Daniel E. Ho
In a substantial contribution to positive political economy, Rogowski and Kayser (2002, p. 526) finds that "systems of proportional representation . . . systematically advantage producers and disadvantage consumers." I find that there is no evidence to sustain that conclusion. The original study extrapolates severely due to the fact that proportional representation countries are systematically different from majoritarian district countries in their background characteristics. Accounting for these differences by matching on the propensity score forces us to discard observations that severely extrapolate from the data, yielding estimates of such high variance that we cannot find evidence for the price-level effect. The only way to reassess the hypothesis without bias or implausible assumptions is to gather a larger dataset, which by including non-OECD democracies increases potential observations. Yet even with this data the price-level effect remains undetectable. The conclusion of a price-level effect thereby rests on modeling assumptions that are theoretically and empirically unjustified.