Making Legibility Work: The Bureaucratic Politics of State Knowledge

The production of state legibility has been a prolific subject of study. However, most works have not paid much attention to the quotidian labor of the street-level bureaucrats that implement legibility projects at a local level. The aim of this article is to explore the implementation of a social registry system at a local level to understand how frontline workers make the population legible. Instead of taking legibility as an object of evaluation or critique, we pay close attention to the inner workings of bureaucracies at the instances in which the sociomaterial conditions of the population are translated into data. Drawing from qualitative research in Peruvian municipalities, we describe the operations of an algorithmic system that classifies the population for the distribution of welfare. We observed how under-resourced bureaucrats were constrained by regulations and technologies of the system. Paradoxically, to make the system work for their local realities, the bureaucrats had to bend the rules and find workarounds. From this perspective, the making of legibility looks less like a top-down exercise of bureaucratic compliance or a story of domination over the population. Instead, we find actors attempting to maintain a delicate balance between inadequate legal rules, scarce resources, and sociopolitical demands.

In decentralized countries, core state functions such as collecting data from the population are delegated to local governments. Which local authorities allocate better human resources for these administrative tasks? To answer this question, I focus on a national social registry system in Peru where legibility activities are carried out by municipalities. I argue that the experience of local appointing officials matters for implementation; specifically, I expect that reelected politicians, who have accumulated knowledge of the formal and informal rules of bureaucracy due to political continuity, hire more qualified bureaucrats than newcomers. I test this argument by leveraging novel administrative data sources that detail municipal officials’ backgrounds and data-entry error rates. I employ a regression discontinuity design as an empirical strategy. The results fail to reject the null hypothesis that mayoral reelection has no effect on the quality of bureaucracy. Considerations about heterogeneity and incumbency disadvantage suggest that these null results are informative: maintaining political continuity does not ensure better personnel quality for datacollecting activities.

This paper studies social embeddedness to assess whether welfare legibility projects, designed to codify and standardize data on the poor, are better implemented by street-level bureaucrats with strong ties to the communities in which they work. I examine the local implementation of a national social registry in Peru, created to collect and process socioeconomic information from households requesting targeted welfare benefits. I leverage large-scale granular panel data on 1,158,604 questionnaires administered to households in the field and subsequently submitted by 1,510 municipal bureaucrats in the office. I analyze data-entry errors made by street-level officials when processing households’ questionnaires digitally in software. To measure social embeddedness, I evaluate bureaucrats' local ties by assessing the concentration of officials’ surnames relative to the general surname population in a given jurisdiction. Additionally, I collect data on officials’ term lengths as a complementary measure of practical knowledge and local expertise. Using a fixed-effects specification, I do not find a substantial association between social embeddedness and information-making performance. Instead, I find disparities in legibility across levels of practical knowledge. These disparities are not fully explained by other officials' or district-level characteristics, showing that these factors play only a partial role. The study suggests that making the poor legible is less about minimizing the information costs of outreach and data collection and more about offering frontline bureaucrats stable conditions that allow them to improve their data-processing skills over time.

This study analyzes the variation in subnational implementation of Peru’s national welfare protection system. We focus on the bureaucratic capacity of local street-level bureaucrats, who are tasked with following the central state's mandate to collect and process socio-economic data from the poor. We argue that in contexts where local authorities have short-term political careers, some exploit the state more than others by appointing unprofessional cadres, particularly in welfare sectors. The degree to which politicians exploit welfare bureaucracies depends on the political strength mayors derive from electoral contests. Specifically, mayors perceived as strong after significant election victories are more likely to extract visible but less salient state resources, such as bureaucratic positions needed for national welfare provision. The study also suggests policy sector heterogeneity: landslide victories negatively affect the bureaucratic quality of welfare officials, while this impact is null for officials in other administrative areas.

Using a mixed-method approach, we present observational evidence showing the relevance of political results in enhancing the opportunistic behavior of local authorities, particularly regarding the appointment of non-professionals working for the implementation of national welfare provision. We utilize novel administrative data from local social registry offices across all municipalities in Peru. This data includes detailed information on the experience and educational credentials of managers, data-entry clerks, and enumerators.  We implement case studies in an urban municipality to examine the causal mechanisms. The study thus concludes that in contexts where social policy provision is less politically salient, local implementation is hindered indirectly by the predatory behavior of empowered politicians.