A Mercantile Theory of Expert Knowledge Utilization in Patrimonialist Bureaucracies: Evidence from the Health Sector in Peru

Abstract

This article formulates a theory of expert knowledge utilization in patrimonialist administrative traditions characterized by politicians’ predominance over bureaucrats. The argument is that in these weak institutional contexts acquiring “expert knowledge” enables politicians to control key bureaucratic functions that facilitates rent extraction to fund their campaigns, and in doing so, they eventually and circumstantially produce some positive welfare outcomes. My findings show that tenured politicians employ expert knowledge to expand their control over previously unknown administrative regulations, temporarily building capacities that enables them to (re)direct budgets to policy areas that report electoral gains, and sometimes incurring in collusion. In contrast, newcomer politicians often fail to take advantage of expert knowledge transferred given their inexperience in office hence producing less welfare outputs. This article sources evidence from a knowledge transfer program designed by the Peruvian central government to enlighten local politicians in the country’s fight against children stunting. Evidence combining a time-varying semiexperimental analysis (Panel DID) and in-depth interviews with regional and local politicians, as well as high level civil servants, largely confirm my theoretical claims.

Resumen

Este artículo formula una nueva teoría sobre la utilización del conocimiento experto en tradiciones administrativas patrimonialistas caracterizadas por la predominancia de los políticos sobre los burócratas. El argumento es que, en estos contextos institucionales débiles, la adquisición de “conocimiento experto” permite a los políticos controlar funciones burocráticas clave que facilitan la extracción de rentas para financiar sus campañas y, al hacerlo, producen eventual y circunstancialmente algunos resultados de política exitosos. Mis hallazgos muestran que los políticos experimentados emplean el conocimiento experto para ampliar su control sobre regulaciones administrativas hasta entonces desconocidas, creando temporalmente capacidades que les permiten (re)dirigir los presupuestos hacia áreas políticas que reportan ganancias electorales, e incurriendo a veces en actos corruptos. En cambio, los políticos nuevos a menudo no aprovechan el conocimiento experto transferido dada su inexperiencia en el cargo, por lo que producen menos resultados de política pública exitosos. Este artículo se basa en la evidencia de un programa de transferencia de conocimientos diseñado por el gobierno central peruano para ilustrar a los políticos locales en la lucha contra la desnutrición infantil crónica en niños. Mis herramientas metodológicas, que combinan un análisis semi-experimental en panel temporal (Panel DID) y entrevistas en profundidad con políticos regionales y locales, así como con funcionarios de alto nivel, confirman en gran medida mis afirmaciones teóricas.

Introduction

In 2020, a scandal broke in Caraveli, Southern Peru. The town’s mayor, Diego Montesinos, instead of investing in the long-awaited potable water system, inaugured a brand-new football camp only five meters away from a principal road (CE 2021). The citizens of Caraveli were upset and argued that the municipal budget was diverted to useless public works, without any consideration of the insufficient access to sanitation or potable water that have been generating many health problems to the district’s most vulnerable: more than half of the children suffered from intestinal infections, 74% were affected by a type of malnourishment, and 8% were chronically stunted (INS 2021).

Asked by the media about his decision, the town’s mayor declared that he had found easier to invest in a public ornament rather than in water and sanitation because the latter demanded more careful “technical knowledge,” “years of planning,” and “additional budget” he did not have (CE 2021). However, the Ministry of Finance of Peru (MEF) later proved that he has only executed an average of 39% of its yearly budget during his mandate (MEF 2021).

Cases like Caraveli are commonplace in Peru. The country’s central government has recently recognized that more than half of local governments’ budgets remain unspent (Gestion 2019). Observers point out that local authorities’ inefficiency responds to their lack of expert knowledge in managing the state: on how to compromise budgets, allocate resources, and manage their local financial administrative systems (Vammalle et al. 2018). Others, in turn, find in local politicians’ lack of tenure in office the principal factor of their failure (Mejía-Acosta and Haddad 2014). Thus, questions must be asked: Does acquiring expert knowledge lead politicians to deliver better policy results? What role does a politician’s tenure play in their receptiveness and utilization of expert knowledge?

The role of expert knowledge and how it informs politicians’ performance in office has been a largely ignored area of research in public policy and administration scholarships. Traditional accounts of knowledge utilization based on the Weberian ideal (Weber 1978, 220) often suggest that bureaucrats rather than politicians are benefited with “informational asymmetry” concerning the functioning of their state’s administrative procedures (Bawn 1995; Volden 2002). This means that bureaucrats are thought to possess the necessary administrative and technical knowledge to transform electoral mandated goals into effective policies. Advocates of this perspective consider bureaucrats’ expertise posit a fundamental challenge to politicians’ ability to implement their preferred policies and advance their careers (Huber and Shipan 2011; Fiorina 1986). While bureaucrats are more interested in maintaining the status quo, often shirking and/or distorting politicians’ preferred policies; politicians must resort to various strategies to limit bureaucratic shirking: micromanaging, legislating agencies’ processes, and also acquiring off-setting information of agencies’ administrative functions (Waterman, Rouse, and Wright 1998).

Arguably, most of this scholarship is based on observations of Global North countries’ administrative traditions where divisions between political and bureaucratic functions are predominant and normatively preferred (Dahlström and Lapuente 2017, 36–46). However, in the Global South, for instance, in most of Latin American countries, political and bureaucratic functions are not clearly divided, thus politicians tend to see bureaucracies as extensions of their own office (Salazar-Morales 2021). In these settings, politicians benefit from the inexistence of independent and meritocratic civil services, low level horizontal accountability institutions and flawed electoral regulations, which enable them to have enhanced and often unchecked access to manipulate and co-opt bureaucrats’ careers (Huber and McCarty 2004).

Moreover, in these patrimonialist contexts, politicians’ predominance over bureaucracies tend to be powered by electoral incentives (Samuels 2003). This is: while electors condition their votes on politicians’ performance in office, they still struggle to obtain reliable information necessary to substantiate their electoral choices (Bratton, Bhavnani, and Chen 2012; Muñoz 2019). Politicians know this, and utilize their advantage to leverage their reputations as “deliverers” by displaying “technical expertise” (Salazar-Morales 2021)—thereby providing “informational shortcuts” to a disaffected electorate close to electoral rallies (Muñoz 2014). Yet, to do so, in contexts marked by party disaffection and weak accountability mechanisms, politicians often need to acquire off-setting information of their bureaucracies with the purpose of extracting rents which they employ to fund their campaigns—often “buying audiences” to propagandize their performance in office, and signaling their “viability” to other offices in their quest for political survival (Boas, Hidalgo, and Richardson 2014; Boas and Hidalgo 2011). Arguably, as Brinks, Levitsky, and Murillo (2020) and Levitsky (2018) sustain, politicians in Latin America, but specially in Peru, have learnt to follow these pre-established career paths.

Studies analyzing how knowledge is acquired, internalized, and utilized by politicians in such patrimonialist institutional contexts, and/or how said knowledge could, under certain circumstances, lead to positive welfare results, are still scarce. This article contributes to fill this gap by documenting the effects of expert knowledge on politicians’ performance.

To achieve this, I propose that in patrimonialist and weak institutional contexts (Brinks, Levitsky, and Murillo 2020; Geddes 1996), politicians’ use of expert knowledge is mercantile as they employ it to leverage their control of key bureaucratic positions to guarantee their political survival and oust contenders. In doing so, and depending on their tenure, politicians might differently produce some positive welfare outputs. I suggest that tenured politicians are best equipped to use “expert knowledge” to simultaneously advance their careers, while eventually producing better welfare outputs. In contrast, newcomer politicians, given their inexperience in office, pay a differential cost of “learning” because they experience multiple lines of pressure for distributing state rents to aides and electors, which due to their inexperience, they struggle to fulfil. To test my claims, I employ an exogenous assignment of expert knowledge developed by the Peruvian government which trained politicians for about a year in key bureaucratic budgetary functions with the aim of reducing children stunting. Evidence combining a time-varying semiexperimental analysis (Panel DID) and in-depth interviews with regional and local politicians, as well as high level civil servants (N = 29), confirm my theoretical claims.

The remainder of this article is organized as follows. I revise existing literature concerning politicians’ career incentives and their relationship with bureaucrats with a special emphasis on their utilization of “expert knowledge” only to later situate this debate in the “weak” institutional contexts and patrimonialist bureaucracies of the Global South. I then next propose my theoretical framework supported by two testable hypotheses. Then I explain the research strategy adopted starting by a quantitative phase, and then a qualitative one. Finally, I present my findings and conclude by discussing the contributions of my study.

Scholarship on Politicians and Bureaucrats Relationship, Their Career Incentives, and Expert Knowledge Utilization

To respond this article’s motivating questions (concerning the effects of expert knowledge on politicians’ performance in office and how it is mediated by their tenure), I start by revisiting two key literature streams exploring politicians’ motivations to acquire off-setting information about the functioning of their bureaucracies, and how this process contributes to their careers’ progression. In doing so, my review also situates politicians’ motivations amid the weak institutional contexts of Global South countries.

Politicians’ Informational Advantage Over Bureaucrats

Informational asymmetry is a critical assumption sustaining “agency theory,” which classical authors have largely employed to comprehend politicians and bureaucrats relationship (Huber and Shipan 2011). The “bedrock” sustaining this theory is that while politicians’ (principals) central motivation is redistributing resources via policies to their constituents in their quest for securing office, bureaucrats (agents) rather prefer maintaining the status-quo and maximizing their institutional budgets (Volden 2002). While this relationship might seem hierarchical and unproblematic, it is marked by a fundamental tension: bureaucrats are thought to be better informed about the procedures, regulations, and ordinances of their institutions, hence possessing a type of “expert knowledge” (information and expertise) which enables them to “manipulate policy outputs”—either producing them at a higher cost, or not producing them at all (Waterman, Rouse, and Wright 1998).

Scholars hence argue that such informational asymmetry place politicians in a trade-off between delegating complex policymaking to expert bureaucrats (who might presumably find “best ways” to produce sound policies) and running the risk of bureaucratic shirking (Huber and McCarty 2004). Consequently, to minimize bureaucratic drift, politicians tend to utilize various oversight tools to police the behavior of agents. They do so either by closely overseeing bureaucratic performance via monitoring the implementation of their preferred policies (Waterman, Rouse, and Wright 1998), by establishing systems of sanctions and rewards for compliant agents (Huber, Shipan, and Pfahler 2001), and/or by acquiring offsetting information about administrative processes which could enable them to actively monitor and lead the bureaucracy toward their preferred goals (Bawn 1995; Fiorina 1986).

In practice, there is an abundance of old (Bawn 1995; Mitnick 1975) and new (Lee 2020; Thatcher 2002) scholarship documenting the abovementioned strategies employed by politicians in safeguarding that their delegated preferences become effective policy outputs specially in Global North countries. Yet, as many other scholars have noted (Brierley 2020; Kappe and Schuster 2021), this conceptualization of politicians and bureaucrats relationship is rooted on the institutional legacies of developed democracies which assume high levels of horizontal accountability, solid civil career systems and strong electoral institutions that keep these actors’ careers incentives as separate—each of them interplaying within clearly different and delimited roles and goals that hardly change over time (Weber 1978, 941–56). Arguably, this situation is different in countries of the Global South (Peru in our case), where political and bureaucratic careers tend to overlap (Dahlström and Lapuente 2017, 36–53). In those settings, bureaucrats tend to be subject of high-powered political incentives because their access to the civil service corpus is neither mediated by closed and institutionalized recruitment systems nor held accountable by a solid judiciary, but rather by the will of principals who exhibit increased leverage over bureaucratic institutions (Salazar-Morales and Lauriano 2020). As Brierley (2020) notes, in Global South bureaucracies, politicians tend to exert informal control over bureaucrats’ careers by resorting to oversight tools to co-opt them and align them with their preferences—sometimes even pressuring them to engage in illicit practices.

Under such conditions, bureaucrats’ cost of drifting is substantially high, hence they rather opt for aligning with politicians’ preferences—acting as their personal staff (Brierley 2020). Scholars suggest that such patrimonialist institutional configuration reduces, in theory, bureaucratic shirking, while benefiting politicians with informational asymmetry (Brierley 2021; Waterman, Rouse, and Wright 1998). Moreover, this process is believed to enhance politicians’ capacities to extract rents from their state apparatuses and distribute them among aides (Grindle 2012, 141).

But also, a direct consequence of this patrimonialist institutional entrenchment is that citizens become aware that politicians dominate any relationship with the bureaucracy (which is perceived to be generally inefficient and colluded), therefore expecting less from bureaucratic institutions, and focusing more on whether politicians’ display any signal or provide “informational insights” that demonstrate their capacity to win office and/or produce better welfare outputs (Bratton, Bhavnani, and Chen 2012). This has electoral consequences as politicians further develop detailed strategies on how and when they will utilize their informational advantage to extract rents, fund their campaigns, mobilize “their agents” and eventually redistribute resources seeking to bolster their reputations via providing citizens with “informational shortcuts” of their performance in office. Yet, as Salazar-Morales (2021) argues, politicians’ tenure also crucially mediates their performance—being more experienced politicians well equipped in utilizing available “expert knowledge” for advancing their careers—as I explain next.

Politicians’ Acquisition of “Expert Knowledge” and Career Progression

The pinnacle of political career’s literature is that politicians’ primordial interest is the retention or obtention of office positions (Schlesinger 1966). They can do so by either advancing their careers horizontally (moving in and out similar offices) and/or vertically (using offices to escalate to higher levels) (Samuels 2003). Whether they chose one, or a combination of these options, depends on politicians’ individual ambitions, but most importantly, on the structure of opportunities of the political system—meaning the strength or weakness of the political institutions, the structure of representation, and the nature of the political organizations in which politicians compete (Borchert 2011).

So far, comparative research on politicians’ careers has noted that the combination of differing structural and institutional factors has led them to pursue varied career tracks. In Canada, for instance, politicians prefer moving “vertically” to executive positions—being thus common that they renounce to their parliamentary seats to tempt decision-making roles in higher offices (Docherty 1997). On the contrary, in Brazil, Samuels (2003) notes, politicians prefer to move “horizontally” among lower municipal offices—hence using their positions to forge alliances with powerful bureaucrats and with other key subnational actors, with the objective of gaining future supporters for their candidacies at the local level.

Common to the abovementioned tracks is however that politicians’ careers progression involves a learning process of preestablished “successful” career paths. This means that across time, politicians acquire relevant knowledge about the informal and formal rules of the institutions (specific office characteristics) and electoral arenas shaping the structure of opportunities that make certain offices available, accessible, and ultimately attractive (Borchert 2011). Arguably, politicians, by gradually acquiring knowledge about the structure of opportunities in which they compete, can preclude successful trajectories, and anticipate which career plans or stablished pathways might be good for their own ambitions.

As for our case situated in the Global South, recent research has started to conceptualize the incentives proper of weak institutional environments in which politicians build their careers. For instance, Brinks, Levitsky, and Murillo (2020) stress that incentives proper of the “endemic” weak institutions of the Global South operate by narrowing politicians’ time horizons drawing them into a political game where, in order to advance their careers, they must learn to strategically dismount rules and procedures “rapidly and wholesale” in a way that undermines democratic stability and economic performance. The authors argue that in following short-term calculations, politicians in these contexts, are bounded by a conjunction of formal and informal practices that render them unable to “redistribute [nor] refract power, authority or expectations in order to produce [effective] institutional outcomes” (Brinks, Levitsky, and Murillo 2020, 8). On the contrary, they tend to be focused on actively gaining access to state resources to guarantee their own survival, in some cases strategically promoting policy forbearance (Holland 2016) by dismantling structures of enforcement and compliance—and/or by actively sabotaging competitors in case they lose office (Lazar 2004).

While of course these are pathways they have learnt to follow, politicians exhibit different degrees of embeddedness and knowledge of the political and bureaucratic apparatus where they compete (Samuels 2003). As recent research shows (Lazar 2004; Salazar-Morales 2021), while some politicians are already tenured and knowledgeable of the availability, accessibility, and overall structural opportunities enabling them better accesses to office—thus rendering them capable of articulating and mobilizing available “expert knowledge” for their own career’s benefits—, others (newer politicians) still pay the differential costs of learning. Arguably, as Borchert (2011) highlights, politicians’ ambitions are shaped by their circumstances, but also by how they anticipate (how much they know of) the pathways beneficial to their own plans.

The Predominance of Patrimonialist Logics in Global South Contexts

Arguably, the institutional context where politicians develop their careers contributes to define their short- and long-term political objectives, cements their electoral decisions and shape their incentives when they gain office. Often, in Global South countries, politicians develop their careers in bureaucracies marked by historical legacies of colonialism, authoritarianism, exclusion/inclusion practices based on bonds of kindship (Brierley 2021; Geddes 1996; Schuster 2016). Scholars have labelled this type of bureaucracies as patrimonialist because they are often characterized by the unchecked and arbitrary influence that politicians (principals) can exert on them in the recruitment and promotion of public employees (Driscoll 2018).

This is confirmed by extensive literature on African and Latin American¹ bureaucracies which recognize them as marked by the widespread presence of patrimonialist recruitment and promotion practices—even in spite of the presence of strict regulations forbidding such appointments (Brierley 2020; Brinks, Levitsky, and Murillo 2020; Driscoll 2018; Geddes 1996; Kappe and Schuster 2021). Such characterization has led scholars to sustain that patrimonialism should be conceptualized as an “structural characteristic” of developing countries’ bureaucracies (rather than as an “outcome” of failed “modernization processes”) (Schuster 2016), and that they must be taken into consideration when studying politicians and bureaucrats’ relationships in these geographic (Global South) settings (Brierley 2020; Driscoll 2018; Geddes 1996).

Under such patrimonialist conditions, to ensure their control over governmental agencies, principals tend to create “pyramids of informal contracts” cemented on relationships of kindship, party allegiance or ideological closeness where they trade offices for loyalties (Geddes 1996; Pereira 2016). These “pyramids” are highly unstable as they rapidly collapse upon the replacement of the incumbent in office consequently weaking a country’s bureaucratic capacity by shortening the horizon of politicians and bureaucrats (Brinks, Levitsky, and Murillo 2020)—who knowing their stay in government is short-term turn onto focusing exclusively on rent extraction also triggering a downward fragmentation of the public service (Brierley 2021), with the consequence of institutional memory loss, discontinuation of policies and projects (Balarin 2008), and the overall dismantlement of state’s infrastructural capacities (Geddes 1996).

Arguably, such patrimonialist hiring and promotion logics have crystallized and adopted new operative forms under competitive electoral conditions (Driscoll 2018). Recent scholars have started to move away from the dichotomic characterization of Global South bureaucracies as either patrimonial or meritocratic, and started to rather recognize that most countries lie in a continuum between “complete meritocracy and partisan interference” (Brierley 2021). This means that it is common to observe developing countries with bureaucratic sectors exhibiting higher autonomy, meritocratic and transparent recruitment processes (labelled as “isles of efficiency”) (Roll 2014), while other sectors showcase high levels of political interference. In this context, as Brierley (2021) has recently shown, principals tend to develop specialized strategies to maximize rent extraction, minimize risk and foster their career development by carefully appointing skilled bureaucrats to high technical positions in government (usually those managing the economy, regulatory sectors, budgetary sectors), while distributing low-ranking public-sector jobs (entailing a lower failure risk) to brokers with the purpose of sustaining party machines. Politicians in these contexts, Brierley sustains, can have the best of two worlds (technocrats and brokers).

Peru is representative of such type of bureaucracy. Scholarly literature typically refers to the country’s public service as marked by legacies of colonialism, racial divide and dominated by closed patrimonial networks (known as “argollas”) (Salazar-Morales and Lauriano 2020). There is an academic consensus referring the country’s institutions are “deeply ingrained by patrimonialism and clientelism” (Crabtree 2010, 371), usually regarded one of the “weakest of the continent” (Iacovello 2015), constructed based on “patrimonial pacts,” and completely dependent on the informal arrangements between “bureaucrats and politicians” (Balarin 2008; Salazar-Morales and Lauriano 2020).

Moreover, in line with recent theoretical developments, scholars such as Dargent (2015) catalogued Peru’s bureaucracy as divided between “isles of efficiency” (economic-oriented and regulatory agencies) characterized by their “technocratic autonomy”, and others generally socially oriented agencies (education, health, security and, the object of this article, subnational governments) which are still dominated by “patrimonialist” recruitment and promotion logics. This is further supported by other scholars who have noted that Peru’s National Civil Service Authority (SERVIR)—which sought to institute meritocratic hiring and promotion practices for public servants in middle and lower ranks—was only implemented at the request of the United States (during the negotiations of a Free Trade Agreement with Peru) in 2008 (Iacovello 2015). Yet in practice, SERVIR has remained dormant over years until 2018 (10 years after its creation) that it has barely managed to run impartial, meritocratic, unbiased recruitment of public servants in 391 state agencies (13%) out of the total 2,900 existing (Mendoza, Laura, and Coronado 2018, 104). Those agencies in which SERVIR reform succeeded are entities typically considered as “technocratic” (regulatory and tax collection agencies), although as Mendoza, Laura, and Coronado (2018) report, other “technocratic” agencies including the Ministry of Finance are still strongly opposing SERVIR reforms and have designed their own recruitment and promotion procedures—thus relegating SERVIR’s authority. As per the agencies this article researches, they are located among the 87% which have not yet been included into SERVIR’s meritocratic recruitment and promotion scheme (SERVIR 2015).

In sum, having revised the centrality of politicians and “narrow” logics governing their careers’ construction in weak and patrimonialist bureaucracies with a special characterization of the Peruvian case, I move on to proposing a novel theory on how knowledge is acquired, appropriated, and utilized by politicians for advancing their careers and, eventually, producing welfare outputs in the abovementioned institutional contexts.

Theory: The Mercantile Uses of Expert Knowledge in Patrimonialist Bureaucracies

Considering the abovementioned discussion, I propose that in weak and patrimonialist institutional contexts where politicians have predominance over bureaucracies, they acquire and utilize expert knowledge with mercantile purposes.² This is: Politicians resort to expert knowledge to obtain “ideas, concepts and information” (Weiss and Gruber 1984) on how to leverage their degree of influence over their bureaucracies, hence gaining control over key administrative procedures such as budget allocation, hiring personnel, and public procurement processes.

The logic is that in acquiring off-setting knowledge of those complex administrative procedures, politicians are more capable to incur into “contract clientelism” (Holland and Freeman 2021) through which they will provide handouts prior elections to powerful local economic actors (who could later make donations to their political campaigns) (Boas, Hidalgo, and Richardson 2014), favorable local media and/or other organizations (Boas and Hidalgo 2011), with the objective of building a supportive network of aides for re-election purposes (Brierley 2020). These actors, in a context like Peru (marked by extreme party fragmentation, lack of stable political organizations, absent long-term political affiliations and short-term careers’ prospects, see Levitsky 2018; Zavaleta 2014), rather than securing direct votes for politicians, disseminate to the electorate “informational value” of the incumbent hence signaling his/her viability during campaign by aggrandizing/exaggerating his/her performance in office (Muñoz 2014, 2019).

It is thus proposed that politicians underlying motivation to acquire “expert knowledge” is to fund circumstantial substitute partisan structures (Zavaleta 2014) that contribute to bolster their reputations and provide informational shortcuts to various “audiences” for re-election purposes (when possible), while signaling their electoral viability to other offices (Muñoz 2014). This is possible because, as scholarship on politicians’ careers and incentives in Peru largely contend, Peruvian politicians have an overall preference for building careers outside parties (hence acting as partisan free agents) because it guarantees that they could construe time-and-again “independent coalitions” with “new faces” and supporters for maintaining or tempting offices (Levitsky 2018; Zavaleta 2014). Said coalitions wok as party substitutes which politicians employ to advance their careers by associating their images to local elites and/or extra political reputed institutions, among them: well-known local businesses, football teams, local private educative institutions, local-media firms, among many others (Barrenechea 2014; Boas, Hidalgo, and Richardson 2014; Zavaleta 2014).

Arguably, these well-documented informal rules and incentives are not only acknowledged by politicians, but also, as Zavaleta (2014), Levitsky (2018), and Muñoz (2019) highlight, by citizens who due to constant rise and fall of political labels and lack of politicians’ consistent party affiliations, suffer from increased informational deficits to evaluate politicians’ trajectories (because they have short political lives and/or constantly create new political organizations) therefore shifting their vote choice judgments toward politicians’ perceived electoral chances and/or potential performance in office rather than assessing their more substantial proposals. To avoid wasting their votes, therefore, citizens “concentrate on” politicians who demonstrate ability to “fare and deliver better,” rather than focusing on those “hopeless” ones (Muñoz 2014, 84–5).

Under these circumstances, acquiring expert knowledge about the cogs and wheels of their bureaucratic apparatuses becomes central for ambitious politicians willing to move their careers forward because it enables them to redirect state rents to construe party substitutes for electoral purposes. Yet, undoubtedly, extracting rents to support campaigns is a dangerous process: It must not raise red flags among the electorate, national media, or principled high-level bureaucrats, which might activate accountability mechanisms that could irreparably harm a local politician’s reputation. Hence politicians are likely to be concerned with balancing poor state performance and rent extraction, because even in patrimonialist polities “voters do consider economic management and public service delivery” (Brierley 2021, 185). This means that citizens pay attention to informational shortcuts that propagandize “small and rapid” marginal welfare gains in their localities, most notably in areas such as local infrastructure, roads, schools, and health clinics (Brierley 2021). Arguably, politicians at different career stages might respond differently to the pressures of providing cues about their viability and performance in office, I formalize my expectations in the following two hypotheses.

H1: In patrimonalist contexts where politicians are predominant over bureaucrats, those who acquire expert knowledge about key administrative sectors of their bureaucracies will best extract rents while eventually producing better welfare policy effects during their time in office.

In patrimonialist settings such as in Peru’s subnational governments, given politicians’ unchecked control over bureaucrats’ careers, in order to effectively obtain welfare outputs that they can instrumentalize to advance their careers, politicians need to become familiar with the informal and formal incentives of their political system. This implies that they must learn how to survive their short-lived political organizations while appealing to a disaffected electorate. I hence expect that by acquiring expert knowledge on the cogs and wheels of their bureaucracies—in special budgetary functions—politicians will have enhanced capacities to extract rents and profit electorally from them, meanwhile, indirectly benefiting constituents with some positive welfare outcomes. Of course, this hypothesis does not imply that politicians’ principal motivation is to deliver policy, but rather it suggests that politicians could double benefit if actively acquiring expert knowledge to advance their own aims, rather than not mobilizing their bureaucracies at all.

H2: In patrimonialist contexts where politicians are predominant over bureaucracies, the effects of expert knowledge in office are augmented by politicians’ experience: experienced politicians (insiders), in contrast to newcomers (outsiders), will better extract rents, produce comparatively better welfare policy effects during their time in office while also propagandizing their office performance or administrative sabotaging competitors.

In settings such as Peru’s subnational governments where bureaucrats are hired and dismissed at the will of the incumbent, and where bureaucracies lack institutional memory, present overwhelmingly weak capacities, and are often structured around pyramids of loyalties, I contend that performance depends on politicians’ experience. This is: newcomer politicians, given their inexperience in office, might struggle to rapidly take control of key positions of their bureaucracies. They might be more prone of making administrative mistakes (raising red flags that might take months to unravel) hence struggling not only to extract and distribute rents among aides but also to counter the administrative “sabotages” of their predecessors (Lazar 2004). In contrast, I expect seasoned politicians to be more “acquainted” about the administrative intricacies of their bureaucracies, therefore knowing when to extract rents, to whom allocate them to construe their circumstantial supportive coalitions, thereby bolstering their reputations, and simultaneously providing crucial “informational value” to their multiple electoral audiences (Salazar-Morales 2021).

In sum, it is my expectation that seasoned politicians might act as “mercantilists”: They will know better when to produce outputs, but only if they are beneficial to their own careers’ ambitions (Schlesinger 1966). Therefore, in line with Lazar (2004), Salazar-Morales (2021), and Levitsky (2018), I also expect seasoned politicians to use their “knowledge advantage” to administratively sabotage competitors by embezzling budgets and incurring in collusion—being more willing to dismount democratic and administrative rules and procedures “rapidly and wholesale” if they are not to their benefit.

Having exposed the theoretical formulation of this article, I next explain my research strategy, the case employed to test my hypotheses, only to then discuss results obtained.

Research Strategy

Following Saetren’s (2005) recommendation that public administration studies require a “total research approach,” this study relies on a comprehensive, pragmatic, and sequential research strategy (QUAN-QUAL) integrating time-varying semiexperimental analysis (Panel DID) and in-depth interviews with regional and local politicians, and other high level civil servants (n = 29). I resort to a sequential strategy because I employ qualitative interviews to corroborate and complement my statistical findings. While my semi-experimental analysis informs about the significance of expert knowledge—as mediated by politicians’ tenure—in the production of better policy outputs, in turn, my interviews reveal unseen rationalities and mechanisms underlying my statistical findings. Having this in mind, I next explain the case chosen, and how I have approached to it.

Case Study: Knowledge Transfer Program to Fight Children Stunting in Peru

Reducing stunting among children has been at the core of the country’s policy priorities for decades (Marini, Rokx, and Gallagher 2017). Over the years the central government effectively garnered political consensus to implement coordinated measures to curb the high ratios of stunting among children (Mejía-Acosta and Haddad 2014). In doing so, they have set national and subnational “stunting reduction” goals, refined their indicators and measurement systems in line with international standards, as well as implemented budgetary incentives for local governments, so they could contribute to fight children stunting in their circumscriptions (Marini, Rokx, and Gallagher 2017). Despite the apparent national consensus on the relevance and need for prioritizing the “stunting reduction program,” the central government yearly reports hundreds of cases of local mayors missing policy goals, unspent budgets and/or their utilization for purposes other than children stunting reduction—as noted in this article’s introduction.

In practice, the Ministry of Finance of Peru (MEF) contends that one of the biggest sources (outside corruption) of policy failure and budgetary mismanagement is local governments’ inefficient budget allocation and expenditure (Gestion 2019). In the latest 10 years, as figure 1 shows, local governments have returned to the MEF nearly 40% of their budgets due to local politicians’ generalized ignorance on how to allocate it through the country’s financial administrative systems (known as SIAF), and/or because they fail to transform their budget into effective policy action (leading procurement processes of health materials, hiring of personnel, supplying local medical posts, promoting children nutrition initiatives locally, etc.) as part of the national stunting fight program.

Figure 1.

Local Governments Yearly Budgetary Performance.

Source:MEF (2021).

The Knowledge Transfer Program

Aware of the local governments’ capacity limitations, in 2014 the MEF, along with the World Bank, developed a knowledge transfer program (KTP) consisting of technical assistance directed at improving local governments’ spending capacities (Marini, Rokx, and Gallagher 2017). The program worked as a “training school” for politicians who received “expert knowledge” about the Peruvian state administrative systems involved in the programming and allocation of budgets (MEF 2017, 11). According to the MEF (2017, 30–5), the program sought to inform local politicians about their legal functions and roles in the local government and taught them how to “coordinate” with local actors, “articulate” the existing social programs of their districts while focusing on the enhancement of their policy outputs, concretely, on reducing children stunting ration. Politicians were taught how to interpret social indicators, such as ratios, percentages, and tendencies—they also received lectures about the long-term detrimental effects of stunting in children (MEF 2016). The goal was to make politicians aware of the problem in their hands, incentivizing them to act.

The program was applied to 197 districts located in three Peru’s regions: Amazonas (Jungle), Cajarmarca (Highlands), and Huanuco (Central Peru). Those regions were selected because MEF officials wanted to ensure the functioning of the program across Peru’s varied cultural and geographical traits, but most notably because MEF officials already had installed capacities and a successful history of working effectively with regional governors and other politicians in those areas (MEF 2017). In those regions, MEF functionaries invited mayors of districts who belonged to the poorest 10%, which at the same time registered more than 600 children under 5 years old. The MEF reports that 160 mayors enrolled in the program, while other 37 came from MEF’s extensive invitations in other districts (not necessarily poor nor having 600 children under five) (MEF 2017). The central government strategy consisted of assigning to each selected district a high-level bureaucrat of the MEF who provided technical assistance, training, and technical support to a local politician and his team intermittently (every month) for about a year (MEF 2016). The program started in January 2014 and expanded over 12 months—no further systematic analysis of its effectiveness has been so far conducted.

Quantitative Stage

To respond this article hypotheses, a robust and gradual analysis is necessary. Thus, I initially use a descriptive OLS analysis, then I gradually increase the complexity of models by including fixed effects and interactions. This permits to efficiently observe marginal changes in the variables of interest. Finally, I will use the differences in differences estimator (DID) in a time varying data set (known as panel DID). This is our preferred model. Unlike canonical DID models which analyze variation across groups (treatment/effect) at different times (pre/post treatment), my preferred model captures the effects of my treatment variable over time (Cerulli and Ventura 2019), allowing us to estimate anticipatory and posterior treatment effects among the threated (ATT) on whether the expert knowledge transferred to politicians led them to better extract rents, while eventually producing better positive welfare consequences, as follows:

$${\displaystyle \begin{array}{l}{\displaystyle Y_{it}=\alpha +{\beta }_{-1}{D}_{it-1}+{\beta }_0{D}_{it}+{\beta }_{+1}{D}_{it+1}+\pi E_{it}+\gamma X_{it}^{\prime }+{\theta }_i+{\epsilon }_{it}}\end{array}}$$

The dependent variable Y is the ration of stunting in children under five. This is the leading indicator officially produced by National Health Institute of Peru. This variable is constructed based on the monthly measures run by the Ministry of Health’s posts located in Peru’s districts which compare the weight/height ration of a kid with the global average set by the Panamerican Health Organization (PHO). A kid is considered stunted if his/her ration marks more than 2σ from the mean. I employ this measure as a proxy of the potential welfare effects produced by the acquisition of “expert knowledge” by politicians who participated in the KTP program.

D is our treatment variable, and it codes 1 if a politician (the mayor) in each district has received the program (KTP), and 0, if otherwise. Following Cerulli and Ventura (2019, 552), I also estimate the effects of D a period before (lag) and after (lead) the year of the treatment to assess whether it presents only contemporaneous effects (during its application), or whether it also exhibits anticipatory of future effects.³ The advantage of Cerulli and Vertura’s estimation method is that it enables us to observe whether certain treatment had an impact on a given target variable with some delay.

Another crucial variable our model assesses is political experience (E), which is a dataset specially built for this article. It codes the years of experience in public office that a local politician (mayor of a district) has accumulated during his/her career, as declared in his/her vitae. This data comes from the National Jury of Elections of Peru. We use the variable experience in two forms: (1) continuous, to assess, in line with our first hypothesis, whether politicians marginally benefit from acquiring expert knowledge; and (2) categorial, where in following Carreras (2012), I classify politicians into “outsiders in office” (those with one or less than one period in office, <4 years), and “insiders” (for those who attain more than one period in office ≥4 years). This categorial division permits us to assess our second hypothesis exploring whether experienced politicians, in advancing their careers, circumstantially produce better welfare policy outcomes during their time in office. Important to mention is that this variable follows the division set by the Peruvian electoral Law which argues that local mayors’ office periods are restricted to 4 years with the possibility of re-election.

The vector Xʹ contains our set of confounding variables. I include other variables that might potentially render insignificant the effects of the treatment. To capture the potential effects of the political context on our dependent variable, I consider testing for the percentage of votes won by the party in office. I use this variable because the Peruvian Electoral Law grants an automatic majority in the regional council to winners of electoral ballots. This variable should inform, in line with existing scholarship, on whether politicians with few electoral supports, once in office, might experience more pressure either coming from losing parties and/or the organized society—which in some cases might lead to recall referendums (Holland and Incio 2019). Moreover, I also control for the type of political organization of politicians in office. This variable codes whether a politician is affiliated to an electoral alliance, regional movement, a political party, or if he runs with his own local organization. The logic is that politicians might be subject to different intraorganizational (e.g., party discipline) incentives motivating them to use the “knowledge transferred” to enhance their office outputs.

In addition, I also incorporate other important social variables. Considering that existing research contents that smaller bureaucracies might exhibit better policy performance, I include the variable “population size” in my model (Jugl 2019). Observers argue that local politicians’ social proximity to citizens might lead them to be more responsive to their constituents’ demands (Jugl 2019). As well, I also include in my model the ration of the population living in poverty. Previous research on stunting fight have found sound evidence signaling that poor populations tend to suffer either from caloric insufficiencies or diarrheic diseases leading to malnourishment (Mejía-Acosta and Haddad 2014). Henceforth the need to include this variable in my model as a key structural confounder.

Finally, in my model, the variable θ_i is a fixed effect per district, while ε_it is the error term. In total my dataset is composed of observations covering 86% of the Peru’s total districts (1,615 of 1,874). My analysis starts in 2006, when the central government started measuring children stunting in line with the technical guidelines of the PHO (INS 2021). Total observations amount to N = 20,995 (T = 13, per each n: 1,615). However, some of my vectors present missing observations (poverty ratio and population size), which provoke drops when running my models. To overcome this limitation, I have employed the Amelia II (Honaker, King, and Blackwell 2011) software, which through an iterative process involving a Multiple Imputation by Chained Equations (MICE) estimates missing values by regressing them based on existing observation of the entire dataset. Unlike other imputation strategies, this method presents the advantage of resampling my entire data set while safeguarding its level distributional structure (Azur et al. 2011). Outcomes coming from this process are employed both in my OLS and semiexperimental analysis. Table 1 presents a close overview to the data employed.

Table 1.

Summary of Variables

Variables	Mean	SD	Min	Max
Ration of stunting in children under 5	27.37	14.32	0.008	100
Political experience	9.225	6.289	0	47
% of poor population (MICE)	45.43	21.24	0.06	99.72
Political organization (ordinal)	2.758	1.050	1	4
Log of the population (MICE)	8.457	1.380	4.095	14

Qualitative Stage

This second stage aims at complementing results obtained from the previous qualitative section with interviews with various actors involved in the KTP. A first group of 20 interviews has been conducted during my fieldwork in Peru between June and September of 2018. Interviewees included regional and local politicians, and higher public officials of the Peruvian government of various institutions (principally the Ministry of Finance and Ministry of Health). A second group of nine of interviews with personnel directly working on the implementation of the KTP was conducted via video calls intermittently between 2019 and 2021. These interviews were also complemented with a detailed documentary research involving minutes, reports, and video records produced by the KTP program. An anonymized list of interviewees can be found in supplementary table A-6.

Drawing on the pragmatic research paradigm, I interpret results stemming from this stage based on an “abductive logic.” This involves a systematic integration of evidence with the theoretical framework of my article by considering the particular institutional and social context of my case study: Peru (Richardson and Kramer 2006). This research strategy is beneficial for situating theories to different contexts, and/or creating new ones. The final aim of this subsection is to complement statistical findings with evidence on how old and new politicians see their role in appropriating, internalizing, and employing expert knowledge during their periods in office.

Quantitative Findings

Politicians General Use of Expert Knowledge

Table 2 presents results by considering more general estimation techniques, and my preferred Panel DID models 3 and 4.⁴ Notably, all models show that politicians receiving the (KTP) from the Peruvian central government generate better welfare outputs during their time in office. In fact, my main model 3 shows that politicians receiving the KTP correlate with about 4% reduction of stunting in children in the first year of the program, with subsequent increased performance over the next years. This finding is important because a positive result in year t₀ indicates that those mayors receiving the treatment were rapidly able to include “stunting fight” activities in their annual budget programs (which are known as PIM⁵).

Table 2.

Models’ Outputs

Variables	OLS	Panel	Panel DID	Panel DID
	Model 1	Model 2	Model 3	Model 4
Output: ration of stunting in children under 5
Interventiont−1			−2.54 (1.794) [.16]	−1.6 (1.79) [.37]
Interventiont0	−0.912** (0.356) [.01]	−9.758*** (0.414) [.00]	−3.93** (1.812) [.03]	−2.16 (1.82) [.24]
Interventiont+1			−5.207*** (1.81) [.00]	−2.65 (1.7) [.14]
Interventiont+2			−7.152*** (1.82) [.00]	−3.53** (1.85) [.05]
Interventiont+3			−9.245*** (1.90) [.00]	−4.53*** (1.90) [.01]
Political experience	−0.149*** (0.0166) [.00]	−0.146*** (0.0143) [.00]	−0.0270* (0.014) [.07]	−0.02 (0.14) [.19]
% of Votes	−7.263*** (1.043) [.00]	−4.43*** (1.033) [.00]	−2.62* (1.532) [.08]	−1.91 (1.48) [.19]
Log of the population (MICE)	−1.187*** (0.0762) [.00]	0.003 (0.103) [.97]	0.0704 (0.106) [.5]	−0.023 (0.105) [.8]
% of poor population (MICE)	0.271*** (0.00497) [.00]	0.0021 (0.0038) [.58]	0.0028 (0.00406) [.4]	0.001 (0.004) [.74]
Political organization
1.Alliance	—	—	—	—
2. Regional organization	−1.453*** (0.373) [.00]	−1.923*** (0.337) [.00]	—	—
3.Local organization	−1.284** (0.607) [.03]	2.955*** (0.501) [.00]	—	—
4. Political party	−0.0817 (0.390) [.83]	2.383*** (0.342) [.00]	—	—
Constant	29.69*** (0.949) [.00]	30.32*** (1.069) [.00]	24.07*** (1.187) [.00]	25.63*** (1.164) [.00]
Observations	17,074	17,074	20,686	20,686
R2	0.190	0.083	0.024	0.083
Fixed effects	No	Yes	Yes	Yes
Year dummies	No	No	No	Yes

Note: Exact p-values are provided in brackets and standard deviations in parentheses. In models 3 and 4, Stata16 drops the variable “political organization” due to collinearity.

***p < .01, **p < .05, *p < .1.

Important to mention is that Peru’s Budget Law stipulates that local government budgets are usually approved the last day of year (e.g., a budget for 2014 must be approved in December 2013), and could be modified by a mayor, according to his/her needs and capacities, until March of the current fiscal year (2014 in our example). Hence, in taking advantage of this three months window of opportunity (January to March), KTP managers interviewed mentioned that they pressured mayors since “…[day one] to focus on including the stunting issue in their annual budget programmes,” which they did so by “helping them to draft their PIM’s while channelling funds from third parties…” (Int. 25). For this reason, some positive welfare outputs can already be seen in t₀ as reported by the INS stunting indicator for 2014.

But there are also other reasons underlying the results shown by the KTP: One of my interviewees noted that “stunting is multicausal and thus could be addressed from various angles” (Int. 19). In some cases, my interviewee, a former MINSA officer now at the World Bank, mentions, “stunting could be caused by the presence of rotaviruses which lead to severe diarrhea; in others, it is produced by insufficient caloric intakes…” (Int. 19). Hence if a major acts quickly, s/he could manage to allocate funds to “hire more nurses, run anti-rotavirus vaccination campaigns, or rapidly acquire and distribute caloric supplements like iron sulphate, therefore having an effective […and rapid] impact on children’s” nourishment (Int. 19, similar sentiment found in 26).⁶

Furthermore, while models 1 to 3 suggest that the KTP effectively obtained results in t₀, when controlling for national trends (via year dummies), I noticed that politicians produce better welfare outcomes over time and close to the electoral years (t₊₂; t₊₃). In practice, my second main model 4 shows that 3 years after the implementation of the KTP, in 2018 (an electoral year), those who received the program (compared to those who did not) tend to reduce stunting at a significant level in about 4.3%—a bit more than the double as when the program started. These results are further supported in a subsequent inspection of the parallel trend assumptions of model 4 presented in supplementary figure A-1 and in its placebo test visible in supplementary figure A-2 both in the supplementary appendix of this article (a more detailed overview of the coefficients of the placebo test for model 4 can also be found in supplementary table A-3). Of course, one way to read these results is that they are cumulative effects of the KTP, however, as exposed in my theoretical section, and considering the stablished career paths of Peruvian politicians (Barrenechea 2014; Zavaleta 2014), I sustain that my results might illustrate politicians’ use of transferred knowledge to strategically showcase effectiveness close to electoral years.

In fact, some of my interviewees support this claim (see R4 in table 5 below) by stressing that treated politicians have organized the delivery of their health outcomes toward the end of their terms by presenting their results in official events ‘thus congregating local and social leaders, like the “neighbors committees,” “loyal media,” among other aides (Int. 24 and also Int. 28). In doing so, of course they intend to retain office, but given the degree of electoral disaffection and lack of party structures in Peru (Muñoz 2019), most local mayors want to use their municipal rents while minimizing risks, hence acting, one interviewee notes, as “passing birds (aves de paso) […]: always looking for higher offices in provinces or regional governments” (Int. 21) or tempting the waters of national politics aiming at “well-paid congressional seats”⁷ (Int. 19).

Consequently, in flagging their policy outputs close to elections, politicians build their personalistic political brands with the objective of, Muñoz (2014) argues, keeping a degree of mediatic presence among the electorate signaling that they are still a “viable option” by constantly earmarking their way to other offices. Overall, evidence stemming from models 1 to 4 seem to confirm in part my first hypothesis contending that politicians acquiring expert knowledge tend produce more positive welfare outputs. Yet, this result cannot be inferred alone from my quantitative evidence: in practice, my interviewees revealed that politicians’ performance respond to the electoral pressures to build personalistic reputational brands to indicate that they are still able to compete, if not for re-election, for other offices. While these results reflect more generally that expert knowledge effectively improves policy performance, it is still important for this article to analyze how politicians at different stages of their careers (either newcomers or seasoned) utilize and mobilize expert knowledge—I explain next.

Politicians’ Tenure and Expert Knowledge

Beyond the general results shown above, I also present a more refined analysis testing whether politicians’ tenure (either newcomer or experienced) might have effects on how they employ expert knowledge during their time in office. To do so, I have recoded, following Carreras (2012), the variable political experience (E) as a dummy whereby “outsider politicians” (those with one or less than one period in office, <4) are noted as 0, while “insider ones” (for those who present more than one period in office, ≥4) coded as 1.

I present my results in two steps. Firstly, in table 3, I assess how tenure moderates the effect of the “expert knowledge” received by politicians via the KTP. Through controlling for fixed effects and national trends (year dummies), model 6 shows that insider mayors who have received the treatment (compared to those who did not) benefit more from expert knowledge: They effectively produce better welfare outputs manifested in a reduction of stunting among children of about −2.9% in t₊₁ (albeit at the p < .05 level), which gradually improves in significance and size until −3.75% by year t₊₃ (at the p < .01 level)_. Conversely, model 5 demonstrates that outsider politicians fail to take advantage of the KTP, hence being unable to significantly use it to produce better welfare outputs. At first sight, this indicates that outsiders’ performance in office after receiving treatment is no different from their peers who did not participate in the program.

Table 3.

Semiexperimental Segmented Analysis (Panel DID)

Output: % Stunting in Children under 5	(Outsiders)	(Insiders)
	Model 5	Model 6
Interventiont−1	−6.24 (9.03) [.49]	−1.27 (1.32) [.21]
Interventiont0	−7.36 (13.70) [.59]	−1.67 (1.3) [.20]
Interventiont+1	−1.84 (4.84) [.70]	−2.90** (1.38) [.05]
Interventiont+2	−13.15 (8.24) [.11]	−3.14*** (1.2) [.05]
Interventiont+3	—	−3.75*** (1.49) [.01]
% of Votes	−0.448 (23.0) [.98]	−0.97 (2.69) [.71]
Log of the population (MICE)	0.583 (1.17) [.62]	−0.29* (0.16) [.07]
% of poor population (MICE)	0.209*** (0.074) [.00]	0.0001 (0.005) [.97]
Constant	8.99 (13.86) [.52]	26.28*** (1.64) [.00]
Observations	5,195	15,800
R2	0.167	0.12
Fixed effects	Yes	Yes
Year dummies	Yes	Yes

Note: Exact p-values are provided in brackets and standard deviations in parentheses. Stata 16 drops t+3 model 5 due to collinearity, and similar to models 3 and 4, also the variable “political organization.”

***p < .01, **p < .05, *p < .1.

Overall, results stemming from table 3 confirm in part my hypothesis 2, contending that seasoned politicians are more sensitive in appropriating expert knowledge. This outcome further confirms what is generally shown in previous models (3 and 4, especially in model 6), which alongside the qualitative evidence stemming from interviews (table 5), led me to sustain that for insider politicians, receiving training only strengthens their scope of control over the local bureaucracies with the aim of securing their political survival—which is why their performance (measured by stunting reduction) almost double when closer to electoral periods (in year t₃). These results are further supported by complementary placebo tests shown in supplementary figures A-3 and A-4 which visually demonstrate the soundness of the parallel trend assumption supporting my DID estimations (also see supplementary table A-4 for a detailed overview of coefficients obtained in running the placebo tests for models 5 and 6).

To conclude with the quantitative evidence, table 4 further complements my previous analysis via naïve interactions avoiding the controlled scenarios of my previous semiexperimental models. Beyond the confirmation that in general treated mayors produce better welfare results in office than untreated ones, model 7 most salient outcome confirms that “expert knowledge” is indeed moderated by politicians’ tenure.

Table 4.

Simple Panel Interaction Analysis (Panel Fixed Effects)

Model 7
Output: % Stunting in Children under 5	No Treatment	Treatment
Outsiders	—	−3.9*** (.66) [.00]
Insiders	−0.922*** (.31) [.00]	−4.39*** (.439) [.00]
Constant	24.38*** (1.37) [.00]
Observations	17,074
R2	0.033
Fixed effects	Yes
Year dummies	Yes

Note: Exact p-values provided in brackets and standard deviations in parentheses.

***p < .01, **p < .05, *p < .1.

For the group of untreated politicians, results indicate that insiders do better in curbing stunting in their districts in about 1%—compared to their outsider counterparts. This result, in line with existing scholarship (Carreras 2012; Salazar-Morales 2021) and interviews carried out (see R3 in table 5 below) indicates that tenure endows insiders with a degree of familiarity with their bureaucracies hence being able to gather support from free-agents (bureaucrats) willing to partake in his/her government (Brierley 2021), effectively countering the administrative sabotage of their predecessors, and rapidly engaging into “contract clientelism” (Holland and Freeman 2021) with key actors that will support their permanence in office (against recall initiatives) and/or in future electoral rounds.

Table 5.

Summary of Qualitative Findings

		Received Treatment?
		No	Yes
Tenure	Outsiders	R1. Less tenured politicians are less familiar with the bureaucratic environment of their offices, hence presenting difficulties to extract rents and profit from them. Eventually, they acquire expert knowledge over time.	R2. New politicians, are subject to greater political jeopardy from citizens’ expectations and patrimonialist pressures, hence experiencing difficulties in utilizing the KTP to produce better welfare outputs. If the opportunity appears, they rapidly shun the “expert knowledge” transferred.
	Insiders	R3. Tenured politicians show more competence in using expert knowledge to deal with “administrative sabotage” of their competitors, and for rent extraction. They frequently use their informational advantage for electoral reasons; and sometimes, for collusion.	R4. Tenured politicians make best use of the transferred knowledge. They are more capable of aligning rent extraction with electoral and developmental gains that bolsters their reputations. Yet they also use their advantage, when losing office, to sabotage competitors, incur in collusion, while dismantling existing local bureaucratic capacities.

In turn, for the group of treated politicians, naïve interactions show that compared to our baseline (nontreated outsiders), outsider politicians who took part in the KTP still manage reduce stunting by −3.9%. While this result does not stem from a semi-experimental analysis as shown in models 5 and 6, it still suggests that the KTP program could be triggering more complex political dynamics: Outsiders might be pressured to re-pay their campaign loans in the form of patrimonialistic appointments and “clientelist handouts” (Brierley 2020), yet face the enhanced scrutiny of the KTP managers, who are more preoccupied in leading mayors to allocate funds toward fighting stunting.

In practice, based on interviews noted in R2 (table 5), it is possible to argue that treated outsiders experience pressure at two fronts: they are trying to keep up with the KTP program demands, while constructing their own political brands in line with their career interests. Of course, for outsiders, such a task is not easy in a context where they run/attain office amid precarious party organizations (proper of Peru’s weak institutional party system and absent civil service career structure), and because they are not yet familiar with the intricacies of their bureaucracies hence facing bigger challenges in congregating loyalist and experienced bureaucrats that could enable them to extract rents from their offices. It is thus no surprise, as R2 signals, that when treated outsiders have the opportunity, they immediately shun some of the lessons of the KTP (exclusively focused on stunting reduction), thus redirecting their efforts toward rents extraction.

In sum, the central lesson here is that politicians never cease to learn—whether directly participating in the KTP or indirectly learning on the go: this is shown by model 7 which only confirms (as previous results) that expert knowledge is best used by experienced politicians who received the KTP.

Finally, as for the variables considered in Xʹ, only “ratio of votes” of politicians that won office appears negatively associated with my dependent variable. This signals that more popular support in their districts, might facilitate politicians’ work—or alternatively, that citizens’ expectations of politicians’ performance in office might pressure them to deliver. Other confounding variables such as the population size per district, the ratio of poverty, and the type of political organization present, in all models, contested and mostly unsignificant coefficients. As well alternative interaction models tested in supplementary table A-5 show mostly contested and unsignificant outputs. Further robustness measures developed in supplementary appendix of this article only account for the strength of the models here discussed.

Qualitative Findings

Results coming from my interviews and documentary revision, further confirm my quantitative findings, yet showing important nuance on how outsider politicians employ expert knowledge in contrast with insider ones. Table 5 summarizes my findings considering my theoretical framework and research design. Most notable findings are noted as #R per quadrant.

Untreated Politicians and Expert Knowledge

R1 reports that outsiders in office undergo a difficult learning process of their bureaucratic functions, being in most cases trapped amid a network of expectations from aides and allies who supported them during electoral times. A local mayor interviewed about his first elected experience argues:

I was before a director of a hospital … but being major is totally different. First time I was elected mayor of the Nazarenas district, I had to learn the administrative processes because the Neighbours Committee [who supported me] wanted their projects to be quickly implemented. I did start from zero, not having the management documents neither personnel … it was hard to find professionals to help you, but if you want to learn a lot depends on your effort and commitment. (Int. 20)

Arguably, newcomer mayors ignore administrative procedures, exhibiting a degree of managerial ignorance in part because, my interviewees contend, the Peruvian government exhibits a myriad of regulations which, for outsiders, are difficult to comprehend. A top MEF functionary further confirms that local politicians who ignore the intricacies of the “11 administrative systems of the state, or at least those involved in budgeting, planning, supply and control” (Int. 1) are unable to execute their programs let alone to respond to those required by the central government. He further argues that most outsider politicians lack capacities to attract personnel or plan their budgetary needs (Brierley 2021). He adds: “there are various ways to execute budget: public auction, simplified assignment, direct services, or in emergencies, direct contract award; newcomers are unaware of those” (Int. 1). Clearly, and in line with statistical evidence, newcomer politicians pay a differential cost of learning initially exhibiting worse office performance. Yet, over time, they become acquainted with the administrative intricacies of their offices hence learning the pre-established career paths involving redistributing rents via “contract clientelism” (Holland and Freeman 2021) through which they could secure a network of aides that enhance their reputations ahead of elections. My interviewee exemplifies:

…mayor Martinelli⁸ joined office without knowing functions, the first year he was lost; the second year he understood how the SIAF⁹ works, the third year was too late…. But then he used the SIAF to benefit his aides, thinking of the next campaign; he blocked projects, tried to hide his “mistakes” … hired personnel only until the end of his term. (Int. 21)

This evidence further confirms that despite exhibiting managerial ignorance, politicians’ motivation to “learn” is conditioned on their need of capturing key budget related bureaucratic positions, extract rents, construe circumstantial support networks, which could aid them to temp other offices where they “can finally showcase some results” (Int. 1 similar sentiments found in Int. 19, 20, 26)—which are well valued by voters of patrimonialist polities (Brierley 2021, 185). Such is the strategy that newcomer politicians must become familiar with in order to succeed in the weak institutional context of Peru (Levitsky 2018).

In turn, R3 documents that insider politicians tend to produce better welfare outcomes. Arguably, insiders have already graduated from the difficulties of being newcomers reported in R1—some of them even exhibit solid “support networks” and are capable of appointing competent and loyal bureaucrats in office. Asked about how he used his experience and knowledge during his mandate, a politician revealed:

…I think learning is a natural process … after my first experience [in office], I knew I needed people knowledgeable about the administrative systems [“capos de la burocracia”]: people informed about the SIAF, SIGA, among others. I have called them; some joined my office … one must be dynamic to attract these people. Like a D.T of a football team—talent is out there. (Int. 23)

Arguably, tenured politicians fill key budgetary positions in office rather quickly. They exhibit larger connections and showcase “informational advantage” (Waterman, Rouse, and Wright 1998) in managing their offices’ administrative and budgetary systems. As well, their tenure, according to my interviews, enables them to better deal with the administrative sabotage of competitors. Soon after they have won the ballot, insiders already expect that their future offices will be sabotaged: budgets will be embezzled, and the personnel dismissed, so they acted. My interview contends:

…when he knew he would lose the election, former mayor Molina compromised 95% of the budget on buying unplanned construction material. Once in office, we cancelled the contract, and used the opportunity of the Modified Budget [PIM] to execute our own works. (Int. 23)

Unsurprisingly, insiders’ informational advantage permits them avoiding the pernicious effects of “past principals” actions (Kappe and Schuster 2021). While this does not necessarily secure that insiders will produce better welfare outputs, their experience and knowledge, certainly helps them to lead the bureaucracy toward advancing their careers. As my interviewees highlight, in some cases, insiders opt for delivering results, so long it helps them to secure their reputations, “propagandize” their office performance, and/or construe new electoral alliances, which could help them to temp re-elections or eventually higher offices (Int. 19). This means that for insiders, external incentives—for instance, reputational ones (Salazar-Morales 2021)—and their personal intentions in office still play a fundamental role. This is further confirmed by my interviewees—and stressed in recent Global South scholarship (Brierley 2020; Brinks, Levitsky, and Murillo 2020; Boas, Hidalgo, and Richardson 2014)—who reveal that in Peru, insiders commonly use their knowledge advantage, “en passant,” for collusion (Int. 21, similarly found in 26 and 27).

Treated Politicians and Expert Knowledge

Unlike the abovementioned cases, politicians who participated in the KTP utilize knowledge differently. The “technical assistance” provided by a high MEF functionary plays a key role in politicians’ performance in office—exhibiting contested patterns of knowledge utilization between outsiders and insiders.

For instance, R2 documents that outsiders receiving treatment are reluctant to “learn”—mostly in part because they experience strong patrimonialist pressures toward paying campaign favors via “contract clientelism” and/or via granting offices to allies (Boas, Hidalgo, and Richardson 2014). A top MEF functionary responsible of training a local politician reveals:

In the [the district] of Panao, the mayor was more worried for his clientele. He has promised giving jobs to the poor people of his town in a construction of road project … also he resented working with the MAIDESOF because they supported his rival.¹⁰ He resisted our suggestion to appoint experienced professionals [tecnicos] in the Local Direction of Social Development—it was a very tense situation. Above all, the mayor did not take our advice in good faith, and it affected the PAN¹¹ execution. (Int. 24)

This evidence further illustrates why outsiders present more difficulties in taking advantage of the KTP. While similar to untreated outsiders (R1), treated ones are experiencing various lines of pressures and expectations as they have recently won elections, being accompanied by a high level MEF functionary, who is guiding and observing their budgetary actions, distorts their administrative behavior, therefore limiting their scope of action. In regular times, outsiders use the “honeymoon period” right after their election for paying favors and delivering the policies they promised during campaign (in our example creating jobs in the road project), and for articulating their network of supporters who will help them to sustain their reputations for future electoral campaigns (Salazar-Morales 2021). Yet, as the program started in 2014, just after local elections, many new politicians might have deemed MEF functionary’s “technical assistance” harmful to their plans. Minutes of the program in Panao further confirm this claim: “the local government resented the technical framework transferred and stopped coordination with local actors for fighting stunting right after the MEF consultant left […] local authorities did not see this as important” (MEF 2017, 53). This only reveals that treated outsiders tend to shun, at the minimum opportunity, the technical assistance provided as an unexpected effect of the program.

On its turn, R4 indicates that insider politicians receiving the KTP from MEF consultants do produce better welfare outputs in office. Given their established network of bureaucracy connoisseurs, and experience in dealing with other instances of the government (either regional and national), they are better equipped to internalize the transferred KTP’s knowledge and utilize it for their own gains. A higher MEF functionary in relation to “experienced mayors” reveals:

…the benefit [for those tenured] is that they have their people [their loyalist bureaucrats]. In Bambamarca, the mayor appointed someone of his trust to learn as much as possible from the PAN. He wanted to show that health networks [to fight against stunting] were working—although everyone knew he was doing it to politically proselytise his activities. (Int. 24)

Like in the “untreated” group (R2), my results show that experienced politicians are more willing to use expert knowledge so long it aligns with their political aims. In fact, a former higher MEF functionary interviewed (now at the World Bank) mentions:

…[experienced] mayors wanted to know how to use the budget, for instance, one asked me how to put in the PAN a “family nurses programme” for early checks to new-borns … I said OK: revise this part of the law … it says you can hire more nurses and paid them more because they will work in the rural area. Then he implemented his programme by calling to the neighbours committee [junta vecinal] and a local NGO for support ... he campaigned it. (Int. 28)

This evidence further confirms our H2 that expert knowledge bestows insiders with off-setting information on how to extend their grips over their bureaucracies by giving them access to previously unknown administrative processes, extracting rents, and redistributing them among aides, who simultaneously will help them to refine their reputations of deliverers. Overall, my interviewees highlight that insiders responded better to the KTP because it enabled politicians to “better use expert knowledge” for improving their own policy programs so they could “[electorally] profit from them” (Int. 24). In the case above examined, further documentary evidence shows that “the mayor [of Bambamarca] was able to programme all health activities in the PAN,” yet toward the end of his term, and for strategical reasons, “he increased his modified budget strongly in 2016 and 2017, after a down period in 2015 [when MEF consultant left the district])” (MEF 2017, 61).

Most notably, my interviews point out that a common factor among insiders’ usage of expert knowledge is their willingness to strengthen their foot in their local bureaucracies. While they circumstantially build capacities by appointing their loyalist expert agents (bureaucrats) in key posts, to mobilize, and lead their bureaucracies toward results, these capacities are not long-lasting—but only serving to their office survival. When politicians know they would lose the ballot, they dismantle the bureaucracy by “dismissing the entire personnel who received the training” (Int. 21). A higher MEF functionary mentions that even sometimes: “[politicians] transfer their prepared technical projects to the regional government so the upcoming [district] mayor will find nothing […].” Moreover, he also notes that some insiders might opt for “deleting the technical profiles of projects” rather than leaving it to their contenders. (Int. 29). He further exemplifies: “When I came back to Santa Maria del Valle to advise the new mayor in 2017, I found only new faces. I had to start again because […the former mayor] used transferred knowledge as a mercantilist, only for his benefit, and to oust his contender, not leaving anything for the municipality nor for citizens” (Int. 29).

Arguably, insiders utilize knowledge for advancing their careers. Given that the Peruvian party system is highly fragmented—where personalistic ties are prevalent (Levitsky 2018)—my interviewees also stress that acquiring administrative expert knowledge facilitates politicians’ use of public projects “as sources to fund their political organisations” (Int. 19). This is: insiders who are in the know and able to attract loyal capable local tecnicos, tend to incur into collusion, so they could obtain resources “to temp other higher offices” (Int. 21).

In sum, R4 documents the nature of expert knowledge utilization in weak and patrimonialist institutional environments. While expert knowledge might circumstantially help politicians to produce better welfare results, their incentives are still cemented in the structural legacies marked by short-term calculations and political survival, proper of the weak and patrimonialist institutional environment of Peru. Hence, expert knowledge, in this case, only serves to strengthen a politician’s grips on office—and most of the times to enable collusion which is used to move their career’s forward.

Conclusion

This article has illustrated the uses of expert knowledge in weak institutional environments characterized by patrimonialist bureaucracies. It is proposed that under such circumstances, politicians employ expert knowledge to strengthen their grips over key bureaucratic (budgetary) functions with the aim of extracting rents to construe political coalitions that could bolster their reputations ahead of elections—via aggrandizing their office performance and/or providing cues about their electoral viability. Results supporting this novel theoretical account present two important implications for public administration scholarship.

First, it situates the debate about politicians and bureaucrats relationship in the institutional settings of the Global South countries. My article hence resonates with emerging evidence (Boas, Hidalgo, and Richardson 2014; Brierley 2020; Oliveros 2016; Salazar-Morales 2021) contesting Global North scholarly preoccupation with politicians’ control over the bureaucracy due to the “informational asymmetry” benefiting independent and unresponsive bureaucratic behavior. On the contrary, my research shows that in patrimonialist bureaucracies like Peru, politicians’ benefit from off-setting informational advantage, and enhanced control over bureaucratic’ careers, thereby dominating policymaking, rent extraction, and distribution.

As a second contribution, my article also expands existing politicians’ career scholarship (Borchert 2011; Brierley 2021; Holland and Freeman 2021; Levitsky 2018; Muñoz 2014; Zavaleta 2014) by illustrating how politicians’ acquisition of expert knowledge, motivations, and career opportunities are mediated by their tenure. It shows that in advancing their careers, new politicians initially struggle to capture state rents to form “independent coalitions” that could help them bolstering their reputations for tempting future offices. In contrast, seasoned politicians better utilize expert knowledge to profit and redistribute state rents among aides. My article shows that insiders best transform bureaucratic rents into political support in order to signal their viability to higher offices, and in doing so, I also show insider politicians produce, en passant, better welfare outputs. My article labels such behavior as “mercantile” because “expert knowledge” is exclusively useful to insiders’ time in office, while it is employed to sabotage competitors, embezzle budgets, and even for collusion purposes.

My research also opens new avenues for exploring the effects of expert knowledge on corruption practices. Recently, scholars such as Brierley (2020, 2021) and Holland and Freeman (2021) show that in weak institutional environments marked by politicians’ predominance over bureaucracies, politicians acquisition of expert administrative knowledge (e.g., budgeting, procurement, etc.) might only facilitate their utilization of office to support corrupt and criminal activities. While my “mercantile theory” provides partial framework to this finding, further analysis on how politicians’ expert knowledge and tenure correlates with corruption practices, as well as how the electorate respond to politicians’ “negative/corrupt” reputational images, is still pending of further scrutiny.

As a final reflection, it is important to mention that the generalizability of my “mercantile theory” to other countries is subject to certain scope conditions as it is originally formulated for subnational Peru. In such context, local politicians display enhanced control over governmental administrative and financial instruments, count with unchecked and uncontested power to appoint, promote, and/or withdraw the careers of local bureaucrats (as of granted by electoral law), and develop their careers in an electorally competitive environment whereby political parties are unimportant—thus focusing more on investing in personalistic political brands. Hence, when utilizing my theory, the reader must carefully consider whether such institutional factors are present, and to which extent. For instance, in a country like Mexico, my “mercantile theory” might need to consider crucial differences such as that Mexican local politicians (unlike Peru) are subject to stricter accountability measures (they are overseen by state governors and by elected fiscal comptrollers known as sindicos), have limited access to financial instruments, and develop their political careers within clearly established party structures. Arguably, such varying conditions might conjure different forms of politicians’ expert knowledge utilization that future research must take into consideration.

Acknowledgments

The author is indebted to Kai Wegrich (Hertie School), Martin Lodge (London School of Economics), Mark Hallerberg (Hertie School), Thurid Hustedt (Hertie School), Richard Shaw (Massey University), and to the three anonymous reviewers and the editors for their valuable comments on an initial version of this article. The author makes especial reference to Luciana Cingolani (Hertie School) and Pedro Pineda (University of Bath) for their kind, human, valuable academic and emotional support during revision of this article. A preliminary version of this article has been presented in the 5th International Conference in Public Policy, Barcelona 2021.

Funding

This research has been funded by the Deutscher Akademischer Austauschdienst (DAAD) Research Grant number 57381412 for the project “Patterns Towards Effective Policy Implementation in Latin America.”

Conflict of Interest

The author declares that he has no conflict of interest in the development, funding, materials employed and/or publication of this article.

Data Availability

Replication material of this article can be found on the following link: https://doi.org/10.7910/DVN/UNC159. Interviews carried out for this article have followed the ethical guidelines of the Hertie School. My informants have been notified about the objectives of my research and the utilization of their data. Participants in this research granted their consent to use their data in line with the Bundesdatenschutzgesetz. All interviews have been anonymized and upon their request their data has been utilized for the purposes of this research only. An anonymized list of interviewees can be found in the supplementary appendix.

Footnotes

1

Thus, the abundance of concepts to nominate to the patrimonialist phenomenon in bureaucratic contexts in the Latin American continent: such as “amiguismo” and “camarillas” (Mexico), “compadrazgo” (Bolivia), “argollas” (Peru), “puntero” (Argentina).

2

According to Encyclopaedia Britannica: “economic doctrine broadly characterised by the monopolistic use of the economy to augment states” power while restricting competition or sabotaging other countries’ economies. “Mercantilists view trade in terms of zero-sum games seeking to keep their governmental balances positive at all costs.”

3

I have opted for this notation for simplicity reasons because as Cerulli explains β₊₁, measures the effects of the treatment in one period before its occurrence (anticipatory effects), while β₋₁, one period after. Although initially confusing, this notation is the outcome of introducing a lag in our model, which renders our dependent vector to Y_t−1 which is caught by β₊₁ while the forward Y_t+1, is caught by β₋₁. For a detailed explanation, please see Cerulli and Ventura (2019, 553–4).

4

Important to mention is that before running my analysis, all variables have undergone a random walk test. Outcomes presented in supplementary table A-1 show that they are stationary and hence present no threat to further statistical analysis. As well, a first inspection to the parallel trend assumptions via lags of all models presented in this article can be found in supplementary table A-2.

5

Institutionally Modified Budget—local government budget modified by the mayor until March of the current fiscal year based on Legislative Decree 1440 § 31.1.

6

There is more nuance on the relationship between mayors and how they help reducing stunting noted in the supplementary appendix “Stunting and Recovery.”

7

My interviewee names these politicians as “passing birds” (aves de paso)—for those who use an office with the unique purpose of getting another at a higher level.

8

Names are fictional upon the request of (some) interviewees.

9

Integrated Financial System of the Peruvian State (SIAF)—A platform where mayors register their “technical projects” to be funded by the Ministry of Finance. Administrative Management System of the Ministry of Finance (SIGA)—digital platform where mayors register and monitor the execution of procurement orders.

10

Committee for the Fight Against Poverty and Malnourishment of the District of Panao.

11

Articulated Nutritional Budget (PAN): Part of the local government budget used in fight against stunting.