Rural agro-dealer input-credit blacklisting and smallholder fertiliser substitution in Tanzanian wards: a regression-discontinuity analysis of credit-scoring thresholds

No observational data are used or claimed. The analysis characterises the design under a declared variance structure with between-cluster SD = 6.0, residual SD = 4.5, and intraclass correlation = 0.640. Power was estimated across 2000 design cells for a before-after-control-impact estimator with clusters of 8, 16, 24, 32 and effect sizes of 3, 5, 8 percentage points. The design requires 54 clusters to detect a 5 percentage-point effect at 80% power. The intraclass correlation of 0.64 imposes a binding constraint on ward-level randomisation and that the before-after-control-impact estimator recovers credible threshold effects only when the number of clusters exceeds 24. The contribution is a design-based account of when regression-discontinuity evidence on input-credit blacklisting can be trusted, and a demonstration that the minimum detectable effect at 80% power falls steeply as clusters rise from 8 to 32. The paper concludes that threshold-based blacklisting studies in Tanzanian wards should be powered for at least 54 clusters if they are to distinguish a 5 percentage-point substitution response from noise.