Jump to ContentJump to Main Navigation
Clients and ConstituentsPolitical Responsiveness in Patronage Democracies$

Jennifer Bussell

Print publication date: 2019

Print ISBN-13: 9780190945398

Published to Oxford Scholarship Online: May 2019

DOI: 10.1093/oso/9780190945398.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2019. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use. date: 13 December 2019



Representation, Distribution, and Constituency Service

(p.3) 1 Introduction
Clients and Constituents

Jennifer Bussell

Oxford University Press

Abstract and Keywords

The introduction to the book presents an overview of the puzzle constituency service presents to our understanding of distributive politics in patronage democracies. It then offers an overview of the book’s argument for why citizens demand assistance from their high-level representatives—individuals with substantially large constituencies such that they cannot know most of their voters personally—and why these politicians respond to such requests in a largely nonpartisan and noncontingent manner. The chapter places this constituency service conceptually within nonprogrammatic politics, alongside more well-studied forms of allocation: clientelism and partisan bias. It then offers an outline of the book’s contents and contributions, including a summary of the data sources used throughout the text.

Keywords:   constituency service, patronage democracy, India, clientelism, democratic representation, nonprogrammatic politics

A NEWLY MINTED legislator in the Indian state of Assam, on the first morning back in his constituency, dreamt that flies circled his head. When he woke, his dream had passed, and so had the flies—but a loud buzzing sound persisted. Curious, he arose and looked out his window at the courtyard of his home, where he saw around one hundred individuals chatting in the dawn light. The din was substantial as they awaited their chance to seek his personal intervention on behalf of their individual claims for public goods and services.1

On a different morning in Delhi, India, another legislator arrived to his office to find multiple visitors already waiting in the intense summer heat. Over the course of the next few hours, he received them in his office and entertained petitions for assistance with school admissions, scholarships, old age pensions, character certificates, problems with the local state-subsidized food shop, opening a government bank account, getting an operation in a public hospital, attaining a “below poverty line” card, and attending to a tree blocking a power line. He listened to each petitioner’s request and made some effort to help, be it by signing the required form, calling a responsible party, or providing additional information on how to acquire the service.2

(p.4) Visitors would encounter similar scenes at the homes and offices of politicians throughout India—and indeed, throughout many parts of the developing world. Legislators often have set times when they welcome petitioners, typically sitting in an open area to receive guests, and keeping an assistant on hand to quickly place a call, draft a letter, or complete a form. Citizens can thus come directly to their representative to make appeals. When their turn arrives, individuals deliver requests; politicians determine the appropriate next steps and often undertake them immediately, so as to work through the multitude of demands in as efficient and responsive a manner as possible.

As I show in this book, providing such assistance is in fact a predominant activity of many elected officials in India. The politicians we will meet in these pages not only pass legislation, receive party workers, and visit sites in their constituencies; to a significant extent, they interact with individual constituents on a regular basis and facilitate citizens’ access to fundamental benefits and services of the state. Perhaps surprisingly, this is true of high-level politicians—defined as those representatives with large constituencies who are unlikely to know most of their constituents personally—as well as local politicians who, as a result of considerably smaller constituencies, are more likely already to know those individuals requesting help. State and national legislators habitually dedicate large portions of a typical day to interacting with individual constituents, and frequently engage in “complaint handling,” such that “when people need help, they go to their legislator.”3 Such direct interactions with citizens—rather than with intermediaries, party workers, businessmen, or bureaucrats—comprise more than two-thirds of politicians’ meetings.4

Direct assistance to constituents is substantially prevalent in many other developing regions as well. As I document, face-to-face contact between citizens and high-level politicians is common across many African and Latin American countries.5 National legislators frequently receive tens or hundreds of requests from individual constituents on any given day.6 Data from these regions also suggest considerable responsiveness to citizens’ requests: politicians report that providing assistance to individual citizens is among their most (p.5) important activities, and some employ multiple individuals to respond to constituents’ appeals.7

For citizens, such personal assistance from high-level politicians can provide a crucial form of access to the state. In many developing countries, large portions of the population are unable to access public services, due both to inefficiencies and irregularities in bureaucratic processes and to discrimination in the assistance offered by politicians and bureaucrats at low levels of government. As a result, many individuals request help from high-level politicians with access to particularistic benefits—often, precisely the same types of services and benefits that other types of intermediaries, often at the local level, are frequently thought to facilitate. Such appeals thus constitute an important element of citizens’ strategies to acquire public goods and services. Politicians not only listen to their constituents’ entreaties but also help them overcome bureaucratic bottlenecks to obtain benefits. In the Indian case, substantial proportions of citizens report success in obtaining their desired benefit when assisted by high-level officials.8 In this manner, tens of millions of citizens receive assistance with acquiring critical basic services.9

The benefits that citizens receive through such petitions can constitute a substantial share of overall public welfare spending. As I describe in this book, high-level politicians in places such as India often influence the allocation of significant resources. Using new data sources, I estimate that the portion of benefits directed to individual citizens via high-level assistance constitutes a similar, and possibly larger, share of overall public welfare spending than those forms of distributive politics on which scholars have predominantly focused—including the clientelism and partisan bias that I will discuss shortly.10 Legislators in many other developing countries have similarly ample personal control over resources, due to influence over local officials and other factors.

Strikingly, as I also show in this book, high-level politicians often respond to constituents’ petitions in a remarkably equitable manner. By this I mean that it is often infeasible for politicians to make their assistance contingent (p.6) on individual citizens’ attributes or their past or future political behaviors. Drawing on evidence from qualitative shadowing of politicians in India, quantitative surveys of Indian politicians, and large-scale field experiments with both citizens and politicians, I demonstrate that in interactions with individual constituents, high-level politicians often do not premise their responsiveness on citizens’ political preferences, ethnicity, or other attributes. To a remarkable degree, even indicators of citizens’ partisan preferences do not affect representatives’ willingness to provide direct assistance. Instead, politicians appear more responsive, and substantially more willing to help citizens of all kinds, than many theories of distributive politics—defined as the process by which decisions are made about the allocation of government resources across a population—would predict.

In this way, I provide in this book an account of the constituency service—noncontingent, nonpartisan attention to the needs of citizens—offered by high-level politicians in many democratic developing countries.11 By intervening in distributive processes, politicians often engage in “mediation from above,” exerting pressure on bureaucratic underlings to facilitate citizens’ access to services and benefits. For citizens, these efforts offer a crucial source of assistance with the often-difficult process of accessing fundamental benefits from state welfare schemes. This assistance is thus an important mode of distributing state resources. Constituency service is also a key element of representation, acting as an important form of “service responsiveness.”12 Those individuals who have difficulty accessing benefits from the state can appeal to their elected representatives for assistance; by responding to the needs of these constituents, high-level politicians “represent” them to the state. In doing so, elected representatives respond to the interests of those citizens who might otherwise be excluded from individual-level public services and bring them within the distributive fold.

These direct exchanges between politicians and their constituents may appear similar to dynamics observed in western democracies. As Fenno notes in his account of state legislators’ behavior in the United States, “Constituent service [is] universally recognized as an important part of the job in its own right.”13 A former Canadian member of Parliament offers his own account: “As citizens showed up in my constituency office with their tales of (p.7) passports delayed, visas withheld, tax files mislaid, my staff and I would pick up the phone and try to help.”14 Similarly, in 1960s Ireland, there was an expectation among citizens that a legislator would, if asked, “give advice on any problem, help prepare applications to public authorities, and make inquiries, submissions, representations or appeals not only for those who vote for him but for anyone in the constituency and particularly his own district.”15

Yet, this type of interaction is also perplexing in a “patronage democracy” such as India—defined as a country where the state is a primary source of resources, but official distribution is often extensive, inefficient, and discretionary.16 A wealth of research on distributive politics in such contexts—most commonly in economically developing countries—suggests that politicians target state goods and services tactically for electoral advantage.17 For example, they may direct group-oriented goods to particular geographic areas, responding to the interests—and electoral support—of specific factions within their constituencies. Politicians in patronage democracies may have substantially greater capacity to influence the distribution of resources than in settings where the rules of distribution are relatively binding; and they may aim to use this discretion to mobilize turnout among core supporters or to persuade swing voters.18 Direct resource distribution by high-level politicians is thus both contingent and partisan.

Alternatively, according to many existing accounts, politicians may allocate resources indirectly, using clientelist strategies. Thus, political parties delegate to local-level intermediaries or “brokers” the responsibility of targeting state services to maximize electoral payoffs. Due to their proximity to citizens, such operatives can facilitate the mobilization of voters and the monitoring of political behavior. Beneficiaries are in turn expected to vote with their local intermediary’s party, either because benefits are granted in an explicit quid pro quo for electoral support or because citizens seek to boost (p.8) their intermediary’s capacity to provide future benefits.19 If brokers can monopolize distribution locally, higher-level politicians may have little choice but to contract with such individuals to ensure a base of electoral support.20

The presence of substantial constituency service in patronage democracies therefore raises a puzzle—and must be explained in light of these alternative tactical uses of politicians’ time and effort. In existing accounts, high-level politicians have little incentive to assist individual citizens directly. After all, national or provincial legislators have large constituencies—an average Indian state assembly constituency has around three hundred thousand people—and they cannot readily verify particular petitioners’ histories of political support. Unlike local politicians, such as a village council president who typically has five thousand constituents living in a few adjacent villages, high-level politicians often cannot have a strong sense of individual citizens’ political preferences or choices. Nor, especially, may they effectively condition the provision of benefits on petitioners’ future behaviors. The direct provision of assistance to individuals likewise does not readily appear to provide the opportunity to target sets of core supporters or persuadable swing voters with group-oriented goods. Providing such assistance may appear to be a highly inefficient mode of targeting, one that does not easily scale from individual voters to aggregate support. The great range of other distributive strategies available in patronage democracies might seem to foreclose the electoral usefulness of such assistance. Perhaps for these reasons, scholars of developing countries have not offered systematic accounts of direct exchanges between high-level politicians and their constituents—which might falsely suggest that their prevalence is low.

The pervasiveness of constituency service in patronage democracies thus raises several key questions. Why do politicians with large constituencies, and many demands on their time, pay substantial attention to the requests of individual citizens? Importantly, how do these efforts relate to more frequently studied forms of distributive politics—such as the targeted allocation of group-oriented goods or clientelism? What is the nature of the democratic representation to which constituency service gives rise in these contexts? In (p.9) western democracies, the provision of constituency service is “recognized as powerful reelection medicine. ‘What political scientists have to understand,’ said one member [of the U.S. Congress], ‘is that an incumbent congressman can get reelected by the services he is in a position to do for the people.’ ”21 Thus, high-level politicians may benefit from offering noncontingent and nonpartisan assistance to their constituents. Yet, in contexts rife with clientelism and patronage, the electoral advantages of constituency service may appear considerably less, and must be subjected to extensive theoretical and empirical investigation.

I argue in this book that high-level politicians in fact have multiple incentives to offer such mediation in a noncontingent manner—and these reasons go beyond those offered to explain constituency service in developed countries. In patronage democracies, the state is a major provider of welfare and other benefits, yet the delivery of such services is often inefficient and highly contingent, making it difficult for even the most needy and eligible individuals to access these resources. If citizens in democracies with less discretionary service delivery still sometimes need the intervention of their representative to speed up a delayed visa or locate a misplaced tax file, this requirement is even greater in patronage democracies, where targeting, exclusion, and inefficiency in the delivery of public services often replaces the rule-bound implementation of public policies.

The characteristics of patronage democracies thus generate strong demand for informal routes to obtaining services, and for interventions by influential intermediaries—even more so than in settings with less discretion in public service delivery. Existing work indeed highlights the prevalence of alternative paths to accessing benefits. Citizens may resort to petty corruption, including the payment of “speed” money.22 Alternately and critically, intermediation via politicians, “fixers,” or organizations at the local level is also thought to be prevalent.23 Citizens in these contexts have difficulty accessing benefits from the state directly and are therefore expected to rely on more powerful or knowledgeable actors to acquire basic benefits.

Yet, there are limitations to these strategies of access from the perspective of citizens. Those individuals who require public benefits the most may well be the least able to offer a compelling bribe to a bureaucrat to acquire (p.10) their desired service. If, alternatively, a citizen turns to a local broker, he may face the expectation of an electoral quid pro quo. For those who are unaligned with local officials, their chances of receiving assistance may be substantially reduced. While these individuals might turn to other nonelected local intermediaries or acquaintances, such alternative facilitators of service delivery are often likely to have relatively less power to offer assistance, again limiting the individuals’ chances of success in acquiring benefits.

For citizens, this suggests the attractiveness of appealing to powerful politicians who are able to smooth the processes of accessing the state. Those who are denied services locally are the likeliest to make such appeals to high-level politicians for mediation from above. Moreover, especially in contexts in which discretionary allocation takes on a partisan hue, local operatives may tend to target assistance to their own co-partisans, while denying or delaying services to citizens with whom they are not aligned. The discretionary, partisan process by which local appeals for service delivery are often denied—what I term local “blocking”—results in a robust demand for assistance from citizens who are not part of local partisan patronage networks.

Local blocking can then, in turn, heighten the incentives of high-level politicians to supply assistance to their constituents. Precisely because of the partisan dynamics that often generate such petitions, constituency service can offer a particularly effective model for appealing to those voters who are not integrated into local clientelist or partisan benefit-delivery networks—and are thus not otherwise able to acquire services. Even if high-level politicians are unable to premise responsiveness on the partisanship or behaviors of individual voters, they may readily make inferences about the types of voters who appeal to them. Constituency service gives them a way to reach potential supporters who are locally blocked by high-level politicians’ non-co-partisans, or even those who are ill-served by their local co-partisan brokers. The uneven reach of partisan bias in these contexts thus engenders the electoral utility of this complementary mode of distribution—making constituency service possibly even more useful to high-level politicians than in the nonpatronage democracies for which the concept was initially conceived.

These dynamics generate specific motivations for responsiveness in another way as well. Provision of aid to citizens who are often substantially in need of benefits from the state but have difficulty accessing them may offer higher levels of appreciation among those who receive help than in other contexts. Moreover, by meeting a substantial unmet demand for assistance, high-level politicians can build a positive reputation among exactly those citizens who are particularly responsive to receiving assistance—including (p.11) potentially persuadable swing voters and those who, having been blocked locally on partisan grounds, may be especially receptive to an inherently nonpartisan act of assistance from a high-level politician. This can increase the chances that an individual will remain supportive of a particular politician, or shift to support that politician, on the basis of receiving assistance. As in many developed countries, constituency service allows politicians to improve their individual reputations and foster a personal vote, or a base of voters who support a politician on the basis of that candidate’s individual character and acts, rather than an association with a particular political party.24 Yet, the potential to build this personal vote through constituency service may be strongest precisely when partisan linkages are also strong and local service denials take place along partisan lines. Helping individual citizens may therefore be even more relevant to high-level politicians’ reputations in such contexts than in less discretionary domains.

The value of constituency service to elected politicians must therefore be viewed alongside other potential forms of distributive politics—and, in particular, the limitations of those other strategies. In the allocation of government funds to support localized projects in a single electoral district, such as roads, community buildings, or water treatment plants—the substance of what we typically understand as pork-barrel politics—politicians have the advantage of being able to reward key groups or areas within their constituency that have offered significant past support. At the same time, few areas are populated solely by core or swing supporters, and it is rare that all supporters live in areas that have voted predominantly for a given politician or party. As a result, politicians cannot ensure that they are targeting only their supporters or all of their supporters when distributing group-oriented goods. Similarly, the uneven nature of local clientelist intermediation—and the tendency of intermediaries to channel benefits to their partisan supporters—implies a set of under-attended individuals in most constituencies. Constituency service can provide a potent means for politicians to target these excluded voters, who are likely to be swing voters or, depending on local electoral dynamics, co-partisans of more senior politicians. Even if they do not and cannot condition assistance on individual petitioners’ partisanship, politicians can realize that in the aggregate, constituency service is an effective way of reaching potentially persuadable voters—and it may provide an alternative or (p.12) complementary strategy that compensates for the weaknesses of other modes of distributive politics.

Thus, constituency service offers electoral benefits to politicians that accrue despite officials’ limited ability to target assistance directly on the basis of individual citizens’ political preferences or behaviors. This is also not to say that the provision of assistance is done only for strategic reasons. Politicians may well offer citizens aid on the basis of altruistic intentions and a sense of representative responsibility. Indeed, the evidence offered in this book suggests that many politicians are less venal and more other-regarding than political scientists may typically perceive them to be. At the same time, I focus here on the specifically electoral motivations, for which there is substantial evidence, and which help to explain several otherwise puzzling empirical patterns. I show that one need not depart from electoral motivations on the part of politicians to understand constituency service in patronage democracies. Yet, this explanatory focus complements my broader finding that politicians often behave in a more responsive and representative fashion than prevailing theories have supposed.

These dynamics of distributive politics also exhibit a much more inherently multilevel nature—with interactions between elected officials and bureaucrats across multiple offices—than has previously been identified. Some research suggests that local-level actors are delegated the authority to make distributive decisions that affect individuals, while high-level politicians are focused on targeting group-oriented goods to particular geographic areas of their constituencies. In either case, the prevalent assumption is that allocations are made in a highly partisan manner.25 We will instead observe in this book a more integrated set of dynamics in which the activities of actors at the local level have direct effects on the demands for assistance from higher-level elected officials—and thereby shape the incentives of these officials to respond to such requests, largely in a noncontingent manner.

Consequently, while the dynamics of patronage democracies might initially seem to make constituency service unlikely, I argue that a particular kind of constituency service emerges in part because of those dynamics. To be sure, politicians who respond to their constituents in a noncontingent, nonpartisan way are also likely to target resources via more partisan strategies. For example, they may prefer to provide individual assistance to known party workers. And as I show, the same politicians who engage in constituency (p.13) service also undertake partisan distribution of group-oriented state resources. Politicians target group-based benefits tactically, especially toward their core supporters; and they take advantage of local networks of partisan brokers when they can. However, they also provide direct assistance to individual constituents, particularly for certain kinds of benefits and especially when they cannot rely on effective delegation to local intermediaries. As I demonstrate in this book, constituency service in fact constitutes a critical element of distributive politics and a key facet of political representation.

This argument, to which I return in more detail in Chapter 4, therefore highlights a considerably more complex politics of distribution than has previously been ascribed to patronage democracies. I offer an amendment to partisanship-driven accounts of democratic representation, suggesting that while contingent exchanges do exist in these contexts, they occur in tandem and in interaction with forms of constituency service that have gone largely unnoticed and untheorized in analyses of distributive politics. High-level politicians, rather than always engaging in partisan distribution of benefits, also dedicate a substantial part of their time and resources to assisting individual citizens in ways that are not dependent on the specific electoral behaviors of those receiving assistance. Because the demand for this assistance arises both from the inefficiencies of the state in which high-level legislators are embedded and from the prevalence of discretionary distribution, we must understand legislators’ incentives for responsiveness as therefore deeply intertwined with the dynamics of contingent allocation. Alongside pork-barrel politics and clientelism, constituency service is an important element of politicians’ overall distributional and representational repertoires, and our accounts of distributive politics must be substantially enriched by consideration of the ways in which politicians provide noncontingent direct assistance to their constituents.

More generally, this book suggests the potential for an underappreciated form of representative democratic politics in patronage democracies. To understand the normative as well as positive consequences of constituency service in patronage democracies, it is critical to assess the kind of responsiveness that constituency service implies. The nature of democratic representation constitutes a core focus in the study of political science. In developed countries, debates often center on the relative allocation of politicians’ time and resources to nonpartisan constituency service versus demands for group-based benefits and attention to specific policy arenas.26 While there is considerable (p.14) debate over the drivers of variation in allocation, the fundamental underlying principle that responsiveness is based on some version of democratic accountability goes frequently unchallenged. In developing countries, by contrast, the prevailing view of allocation is one based on partisan ties and, often, clientelism, resulting in a form of “perverse accountability” that threatens the foundations of democratic representation.27 I rather suggest that politicians in these latter contexts engage in practices that generate not only perverse but also genuinely democratic—if constrained—accountability of representatives to their constituents. The implications for our understanding of democratic practice merit substantial attention that they have not yet received.

The concept of constituency service

The analysis in this book requires a novel set of conceptual, theoretical, and empirical approaches. Conceptually, in treatments of distributive politics, constituency service has effectively been assigned to a “residual” category of politician behavior; and its status as a form of representation has been unclear.28 Yet, successful theoretical and empirical study of constituency service requires a prior conceptual approach that situates constituency service in relation to other modes of representation and distribution. I offer here a new conceptualization of distribution types that incorporates the manner by which politicians offer noncontingent assistance to voters. I focus centrally both on understanding the nature of the political responsiveness that constituency service implies, and on its relationships to other forms of distributive politics. Thus, I ask: what kind of representation is constituency service in a patronage democracy? How does the activity of providing noncontingent assistance to citizens—through which individuals gain access to basic benefits and services of the state—compare conceptually to other modes of distributing valued resources?

In the classic conceptual account, Pitkin defines representation as “acting in the interest of the represented, in a manner responsive to them. The representative must act independently; his action must involve discretion and judgment; he must be the one who acts. The represented must also be (conceived as) capable of independent action and judgment, not merely being (p.15) taken care of.”29 For a political representative, then, the act of representing may potentially encompass a wide range of activities, insofar as those activities involve “acting in the interest of the represented.”30 To be sure, one might constrain a political representative’s actions to those defined by a constitution or other laws setting out the roles and responsibilities of such actors. Within legal limits, however, there seem to be a host of actions that might feasibly fall under the category of “representation.”

As others have noted, Pitkin did not define what makes up the “responsiveness” she emphasized, leaving empirical researchers to determine an appropriate operationalization of the term.31 Most analyses of representative behavior on the part of elected legislators, in both developed and developing countries, focus on the role of these individuals in making or influencing policy—so-called policy responsiveness. Landmark work by Verba and Nie, as well as Miller and Stokes, focuses on public policies as the primary forum for representatives to engage in responsiveness, and subsequent empirical work has largely followed this lead.32 Yet, the critique offered by Eulau and Karps is a striking one: “In focusing exclusively on . . . policy attitudes or preferences, [authors] ignored other possible targets in the relationship between representatives and represented which may also give content to the notion of responsiveness. By emphasizing only one component of responsiveness as a substantive concept, they reduced a complex phenomenon like representation to one of its components and substituted the component for the whole.”33

This concern with the potential elements of responsiveness is important both conceptually and empirically. Conceptually, the idea of responsiveness for elected officials requires a delineation of all the most likely activities that may reasonably merit inclusion as representative acts. Empirically, “if responsiveness is limited to one component, it cannot capture the complexities of the real world of politics.”34 At least in part, this is because existing research tells us how difficult it is for individual voters to know and communicate their (p.16) preferences about every potential policy.35 As a result, politicians will never have sufficient information about these preferences to respond with policies in perfect accordance with their constituents’ wishes. Relatedly, Achen and Bartels have recently criticized what they call the “folk theory” of democratic accountability: the idea that voters formulate preferences and communicate them to politicians, who in turn act on those aggregated preferences to enact public policy.36

Purely policy-oriented perspectives additionally make at least two implicit assumptions about the nature of policymaking and policy implementation. First, a policy orientation to representation assumes that all representatives have equal ability to influence the character of policy and, thus, can be equally responsive to their constituents’ interests. Yet, a significant majority of legislators typically have only limited, if any, influence over the design of policy, either because they are in the opposition or because they are not on the committee or cabinet that is primarily formulating policy. For this majority, policy responsiveness amounts to posing questions about potential policies during public debates and voting on those policies; the scope to vote in a way that is responsive to constituents may be limited even further by party discipline rules.

Second, policy-focused views of responsiveness implicitly assume that once a policy is passed, there is no further opportunity to affect the outcomes of that policy. While this may seem reasonable theoretically, it does not reflect the realities of policy implementation in most, if not all, representative democracies. Public bureaucracies are tasked with the implementation of policies, and there may be many reasons why politicians might want, or have the opportunity, to influence the character of that implementation in the interest of their constituents. Particularly when a bureaucracy often does not implement policies according to the letter of the law, or when the law has implicitly (or explicitly) allowed for discretion in the implementation of policy, politicians may have opportunities to address inefficiencies—or take advantage of discretion—in policy implementation. In India, for example, there are multiple policy channels for distributing benefits, including subsidies on products such as electricity, provision of low-cost consumables, and creation of employment for millions of individuals through a national work program. Yet it is often in the execution of these policies, rather than in their design, (p.17) that decisions over access to benefits are made.37 As Scott noted, “Between the passage of legislation and its actual implementation lies an entirely different political arena that, in spite of its informality and particularism, has a great effect on the execution of policy.”38

If elected representatives typically have little room to influence the character of policies, but instead have substantial opportunities to influence how they are enacted, then we might expect a large portion of responsiveness to involve actions in the realm of shaping implementation. For instance, other recognized forms of responsiveness look very much like strategies for influencing policy implementation. Thus, “service responsiveness” refers to “the efforts of the representative to secure particularized benefits for individuals and groups in his constituency.”39 These efforts most typically concern casework and the solving of problems for individuals in dealing with the bureaucracy. In contrast, “allocation responsiveness” entails interventions in the “allocation of public projects” in ways that proactively result in benefits that may advantage a politician’s constituency.40 Allocation responsiveness is then group-oriented distribution that is closely related to the localized, district-specific projects we commonly refer to as pork-barrel politics.

The form of responsiveness at the heart of the discussion in this book is constituency service, which I define, in line with Fenno, as service responsiveness that does not involve attention to the partisanship or history of political support of the individual or group making a request.41 An offer of constituency service from a politician is contingent on a request being made, a characteristic to which I return in the theoretical discussion of Chapter 4, but it is not contingent on the political behavior of the individual making the request. It is also crucial to underscore that politicians may choose to engage in constituency service in order to advance their broader electoral or partisan goals—as indeed they do in the theory that I develop and test in this book. Yet in facilitating the delivery of services to particular citizens, politicians provide constituency service whenever they do not condition their assistance on the partisanship, history of political support, or other attributes of petitioners. While this conceptualization differs from a broader understanding of the term (p.18) occasionally used in comparative politics, my usage retains the noncontingent basis upon which constituency service is traditionally understood to rest.42

The noncontingency of constituency service is important for both normative and positive reasons. Normatively, constituency service involves direct exchanges between politicians and citizens but need not involve the “perverse accountability” of clientelism, in which politicians hold citizens to account for their behavior rather than the other way around. In contrast, constituency service acts as a conduit for any individual citizen to make demands on their elected representatives. Moreover, if, in assisting constituents, representatives are “acting in the interest of the represented, in a manner responsive to them,” then constituency service is also a key aspect of democratic representation.43

Yet, the noncontingent nature of constituency service is also important for positive reasons—because it raises the empirical puzzles I introduced above. Through constituency service, politicians allocate access to welfare benefits and services of the state. Thus, constituency service is also a form of “distributive politics:” it reflects choices and struggles over the allocation of state resources to groups and individuals within society. Previous treatments of distributive politics, however, often suggest that politicians act strategically and tactically to advance their positions through the targeting of material resources along electoral and often partisan lines. Specifically, existing work links receipt of public benefits and services fundamentally to support of specific political parties, or to the potential for supporting a particular party or politician. This can be true whether the distribution in question arises from policies—such as market regulations or the design of welfare programs—or from their de-facto implementation, e.g., the discretionary allocation of projects to particular geographic areas or the transfer of individual benefits to political supporters. The fact that politicians assist citizens in large numbers—but do not, in so doing, appear to target those they help on electoral grounds—therefore flies in the face of a large literature on distribution, especially in patronage democracies.

This raises a further question: how should we conceive of different types of distributive politics—and where does constituency service fit in such a conceptualization? One useful view, offered by Stokes, Dunning, Nazareno, and Brusco, distinguishes primarily between programmatic (p.19) and nonprogrammatic modes of distribution.44 In programmatic contexts, there are 1) public rules that 2) actually shape the character of allocation. Nonprogrammatic politics, in contrast, predominates where only one or neither of these conditions holds. Within nonprogrammatic politics, benefit receipt is then either contingent on direct political support (the presence of a quid pro quo) or not. Contingent provision of benefits is termed “clientelism,” whereas noncontingent distribution is conceived of as “partisan.” In other words, if distribution is nonprogrammatic, it must in some way, by Stokes et al.’s definition, relate to the targeting of benefits to supporters of a political party: either distribution requires explicit demonstration of support for that party or it is influenced by partisan bias in a more general manner.

Yet, it is possible for distribution to be both nonprogrammatic and nonpartisan. This is the case when state resources are distributed in a nontransparent manner, but the choices over that allocation are not explicitly based on partisan ties or electoral behavior. Take, for example, an individual who requests assistance from a politician for accessing a public service. Suppose that politician, based only on the request (or perceived need), asks that the relevant bureaucrat give additional attention to that person’s application, and the bureaucrat then prioritizes the application in a way that he would not have otherwise done. This would be nonprogrammatic—because binding, public rules are not guiding all aspects of distribution—but also nonpartisan, or more generally noncontingent, distribution.

Building on this idea, consider a different case in which a politician is faced with two individuals who need similar assistance, and the representative has information that one of these individuals is her co-partisan (or, more generally, a supporter of the candidate). If she is influenced by this information and, as a result, is more willing to provide assistance to that political supporter than to a person for whom she has no information on partisanship or patterns of previous political support, this shifts into the realm of partisan bias.45 She may still provide assistance to the second person, but she does so in a less speedy or less aggressive manner. This politician, then, is not engaging in constituency service with regard to her supporter, but rather is influenced in her actions by partisan bias. Thus, it is very feasible to imagine a politician who is regularly engaging in both partisan and nonpartisan intermediation—the (p.20) latter being constituency service—even where the two actions look, on their face, quite similar.

In order to account for such forms of noncontingent distributive politics as highlighted in the first example, I shift the focus in the conceptual framework for nonprogrammatic politics offered by Stokes et al. from beginning with the question, “Is receipt of benefits contingent on an individual’s political support?” to instead asking, “Are benefits to citizens targeted to political supporters?” (By this I mean that patterns of potential political support influence the allocation; the conceptualization does not make a commitment to a particular form of targeting, for example, the privileging of core or swing voters). A “yes” response to this question indicates forms of partisan distribution and then replicates a portion of the Stokes et al. framework, beginning with the modified question: is political support a quid pro quo for receipt of benefits? This branch of the conceptual framework thus distinguishes between cases in which there is enforcement of a quid pro quo—clientelism—and those where there is not—partisan bias, including forms of pork-barrel politics and benefits targeted to individuals who are thought to be supporters.46 The important distinction between these forms of distribution is that with partisan bias, “recipients who defect and vote for a different party suffer no individual punishment.”47 A “no” response to my initial question on political targeting instead implies nonpartisan distribution and leads to the broad category of “constituency service,” as shown in Figure 1.1. This is in line with (p.21) Fenno’s original characterization that “[c]onstituency service is totally nonpartisan and nonideological.”48

IntroductionRepresentation, Distribution, and Constituency Service

Figure 1.1 Conceptualizing Constituency Service in Nonprogrammatic Politics

While constituency service has at times been thought to span the line between programmatic and nonprogrammatic politics, I suggest that at any point when an official intervenes with some nonpolicy-based discretion on behalf of a citizen, this can be considered a shift into nonprogrammatic distribution. This can occur in contexts traditionally associated with constituency service—such as when a politician in the United States intercedes with a bureaucrat on behalf of a constituent who needs assistance acquiring veterans’ or Social Security benefits—as well as in patronage democracies.49

Note also that that the petitioner in question could be an individual or a group. In practice, groups may be more likely to request group-oriented goods—i.e., goods for which more than a single individual or family unit is likely to benefit. I consider in the theoretical discussion of Chapter 4 why such requests may be more likely to offer politicians opportunities for partisan bias in distribution. Conceptually, however, a politician may feasibly respond to both individuals and groups in either a contingent or noncontingent manner.

Indeed, this book documents the prevalence of assistance to individual citizens by high-level politicians. Yet, whether that assistance takes the form of constituency service or instead constitutes partisan distribution is a variable. The theory and empirical tests offered later in the book thus help to explain the conditions under which politicians offer partisan bias, rely on clientelism, or instead provide constituency service.

The conceptual difference between partisan distribution and constituency service is an important one because it allows us to incorporate into our understanding of distributive politics those forms of behavior that may look partisan—as they operate within a nonprogrammatic context—but that do not exhibit key characteristics associated with typical conceptualizations of partisan allocation. When a politician helps an individual citizen access the state, we might assume, based on our theoretical priors, that she does so only because she knows the political affiliation of this individual or can enforce future electoral support. Yet, that assumption may be fundamentally inaccurate in contexts where it is simply infeasible for politicians to infer an individual’s partisanship accurately or to make the receipt of assistance conditional on future support.

(p.22) These distinctions are particularly critical to make in the context of patronage democracies, where the state is an important provider of resources but the allocation of those resources is often discretionary. Here, it is also necessary to come to conceptual grips with the notion of patronage relative to constituency service. While recent empirical work on the subject of patronage takes the approach of not defining the term, a view on how this concept is used, particularly in developing countries, is necessary here.50 Within political science, patronage is often equated with clientelism, highlighting a power differential between the patron and the client, and the exchange of material goods for electoral promises.51 In contrast, a current anthropological view suggests a broader understanding, one that effectively sees patronage on moral terms as reflecting an expected mode of interaction between those individuals with greater access to resources—specifically, politicians—and those with less.52 From this perspective, patronage relations are not simply about the buying of votes, but also incorporate voters’ views of whether politicians are effectively utilizing the power of their position to provide benefits to their constituents.53

The conceptual approach that I take in the theory and analyses of this book has to do with this latter point. If citizens are free to change their vote, to hold politicians accountable for their performance in office, and, in so doing, these same voters do not directly threaten their ability to access public benefits, then what we are observing is not a form of clientelist politics. In this view, those politicians who make the receipt of goods conditional on electoral behavior are acting as clientelist patrons, while those who do not are serving as representatives. If these two activities coexist, as I suggest in my theoretical argument and as is emphasized in the view of patronage held by Piliavsky, Price, and others, then this reflects a more diverse set of political dynamics than have been considered in most political science work on distributive politics.54 For practical purposes, and in line with the broader literature on representative democracy and constituency service, I use the terms clientelism and constituency service as defined by Figure 1.1, while keeping in mind that (p.23) both may refer to specific kinds of activities that fall within a broad rubric of “patronage” as used in recent anthropological work.

The term “patronage” is also often used in a narrower sense to refer specifically to the discretionary allocation of state employment, while usage of the word as an adjective modifying democracy has typically been more general.55 Thus, in coining the term “patronage democracy,” Chandra emphasized public employment as a key measure of a democracy’s patronage character, but she also highlighted more generally the role of the state in providing social services and other benefits relevant to the general well-being of the public.56 This implies that not all developing countries are patronage democracies—the poorest states have only limited resources to allocate to the public—but a large portion of developing countries are likely to fit within this category, an expectation I test explicitly in Chapter 11.57 Thus, in the many national contexts where a significant portion of resources are allocated by the state in a discretionary—and so nonprogrammatic—manner, differentiating between contingent and noncontingent allocation should be core to our analyses of distributive politics.

With the typology in Figure 1.1—and with the integration of constituency service as a key nonprogrammatic, yet nonpartisan, form of distribution—we can link ideas about “representation” and “responsiveness” to conceptions of “distributive politics.” Notions of policy responsiveness often correspond, in general, to programmatic politics. Forms of partisan bias—including pork-barrel politics—might reasonably fit into an understanding of allocation responsiveness. Yet, should politicians who engage in nonprogrammatic but noncontingent distribution be properly conceived of as responsive to their constituents? And how does this form of distribution align with the different concepts of responsiveness previously discussed?

I posit that constituency service is an important form of service responsiveness that, at least theoretically, can serve as a key tool for representation. High-level politicians who solve problems for their constituents clearly act in the interest of the represented, and in a manner responsive to them. Perhaps most critically, following Pitkin’s previously discussed formulation, citizens who petition their legislators exert independent action: they are not “merely being taken care of,” nor is their autonomy subverted by politicians (p.24) who demand political support in exchange for benefits.58 In this key respect, constituency service contrasts with clientelism—in which an individual is held accountable to the person or party providing them a benefit. As Stokes explores, quid pro quos generate a perversion of accountability in which voters become accountable (or responsive) to their elected officials, rather than the reverse.59 Thus, such forms of contingent allocation—while a central aspect of distributive politics—cannot be understood as compelling forms of representation or adequate political responsiveness.60

In sum, distributive politics intersects in important ways with the concept of representation and notions of responsiveness. Both contingent and noncontingent tools of distribution—targeted policies and constituency service—can be seen as forms of political representation, while clientelism instead subverts the direction of accountability between representatives and the represented. Most importantly for current purposes, constituency service is both a form of representation and a potentially important strategic tool for distribution. The prevalence of direct assistance from high-level politicians—especially in relation to policy allocation and local-level intermediation—thus suggests a rethinking of representation in patronage democracies and of the diverse forms of distribution that can emerge in democratic contexts. I return in the concluding chapter to the normative implications of the form of responsiveness I document here.

Outline and contributions of the book

These conceptual preliminaries provide the basis for an empirical and theoretical study of constituency service in patronage democracies. I turn to this topic in the rest of this book.

In Part I, I present the puzzle of constituency service in India, a paradigmatic patronage democracy. Despite existing expectations that high-level politicians should allocate little time to the affairs of individual citizens, I show that interactions with constituents merit a large portion of these politicians’ attention. These exchanges constitute a major channel for the distribution (p.25) of welfare benefits. In addition, and surprisingly, this helpfulness is generally offered without regard to a person’s political behavior or preferences. Scholars have devoted relatively little attention to these important constituent-facing activities of high-level politicians. The novel descriptive and causal evidence I offer in Part I thus provides a central contribution of the book.

In Chapter 2, then, I draw on accounts from sustained, in-depth shadowing of politicians, as well as large-scale politician surveys, to characterize the nature of politicians’ engagement with their constituents. I highlight the importance these politicians place on making time for citizen interactions and responding to requests—to the extent that high-level politicians spend, on average, a quarter of their time interacting with individual citizens. Critically, the primary focus of these contacts is requests for assistance with the same types of goods and services that are typically also requested of local politicians such as village council presidents, providing preliminary evidence that demands may at least partially originate from individuals’ failure to acquire these benefits at the local level—a topic to which I turn later in the book. I also show in Chapter 2 that the individual benefits directly provided to citizens by high-level politicians are substantial. Thus, the phenomenon I study here is a central element of distributive politics in an emblematic patronage democracy.

While these analyses suggest that high-level politicians spend significant time assisting individual voters, the evidence in Chapter 2 cannot readily show whether and when this assistance is nonpartisan and noncontingent in nature. In Chapter 3, I draw on a field experimental audit of politicians with a near census of Indian state and national legislators to show that, on the whole, politicians do not take indicators of partisanship into account when responding to individual-level requests. Specifically, this national field experiment—which I believe is the largest of its kind, and the first conducted with state and national legislators—shows that for India’s high-level politicians, information on electoral preferences does not affect the willingness of representatives to respond to an individual’s request for assistance. In addition, indications of shared ethnicity, e.g. caste, which may be closely tied to political preferences, do not result in preferential treatment. Overall, these findings offer strong evidence that the aid high-level politicians offer to individuals requiring assistance navigating the state is often noncontingent in nature, taking the form of constituency service.

Taken together, Chapters 2 and 3 therefore demonstrate the importance of constituency service in a patronage democracy. Noncontingent assistance is critical both for the high-level politicians for whom providing individual aid (p.26) is a major component of their representative activities, and for the millions of citizens who thereby gain critical welfare services and benefits. The evidence in Part I makes an important contribution, by showing the surprising salience of this representative and responsive behavior in a context in which politicians are often thought to rely mostly on partisan bias and clientelism. This descriptive evidence therefore underscores the puzzle of constituency service, motivating the theoretical and empirical contributions in the remainder of the book.

With this evidence in mind, Part II turns to explaining the sources of constituency service. In Chapter 4, I offer a theoretical explanation for why we should expect to see constituency service in patronage democracies, despite existing expectations to the contrary. In Chapters 5 through 10, I then provide substantial empirical evidence to support this argument in the Indian case. This Part II focuses substantially on the nonprogrammatic aspects of democracy in India and, especially, on how demands for assistance from high-level politicians originate partly from the partisan dynamics of local distribution. Thus, though this book contributes most centrally to understanding direct assistance by high-level politicians, it necessarily expends substantial attention on local politics, using original qualitative and quantitative data to illustrate the nature of local targeting and how this generates appeals for assistance from high-level politicians. This account is critical for the book’s argument, because as previously discussed, it is at least in part the multilevel nature of politics and distribution that generates both a demand for and the supply of constituency service. This focus on local distribution also allows for a more holistic depiction of the nature of political responsiveness in India.

In Chapter 4, I discuss in greater detail my argument for why constituency service constitutes a key element of distributive politics, alongside forms of locally brokered clientelism and contingent allocation of group-oriented goods. I elaborate the sources of demand for, and supply of, constituency service in a patronage democracy. I then consider the limitations of more widely studied forms of distributive politics and outline the ways in which constituency service offers a compelling alternative to politicians for pursuing their desired electoral ends.

This discussion suggests a number of empirical implications, laid out in detail in Chapter 4, which guide the analyses in subsequent chapters. In particular, I consider 1) the predictions of my argument for the relationship between citizens and both local- and high-level officials, including the character of demands for intermediation and assistance; 2) the partisan character, or (p.27) lack thereof, of distribution related to public goods and services; and 3) the tradeoffs high-level politicians make between different distributive strategies.

In Chapter 5, I present a discussion of service provision in India to set the context for an analysis of distributive politics and, in particular, constituency service, in a patronage democracy. I lay out the features of India’s political economy that continue to lead us to describe it as a patronage democracy. I then outline the characteristics of India’s primary political, bureaucratic, and social institutions, considering the formal roles and responsibilities of key actors at all levels of government with regard to distributive politics and their informal powers over distribution, so as to provide a comprehensive overview of the Indian political system at the national, state, and local levels. I show that, while local political actors often have substantial power over allocation of benefits from important development programs, high-level officials also have the capacity to influence the distribution of both these and other resources from the state. Thus, these senior officials can feasibly shape access to benefits and services in quite important ways for individual citizens. I draw on these accounts in later chapters, so as to showcase the ways in which the dynamics I describe and document provide key insights into the overall nature of distributive politics in India and, I argue, other patronage democracies. I conclude with a discussion of the potential for constituency service in India from the perspective of prevailing theories, considering in particular the character of political institutions, the nature of electoral and party politics, and the dynamics that may—or may not—encourage politicians to build personal reputations for responsiveness.

The discussion in Chapter 5 also helps establish the utility of the Indian case, and of particular states within India, for the study of constituency service. As I explore in additional detail in that chapter, India is a paradigmatic case of patronage democracy, in which the state is an important provider of employment and, even more so, of significant welfare benefits to large portions of the population. Yet, the allocation of these benefits is often highly discretionary, resulting in a distributive politics characterized by nonprogrammatic policy implementation and, in particular, by partisan bias, as I document in later chapters. Within India, moreover, a substantial portion of my empirical evidence comes from four states in which discretionary distribution is also characteristic.61 Three of these states—Bihar, Jharkhand, and Uttar Pradesh—are situated in the Hindi-speaking belt of north India and represent perhaps the (p.28) least developed region of India, where state inefficiencies and corruption are rife and clientelism of various forms is documented in public services.62 The fourth, Karnataka, is a more economically developed state in southern India. Yet, recent analyses suggest that patronage politics and the quid pro quo allocation of state benefits occur here as well suggesting, per existing accounts, that high-level politicians would have little incentive to engage in noncontingent distribution.63 From the perspective of received theory, India, and these states within India, therefore appear a surprising context in which to find substantial noncontingent, nonpartisan delivery of services directly to citizens by high-level politicians.

In short, India appears to be a “hard case” for constituency service. That constituency service would thrive in this patronage democracy—as I show it does—suggests the possible relevance of this book’s argument to nonprogrammatic democratic contexts elsewhere in the developing world—and thus makes it an opportune case for exploring the sources of constituency service in an in-depth and sustained manner.

In Chapter 6, I present the first set of empirical analyses focused on the dynamics of politics at the local level. Drawing on close-range qualitative shadowing of local council presidents and surveys of both presidents and citizens, I document the intimate relationships that these elected officials have with their constituents. This analysis suggests considerable responsiveness of local politicians to citizens, often highlighting a form of local constituency service that has not been sufficiently described in many prior accounts. However, I also underline the partisan nature of local politics, and the ways in which partisanship is emphasized, especially by those presidents who share the party of the more senior state legislator in their area. I then document the implications of this partisanship for contingent distribution, showing that individuals who do not share partisan ties with their village council president tend to be locally “blocked”—in particular, they are less likely to receive benefits from the local council.

The analysis of local politics continues in Chapter 7, where I examine the implications of this dynamic of local blocking for the character of citizen strategies to access state services. I show that non-co-partisans of local officials, who are more likely to be denied services locally, are also expected to make appeals to a larger number of potential intermediaries when attempting to access (p.29) benefits and services from the state, than do co-partisans of local officials. Using detailed survey data from three Indian states—which complements India-wide analysis elsewhere in the book—I show that these findings are particularly strong in Uttar Pradesh, where a long history of village council elections has enabled political parties to become more entrenched in local elections than in the neighboring states of Bihar and Jharkhand. I then establish that high-level politicians are important alternative sources of assistance, particularly when individuals have difficulty accessing public benefits from their local elected official. Importantly, I use survey data from the state of Karnataka to show that those individuals who appeal to high-level politicians for assistance are, on average, more successful in acquiring their desired service than those who appeal to local politicians. Thus, local blocking is associated with an increased demand for assistance from high-level politicians.

Chapter 8 shifts attention back to the perspective of high-level politicians and evaluates in greater detail the incentives they may have to offer constituency service to petitioners. In particular, I test observable implications of my theory on constituency service’s supply, using the field experiment introduced in Chapter 3 to assess what motivates politicians to respond to petitions. A further analysis of the experiment finds that indicators of a personal vote—that the petitioner has voted for the politician in the past—can have a small effect on the quality of a politician’s response, if not the baseline rate at which they respond.

In this eighth chapter, I also investigate whether information on local blocking affects politicians’ willingness to respond. I show that, in general, politicians are slightly less willing to respond when an individual indicates that they have attempted to acquire assistance at the local level and have failed to do so. However, this occurs only when politicians are not given additional electoral information about partisanship or patterns of past support; given the plausibly predominant presence of co-partisans among local officials in many legislators’ constituencies, legislators may reasonably infer that the petitioner is a less persuadable voter in this context. Moreover, this finding is driven by the behavior of politicians in those states where there is a long history of local council elections, and, thus, political parties have had the opportunity to become entrenched locally. In these states, when information on past local appeals is combined with information that the petitioner shares the politician’s electoral preferences, the negative effect of information on local denial of service disappears. This implies that politicians in those states with strong local party penetration interpret information about a failure to receive assistance locally as an indicator of local partisan blocking and, combined (p.30) with information on electoral history, an indication that it is a supporter or persuadable voter who requires assistance. In the aggregate, politicians do not premise the provision of assistance on indicators of individual citizens’ partisanship; yet, consistent with qualitative evidence from politician shadowing, this experimental evidence substantiates the argument that politicians take the opportunity to reach potential supporters directly and helps motivate politicians’ responsiveness.

I expand on these analyses of the field experiment in Chapter 9, where I explore additional variation in the characteristics of politicians’ responsiveness. First, I investigate the degree to which politicians’ responses reflect state- and individual-level characteristics that may be associated with incentives to cultivate a personal vote. These analyses highlight, in particular, that—consistent with the theoretical discussion in Chapter 4 and existing work on the personal vote—electoral politics play a key role in affecting the degree to which politicians attempt to build their individual reputations via provision of assistance to individual constituents. Chapters 8 and 9 therefore collectively explore the conditions under which constituency service does and does not occur.

A major challenge for the study of constituency service, however, and an important contribution of this book, is to situate noncontingent assistance in relation to better studied forms of representation or distribution—such as partisan bias. The theoretical discussion in Chapter 4 anticipates that contingent allocation should be more likely where the cost of allocation is high and it is relatively easy to determine the electoral preferences of likely beneficiaries. Thus, the same politicians may engage in both constituency service and partisan bias, depending on the nature of the benefit. In Chapter 10, I evaluate the conditions under which politicians will allocate benefits in a contingent, versus noncontingent, manner, using evidence from politician surveys, experimentally induced variation in the type of good for which citizens request assistance in my audit study, and data on development spending by politicians.

I show that—while less frequent than particularistic requests—citizens do ask politicians for group-based goods, and these are largely the same types of goods that state legislators allocate using their proprietary constituency development funds (CDFs).64 By mapping the locations of CDF projects and (p.31) matching these to the locations of polling stations, I then show that citizens living in areas that offered strong support to a politician in the last election are much more likely to receive spending from that politician’s CDF.

Drawing on data from my experimental audit, which includes the set of politicians from Karnataka for whom I have CDF spending data, I then compare the behavior of politicians spending their CDF funds with responses to the experimental audit of responsiveness. This comparison shows that, while patterns of electoral support do predict behavior with regard to partisan targeting, they offer relatively less explanatory value for understanding patterns of constituency service. Thus, the same factors cannot explain both partisan bias and constituency service, and the same individuals who engage in noncontingent individual assistance may also target group-based benefits in a largely partisan manner.

What are the implications of these findings in India for our understanding of distributive politics and representative democracy more generally? These are the questions I take up in the final Part III of the book. In Chapter 11, I consider the extent to which we should expect to observe similar dynamics of distributive politics in other parts of the world. I draw on a range of cross-national data to show that the contextual characteristics supporting constituency service—the dynamics of patronage democracy, difficulty in access to public benefits, and partisan allocation of benefits at local levels, accompanied by the presence of high-level representatives with little ability to monitor individual electoral behavior—coexist across a range of democracies around the world. I offer evidence to suggest that high-level politicians in countries across Africa, Asia, and Latin America also engage in individual-level distribution with an eye toward building a personal vote, rather than support for their party, and that highly partisan distribution by local operatives may ironically heighten their incentives to assist constituents in a nonpartisan manner. Thus, India, as a patronage democracy and place where high-level politicians offer assistance to individual citizens, is an exemplar of a common trend, rather than a global outlier. My findings in this book therefore point toward a broader research agenda focused on constituency service in the developing world.

Finally, in Chapter 12, the conclusion to the book, I address the broader implications of my findings for our view of representative democracy in many parts of the world. I discuss in greater detail the implications of my analyses of constituency service for understanding the “repertoires of distribution” used by politicians. I then consider the normative implications of my findings for our understanding of democratic practice in patronage democracies. I posit (p.32) that the form of representation present in these contexts is characterized by “constrained accountability.” High-level politicians in patronage democracies are, I argue, substantially more accountable to their individual constituents than we have been led to believe by the existing literature. However, the nature of this accountability remains, in multiple ways, limited. Citizens are hindered in their ability to access high-level politicians, relative to local intermediaries. Politicians’ logic for offering constituency service is influenced by its relationship to the nature of local, contingent distributive politics. And, finally, the long-term responsiveness of politicians to the inefficiencies in public service delivery brought to their attention by citizens’ requests is hampered by their short-term electoral incentive to remain relevant intermediaries for their constituents. Constituency service, then, is an important conduit for achieving democratic responsiveness in contexts otherwise characterized by discretionary targeting and perverse accountability, but it is a particular form of representation that is still shaped and constricted by the environment from which it emerges.

In providing this form of limited accountability, constituency service also serves to support the functioning of democracy in patronage contexts. While the targeted nature of clientelist and partisan distribution excludes a large portion of voters from the significant resources of the state in patronage democracies, constituency service offers those same voters a potential resource for accessing the state. High-level politicians, who play important roles in partisan distribution, also provide direct assistance to their constituents in a nonpartisan manner. This responsiveness makes available the resources of the state to a much wider swath of voters than would otherwise be the case and, in doing so, contributes to the functioning, and persistence, of patronage democracy.

Methodological approach

Studying the prevalence and determinants of constituency service is complicated by both empirical and methodological challenges. The absence of data, particularly data that match legislators to individual constituents, makes difficult the evaluation of claims about the link between partisanship and provision of assistance. In addition, analyses of distributive politics are often complicated by social desirability bias, in particular, the potential perception that clientelistic or other relationships between politicians, brokers, and citizens are undesirable or illicit. Furthermore, many inferential difficulties arise in interpreting responses to observational survey questions in this context.

(p.33) In this book, I therefore use a range of novel data and distinct methodological techniques to document the presence of constituency service in India and patronage democracies more generally, and to test my arguments about the character of contingent and noncontingent distribution in the Indian case. In particular, I triangulate evidence for claims about the behavior of politicians, and their interactions with citizens, by utilizing in-depth qualitative observations in the field, surveys with a range of citizens and state actors, administrative data, and field experimental techniques. This diverse set of data sources and analytic approaches offers a unique and comprehensive view of the behavior of politicians in their constituencies and, in particular, the strategies they use to affect the distribution of state resources.

To set the stage for these empirical analyses, I present here an overview of my data sources as a whole, to serve as a reference for later discussions throughout the text; details of additional methods and data collection are addressed in the relevant chapters and in the Appendix. In Table 1.1, I provide a summary of all original data sources upon which I draw in the empirical analyses.65 These are qualitative interviewing and shadowing of politicians; surveys of politicians, bureaucrats, and citizens; administrative data on allocation of state benefits; and a large-scale field experiment with state- and national-level politicians. For reference in later chapters, I refer to the study name in the leftmost column when discussing each individual data source.

These data combine to offer a comprehensive view into the lives and activities of politicians in their constituencies and in interaction with constituents. Not only can we examine the minutia of the daily work that occupies so much of their time, but we also gain insights into the ways in which they perceive, and report on, their work as a whole. In addition, we are able to examine objectively the ways in which they respond to citizens when unaware that they are being observed. While no single data source provides a complete understanding of politician behavior, I suggest that the combination of these data offers new and important insights into the ways in which Indian politicians, and their peers in other patronage democracies, engage in representative behavior on a day-to-day basis. (p.34) (p.35) (p.36) (p.37)

Table 1.1 Original Data Sources




Uses of the Data

Politician Survey #1a

Face-to-face surveys and survey experiments with a near census of national and state legislators and probability samples of district, block, and village councilors in three Indian states: Bihar, Jharkhand, and Uttar Pradesh


  • Assess prevalence and role of direct assistance to individual citizens within a portfolio of politician activities (Chapters 2, 6, 7, 10)

  • Analyze how attributes of petitioners affect politicians’ responsiveness (Chapters 3, 6)

Politician Survey #2b

Online survey of state and national legislators drawn from nearly all of India’s states; recruited via email and WhatsApp


  • Evaluate generalizability of findings from Politician Survey #1 (Chapter 2)

  • Assess impact of wider range of petitioner attributes on responsiveness, using conjoint experiment (Chapter 3)

Citizen Survey #1c

Face-to-face surveys with random sample of “service seekers” outside government offices in state of Karnataka


Assess the service-seeking strategies and success rates of people who desire to acquire public services (Chapters 2, 7)

Citizen Survey #2d

Face-to-face surveys and survey experiments with probability samples of citizens in Bihar, Jharkhand, and Uttar Pradesh villages


Evaluate variations in citizen claim-making and access to public services, as well as the degree to which provision of assistance is perceived to be conditional on citizen attributes and behaviors (Chapters 2, 5, 6, 7)

Citizen Survey #3e

Face-to-face household and government office surveys with citizens in thirty-one states and union territories


Evaluate variations in citizen claim-making and use of bribery (Chapter 2)

Bureaucrat Survey #1g

Face-to-face surveys and survey experiments with district, block, and village bureaucrats in three Indian states: Bihar, Jharkhand, and Uttar Pradesh


Assess relationship between appointed officials and elected counterparts in government (Chapter 5)

Politician and Bureaucrat Interviewsh

Face-to-face interviews with public officials


Evaluate the characteristics of public service delivery and the relevance of political intervention

Politician Shadowingi

Sustained observation of daily activities of state legislators in five Indian states (Assam, Bihar, Delhi, Rajasthan, and Uttar Pradesh) and village council presidents in Uttar Pradesh

Fourteen state legislators and fourteen village council presidents

  • Observe the quality, content, and interpersonal dynamics of legislator-constituent interactions (Chapters 2, 8)

  • Secondarily, assess the types and provenance of citizen petitions (Chapters 6, 10)

Politician Spending Data

Polling-station-level data on spending of a state-level constituency development fund (CDF) by legislators in the state of Karnataka

224 legislators, across 52,034 polling stations

Assess the partisan logic of allocation of constituency development funds (Chapter 10)

Politician Field Experiment and Dataset on Politician Characteristics and Responsivenessj

Experimental audit in which fictitious constituents sent messages to legislators requesting assistance with specific government services. In the broader dataset, audit responses are merged with data on state- and individual-level demographic and electoral characteristics.k

3,936 state and national legislators in nearly all Indian states

  • Evaluate whether experimentally manipulated attributes of constituents, such as past support, partisanship, and experience with local blocking, affect responsiveness (Chapters 3, 9, 10)

  • Test the relationship between state and individual-level characteristics and responsiveness (Chapter 9)

Cross-National Dataset on Patronage Democracy

National-level data from multiple public sources on characteristics related to the presence of patronage democracy and constituency service

Sixty-four countries

Evaluate the likely presence of high-level constituency service in global patronage democracies (Chapter 11)

(*) N=Actual number of respondents or subjects for which data was collected.

(a) This study was approved by the University of California, Berkeley, Committee for the Protection of Human Subjects under protocol no. 2013-07-5471.

(b) This study was approved by the University of California, Berkeley, Committee for the Protection of Human Subjects under protocol no. 2013-07-5471, as amended on June 8, 2016. The analysis was preregistered with Evidence in Governance and Politics (EGAP) under registration ID 20160921AB.

(c) This study was approved by the University of California, Berkeley, Committee for the Protection of Human Subjects under protocol no. 2005-12-33.

(d) This study was approved by the University of California, Berkeley, Committee for the Protection of Human Subjects under protocol no. 2013-07-5471.

(e) Conducted by Transparency International India and the Centre for Media Studies. Analyses described in this book are new and original work by the author.

(f) Refers to the number of sampled households, not individuals.

(g) This study was approved by the University of California, Berkeley, Committee for the Protection of Human Subjects under protocol no. 2013-07-5471.

(h) This study was approved by the University of California, Berkeley, Committee for the Protection of Human Subjects under protocol no. 2005-12-33.

(i) This study was approved by the University of California, Berkeley, Committee for the Protection of Human Subjects under protocol no. 2015-12-8200.

(j) This experimental audit was approved by the University of California, Berkeley, Committee for the Protection of Human Subjects under protocol no. 2016-02-8365. The analysis was pre-registered with Evidence in Governance and Politics (EGAP) under registration ID 20160926AA. The dataset on Politician Characteristics and Responsiveness merges legislator data from multiple sources.

(k) Individual electoral data from Jensenius et al. (no date). Individual caste data from Kumar et al. (no date).

Conclusion: Integrating constituency service in distributive politics

Constituency service—defined as noncontingent assistance to citizens—is not a prevalent theme in studies of distributive politics in patronage democracies. Yet this is not because it does not exist or does not constitute a significant form of distribution. In this book, I provide substantial evidence that constituency service is a common and important form of allocation in such contexts. However, existing theories of allocation have focused so explicitly on contingency, and on how partisan bias affects patterns of distribution, that they have missed the potential relevance of this alternative strategy for allocation. Perhaps most importantly, existing scholarship has ignored the ways in which the very dynamics it describes—the targeting of specific voters over others for receipt of fundamental government resources—may generate a demand for assistance that engenders the supply of noncontingent allocation.

On the pages that follow, I set forward an agenda for understanding the ways in which such forms of assistance constitute a key element of political representation, as well as distribution, in patronage democracies. As a form of service responsiveness, constituency service potentially exemplifies a model of representation that has been assumed not to exist, or exist in a very limited way, across much of the world. Yet, as I consider theoretically and explore empirically, there is substantial reason to believe that politicians in contexts characterized by nonprogrammatic distribution are remarkably more responsive than we have previously believed. While there are important normative caveats to these claims that I explore further, a key implication of this book is that democracy in nonprogrammatic contexts may be far more representative than has previously been claimed.

Constituency service therefore helps to make patronage democracy work. In this book, I aim to show how and why this occurs.


(1.) Personal interview, April 22, 2016.

(2.) Observations from shadowing of politicians, Respondent F. I describe this methodology in Chapter 2 and Online Appendix B.

(3.) The first quotation is from Mohapatra (1976), the second from Chopra (1996: 102). See also Maheshwari 1976.

(4.) See “The burden of responsiveness,” Chapter 2.

(5.) See “Politician-citizen interaction in patronage democracies,” Chapter 11.

(6.) Afrobarometer 2008, Latinobarómetro 2008; see “Politician-citizen interaction in patronage democracies,” Chapter 11.

(7.) UNDP and IPU 2012; Barkin and Mattes 2014; personal interview with former staff member of Colombian senator.

(8.) See “Do appeals to high-level politicians resolve citizens’ problems?” in Chapter 7 and also Gupta 2017.

(9.) See “The value of responsiveness,” Chapter 2.

(10.) See below and Chapter 2.

(11.) Fenno 2003 (1978).

(13.) Fenno 2003 (1978): 101.

(15.) Chubb 1963: 276.

(16.) Chandra 2004, 2009.

(18.) See, e.g., Cox and McCubbins 1986, Lindbeck and Weibull 1987, Dixit and Londregan 1996, Porto and Sanguinetti 2001, Chandra 2004, Chhibber and Nooruddin 2004, Wilkinson 2007, Arulampalam et al. 2009, Keefer and Khemani 2009, Baskin and Mezey 2014, and Ejdemyr et al. 2017. Politicians in advanced countries may have substantial capacity to engage in targeted distribution as well, as discussed later; yet the contexts I discuss are characterized by a greater extent of political discretion in policy implementation.

(19.) On clientelism and partisan linkages, see inter alia Auyero 2000, Stokes 2005, Kitschelt and Wilkinson 2007, Nichter 2008, Stokes et al. 2013, and Weitz-Shapiro 2014. For a capacity argument emphasizing the role of traditional authority, see Baldwin 2013.

(20.) Delegation is not costless for politicians; as Stokes et al. 2013 describe, there are multiple downsides to the principal-agent dynamic associated with brokers, but in many contexts the benefits may be sufficient to offset the costs. I consider the broader implications of these limitations below and in Chapter 4.

(21.) Fenno 2003 (1978): 101.

(22.) See, e.g., CMS/TII 2005, 2008; Bussell 2012.

(28.) See, e.g., Stokes et al. 2013, Chapter 1.

(29.) Pitkin 1967: 209.

(37.) See, e.g., Khera 2011.

(38.) Scott 1969: 1142.

(41.) Paraphrasing Fenno 2003 (1978): 102.

(42.) For alternate usage of the term see, inter alia, Lindberg 2010, UNDP and IPU 2012.

(43.) Pitkin 1967: 209.

(45.) In general, and throughout the text, I use feminine pronouns to refer to politicians and masculine pronouns to refer to bureaucrats.

(46.) What I define here as partisan distribution, including both clientelism and partisan bias, is akin to a broad definition of clientelism as used by Kitschelt and Wilkinson (2007) and Ziegfeld (2016), inter alia.

(48.) Fenno 2003 (1978): 102.

(49.) The US example is from Fenno (2003 [1978]: 101).

(51.) Piliavsky 2014a: 7–8.

(52.) See Piliavsky 2014a: 16–17.

(56.) Chandra 2004: 115–29.

(57.) See “The global prevalence of patronage democracies,” Chapter 11.

(58.) Pitkin 1967: 209.

(60.) Whether this perverse accountability is a problem for representation empirically is a different question. If, as recent literature suggests, local clientelist brokers are largely targeting for benefits distribution those individuals who are already co-partisans—their core voters—then there is little empirical effect of the quid pro quo on behavior.

(61.) However, I draw on additional evidence from across India to examine trends and variation in the behavior of high-level politicians.

(64.) As I describe in greater detail in Chapter 10, constituency development funds provide politicians with a lump sum amount they can use for development projects in their constituency. These funds are an increasingly common around the world (UNDP and IPU [2012], Baskin and Mezey [2014]).

(65.) See Online Appendix B for additional details.