My last post talked about private-sector job growth (or decline, in one case) under different presidents. But since the presidents came to office under different circumstances, that's not really a fair comparison. To adjust for circumstances, I estimated an "autoregressive" model, in which job growth is predicted from past job growth. Over the period from 1950-2002, the estimated model is
g=.03356+.1904x+.2676y+.1418y+.1402z, where g is predicted percentage change in private yemployment, x is actual percentage change last month, y is percentage change two months ago, and z is percentage change three months ago. Even though the model includes only the last three months, the more distant past still has an effect, because it influenced the values for the recent past. In fact, the effects of the past never completely disappear, they just fade away. I estimated the total predicted job growth over each president's time in office, given the job growth in the three months before he took office and subtracted that amount from the actual amount. Then I divided that figure by the length of time he was in office to put it in annual percentage terms. The resulting figure can be understood as the rate of job growth relative to what would have been expected.
Two presidents come out substantially better under the adjusted rankings: Obama and Kennedy. Of course, the adjusted rankings aren't completely fair as a "grade"--employment changes are affected by circumstances beyond the president's control, and there are other important economic conditions. But the record of private-sector job growth under Obama is not unusually weak by historical standards--the recession was just unusually deep. Also, the difference in private-sector job growth between Democratic and Republican presidents is even stronger when you adjust for the conditions they faced when they took office.