Democrats have reason to be encouraged by this morning’s report that that the economy added 243,000 jobs in January and the overall unemployment rate has dropped to 8.3 percent. Of course the Administration should vigorously exercise its bragging rights concerning the monthly report, and especially the overall favorable employment trend of recent months.
For those who want a more nuanced understanding of what the latest employment numbers may mean for the 2012 elections, however, Nate Silver’s “Obama’s Magic Number? 150,000 Jobs Per Month” at his FiveThirtyEight NYT blog may be the most incisive data-driven analysis yet published on the relationship of employment to presidential politics. Silver takes a sobering look at the connection, and explains:
No economic indicator is the holy grail…And there are a number of non-economic variables pertinent to predicting presidential elections — wars, candidate quality and ideology, turnout, scandals and so forth…But if you want to focus a single economic indicator, job growth during the presidential election year — especially as measured by the series called nonfarm payrolls — has a lot going for it.
…Data related to the change in the level of employment have had among the highest correlations with electoral performance in the past. The correlations aren’t perfect by any means. But if you perform a true apples-to-apples comparison (that is, looking at the economic indicators alone rather than muddying them with other sorts of extraneous variables), they do at least as well as anything else in predicting elections, and slightly better than some other commonly used metrics.
Just as important, there are a lot of qualitative reasons to focus on the jobs numbers. They measure something tangible and important. They receive much attention from economists, investors, political campaigns and the news media, and therefore inform the public discussion. They are released every month after only a minimal lag. They are subject to revision, and the revisions can be significant, but they aren’t quite as bad as those for other economic series like G.D.P. or personal income growth. The jobs numbers are calculated in a comparatively straightforward way, and are usually in pretty good alignment with other economic measures. They don’t need to be adjusted for inflation.
Silver then taps some creative methodology to correlate the nonfarm payroll growth rate with the popular vote margin of defeat or victory for the incumbent party in 16 post WWII presidential elections, and he comes to some interesting conclusions, including:
Overall, the relationship between job growth and electoral performance is good but not great…Roughly speaking, there were 10 election years in which you could make a pretty good prediction about the election outcome from knowing the jobs numbers alone: 1948, 1960, and then the eight elections from 1980 onward…In six other elections, you would have needed to look beyond the jobs numbers to come to a good prediction about the outcome.
Citing some of the complicating factors that can cloud his data-driven analysis, such as Eisenhower’s charisma, the Watergate scandal and foreign policy debacles. Regarding a possible Obama-Romney race, Silver argues,
…If we knew nothing else about the election but how many jobs were created between January and October 2012, we would deem Mr. Obama to be a favorite if the economy created more than 107,000 jobs per month and an underdog otherwise. Basically, this would represent job creation at about the rate of population growth.
That’s good news for Obama. The “what have you done for me lately?” factor may signal even better news:
…The public has tended to give greater weight to recent job growth, discounting earlier performance when the trajectory seems positive…If you break it down in more detail, you’ll find that job growth during the third year of a president’s term has a positive effect on his re-election odds, while the coefficients attached to the first two years are negative.
But none of these results are statistically significant or particularly close to it; only job growth during the fourth year of a president’s term has a clear effect.
Silver then factors in presidential approval ratings into his calculations, which indicate:
Mr. Obama’s approval rating is now 46.5 percent, according to the Real Clear Politics average…That isn’t terrible — it’s in the range where Mr. Obama might be able to eke out a victory in the Electoral College — but it’s somewhat below average. From 1948 through 2008, the average president had an approval rating of 52 percent as of Feb. 1 of the election year, according to the Roper Center archives.
If Mr. Obama has an approval rating of 52 percent by November, he will almost certainly win re-election. He’d also be a favorite if he’s at 50 percent. And 48 percent or 49 percent might also do the trick, since at that point Mr. Obama’s approval rating would likely exceed his disapproval rating.
But Mr. Obama is not quite there yet. The surest way for him to improve his approval rating will be to create jobs at a rate that exceeds the rate of population growth.
We can come up with an estimate of just how many jobs this might be if we put a president’s approval rating as of Feb. 1 and the payrolls numbers into a regression equation…I’ll spare you the math (although it is straightforward), but this works out to a break-even number of 166,000 jobs per month — not a huge number, but more than the 107,000 that we had estimated before accounting for Mr. Obama’s middling approval rating.
…If you run another version of the analysis that considers a president’s net approval rating, along with the rate of payroll growth net of population growth, you come up with a break-even number of 151,000 jobs per month.
The Wall St. Journal is predicting an average of 155K jobs being added per month in 2012, notes Silver. But he adds that forecasting track records are “frankly pretty mediocre.” Taking all of the factors into consideration, Silver ventures, ” If payrolls growth averages 175,000 per month, Mr. Obama will probably be a favorite, but not a prohibitive one. If it averages 125,000 per month, he will be a modest underdog.”
Silver’s numbers appear to be sound enough, and 150K jobs per month seems like a good guidepost. Rachel Weiner cautions at WaPo’s The Fix, however, that “No president in recent history has been reelected with unemployment above 8 percent, and analysts suggest it would take growth of between 167,000 and 260,000 jobs a month to get there by November.”
It would be interesting to see what Silver’s analysis could do scaled down to the state level, taking into consideration Geoffrey Skelley’s point at Sabato’s Crystal Ball that “after all, presidents are elected in 51 individual battles (50 states plus Washington, D.C.).” It might be worthwhile to look at needed job growth and margins of victory in the half-dozen most volatile swing states. That could be helpful to Dems in terms of laser-targeting resources.