A Cost-Modeling Tool for a Meta-analysis of Entrepreneurship Interventions

Study Context

The inconsistent reporting of cost data across randomized controlled trials (RCTs) in global development research has prevented the creation of credible cost-effectiveness estimates for meta-analyses and other efforts to systematically compare the costs and effectiveness of development interventions. 

Study Design

Leveraging its extensive costing expertise, CEGA developed a framework for reconciling disparate and often incomplete cost information, and piloted a comprehensive model to consistently and reliably predict the costs of entrepreneurship and business training interventions implemented in low- and middle-income country (LMIC) settings. The project focused on improving the extraction of relevant contextual information from the RCTs, utilizing costing vignettes, globally-available data sets, and study-reported data to analyze and predict the costs of the training programs. 

As of September 2025, CEGA has created the first iteration of a cost prediction tool in collaboration with the Global Poverty Research Lab (GPRL) at Northwestern University. Together, we have piloted our cost prediction model using a data set of 59 business training, consulting, graduation, and mentoring interventions included in a meta-analysis of 31 studies published after 2010. Our modeling approach builds individual cost components first and then combines them to arrive at a harmonized definition of cost per participant. We then predict cost per participant at the treatment-arm and program activity level, even for cases where costs were unreliably reported (or omitted). This bottom-up approach encourages more useful and accurate cost predictions, most especially by building-in activity costs for each intervention across settings.

Results and Policy Lessons

Forthcoming deliverables include the development of a general methodology for costing interventions included in past RCTs; creation of a cost prediction model to explain observed costs and to predict costs using a sample of 31 published randomized controlled trials (RCT) of business and entrepreneurship training programs; and a narrative synthesis of the model fit and residual analysis.