Who Knows How to Govern? Procedural Knowledge in India’s Small-Town Councils

Abstract

Governments across the Global South have decentralized a degree of power to municipal authorities. Are local officials sufficiently knowledgeable about how to execute their expanded portfolio of responsibilities? Past studies have focused on whether citizens lack the requisite information to hold local officials accountable. We instead draw on extensive fieldwork and a novel survey of small-town politicians in India to show that local officials themselves have distressingly low levels of procedural knowledge on how to govern. We further show that procedural knowledge shapes the capabilities of officials to represent their constituents and that asymmetries in knowledge may blunt the representative potential of these bodies. Finally, we show that winning office does not provide an institutionalized pathway to knowledge acquisition, highlighting the need for policy-based solutions. Our findings demonstrate the importance of assessing knowledge deficits among politicians, and not only citizens, to make local governance work.

Khejri Nagar, a small town of 25,000 people in the north Indian state of Rajasthan, sits just off the main highway connecting Jaipur and Agra. Compared to the two larger cities it links, Khejri Nagar is a quiet, peripheral place. Modest houses make up most of the built space. A few markets line Khejri Nagar’s streets, offering residents basic provisions and wares. Like many towns in Rajasthan, Khejri Nagar has a crumbling, colonial-era fort perched on a nearby hilltop. Public infrastructure in town is in a similar state of disrepair. Paved road coverage is spotty, leaving the town’s streets rugged and flood prone. Trash clogs gutters and fills vacant plots. Unchecked construction pushes the town’s boundaries into the countryside, mocking Khejri Nagar’s master plan. Service delivery challenges can be seen at every turn (See Figure 1 for example images from other small towns in Rajasthan).¹

Figure 1. Governance in India’s Small Towns

The main set of actors charged with addressing these development challenges are Khejri Nagar’s ward councilors. Under a constitutional amendment passed in the early 1990s, Khejri Nagar and other towns in India have been devolved a list of political and fiscal powers and are required to hold local elections every 5 years. As part of these decentralization reforms, Khejri Nagar has been carved into 25 wards, each of which directly elects a councilor. Councilors have a range of tasks, including representing their constituents in the nagar palika (municipal council), requesting public services for their ward, participating in meetings to craft and approve the annual budget, spearheading efforts to boost revenues through tax and fee collection, and ensuring town development aligns with stated guidelines and priorities.

Most councilors in Khejri Nagar, however, lack knowledge of how to carry out these responsibilities. While decentralization is anticipated to empower local officials who know what outcomes residents want, few scholars have asked if these officials know how to produce such outcomes. Governance is, in large part, a technical undertaking: officials need to know how to get their requests incorporated into town budgets, initiate procurement processes for projects in their constituencies, ensure routine service maintenance, bolster tax and fee collection, and follow zoning regulations.

Yet most councilors in Khejri Nagar we spoke to are unaware of these responsibilities, let alone know how to fulfill them. Councilors exhibit distressingly low levels of procedural knowledge: information regarding the rules and procedures related to core governance activities, including revenue-raising, spending, and institutional protocols. Most did not know how the town budget is made, when the council last met or is supposed to meet next, how to submit requests for public services, or how procurement decisions are made. Many complained that even after a few years in office, they know little of such processes.

According to these same councilors, such deficiencies in procedural knowledge have consequences for their ability to govern. As one councilor stressed, procedural knowledge “is very important. Councilors are the first step in the governance ladder. If the first step is not capable, then the whole ward falls apart. A councilor has a lot of responsibility. It is important for them to be educated and informed” (Interview 3). Another councilor, describing the consequences of poor training and low procedural knowledge among his colleagues, told us, “If councilors do not receive proper training, they will waste time. If they are not able to do work, then citizens will be upset. Development work will be slow” (Interview 4).

Are procedural knowledge deficits broadly observed outside of Khejri Nagar? If so, do such deficits constitute a significant obstacle to the functioning of urban governments? For local representatives to work in the manner expected by champions of decentralization, they must have an understanding of government procedures and responsibilities.

Studies on decentralization across the Global South implicitly assume that local officials possess information on how to govern. Underperformance for governance outcomes is typically attributed to weak political incentives to act on this knowledge (Grindle 2009). Weak incentives in turn are tied to insufficient accountability pressures from citizens below. Consequently, studies assess and intervene in improving citizen knowledge about politician performance to strengthen the latter’s incentives (Dunning et al. 2019).

By contrast, there have been few, if any, attempts to systematically assess politician knowledge about how to govern. This lacuna is buttressed by the presumed dominance of non-programmatic politics, which prioritizes the ability of officials to work outside of formal rules and regulations rather than know and follow them (Stokes et al. 2013). While such descriptions resonate in our study setting, they do not preclude the role of formal rules in shaping the pace and bent of governance. Instead, our discussions with councilors in towns across Rajasthan underscored the importance of a basic understanding of key protocols of local government.

In this article, we draw attention to a crucial and overlooked set of questions for local governance. Are the officials empowered by decentralization sufficiently informed to carry out the duties they are entrusted with? Does a lack of procedural knowledge impact their ability to govern and represent constituents? If so, can such knowledge be acquired?

Answering these questions requires assessing whether officials possess procedural knowledge. We develop and implement such an assessment across 60 municipal councils in the Indian state of Rajasthan, a large province with a population roughly the same as Germany or Turkey. We first draw on 4 months of qualitative fieldwork, conducted between 2019 and 2022, across eight municipalities, to develop contextually specific measures of procedural knowledge. The materials we draw on include interviews with elected officials and unelected bureaucrats and council staff, and an extensive reading of official government documents, which we hand-collected from the Rajasthan Department of Local Self Government. Next, we use these measures to conduct a survey-based knowledge assessment of a representative sample of 1,142 ward councilors and 923 runner-up candidates from the most recent municipal elections (2019–20) across these 60 towns.

We situate our empirical efforts within India’s massive constellation of small towns. India’s small towns are substantively and theoretically productive spaces for our study. First, urban governance is an especially technical and multidimensional context where procedural knowledge is important. Second, the face of local government across India and the Global South is increasingly urban. Urban India is now home to half a billion people and will increase by roughly three hundred million people in the next 30 years (UN 2015, 219). Concurrently, the global urban population is expected to rise by 2.2 billion people, almost all within Asia and Africa (UN-Habitat 2022, xv).

Much of this urbanization will happen in small towns. Despite a scholarly focus on the India’s largest cities like Bangalore, Delhi, and Mumbai, only about 23% of India’s urban population resides in such cities with over five million people. By contrast, 5 in 10 Indian urbanites live in towns with under five hundred thousand people.² While much of India and the world’s urban population resides in small towns, most scholarly analyses and policy debates focus on large cities (Kumar and Stenberg 2022).³

Our study yields four key findings. First, we document that small-town officials are alarmingly ill-informed about how to govern. The average respondent was only able to answer about 40% of our questions correctly. In half of the 10 domains we examine, fewer than one in four respondents were able to answer our questions correctly. These results were sobering given our assessment’s rudimentary questions about local government. Our results are also striking given that Rajasthan’s small towns exhibit several features anticipated to strengthen incentives for politicians to acquire procedural knowledge: small constituencies, high voter turnout, and frequent party turnover (Mansuri and Rao 2012; Weaver 2021).

Second, we combine qualitative and survey evidence to show the substantive importance of procedural knowledge. Specifically, we show that such knowledge informs the efficacy of representatives in fulfilling several key duties, including involvement in crafting the budget, securing goods for their ward, holding connections to higher-level officials, and responding to constituents.

Third, we document stark inequalities in how procedural knowledge is distributed across elected officials. Our analysis finds women, inexperienced politicians, and members of marginalized ethnic groups display especially low levels of knowledge. Higher levels of procedural knowledge are concentrated among politically seasoned men of a high ethnic status, constituting an unacknowledged mechanism for entrenching elite dominance.

The uneven distribution of procedural knowledge across different social groups has important implications for democratic representation. Scholars have found that when women and marginalized ethnic groups hold office, public spending can bend more toward the preferences and material well-being of those same groups (Chattopadhyay and Duflo 2004; Gulzar, Haas, and Pasquale 2020). Such descriptive representation can also reduce social and political discrimination (Chauchard 2017; Jensenius 2017). Procedural fairness also undergirds an intrinsic justification for democracy, not just an instrumental one (Christiano 2002). The asymmetries in procedural knowledge we document in this study, however, may undercut procedural fairness and hamper the translation of descriptive representation into substantive gains by blunting the abilities of elected representatives from disadvantaged groups to advocate for their communities.

Finally, we show that holding office does not provide an institutionalized pathway to knowledge acquisition. A close-elections regression discontinuity (RD) design shows that narrowly winning office provides, at best, modest increases in procedural knowledge that are inadequate to remedy the alarmingly low baseline levels we document. Given the complexities of municipal governance, acquiring procedural knowledge often takes sustained experience on the job. Yet high rates of electoral competition, rotating quotas, and candidate turnover ensure most councilors in our study setting are first-time representatives, and very few last beyond a single electoral cycle.

Our article contributes to scholarship on decentralization, local governance, and urbanization. First, our findings are a reminder that formal processes matter, even in worlds often assumed to be dominated by informality and political discretion. Our focus on procedural knowledge increases attention to the mundane rules and procedures that are at the heart of decentralizing reforms. Second, we demonstrate the importance of assessing knowledge deficits among politicians, and not only citizens, to make local governance work. We show that procedural knowledge shapes the capabilities of officials to represent their constituents and that asymmetries in knowledge carry representative and distributive consequences. Third, we highlight the need for policy-based solutions for addressing knowledge deficits. Such deficits cannot be remedied merely by winning office or by institutional rules that ensure representation for vulnerable groups. Instead, we suggest the need for institutionalized training at the outset of an official’s tenure and reveal widespread enthusiasm for such efforts among officials. Finally, we advance scholarship on urbanization by centering small towns, which remain understudied despite housing much of the world’s urban population.

THE KNOWLEDGE AND PERFORMANCE OF LOCAL OFFICIALS

During the past four decades, a degree of authority has been devolved to local governments across countries in Africa, Asia, and Latin America (Rodden and Wibbels 2019).⁴ In many of these countries, local elections are now regularly held. Decentralization is expected to strengthen state responsiveness by electing local representatives who, because of their embeddedness in communities, intimately know local problems, can be more easily monitored by constituents, and produce better policy outcomes, especially with respect to local public goods (Crook and Manor 1998). Decentralization is also viewed as a political reform that can, alongside bottom-up citizen participation, promote democratic deepening and human development (Baiocchi, Heller, and Silva 2011; Heller 2001). However, evidence that decentralization corresponds to improved outcomes in governance and development is mixed (Rodden and Wibbels 2019). Such findings have prompted a large literature on why political decentralization does not always generate effective local government.⁵

Demand-side explanations center on citizen pressures and incentives to make local officials perform. Scholars have found that when citizens lack information about government performance, the attendant loss of oversight hampers accountability and development (Pande 2011). This observation has animated a flurry of experimental work on how information and social accountability tools—citizen scorecards, social audits, and grievance redressal systems—can empower citizens to demand more from officials (Dunning et al. 2019; Gottlieb 2016; Lieberman, Posner, and Tsai 2014; Pande 2011).⁶ Related studies examine the conditions under which citizens participate in local decision-making and mobilize to hold local governments accountable (Mansuri and Rao 2012).

An implicit assumption in demand-side arguments is that if local officials face strong incentives to perform, they will. Do local governments, though, have the supply-side capacity to deliver? A long-held concern among scholars of decentralization is whether local governments are given sufficient fiscal and administrative support to carry out their devolved functions. Across a wide range of contexts, studies document that resources and personnel are often lacking, undermining local government performance (Faguet 2014; Manor 1999). Studies also demonstrate that weak state capacity due to “overload”—bureaucrats given long lists of tasks with little support to carry them out—undercuts service delivery (Dasgupta and Kapur 2020).

Other supply-side explanations focus on the monitoring capabilities of senior officials placed above local government actors. Scholars have investigated how information technologies like smartphone apps can assist extra-local bureaucrats in tracking the performance of local public institutions and front-line service providers (Callen et al. 2023; Dal Bó et al. 2021). Scholars have also examined the impact of political oversight over local bureaucrats, finding that monitoring improves service delivery, especially in electorally competitive settings (Raffler 2022). Other studies show how senior officials who make space for iterative learning among their local subordinates generate governance environments conducive to improved service delivery (Andrews, Pritchett, and Woolcock 2017; Mangla 2022).

Related supply-side studies focus on concentrations of power among local elites. A large literature investigates elite capture, where locally dominant individuals and groups control decision-making and steer resources toward advancing their own interests, at the expense of those from marginalized social groups (Bardhan and Mookherjee 2006; Olken 2010; Mohmand 2019). In a similar vein, scholars have examined agency problems among local elites—bureaucrats and politicians—as a frequent source of underperformance (Grossman and Slough 2022; Pepinsky, Pierskalla, and Sacks 2017). These problems include moral hazard and adverse selection and often inform calls for oversight of one class of elites by another. Other studies reveal that a lack of cohesion among local elites can weaken bargaining with higher tiers of government (Wilfahrt 2018).

Across this expansive literature, little attention has been paid to whether local government actors have the requisite personal capacities to perform. This is particularly true for local representatives. We have a poor understanding of how qualified local leaders are for performing the duties given to them by decentralizing reforms (Carreri and Payson 2021; Raffler 2022). Scholars often rely on indirect measures such as education and occupational prestige as imperfect proxies for leader competence (Dal Bó et al. 2017). Only recently have scholars developed sharper measures of state actor quality (Carreri 2021). Much of what decentralization does is empower local representatives with new responsibilities, provoking the question of whether these actors have the knowledge to make the gears of the state turn for themselves and their constituents.

Procedural Knowledge

Our focus on the procedural knowledge of local elected representatives—the depth and breadth of information these actors have about their formal responsibilities, and how to carry them out—emerged from our reading of official municipal documents and our fieldwork in the small towns of Rajasthan. Two key documents, India’s 74th Constitutional Amendment Act as well as the 2009 Rajasthan Municipal Act, outline a range of responsibilities that are to be carried out by municipal governments, within which councilors are afforded key powers and responsibilities. Stipulated responsibilities for municipalities include land regulation and construction, roads and bridges, water supply, and providing public amenities including street lighting, parking lots, and bus stops. The municipal council is entrusted with deciding how to apportion the town budget to meet these responsibilities, which must be approved by councilors in an annual budget meeting. Councilors are also given the responsibility of electing a chairperson from within their ranks, who heads council proceedings. Councilors have the right to raise questions regarding demands for their ward and town to the council chairperson and senior town bureaucrat, move resolutions on matters connected with municipal governance, and inspect municipal budget and spending records. They are also empowered to levy certain forms of taxes and user fees from residents.

These functions are not marginal, nor do they simply exist on paper. Instead, they are core to public service delivery. The ability to carry out basic government functions is expected by many constituents who routinely turn to councilors to solve problems that require submitting paperwork for public works projects, engaging in budget meetings, and following the procedures that make municipal government tick. Importantly, these duties are not the sole purview of the chairperson or the senior town bureaucrat, but are given to the entire city council, and individual councilors within these bodies.

Many councilors recognized that procedural knowledge is crucial to getting things done. For example, a councilor from Rohida Nagar, a town of 40,000 people, told us, “If you don’t have knowledge about development-related matters then you cannot get anything done for your ward. What is the rule, what are the regulations? [If you don’t know them] you can get nothing done” (Interview 6). Another councilor concurred, noting, “knowledge of rules is very important. If someone doesn’t have knowledge of rules and regulations, they should not become a representative…They should first study the rules and then contest an election” (Interview 7).

Many of the councilors we interviewed discussed the need for training to fill gaps in procedural knowledge. One councilor in Rohida Nagar noted, “There should be some training by the government. I am new to politics, and do not have much knowledge about how to do works. Training will help me do a two-hour-long job in one hour’s time” (Interview 5). Other councilors discussed how greater knowledge of rules and responsibilities would yield efficiency gains in securing local public goods (Interview 8) and help them more effectively approach bureaucrats (Interview 9). A former councilor in Khejri Nagar framed the importance of gaining procedural knowledge as a matter of rights and accountability:

[Councilors] do not know anything about the municipality or their rights. When I was a councilor, I could not understand my rights for two years. But once I read all the books from the Directorate of Local Bodies, I went behind [the chairperson]. I filed many complaints against [the chairperson]…If the chairman and E.O. [senior town bureaucrat] do not make working committees then it is a violation of the government guidelines. Councilors here do not even know that there are currently no working committees (Interview 2).

In the town of Jamun Nagar, a councilor similarly reflected on how a lack of procedural knowledge inhibits his colleagues from exercising their statutory rights as elected representatives:

I know that a councilor has many rights, but I do not know what my rights and powers are. There should be some training. There is immense loss to not having training. Suppose someone in my ward says there is no bulb in a streetlight, then I will call the chairman who will tell me to file a complaint. I do not know what my rights are, how should I fight, and who should I contact for a particular problem (Interview 10).

In sum, our qualitative fieldwork in Rajasthan’s small towns emphasized the pivotal role of procedural knowledge in enabling councilors to carry out the governance tasks devolved to them under decentralization reforms. Despite this widespread acknowledgement of the importance of procedural knowledge for local governance, and the demonstrated lack of such knowledge among many local representatives, scholars have yet to systematically measure levels of procedural knowledge, assess its relationship with political efficacy, and examine why it varies across these political actors. Before turning to our research design, we motivate the importance of studying small towns and describe our study setting.

STUDY SETTING: THE IMPORTANCE OF SMALL TOWNS

Nearly half of the world’s urban population (48%) lives in cities with fewer than five hundred thousand people, far more than the proportion living in five-million plus megacities (20%).⁷ This trend rings true for India: just under half of India’s urban residents live in towns with populations below five hundred thousand people.⁸ Despite this demographic distribution, smaller towns in India and across the Global South are neglected in terms of both scholarship and public policy, relative to both rural villages and large cities.

Small towns in India and elsewhere tend to differ from larger cities in several key respects. The former tend to be sidelined in terms of economic investments next to larger cities (UN-Habitat 2022, 116). Scholars have found evidence of higher poverty rates in smaller cities across a diverse set of countries (Ferre, Ferreira, and Lanjouw 2012). A study of urban Brazil and Indonesia finds that non-state service providers also disproportionally cluster in larger cities (Post and Kuipers 2022). Scholars further point to low levels of state capacity in small towns, undermining their ability to govern (Kumar and Stenberg 2022; Kundu 2014). Small cities and towns thus tend to be less developed and have weaker state capacity than their larger counterparts. They demand systematic inquiry as distinct political spaces.

A nascent literature centers on India’s small towns. Mukhopadhyay, Zerah, and Denis refer to these towns as examples of “subaltern urbanization,” which exist “outside the metropolitan shadow” yet reflect the experience of a substantial percentage of India’s urban citizens (Mukhopadhyay, Zerah, and Denis 2020). The authors argue that small towns are not peripheral extensions of their larger counterparts and should be approached as having their own political economies. Recent research on small-town India examines several themes, including women’s employment (Naik 2022), social change and segregation (Sharma 2003; Sircar 2018), and agricultural markets (Krishnamurthy 2012). We advance this small but growing literature on India’s small towns by focusing on the performance of elected officials within them.

The Institutional Contours of India’s Small Towns

Urban spaces in India are divided into two broad categories⁹: statutory towns with a recognized form of urban government and census towns, which meet the demographic criteria for being urban but have rural forms of government. In Rajasthan, statutory towns with populations greater than five hundred thousand are usually governed by municipal corporations (nagar nigams), those between one hundred thousand and five hundred thousand are usually governed by municipal councils (nagar parishads), and those with less than one hundred thousand people are usually governed by municipalities (nagar palikas). Our study centers on nagar palikas and nagar parishads.

While municipalities cannot match the powers and resources of the state and central governments in India’s asymmetric federal structure (Jacob and Jacob 2022), they nevertheless have significant resources to carry out their responsibilities (details in Appendix C of the Supplementary Material). Between 2016 and 2020, a typical town in Rajasthan had, on average, Rs.102.3 million (1.3 million USD) every year for capital expenditures, over which they exercise significant control. These funds come from various sources, including unconditional fiscal transfers from the state and central finance commissions. In Rajasthan, municipalities also receive compensation for the removal of the Octroi (point of entry) tax in the early 2000s. Municipalities are further empowered to levy local taxes and fees, most importantly on property and land conversions. Moreover, they receive funding for specific central and state programs implemented at the local level.

A key feature of decentralization in India was instituting competitive elections across urban local bodies. In Rajasthan, councilors (parshads) are elected to represent municipal wards. Elections for councilors are held every 5 years, as required by India’s 74th Constitutional Amendment. The most recent municipal elections were held in Rajasthan in 2019–20. Alongside elected councilors are a small number of nonelected bureaucrats, headed by an executive officer or commissioner, who are posted in towns and given a variety of important governance tasks.

Ward councilors in Rajasthan’s small towns face incentives and pressures to perform for their constituents. They are embedded political actors, with most living in the tiny, roughly one-thousand-person ward they represent. The average number of wards (and by extension, councilors) in our sample of 60 towns is 36. The average population of a ward across our 60 sampled towns is 1,086 people, with a standard deviation of 468 people.¹⁰ The elections that bring councilors to power have high turnover. Nearly 80% of elected councilors in our sample have never won a prior election. This combination of embeddedness and electoral competitiveness is expected to yield politicians who are knowledgeable about, and responsive to, citizen needs.

SURVEYING SMALL-TOWN POLITICIANS

In the fall of 2021, we conducted a survey of 2,065 politicians across 60 small towns in Rajasthan (Figure 2). Our sample frame of towns covered all of Rajasthan’s nagar palikas and nagar parishads (185 towns in total). We took a simple random sample of 60 towns. The average sampled town has a population of 43,945 people, with a one standard deviation of 34,758 people. The most populous town in the sample has 165,294 people, and smallest has 10,000 people.¹¹ The smallest number of wards in our town sample is 20; the largest, 60. There are 2,160 wards across the 60 towns, which served as our sample frame of wards.

Figure 2. Rajasthan’s Small Towns

Note: Sampled towns in black points and non-sampled towns in white points. Shape files used in the map are from DataMeet and available here: http://projects.datameet.org/maps/.

We created wardwise lists of all winners (current incumbents) and runners-up across the 60 towns from the 2019–20 Rajasthan municipal elections. Next, we digitized information from affidavits that electoral candidates must submit to the state election commission to generate a comprehensive list of cell phone numbers for all winners and runners-up.¹² We then randomly selected 20 wards in each town for interviews.

A survey firm based in north India carried out the interviews over the phone (details in Appendix A of the Supplementary Material). The team of enumerators included both men and women, with the latter assigned to interview female politicians. The final number of surveyed councilors was 1,142, 35% of whom were women. The final number of surveyed runners-up was 923, 34% of whom were women. The percentage of female politicians in our sample closely matches the percentage of wards in each town that are “reserved” for women candidates—one-third—as mandated in the 74th Constitutional Amendment Act.

Before proceeding to the analyses of procedural knowledge, we first describe our 1,142 surveyed ward councilors and their governance activities. Doing so highlights the centrality of these actors in processes of urban governance, as well as their accessibility to citizens. Table 1 provides general descriptive statistics for the sampled councilors. Where possible, we also provide relevant comparative statistics from a sample of 201 ward councilors from Rajasthan’s largest city and provincial capital, Jaipur.¹³ Doing so helps us parse those features which distinguish small-town councilors from their counterparts in larger cities.

Table 1. Descriptive Statistics of Elected Representatives

Note: Table provides descriptive statistics from surveyed small town ward councilors ( $ N=1,\hskip-3pt 142 $ ).

Our survey reveals small-town officials to be approachable, socially diverse, and fairly politically inexperienced. First, councilors are important ports of call for citizen claims. Consistent with our field observations of routine citizen engagement, nearly 90% of our sample of councilors report receiving daily requests for help from constituents. They are expected to take up these requests, which requires engaging the municipality. To understand the nature and distribution of citizen requests, we asked respondents to name up to three of the most common types of constituent requests they recieved. The five most frequently listed request types were sanitation and solid waste removal (76%), water provision (58%), electricity provision (41%), accessing welfare benefits (42%), and infrastructure projects like paving roads and drainage construction (26%).

Councilors are also highly approachable. 82% of them live within the ward they represent. Given the small size of wards that councilors live in, most of them are approached directly by citizens and without the assistance of local intermediaries. 86% of councilors say citizens never approach them with the assistance of any intermediary. By contrast, only 28% of councilors surveyed in the capital city of Jaipur (with an average ward size of 16,000) said residents never approached them with intermediaries (Auerbach 2020).

Small-town councils are also diverse. Among those who sit on their council, 35% of councilors are women, almost all of whom are elected through gender quotas. 24% of councilors come from disadvantaged Scheduled Castes and Scheduled Tribes (who also receive electoral quotas), almost exactly the same proportion (25%) as those who come from privileged castes.¹⁴ This level of diversity is higher than what was found among Jaipur’s councilors. In Jaipur, 49% of the council came from privileged caste Hindus.

Finally, small-town councilors lack formal political experience. Just under 70% of councilors were contesting their first election, 20 percentage points higher than among our Jaipur sample. Only 30% have official party positions, compared to 73% of our Jaipur sample. Finally, 24% of councilors ran as independents, three times the proportion in Jaipur.¹⁵ Councilors are evenly split between the two major parties in Rajasthan, the BJP (37%) and Congress (38%), with the remainder as independents and others (25%).

MEASURING PROCEDURAL KNOWLEDGE

Our survey included a battery of 10 questions that jointly comprised a knowledge assessment of local politicians (see Table 2). We identified 10 knowledge domains from our reading of key central and state government documents collected during fieldwork that outline the workings of municipal governments, and a handbook for elected representatives written by the government-run City Managers Association of Rajasthan. We then verified the importance of these domains through interviews with officials working in Rajasthan’s Directorate of Local Bodies, which oversees the workings of urban local bodies across Rajasthan, and interviews with councilors and town bureaucrats (for more details, see Appendix B.1 of the Supplementary Material).

Table 2. Measuring Procedural Knowledge

Two points bear noting about the measures. The first is methodological. These ten measures fall into three categories, which motivate the construction of indices and data aggregation for the empirical analysis. Four questions measure general information about the rules and protocols of local government. These form an index that captures general knowledge. Two questions focus on the municipal government’s spending rules. These questions form an index that measures knowledge of spending rules. Decisions on how to spend municipal funds are a core responsibility of the council. About three-fourths of the funds each town spends come from state and central transfers, with the bulk of these funds relatively untied. Bureaucrats are involved in the implementation of spending decisions, but councilors are given the power to allocate funds to particular budget line items, especially with respect to capital expenditures. Four of the survey questions are on the municipal government’s revenue-side functioning. Together, these form an index that measures knowledge of revenue or taxation powers. While less than a quarter of town budgets come from own source revenue, the ability to raise such revenues is seen as an important power that has been devolved to municipal governments. For each domain, we take a simple average of constituent items to construct an index. We also construct an overall index using all 10 measures. In Appendix B.2 of the Supplementary Material, we show that an exploratory factor analysis corroborates the underlying dimensionality of our theoretically guided approach to constructing indices.

The second point is substantive, that these measures were constructed to capture rudimentary levels of procedural knowledge. In most instances, they measure a respondent’s basic awareness, rather than more technical questions about rules and responsibilities regarding their office. For example, respondents are asked if they are aware of their town having a master plan, and a simple piece of information prominently displayed on the cover of these documents (the year range of the plan), rather than asked to report details included within the plan itself.¹⁶ Similarly, they are asked if their council can levy new local taxes and not the exact set of steps needed to do so. Our measure therefore constitutes a low bar for respondents to cross, making a failure to do so especially sobering.

RESULTS

We present four main findings. First, we find that local politicians exhibit low levels of procedural knowledge. Second, we triangulate evidence from across our survey and fieldwork to show that procedural knowledge is associated with the everyday effectiveness of local representatives. Third, we show that women politicians, politicians from marginalized groups, and those with no prior electoral wins exhibit lower procedural knowledge. Finally, we demonstrate that winning office (incumbency) is insufficient for remedying procedural knowledge deficits.

Local Politicians Lack Procedural Knowledge to Govern

Our survey provides compelling evidence that municipal politicians have low levels of basic procedural knowledge. Figure 3 shows knowledge levels by domain and constituent measures. It is clear from the top panel that local politicians are particularly underinformed about spending-related matters. On average, local politicians correctly answered only 16% of questions relating to municipal spending, specifically the cost of projects that the head town bureaucrat can approve unilaterally, without engaging elected representatives, and the date by which budget estimates must be finalized by the municipality. Just 9% of politicians knew of the budget date, and only 24% correctly identified the senior bureaucrat’s spending power.

Figure 3. Small Town Politicians Display Low Procedural Knowledge

Note: The top panel reports the percentage of procedural knowledge questions pertaining to a domain that are correctly answered by local politicians. The bottom panel shows the percentage of local politicians that correctly answer individual questions that measure procedural knowledge. The figure shows 95% confidence intervals for every estimate, constructed using heteroskedasticity-robust (HC2) standard errors.

Local politicians appear to be somewhat more aware of revenue powers. On average, a local politician correctly answered about 60% of questions relating to taxation, specifically the municipal body’s power to raise money through waste fees, land use taxes, and new local taxes. Collecting taxes and fees are crucial for towns, but many officials in these towns are unaware of this key duty. Only 43% know that municipal bodies can raise revenue by collecting waste fees. Similarly, only half of the politicians know that municipal bodies can raise money by imposing new local taxes or fees.

Local politicians display a similarly low level of knowledge about constitutional and legal provisions that govern the functioning of municipal bodies. An average local politician correctly answered 32% of questions pertaining to constitutional and legal provisions. Nearly all of them (99%) correctly identify their town’s classification, which determines a number of structural features of local government. However, only 2% know of the constitutional provision that formally instituted elections for local government. Furthermore, only 7% know of their town’s master plan and its valid year range, a crucial document that shapes how infrastructure development, land use, and resource collection should unfold. Similarly, knowledge about the town council’s rules and procedures is low. Only about one-fifth (21%) are aware that the council must meet at least six times in a year. These meetings are supposed to be important occasions for councilors to pursue their constituents’ interests and hold the chairperson accountable.

Across domains, a typical local politician only answered 40% of our procedural knowledge questions correctly. This is a low figure for elected representatives and runner-up candidates who are (or could be) entrusted with municipal governance, especially given the relatively low bar set by the assessment. This figure is also striking given that our measures are based on questions that relate to core mandates: regularly meeting as a body, raising revenues from residents, and shaping spending decisions on development and infrastructure.

We observe marked variation across the three knowledge domains we assess, with knowledge of expenditure procedures far lower than that of revenue powers. Interviews with elected officials suggest one reason may stem from the incentives of the chairperson who leads each council. Interviewees suggested chairpersons may prefer to keep councilors in the dark around expenditure rules, to allow the chair themself to retain greater control over spending decisions that may offer rent-seeking opportunities. By contrast, council revenue powers involve responsibilities that are politically unpopular (charging fees and imposing land taxes) and afford less scope for rents. Hence, the chairperson faces stronger incentives to involve councilors in those activities.¹⁷

Finally, we find that low procedural knowledge is a problem in all 60 towns of the survey. Figure 4 shows that there is limited variation in town-level scores. The median town-level score is 40% (0.4 out of 1). The 75th percentile town-level score is 42%, just 5 percentage points higher than the 25th percentile score (37%). At the extremes, the highest town-level score is 51% and the lowest is 32.7%. In other words, a typical politician in the “best” town only answers 51% of our procedural knowledge questions correctly. In the “worst” town, a typical politician correctly answers only 33% of questions. From the perspective of municipal governance, procedural knowledge appears distressingly low in either case.

Figure 4. Town-Level Variation in Procedural Knowledge

Procedural Knowledge Informs Governing Efficacy

An obvious concern with our assessment is the degree to which the knowledge we measure informs the governing capabilities of small-town officials. Does procedural knowledge matter? Evidence from our survey and interviews increases confidence in the importance of procedural knowledge. For example, a councilor who scored a 4 out of 10 on our knowledge index noted the following when asked what the main sources of revenue and expenditure in the town budget were:

Hum faltu hain [We are useless]…Whenever I go to ask for some work, they tell me there is no budget. It’s been over a year but nobody listens. They just tell us to wait for the budget to get works done but we do not know when the budget is supposed to come. I do not even know what the budget of the council is…Nobody tells me about the procurement process and what is the number of tenders that have gone public. How do I complain about this? I don’t know how to raise points in front of the chairman and the secretary. I don’t have enough knowledge about government programs and my rights as a councilor (Interview 19).

Contrast this response with the following from another councilor, who scored a 7 out of 10 (putting him in the top 4% of respondents):

Our source of income is the Urban Development (UD) tax taken from residential and commercial properties. It is about 20 lakhs (roughly USD 30,000)¹⁸ per annum. Other than that, we have municipal shops given on rent. But our main income source is conversion of agricultural land to non-agricultural land…Earlier all towns had a octroi [point of entry tax] system. Now that has been abolished. Instead, the government gives us a compensation grant. The rest of the money comes from the state and central finance commissions. Nowadays money also comes for the Swacch Bharat Mission [National Sanitation Mission]…(Interview 12).

The differences in responses between councilors at opposite ends of our procedural knowledge measure are stark. The first respondent appears to lack even basic knowledge of what the town budget’s size is, what major revenue sources are, and how spending on infrastructure occurs. By comparison, the second respondent offers granular information on different sources of revenue.

These responses suggest how procedural knowledge underwrites the effectiveness of elected officials in dispensing their duty. The first respondent’s lack of knowledge about the town budget impacts his stated ability to pursue projects for his ward, so much so that he openly states that he feels “useless.” The interview revealed he was unaware that municipal law in Rajasthan empowers him to inspect all documents kept at the municipality. By contrast, the final respondent not only identifies revenue sources but sources for spending for development projects in his locality.

Our survey evidence aligns with this qualitative evidence. We asked respondents a series of questions regarding their involvement in the town’s annual budget-making process, citizen engagement, and connections in municipal government. Figure 5 shows the partial relationship between procedural knowledge (on the x-axis) and these measures of efficacy (on the y-axis). We residualize both variables, netting out any variation explained by confounding background characteristics like age, gender, religion, caste, education, household income, political experience, and party affiliation.¹⁹

Figure 5. Procedural Knowledge and Governing Efficacy

Note: This figure shows the partial correlation between procedural knowledge and four measures of representational efficacy. The slopes here correspond to the regression coefficient for procedural knowledge reported in Appendix D of the Supplementary Material.

The top-left panel focuses on elected representatives’ involvement in the town’s budget-making process. The budget process is a key channel through which councilors can request goods and services for their ward. We asked respondents a series of questions including whether their town council had worked on the budget since they became a councilor, whether they personally made a suggestion or proposed change during budget discussions, whether they were able to secure funding for a project in their ward in the town budget, and their general assessment of the degree to which councilors in their town are involved in coming up with the budget. We create an index of these items by taking an average of all measures. We find a positive association between procedural knowledge and involvement in the budget-making process. Fixing all other factors, a councilor with perfect procedural knowledge is 0.09 scale units (s.e. $ =0.07 $ , $ p=0.16 $ ) more involved in the budget-making process, compared to a councilor with no procedural knowledge. Since this is an extreme comparison, we focus on the top 5% (65% knowledge score) and bottom 5% (20% score). Substantively, going from the top 5% of councilors to the bottom 5% decreases budget involvement by about one-sixth of a budget activity.

As a robustness check, we also construct a more conservative index of budget involvement that only includes the two most theoretically relevant items, both of which examine reported individual actions (making a suggestion during a budget meeting and securing funding for a project in their ward during budget discussions). Using the same model specification and right-hand-side variables, the coefficient on procedural knowledge increases to 0.21 and is more precisely estimated (s.e. $ =0.09 $ , $ p=0.02 $ ).

The next two panels focus on engagement with citizens. We measure politicians’ self-reported engagement with their constituents using two different indices. The first index (top-right panel) measures how “proactive” local politicians are in helping constituents. We focus on a range of common problems, including arranging for cleaning in the ward, helping residents request government documents, resolving disputes among residents, helping residents obtain COVID-19 tests or vaccines, and requesting work on roads, sewers, or drains within the ward. We find a positive association between procedural knowledge and being proactive. Fixing all other considerations, a politician who correctly answers all the knowledge-related questions is 0.12 scale units (s.e. $ =0.03 $ , $ p<0.001 $ ) more proactive than a politician who does not answer any of them correctly. When comparing the top 5% to the bottom 5%, this corresponds to helping constituents with roughly one-fourth an additional activity out of five.

The bottom-right panel focuses on another measure of engagement with citizens: an index to measure how much politicians report being contacted by their constituents to address local problems. If knowledge is correlated with being an effective representative, one can imagine citizens are more likely to go with their problems to a politician who is knowledgeable. This is indeed what we find. We measure the extent to which citizens rely on local politicians to address their problems using a seven-point scale, where 0 is when residents never contact a politician asking for help and 6 is when they do so on a daily basis. Higher values on this scale imply greater engagement with citizens who are seeking help with something. Fixing all other factors, a politician who correctly answers all the knowledge-related survey questions is 1.22 scale units (s.e. $ =0.22 $ , $ p<0.01 $ ) more likely to be approached by a resident than a politician with no knowledge. When comparing the top 5% to the bottom 5%, this corresponds to over half a scale unit increase or 10% of the full scale. It is worth noting that self-reported citizen engagement, using either measure (proactive behavior and citizen engagement), is high in the entire sample. Despite the possibility of ceiling effects, we observe substantively meaningful differences between high and low knowledge politicians.

All three of these measures are self-reported by respondents. We therefore cannot rule out our results potentially reflect the fact that politicians with better knowledge might simply be savvier in reporting high levels of engagement with citizens. That said, the associations reported in Figure 5 net out the variation explained by a number of predictors of being savvy—for example, education and political experience.²⁰

We also note our fourth and final test, reported in the bottom-left panel, which focuses on political connections. Our interviews suggest such connections are critical for getting work done. We measure political connections using three measures: whether the local politician knows the mobile number of the EO, chairperson, and member of the state legislative assembly. We find a statistically significant, positive association between knowledge and political connections. Fixing all other factors, a typical local politician with perfect knowledge is 54 percentage points (s.e. $ =0.06 $ , $ p<0.001 $ ) more likely to have the contact details of all local elites, compared to a local politician with no knowledge. Substantively, this corresponds to the top 5% of respondents knowing three-fourths of an elite phone number more than the bottom 5%. In summary, less knowledgeable politicians are also less able to access elites. This outcome provides a more stringent test because it is based on a councilor’s ability to provide a mobile number for each official, and not a fully self-reported behavior.

Overall then, politicians in Rajasthan’s small towns display low levels of procedural knowledge. Qualitative and survey evidence suggests such knowledge informs the efficacy of these politicians in fulfilling their proscribed duties. While these associations appear modest, it is worth noting that we use basic measures of procedural knowledge that potentially underestimate the difference between highly and less knowledgeable politicians. Future scholarship can employ measures that focus on technical details about municipal spending, revenue powers, and relevant administrative procedures to capture even greater variation in knowledge. Further, it is possible that knowledge affects governing efficacy in more nuanced ways, for instance, the amount of time and effort it takes to get work done.

Together, these findings highlight a significant concern for urban development. In empowering local actors like India’s municipal councilors, decentralizing reforms aim to build local state capacity, while making government more accountable and responsive. Yet such goals appear unlikely to be met if most of these actors have scant knowledge about their key duties and the core processes through which they can deliver development for citizens.

Marginalized Groups and New Entrants Exhibit Lower Procedural Knowledge

We earlier presented evidence from our survey which spoke to the inclusiveness and accessibility of India’s small-town councils: councilors are frequently approached directly by residents, and a high number are women, members of socially disadvantaged groups, and newcomers to electoral politics.

These descriptive patterns raise the specific question of how evenly procedural knowledge is distributed across politicians of different genders, ethnicities, and levels of political experience. Given the association with efficacy we documented, if procedural knowledge is concentrated among certain types of councilors, it may affect the degree to which high levels of descriptive representation in town councils translate into substantive efficacy in serving town residents. Troublingly, we find that procedural knowledge is unevenly distributed across men and women, ethnic groups of varying social status, and across local politicians with varying political experience.

Table 3 reports the results from a multivariate ordinary least squares regression in which respondent characteristics predict their procedural knowledge. The reported specification includes town fixed effects so that all comparisons are within-town. The first column of the table reports the regression coefficients for the overall index of knowledge, and the next three columns report the coefficients for subindices: an index with items that measure knowledge of spending rules, an index with items that measure knowledge of revenue powers, and an index with items that measure knowledge of relevant administrative rules and constitutional provisions.

Table 3. Correlates of Procedural Knowledge

Note: This table reports the coefficient estimates from a multivariate regression in which individual-level characteristics predict procedural knowledge. Column 1 reports the result for the overall index. Columns 2–4 report the result for subindices: knowledge of spending rules (column 2), revenue powers (column 3), and constitutional provisions (column 4). The model includes town fixed effects. HC2 robust standard errors are reported in parentheses. ***p<0.001, **p<0.01, *p<0.05.

With respect to gender, recall roughly one in three councilors are women. The analysis suggests that gender-based differences in procedural knowledge are pronounced. Fixing all other attributes, female politicians in our sample score 6.6 percentage points lower on knowledge questions compared to male politicians. This gender gap in knowledge is even larger for spending rules (7.8 percentage points) and revenue powers (9.9 percentage points).

Next, we find that political experience is associated with significantly greater procedural knowledge.²¹ This conforms to our expectation that local politicians might be learning on the job. The regression specification includes a numeric variable that captures the number of elections won by a local politician, ranging from 0 to 6.

Table 3 shows that political experience (number of election wins) is associated with large, statistically significant changes in procedural knowledge.²² Holding constant all other politician attributes, an additional election victory increases procedural knowledge by 3 percentage points. This difference is largest for knowledge of spending rules: an additional election victory is associated with a 6.4 percentage point increase in the knowledge score, fixing all other attributes. Put another way, a local politician who has won three elections is 19.2 percentage points more knowledgeable about spending rules than a local politician who has not won any election. However, this is a pathway of knowledge acquisition that is available to relatively few people. Only 7.8% of politicians in our sample have won two or more elections prior to the most recent (2019–20) contest.

We also find unevenness in procedural knowledge for politicians of ethnic groups of different levels of privilege. The regression includes a variable indicating marginalized group status that takes a value of 1 if the respondent is from a disadvantaged ethnic group (Dalits, Adivasis, and Muslims), and 0 otherwise. The third row of Table 3 shows that members of marginalized groups have consistently lower levels of procedural knowledge compared to members of elite or dominant groups (general and OBC caste categories). Fixing all other attributes, there is a 3.3 percentage point gap in the overall knowledge index between these two groupings, reflecting a 5.8 percentage point gap in knowledge relating to spending rules, a 3.4 percentage point gap in revenue powers, and a 1.9 percentage point gap in knowledge of constitutional and legal provisions.

Predictably, we find standard measures of candidate quality like age, education, and income to be positively associated with knowledge. For example, every 5 years of age is associated with a half a percentage point increase in knowledge scores, while every 5 years of education is associated with a 2 percentage point increase in scores. Counterintuitively, candidates that contest from the ward they reside in are less knowledgeable than those who contest from a different ward.²³

Appendix E of the Supplementary Material shows that these empirical associations are robust to machine learning algorithms like LASSO that drop non-predictive covariates. Supplementary Figure 8 reports these results.

Imbalances in procedural knowledge may blunt the quality of representation offered by local governments to constituents. The presence of women and members of marginalized councilors is often viewed as helpful for these communities to have their demands raised and met. Yet the lack of procedural knowledge among politicians from these very communities can hamper their abilities to competently deliver on this potential. While noting that subsamples within our survey cannot be regarded as representative, it is nevertheless instructive that the knowledge score for surveyed privileged caste men who had contested at least two prior elections was 0.586, compared to just 0.327 for non-upper caste women contesting their first election. However, the latter population was 9.6 times larger than the former (46–440) within our sample. Imbalances in procedural knowledge might therefore serve as an important channel through which governing power remains concentrated among a small elite.

Winning Office Is Insufficient to Reduce Knowledge Deficits

Both our key findings, low overall levels of procedural knowledge coupled with asymmetries in such knowledge, compel thinking through how such knowledge might be acquired. The most crucial pathway to assess is whether winning office provides an institutionalized way to acquire knowledge. If so, some of the pessimism of this analysis might be blunted.

To get at the causal effect of winning, we employ a close-elections RD design focusing on politicians who have not won any prior election. Typically, ward-level municipal elections have small electorates and the average winning margin in 2019–20 was 106 votes. Our sample includes 1,660 politicians who have not won any prior election, of which 918 were elected as councilors in 2019–20, while the remainder are runners-up.²⁴ The RD analysis estimates the short-term or immediate impact of incumbency on politicians’ knowledge levels. Did narrowly winning an election cause an increase in procedural knowledge 18 months later? Figure 6 reports the difference in knowledge levels at the discontinuity separating local politicians that narrowly win a municipal election from those that narrowly lose. The figure suggests that winners accrue greater procedural knowledge 18 months into the job, compared to losers. However, these differences are modest and often statistically insignificant.

Figure 6. Incumbency Does Not Remedy Knowledge Deficits

Note: The top panel shows regression discontinuity plots using a linear specification, triangular weights, and MSE optimal bandwidth. The plots zoom-in on data around the cut point ( $ \pm $ 25 percentage points). The bottom panel reports the difference at the cut point, specifically the robust estimate, standard error, and confidence interval generated by rdrobust. Estimates in Table 9 in Appendix F.4 of the Supplementary Material. Specification curves in Supplementary Figure 10.

For instance, winners have an overall knowledge score that is 3 percentage points higher than losers. This difference is statistically significant at the 0.1 level and corroborates the point estimate from the multivariate regression we reported earlier. Winners are also 7 percentage points more likely to correctly answer questions about spending rules, compared to losers. The spending rules difference approaches statistical significance at the 0.05 level. Winners are approximately 2 percentage points more likely to correctly answer questions on revenue powers or constitutional and legal provisions, compared to losers. These differences are not statistically significant at conventional levels, although we note that this might simply reflect inadequate statistical power. Even if that were the case, the effect sizes indicate winning office leads to at best modest improvements to low levels of baseline knowledge. Appendix F.4 of the Supplementary Material shows that these results are not particularly sensitive to different RD specifications. Appendix F.3 of the Supplementary Material shows that the results are also not sensitive to the exclusion of observations near the cut point.²⁵

One limitation of our RD design is that 18 months might be insufficient time for a councilor to acquire knowledge. Our results do not preclude a more gradual accumulation of procedural know-how. Indeed, our prior correlational evidence does suggest the possibility of gradual learning, as we note a modest and statistically significant difference of 3 percentage points in knowledge levels for every election won.²⁶

However, this gradual learning process is limited by two factors. First, the modest associations we document appear insufficient for producing highly informed officials, given the low levels of knowledge most officials in our sample have.²⁷ Second, winning multiple terms is a pathway open for very few politicians. Due to the high rates of turnover, most office holders are new to office, and many will not last past their current term. Recall that only 20% of our sample had won a prior election and that only 7.8% of our sample have won two or more elections prior to the most recent contest. Consequently, if holding office is to constitute a broadly accessible pathway for knowledge acquisition, it will have to work for first-time representatives and work relatively soon into their first (and often their only) term.

Instead of incumbency, perhaps proximity to experienced politicians like the chairperson might confer knowledge. Is this the case? We find that politicians from the chairperson’s party are as knowledgeable as those in the opposition camp. A typical “aligned” politician scores 40.2 out of 100 on our knowledge items, while “unaligned” (opposition) politicians score 39.6. There is no statistically significant difference between the two groups ( $ t=-0.661 $ , $ p=0.51 $ ). In other words, neither incumbency nor partisan alignment with the chairperson improves knowledge.

DISCUSSION

This article is motivated by two widely observed transitions across the Global South: urbanization and the devolution of key governing responsibilities to local elected officials. The intersection of these phenomena brings into focus the municipal governments tasked with overseeing development for a growing share of the world’s population. Prior studies have often pointed to the need to equip and inform citizens to hold these local governments accountable for executing their widened responsibilities. But none have systematically examined how informed local officials are to dispense their duties.

Our article introduces a key concept, procedural knowledge, designed to capture information that officials must have to govern effectively. We then conduct an intensive assessment of procedural knowledge in India’s small towns. We find that local politicians display strikingly low levels of knowledge, that this knowledge informs their governing efficacy, that it is unevenly distributed in ways that weaken the representative potential of local government, and that holding office is insufficient in remedying knowledge deficits.

These results suggest the need for scholars to rigorously assess procedural knowledge in other settings. Prior work tends to assume either that officials have such information or that formal procedures do not matter in settings dominated by informal, discretionary, or clientelistic politics. By contrast, we find procedural knowledge to be crucial in the multidimensional and often technical world of local urban governance. Further, our article provides a template for how to combine intensive qualitative fieldwork with surveys to conduct contextualized knowledge assessments at scale.

Our article also has important implications for urban policymakers. There is a growing emphasis in international development on strengthening the capacity of urban local governments to deliver public services (Carter and Post 2019; Obeng-Odoom 2013; UN-Habitat 2022). In India, there are serious debates regarding the inability of small towns to execute basic tasks, from implementing local projects to align with master plans to collecting revenue from residents. Urban governance challenges in other Global South countries have generated similar policy debates. These challenges suggest the need for sweeping institutional reforms that expand resources, administrative personnel, and the policy-making powers of municipalities. Our findings point to a complementary, but under-explored area for capacity building: the need to deepen the procedural knowledge of local representatives regarding their spending and revenue-raising powers, and the regulations meant to guide local development.

One question our findings provoke is why politicians do not make more efforts to acquire procedural knowledge, especially given several interviewees expressed an understanding of the value of such knowledge and a desire to obtain it.²⁸ Such enthusiasm is also reflected in the considerable appetite we find for institutionalized training among elected officials. Only 5% of councilors in the survey reported receiving any formal training by the state government about their roles and responsibilities. Among untrained councilors and all runners-up, 96% stated that their provincial government should provide training to winning candidates. As one experienced politician notes:

In reality, elected representatives do not know about the [local government] Act. Everyone needs training. As soon as someone gets elected, they should undergo training on matters related to the budget, even on the Act…They should be told about all these rules and processes (Interview 13).

The importance of institutionalized training is reinforced by three constraints to knowledge acquisition: high turnover, intersectional inequalities, and weak party organizations. First, we document that winning office does not constitute an effective path for knowledge acquisition, especially for the vast majority of officials who will not serve multiple terms. High turnover in our setting is underwritten by the large percentage of seats assigned to quotas for gender and caste, and the rotating nature of several of these quotas, which ensure few seats are “safe” for an incumbent across multiple election cycles. Second, councilors who most lack knowledge also lack means to obtain it. For example, politicians with less procedural knowledge are also less connected to political elites. Many of these politicians face intersectional forms of marginalization—due to gender and caste—that makes knowledge acquisition hard. Finally, an alternative pathway for knowledge acquisition through political parties is undercut by weak party organizations in our setting. Only 27% of respondents ever held a formal party position, and 74.5% attend party meetings infrequently (a few times a year or less). We also find that procedural knowledge is not positively correlated with shared partisan ties to the chairperson.

More optimistically, our survey points to high demand for specific areas of formal training, were it to be offered. We probed this point by asking what kinds of training would be most helpful. 73% of respondents sought training on the rules and procedures that govern council meetings. 70% wanted to know more about spending rules, while 68% wanted more information on how to issue work orders. Similarly, 66% wanted training on budget-making, while 57% wanted to know more about the revenue powers of municipal bodies.

This evidence points to productive extensions of our work, including in implementing and evaluating efforts to train local representatives, especially those from marginalized groups like women and disadvantaged castes. Further analyses could also look to explore whether features of the political, electoral, or institutional environment inform variations in levels of procedural knowledge among local officials. More broadly, our article underscores that processes of decentralization cannot end with the formal devolution of power. Realizing the intended goals of decentralization will require efforts to assess and inform the local officials entrusted to govern.