CAF - Performance Incentives and Managerial Practices Experimental Evidence from Local Governments in Peru

C A F - W O R KI N G P AP E R # 2 0 2 5 / 1 5

T h i s v e r s i o n : D e c e m b e r 1 5 , 2 0 2 5

Performance Incentives and Managerial

Practices: Experimental Evidence from Local

Governments in Peru Sarita Oré-Quispe1 1Teachers College, Columbia University. spo2111@tc.columbia.edu This paper examines how public organizations respond to institutional incentive design by comparing an input-based contract with a mixed inputand outcome-based contract. Using a clustered randomized controlled trial across 539 rural municipalities in Peru, I study how these contracts shape managerial practices in the implementation of a national Home Visit Program. While the mixed incentive did not alter home-visit coverage, it produced clear shifts in managerial behavior: municipalities expanded their supervisory staff, rewarded Community Health Workers more frequently, and intensified monitoring of the anemia indicator tied to the contract. They also adopted more targeted innovations for children at risk of anemia, although broader processes—such as training and supervision intensity—remained largely unchanged. K E Y W O R D S Performance-Based Incentives, Public Service Delivery, Health, anemia Small sections of text that are less than two paragraphs may be quoted without explicit permission as long as this document is acknowledged. Findings, interpretations and conclusions expressed in this publication are the sole responsibility of its author(s) and cannot be, in any way, attributed to CAF, its Executive Directors or the countries they represent. CAF does not guarantee the accuracy of the data included in this publication and is not, in any way, responsible for any consequences resulting from its use. ©2025 Corporación Andina de FomentoC A F - D O C U M E NT O D E T R AB A J O # 2 0 2 5 / 1 5

E s t a v e r s i ó n : 1 5 d e d i c i e m b r e d e 2 0 2 5

Incentivos al desempeño y prácticas de gestión: Evidencia experimental de gobiernos locales en Perú Sarita Oré-Quispe1 1Teachers College, Columbia University. spo2111@tc.columbia.edu Este artículo examina cómo las organizaciones públicas responden al diseño de incentivos institucionales comparando un contrato basado en insumos con un contrato mixto basado en insumos y resultados. Utilizando un ensayo controlado aleatorizado en 539 municipalidades rurales del Perú, analizo cómo estos contratos moldean las prácticas de gestión en la implementación de un programa nacional de visitas domiciliarias. Si bien el incentivo mixto no alteró la cobertura de visitas domiciliarias, este produjo cambios en el comportamiento gerencial: las municipalidades ampliaron su equipo de supervisores, recompensaron con mayor frecuencia a los Agentes Comunitarios de Salud y reforzaron el monitoreo del indicador de anemia vinculado al contrato. También adoptaron innovaciones más específicas para niños en riesgo de anemia, aunque los procesos más amplios —como la capacitación de los agentes y la intensidad de la supervisión— se mantuvieron en gran medida sin cambios. K E Y W O R D S Incentivos basados en el desempeño, provisión de servicios públicos, salud, anemia Pequeñas secciones del texto, menores a dos párrafos, pueden ser citadas sin autorización explícita siempre que

se cite el presente documento. Los resultados, interpretaciones y conclusiones expresados en esta publicación son de exclusiva responsabilidad de su(s) autor(es), y de ninguna manera pueden ser atribuidos a CAF, a los miembros de su Directorio Ejecutivo o a los países que ellos representan. CAF no garantiza la exactitud de los datos incluidos en esta publicación y no se hace responsable en ningún aspecto de las consecuencias que resulten de su utilización. ©2025 Corporación Andina de FomentoORÉ-QUISPES. 2 1|INTRODUCTION Governments play a central role as service providers and in facilitating economic development (Finan et al., 2017). What can be done to motivate public providers1 to deliver quality services? A growing body of research focuses on the importance of Pay-for-Performance (P4P) monetary incentives for inducing agents’ effort (Hasnain et al., 2014; Finan et al., 2017). When it comes to incentive design, the decision of whether to contract based on inputs or outcomes is not trivial (Mohanan et al., 2021). For instance, if the primary objective of a program is to achieve a specific outcome —such as improving health— it may seem logical for performance incentives to reward those outcomes rather than service utilization or other inputs (e.g., home visits). However, there is a trade-off between innovation and risk. While contracting on outcomes can provide strong incentives for agents to put forth effort and develop efficient, context-appropriate strategies to maximize desired results, this may be ineffective if agents lack control or doubt their ability to influence the outcome (Miller and Babiarz, 2014; Abraham and Cooper, 2019). In such cases, where the contract may be perceived as risky, it might be more effective to focus on contracting inputs instead, as these

are generally more responsive to agents’ effort. In this paper, I examine the effects of input-based incentives compared to a middleground approach that combines both inputand outcome-based incentives, which I denominate asMixed incentive. In particular, I compare the effects of both incentive schemes on public service delivery and managerial practices in the implementation of the home visits. To this end, in partnership with the Peruvian Ministry of Economy and the Ministry of Health, we designed and implemented both incentive schemes targeting the rural municipalities. The input-based incentive scheme considers two input indicators: i) the percentage of children under 12 months old who receive home visits, and ii) the percentage of pregnant women who receive home visits. The Mixed incentive extends the input-based contract by incorporating an additional outcome indicator: the percentage of children who are not anemic or have improved their hemoglobin levels at six and/or twelve months of age. I evaluate the effects of the mixed incentive with a clustered-randomized controlled trial involving 539 municipalities (85% of the national group of rural municipalities). From these, 270 municipalities were randomly assigned to input-based contracts—i.e., continued under thestatus quocontract existing since 2019—while 269 received the mixed incentive contract. There is no pure control group without any type of incentive scheme. These incentive schemes were applied to the municipalities during the period of August to December

2024. Baseline data collecting information on managerial practices was implemented in March–April 2024, approximately three months before the incentive intervention was introduced. Endline was collected during June-September 2025. To shed light on whether there is a differential effect between the input-based and mixed incentive contracts on service delivery, I use administrative data from the Home Visit Program.

Municipalities demonstrated a strong understanding of the incentive schemes to which they were assigned. By October 2024, 98.2% of municipalities correctly identified whether 1The public sector organizations differ from the private sector in multiple aspects. Unlike the private sector, public sector objectives are often more long-term, complex and diffuse, making performance difficult to measure (Xu et al., 2024). In addition, employees in the public sector are often driven by intrinsic motivations, making financial incentives less effective (i.e., motivational crowding out). Public agencies also face multiple principal–agent problems (Dixit, 2002; Pepinsky et al., 2017) due to complex hierarchies, and are often subject to political and institutional constraints that do not apply to private sector companies.ORÉ-QUISPES. 3 they were operating under an input-based incentive or a mixed incentive. The introduction of a mixed input-and-outcome incentive contract generated clear shifts in managerial behavior, even though frontline service delivery remained unchanged. Municipalities assigned to the mixed incentive expanded their supervisory teams by 61% and narrowed supervisors’ scope of responsibilities, indicating greater specialization in oversight functions. Managers also became substantially more likely to reward Community Health Workers —an increase of 17 percentage points (p.p.). Managers were also more likely to monitor the anemia indicator tied to the contract, a 45% increase relative to the input-based group. Beyond these internal management adjustments, municipalities under the mixed incentive selectively adopted innovations targeted at children at risk of anemia, while broader processes such as CHW training, supervision intensity, and general monitoring systems remained largely unaffected. This study makes several contributions to the literature. To the best of my knowledge, this study will be the first to compare input-only incentives to a scheme combining input incentives with output incentives in the context of institutional incentives, contributing

to a better understanding of service delivery in the developing world (Pepinsky et al., 2017). It will also contribute to the literature on the personnel economics of the state and bureaucracies and on how to increase civil servant performance (Celhay et al., 2019; Huillery and Seban, 2021; Luo et al., 2019; Miller and Babiarz, 2014; Mohanan et al., 2021; Olken et al., 2014). Finally, it will contribute to the literature on home visits and volunteer health workers (Lehmann and Sanders, 2007; Caria et al., 2024; Fracchia, wp; Fracchia et al., 2023). The rest of the document proceeds as follows: in Section 2 and 3 I describe the setting and the home visit program. In Section 4 I describe the experimental design. Section 5 and 6 present the data and main results. Finally, Section 7 concludes. 2|BACKGROUND This section provides essential context for understanding the intervention analyzed in this study. It begins by outlining the country’s administrative structure and the role of municipalities in public service delivery, with a focus on the Municipal Incentive Program—a national performance-based scheme that allocates additional budget to municipalities. It then describes the prevalence and consequences of childhood anemia in Peru. 2.1|Administrative organization and municipal incentives Peru is an upper-middle-income country with a per capita income of 8,452 USD in 2024.2 Administratively, the country is divided into 25 regions, which are further subdivided into 196 provinces and 1,891 districts. Regional governments oversee the regions, while local governments –known as municipalities– are responsible for district-level administration. In practice, municipalities primarily manage basic local services such as water and sanitation,

waste collection, maintenance of public spaces, local infrastructure, business licensing and local tax collection. The decentralization of the health sector, however, remains limited. The primary responsibility for both primary healthcare and hospital services in Peru lies at the regional level, with little participation from local governments (Alves et al., 2025). In 2009, the Ministry of Economy and Finance launched the Municipal Incentive Pro2World Bank (November 29th, 2024). Peru Overview. https://data.worldbank.org.ORÉ-QUISPES. 4 gram3 to improve the quality of local service delivery. The incentive program operates as an institutional performance-based financing scheme, allowing municipalities to access additional budget conditional on achieving specific service delivery targets. Performancebased transfers of this kind have become increasingly common across Latin America and the Caribbean (Alves et al., 2025).4 On top of that, governments often favor institutional over individual performance incentives, as they limit discretion, enhance transparency, and reduce corruption risks.5 Participation in the Municipal Incentive Program is formally voluntary; however, in practice, all municipalities participate, as the potential additional transfers represent. On average, municipalities can obtain resources equivalent to 3.4 times their expected annual tax revenue (Figure A.2), with a median ratio of 15.7. The incentive payments are disbursed after performance verification, typically during the following fiscal year, and can be used freely to finance the provision of public services. 2.2|Children’s anemia in Peru Anemia is a condition in which the number of red blood cells or the hemoglobin concentration is lower than normal. Anemia during early childhood is broadly associated with growth retardation, reduced resistance to infections, delayed psychomotor development, lower language acquisition, and impaired cognitive and socioemotional development (INS, 2024). These developmental deficits can become a significant deterrent to human capital accumulation in developing countries, and hence a potentially important barrier to economic development (Chong et al., 2016). Yet anemia continues to be a pressing global concern. Nearly half of all children under five worldwide are affected (42.8%), and the burden is even more severe in lowand middleincome countries, where prevalence reaches 56.5% (WHO, 2025; Sun et al., 2021). In Peru, the prevalence of anemia among children under five years of age is 33.4%, and 43.1% among those aged 6 to 35 months (INEI, 2023). This prevalence is consistently higher in rural areas.6. Anemia is primarily caused by iron deficiency (WHO, 2023) and it is the most prevalent micronutrient deficiency in the world (Chong et al., 2016). For children under one year of age, the most appropriate interventions are iron supplementation and caregiver’s nutritional education, given their age-specific nutritional needs and feeding practices. 7 For infants under one year of age, daily iron supplementation is administered in drops or syrup form 3In Spanish,Programa de Incentivos a la Mejora de la Gestión Municipal. Before 2016, it was known asPlan de Incentivos a la Mejora de la Gestión Municipal. For more information, see www.mef.gob.pe. 4A regional survey covering 24 countries found that 14 of 17 respondents use some form of performance-based financing (Pimenta et al., 2023). 5However, pooling incentives at the organizational level can also generate free-riding and weak internal accountability, since the rewards may not translate into direct benefits for the frontline workers (Miller and Babiarz, 2014). 6Between 2009 and 2023, the prevalence of anemia in rural areas has been approximately 10 p.p. higher than

in urban areas. For instance, in 2023, anemia among children aged 6 to 35 months was 50.3% in rural areas, compared to 40.2% in urban areas (INEI, 2023). 7The World Health Organization recommends four strategies to combat anemia: iron supplementation, food fortification, nutrition education, and dietary diversity (WHO, 2006). Although food fortification is generally considered the most effective and cost-efficient strategy (Mithra et al., 2020), it is not ideal for infants under one year. Children under six months are typically exclusively breastfed, and even after the introduction of complementary feeding at six months, the small portion sizes limit the intake of fortified and diverse foods.ORÉ-QUISPES. 5 for at least three consecutive months, with consistent adherence being crucial to ensure its effectiveness (WHO, 2016). 3|THE HOME VISIT PROGRAM In 2019, the Municipal Incentive Program began incentivizing the implementation of a Home Visit Program "Commitment 1" aimed at increasing adherence to iron supplementation among children under one year of age. It constituted a first step toward the decentralization of public health functions, engaging local governments in the delivery of preventive and community-based health interventions.8 The program targets 879 municipalities, including all 245 urban municipalities located in main cities and 637 of the most rural municipalities (Figure 1).9 F I G U R E 1Geographic Distribution of the 879 Municipalities Participating in the Home Visit Program Notes:The figure shows the districts included in the implementation of the Home Visit Program. according to Supreme Decree No. 072-2019-EF. The Home Visit Program is a community education initiative that delivers information 8This program was designed by the Ministry of Health. The full name of the program in Spanish isCompromiso 1: Mejora del estado nutricional y de salud de los niños. Before 2023, it was calledMeta 4: Acciones de los municipios

para promover la adecuada alimentación, la prevención y la reducción de la anemia. 9Municipalities are classified into seven types (A to G) based on their urban status, whether they are the main district of a province, and their share of urban population (Table A.1). According to this classification, the Home Visit Program targets the urban municipalities from main cities (types A, C, and D) and the group of the most rural municipalities (type G), with less than 35% of urban population.ORÉ-QUISPES. 6 to caregivers mainly on how to reduce anemia in their children. As part of this initiative, municipalities, in collaboration with healthcare staff, conduct home visits to low-income10 households with children aged 12 months or less. Months 4 and 5 mark a critical developmental window since neurological development accelerates and iron needs rise sharply with it (MIDIS, 2018). Accordingly, home visits for children aged 4–5 months are preventive and aim to reduce the likelihood of anemia by six months. For children aged 6–12 months, visits may be preventive or recuperative, depending on whether the child has already been diagnosed with anemia.11 Home visits are performed three times a month, each with an average duration of 20 minutes. In 2024, during these home visits, standardized messages promoting essential health practices for maternal and child well-being were delivered, focusing on nutrition, hygiene, preventive care, and early childhood development.

Home visit organization: Municipalities are vertical organizations with three layers of hierarchy (see Figure 2). Each municipality appoints a Manager (top layer) who oversees the implementation of the Home Visit Program and acts as the principal to a team of supervisors.

Just over half of managers (55%) report being motivated by the program’s contribution to

population wellbeing, while the remainder are primarily motivated by meeting performance targets to secure the incentive payment. A large majority (87%) believe that the program’s success depends on the effort they exert. Each supervisor (intermediate layer) has a group of 20 service delivery agents 12 or Community Health Workers - CHW (bottom layer) as his own lower-tier agents. Each CHW works with 15 families. The personnel size depends on the size of the target population in their jurisdiction. 10The Program targets low-income families identified by their type of health insurance. The children targeted are those who do not have any health insurance or have government-sponsored health insurance for low-income populations calledSeguro Integral de Salud (SIS).SISis equivalent to Medicaid in the United States context. 11From 2019 to 2022, home visits targeted households with children aged 4 to 11 months. Until 2022, children 4-5 months would receive home visits regardless of their nutritional status; while children 6-11 would receive them only if they had been diagnosed with anemia at least once. In 2023, due to budget constraints, the focus shifted to households with children aged 3 to 5 months. In 2024, home visits were expanded to include households with children aged 1 to 12 months and pregnant women. Since 2023 home visits are performed regardless of the nutritional status of the child or the pregnant woman. 12The majority of service delivery agents (80%) are CHWs, so, for simplicity, I use the termsservice delivery agents andCHWinterchangeably.ORÉ-QUISPES. 7 F I G U R E 2Hierarchy Illustration for Municipality 1 and 879 Central Government Municipality Manager 1 Municipality Manager 879 Supervisor 1 Supervisor 2 Supervisor n Supervisor 1

Supervisor 2 Supervisor n Community Health Worker 1 Community Health Worker 20 Community Health Worker 1 Community Health Worker 20 15 families 15 families 15 families 15 families Notes:Municipalities have a vertical organization in developing the Home Visit Program. Municipalities designate one manager to oversee the whole Program. The manager is the principal with the team of supervisors as his agents. At the same time, these supervisors are in charge of the service delivery agents. Mostly, the service delivery agents are Community Health Workers. The personnel size ultimately depends on the size of the target population in their jurisdiction. Each service delivery agent works with 15 families, and there is a supervisor for every 20 service delivery agents. Personnel from the health facilities train the service delivery agents.

Role of the managers: The municipal manager oversees all health and social programs within the municipality, including the implementation of the Home Visit Program. As part of this role, the manager plans and assigns tasks to supervisors and makes key personnel decisions such as hiring, dismissals, promotions, and setting incentives or compensation for supervisors. In coordination with health facility staff, the manager may also participate in similar personnel decisions concerning CHWs.

Overall, autonomy in these areas is high: 46% of managers report a high level of autonomy, 41% a medium level, and only 11% a low level, indicating that most managers exercise substantial discretion over the program implementation. Autonomy, however, varies across specific tasks. Managers report particularly high autonomy in assigning tasks to supervisors (55%), hiring CHWs (52%), and managing personnel payments (47%). Similarly, 44% report high autonomy in planning activities, while 42% report high autonomy in incentivizing CHWs and 42% in hiring supervisors. Taken together, these patterns suggest that managers

hold significant decision-making authority across multiple dimensions of the program management (Figure A.3).

Role of the supervisors: Supervisors play a central role in supporting CHWs by coordinating with health facilities to ensure that staff provide regular training sessions and that referred children receive appropriate care. They also monitor CHWs’ productivity and the quality of their home visits. To maintain high standards of service delivery, supervisors accompany CHWs on home visits and use a checklist to assess performance. This allows them to provide constructive feedback and guide health workers on areas for improvement.

Additionally, supervisors offer ongoing, on-site training to further develop CHW’s skills and to enhance their performance. Supervisors exhibit a high degree of autonomy in theseORÉ-QUISPES. 8 activities: 71% report high autonomy in coordinating with health facilities and 59% in training CHWs. These patterns underscore supervisors’ pivotal role in maintaining service quality and supporting CHW performance (Figure A.3).

Role of the CHWs: CHWs are responsible for scheduling and conducting the home visits, during which they provide health education to families using standardized messages.

For doing so, they receive 2-days in-person training from the Health Facility staff on essential topics such as iron supplementation, exclusive breastfeeding, iron-rich complementary feeding, vaccinations, the importance of attending health center growth and development check-ups, handwashing, safe water consumption, and early bonding. They are also trained to recognize children’s danger signs for referral to health facilities. In addition, CHWs receive discretionary "on-the-field" training from supervisors. To ensure high-quality visits, CHWs follow a printed checklist format, and use a flipchart that provides a structured guide on how to effectively deliver each topic to families.13 The municipality’s performance ultimately depends on the CHW’s effort since they have the most direct contact with the target population of the Home Visit Program.

Performance verification and bonus payment:The Ministry of Health verifies municipal performance using administrative data (details in Section 5.1). Each municipality d participates in the incentive program and is eligible for a total potential bonus Bd, distributed equally across K performance indicators. Hence, each indicator k carries a potential bonus of Bdk =B d/K. Municipality’s payment for indicator k, denoted Pdk, depends on the municipality’s performance xdk relative to the pre-defined performance floor ( ¯xdk) and performance ceiling (¯xdk): Pdk =    0, forx dk < ¯xdk 0.75∗B dk, for ¯xdk ⩽x dk <¯xdk Bdk, forx dk ⩾¯xdk (1) The municipal incentive payment Pm = PK k=1 Pdk is disbursed after performance verification and at the beginning of the following fiscal year, without affecting the ongoing implementation. Before the trial began, from January to July 2024, municipalities were evaluated solely on whether they met the predetermined target for the indicator measuring the share of children under 12 months who received a home visit. Consequently, the total bonus that municipalities could obtain depended entirely on their performance in that single indicator. Meeting the performance targets in the incentive scheme is highly salient for managers. They are generally optimistic about their ability to do so, they estimate that the probability of meeting the performance targets is 82.6% (Table 1). At the same time, managers anticipate meaningful accountability consequences if the target is not met. Among rural managers, the perceived probability of being dismissed is 56%, and the perceived probability of being reassigned to another municipal department is 31%. Managers also expect consequences for their subordinates, though of lower magnitude. If the targets are not achieved, they assign

a 50% probability that supervisors would be dismissed and a 36% probability that CHWs would face dismissal. 13More details on the materials used by the CHWs and supervisors during the home visits can be found in Appendix C.ORÉ-QUISPES. 9 TA B L E 1Managers’ Expectations and Perceived Consequences Related to the Incentive Scheme Total Urban Rural Municipality Municipality (%) (%) (%) Expected probability of meeting the target 81.19 77.62 82.58 Expected probability of consequences if failing the target Manager dismissed 59.11 66.26 56.32 Manager reassigned to another department 30.06 26.85 31.31 Supervisors dismissed 54.25 65.70 49.79 CHWs dismissed 37.48 41.49 35.92

Notes: Expected probability of meeting the targetis the supervisor’s subjective probability that the municipality will meet the performance thresholds.Manager dismissed,Manager reassigned, Supervisors dismissed, andCHWs dismissedare dummies equal to 1 if the supervisor reports that each respective position would face that consequence should the municipality fail to meet all indicators. Data come from the baseline survey to managers. 4|EXPERIMENTAL DESIGN The incentive scheme for August – December 2024 was developed in joint collaboration with the Ministry of Economy and the Ministry of Health. It was formalized through a series of legal resolutions.14 The incentive scheme considers up to three indicators:

(i)The percentage of children under 12 months old who receive home visit (ii)The percentage of pregnant women who receive home visits (iii) The percentage of children who are not anemic or have increased their hemoglobin levels at six and/or twelve months old Based on these indicators, municipalities were offered one of two alternative incentive contracts: • Under theinput-based incentive contract, municipal performance is evaluated exclusively

on the two input indicators (i and ii). Municipalities are eligible for a total potential bonus of Bm, which is divided equally across the two indicators, so that each indicator carries a maximum payment ofB mk =B m/2. • Under themixed incentive contract, municipal performance is evaluated on the same two input indicators (i and ii) as well as the outcome indicator (iii). The total potential bonus Bm is held constant across contracts but is divided equally across the three indicators, such that each indicator carries a maximum payment ofB mk =B m/3. I use a clustered randomized controlled trial to evaluate the impacts of the mixed incentive on municipality’s service delivery and managerial practices. Randomization was 14See, among others: Law 31953, DS 318-2023-EF, RD 0006-2024-EF/50.01, DS 132-2024-EF, and RD 0027-2024EF/50.01.ORÉ-QUISPES. 10 done on the group of rural municipalities at the district level.15 For evaluation purposes, I restrict the sample to rural municipalities with at least 15 children, resulting in a final group of 539 municipalities. Using a random number generator in STATA, 270 were randomly assigned to input-based contracts, while 269 received an inputand outcome-based incentive contract or"Mixed incentive"(See Figure 3). 16 F I G U R E 3Experimental Design

Period:

STATUS QUO RCT PHASE

[May 2019 – Jul 2024] [Aug 2024 – Dec 2024] Input-based incentives (N=539) Input-based incentives (N=270) Mixed incentive (N=269) Notes:The figure summarizes the experimental design. Municipalities receiving input-based incentives from May 2019 to July 2024 were randomly assigned to either continue with the same

scheme (N=270) or to switch to a mixed incentive during the RCT phase (N=269). The treatment assignment was disseminated through multiple institutional channels to ensure transparency and municipal awareness. In late July 2024, the Ministry of Economy and Finance published the legal resolutions in the national newspaper and on its website. These resolutions included an extensive table listing all participating municipalitie