Earlier this year, just before the government’s January employment report came out, I wrote that creating 150,000 jobs per month seemed to be a rough breakeven number for President Obama. If payroll growth came in above that number, he looked to be a favorite for re-election and an underdog otherwise.
The January employment report turned out to be a very good one, initially reporting that 243,000 jobs were created that month. (The figure has since been revised up further, to 275,000 jobs.) The February jobs number — originally 227,000 and now 259,000 jobs after revisions — was likewise quite strong.
But figures in the last two months have been more sluggish. The March jobs report initially showed that 121,000 jobs were created — although like other recent reports, it has been revised upward, and now shows a gain of 154,000 jobs. And the April jobs report, which was released on Friday, showed a gain of only 115,000 jobs.
If you look at the four months as a whole, Mr. Obama is still somewhat ahead of pace. An average of about 200,000 payroll jobs were created in the first four months of the year. What that means is that the May through October jobs reports would only need to show about 120,000 jobs being created per month for Mr. Obama to reach the 150,000-job “magic number” throughout the election year.
But let’s make this analysis more forward-looking. The jobs numbers for the past five months or so, taken as a whole, have been fairly decent, especially relative to where the figures had been throughout most of the last three years. The concern economists have, however, is that the March and April figures were less solid than those in December, January and February, which could presage another summer slump in job growth, as occurred in both 2010 and 2011.
My original analysis was based on part, however, on what Mr. Obama’s approval rating had been prior to the January jobs report being released. Presumably, if that ahead-of-the-pace jobs creation in January and February has helped him, it would already be reflected in his approval numbers by now.
Mr. Obama’s approval rating is in fact modestly higher than it was several months ago. In the Real Clear Politics average, it now averages 47.9 percent, versus a 47.4 percent disapproval rating.
In the table below, I’ve shown the annualized rate of nonfarm payrolls growth from May through October of the election year for incumbent presidents running for re-election since 1948. As in the original analysis, the rates are net of population growth. Specifically, they are net of the growth rate in the overall adult population and not the growth in labor force, which is less exogenous since the number of people seeking jobs can be affected by economic conditions.
(The unemployment rate uses the latter definition, which is why it declined to 8.1 percent last month despite sluggish job creation. Although the unemployment rate is widely reported upon and can therefore plausibly affect a president’s re-election chances, I consider it less reliable as an economic benchmark and therefore prefer to work with versions of the payrolls numbers when given the choice.)
Because these job growth rates are not commonly reported, I’ve used a little algebra to translate them into something I call “equivalent monthly jobs.” This represents how many jobs would be added or subtracted per month using today’s figure of 133.0 million payroll jobs as a baseline and adult population growth of 0.8 percent, as it has averaged so far this year.
For instance, in 1984, Ronald Reagan’s administration created an average of about 306,000 jobs per month from May through October. However, that is equivalent to 426,000 in today’s terms because the population is now larger.
In the table, I’ve also listed the incumbent’s net approval rating as of May 1. Between an incumbent’s approval rating as of May and the job growth for the rest of the year, you usually get a pretty good story about what his ultimate fate will be.
Harry Truman was rather unpopular at this point in 1948, for instance, but job growth was quite robust in the second half of the year, helping him to win re-election. Job growth was very marginal in the second half of 1956, meanwhile, but Dwight D. Eisenhower was starting from such a high baseline of popularity that it didn’t matter much. In 1980, Jimmy Carter was already somewhat unpopular, but a sharp loss of jobs in the second half of the year made things much worse for him, leading to a landslide defeat. And in May 2004, George W. Bush’s approval ratings were about where Mr. Obama are now, but just enough jobs were created for the rest of the year to help him get over the finish line.
If you put these figures into a regression equation, you can use them to formulate a projection about Mr. Obama’s fate going forward. For instance, if 150,000 jobs are created in an average month from now through October, the regression equation projects a very close result, showing a victory for Mr. Obama in the popular vote by about 1 percentage point. That would translate into about a 60 percent chance of his winning the popular vote given the historical error in the formula when fit to past elections.
This analysis does suggest that Mr. Obama’s breakeven number on job creation is now slightly lower than when we had calculated it a few months ago — about 125,000 jobs per month rather than 150,000.
Still, I don’t want to create the impression that the relationship is all that precise. Certainly, if the economy creates 200,000 or more jobs per month throughout the rest of the year, as it did in January and February, then Mr. Obama is likely to be in pretty good shape. And if it averages 75,000 or fewer jobs per month — that’s about the point at which the rate of job growth will lag that of population growth — he is likely to be in some trouble.
Anything between about 100,000 and 175,000 jobs per month, however, should leave both Mr. Obama and Mitt Romney with ample room to debate the nuances of job creation and the rest of the economy.
One thing you’ll notice, however, is that unlike in the original analysis, I excluded years in which the incumbent retired or was term-limited and did not run for re-election. Job growth has been a much less reliable guide to the general election result in these years. For instance, very robust job growth in the second half of 1952 did not help Truman’s Democratic successor, Adlai Stevenson. Nor did excellent job growth in late 1968 help Hubert H. Humphrey.
There is a tendency in analyses like these to take a one-size-fits-all approach that lumps all incumbent parties together, whether or not the incumbent president himself was actually running for re-election. So, for instance, Richard Nixon running in 1960 is treated as Eisenhower’s surrogate, equivalent to Mr. Carter in 1980 when he was running for himself.
There are arguments for and against doing this, but after having looked at this sort of data more thoroughly as I work to get our general election model ready, I question whether this is the right approach. The dynamics when a true incumbent is running seem to be a little different than when he retires, and there is slightly more ambiguity about who gets credit or blame for economic performance.
Anyway, if you do re-run the numbers based on all years, and not just those where there was a true incumbent running, it suggests that Mr. Obama’s breakeven number is lower — about 75,000 jobs per month rather than 125,000. It also suggests that the job figures will make less difference either way — for instance, even creating 225,000 jobs per month would only make Mr. Obama about an 80 percent favorite to be re-elected under this version of the model, versus about 90 percent under the version that looks at true incumbents only.
Intuitively, that doesn’t seem right to me. Given the importance attached to the jobs report and its salience as an overall indicator of economic health, it seems likely to be that it will be a fairly significant factor. Moreover, it seems that 75,000 jobs per month would not be enough to give Mr. Obama break-even odds at another term; he’d need a lot of other factors to go his way to win re-election under those conditions.
At the same time, it should be remembered that while financial reporters tend to make a big deal of any “miss” in the jobs report — for instance, that the April report came in at 115,000 jobs rather than about 160,000 as economists had projected — these differences are fairly minor in the grander scheme of things: about 35,000 jobs in a country of about 240 million adults. Moreover, these types of “surprises” are common. On average, economists’ initial estimates of job creation miss by about 70,000 jobs in one direction or another.
Also, the revisions to jobs figures can be fairly substantial. For instance the August 2011 jobs number — initially reported at zero net job creation — has since been revised upward to a gain of 85,000 jobs. Historically, it has often been the case that initial estimates of job growth come in too low during recoveries (this has been happening recently; figures have been revised upward for the past 10 consecutive months) but underestimate the degree of deterioration in the jobs market when we are in an economic slump. (The chart below shows the magnitude of revisions from January 2000 through July 2011.)
In the big picture, the job creation numbers through the first four months of the year have been helpful on balance to Mr. Obama. However, the notion that the economy would just zoom forward, giving him a relatively easy path toward re-election, is now looking less likely given the April numbers and some negative “surprises” in other economic figures.
At the same time, these are far from recessionary figures, and some of the macroeconomic risks stemming from the European debt crisis or oil price instability in the Middle East seem to have lessened.
For the time being, this election looks to be roughly on the 2004 track, where a president with very average approval ratings and very average job growth was in a very close contest all year, and perhaps just put over the finish line by running a strong campaign. Still, economists have a long history of badly underestimating the chances of both positive and negative economic surprises.