Who Enters Politics and Why?

Despite the importance of politicians, empirical work rarely examines who decides to enter politics and why. This survey presents conceptual issues in measuring political entry; reviews work on individual,organizational,and institutional determinants of political entry; and summarizes the main findings and puzzles related to the representation/competence trade-off in recent mi-crocensus studies on who runs for office. Fruitful directions for future work are highlighted throughout the article.


INTRODUCTION
Perhaps the most important agents of change in modern democracies are politicians.Politicians aggregate citizen preferences in legislatures, work for the development of their constituencies, and oversee the executive arm of the state.Despite their importance, who decides to enter political life and what consequences those decisions carry for democratic performance are questions rarely studied empirically.
In models of electoral accountability where voters are able to perfectly control those they elect, the identity of politicians should not matter for the outcomes we observe.But, of course, voters exert only imperfect control over who is elected to office, so the identity of those who enter politics should impact the outcomes of democracies. 1 This article reviews the literature on political entry with the aim to synthesize patterns where evidence is available, while identifying areas where research is particularly thin.
The survey begins with Section 2, which discusses conceptual issues related to measuring political entry.I argue that the empirical study of political entry faces a rare-event problem: Who runs for office is a minuscule fraction of the total set of people who are eligible to enter politics.Earlier work disregarded this problem in its focus on people who had already entered politics.More recent work, aware of the inferential challenges of that approach, adopts two coping strategies: First, scholars tend to examine latent political ambition, on which there is greater variation, rather than formal political entry.Second, scholars focus on professions that already give rise to many politicians, allowing them to examine the decisions of people who do not enter politics as well as those of people who do.I discuss the costs and benefits of these two approaches and point out how work that makes use of behavioral data and large data sets is able to overcome some of the challenges.
Next, I examine the literature on the determinants of political entry in Section 3. I outline a simple framework for thinking about the interactions between the benefits of office, the cost of contesting, and the probability of getting elected, and how these may determine whether a person decides to enter a political race.I then survey the literature on these factors and relate them to organizational determinants of the decision calculus of political parties and other organized groups, as well as institutional determinants such as electoral rules and other institutional arrangements that impact political entry.
Finally, in Section 4, I consider whether there exists a trade-off between having a political class that is competent and one that is representative of the electorate.To examine this debate, I synthesize recent evidence from studies that employ microcensus data to provide a rich description of who runs for office from the entire population of eligible office holders.I highlight common patterns in representativeness along the dimensions of gender, race and ethnicity, wealth and class, and age.I also discuss measures of the competence of the political class, including cognitive ability and education.I argue that the parameters of the debate over the competence/representativeness trade-off remain muddled, and evidence on the question remains thin.
I offer a few caveats before we proceed.First, this article takes a broad view of political entry, examining work on selection by parties, formal political candidacy, and the earlier formation of political entry ambition.Second, the focus of this review is work in political science, with occasional reference to work in other disciplines. 2Third, this review mostly analyzes empirical rather 1 Ashworth (2012, p. 186) discusses the difference between pure moral hazard models, where all candidates are identical (making voters indifferent between them), and models that allow for candidate heterogeneity. 2 See Besley (2005) and Dal Bó & Finan (2018) for reviews on this topic in economics.

Political Ambition as a Latent Measure of Political Entry
One way to resolve the empirical challenge of political entry being a rare event is to focus on political ambition: the innate probability that a person will enter politics.In our simple exercise above, it is easy to imagine that even if eight candidates are identified in a random sample of 1,000 citizens in Khyber-Pakhtunkhwa, potentially many more people carry ambitions to run.
The examination of political ambition emerged with the classic study by Schlesinger (1966, p. 4), who described "political ambitions" as being shaped by a person's "office goals."This conceptualization of political ascent argues that people take actions which enable them to reach their goals of advancing in their political careers.This has since come to be known as progressive ambitionthat is, the ambition to progress up the political ladder-as opposed to nascent ambition, which is ambition associated with first formal political entry.The underlying logic is that political aspirants, both those already in politics and those hoping to join for the first time, are constantly calculating what actions are likely to yield the highest return to time and resources when deciding to enter a race for political progress.
2.1.1.Ambition in earlier work: desire and ability.This definition of political ambition portrays each office-eligible person in the population as a rational agent weighing the costs and benefits of political entry.The latent probability of entry at any one point is therefore the person's political ambition.Indeed, this is how the term was first formalized by Black (1972), who presented a simple cost/benefit calculation that a prospective politician carries out: u = PB − C, where P is the "candidate's estimate of the probability that [they] can obtain an office should [they] attempt to seek it" (p.146), B is benefits that accrue from winning office, and C is the cost of running for office.If the probability-weighted benefits are larger than costs, and if these net benefits are larger than those offered by alternative options in front of the person, they will decide to run for office.(I review the determinants of these terms in Section 3.2.)Ambition thus included not only the desire to enter politics but also the ability to do so.Additionally, both of these aspects of political ambition were seen as malleable based on changing circumstances.For Black (1972, p. 145), office-seekers, "rather than being driven by excessive ambition. ..tend to develop ambition slowly as a result of their changing circumstances. . . .Thus, the system does not cause either ambition or success in a direct sense; what it does is to determine indirectly the kind of [people] whom we will find in various types of offices."2.1.2.Ambition in recent work: desire.More recent work conceptualizes political ambition somewhat differently.It focuses on the desire for political entry as the more important object of study.For example, Lawless & Fox (2005, p. 3) express the more recent meaning of political ambition as "the desire to acquire and hold political office through electoral means."4Although I argue below that empirical researchers do still consider ability to run, recent work does not directly discuss ability.
Typically, ambition is measured through survey questions that ask respondents about their desire to enter politics.For instance, Lawless & Fox (2010, p. 195) ask the following question to measure ambition: "If you have never run for office, have you ever thought about running for office?"Responses are coded as "Yes, I have seriously considered it," "Yes, it has crossed my mind," or "No, I have not thought about it."This broad way of measuring ambition has been almost universally adopted.As an example, a recent survey by Crowder-Meyer (2018) asks: "In the last few years, have you ever thought about running for political office?" One reason for thinking of political ambition as a desire for political entry relates to the muchneeded focus of recent work on understanding reasons why political leadership remains noninclusive despite large gains in political participation among women and other minorities across the globe.For instance, research on the gender gap in political candidacy in the United States considers whether women are differentially desirous of political entry before considering whether they are able to run (Lawless 2015).One takeaway from this line of argument is that spurring women's desire to enter will reduce the gender gap in candidates.Some recent work, however, shows that only the interaction of desire and ability can produce meaningful changes in who we see in political office (Bernhard et al. 2019).

Narrowing the Study Population Mitigates the Rare-Event Problem of Political Entry
Even though ambition has been studied in terms of the desire to enter politics in recent work, I argue in this section that scholars nevertheless circumscribe their research in terms of an individual's ability to enter-the second aspect of ambition recognized in earlier work.Scholars select a study population that is already a high-ability one relative to the entire population of officeeligible citizens.This is done because it helps resolve the issue of studying an empirically rare event.
Comparing the study population to the complete office-eligible reference population was not a major concern in earlier work that examined political careers among people already in politics.Schlesinger (1966), for instance, drew empirical inferences by studying the careers of existing politicians with what were called "career data" or biographies of popular political personalities.As has now been well documented, inferences about the decision to enter politics are vulnerable to bias if derived from studying only people who have already entered (see Coppock 2019 for a discussion of this in the audit experiment literature).
Fowler & McClure (1990), recognizing this issue, argue that there exist "unseen candidates" who never make it into the study of political entry because they decide not to enter, and that their decision merits closer examination.The way Fowler & McClure define these unseen politicians is important.They argue that unseen congressional candidates have three traits: "First, none is obviously a political crank.Second, each has given some serious thought, at least briefly, to running for Congress.And third, each has some real prospect, however slim, of actually ending up [in office]. . . .As a result, all that first appears to distinguish the unseen candidates from the declared ones is that the latter publicly announce their intention to campaign for a seat" (Fowler & McClure 1990, p. 2).In other words, unseen candidates are politicians who were on the margin of running, but ended up not throwing their hat into the ring.
One way of interpreting these marginal types is that their ability to run for office is similar to that of actual candidates, but some barrier has prevented them from running.That barrier could be related to personal circumstance or could be institutional in nature.By studying the decision to enter for both the marginal unseen candidates and the ones who barely decided to run for office, we can infer what made some of the candidates emerge into politics from the initial pool that included both types.
Several scholars have built on this work by taking a similar approach to defining a study population.Most famously, Lawless and Fox established the Citizen Political Ambition Panel Study to understand gender differences in the propensity to run for office.They survey 6,800 professionals, equally divided between men and women, in the fields of law, business, and education (Fox & Lawless 2004).Instead of focusing on specific professions, Gulzar & Khan (2018) carry out a field experiment encouraging candidacy among "marginal politicians," whom they identify by asking a random subset of the electorate to nominate people who are on the margin of running for office but may or may not actually run.
The method of defining the study population used in these studies is essentially zooming in on the part of the population that has a relatively high ability of running for office.By holding ability somewhat fixed at this high level, studies of political ambition within this group can focus on the desire to enter politics while still incorporating some notion of an individual's ability to run.Although the inferences drawn are valid for the specific population being studied (e.g., people from a particular profession), there are two potential issues with this approach.First, in focusing on the marginal pool, scholars are unable to speak to how political entry operates in the entire population.We might need to know the latter if we are to understand how politics can include people from nonelite backgrounds.Second, the examination of entry decisions only among high-ability types may cause scholars to miss how factors boosting political entry can have larger marginal impacts on entry among those not studied.For example, if a study finds that offering financial support to lawyers does not increase candidacy among lawyers, we may not want to conclude that offering financial support to other people will similarly not work.Some recent work moves beyond examining political entry through what scholars have termed marginal, unseen, or eligible candidates.Crowder-Meyer (2018) studies political ambitions among people who are not typically from elite professions in the United States.Gulzar & Khan (2020) study candidacy among all office-eligible men in their study area in Pakistan.Cruz et al. (2017) examine complete network data in the Philippines to see if central individuals are more likely to run.All these studies highlight that political entry exists and can be improved among people who are traditionally thought of as carrying lower ability.An open avenue for research is therefore the exploration of the interaction between ability to enter politics and desire to run.

Behavioral Measures of Political Entry and the Political Entry Pipeline
Do stated measures of political entry like political ambition correlate with eventual candidacy?That is, do people who express high ambition on a survey actually end up running for office, and vice versa?Addressing this question is challenging in the absence of exogenous variation in political ambition because ambition itself is correlated with many other covariates we can observe among prospective political candidates.
Evidence to date is mixed.On one hand, the analysis by Lawless & Fox (2005, ch. 7) suggests that, controlling for some observables, political interest is not correlated with actual candidacy, perhaps due to structural barriers.On the other hand, Green & Conroy (2020) show that initial statements of interest from political aspirants predict eventual candidacy.
Several recent papers have attempted to focus on behavioral measures to circumvent the potential issue of cheap talk in stated measures of political ambition.For instance, Preece & Stoddard (2015) test whether people attend a training session organized by a political party, Broockman (2014) codes whether subjects respond to an email to measure interest in entering politics, and Gulzar & Khan (2020) trace actual candidacy in official election data for their experimental subjects.Relatedly, some studies focus on intermediate behavioral measures of political participation as early stages of formal political entry.Lundin et al. (2016) study whether winning an office in high school affects running for a formal political office in the future.Weghorst (2021) makes the case that opposition candidacy in authoritarian regimes is the result of early-life civic activism instead of party activism.
The questions that remain particularly fruitful for future work relate to tracing political entry from ambition to formal candidacy.How does incipient interest in running (i.e., ambition) relate to specific actions (e.g., participation in a training session) and lead people to decide to run for a formal political office (Ravanilla 2020)?More broadly, scholars may consider connecting political candidacy with the literature on political participation.Are there complementarities between being an active political participant more generally and political entry?Thinking of the political entry decision as a spectrum with intermediate behavioral outcomes could be a particularly powerful way of assembling evidence on future candidates from incipient interest to their actual candidacy, and on the steps in between where this progression might break down.

THE CALCULUS OF POLITICAL ENTRY
What factors affect people's decision to enter politics?I divide the literature on the determinants of political entry into three levels: individual, group, and institutional.In the formulation proposed by Norris (1997, p. 1), individual factors relate to the supply side of political entry, group factors relate to recruitment and the demand of gatekeepers, and institutional factors relate to the political system.

How Do Individuals Decide to Enter the Political Race?
Recall Black (1972)'s view of a prospective politician's cost/benefit calculation, introduced in Section 2.1.1:u = PB − C, where P is the probability of winning office if one runs, B is the benefit of winning office, and C is the cost of running.The person will decide to run if the probabilityweighted benefit is greater than the cost and the net benefit is greater than the perceived alternative options. 5n costs, Hall (2018, p. 175) writes that in the United States, it is impossible to quantify in any precise manner the full costs of running for office.Certainly, some are numeric, like the amounts a candidate must raise, the number of hours a candidate is expected to spend campaigning, and the amount of salary a candidate foregoes while running.Others, like the stress a campaign places on a candidate's family, the personal distaste and shame that comes along with the incessant public pandering candidates must do, and the sheer boredom of the endless banquets that candidates must attend, are harder to quantify.
As a consequence, high costs may prevent nonelites from running for office.A field experiment in Pakistan finds that help with procedural costs of filing papers through the services of a lawyer can enable participation and election in political office, suggesting that nonelites who are not competing in elections may be electable if they enter the race (Gulzar & Khan 2018).Prospective candidates with an advantage in raising money for campaigns, therefore, may also be more likely to run for office (Bonica 2020).
Prospective politicians may also decide to enter the race for policy and private returns.Indeed, in citizen-candidate models, political aspirants represent certain policies that voters can elect to office (Besley & Coate 1997, Fearon 1999).In addition, Fisman et al. (2014), Truex (2014) and Eggers & Hainmueller (2009) show that there exist substantial private returns to holding public office even after people leave office.These long-term benefits can drive the initial decision to run.Fisman et al. (2014, p. 807) note that "official salaries, private sector opportunities after leaving office, and also nonsalary earnings while in office, legal or otherwise" may motivate an individual to run.
In many countries, elected politicians are also paid wages.Carnes & Hansen (2016) consider data from the United States to argue that higher wages for local and state level politicians are not associated with more nonelite politicians.They conclude that "activists and political observers should stop saying that raising legislative salaries would make holding office more accessible for middle-and working-class Americans or that it would reduce class-based political inequalities" (Carnes & Hansen 2016, p. 709).Theoretical work suggests that higher salaries might induce higher-quality politicians to enter-although, as I discuss in Section 4.2, we might need more work to unpack what we mean by "quality" in order to reconcile conflicting empirical findings.Caselli & Morelli (2004) argue that when the returns from holding office are sufficiently large, high-quality citizens run for office; however, "when these returns are low, high-quality citizens choose to lead private lives, and voters are forced to make do with low-quality candidates" (p.760).Mattozzi & Merlo (2008) show that a higher salary decreases the average quality of politicians.Interestingly, in Peru, Bandiera (2020) provides evidence that instead of enticing more corrupt politicians to seek office, the opportunity to extract illegal mining rents corrupts the existing political class.
Pecuniary benefits can yield different responses from prospective politicians because other motivations might substitute or complement monetary returns from office (Benabou & Tirole 2003).Unfortunately, empirical work on how such motivations affect the decision to enter remains rare.For instance, Broockman (2013) shows that intrinsic motivations, beyond electoral concerns, can drive in-office behavior of politicians.In a field experiment, Gulzar & Khan (2020) show that encouraging people to run for office for prosocial reasons changes who runs and improves the alignment of policy with citizens' preferences.Weghorst (2021) argues that in the absence of officerelated benefits-those that accrue after political office is won-people may seek candidacy with parties that have little chance of winning because of intrinsic benefits they glean from doing so.
Examining how incentives interact with outside options individuals possess offers opportunities for further work.For instance, expanding the set of studies that examine nonpecuniary incentives to seek political office is perhaps particularly important in contexts where political jobs do not pay a competitive salary that is sufficient to attract competent politicians to office (the meaning of "competent" is discussed in Section 4.2).Low salaries could attract people looking to extract personal rents from office or those who are already rich and are able to support themselves from personal income while in office, but if the political job entails intrinsic incentives, then people with a prosocial orientation might be encouraged to enter (Gulzar & Khan 2020).
Research shows that, besides the benefits and costs of competing for office, subjective beliefs about the political environment can shape a person's proclivity to enter.Recent work shows that behavioral factors can create competition aversion (Kanthak & Woon 2015) and lead people to undervalue their own electability (Lawless 2015).As a consequence, people may be less likely to run for office because they underestimate their chances of winning or because their subjective evaluations of benefits and costs are skewed.
These three factors can interact in interesting and important ways that influence the kinds of people who put themselves forward for political office.For instance, in the context of American elections, Hall (2018) and Thomsen (2017) argue that as the cost of running for office increases, the candidates contesting are more likely to be ideologically extreme because their net policy payoffs from running are higher than those of moderate candidates.

Recruitment and Institutional Determinants
Given the simple cost/benefit framework outlined above, it is easy to conceptualize how recruitment and institutional factors-those that are external to the person deciding to run-can manipulate certain parameters of the calculus, or at the very least, the perception of those factors by a potential political entrant.In this sense, there is a direct connection between the internal decision calculus (one that influences the supply of politicians) and the external calculus (one that influences the demand).Who we see in office is, of course, a product of those two forces.

Determinants of entry at the group level.
Recruitment by political parties and other groups impacts political entry (Norris 1997, Smith 2018).For instance, gatekeepers in parties often identify and recruit candidates from their own networks, a process that disadvantages women candidates (Cruz et al. 2017, Fox & Lawless 2010, Karpowitz et al. 2017).Party elites actively alter the choice of candidates available to voters by directing resources and attention to their preferred candidates (Cohen et al. 2009, Dancygier et al. 2015, Galasso & Nannicini 2011, Gulzar et al. 2021a, Hassell 2017, Shaukat 2019).In the absence of party support, candidates struggle to compete and may be more likely to drop out of the race (Broockman 2014).The organizational structure of political parties also matters: More centralized parties are better able to influence and control candidate selection decisions (Rahat & Hazan 2001).
Overall, candidate selection is a complex decision process for elites, and it is not obvious that party leaders are always looking to maximize the chances of victory when selecting candidates.Additional considerations may also play a role.Party leaders may prefer more loyal candidates, perhaps those with a longer tenure in the party, over those who may do well on policy once in office (Auerbach & Thachil 2019, ch. 5).Cirone et al. (2020) show, for example, that party leaders like to renominate incumbents and senior party members, while Fiva et al. (2020) argue that parties aim to balance candidate lists geographically.Even when these decisions are taken with electability as a chief concern, party leaders' beliefs about who is electable might not be accurate.However, Gulzar et al. (2021a) and Casey et al. (2019) show in field experiments in Nepal and Sierra Leone, respectively, that party leaders may be responsive to polling information when it exists and that this responsiveness may improve electoral returns for both parties and voters.Smith & Tsutsumi (2016) show that in Japan, recruitment by party elites that allowed party outsiders to apply led to candidates who were closer to voter preferences.
From candidates' perspective, an expectation of bias in party recruitment can deter some people from putting themselves forward (Butler & Preece 2016).For instance, women often believe that as recruits they will receive less financial and strategic support from party leaders than male candidates will and are thus less likely to respond to recruitment efforts.Even among highly active party members, there can be systematic differences in the degree to which certain groups, particularly women, respond to recruitment efforts (Preece et al. 2016).On the other hand, there is evidence that outreach efforts and training by elites and other organizations can substantially improve the candidate pool (Lundin et al. 2016, Preece & Stoddard 2015).A field experiment in the United States shows that encouragement by party leaders can motivate people to seek and get elected to higher-level offices (Karpowitz et al. 2017).A field experiment with real election outcomes in the Philippines shows that leadership training can be a particularly effective method to boost candidacy among civic-minded people (Ravanilla 2020).
Another determinant of candidacy is historical persistence, which has given rise to dynastic political families around the world (Chandra 2016, Cruz et al. 2017, Dal Bó et al. 2009, Querubin 2016, Smith 2018).While legacy candidates come from privileged backgrounds and represent a narrow range of occupations, resulting in adverse selection of candidates, there are both demandand supply-side reasons for their preponderance in politics today.There is evidence that voters may carry a preference for dynastic leaders over nondynastic ones (Chandra 2016).Smith (2018) argues that not only are dynastic candidates more likely to put themselves forward for candidacy but perceived inherited incumbency value may also make their selection by party leaders more likely.There is evidence from India that dynastic candidates negatively impact GDP growth (Dar 2018) and public goods provision (George & Ponattu 2019).Cruz et al. (2017) empirically isolate the political importance of family networks for electoral outcomes using a data set of over 20 million individuals in 15,000 villages in the Philippines; they find that family networks matter for the "organizational and logistical" advantages that they confer and have "less to do with elite status, wealth or name recognition" (p.3034).Consequently, family networks "create barriers to entry for candidacy, impede political competition, and weaken mechanisms of electoral accountability." Finally, studies also demonstrate that social incentives and peer effects matter for candidacy decisions.For instance, Grossman & Hanlon (2014) demonstrate that the extent to which local communities can monitor their leaders affects whether high-quality candidates put themselves forward for candidacy.In a field experiment in Pakistan, Gulzar & Khan (2020) show that publicly highlighting political office as a way to improve community welfare increases candidacy, but highlighting that becoming a politician can yield respect and status depresses candidacy.Grossman et al. (2020) present field experimental evidence from Uganda that term-wide transparency initiatives around incumbent performance create space for challengers to enter the race in subsequent elections.

Institutional determinants of entry.
The way elections are set up can also affect political entry.Research on electoral rules shows that selection rules affect candidacy decisions (Arora 2020, Avis et al. 2017, Beath et al. 2016, Bueno & Dunning 2017, Grossman 2014).For instance, Rule (1987) presents evidence from several countries to show that women are more likely to get elected under proportional representation systems than under single-member district majority systems.In Brazil, Bueno & Dunning (2017, p. 347) show that "the reduction in the descriptive gap [between whites and nonwhites] appears larger for offices elected through proportional representation, such as federal and state deputies, senators, and city councilors, than for executive offices elected through winner-take-all systems, such as governor and mayor."Arora (2020) notes that in villages in Gujarat, India, politicians elected via community consensus are younger and more educated than those elected by secret ballot.However, these politicians tend to have worse governance outcomes in terms of reduced local expenditure and reduced targeting of workfare employment.
Political affirmative action through electoral reservations has been adopted around the world, meeting broad success.For example, political reservations can set certain areas on an inclusive equilibrium where the candidate pool continues to be representative of the underlying populations even after reservations are withdrawn (Besley et al. 2017;Bhavnani 2009Bhavnani , 2017;;O'Brien & Rickne 2016).Lawless (2015, p. 352) contends that in countries where quotas have been instituted to increase the representation of women, the legislatures see substantial increases in women's political representation.Franceschet & Piscopo (2008) present evidence of this from Argentina, while Goyal (2020) presents evidence of this from India.Besley et al. (2017) andO'Brien &Rickne (2016) show that the introduction of a "zipper-quota" mandating women's representation on the candidate list created space for women to be elected.Gulzar et al. (2020) show that political quotas in India improved economic outcomes for the targeted minorities without hindering outcomes for nontargeted minorities, or overall development.
Campaign spending rules may also have an impact on political entry and selection.Avis et al. (2017) show that imposing limits on campaign spending creates a larger pool of candidates that are on average less wealthy.However, Gulzar et al. (2021b), who study the case of Colombia, do not find similar effects.

WHO ENTERS POLITICS? AN ANALYSIS OF POLITICAL COMPETENCE AND REPRESENTATION IN RECENT MICROCENSUS STUDIES
I have argued that the literature's focus on what various authors call unseen, marginal, and eligible politicians is an empirical shorthand for examining political entry in a population where the probability of political entry is already relatively high.This means that in describing the political class as a whole and understanding who enters politics, we still need research that compares politicians and political aspirants with the entire reference population.One conception of the reference population is the entire pool of office-eligible people.Comparing with this benchmark allows us to understand the degree to which democracy delivers representative and competent leaders.This approach also avoids some of the problems discussed in Section 2.2.

Income level
Population Candidates Elected

Figure 1
Example of figures that can be built using census data (the data here are fictitious).The income distribution across seven bins for political candidates is skewed to the right of the modal income of the entire population; almost no one from the lowest income bracket runs for office.This skew is further exacerbated when we consider who is actually voted into office from among the candidates.
In this section, I review descriptive evidence on who enters political office, focusing in particular on recent papers that make use of microcensus data to compare the political class against the population they are representing.This new line of research is inspired by Dal Bó et al. ( 2017), who compiled administrative register data on the entire population of Sweden greater than 16 years of age between 1979 and 2012.These data include not only standard census-type characteristics but also a host of other variables, like personality traits and cognitive scores, that enable the analysis of specific traits that the political class might possess.
Others have used similar data in other contexts.There are studies on the trade-off between representation and competent leadership in Denmark, using detailed administrative data for the Danish population and the candidates in the local and national elections (Dahlgaard & Tue Pedersen 2019); on the impact of political revolutions on political selection and associated policy outcomes in Nepal, using census data, party nomination lists, electoral data, and information on conflict incidence (Bhusal et al. 2019); and on the economic underrepresentation of American politicians, using a deanonymized census data set from 1940 (Thompson et al. 2019).One of the earlier studies on political entry made use of the self-reported candidate data of Tanzania's parliamentary election of 1970 (Kjekshus 1975).
I now review the three broad questions asked in these and other related studies: How representative is the political class?How competent is it?Do societies face a trade-off between competence and representation?Section 6 summarizes the findings from these studies in a table.

How Representative Is the Political Class?
The data from the studies referenced above allow comparison of some important covariates, such as education, across candidates, politicians, and the entire population across the range of that covariate.Dal Bó et al. (2017) do this by matching the Swedish register data to politicians so that they can create figures like Figure 1, comparing the political class against the entire population.
As a caveat, while much of the evidence below is from national legislatures, there is some limited evidence that local politics seems to be more representative than national politics.Bueno & Dunning (2017, p. 347) note that in Brazil, "although elected federal deputies are about thirty percentage points more likely to be white than the population, the disparity falls to about ten percentage points among nonelected candidates for federal deputy."In Nepal, while politicians are positively selected relative to the population, the extent of positive selection is greater for municipal than ward (local) politicians (Bhusal et al. 2019).Similarly, Carnes (2013) and Carreri & Payson (2020) note that in the United States, even though policy makers at the state and local levels are considerably better off than the general population, they are less privileged than the politicians at the federal level.

On gender.
There is historical underrepresentation of women in elected office.Lawless & Fox (2005) observe the pervasiveness of underrepresentation of women in American politics and note that although the number of female candidates seeking elected office increased in the 1980s and early 1990s, this trend largely plateaued in the late 1990s and early 2000s [Shames et al. (2020) discuss how this trend is reversing since the 2016 presidential election]. 6Women's participation in politics remains a concern even in established democracies in Scandinavian countries.Dahlgaard & Tue Pedersen (2019) note that in Denmark, even though women comprise slightly more than half of the adult population, they make up only 28% of the elected officials at the municipal level and 37% of the elected politicians at the national level.
Political entry is also largely a male undertaking in the developing world.Bhusal et al. ( 2019) use census and local elections data from 2017 to document patterns of political selection in Nepal after the Maoist revolution of 2015.They find that women were more likely to receive party nominations when the law specifically mandated that women be nominated.However, they also report variation in the degree to which parties are inclusive, regardless of mandated representation for women.Iyer (2019) estimates that in the 2019 national elections in India, while 14% of elected members of Parliament (MPs) were women, only 9% of the candidates were women.In an older study, Kjekshus (1975) examines the socioeconomic profile of candidates in Tanzania's parliamentary election of 1970 to find that only 4.2% of the candidates in the national election were women.4.1.2.On race, ethnicity, and caste.Bueno & Dunning (2017) compare the race of state and federal deputies, senators, and governors elected in 2014 to the racial distribution of the Brazilian population and find substantial racial discrepancies: While browns and blacks comprised more than 50% of the Brazilian population in 2014, they accounted for less than 25% of elected politicians.
Similarly, Bhusal et al. (2019) show that Dalits are underrepresented relative to the population in Nepal in the absence of electoral quotas.Looking at all constituency seats (including Dalit reserved seats), it is striking that the caste distribution of candidates is largely representative of the population distribution, with overrepresentation of Dalits.After removing the Dalit reserved seats, the caste distribution of candidates is still representative of the underlying caste distribution of the population, except for underrepresentation of Dalits.These results are encouraging to the extent that they suggest that Nepal was able to elect candidates who were ethnically representative of the population in the first local election since democratization.
4.1.3.On wealth and class.In most democracies, there is a longstanding phenomenon of politicians being vastly better-off than the citizens they represent (Carnes & Lupu 2016a).Thompson et al. (2019) link the future members of the US Congress data to deanonymized 1940 census data and find that the future members of Congress who were below age 18 in 1940 came from affluent socioeconomic backgrounds.Their parents earned more than twice the population average and were six times more likely than the general population to hold college degrees.This is consistent with findings by Carnes (2013), who notes, for instance, that while lawyers and business owners comprise about 10% of the population, they make up at least 50% in both chambers of Congress.Similarly, legislators from working-class jobs make up less than 2% of Congress.In Latin America, too, Carnes & Lupu (2015) find that legislators are overwhelmingly white-collar professionals.Only about 5-20% of the legislators in these countries belong to the working class.Thus, these politicians are not representative of the population, as the working class in Latin America (manual laborers or service industry workers) comprises more than 60% of the labor force (Carnes & Lupu 2015, p. 6).
4.1.4.On age.An aspect of representation that is often understudied is the difference in the age of the represented and the representatives.In the United States, this issue has particularly come to prominence with the election to Congress of high-profile and relatively young representatives like Alexandria Ocasio-Cortez.Schlesinger (1966), upon examining the age of the members of the 85 th US House of Representatives in 1956, notes that the largest proportion of first-time elected members are in the age group between 35 and 40.However, the age level for entry rises progressively from the House to the office of governor and then to the Senate.
McClean ( 2019) examines the underrepresentation of young politicians in Japan and notes that only about 6% of the elected officials at the municipal level are under 40, while the average elected official is over 60 years old.This has substantial repercussions in terms of the age orientation of the social welfare programs.For instance, younger Japanese politicians tend to allocate higher resources for child welfare than for elderly welfare.
The mismatch in the age profile is particularly a concern in developing countries, where the majority of the population is very young.For instance, in Tanzania's parliamentary election of 1970, Kjekshus (1975) contrasts the age profile of MPs and candidates with the age profile of all Tanzanians above age 20 and finds that only 11.8% of the candidates, 6.5% of the nominees, and 5.1% of the elected MPs are below the age of 30.7 Unfortunately, not many authors decompose their analysis by this descriptive variable.
The analysis of age is also theoretically interesting because many models of political accountability proxy political competence with experience accrued over time in office.If the political class is overwhelmingly older than the electorate, to what extent is unrepresentativeness on this dimension compensated by the longer life experience of the political class?

How Competent Is the Political Class?
Getting talented and skilled individuals into office is important.Broadly, political scientists have conceptualized political competence in two ways.The first approach considers outcomes and effort, along the lines of Pitkin's (1967) "substantive representation."This work includes outcomes such as the number of bills sponsored in the legislature (Volden & Wiseman 2014); speaking in legislative session and writing bills (Grimmer et al. 2012, Parthasarathy et al. 2019); the number of Annu.Rev. Polit.Sci.2021.24:253-275.Downloaded from www.annualreviews.orgAccess provided by Stanford University -Main Campus -Robert Crown Law Library on 08/30/21.See copyright for approved use.
visits to home districts (Bussell 2019); acquiring funds for one's constituency (Anzia & Berry 2011); and monitoring the performance of the executive branch of government (Gulzar & Pasquale 2017).Moving beyond outcomes, recent work on Italian (Carreri 2021) and American (Carreri & Payson 2020) mayors also shows that administrative competence-the rules and regulations politicians set up to perform their duties-also affects how well mayors are likely to perform on the job.
A drawback of this first approach is that it only allows us to compare the competence of politicians who are already elected to office.For studies that seek to explain how competence relates to people's decision to put themselves forward for political office, having a measure of competence for both elected and unelected politicians is important.The second conceptualization of competence, therefore, focuses on the innate measures of the capacity of politicians to do their job well, arguing that this innate capacity should translate into better outcomes once the politicians are elected.This is one way of measuring what theories of political selection term politician "types" (Caselli & Morelli 2004).Below, I review some ways scholars have measured the innate competence of politicians.4.2.1.Leadership, personality traits, and cognitive abilities.One way of measuring innate competence relates to personality traits.For instance, the pioneering work of Lasswell (1986) established the behavioral foundations of political leadership by examining the relationship between personality types and leadership.More recently, Dynes et al. (2019) compare the Big Five personality traits of a sample of the US population and elected municipal officials to show that personality profile differences between men and women emerge at the election stage but not before.
Dal Bó et al. (2017) construct "leadership scores" of the general male population, nominated but not elected politicians, elected politicians, mayors, and MPs in Sweden.Their leadership score summarizes four personality traits-social maturity, psychological energy, intensity, and emotional stability-that are available through the Swedish register data.The study finds that mean leadership scores increase up the career ladder from nominated politicians to elected ones, and from mayors to MPs, suggesting that more competent people rise through the political career ladder.
Data in Sweden also allow Dal Bó et al. (2017) to measure cognitive abilities.These are quantified through a written test, which assesses ability in problem solving, induction capacity, and numerical, verbal, spatial, and technological comprehension.It can also be thought of as an IQ score.The authors find that "politicians score higher than the population, more strongly so when elected to office, and particularly so when selected for top municipal office and parliament" (Dal Bó et al. 2017, p. 1891).

Education.
Political skills are also often proxied by education.Dahlgaard & Tue Pedersen (2019) find that candidates nominated for local government are likely to be more educated than the population of Denmark.In addition, education levels increase at each step in the political career ladder.A comparable pattern is observed in the case of Sweden, where Dal Bó et al. (2017) show that politicians are underrepresented at the bottom levels of education and overrepresented at higher levels.Thompson et al. (2019) find that in the United States, "compared to siblings who did not become politicians, future members of Congress between the ages of 18 and 40 in 1940 were higher-earners and more educated, indicating that socioeconomic background alone does not explain the differences between politicians and nonpoliticians."Carreri & Payson (2020), examining a more recent period in the United States, show that even mayors exhibit far greater levels of education than the underlying populations they represent.
While extensively used as a measure of politician competence, Carnes & Lupu (2016b) contend that education might not be a good predictor of actual political performance.Analyzing data from the United States and Brazil, they find little evidence of a link between education and leadership quality and maintain that politicians with college degrees perform the same as or worse than politicians without.They measure political performance through a wide range of outcomes, such as economic growth, inequality, social unrest, interstate conflict, unemployment, inflation, re-election, legislative productivity, and corruption.Besley & Reynal-Querol (2011, p. 552) take a different view on this issue with data on 1,400 world leaders across the world from 1848 to 2004.They argue that "education is a particularly interesting aspect of political selection in view of the strong correlation found between educational attainment and earnings, which is consistent with education either enhancing skills or signaling ability.Education is also strongly correlated with civic engagement.Education is thus a compelling indicator of a leader's quality." Dal Bó et al. ( 2017) also raise the possibility of education simply reflecting elite membership.In such a situation, education becomes a poor marker of positive political selection.They argue that "if parental human-capital investments shape individual competence, a strictly meritocratic system might still favor elites.Meritocracy could then favor the competent within a family, but still be elitist across families" (Dal Bó et al. 2017, p. 1902).
One way researchers have sought to circumvent this problem is by the creative construction of counterfactuals that compare politicians with their parents, siblings, and other elite professionals.Dal Bó et al. (2017) compare the individual characteristics of Swedish politicians not only against the population, but also against siblings, parents, and members of other elite professions.Elected politicians show higher cognitive, leadership, and earnings scores than their siblings.This strongly indicates that ability and not family background is the key selection criterion.Next, the authors measure social background by parental incomes and note that the income distribution of fathers of politicians across income percentiles is largely uniform for each of the three types of politicians: those elected to a municipal council, mayors, and MPs.Further, the earnings of fathers of individuals in other elite occupations are compared to the earnings of fathers of politicians, and earnings for fathers of other elite professions, including doctors and chief executive officers, skew much higher than earnings of fathers of politicians.Dal Bó et al. (2017Bó et al. ( , p. 1900) ) conclude that, as measured by intergenerational earnings differences, social mobility into a political career seems to be high in absolute as well as relative terms.Similarly, Dahlgaard & Tue Pedersen (2019) compare politicians to their nonpolitician siblings in Denmark and find that politicians not only have a greater income score than their nonpolitician sibling but are also better educated.

Ability beyond education and income.
Another way in which we might make progress on the issue of education as a measure of competence is to focus on the innate ability of a prospective politician beyond their socioeconomic and educational status.This is not straightforward, however, as a large amount of data is needed to run regressions that enable the computation of this variable.Recent papers that use census data are able to make significant progress.Besley et al. (2017) compute an earnings score through the residual of a Mincer equation that has been deployed extensively in the education economics literature.They run models of the following form: where y i, m, t is the disposable income for person i in municipality m in year t.The independent variables include socioeconomic characteristics such as age, education, and employment sector.
In order to capture income differences across regions and between rural and urban areas, Besley et al. (2017) include α m, t , which represents municipality fixed effects.The authors define a measure of ability, labeled an earnings score, for each individual by computing the residual ε i, m, t for each available year and then averaging it across years.It is expected that more competent people will have a positive residual if the labor market abilities are not accounted for by the variables in the regression.Besley et al. (2017) show, for example, that earnings scores are correlated with cognitive and leadership ability as well as various measures of political and policy success.
Dal Bó et al. (2017) note that the earnings scores of mayors and MPs surpass those of the population in Sweden.Similarly, Dahlgaard & Tue Pedersen (2019) in Denmark use a similar measure of the earnings score and note that candidates nominated for local government are likely to have an above-average earnings score, while those who win a seat in local government and those nominated for the national parliament have higher scores, and those elected to the parliament score higher still.Using a similar measure, however, Bhusal et al. (2019) do not find evidence of positive selection of politicians with respect to ability in Nepal: Politicians are largely representative of the population and about 0.2 standard deviations higher on a similar score proxying for ability.

Is There a Trade-Off Between Representation and Competence?
The general debate around all the dimensions of representation discussed above concerns the idea that, as politics becomes more representative, politicians might be better able to convey the preferences of a broader set of voters in the policy-making process.However, there is a concern that an improvement in this dimension might come at the cost of competence-that is, politicians from underprivileged backgrounds might not be able to effectively formulate policy, leading to a deterioration in policy outcomes.
The evidence so far, though limited, suggests that the trade-off might not be as stark as may commonly be assumed.Dal Bó et al. (2017) find that, in Sweden, politicians on average are more competent than the population they represent, and this relationship holds true even after accounting for socioeconomic background.Consequently, the authors characterize Sweden as an "inclusive meritocracy."Dahlgaard & Tue Pedersen (2019) similarly find an inclusive meritocracy in Danish politics; they conclude that both candidates and election winners are competent and representative of a broad segment of Danish population and that there is no trade-off between "choosing from and electing representative and competent candidates" (p.24).
However, both these studies report results from Scandinavian democracies, and the lack of trade-off between competence and representation could be because of high economic mobility within these countries or relative ethnic homogeneity.For instance, Thompson et al. (2019, p. 3) study representation and selection in American politics and note that compared to the United States, In Sweden, there are similarly strong patterns of political selection for higher-earning and moreeducated individuals. . . .However, in Sweden, this process of selection does not lead to a political class that is economically unrepresentative of the population in terms of parental earnings.This key difference in our results versus the Swedish results may be because economic mobility is higher in Sweden than in America.In America, political selection for individuals with higher-earning ability and education is likely to lead to a political class with substantially unrepresentative family backgrounds, because there is a strong correlation between parental earnings and child earnings ability.This correlation is weaker in Sweden, allowing for the possibility of political selection for these skills without a cost in terms of representativeness.
The trade-off between representation and competence of the political class is therefore a particularly ripe topic for further research.One complication of this issue is that political competence is typically not measured in terms of what policies are actually carried out.Instead, as I discuss at the beginning of Section 4.2, scholars often use the attributes and ability of prospective politicians as a proxy for their competence in office.However, different margins on policy could be important if one were to adjudicate between the representation and competence of the political class.Scholars have tended to examine political competence along an extensive margin of policy efficacy, where politicians who can translate the same amount of effort and resources into more outcomes, such as sponsoring are defined as more competent.
What remains underexplored is whether and how the benefits of political effort may vary on the intensive margin.For instance, it could be the case that a particularly competent politician does not sponsor many bills, but the ones they do sponsor have a large positive impact on the welfare of their constituents.Gulzar et al. (2020) show in India that improving representation may not change outcomes on the extensive margin, but it may better align the distribution of resources to underlying population shares of groups.Szakonyi (2021) shows that politicians with a private sector background in Russia improve policy efficiency while leaving redistributive margins unaffected.These studies suggest that examining both the extensive and intensive margins of policy changes may be important when evaluating the trade-off between competence and representation.

SUMMARY AND FUTURE DIRECTIONS
This article surveys the literature on political entry with the aim of identifying areas that are particularly ripe for future work.I clarify conceptual issues related to measuring political entry, review research on the determinants of political entry, and identify and synthesize key findings from the recent literature on who runs for office and what that tells us about the representativeness and competence of the political class.
Who decides to become a politician influences the performance of democracies in important ways.Yet, the empirical examination of this decision remains relatively rare in political science.The current literature consists disproportionately of studies from advanced democracies.There is a particular need to bring in evidence from a broader set of cases to enrich our understanding of the key determinants of political entry in countries where democracies are not as well established, where parties are weaker, and where political divisions on ethnic, gender, and other dimensions are particularly salient.
While researchers often focus on work on political entry or leadership that has a formal political flavor, more work is needed on the antecedents of political entry.What activities and jobs increase the likelihood of future participation in politics as a politician?Tracing the pipeline of political entry before it formally occurs could provide insights into how the political class may be broadened, for instance.
One reason for examining the entry of key political personnel is the importance of the work they do once in office.The question of trade-offs between political competence and representativeness is open along several dimensions.Much work remains both on conceptualizing this trade-off and on assembling evidence for or against it.

DISCLOSURE STATEMENT
The author is not aware of any affiliations, memberships, funding, or financial holdings that might be perceived as affecting the of this review.