When the Benchmark Moves with the Policy: Measuring Housing-Access Instruments in Allocation Simulations

An allocation policy changes which agents remain available for later decisions. A benchmark built from that changing pool can therefore move with the policy. I show how this and a second measurement choice, the population over which access is counted, change the ranking of housing-access instruments in an agent-based market calibrated to New York proxies. A ratio identity and an exact two-period example give the mechanism, in which the moving-benchmark ratio rises from 0.80 to 1.00 while realised value falls from 16 to 11. In the calibrated market a contention-linked duty appears to add value above a 2.2 percent offer discount under the moving benchmark, while a fixed benchmark shows a loss at every discount up to 20 percent. A deadline queue produces the opposite error. Following a fixed entry cohort changes measured access gains by factors of 0.7 to 2.5 and reorders the instruments. In a published model of Brazilian poverty policies, counting distinct families instead of monthly payments cuts the advantage of cash transfers over house purchases from about 170-fold to 2.3-fold, and in a published liver-allocation simulator broader sharing widens geographic differences among transplanted patients and new listings while narrowing them among patients already waiting. The certainty discount itself explains little of the access shortfall.