When Does the Preferred Supply-Chain Configuration Change? A Switching-Boundary Decision Analysis of a Smallholder Rubber Case

Static supply-chain optimization identifies a preferred configuration under fixed conditions but does not indicate when that recommendation should be reconsidered as modeled conditions deteriorate. This study develops a Switching-Boundary Decision Analysis (SBDA) layer over three representative linear-programming (LP) instances previously reported for smallholder rubber marketing channels in Musi Rawas, South Sumatra, Indonesia. The published instances were independently reproduced and evaluated through 82 predefined deterministic scenario states across three configurations, yielding 246 LP solves. Stress tests examined configuration-specific and system-wide changes in modeled distribution-profit coefficients, transportation capacity, labor/time capacity, product coefficients, and compound conditions, supplemented by an equal-throughput benchmark. Baseline reproduction confirmed C3, the UPPB-based configuration, as preferred under native-capacity modeled profit, at approximately Rp5.567 million. A C3-specific deterioration of approximately 7.74% produced the first C3-to-C1 crossover under native-capacity profit comparison, whereas the unit-normalized crossover occurred at approximately 16.96%. Under labor/time deterioration, the C3 labor/time constraint became binding at approximately 41.41%, while preference switching occurred later, at approximately 45.94%. Stress testing also identified meaningful optimal-allocation changes: a 50% labor/time-capacity reduction in C2 changed the solution from all-A to a mixed 230-A/300-B allocation, while equal-throughput normalization changed C3 to a mixed allocation but preserved the C3C1C2 ranking. SBDA therefore distinguishes performance deterioration, constraint activation, solution-structure change, and configuration switching, and connects these states to a parameterized Stable-Watch-Re-evaluate decision layer. The numerical boundaries are conditional on reconstructed historical instances and should not be interpreted as population-level or current-market thresholds. They provide analytical decision points rather than universal prescriptions for supply-chain management in practice