Estimating the Impacts of Volunteer vs. Paraprofessional Community Worker Interventions among Conditional Cash Transfer Recipients in the Dominican Republic

The project interventions are administered within the framework of the national conditional cash transfer system and include i) providing cash transfers subject to fulfillment of educational and health conditions and ii) monthly household visits from volunteer or paid community workers trained in a 12 month social-familiar curriculum. Cash transfers are provided in the form of earmarked vouchers which may be spent through the use of debit cards and may may be used to purchase permitted food and educational items from a national network of retail shops. Some households will only receive the cash transfers. Other households will receive both the cash transfers and monthly home visits. The monthly home visits will be conducted by community workers who are assigned to 20-50 households each, consistent with the current operations of the conditional cash transfer program. Home visits are designed to provide beneficiary households with social-familiar support, to provide information about public resources, to encourage fulfillment of program conditionalities, and to provide 12 months of instruction on a wide range of wellness, health, citizenship, and job market preparedness topics through a structured curriculum specially designed to be used during the home visits.

Community workers providing home visits were recruited through three different methods each of which constitutes its own intervention. One arm of the study recruited candidates through posting a job announcement for a part-time paid position on the Facebook page of public colleges in the target areas. These candidates were screened, assigned to geographic and personality characteristics strata, and randomly select to be offered the job from a pool of candidates within the same strata. A second arm of the study recruited volunteer (not paid) candidates in a method designed to resemble the status quo process as much as possible. Candidates were referred for volunteer community worker positions in specific locations by supervisors or other more senior managers from the regional CCT operations staff. Candidates were then screened and ranked according to cognitive characteristics and randomly selected for job offers using a pair-wise matching randomization among the top two candidates in each location. A third arm of the study recruited volunteer (not paid) candidates through public advertisements for the position distributed by hand as posters and flyers in targeted neighborhoods. Candidates were ranked by cognitive characteristics and randomly selected for job offers using a pair-wise matching randomization among the top two candidates.

From the perspective of estimating the impacts of working as a community worker on the lives of candidates for the new community worker positions, the interventions are defined as being offered a volunteer or paid job.

2016-10-01

2018-03-31

Key beneficiary household outcomes of interest can be divided into operational outcomes, knowledge outcomes, and behavioral outcomes.

Operational outcomes: the rate of compliance with conditionalities by cluster every quarter/half year (depends on conditionality), percentage of cash transfer used every month, coverage under public health insurance, and personal knowledge of community worker.

Knowledge Outcomes: knowledge of content of social-familiar educational curriculum, location of local health clinics and schools, requirements and prohibitions about CCT program participation, identity of local politicians and public servants, availability and content of vocational training programs.

Behavioral Outcomes: level, diversity, and quality of food consumption, education and health related expenditures, level of credit and savings, entrepreneurship, labor market participation of household head and other household members, political and community volunteering and civic involvement, use of public health services, participation in vocational training programs, experience with and cause of recent household shocks, in/outmigration among household members, and subjective well-being.

Among community workers, key outcomes include:

Operational outcomes: the fraction of households visited every month and personal knowledge of beneficiaries.

Knowledge Outcomes: knowledge of content of social-familiar educational curriculum, location of local health clinics and schools, requirements and prohibitions about CCT program participation

Behavioral Outcomes: Personal progress and performance in university (if applicable), income and other employment for self and other members of household, participation in other community and political activities, volunteerism and leadership, career aspirations and orientation to public service, views on poverty and government, enrollment of family members in the CCT program, and subjective well-being.

The sample includes 21,679 households organized into clusters of 20-50 households across 5 experimental arms each with 107-108 clusters. One experimental arm will remain a pure control that does not receive either intervention of cash transfer or monthly visits from community workers. A second arm will received only cash transfers and no monthly visits. The remaining arms will receive both interventions, but will differ in the means by which the community workers conducting the home visits were recruited and incentivized. The third arm will receive visits from paid community workers recruited from universities. The fourth arm will receive visits from volunteer community workers recruited through referrals by existing program staff. The fifth arm will receive visits from volunteer community workers recruited through public advertisement. The sample of households includes 12,619 new beneficiary households and 9,060 existing beneficiary households.

A second experiment looking at impacts among the community workers themselves is designed within the overall experiment. Among the pool of candidates recruited to work as community workers for each of the 322 community worker clusters (paid and volunteer), individual candidates were randomly selected to receive job offers.

Randomization conducted in Stata.

Randomize assignment of households to experimental arm by cluster. Randomize selection of community workers from among pool of qualified candidates at individual level.

Yes

537 clusters of CCT beneficiaries

21,679 households

107 control clusters, 107 paid university recruited community worker clusters, 107 referred community worker clusters, 108 publicly recruited community work clusters, 108 cash transfer only clusters