Overview
This series explains why positive average effects from formal credit do not provide a universal lending rule for small and medium-sized enterprises. The IDB review reports about 12% average employment gains and about 18% average sales and profit gains across varied impact evaluations, while an Egyptian randomized comparison of larger enterprise loans among approved Tamouh applicants at the Alexandria Business Association finds a small average impact but meaningful differences in predicted response. It moves from population averages through sample-split prediction and corrected estimates to the limits of mechanisms, transport, fairness, and market spillovers. Its conclusion is conditional: prediction can motivate prospective allocation tests, but it is not an explanation, harm detector, or proven optimal policy.
Learning outcomes
- Formal credit improves SME employment, sales, and profits on average in the reviewed evidence, but effects vary by program and setting.
- In the Egyptian experiment, larger enterprise loans had small average impacts but materially different predicted responses across approved applicants.
- The corrected sample-split analysis supports meaningful heterogeneity, while the bottom group’s negative estimate is not individually statistically significant.
- A predictive ranking can inform prospective allocation tests, but it is not a causal explanation, harm detector, or universal policy.
Keywords
Preview every chapter · Unlock full text, audio, and video from chapter 4
PlansFAQ
What does the evidence say about formal credit and SME performance?
The IDB review reports about 12% higher employment and about 18% higher sales and profits on average across its assembled impact-evaluation evidence. These are population-level means across varied programs, settings, and outcome horizons, not forecasts for every firm.
Why is a positive credit average not an allocation rule?
A positive credit average combines firms with different responses, so it cannot identify which applicant will benefit from a larger loan or face greater repayment exposure. Program design, targeting, local financial architecture, and outcome horizon also change what the average means.
What did the Egyptian larger-loan experiment find?
Among approved applicants to the Tamouh program at the Alexandria Business Association, relatively large enterprise loans had small average impacts but materially different predicted responses. The corrected analysis found +6,534 EGP for the top predicted profit group and −4,571 EGP for the bottom, with the top-versus-bottom difference remaining statistically strong at p = 0.002; the bottom estimate itself was not individually statistically significant.
Can a predictive credit score identify who will benefit?
It can identify groups with different predicted responses in the Egyptian study, but it cannot identify an individual beneficiary with certainty. The ranking still needs prospective validation under the intended contract and market, and it does not explain causation or prove an optimal policy.