limits diﬀer from standard succes-

sions as they likely aﬀect politicians at a younger age. As a consequence, their children might not be

old enough to run for oﬃce and thus, their wives may become important (or the only) contenders

in the succession. As politicians get older, they become more likely to have older children who

can run to replace them. To illustrate this point, assume the same male-bias applies to dynastic

successions following binding term limits. In other words, term-limited incumbents select an

eligible male relative when available, otherwise they select their wives or eligible daughters. In

order to account for the fact that sons may not yet be of eligible age, we introduce parameters p

and p

, as the probability that, respectively, the ﬁrst, second and third child is eligible to run.

Thus, the share of potential male successors becomes: p

∗ 1/2 + p2 ∗ 1/4 + p

∗ 1/8.

Using statistics

on the age distribution of fathers from Oﬃcial Vital Statistics of the Philippines at the end of the

70s, we can estimate p

, p

and p

for diﬀerent age values of the term-limited politician, given a

representative politician with three kids and an age diﬀerence of two years between the kids.

can then simulate the probability of having a male relative of eligible age for incumbents aged 45

to 64 years old. Figure 3 shows the simulated (predicted) share of male relatives running in our

sample. As the age of the term-limited incumbent increases, the share of male relatives running

converges towards the share in other non-term-limited successions. Unfortunately, we do not have

data on the age of incumbents in our sample. However, assuming that politicians are uniformly

distributed between 45 - 64 years old at the moment of ﬁrst becoming term-limited, the probability

of having an eligible male relative would be, on average, 54% which coincides with the actual

(observed) fraction of male relatives running for oﬃce in term-limited successions.

the Total Fertility Rate of women with at least some College education (as a proxy for wealth) was 3.08. 87.5% comes

from the probability of having at least one son, conditional on having three children: i.e., 1 − (1/2)

The formula comes from the probability of having a ﬁrst son weighted by the probability of the ﬁrst son being of

eligible age (p

∗ 1/2), the probability of having a middle son weighted by the probability that the middle son is of eligible

age (p

∗ 1/4), and the probability of having a third son weighted by the probability that the third son is of age (p

∗ 1/8).

See https://psa.gov.ph/sites/default/files/1978%20Vital%20Statistics%20Report.pdf accessed on

September 5, 2019.

These numbers would be in line with mayors being ﬁrst elected when they are 35 – 45 years old, which is consistent

Figure 3: Simulation probability of having a male relative running

It is worth emphasizing the importance of wives for this channel, which is a by-product of

the high share of male incumbents in the Philippines. For example, close to 92% of term-limited

incumbents in 1998 were men. This high share of male politicians elected in unconstrained races

is necessary for term-limits to generate a discrete jump in the gender composition of dynastic

successions. Had the majority of term-limited incumbents in 1998 been female, then term-limits

would have increased the likelihood of having male mayors, since term-limited female incumbents

would have had eligible male relatives (their husbands) available for the succession.

4.2.2 Women as Benchwarmers

For our second explanation behind the change in the gender composition of dynastic successions,

recall from Section 2 that term limits in the Philippines impose a short-term constraint on the

incumbent, forcing the politician out of oﬃce for only one term. This contrasts with other standard

dynastic successions (following retirement or death) where the incumbent’s exit is more permanent.

Thus, since term limits generate only a short-run (i.e. one term) constraint on the incumbent, those

interested in running again after one term may be more likely to appoint a female relative in oﬃce

with statistics for other countries.

to serve as benchwarmers. This is a Filipino expression for incumbent relatives who serve for only

one term after the forced open-seat race, and retire immediately after to allow their relative to run

again (Coronel et al., 2004). Elenita, the wife of Jejomar Binay, is an example of a benchwarmer.

We argue that female relatives might be more attractive than male relatives as benchwarmers due

a number of reasons. First, consider a term-limited incumbent who wants to remain inﬂuential

during his term out of oﬃce and to run again immediately after. Given gender norms in the

Philippines, women may be more willing to comply with the relatives’ directives while in oﬃce

and to retire immediately after to allow his return. Similarly, in a context where the careers of

men are perceived as more important than those of women, short-term distortions on the life (and

career) of the female relatives might be perceived as less costly. For example, according to the

2009 Labor Force Survey, female labor force participation is 19 percentage points lower for women

than for men (gap amongst those with college education is similar). Moreover, 60% of Filipino

respondents in the 6

wave of the World Values Survey (2010-2014) agree that whenever jobs are

scarce, men should have more right to a job than women. Finally, 56.4% of respondents agree with

the statement that “men make better political leaders than women do” (fraction who agree amongst

male respondents is 63.7%). This is also consistent with ﬁndings by Fox and Lawless (2014) and

Lawless and Fox (2005) that women are signiﬁcantly less likely than men to demonstrate ambition

to run for elective oﬃce. If incumbents are interested in returning to oﬃce after one term, they

may prefer to select female relatives who will be less inclined to remain in oﬃce and continue a

political career of their own.

In line with this explanation, we ﬁnd that female mayors who replaced their term-limited

relative are more likely to be one-termers – that is, to exit after serving only one term – than other

mayors. In particular 57.7% of women who replaced their term-limited relative serve for only one

term. This is higher than the share of one-termers amongst non-dynastic women (44.5%) or men

(38%) elected in forced open-seat races. It is also higher than the share of one-termers amongst

women elected in non-forced open-seat races (37%).

Female mayors who replaced their term-

limited relative are also three times more likely to be benchwarmers than their male counterparts

(that is, to retire after one term in order to allow the previously term-limited incumbent to return

These diﬀerences in the share of one-termers between female relatives of the incumbent elected in forced open-seat

races and other mayors are statistically signiﬁcant at the 1% level

to oﬃce). While only 10% of male relatives are benchwarmers, the corresponding ﬁgure for female

relatives is 31%. As a consequence, male relatives of the incumbent serve on average a larger

number of terms (2.23) than female relatives (1.81). This may also explain the small coeﬃcient on

the one and two period lags of the term-limit dummy reported in Figure 2. The high prevalence

of benchwarmers amongst dynastic women is in line with the eﬀect of term-limits on female

representation being short-lived.

The benchwarmer eﬀect and the lower average number of terms served by female politicians

elected in forced open-seat races, point at two important forces behind the steady increase in the

share of municipalities with a female mayor in the Philippines in the last 30 years. While there is

an increase in the number of female mayors accessing oﬃce during forced open-seat races, there

is a decrease in the average number of terms these women stay in power (as well as an increase

in the number of one-termers), compared to other races. In other words, while following binding

term limits a larger number of municipalities experience a female mayor, these female mayors are

on average less likely to be re-elected. These two forces, the number of forced open-seat races and

the share of women exiting after one term, generates a delicate balance of inﬂows and outﬂows

that translate into the observed increase in the share of municipalities with a female mayor. For

example, the number of female mayors in 1995 was 147 and it increased to 237 in 1998. This net

increase in 90 female mayors consists of 170 new female mayors (of which 133 were elected in forced

open-seat races ) and 80 outgoing female mayors who either did not run or were not re-elected.

A similar pattern can be observed in 2001: out of the 237 female mayors, 94 decided not to run,

while 40 were term-limited (for an outﬂow of 134 female mayors). Importantly, 59 out of the 94

female mayors who decided to not run another term, were relatives of the term-limited incumbent

in 1998 and served for only one term. Despite the higher share of female mayor outﬂows in 2001

(40% or 94 out of 237) than in 1998 (23% or 30/133) there was still a net increase in the total number

of female mayors to 246 due to the inﬂow of 143 new female mayors (83 of them elected in forced

open-seat races). The key observation is that absent the benchwarmer eﬀect (or higher likelihood of

one-termers amongst female relatives of term-limited incumbents), the net increase in the number

of municipalities with a female mayor would have been much larger.

To sum up, in this section we provide evidence that the age-proﬁle of term-limited incumbents

and the use of benchwarmers by term-limited incumbents interested in returning to oﬃce after one

term can help us explain the change in the gender composition of dynastic successions in forced

open-seat races. Returning to our original question, what fraction of the change in the fraction of

women running for oﬃce can be explained by the increase in the fraction of women in dynastic

successions? We conduct a similar exercise to the one reported in Section 4.1, but use the baseline

(1988-1995) fraction of incumbent relatives running (Prob(Rel) = 0.08) and predict the change in the

likelihood of women running for oﬃce, using the change in the fraction of incumbent relatives that

are female: ∆Prob(FemRel) = 0.08 × 0.3 = 0.024. Thus, the change in the female ratio of dynastic

successions by itself can account for about 10% (0.024/0.24) of the change in the likelihood of women

running for oﬃce. This suggests that the interaction of both forces – an increase in the likelihood of

dynastic successions and in the female share of these successions– provides a compelling account

of how term limits led to increased female political representation in the Philippines, as each force

by itself is not enough to explain the jump in the fraction of women running for oﬃce.

5 Conclusion

In this paper we show that term limits - a widely used institution across several democracies -

can lead to an increase in female representation and explain the dramatic rise of female mayors

in the Philippines. However, we show that in the Philippines this eﬀect was mostly restricted to

female relatives of the incumbent. This ﬁnding has potential implications for the extent to which

an increase in female descriptive representation translates into substantive representation or the

empowerment of other women in dynastic contexts. Previous research has shown that women

who were elected to oﬃce via reservations or gender quotas have enacted pro-women policies

and have had demonstration or role model eﬀects on the political participation of other women

(Chattopadhyay and Duﬂo, 2004; Bhavnani, 2009). However, it is likely that the channel through

which women access oﬃce mediates their impact on other policy or electoral outcomes. Women

elected through the dynastic channel following binding term limits may be unwilling to steer

policy away from their family’s interests and closer to their own preferences (or the preferences

of other women) or may be unable to do so as they are often ﬁgureheads or benchwarmers for

their relatives, who constrain their decisions and retain de facto power. As argued by Jalalzai

(2013), women might derive their political identities through their close male relatives and may be

expected to further the political goals of the male relatives they succeeded. Similarly, if women are

elected through the dynastic channel, they may not have a mandate that enables (or encourages)

them to prioritize women’s needs and preferences, in contrast to contexts where women are elected

via gender quotas (Franceschet and Piscopo, 2008). Third, the new pool of elected dynastic women

may not be representative of the broader pool of women which may also impact whether their

policy choices will beneﬁt the majority of women. For instance, Clots-Figueras (2011) ﬁnd that in

India, female legislators of lower castes, but not those of higher castes, are more likely to enact

female-friendly policies. Finally, the fact that dynastic women elected following binding term limits

often serve for only one term, may not only limit their substantive impact on economic policies but

may also attenuate their “role model eﬀect” if other women in society perceive that women, once

in oﬃce, cannot pursue a political career of their own and must retire to promote the career of their

male relatives.

While the evidence presented in this paper is speciﬁc to the Philippines, we believe that many of

the insights are relevant for other democracies where term limits are common and where parties are

weak and/or political dynasties play an important role. Moreover, some of the speciﬁc channels we

have explored (such as the diﬀerential age proﬁle induced by term limits) likely apply to many other

contexts. Our ﬁndings, however, highlight the importance of understanding how the cultural and

social context mediates the eﬀect of widely used political institutions such as term limits. In dynastic

contexts, term limits may, perhaps inadvertently, increase female representation but restrict this

higher access to oﬃce to dynastic women. And this may have important consequences for the role

that these newly elected female politicians can play. As stated by Jalalzai (2013), “in spite of the

rising numbers of women executives, we should question women’s ultimate progress in achieving

powerful positions” (p. 114). Where women’s increased access to elected oﬃce is mainly driven

by family connections, the consequences for substantive representation may be very diﬀerent.

References

Arriola, Leonardo R. and Martha C. Johnson. 2014. “Ethnic Politics and Women’s Empowerment

in Africa: Ministerial Appointments to Executive Cabinets.” American Journal of Political Science

58(2):495–510.

Ashraf, Nava, Natalie Bau, Nathan Nunn and Alessandra Voena. 2016. Bride Price and Female

Education. Working Paper 22417 National Bureau of Economic Research.

URL: http://www.nber.org/papers/w22417

Bhavnani, Rikhil R. 2009. “Do electoral quotas work after they are withdrawn? Evidence from a

natural experiment in India.” American Political Science Review 103(01):23–35.

Cain, Bruce, John Hanley and Thad Kousser. 2006. Term Limits: A Recipe for More Competition?

Brookings Institution Press pp. 199–221.

Cassan, Guilhem and Lore Vandewalle. 2017. Identities and public policies: Unintended eﬀects

of political reservations for women in India. Technical report Economics Section, The Graduate

Institute of International Studies.

Chandra, Kanchan. 2016. Democratic Dynasties: State, Party, and Family in Contemporary Indian

Politics. In Democratic Dynasties: State, Party, and Family in Contemporary Indian Politics, ed.

Kanchan Chandra. Cambridge University Press pp. 10–50.

Chattopadhyay, Raghabendra and Esther Duﬂo. 2004. “Women as Policy Makers: Evidence from

a Randomized Policy Experiment in India.” Econometrica 72(5):1409–1443.

Clots-Figueras, Irma. 2011. “Women in politics: Evidence from the Indian States.” Journal of Public

Economics 95(7-8):664 – 690.

Corno, Lucia, Nicole Hildebrandt and Alessandra Voena. 2017. Age of marriage, weather shocks,

and the direction of marriage payments. Technical report National Bureau of Economic Research.

Coronel, Sheila S., Yvonne T. Chua, Luz Rimban and Booma Cruz. 2004. The Rulemakers: How the

wealthy and well-born dominate Congress. Manila: Philippine Center for Investigative Journalism.

Cruz, Cesi, Julien Labonne and Pablo Querubin. 2017. “Politician Family Networks and Electoral

Outcomes: Evidence from the Philippines.” American Economic Review 107(10):3006–3037.

Dal Bo, Ernesto, Pedro Dal Bo and Jason Snyder. 2009. “Political Dynasties.” Review of Economic

Studies 76(1):115–142.

Dolan, Kathleen A. 2014. When Does Gender Matter?: Women Candidates and Gender Stereotypes in

American Elections. Oxford University Press.

Escobar-Lemmon, Maria and Michelle M. Taylor-Robinson. 2005. “Women Ministers in Latin

American Government: When, Where, and Why?” American Journal of Political Science 49(4):829–

844.

Escobar-Lemmon, Maria and Michelle M. Taylor-Robinson. 2009. “Getting to the Top: Career Paths

of Women in Latin American Cabinets.” Political Research Quarterly 62(4):685–699.

Fafchamps, Marcel and Julien Labonne. 2017. “Do Politicians’ Relatives Get Better Jobs? Evidence

from Municipal Elections in the Philippines.” Journal of Law, Economics & Organization 33(2):268–

300.

Folke, Olle and Johanna Rickne. 2016. “Electoral Competition and Gender Diﬀerences in Political

Careers.” Quarterly Journal of Political Science 11:59–102.

Folke, Olle, Johanna Rickne and Daniel M. Smith. 2016. “Gender and Dynastic Political Recruitment:

Theory and Evidence”. Working Paper. https://projects.iq.harvard.edu/files/pegroup/

files/folkeetal2016.pdf.

Fox, Richard L. and Jennifer L. Lawless. 2014. “Uncovering the Origins of the Gender Gap in

Political Ambition.” American Political Science Review 108(3):499–519.

Franceschet, Susan and Jennifer M Piscopo. 2008. “Gender quotas and women’s substantive

representation: Lessons from Argentina.” Politics & Gender 4(03):393–425.

Hirschman, Charles and Philip Guest. 1990. “The Emerging Demographic Transitions of Southeast

Asia.” Population and Development Review 16(1):121–152.

Jalalzai, Farida. 2013. Shattered, Cracked, Or Firmly Intact?: Women and the Executive Glass Ceiling

Worldwide. Oxford University Press, USA.

Jalalzai, Farida and Meg Rincker. 2018. “Blood is Thicker than Water: Family Ties to Political Power

Worldwide.” Historical Social Research / Historische Sozialforschung 43(4):54–72.

Krook, Mona Lena. 2009. Quotas for Women in Politics: Gender and Candidate Selection Reform

Worldwide. Oxford University Press.

Lawless, Jennifer L. and Kathryn Pearson. 2008. “The Primary Reason for Women’s Underrepre-

sentation? Reevaluating the Conventional Wisdom.” The Journal of Politics 70(1):67–82.

Lawless, Jennifer L. and Richard L. Fox. 2005. It Takes a Candidate: Why Women Don’t Run for Oﬃce.

Cambridge University Press.

McCoy, Alfred. 2009. An Anarchy of Families: The Historiography of State and Family in the

Philippines. In An Anarchy of Families: State & Family in the Philippines, ed. Alfred McCoy. Madison,

WI: University of Wisconsin Press pp. 1–32.

Montinola, Gabriella. 1999. “Parties and Accountability in the Philippines.” Journal of Democracy

10(1).

O’Brien, Diana Z. and Johanna Rickne. 2016. “Gender Quotas and Women’s Political Leadership.”

American Political Science Review 110(1):112–126.

Pande, Rohini and Deanna Ford. 2012. “Gender Quotas and Female Leadership”. World Bank

Working Paper. http://hdl.handle.net/10986/9120.

Querubin, Pablo. 2012. “Political Reform and Elite Persistence: Term Limits and Political Dynas-

ties in the Philippines”. Technical report. https://sites.google.com/site/pabloquerubin/

research/Querubin_Term_Limits.pdf.

Querubin, Pablo. 2016. “Family and Politics: Dynastic Persistence in the Philippines.” Quarterly

Journal of Political Science 11(2):151–181.

Robinson, Amanda Lea and Jessica Gottlieb. 2019. “How to Close the Gender Gap in Political

Participation: Lessons from Matrilineal Societies in Africa.” British Journal of Political Science

pp. 1–25.

Rosen, Jennifer. 2013. “The Eﬀects of Political Institutions on Women’s Political Representation: A

Comparative Analysis of 168 Countries from 1992 to 2010.” Political Research Quarterly 66(2):306–

321.

Schwindt-Bayer, Leslie A. 2011. “Women Who Win: Social Backgrounds, Paths to Power, and

Political Ambition in Latin American Legislatures.” Politics and Gender 7(1):1–33.

Smith, Daniel M. and Shane Martin. 2017. “Political Dynasties and the Selection of Cabinet Minis-

ters.” Legislative Studies Quarterly 42(1):131–165.

van Coppenolle, Brenda. 2017. “Political Dynasties in the UK House of Commons: The Null Eﬀect

of Narrow Electoral Selection.” Legislative Studies Quarterly 42:449–475.