| Sumario: | Value chain interventions are rarely evaluated as rigorously as interventions in agricultural production or health.
This is due to various reasons, including the intrinsic complexity of value chain interventions, intricate contextual
support factors, presence of multilevel system actors, constant adaption to market and nonmarket forces and the cost
associated with conducting an evaluation. This paper discusses a range of approaches and benchmarks that can guide
future design of value chain impact evaluations. Twenty studies were reviewed to understand the status and direction
of value chain impact evaluations. A majority of the studies focus on evaluating the impact of only a few interventions,
at several levels within the value chains. Few impact evaluations are based on well-constructed, well-conceived
comparison groups. Most of them rely on use of propensity score matching to construct counterfactual groups and
estimate treatment effects. Instrumental variables and difference-in-difference approaches are the common empirical
approaches used for mitigating selection bias due to unobservables. More meaningful value chain impact evaluations
should be prioritized from the beginning of any project and a significant amount of rigor should be maintained;
targeting a good balance of using model-based and theory-based approaches.
|
|---|