This New York Times blog post asks “Can We Judge Economic Success? Perry Versus Obama.” The author rightly points out that distinguishing competence from luck is apparently the critical task of voters, also emphasizing repeatedly that this is a problem of causal inference.

Even if we agree that the person in charge deserves some credit (or blame) for the state of the economy, it doesn’t follow that different policies would have produced a significantly different result. Obama’s stimulus and bailout programs may have helped reverse the decline in gross national product, but that doesn’t mean that conservative tax-cuts and relaxed regulation would not have had a similar effect. Conversely, Obama’s policies might have worked in Texas as well as Perry’s.

Of course, it is often easy to see that a particular policy produced a short-term economic effect: Obama’s cash-for-clunkers program caused an increase in car sales. But it is much more difficult to determine what effect an administration’s overall policies had on long-run economic health. Expert economic analysis might in principle produce some creditable conclusions. But there is little chance of a consensus among economists on such an analysis, particularly when political stakes are high. We ordinary citizens are left with no reliable professional guidance.

The rest of the post, I think, is a little too committed to the idea that luck plays a disproportionate role in determining performance.

(So higher political stakes make for poorer research? Hm…)

The very same insight is made in this 1993 paper by economist Alberto Alesina and political scientists John Londregan and Howard Rosenthal:

Administrations vary in their degree of “administrative competence.” For a given rate of inflation (expected or unexpected), a more competent administration is likely to produce more growth than a less competent one (Persson and Tabellini 1990). Since voters prefer more competence to less, elections will turn on not only partisan preferences but also efficiency arguments. However, a voter can not observe “competence” directly but only its effect on the economy. Since the economy is also affected by technological innovations, oil price changes, wars, and other matters that have little to do with administrative competence, voters cannot immediately distinguish competence from “luck.” A “rational”retrospective voter can only use available information to make a forecast of the incumbent’s post electoral competence. This forecast, as we shall show, leads economic growth to affect electoral results in a manner distinct from “naive” retrospective voting, where no attempt is made to distinguish good luck from good government.