You are on page 1of 35

Family and Politics: Dynastic Incumbency Advantage in the Philippines

Pablo Querubin

June, 2013

Abstract In many democracies a small subset of individuals enjoy, de facto, an electoral advantage. The existence of political dynasties, where individuals from a narrow set of families obtain larger vote shares and are more likely to access oce illustrates this phenomenon. In this paper, I study political dynasties in the Philippines and provide evidence of what I dene as dynastic incumbency advantage. More precisely I provide evidence of a causal eect of incumbency on the probability of having future relatives in oce. Using a regression discontinuity design based on close elections, I nd that candidates who barely win their rst election by a small margin are around 5 times more likely to have a relative in oce in the future, than individuals who barely lose and do not serve.

A previous version of this article circulated with the title Family and Politics: Dynastic Persistence in the Philippines. I would like to thank Daron Acemoglu, Esther Duo, James Robinson and James M. Snyder Jr. for all their comments and support. I would also like to thank participants in seminar presentations at Harvard, LACEA, MIT, NEUDC, NYU, UCLA and Yale for very good comments. Julien Labonne, Horacio Larreguy, Sahar Parsa and Roman Zarate also gave me very useful feedback. This paper wouldnt have been possible without the hospitality and generosity of many people in the Philippines during my visit in 2009. I thank Rep. Juan Romeo Acosta, Arsenio Balisacan, Emmanuel de Dios, Jose Ferraris, Rep. Risa Hontiveros, Nico Ravanilla, Juan Rafael Supangco, Jaime Veneracion and Sen. Juan Miguel Zubiri and the sta at CenPEG, the Institute for Popular Democracy, Innovations for Poverty Action and the Philippine Center for Investigative Journalism. The nancial support of Banco de la Republica and the Schultz Fund at MIT is gratefully acknowledged. Assistant Professor of Politics and Economics, New York University, E-mail: pablo.querubin@gmail.com

Introduction

Political power in most contemporary democracies is not equally distributed. Existing evidence suggests that some individuals enjoy a larger de facto electoral advantage relative to other candidates, which gives them greater access to elected oce. Examples include the electoral advantage enjoyed by incumbents, or candidates from incumbent parties (see for example Lee, 2008). The electoral advantage of some groups raises concerns to the extent that it can create barriers to entry into the political system, leading to the under-representation of some groups in society and to political capture by a narrow set of interests. The existence of political dynasties in many contemporary democracies such as India, Japan, the Philippines and the United States, amongst others, is another example of this broader phenomenon. Candidates from a narrow set of families enjoy an electoral advantage relative to other candidates, which allows members from these families to hold elected oce for many generations. Notable examples include the Kennedy family in the United States, the Gandhi family in India, and the Aquino family in the Philippines. But where does the electoral advantage of these families come from? It is possible that their success is rooted in sources completely outside the political system, such as the ownership of land, in wealth, social networks etc. It is also possible that their power, to some extent, ows from access to the political system by previous members of the family. Once in power a member of a family can use the instruments of oce to increase the political power of his or her family, for example through the use of public resources for personal enrichment or to gather political support for other relatives. Conceptually, one would like to disentangle the role of incumbency by previous relatives, from other observable and unobservable characteristics of the family correlated with political power. The conceptual challenge is similar to that underlying the electoral advantage of candidates from an incumbent party. Candidates from the party in power may enjoy an electoral advantage due to other characteristics of the party that make it popular in a given locality (and thus more likely to access oce) and not due to the incumbency status of the party per se. Understanding the extent to which political dynasties persist due to an incumbency advantage that spills over to other members 1

of the family, is important in order to assess the extent to which they may create implicit barriers to entry into the political system. In this paper, I study political dynasties in the Philippines and provide evidence of what I dene as dynastic incumbency advantage. More precisely, I provide evidence of a causal eect of incumbency on the probability of having relatives in oce in the future. The Philippines constitutes an interesting setting in which to address these questions as political dynasties are prevalent in many elected oces. For example, in the most recent 2010 election, roughly 50% of elected congressmen and governors were dynastic (had a previous relative in oce).1 Moreover, in 35 of the 80 provinces, the governor and congressman are related. I classify every congressional and gubernatorial candidate between 1946 and 2010 as dynastic or non-dynastic. A candidate is classied as dynastic if he/she had a previous relative serving in congress or as governor prior to the election. I document a substantial electoral advantage of dynastic candidates; their vote share is on average 16 percentage points larger relative to other non-dynastic candidates. The electoral advantage is even larger for dynastic candidates who are relatives of an incumbent in power at the time of the election; their vote-share is on average 26 percentage points larger relative to non-dynastic candidates. These simple comparisons however, confound the eect of incumbency by previous members of the dynasty with other characteristics of the family. To estimate the dynastic incumbency advantage, I use a regression-discontinuity design based on close elections to estimate the causal eect of holding oce on the probability of having a relative in power in the future. The main result of the paper is illustrated in Figure 1. Candidates who barely win their rst race (positive win margin) are roughly 12 percentage points more likely to have a relative serve in oce in the future than runner-ups (negative win margin). These results are robust to the inclusion of province and year xed-eects, to controlling for candidate characteristics such as party, gender and previous political experience and to alternative sub-samples. Establishing the existence and magnitude of a dynastic incumbency advantage contributes to our understanding of the sources of power and the determinants of electoral performance
1 In comparison, only 7% of current U.S. congressmen had a previous relative in congress. See Dal Bo, Dal Bo and Snyder (2009).

in democracies. The importance of access to oce suggests that the prevalence of dynastic politicians in the Philippines does not simply reect the existence of a xed group of powerful elites. De jure political power (holding elected oce), is an important determinant of electoral success of other members of the family and it allows new incumbents to entrench their family in power enacting de facto entry barriers into the political system. Identifying the exact mechanisms through which incumbency benets future relatives in the Philippines is beyond the scope of this paper. Nonetheless, I discuss alternative channels through which families reproduce their power across time. In particular, existing evidence by academics and journalists in the Philippines points to the use of public oce for dispensing patronage and for personal enrichment. This paper is related to various strands of literature. Previous papers have studied political dynasties in dierent contexts. In a closely related paper, Dal Bo, Dal Bo and Snyder (2009) nd that U.S. congressmen who barely win their rst reelection attempt and serve for more than one term are 6 percentage points more likely to have relatives serving in Congress after them, than those who barely lose their rst reelection attempt. They focus on incumbents and thus, they cannot estimate the eect of incumbency relative to other individuals who run but do never serve. Rossi (2011) studies a similar phenomenon in the Argentinean case. Other papers have documented the electoral advantage of dynastic candidates. See for example Feinstein (2010) for the U.S., Bjolken and Chandra (2012) for the case of India, and Asako et.al. (2012) for Japan. For the case of the Philippines the work by Lande(1965), Anderson (1988), McCoy (1994), Hutchcroft (1998), Sidel (1999), De Dios and Hutchcroft (2003), Simbulan (2005), Coronel et.al. (2007) and De Dios (2007), amongst others, provides an interesting assessment of the historical origins of political dynasties and of their consequences on the political system. To my knowledge however, this is the rst paper to estimate the causal eect of incumbency on the electoral success of family members. This paper is also related to the academic literature on estimating incumbency advantage for individual incumbents or candidates from the incumbent party (see Erikson, 1971, Gelman and King (1990), Levitt and Wolfram, 1997, Lee, 2008, and Ansolabehere and Snyder, 2004).

However, rather than estimating the eect of incumbency on the incumbent herself or on other party members, I estimate the eect of incumbency on other members of the family. Finally, this paper is related to the literature on the interaction of de jure and de facto political power (see for example Acemoglu and Robinson, 2008 and Baland and Robinson, 2008). The rest of the paper is organized as follows. In section 2, I provide some brief background on the political system in the Philippines. In section 3, I describe data sources and provide some descriptive statistics. In section 4, I present the main results of the paper together with robustness and validity checks. In section 5, I discuss alternative channels that may explain dynastic incumbency advantage. Finally, section 6 concludes.

Historical and Political Background

During almost 400 years of Spanish control, economic and political power was restricted to a small mestizo elite known as the principalia. In 1899 with the arrival of the United States, the power of these families was further consolidated. In order to gain their support and loyalty (necessary to pacify the islands), Americans introduced local elections in 1901, elections for a national legislature from single-member districts in 1907, and nally elections for the senate in 1916. It was precisely the introduction of positions of power initially at the local level that gave principalia families substantial economic and political power. The subsequent introduction of elections at higher levels of government (provincial and congressional district level), increased these families sphere of inuence. National politics and the central state became subordinated to the local dynamics of power. This power structure prevented the emergence of strong political parties with national platforms. The weakness of parties is often mentioned as an explanation for the importance of the family as a unit of political organization: parties did not emerge due to the local concentration of power and party weakness further consolidated the power of elite families (Anderson, 1988; Sidel, 1999 and Hedman and Sidel, 2000). A common argument, thus, is that access to public oce was an important source of

power of these families, an argument closely related to what I explore in this paper. In 1946, following a brief period of Japanese invasion during WWII, the Philippines became independent from the United States. There were no major changes to the electoral system: 24 Senators were elected every 6 years from the country at large, and elections were held every 4 years for provincial governors and congressmen. The House of Representatives was composed of members from single-member congressional districts. Seven congressional and gubernatorial elections took place between 1946 and 1972, when Ferdinand Marcos declared martial law and closed congress.2 In 1987, following the return to democracy, a new constitution introduced some changes to the political system. Congressional districts were reapportioned and term length was reduced from 4 to 3 years for congress, governors and other provincial and local oces. Similarly, the 1987 constitution introduced term-limits for all elected oces. Senators can now only be elected to two consecutive 6-year terms while congressmen and all other local ocials can only be elected to three consecutive 3-year terms. This reform however, has not been successful in limiting the political power of dynasties. Elected ocials are often replaced by their relatives after reaching the term-limit or switch to other elected oces. For more on the adaptive strategies of dynasties to term-limits, see Querubin (2012). Finally, parties have historically played a relatively minor role in Philippine politics. They are often personality based and only play a relevant role during elections in order to establish electoral alliances with provincial and local politicians. There are no major programmatic dierences. Party-switching (locally known as turncoatism) is a common phenomenon and opportunistic (often personal-based) coalitions are made in every legislature in order to secure support from the executive (see Hutchcroft and Rocamora, 2003).
The only gubernatorial elections during the period of martial law took place in 1980. Similarly, elections for a new unicameral parliament, the Batasang Pambansa, took place in 1978 and 1984. However, there is wide evidence that elections under martial law took place in the midst of widespread intimidation and persecution of political opponents. As a result, most races were dominated by Marcos allies from the Kilusang Bagong Lipunan (KBL) party.
2

3
3.1

Data and Descriptive Statistics


Data Sources

In this paper I focus on two elected oces: the House of Representatives and provincial governors. These are the most inuential oces at the provincial level of government.3 The Philippines is currently divided into 80 provinces that are headed by a provincial governor (top executive position). Provinces and cities are divided into multiple congressional districts, each electing a member of congress to the House of Representatives (lower chamber).4 There are currently 229 congressional districts in the Philippines, each composed of approximately 250,000 inhabitants.5 I collected the names of incumbents going back to 1901 from various archival sources.6 I also constructed a dataset with the name and number of votes received by all congressional and gubernatorial candidates for the period 1946-2010.7 To my knowledge, this is the rst paper to bring together these electoral data and to analyze them in a systematic way. The dataset on incumbents includes 2,863 individuals who served as governors during the period 1901-2010 or as congressmen during the period 1907-2010. The dataset on candidates covers 15 congressional and gubernatorial elections during the period 1946-2010, corresponding to 3,376 dierent races and 7,386 candidates.
The main sub-national level of government with elected ocials in the Philippines is the province which is the equivalent of a U.S. state. The top executive position in the province is the governor followed by a vice-governor and a provincial board (equivalent of a U.S. state legislature). The next sub-national level is the city/municipality (equivalent to a U.S. city/town) headed by an elected mayor, vice-mayor and body of councilors. 4 Nonetheless, currently 28 provinces have lone congressional districts and elect only one congressman from the province at large. Cities are entitled to elect at least one congressman to the House of Representatives. 5 The number of provinces and congressional districts has been increasing since 1907 due to reapportionment and the creation of new cities and provinces. There were originally 33 provinces and 80 congressional districts in the rst legislative elections in 1907. At the time of independence in 1946 the number of provinces had increased to 50 and the number of congressional districts to 133. 6 The names of provincial governors for the period 1901-1935 come from the Roster of Public Ocials available in the National Archives in Manila. Names of congressmen for the period 1907-1972 come from the Congressional Directories available in the House of Representatives in Quezon City. Data for the period 1987-2010 comes from the Commission of Elections and Coronel et.al (2007). 7 Electoral data for the period 1946-1972 were collected by hand from the original Canvass of Votes of the Commission of Elections available in microlm at the Center for Research Libraries. Data for the 1987 congressional and 1988 gubernatorial elections are available in Gutierrez, Torrente and Narca (1992). Electoral data for the period 1992-2010 were provided by the Commission on Elections and complemented with data from the Institute for Popular Democracy.
3

3.1.1

Dynastic Measures

As a rst step I establish the number of prior and posterior relatives in oce for all candidates in the dataset. This is done by matching the candidates family names, with the family names of prior and posterior incumbent governors and congressmen within the same province.8 However, bilateral descent in the Philippines implies a particular structure of family names that must be taken into account. The name of a Filipino male or single female takes the form: rstname midname lastname where midname corresponds to the mothers family name and lastname corresponds to the fathers family name. In the case of married women, names take the following form: rstname midname lastname-lastnamehusb where again midname corresponds to the mothers family name, lastname corresponds to the fathers family name and lastnamehusb corresponds to the husbands lastname. Thus, relatives are identied by nding a match of the midname, lastname or lastnamehusb within the same province.9 Relatives traced only by lastname would fail to identify wives, and some grandchildren. However, the results presented in this paper are similar if relatives are traced using only lastname. A natural concern with the above matching procedure is that individuals from the same province who share a midname, lastname, or lastnamehusb may not necessarily be related biologically to each other.10 While this is certainly a possibility, this is less of a concern in the Philippines than in other countries due to the peculiar way in which family names are
In the case of provinces that split into multiple provinces, posterior relatives for candidates in the original province are traced in all the dierent provinces into which the original province splits. 9 Several biographical sources were used to nd the midnames of as many incumbents as possible. For most of the post-1946 congressmen, midnames were found in the Congressional Directories available at the House of Representatives in Quezon City. In the case of new provinces, relatives are traced back to incumbents in the original province. 10 This matching procedure will identify almost all existing relatives in the dataset. The main concern is the existence of false positives, or matches that do not correspond to actual relatives.
8

distributed across the dierent provinces. In 1849, concerned with the arbitrary way in which Filipinos chose their surnames, Governor Narciso Claveria y Zaldua created a catalog with a list of 61,000 dierent surnames.11 A dierent set of surnames (often starting with the same letter) was assigned to each town and local ocials had to assign a dierent surname to the dierent family heads. As a consequence, common lastnames (such as Smith in the U.K. and U.S. or Gonzalez in Latin America) are not as prevalent in the Philippines. I also used various biographical sources to verify the accuracy of the relatives identied by the matching procedure for a sample of individuals.12 Nonetheless the possibility of some measurement error remains. Based on the above procedure, I construct several measures for each congressional and gubernatorial candidate in the dataset. The rst measure is a Dynastic Ever dummy which takes a value of one if the individual had a relative who served as governor or as congressman at any time prior to the election year. In order to minimize the likelihood of matches not corresponding to relatives I also create the dummy Dynastic Recent that takes a value of one if the individual had a relative who served as governor or as congressman in the 20 years prior to the year of election. Finally, I construct the dummy Incumbent Relative that takes a value of one if the candidate is related to an incumbent at the time of the election. Naturally: Incumbent Relative Dynastic Recent Dynastic Ever. In addition to the dynastic dummies, I also construct post-relative dummies that take a value of one if the candidate has any relatives who rst enter the House of Representatives or serve as governors at any time after the election (Post Relative Ever ) or in the 20 years following the election (Post Relative Recent ).
Claveria complained that the natives arbitrarily adopt the names of saints and this practice has resulted in the existence of thousands of individuals having the same surname. He added: I saw the resultant confusion with regard to the administration of justice, government, nance, and public order, and the farreaching moral, civil and religious consequences to which this might lead, because the family names are not transmitted from the parents to their children, so that it is sometimes impossible to prove the degrees of consanguinity for purpose of marriage, rendering useless the parochial books which in Catholic countries are used for all kinds of transactions. See National Archives of the Philippines (1973). 12 In particular, Coronel et.al. (2007) provide a list of current and previous relatives in oce for congressmen elected in 1992, 1998, 2001 and 2004 and governors elected in 2001 and 2004. This information is self-reported by the politicians in their Sworn Statement of Assets and Liabilities and was veried by the Philippine Center for Investigative Journalism.
11

3.2

Descriptive Statistics

The main descriptive statistics are presented in Figure 2, which illustrates the evolution of the dierent dynasty and post-relative measures across time and for each oce separately (House of Representatives and Provincial Governorship). The gures in Panel A reveal an increasing trend in the fraction of dynastic candidates in both oces. Such an increasing trend is not surprising for the Dynastic Ever measure given that the pool of potential previous relatives becomes larger with time. However, the gures in Panel B reveal a similar increasing trend for the Dynastic Recent variable. Recall that the Dynastic Recent only considers relatives in the 20 years prior to the election and thus, the pool of potential relatives does not mechanically increase with time. The most striking feature is the dierence in the fraction of dynastic candidates amongst losing and winning candidates. The prevalence of dynastic candidates is, on average, 20 percentage points larger amongst winning candidates than amongst losing candidates. This dierence is roughly similar across time, oces and the dierent dynasty measures. Such dierence suggests that dynastic candidates have a higher probability of being elected, a fact I document more systematically in the next section. Finally the gures in Panel C plot the evolution over time of the Post Relative Recent measure. The fraction of candidates with posterior relatives in government is almost three times larger amongst elected candidates than amongst losing candidates. Naturally, the fraction of candidates with posterior relatives in oce is declining in the most recent years as there are fewer elections for relatives of new incumbents to access oce. Nonetheless any time trends in the variables of interest will be fully controlled for in the analysis with year xed eects. Figure 2 illustrates some basic patterns in the data but does not establish any causal relation between dynastic status and electoral success. We explore these patterns more systematically in the next section.

Results

Given the similarity in the patterns of the dynastic and electoral variables in both oces, in the remaining analysis I pool congressional and gubernatorial elections. Similarly, for conciseness I only report results using the narrower (and more conservative) denitions of the dynastic and post-relatives measures that I will refer to as Dynastic and Post Relatives respectively. However, the results are similar if I use the measures based on the broader denition or if I report results for congress and governors separately.

4.1

Electoral Advantage of Dynastic Candidates

The rst set of results focuses on the electoral advantage of dynastic candidates. The results are presented in Table 1, where I report standard OLS regressions of a candidates vote share against dierent candidate characteristics. Regressions in all columns include a full set of province and year xed eects. Column 1 includes all candidates in every election between 1946-2010. Since the are multiple observations for some candidates (candidates who run in multiple election years), I cluster standard errors at the candidate level. The results show that dynastic candidates enjoy a substantial electoral advantage; they receive an average vote share 17 percentage points larger relative to other non-dynastic candidates. This dierence is statistically signicant. The result in column 1, however, may confound the eect of being dynastic with the electoral advantage of being an incumbent running for reelection (as seen in Figure 2, dynastic status is more prevalent amongst elected candidates). Thus, in column 2 I focus only on each candidates rst election attempt; that is, the regression in column 2 excludes all incumbents running for reelection. The coecient on the dynastic dummy remains essentially unchanged. In column 3, I dierentiate between two types of dynastic candidates, (i) those whose previous relative does not hold oce at the time of the election, and (ii) those whose previous relative is the incumbent congressman or governor at the time of the election. I label the latter cases Incumbent Relative. The results reveal a substantial electoral advantage for all types of dynastic candidates, but particularly for incumbent relatives. Candidates whose relative

10

is the incumbent at the time of the election, receive an average vote share 26 percentage points larger relative to other non-dynastic candidates. Dynastic candidates whose previous relative does not hold oce at the time of the election also enjoy a considerable advantage relative to non-dynastic candidates (11 percentage points) but this advantage is less than half the one observed for incumbent relatives. Finally, for comparison, in column 4 I include all candidates and add an incumbent dummy in order to compare the electoral advantage of dynastic candidates relative to that observed for incumbents seeking reelection. The estimates reveal a large electoral advantage of incumbents; they receive an average vote share 35 percentage points larger relative to other non-dynastic candidates. Nonetheless, this advantage is not much larger relative to that enjoyed by incumbent relatives which suggests the existence of considerable incumbency advantage spillovers from an incumbent to his/her relatives. The results in Table 1 are presented only for descriptive purposes, in order to illustrate the electoral advantage of dynastic candidates. However they do not necessarily provide evidence of a causal eect of incumbency on the electoral success of an incumbents relatives. The dynastic status may confound the incumbency of previous relatives with other characteristics of dynastic candidates such as wealth, access to land, networks or popularity that lie outside the political system and may be transmitted across generations.

4.2

Regression Discontinuity: Estimating the Dynastic Incumbency Advantage

In order to estimate the causal eect of incumbency on the electoral success of a politicians relatives, I follow a more direct approach. I compare the probability of having future relatives in oce for incumbents and candidates who run but do not serve. This approach is closely related to the regressions reported in Table 1. If candidates with previous relatives in oce (dynastic) are more likely to win an election, this implies that individuals who win an election are more likely to have future relatives in oce. However, a naive comparison of the electoral success of relatives of election winners and losers will lead, potentially, to misleading inference on the eect of incumbency. In practice, 11

winning and losing candidates are dierent along various characteristics. Thus, this naive comparison will confound the eect of incumbency with other characteristics of winning candidates. In order to estimate the causal eect of incumbency on the electoral success of a politicians relatives, one would like to compare two individuals who are identical in every respect (wealth, charisma, connections, interest in politics, etc.) but due to random reasons, only one of them becomes an incumbent. The dierence in the probability of having future relatives in oce between these two hypothetical individuals I dene the dynastic incumbency advantage. A regression discontinuity design based on close elections provides an empirical counterpart of this ideal counterfactual. The underlying identication assumption is that the outcome of close races is as good as random and thus, the only dierence between winners and losers of close elections is incumbency status. The regression discontinuity design I use in this paper follows Hahn, Todd and Van der Klaauw (2001), Lee (2008), Imbens and Lemieux (2008) and Meyersson (2012) and is based on estimating a regression of the following form:

P ost Relativeijt = + W innerijt + f (xijt ) + i s.t. xijt (h, h)

ijt

(1)

where P ost Relativeijt is a dummy variable that takes a value of 1 if candidate i in province j has any relatives in oce in the 20 years following the election at time t, and W innerijt is a dummy variable that takes a value of 1 if candidate i wins the election. The control function f (xijt ) corresponds to an n-th order polynomial of the forcing variable xijt which in the context of this paper, corresponds to the winning margin between the winner and runner-up of the election. The winning margin takes values between -1 and 1 and is positive for election winners and negative for runner-ups.13 The coecient of interest is which is the estimate of the dynastic incumbency advantage. Most approaches to the estimation of regression (1) rely on dierent congurations of
13 For example, a race in which the winner receives 40% of the votes and the runner-up receives 37% of the votes, the winning margin is 0.03 for the winner and -0.03 for the runner-up.

12

the control function f (xijt ) and dierent choices of the bandwidth h which determines the estimation sample. With smaller values of h the sample is restricted to individuals who win or lose the election by a very narrow margin. This resembles more closely the empirical counterfactual but comes at the expense of eciency due to small samples. In the benchmark specications, I follow Imbens and Lemieux (2008) who propose the estimation of local linear regressions. For the choice of h, I follow the optimal bandwidth choice rule proposed by Imbens and Kalyanaraman (2012). I also show the robustness of the results to alternative choices of bandwidth and of the control function f (xijt ). Ideally, estimates of should not rely much on the choice of bandwidth or in the specication of the control function. Throughout the analysis I focus only on the rst election of non-dynastic candidates (winners and runner-ups). The focus on the rst election is important for the identication assumption. The outcome of a close election involving an incumbent running for reelection or a seasoned candidate with previous electoral experience is less likely to be as good as random.14 The focus on non-dynastic candidates is important in order to avoid confounding the eect of incumbency of a candidates previously elected relatives with a candidates own incumbency eect on future relatives. For simplicity, throughout the analysis I only report estimates of a linear probability model of regression (1) (see Angrist and Pischke, 2009). However, all the results are qualitatively similar with Probit regressions. Finally, I defer a detailed assessment of the validity of the assumptions for the regression discontinuity design to section 4.3, after presenting the main results of the paper. 4.2.1 Reduced Form Estimates

Figure 1 shows graphically the main result of the paper. The dots show the fraction of candidates with future relatives in oce, averaged in 2% bins of the winning margin in the rst election. Local polynomial smoothing regressions using the raw (unbinned) data are shown, together with 95% condence intervals. The gure reveals a discontinuous jump at
I drop candidates whose rst election was in 2010 (last election in my sample) as they cannot have any relatives entering politics after them.
14

13

the threshold in the fraction of future relatives in oce. The magnitude of the discontinuity suggests that candidates who win their rst election by a small margin are roughly 12 percentage points more likely to have future relatives in oce than candidates who barely lose. Table 2 reports estimates of based on equation (1). For reference, in columns 1 and 2 I report standard OLS estimates on the full sample of the naive comparison between winning and losing candidates. Standard errors are clustered at the candidate level. The specication in column 2 includes province and year xed eects. The OLS estimates suggest that winning candidates are 24 percentage points more likely to have future relatives in oce than losing candidates. However, as discussed earlier, these cannot be interpreted as causal estimates of the eect of incumbency on the electoral success of future relatives. In columns 3 and 4, I report the benchmark regression discontinuity estimates based on local linear regressions using the Imbens and Kalyanaraman (2012) optimal bandwidth (in this case h = 0.27). Standard errors are clustered at the province level. The estimates provide evidence of a causal eect of incumbency on the probability of having future relatives in oce. Candidates who win their rst race by a small margin are 12 percentage points more likely to have future relatives in oce. The estimates are statistically signicant at conventional levels and remain essentially unchanged with the inclusion of province and year xed eects in column 4. In Table 3, I show the robustness of these estimates to alternative bandwidth choices and control functions. All standard errors are clustered at the province level and regressions in columns 2, 4 and 6 include province and year xed eects. In columns 1-4, I follow Angrist and Lavy (1995) and focus on a small discontinuity sample and drop the control function. Columns 1-2 report estimates for a 5% bandwidth and columns 3-4 for a 2.5% bandwidth. The estimates remain essentially unchanged and statistically signicant in spite of a sizable reduction in sample size. In columns 5-6, I follow Lee (2008) and consider the full sample of candidates but include a fourth order polynomial in the winning margin in the rst race (allowing for a dierent polynomial at both sides of the threshold). Again, the estimates

14

remain essentially unchanged and statistically signicant. The stability of the estimates to the choice of bandwidth and control function is reassuring. 4.2.2 2SLS Estimates

The results reported in Tables 2 and 3 correspond to reduced form estimates of winning or losing the rst election by a small margin. Nonetheless, some candidates who lose their rst election may run again and eventually win. In this case, the reduced form regressions underestimate the causal eect of incumbency since I am classifying as non-incumbent, losing candidates who eventually serve. To address this issue, I use a fuzzy regression discontinuity design (see Hahn, Todd and Van der Klaauw, 2001; Imbens and Lemieux, 2008). Essentially, this consists in using the outcome of the rst close election as an instrument for whether the candidate serves in elected oce. The rst stage regression corresponds to:

Servedijt == + W innerijt + f (xijt ) + uijt i s.t. xijt (h, h)

(2)

where Servedijt is a dummy that takes a value of one if candidate i eventually serves in congress or as provincial governor, and W innerijt , as before, is a dummy variable for whether candidate i wins his rst election. The second stage regression is given by:

P ost Relativeijt = + Servedijt + vijt where Servedijt is the predicted value from equation (2).

(3)

Table 4 reports the 2SLS estimates of for the benchmark local linear regression specication reported in Table 2 and for the alternative bandwidths and control functions reported in Table 3. I include province and year xed eects in all regressions and standard errors are clustered at the province level. As predicted, the point estimates are higher relative to the reduced form estimates. The estimate in column (1) suggests that candidates who serve in congress or as provincial governors are 17 percentage points more likely to have their relatives become congressmen or governors in the future, relative to candidates who run for these 15

positions but do not serve. The estimated eect is remarkably large. Given the average value of Post Relatives Recent for losing candidates who never serve (roughly 4%), the estimates imply that incumbency makes a candidate 5 times more likely to have a relative serve in oce in the future.

4.3

Robustness and Validity Checks

The regression discontinuity estimates presented in Tables 2-4 are only valid to the extent that some basic identication assumptions are satised. The underlying assumption is that the outcome of close elections is as good as random and does not depend on any other underlying characteristics of the candidates. In this section I present some basic validity checks of this assumption. One useful exercise proposed by McCrary (2008) consists in testing for a discontinuity in the density of observations around the threshold. For example, to the extent that close elections are manipulated and candidates in the sample lose more than 50% of the close races, we should observe a larger fraction of observations concentrated to the left of the threshold. In Figure 3, I plot the density of observations, averaged over equal sized bins. There is no evidence of sorting around the threshold. This is conrmed by a formal test reported at the bottom of the gure, that estimates the magnitude of the discontinuity. The estimate is very close to zero and is not statistically signicant. Another important validity check is to test for balance of candidate characteristics across the threshold. Unfortunately, very little information is available on candidate characteristics. Nonetheless, for every election for the period 1995-2007 I have information on the party of every single candidate. This allows me to test whether candidates from certain specic parties are more or less likely to win close races. I also compute a coalition dummy that takes a value of one if the candidates party belongs to the presidents coalition at the time of the election.15 For the period 1988-2007 I also use electoral data for other oces and create a previous experience dummy that takes a value of one if the candidate held a dierent
Coalition parties for the dierent election years are: LAKAS and LDP for 1995; LAKAS for 1998; LAKAS, NPC, LP, AKSYON and PDP for 2001; LAKAS, KAMPI, NP, LP and NPC for 2004; and LAKAS, KAMPI and LDP for 2007.
15

16

provincial or local elected oce prior to the election. Finally, for the whole sample (19462007) I use rst names to create a female dummy to test for a discontinuity in gender across the threshold. Table 5 reports regression discontinuity estimates, based on the local linear specication, to test for balance in these candidate characteristics across the threshold. In columns 1 and 2, I focus on elections for the 1995-2007 period. There is no evidence that candidates from parties in the presidents coalition or with previous political experience are more or less likely to win close races. The estimates are small and are not statistically signicant.16 Similarly, in column 3, I nd no evidence of sorting based on the candidates gender. While the results in columns 1-3, suggest no dierence in available candidate characteristics across the threshold, for robustness in Table 6 I report OLS and 2SLS regression discontinuity estimates after controlling for a full set of party dummies, as well as the coalition, previous experience and female dummies. Naturally, sample sizes fall considerably since these candidate characteristics are only available for 1995-2007.17 . Reassuringly, the magnitude and statistical signicance of the coecients remain essentially unchanged. Incidentally, by focusing on a shorter and more recent sub-period, the results in Table 6 also show that dynastic incumbency advantage operates in the short run. Incumbents who rst enter oce in 1995 or later are able to increase the electoral success of their relatives in a relatively short span of time. I will get back to this issue in the discussion in section 5. Finally, another potential concern is whether the outcome of close races depends on the identity of the opponent. Non-dynastic candidates in my sample who are running for these oces for the rst time, may be less likely to win a close race when facing an incumbent, a dynastic opponent, or a seasoned candidate who has run for these provincial oces in the past.18 In Table 5, I rule out this concern and nd no evidence of sorting around the
Similar regressions were performed using dummy variables for each individual party as dependent variables. Estimates are close to zero and are not statistically signicant. Results available from the author upon request. 17 I do not report specications based on a 2.5% bandwidth since sample sizes become prohibitively small. 18 For the U.S. case, recent papers by Snyder (2005), Caughey and Sekhon (2011) and Carpenter et al. (2011) criticize RDD studies arguing that incumbents and candidates from the party in control of state oces win noticeably more than 50% of close races.
16

17

threshold for candidates facing an incumbent (column 4), a dynastic candidate (column 5) or a seasoned candidate (column 6). All estimates are small and statistically insignicant. This suggests that candidates in my sample win on average 50% of close races, irrespective of the opponent they face. Nonetheless, in Table 7 I show the robustness of the results to considering alternative sub-samples based on the type of opponent. In Panel A, I exclude races where the candidates in my sample face an incumbent or a dynastic candidate in their rst race. Both the OLS and 2SLS estimates are very similar to those reported in Tables 2-4 for the whole sample. In Panel B, perhaps the most demanding specication, I restrict the analysis to races in which both the winner and runner-up are non-dynastic candidates running for the rst time. By construction, exactly 50% of the individuals in this sample win the race while the remaining 50% lose. The estimates are in line with those reported for previous samples and specications.

Discussion

The estimates in Tables 2-7 provide robust evidence of the existence of a dynastic incumbency advantage in the Philippines. A natural question in light of this result regards the mechanisms by which an incumbent is able to reproduce his or her political power and enhance the electoral performance of other relatives. Identifying, quantitatively, these exact mechanisms is beyond the scope of this paper. Nonetheless, as motivation for future research, in this section I discuss the plausibility of potential channels in light of existing qualitative evidence. A simple channel through which incumbency may aect the political success of future relatives is occupational choice. Just like with many other occupations, a politicians son or daughter may be more likely to follow the same career of his/her parents. Relatedly, it is plausible that relatives of former incumbents are able to inherit the experience, and political knowledge and skills required to run a province. This would make dynastic candidates better suited for public oce in the eyes of the voters and would explain their electoral success. While plausible, these explanations are unlikely to fully explain dynastic incumbency advantage in the Philippines. Channels operating through occupational choice or acquisition

18

of experience/skills occur in the long run. However, results that focus on the most recent 1995-2007 sub-period, reported in Table 6, suggest that incumbents are able to bring their relatives into politics a couple of years after entering oce for the rst time. Coronel et.al. (2007) report that dynastic candidates are very diverse in their professions and occupations prior to running for oce, and often run reluctantly to secure the political dominance of the family. Thus, many relatives of incumbents enter politics after their occupational choice has been made, and after years of experience in non-political occupations. In addition, there is no evidence that dynastic candidates have greater political experience in local oces prior to running for congress or the provincial governorship. For the 1995-2010 period, the fraction of dynastic candidates with previous experience in other oces (mayor, vicemayor, councilor, vice-governor or provincial board) is 24%. The corresponding fraction for non-dynastic candidates is essentially identical, 25%. This may simply reect that dynastic candidates are able to run directly for higher oce without having to go through local positions. Nonetheless, this is inconsistent with dynastic candidates being advantaged due to greater formal experience in other oces. Another potential mechanism behind dynastic incumbency advantage is name recall. In the absence of coherent and reliable parties, family names play the role of brands that provide information to the voters. Incumbents have greater exposure (both through the media and direct interaction with the voters) making it easier for them to create a brand name, relative to other candidates. Assesing the relative relevance of this mechanism is an interesting topic for future research. Perhaps the mechanism most often cited in the media and academic literature is the use of oce, in particular patronage and public resources, in order to entrench the family in politics. This can be inferred from the estimates reported in Table 1. Recall that the largest electoral advantage is enjoyed by candidates whose relative is an incumbent (and in control of spending, policies and employment) at the time of the election. The electoral advantage of candidates whose previous relative is out of oce at the time of the election is less than half the advantage enjoyed by incumbent relatives.

19

Incumbency and access to public resources is fundamental in order to fund the clientelistic networks and machinery that deliver votes. A politician can then bequeath these networks to her relatives in order to expand the political dominance of the family. In fact, clientelistic networks enjoy substantial economies of scale and they can support two or more members of the same family who run for dierent oces (i.e. congress and provincial governor). The use of oce to further the private economic interests of the family is also widely cited as a channel behind dynastic entrenchment. In the context of a renowned dynasty in the province of Cebu, Cullinane (1994, p.187) mentions,all the assets of the familys domain -revenues, land, agricultural commodities, industries, power and inuence - were derived from success at the polls. He then argues that much of the prots from their enterprises were invested in elections to guarantee the familys continued dominance. In the context of the 2007 elections, Fafchamps and Labonne (2012) also show that relatives of an incumbent are disproportionately more likely to be employed in the public sector. Similarly, they nd that relatives of unsuccessful candidates are disproportionately less likely to work in the public sector. This illustrates the way in which patronage can be used to benet an incumbents relatives, and punish members from the opposition. Finally, incumbents can also bequeath to their relatives their connections to the central bureaucracy, which allow them to secure the ow of pork barrel funds to their provinces and districts. In this sense, voters may nd themselves stuck in an equilibrium in which electing dynastic politicians is the best strategy when other districts are also electing dynastic (and connected) politicians. Otherwise, they risk losing their connections to the central bureaucracy and cutting the ow of funds to their district. This argument is similar to the one made by McKelvey and Riezman (1992) for seniority advantage.

Conclusions

In this paper I present evidence of a dynastic incumbency advantage. Incumbent congressmen and governors are roughly 5 times more likely to have a relative serve in these oces in the future, relative to similar candidates who run but do not serve.

20

These results provide important insights regarding the persistence of political elites across time. Political power does not ow uniquely from sources of power outside the political system such as landownership or wealth. De jure political power, that is, control of elected oce, is an important determinant of electoral success of other members of the family. Access to oce allows previously unconnected individuals (in my sample, non-dynastic candidates) to amass political power that they can bequeath to their relatives. As argued by Coronel et. al (2006, p.50) new families join the gilded circle with each election, and they and their kin have to keep winning at the polls in order to stay in that circle. At the same time, older dynasties fail in elections and they fade out of the political scene sooner or later. Establishing the exact mechanisms through which incumbency benets other relatives is beyond the scope of this paper and is an important question for future research. Some of the plausible mechanisms outlined in section 4, are in fact similar to those that allow party incumbency to benet other party members. However, the consequences in the case of dynasties are, potentially, very dierent. A crucial dierence between dynasties and parties is that membership in the former, by denition, is restricted to those related by blood (the one exception being marriage). Thus, a political system organized around dynasties creates tighter entry barriers into the political system and creates the potential for capture by a narrow set of interests. Similarly, future research should study the consequences of dynastic persistence on policy outcomes and public goods provision. This is important in order to determine whether reforms aimed at preventing the entrenchment of political dynasties are desirable.

References
Acemoglu, Daron and James A. Robinson (2008) Persistence of Power, Elites, and Institutions, American Economic Review, 98(1), March. Anderson, Benedict (1988), Cacique Democracy in the Philippines: Origins and Dreams, New Left Review, 169, May-June. Angrist, Joshua and Victor Lavy (1999) Using Maimonides Rule to Estimate

21

the Eect of Class Size on Scholastic Achievement. The Quarterly Journal of Economics. 114(2): 533-575. Angrist, Joshua and Jorn-Steen Pischke (2009) Mostly Harmless Econometrics. Princeton University Press, Princeton, NJ. Ansolabehere, Stephen and James M. Snyder Jr. (2004) Using Term Limits to Estimate Incumbency Advantages when Oceholders Retire Strategically, Legislative Studies Quarterly, 29(4): 487-515. Asako, Yasushi, Takeshi Iida, Tetsuya Matsubayashi and Michiko Ueda (2012), Dynastic Politicians: Theory and Evidence from Japan, Unpublshed Manuscript. Baland, Jean Marie and James A. Robinson (2008), Land and Power: Theory and Evidence from Chile, American Economic Review, 98(5): 17371765. Bjolken, Andali and Kanchan Chandra (2012), Dynastic Politics and Party Organizations: Why Family Ties Improve Electoral Performance in India, Unpublished Manuscript. Carpenter, Daniel, Brian Feinstein, Justin Grimmer and Eitan Hersh (2011) Are Close Elections Random? Unpublished manuscript. Caughey, Devin M. and Jasjeet S. Sekhon (2011) Elections and the Regression Discontinuity Design: Lessons from Close U.S. House Races, 1942-2008, Political Analysis, 19(4): 385408. Coronel, Sheila, Yvonne T. Chua, Luz Rimban and Booma B. Cruz (2007) The Rulemakers: How the Wealthy and the well-born dominate Congress, Philippine Center for Investigative Journalism, Quezon City. Cullinane, Michael (1994) Patron as Client: Warlord Politics and the Duranos of Danao in Alfred McCoy (Ed.) An Anarchy of Families: State and Family in the Philippines, Madison, WI, University of Wisconsin Press. Dal Bo, Ernesto, Pedro Dal Bo and Jason Snyder (2009) Political Dynasties, Review of Economic Studies, 76(1), January. De Dios, Emmanuel (2007) Local Politics and Local Economy in The Dynamics of Regional Development: The Philippines in East Asia, Arsenio Balisacan and Hal Hill (eds.),

22

Quezon City, Ateneo de Manila Press. De Dios, Emmanuel and Paul Hutchcroft (2003), Political Economy in The Philippine Economy: Development, Policies and Challenges, Arsenio Balisacan and Hal Hill (eds.), Quezon City, Ateneo de Manila Press. Erikson, Robert (1971) The Advantage of Incumbency in Congressional Elections, Polity, 3(3): 395-405. Fafchamps, Marcel and Julien Labonne (2012) Nepotism and Punishment: The (Mis)Performance of Elected Local Ocials in the Philippines. Unpublished Manuscript. Feinstein, Brian (2010) The Dynasty Advantage: Family Ties in Congressional Elections, Legislative Studies Quarterly, 35(4): 571-98. Gelman, Andrew and Gary King (1990) Estimating Incumbency Advantage Without Bias, American Journal of Political Science, 34: 1142-64. Gutierrez, Eric, Ildefonso Torrente and Noli Narca (1992) All in the Family: A Study of elites and power relation in the Philippines, Quezon City, Institute for Popular Democracy. Hahn, Jinyong, Petra Todd, and Wilbert Van der Klaauw (2001) Identication and Estimation of Treatment Eects With a Regression Discontinuity Design, Econometrica 69(1), 201-9. Hedman, Eva and John Sidel (2000) Philippine politics and society in the twentieth century: colonial legacies, post-colonial trajectories, London, Routledge. Hutchcroft, Paul (1998), Booty Capitalism: The politics of banking in the Philippines, Quezon City, Ateneo de Manila Press. Hutchcroft, Paul and Joel Rocamora (2003), Strong Demands and Weak Institutions: The Origins and Evolution of the Democratic Decit in the Philippines, Journal of East Asian Studies, 3, pp. 259-292. Imbens, Guido and Thomas Lemieux (2008) Regression Discontinuity Designs: A Guide to Practice, Journal of Econometrics, 142(2), pp.615-635 Lande, Carl (1965) Leaders, Factions and Parties: The Structure of Philippine Politics,

23

Yale University South East Asian Studies, New Haven, CT. Lee, David (2008) Randomized Experiments from Non-random Selection in U.S. House Elections, Journal of Econometrics, 142(2), pp.675-697. Levitt, Steven and Catherine Wolfram (1997) Decomposing the Sources of Incumbency Advantage in the U.S. House, Legislative Studies Quarterly, 22(1): 45-60. McCoy, Alfred (1994) An Anarchy of Families: State and Family in the Philippines, Madison, WI, University of Wisconsin Press. McKelvey, Richard and Raymond Riezman (1992) Seniority in legislatures. American Political Science Review 86(4): 951-965 Meyersson, Erik (2012) Islamic Rule and the Emancipation of the Poor and Pious?, Unpublished Manuscript. National Archives of the Philippines (1973) Catalogo Alfabetico de Apellidos, Manila. Querubin, Pablo (2012) Political Reform and Elite Persistence: Term Limits and Political Dynasties in the Philippines. Unpublished Manuscript. Rossi, Martin (2011) Family Business: Causes and Consequences of Political Dynasties, Universidad de San Andres, Unpublished Manuscript. Sidel, John (1999), Capital, Coercion and Crime: Bossism in the Philippines, Stanford, CA, Stanford University Press. Simbulan, Dante (2005), The Modern Principalia, Quezon City, University of the Philippines Press. Snyder, Jason (2005) Detecting Manipulation in U.S. House Elections. Unpublished Manuscript.

24

Figure 1 Posterior Rela0ves in Oce and Win Margin in First Elec0on


.3 0 Posterior Relative in Ofce .1 .2

-.3

-.2

-.1 0 .1 Win Margin in First Race

.2

.3

Horizontal axis shows the winning margin of non-dynas0c candidates (winners and runner-ups) in their rst race. Candidates with a posi0ve winning margin win the elec0on while those with nega0ve winning margins are runner-ups. Figure focuses on candidates within the op0mal bandwidth proposed by Imbens and Kalyanaraman (2012) (h=0.27). The dots show the frac0on of candidates with future rela0ves in oce, averaged in 2% bins of the winning margin. Local polynomial smoothing regressions using the raw (unbinned) data are shown (solid lines), together with 95% condence intervals (doVed lines), at both sides of the threshold.

Figure 2 Fraction of Candidates that are Dynastic, Dynastic Recent and with Posterior Relatives in Power By Election Year: 1946-2010 A. Fraction Dynastic Congress Governors
0.800# 0.700# 0.600# 0.500# 0.400# 0.300# 0.200# 0.100#
1949# 1957# 1961# 1965# 1969# 1987# 1995# 1998# 2001# 1946# 1953# 1992# 2004# 2007# 2010#

0.700# 0.600# 0.500# 0.400# 0.300# 0.200# 0.100# 0.000#

1947#

1951#

1955#

1959#

1967#

1971#

1988#

1995#

1998#

2001#

1963#

1992#

2004#

2007#

Losing#Candidates#

Elected#

B. Fraction Dynastic Recent


0.700# 0.600# 0.500# 0.400# 0.300# 0.200# 0.100#

Losing#Candidates#

Elected#

2010#

0.000#

Congress
0.600# 0.500# 0.400# 0.300# 0.200# 0.100#
1949# 1957# 1961# 1965# 1969# 1987# 1995# 1998# 2001# 1946# 1953# 1992# 2004# 2007# 2010#

Governors

0.000#

1947#

1951#

1955#

1959#

1967#

1971#

1988#

1995#

1998#

2001#

1963#

1992#

2004#

2007#

Losing#Candidates#

Elected#

C. Fraction Posterior Relatives Recent


0.500# 0.400# 0.300# 0.200# 0.100# 0.000#

Losing#Candidates#

Elected#

2010#

0.000#

Congress
0.600# 0.500# 0.400# 0.300# 0.200# 0.100# 1949# 1957# 1961# 1965# 1969# 1987# 1995# 1998# 2001# 2007# 1946# 1953# 1992# 2004# 0.000#

Governors

1949#

1957#

1961#

1965#

1969#

1987#

1995#

1998#

2001#

1946#

1953#

1992#

Losing#Candidates#

Elected#

Losing#Candidates#

Elected#

2004#

2007#

Figure 3 McCrary Density Test

1.25

Density 2.5

-.3

-.2

-.1 0 .1 Win Margin in First Race

.2

.3

Discontinuity est. = -0.026, s.e. = 0.137


Figure shows McCrary (2008) test for discon0nuity in the density of observa0ons around the threshold. Sample includes the rst race of non-dynas0c congressional and gubernatorial candidates for the period 1946-2007. Plot show observa0ons within the op0mal bandwidth proposed by Imbens and Kalyanaraman (2012) (h=0.27).

Table 1 OLS Regressions for Vote Share and Type of Candidate Dependent Variable is Vote Share (1) (2) (3) Dynastic 0.170 0.153 0.114 (0.007) (0.007) (0.007) Incumbent Relative 0.152 (0.017) Incumbent Province Fixed Effects Year Fixed Effects Observations R-squared YES YES 12,564 0.151 YES YES 7,158 0.148 YES YES 7,158 0.168

(4) 0.098 (0.006) 0.154 (0.012) 0.345 (0.006) YES YES 12,564 0.404

Robust standard errors, reported in paretheses. Standard errors in columns (1) and (4) clustered at the candidate level. Standard errors in columns (2)-(3) clustered at the province level. Sample in columns (1) and (4) includes all congressional and gubernatorial elections for the period 1946-2010. Sample in columns (2) and (3) includes only the first race of all congressional and gubernatorial candidates for the period 1946-2010. Dynastic is a dummy variable that takes a value of 1 if the candidate had a relative who served as congressman or governor in the 20 years prior to the election. Incumbent Relative is a dummy that takes a value of 1 if the candidate has a relative who is the incumbent at the time of the election. Finally, Incumbent is a dummy that takes a value of one if the candidate is an incumbent running for reelection.

Table 2 OLS Reduced Form Regressions Benchmark Local Linear Regressions Dependent Variable is Posterior Relatives Dummy (1) (2)

(3)

(4)

Winner Bandwidth (h) Control Function f(Win Margin) Province Fixed Effects Year Fixed Effects Observations R-squared

0.244 (0.013) 1 None NO NO 11,499 0.096

0.237 (0.013) 1 None YES YES 11,499 0.148

0.122 0.126 (0.028) (0.029) 0.27 0.27 Local Linear Local Linear NO NO 1,421 0.043 YES YES 1,421 0.116

Robust standard errors in parentheses. Standard errors in columns (1) and (2) clustered at the candidate level. Standard errors in columns (3)-(4) clustered at the province level. Sample in columns (1) and (2) include all Congressional and Gubernatorial candidates for the period 19462007. Sample in columns (3) and (4) include only the first race of all non-dynastic congressional and gubernatorial candidates (winners and runner-ups) for the period 1946-2007. Dependent variable is a dummy that takes a value of 1 if the candidate has any relatives entering congress or the provincial governorship in the 20 years following the election. Winner is a dummy that takes a value of 1 if the candidate wins the election.

Table 3 OLS Reduced Form Regressions Alternative Bandwidth and Control Functions Dependent Variable is Posterior Relatives Dummy (1) (2) (3) (4)

(5)

(6)

Winner Bandwidth (h) Control Function f(Win Margin) Province Fixed Effects Year Fixed Effects Observations R-squared

0.114 (0.033) 0.05 None NO NO 403 0.030

0.120 (0.038) 0.05 None YES YES 403 0.266

0.115 (0.052) 0.025 None NO NO 203 0.030

0.152 (0.070) 0.025 None YES YES 203 0.437

0.104 (0.042)

0.112 (0.042)

1 1 4th Order Pol 4th Order Pol NO NO 1,918 0.063 YES YES 1,918 0.124

Robust standard errors in parentheses. Standard errors clustered at the province level. Sample includes the first race of all non-dynastic congressional and gubernatorial candidates (winners and runner-ups) for the period 1946-2007. Dependent variable is a dummy that takes a value of 1 if the candidate has any relatives entering congress or the provincial governorship in the 20 years following the election. Winner is a dummy that takes a value of 1 if the candidate wins the election.

Served Bandwidth (h) Control Function f(Win Margin) Province Fixed Effects Year Fixed Effects Observations R-squared

(1) 0.168 (0.022)

Table 4 2SLS Regressions Dependent Variable is Posterior Relatives Dummy (2) (3) (4) (5) 0.160 0.156 0.163 0.163 (0.023) (0.045) (0.050) (0.074) 0.05 None NO NO 403 0.048 0.05 None YES YES 403 0.294 0.025 None NO NO 203 0.032

(6) 0.218 (0.101) 0.025 None YES YES 203 0.444

(7) 0.179 (0.017)

(8) 0.167 (0.021)

0.27 0.27 Local Linear Local Linear NO NO 1,421 0.038 YES YES 1,421 0.117

1 1 Pol Order 4 Pol Order 4 NO NO 1,918 0.054 YES YES 1,918 0.120

Robust standard errors in parentheses. Standard errors clustered at the province level. Sample includes the first race of all non-dynastic congressional and gubernatorial candidates (winners and runner-ups) for the period 1946-2007. Dependent variable is a dummy that takes a value of 1 if the candidate has any relatives entering Congress or the Provincial Governorship in the 20 years following the election. Served is a dummy that takes a value of 1 if the candidate ever serves in congress or as provincial governor.

Winner Bandwidth (h) Control Function f(Win Margin) Province Fixed Effects Year Fixed Effects Observations R-squared

Table 5 OLS Reduced Form Regressions Alternative Bandwidth and Control Functions Dependent Variable is: Coalition Previous Exp. Female Against Incumb. (1) (2) (3) (4) -0.115 -0.008 -0.011 -0.040 (0.118) (0.127) (0.028) (0.040) 0.27 Local Linear YES YES 425 0.207 0.27 Local Linear YES YES 425 0.192 0.27 Local Linear YES YES 1,421 0.105 0.27 Local Linear YES YES 1,421 0.377

Against Dynast. (5) -0.026 (0.036) 0.27 Local Linear YES YES 1,421 0.199

Against Seasoned (6) 0.022 (0.043) 0.27 Local Linear YES YES 1,421 0.278

Robust standard errors in parentheses. Standard errors clustered at the province level. Sample in columns (1)-(2) includes the first race of all non-dynastic congressional and gubernatorial candidates (winners and runner-ups) for the period 1995-2007. Sample in columns (3)-(6) includes the first race of all non-dynastic congressional and gubernatorial candidates (winners and runner-ups) for the period 1946-2007. Dependent variable in column (1) is a dummy that takes a values of one if the candidate's party is part of the president's coalition at the time of the election. Dependent variable in column (2) is a dummy that takes a values of one if the candidate held a provincial or local elected office prior to the election. Dependent variable in column (3) is a dummy that takes a values of one if the candidate is female. Dependent variable in columns (4)-(6) is a dummy that takes a value of one if the candidate is running against an incumbent, a dynastic candidate or a seasoned candidate, respectively. The independent variable, Winner, is a dummy that takes a value of 1 if the candidate wins the election.

Winner Served Bandwidth (h) Control Function f(Win Margin) Province Fixed Effects Year Fixed Effects Observations R-squared

Table 6 Robustness Checks I Controlling for Candidate Characteristics Dependent Variable is Posterior Relatives Dummy OLS Regressions 2SLS Regressions (1) (2) (3) (4) (5) (6) 0.181 0.170 0.164 (0.061) (0.082) (0.071) 0.178 0.229 0.190 (0.043) (0.110) (0.035) 0.27 Local Linear YES YES 421 0.250 0.05 None YES YES 110 0.536 1 4th Order Pol YES YES 710 0.232 0.27 Local Linear YES YES 421 0.232 0.05 None YES YES 110 0.537 1 4th Order Pol YES YES 710 0.198

Robust standard errors in parentheses. Standard errors clustered at the province level. Sample includes the first race of all non-dynastic congressional and gubernatorial candidates (winners and runner-ups) for the period 1995-2007. All regressions control for a female dummy, a full set of party dummies, a dummy for whether the candidate served prior to the election in a provincial or local office and a dummy for whether his party belongs to the president's coalition at the time of the election. Dependent variable is a dummy that takes a value of 1 if the candidate has any relatives entering Congress or the Provincial Governorship in the 20 years following the election. Winner is a dummy that takes a value of 1 if the candidate wins the election. Served is a dummy that takes a value of 1 if the candidate ever serves in congress or as provincial governor.

Table 7 Robustness Checks II: Alternative Samples Dependent Variable is Posterior Relatives Dummy OLS Regressions 2SLS Regressions (1) (2) (3) (4) (5) Panel A. Excluding races with an incumbent or a dynastic candidate Winner Served Bandwidth (h) Control Function f(Win Margin) Province Fixed Effects Year Fixed Effects 0.27 Local Linear YES YES 0.05 None YES YES 1 4th Order Pol YES YES 0.119 (0.043) 0.145 (0.054) 0.117 (0.067) 0.166 (0.031) 0.27 Local Linear YES YES 0.181 (0.067) 0.05 None YES YES

(6)

0.164 (0.028) 1 4th Order Pol YES YES

Observations 662 200 787 662 200 787 R-squared 0.180 0.408 0.175 0.181 0.408 0.177 Panel B. Keeping races where both winner and runner up are non-dynastic candidates running for the first time. Winner Served Bandwidth (h) Control Function f(Win Margin) Province Fixed Effects Year Fixed Effects Observations R-squared 0.27 Local Linear YES YES 478 0.225 0.05 None YES YES 142 0.443 1 4th Order Pol YES YES 568 0.265 0.125 (0.055) 0.127 (0.059) 0.119 (0.077) 0.177 (0.044) 0.27 Local Linear YES YES 478 0.221 0.167 (0.077) 0.05 None YES YES 142 0.444 0.232 (0.046) 1 4th Order Pol YES YES 568 0.259

Robust standard errors in parentheses. Standard errors clustered at the province level. Sample includes the first race of all non-dynastic congressional and gubernatorial candidates (winners and runner-ups) for the period 1946-2007. All regressions control for a female dummy, a full set of party dummies, a dummy for whether the candidate served prior to the election in a provincial or local office, and a dummy for whether his party belongs to the president's coalition at the time of the election. Dependent variable is a dummy that takes a value of 1 if the candidate has any relatives entering congress or the provincial governorship in the 20 years following the election. Winner is a dummy that takes a value of 1 if the candidate wins the election. Served is a dummy that takes a value of 1 if the candidate ever serves in congress or as provincial governor.