Evaluating the impact of the Ghana Productive Safety Nets Project (GPSNP)'s productive inclusion program

The Complementary Livelihood and Assets Support Scheme (CLASS) facilitates income generation activities (IGAs) for 30,579 beneficiaries from the Labor-Intensive Public Works (LIPW) program and the Livelihood Empowerment Against Poverty (LEAP) cash transfers program. Under CLASS, the Ministry of Local Government, Decentralization and Rural Development (MLGDRD) will first carry out market feasibility studies of various livelihoods projects and through its zonal offices, the ministry will facilitate the selection of communities and households to undertake productive activities in selected projects. Selected individuals will benefit from life skills and micro-enterprise skills trainings; provision of start-up lump sum cash grants; and mentoring and coaching.

The Complementary Livelihood and Assets Support Scheme (CLASS) facilitates income generation activities (IGAs) for 30,579 beneficiaries from the Labor-Intensive Public Works (LIPW) program and the Livelihood Empowerment Against Poverty (LEAP) cash transfers program. Under CLASS, the Ministry of Local Government, Decentralization and Rural Development (MLGDRD) will first carry out market feasibility studies of various livelihoods projects and through its zonal offices, the ministry will facilitate the selection of communities and households to undertake productive activities in selected projects. Selected individuals will benefit from life skills and micro-enterprise skills trainings; provision of start-up lump sum cash grants; and mentoring and coaching.

Per guidelines prepared by MLGDRD, district authorities identify and rank all communities with LEAP and LIPW households based on the district’s poverty profile. Each district is then required to select a minimum of 4 and a maximum of 7 communities. Preparatory activities and community fora are then set up prior to beneficiary selection. Information and sensitization campaigns are carried out by the District CLASS Implementation Teams (DCIT) under the guidance of the Zonal Office. The sensitization programs are aimed at informing all potential beneficiaries about the purpose and objective of CLASS and its implementation modalities. Within selected communities, at most 1 adult per household will be eligible only if they are LEAP or LIPW beneficiaries; and that their poverty levels are within the extremely poor category. Individuals indicate three preferred IGAs as part of enrolment by the Rural Development Coordination Unit (RDCU). Prospective beneficiaries must either be interested in commencing a new IGA or seeking additional investment for an existing IGA that is on the DA’s positive list.

Once beneficiaries are selected and enrolled, they are put into groups of 10-25 across communities based on their IGAs. These groups are to be formed for the purposes of training as beneficiaries will not be compelled to work together on collective income generating activities (although they may choose to do so of their own volition to enjoy the benefits of economies of scale). Upon completion of all required trainings and other preparatory activities, each beneficiary will be provided with a lump sum cash grant up to the cedi equivalent of US$200 to assist beneficiaries in setting up, operating, and/or expanding their enterprise activities. The cash grant ceiling is informed by national and international experiences. The cash grants will be disbursed in two to three tranches with the understanding that the release of subsequent tranches will be conditional on satisfactory utilization of the previous tranche(s) and the completion of specific milestones as determined by the supervising DCIT and the approving authority. All beneficiaries who receive the first tranche will be required to either enroll with a Village Savings and Loans Association (VSLA) or show evidence of savings to qualify for subsequent tranches. All CLASS grants will be disbursed through an electronic payment platform. In this regard an e-payment Service Provider (with offline and biometric capabilities) will be engaged by MLGDRD.
The intervention’s theory of change assumes that immediate outputs from the provision of start-up grants and skills training include enterprise formation, and that outputs from coaching and mentoring include expanded sales and increased market share. The overall development objective or outcomes from the project include (i) increased productivity and incomes; (ii) increased local employment; (iii) increased health and education expenditures; (iv) women’s empowerment; and (v) falling poverty, among other things. The novelty of the productive inclusion intervention lies not in its constituent components, but in the way that components are combined in a holistic way to lift households out of extreme poverty while ensuring they do not slip backward from shocks along the way.

2021-11-30

2021-12-31

Micro-enterprise productivity and incomes; Local employment; Household expenditure in health and education expenditure; and Women’s empowerment.

Since the intervention offers cash grants to invest in income generating activities (IGA) such as goat rearing, individual enterprise productivity will be measured through marginal profit from each dollar invested in an IGA. Beneficiary and household incomes will be measured through self-reported surveys and monitoring reports from the implementing agency. Local employment will be measured through comparisons between baseline and endline, non-household employment shares of the average micro-enterprise. Women’s empowerment will be measured through participation in community leadership and household decision-making before and after the intervention.

The intervention will be implemented in three phases. Phase one will target 136 (out of 455) communities to receive the treatment while phases 2 and 3 each target 153 and 206 communities, respectively. Phase 2 and 3 communities will serve as the evaluation communities from which baseline community, individual, and household surveys will be conducted. Per sample and power calculations, the experiment requires a minimum of 180 communities (and 15 individuals per community). As such, 180 communities are randomized into the experiment - 90 treated (batch 2) and 90 control (from batch 3). The baseline survey is conducted from all 180 communities (2,700 individuals). After the baseline surveys are completed, the intervention can be initiated in 90 of them (batch 2) towards the end of 2021. The remaining 90 experimental communities are left untreated until at least the end of 2022 (or 1 year after the end of the treatment of batch 1 communities).

Having determined the minimum required cluster and sample sizes, we implemented the randomization and sampling as follows:
• We started off with 311 communities for batches 2 and 3 that the government has gathered necessary data for and shared with the Bank.
• We pooled these 311 communities together. We then removed some communities for various analytical and survey reasons from consideration for the evaluation sample.
• After eliminations, we obtained a sample of 276 communities that make up our potential evaluation sample.
• Taking the potential evaluation sample, we randomly drew 180 communities from 276, and then randomly divided the 180 into “batch 2-treated” (90 communities) and “batch 3-untreated” (90 communities).
• The communities not in the evaluation sample, 131 of them, were marked as “unbatched.”
The experimental communities are spread throughout the study region. Within each of the 180 communities, 15 main interviewees are drawn at random and another 5 are drawn as replacements. The intervention will be implemented in three phases. Phase one will target batch 1 communities to receive the treatment while phases 2 and 3 each target 153 and 206 communities, respectively. 180 phase 2 and 3 communities will serve as the evaluation communities from which baseline community, individual, and household surveys will be conducted. The baseline survey is conducted from all 180 communities (2,700 individuals). After the baseline surveys are completed, the intervention can be initiated in 90 of them (batch 2) towards the end of 2021. The remaining 90 experimental communities are left untreated until at least the end of 2022 (or 1 year after the end of the treatment of batch 1 communities).

Randomization done in STATA 16.

Community randomization into treatment/control areas.

Yes

180 communities

2,700 households

90 treated communities (1350 households), 90 control communities (1350 households)

Using the latest round of the Ghana Living Standard Survey (GLSS 7), we estimate the control-group mean household food consumption in the northern regions of Ghana to be GHC349.09/month with a standard deviation of GHC164. After the intervention, we expect our sample to detect a minimum of 6 percent increase in household food expenditure among households in treated communities. The intra-cluster correlation used corresponds to 0.016, which was used in an impact evaluation of a related program in the same region in 2016 (see pre-analysis plan for details).