CAF - Labor Market Impacts of an Internship Program on People with Disabilities

DEVELOPMENT BANK

OF LATIN AMERICA

AND THE CARIBBEAN

CAF - WORKING PAPER #2026/06

First version: April 28, 2026

Labor Market Impacts of an Internship Program on People with Disabilities Inés Berniell! | Belén Cafiuelo | | Néstor Grién” | ! CEDLAS, IIE, Universidad Nacional deLa Plata. miberniell0gnail .con CEDLAS, CEPE. luberniellognail.con OEDE Ministerio de Capital Humano de Argentina. veastil10trabajo. gob.ar 4CAF-Development Bank of Latin America and the Caribbean. ncanueloteaf .con SCAF-Development Bank of Latin America and the Caribbean. ndelanatatcaf. con “Gobierno de la Provincia de Cordoba, Universidad Nacional de Cérdoba. ggarciaorotgnail .con “Gobierno de la Provincia de Cordoba, Universidad Nacional de Cérdoba. ngriontgnail .con Inter-American Development Bank. carlosjutiadb. org °CEDLAS, IE, Universidad Nacional deLaPlata. luislaguingeatgmail .con Lucila Berniell® | Dolores de la Mata® | Federico Juncosa Victoria Castillo® | Gerardo Garcia Oro® 8 | Luis Laguinge’ This paper examines the impact of an internship program in Argentina (Programa Primer Paso; PPP) on the formal employment trajectories of people with disabilities (PwD). Using administrative data from 2002 to 2023 and a staggered differencein-differences design, we estimate that program participation increases formal employment and improves employment stability over time. The average probability of being formally employed rises 8 percentage points within three years after treatministrative data from 2002 to 2023 and a staggered differencein-differences design, we estimate that program participation increases formal employment and improves employment stability over time. The average probability of being formally employed rises 8 percentage points within three years after treatment, with the effect growing over time. The program also increases sustained employment and total months in formal jobs. About three-quarters of the employment gains arise from jobs at firms other than the initial internship host, indicating broad integration into the formal labor market, not just formalization of preexisting ties. The findings suggest that internships can reduce labor market gaps for PwD in developing countries, with patterns consistent with three mechanisms: alleviating informational frictions, raising expectations about employment prospects, and acquiring transferable skills. KEYWORDS disability, labor markets, internship programs, active labor market policies, employment formality This paper was originally published as an Inter-American Development Bank (IDB) Working Paper. This CAF Working Paper version reproduces the original content under the Creative Commons Attribution 4.0 International License (CC BY 4.0). The IDB is acknowledged as the original publisher and funding institution. 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 useDEVELOPMENT BANK

OF LATIN AMERICA

AND THE CARIBBEAN

CAF - DOCUMENTO DE TRABAJO #2026/06

Esta versi6n: 28 de abril de 2026 Impactos en el Mercado Laboral de un Programa de Pasantias para Personas con Discapacidad Inés Berniell! | Belén Cafiuelo |

Esta versi6n: 28 de abril de 2026 Impactos en el Mercado Laboral de un Programa de Pasantias para Personas con Discapacidad Inés Berniell! | Belén Cafiuelo | | Néstor Grién” | ! CEDLAS, IIE, Universidad Nacional deLaPlata. niberniellognail. con CEDLAS, CEPE. luberniellognail.con OEDE Ministerio de Capital Humano de Argentina. veastil10trabajo. gob.ar 4CAF-Development Bank of Latin America and the Caribbean. ncanueloteaf .con SCAF-Development Bank of Latin America and the Caribbean. ndelanatatcaf. con “Gobierno de la Provincia de Cordoba, Universidad Nacional de Cérdoba. ggarciaorotgnail . con “Gobierno de la Provincia de Cordoba, Universidad Nacional de Cordoba. ngricn@gnail.con Inter-American Development Bank. carlosjutiadb. org °CEDLAS, IE, Universidad Nacional deLaPlata. luislaguingeatgmail .con Lucila Berniell® | Dolores de la Mata® | Federico Juncosa Victoria Castillo® | Gerardo Garcia Oro® 8 | Luis Laguinge’ Este articulo examina el impacto de un programa de pasantfas en Argentina (Programa Primer Paso; PPP) en las trayectorias de empleo formal de personas con discapacidad (PcD). Utilizando datos administrativos de 2002 a 2023 y un disefio de diferencias en diferencias escalonado, se estima que la participacion en el programa aumenta el empleo formal y mejora la estabilidad laboral. Tres afios después del tratamiento, la probabilidad prodatos administrativos de 2002 a 2023 y un disefio de diferencias en diferencias escalonado, se estima que la participacion en el programa aumenta el empleo formal y mejora la estabilidad laboral. Tres afios después del tratamiento, la probabilidad promedio de empleo formal aumenta 8 puntos porcentuales, y el efecto crece con el tiempo. El programa incrementa el empleo sostenido y los meses trabajados formalmente. Tres cuartas partes de las mejoras en el empleo provienen de empresas distintas ala dela pasantia inicial, indicando una integracién amplia en el mercado laboral formal y no solo la formalizacién de vinculos preexistentes. Los hallazgos sugieren que las pasantias pueden reducir las brechas laborales de las PcD en los paises en desarrollo mediante la mitigacion de fricciones informativas, las mayores expectativas sobre las perspectivas de empleo y la adquisicion de habilidades transferibles. KEYWORDS discapacidad, mercados laborales, programas de pasantfas, politicas activas del mercado de trabajo, formalidad del empleo Este documento fue publicado originalmente como documento de trabajo del Banco Interamericano de Desarrollo (BID). Esta versién de Documento de Trabajo CAF reproduce el contenido original bajo licencia Creative Commons Attribution 4.0 International (CC BY 4.0). Se reconoce al BID como institucién publicadora original y financiadora. Pequerias secciones del texto, menores a dos pérrafos, pueden ser citadas sin autorizacién explicita siempre que se cite el presente documento. Los resultados, interpretaciones y conclusiones expresados en esta publicacion 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 paises que ellos representan. CAF no

en esta publicacion 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 paises que ellos representan. CAF no garantiza la exactitud de los datos incluidos en esta publicacion y no se hace responsable en ningtin aspecto de las consecuencias que resulten de su utilizacionBERNIELL ET AL. 2 1 | INTRODUCTION People with disabilities (PwD) face significant barriers in the labor market worldwide, resulting in lower participation rates, higher unemployment, and greater informality compared to people without disabilities. In Latin America and the Caribbean (LAC), PwD experience approximately 20 percentage points lower employment rates and 10 percentage points higher informality rates than the general population (Berlinski et al., 2021). With approximately 14% of the population in LAC currently living with some form of disability, addressing these labor market gaps represents a significant economic and social challenge. Rigorous evidence on active labor market policies (ALMPs) that improve employment outcomes for PwD remains scarce, especially in developing countries (Duryea et al., 2023a; Rucci, 2025b). Internship programs, which combine on-the-job training with work experience, are a particularly underexplored intervention for this group. We analyze the First Step Program (Programa Primer Paso; PPP), a long-standing largescale internship program in the province of Cérdoba, Argentina, with quotas for PwD since 2012. The program provides beneficiaries with 12-month formal employment in registered firms in the private sector. We combine administrative records from the program with formal employment data from Argentina’s social security system (Sistema Integrado Previsional Argentino; SIPA) to create a comprehensive dataset spanning 2002 to 2023. This allows us to track individual employment trajectories both before and long after program participation. We measure the program’s impact on the labor market trajectories of

Previsional Argentino; SIPA) to create a comprehensive dataset spanning 2002 to 2023. This allows us to track individual employment trajectories both before and long after program participation. We measure the program’s impact on the labor market trajectories of beneficiaries, specifically the likelihood of securing formal employment, and examine the underlying mechanisms. Our empirical strategy exploits variation in the timing of program entry and leverages recent advances in difference-in-differences (DiD) estimation under staggered treatment adoption. We rely on the estimator proposed by Callaway and Sant’Anna (2021), which addresses the limitations of traditional two-way fixed effects in the presence of treatment-effect heterogeneity. We use not-yet-treated PwD that apply to the program as a comparison group and conduct robustness checks using other control groups to validate our identification. Our results show that participation in the PPP substantially improves the employment trajectories of PwD. The probability of being formally employed for at least one month in a given year increases on average by 8 percentage points in the post-treatment period, reaching nearly 11 points by year three. The program also fosters sustained employment: The probability of working at least six months in a year increases by 6.6 points, while the chance of twelve consecutive months of employment rises by 4.2 points. In addition, participation boosts total months in a formal job in a calendar year by an average of 0.7 months, with effects accumulating over time. These gains are particularly meaningful given the low baseline employment levels among program applicants and the absence of restrictions on re-enrollment in the program. These improvements are not simply the result of continued employment with the participating firm: At least three-quarters of the gains come from jobs at other firms, underscoring broader integration into the formal labor market. These results are robust to the use of the Borusyak et al. (2024) estimator, a sensitivity analysis following the method proposed by Rambachan and Roth (2023), and alternative control-group definitions.

labor market. These results are robust to the use of the Borusyak et al. (2024) estimator, a sensitivity analysis following the method proposed by Rambachan and Roth (2023), and alternative control-group definitions. We contribute to the literature in four main ways. First, we provide novel causal evidence from a developing country on the medium-term effects of internship programs for PwD—a population typically underrepresented in ALMP evaluations. Most studies focus on highincome countries and reach mixed results, often showing limited or short-lived impacts. In lowand middle-income countries, where formal jobs are scarce and public budgets are constrained, identifying effective tools to integrate PwD into the formal sector is even moreBERNIELL ET AL. 3 critical. Yet empirical evidence from these settings remains scarce, and few studies have examined impacts beyond the short term. Our study helps fill this gap and shows that participation in the PPP leads to a sustained and growing increase in formal employment. Second, we examine heterogeneous effects across gender, age, type of disability, firm size, and place of residence. We find larger impacts for individuals residing in the capital city (also called Cordoba). Effects are smaller for those placed in small firms and for individuals with mental disabilities. We find no statistically significant differences by age, and observed gender differences in absolute terms over the medium term should be interpreted in light of baseline differences in formal employment levels. Third, we investigate the mechanisms and dynamics behind the results. We show that many beneficiaries participate in the PPP more than once. This extended participation, instead of discouraging the search for unsubsidized jobs (as one might expect), may foster integration into the labor market, giving beneficiaries the time needed—probably more than 12 months—to transition from subsidized internships to unsubsidized formal employment. As noted, we find that formal employment gains are primarily concentrated in firms other than the internship providers, indicating broader improvements in employability and access to the labor market. This finding is consistent with the idea that the program alleviates the barriers workers face, such as informational frictions (including expectations about

As noted, we find that formal employment gains are primarily concentrated in firms other than the internship providers, indicating broader improvements in employability and access to the labor market. This finding is consistent with the idea that the program alleviates the barriers workers face, such as informational frictions (including expectations about employment prospects and the benefits of formal work), and facilitates acquisition of transferable skills. Fourth, we examine the effects on participating firms. The evidence on firm behavior is less conclusive: On average, we do not find significant changes in the share of firms that hire at least one PwD after hosting a PPP intern with a disability. (We do observe a statistically significant but modest impact on small firms.) This finding should not be interpreted as a shortcoming of the program: While programs of this kind could help shift employer perceptions and reduce stigma, thereby expanding access to formal jobs for other PwD, the primary objective of the PPP is to improve the employment prospects of workers rather than altering firms” hiring practices. Our findings suggest that if the program aims to influence firm behavior, its design needs to incorporate this objective explicitly. More broadly, our paper highlights the need for further research on how ALMPs can shape employer behavior and reduce PwD’s barriers in the labor market. Our study not only contributes to the literature but provides actionable recommendations for policy makers to improve the design and implementation of internship programs tailored to PwD. The rest of the paper is organized as follows. Section 2 reviews the relevant literature on ALMPs and labor market integration of PwD. Section 3 describes the institutional context of the PPP and the features of its disability component. Section 4 presents the data, and Section 5 outlines the empirical strategy. Section 6 reports the main results, while Section 7 presents robustness checks and discusses potential mechanisms. Section 8 concludes. 2 | RELATED LITERATURE Our paper contributes to the literature on ALMPs for PwD by analyzing the causal effects of an internship program—a type of intervention that has received limited attention in the literature, as existing studies have focused primarily on other types of ALMPs. In addition,

2 | RELATED LITERATURE Our paper contributes to the literature on ALMPs for PwD by analyzing the causal effects of an internship program—a type of intervention that has received limited attention in the literature, as existing studies have focused primarily on other types of ALMPs. In addition, we provide evidence from a developing country context, which remains underrepresented in this literature. Internships differ from other employment programs for PwD, such as supported employment or customized employment, in that they emphasize time-limited, subsidized work experience in firms operating in the open labor market, combined withBERNIELLET AL. | 4 on-the-job learning, rather than ongoing individualized support or tailored job design. This section reviews evidence from other ALMPs, highlighting their impacts and contrasting their design features with those of the PPP. This comparison helps clarify which mechanisms identified in prior ALMP studies are likely to apply in the context of the PPP. It also sheds light on why an internship-based intervention may generate different effects from those documented previously. Previous research has primarily focused on developed countries, with a concentration on Europe and North America. These studies have evaluated various interventions, including quota systems (Krek6 and Telegdy, 2025; Mori and Sakamoto, 2018; Wuellrich, 2010; Lalive etal,, 2013; Malo and Pagén, 2014; Barnay et al., 2019), wage subsidies in general (Angelov and Eliason, 2018; Baert, 2016; Gupta et al., 2015; Deuchert and Kauer, 2017; JiménezMartin et al., 2019; Gupta and Larsen, 2008), short-term wage subsidies conditional on worksite accommodation (Aizawa et al., 2026), vocational rehabilitation (Dean and Dolan, 1991; Yin et al., 2023), supported employment programs (Kinoshita et al., 2013; Dieterich etal,, 2017; Fogelgren etal., 2023), and customized employment (Domenella and Bentolila,

1991; Yin et al., 2023), supported employment programs (Kinoshita et al., 2013; Dieterich etal,, 2017; Fogelgren etal., 2023), and customized employment (Domenella and Bentolila, 2025). The literature on labor market programs in developing countries has expanded substantially in recent years. However, to the best of our knowledge, with the exception of the literature on disability quota systems (Bosch et al., 2021; de Aratjo et al., 2022; Szerman, 2024; De Souza, 2025; Duryea et al., 2023b, 2024; Berlinski and Gagete-Miranda, 2025) and vocational training (Takasaki, 2024), studies evaluating ALMPs in developing countries rarely examine outcomes for PwD in particular. This leaves a substantial knowledge gap regarding which policy approaches can improve labor market prospects for PwD in lowand middle-income settings. Additionally, to our knowledge, the economics literature has not examined the effects of internship programs targeted at PwD." The research on disability quota systems has yielded divergent results. While several studies document moderate positive employment effects (Lalive et al., 2013; Malo and Pagén, 2014; Mori and Sakamoto, 2018; Duryea et al., 2024), others find null or even negative impacts (Wagner et al., 2001; Verick, 2004; Barnay et al., 2019). Even when positive effects are observed, they may partly reflect reclassification of existing employees rather than genuine job creation (Lalive et al., 2013; Duryea et al., 2024). Another strand of research highlights that compliance incentives can shape the effectiveness of quotas. Higher fines, more inspections, and informational nudges have been shown to improve compliance and strengthen employment effects (Wuellrich, 2010; Bosch et al., 2021; Szerman, 2024; De Souza,

highlights that compliance incentives can shape the effectiveness of quotas. Higher fines, more inspections, and informational nudges have been shown to improve compliance and strengthen employment effects (Wuellrich, 2010; Bosch et al., 2021; Szerman, 2024; De Souza, 2025; Duryea et al., 2023b; Krek6 and Telegdy, 2025). Nonetheless, even strong financial incentives may fall short of their goals if other frictions remain unaddressed, such as weak matches between workers and firms (Kreké and Telegdy, 2025). The literature on quotas underscores that the effectiveness of employment policies depends critically on the incentive environment they shape. In this respect, an internship program offers a distinct advantage over quota-based schemes. Because it is not designed to enforce numerical targets, it does not create strong incentives for firms to reclassify existing workers or to displace non-PwD employees. Moreover, matches between firms and beneficiaries are formed in a decentralized and voluntary manner prior to the application process, as we describe in Section 3, reducing the risk of inefficient or forced placements—one of the key concerns highlighted in the literature. This higher-quality matching may help explain the PPP’s notably high internship completion rate. Regarding wage-subsidy programs, research on developed countries has generally not found the expected positive effects on the employment outcomes of PwD. At least two IThere is, however, a related line of research in the medical and rehabilitation sciences evaluating supported or transitional employment models for youth and adults with disabilities (Wehman et al., 2014, 2017, 2020; Ligon and McKelvey, 2025).BERNIELLET AL. |s explanations have been proposed. First, disclosing a disability may create a negative signaling effect, whereby employers interpret disability status as an indicator of lower productivity and higher costs associated with potential workplace accommodations. Correspondence studies show reduced interview callback rates for applicants who reveal a

explanations have been proposed. First, disclosing a disability may create a negative signaling effect, whereby employers interpret disability status as an indicator of lower productivity and higher costs associated with potential workplace accommodations. Correspondence studies show reduced interview callback rates for applicants who reveal a disability (Deuchert and Kauer, 2017; Baert, 2016; I'Horty et al., 2022), with no improvement when wage subsidies are mentioned (Baert, 2016; 1'Horty et al., 2022). This mechanism might not be a concern in the context of the PPP. One reason is that the program is not restricted to PwD: Non-PwD applicants can also participate—although they face a lottery—so employers do not infer disability status from participation alone. Also, firms know in advance when a candidate has a disability and voluntarily choose to work with that individual, limiting the scope for negative signaling or unintended inference at the matching stage. A second reason why wage-subsidy programs often fail to generate large effects is that participants may remain in subsidized positions rather than transitioning to unsubsidized formal employment. This can occur when programs allow subsidies to be granted or extended for long periods, potentially creating disincentives for participants to search for unsubsidized jobs (Angelov and Eliason, 2018). In the case of the PPP, individuals are allowed to apply for subsequent internships, which—given this evidence on wage-subsidy programs—raises a concern about whether extended participation may hinder the transition to unsubsidized formal employment. However, because firms are under no obligation to retain interns after the program ends, they face no costs of terminating the relationship once the internship period concludes. This substantially reduces the firms’ incentive to keep participants in subsidized positions for extended periods, attenuating a key mechanism underlying the limited effectiveness of wage-subsidy programs. In other contexts, when subsidies include clauses that allow firms to employ subsidized workers only for a limited period, employers may become less willing to offer permanent contracts. Additionally, restrictions on terminating open-ended employment relationships

underlying the limited effectiveness of wage-subsidy programs. In other contexts, when subsidies include clauses that allow firms to employ subsidized workers only for a limited period, employers may become less willing to offer permanent contracts. Additionally, restrictions on terminating open-ended employment relationships can make firms favor temporary contracts to avoid long-term commitments, ultimately leading to more precarious employment for PwD rather than the stable jobs these programs are intended to promote (Jiménez-Martin et al., 2019). Similar concerns could arise in the context of the PPP; however, they are likely to be weaker, given that the number of subsidized positions is capped each year by a quota. Regarding the PPP itself, previous research has shown positive labor market effects for the general youth population. Berniell and de la Mata (2017) study the employment outcomes of the general track of the program’s 2012 edition, targeting people between 16 and 25 years old. This version differs from the disability track in that excess demand led to a lottery-based selection of beneficiaries. The authors exploit this random assignment to estimate the program’s impact and find that beneficiaries are 5.1 percentage points more likely than non-beneficiaries to have a registered job one year after program completion. These positive effects persist in the medium term, with a 4.6 percentage point difference observed 4.5 years later. They also find that the effect is stronger for women (5.9 pp. one year after the program) than for men (4.2 pp.) and that impacts are more pronounced among, individuals with lower levels of education and those with higher cognitive skills. All these differences across subpopulations persist in the medium term. Additionally, they provide evidence that the main mechanism behind these results is a reduction in informational barriers rather than an increase in skills formation. Berniell and de la Mata (2017) find no aggregate displacement effects of the PPT, a general concern in the ALMP literature (Crépon et al., 2013; Marinescu, 2017; Abebe et al., 2021).BERNIELL ET AL. 6

aggregate displacement effects of the PPT, a general concern in the ALMP literature (Crépon et al., 2013; Marinescu, 2017; Abebe et al., 2021).BERNIELL ET AL. 6 3 | INSTITUTIONAL FRAMEWORK 31 | First Step Program (PPP) The PPP is a large-scale internship initiative administered by the Cérdoba provincial government that began in 1999, with annual application cycles (interrupted in 2008-11, 2020, and 2022). The program offers participants 12-month part-time paid placements in formally registered private firms. There are no restrictions on the type of occupation or activities that interns must do in the firms. The PPP includes a quota for PwD and has allocated 6,379 internship positions to PwD from 2012 to 2018.2 Initially fully government subsidized, it transitioned in 2016 to a cost-sharing arrangement in which firms contribute between oneand two-thirds of costs, depending on their size. The benefit amount that beneficiaries receive is equivalent to 90% of the hourly minimum wage. Participating firms are not required to make any workplace adjustments and have full discretion over whether to retain interns as employees once the program ends. (They face no costs or obligations if they choose not to continue the employment relationship.) | Eligibility Criteria The eligibility requirements for workers and firms are established by Provincial Law 10,236. Regarding workers, eligible candidates for the general-population track must not have been employed in the six months prior to the application, must have a legal address in the province of Cérdoba, and must be within the 16-25 age group. For the disability track, the age limits are 1649 years, and the program allows for repeated participation. The lack-of-employment requirement is verified through administrative records and thus only conditions on the lack of formal employment. Additionally, while candidates on the general-population track cannot be beneficiaries of other national or local employment or social programs, with the sole exception of the national conditional cash transfer program,

The lack-of-employment requirement is verified through administrative records and thus only conditions on the lack of formal employment. Additionally, while candidates on the general-population track cannot be beneficiaries of other national or local employment or social programs, with the sole exception of the national conditional cash transfer program, Asignacién Universal por Hijo, these restrictions do not apply to the disability track: Those participants can be beneficiaries of benefits established under the national social protection system such as disability pensions. This aspect of the program’s design is particularly relevant, as the literature shows that the structure of incentives surrounding social security benefits for PwD can discourage labor supply among this population (Kostel and Mogstad, 2014; Maestas et al., 2013; Levere et al., 2025). Regarding firms, eligibility requires them to be formally registered—that is, they must have a current business register—and to have at least one formally registered employee. Additionally, internship applicants cannot be related to the owner of the firm. All these requirements are cross-checked by program administrators using administrative data. | Application Process To apply to the PPP, applicants need the endorsement of a firm that is willing to hire them as a PPP intern. The endorsement requirement is intended to show, ex ante, the interest of both parties in participating. In other words, applications reflect already-formed matches between a candidate and a firm. Before a stipulated deadline (usually around the end of May), applicants must submit an enrollment form that collects their personal data and data on the firm they intend to work for (see Figure A.1). In early editions of the program and for paper submissions, the form also carried a signature from a firm representative; for online 2This figure represents approximately 4.6% of all benefits granted by the program, which totaled around 140,000.BERNIELLET AL. |7 submissions, firms must reaffirm their consent by reviewing and accepting the application form pres