Tag Archives: politics

Revisiting Our February Jobs Prediction

3 Nov

By NATE SILVER

 

Friday’s jobs report was a reasonably strong one, economically speaking. The economy added 171,000 jobs in October, according to the government’s survey of business establishments. In addition, estimates of jobs growth were revised upward for August and September.

The unemployment rate, which is calculated through a separate survey of households, ticked up to 7.9 percent. But this was because it was estimated that more workers, 578,000, entered the labor force in October, outweighing what it said were 410,000 people who found jobs.

Is the report good enough to have an impact on the waning days of the campaign? There is a dispute in the political science literature about whether voters react to underlying economic conditions, or rather, to the news media’s coverage of the economy.

If it’s the real-world conditions that count, the actual act of the government publishing the jobs figures is unimportant. People will already have observed local economic conditions and incorporated them into their decision of who they might vote for.

Indeed, measures of subjective economic attitudes, like consumer confidence, have shown more strength in the last two months, suggesting that the public has already “priced in” the idea that the economy is slowly returning back to normal.

If the news coverage matters instead, and perceptions are more important than the reality on the ground, then a report with good headline numbers (for example, the one last month, which had unemployment dropping to 7.8 percent) will have more impact than one where the strength is in the fine print (as it was in this latest report). Furthermore, it could matter which other stories the economic news is competing against. Right now, we’re already in a very busy news cycle, between Hurricane Sandy and the end of the presidential campaign.

These ideas are not mutually exclusive, of course. It’s very likely that both the reporting on the economy, and actual conditions on the ground, matter to some degree.

In this case, however, neither would be suggestive of much last-minute political impact. Subjective perceptions of the economy, and the statistical data we have about it, already seemed to have been roughly in line with one another. And the headline numbers in this jobs report were not quite strong enough to break through and dominate the news coverage over the final weekend.

This may be a case of two wrongs making a right. This report is a little stronger than it appears, but reporters also tend to over-interpret the meaning of the monthly swings in the jobs numbers, which are subject to a fair amount of statistical error.

Still, this month’s numbers are part of a longer-term story. With the relatively strong October numbers, and the upward revisions to August and September, the economy has now created an average of 157,000 jobs per month so far this year. This may be a slight underestimate, in fact. The government has announced, but not yet officially incorporated into the numbers, its estimates of annual benchmark revisions, which would add to the jobs numbers in January through March.

Those revisions would bring average jobs growth throughout the year to about 165,000 jobs per month.

In February, just before the jobs figures for January were announced, we published a simple statistical projection that forecast how the election would turn out based on President Obama’s approval ratings at that time, and the jobs numbers over the rest of the year.

Our conclusion was 150,000 represented the over-under line. If more than 150,000 jobs were created per month, then Mr. Obama would be a favorite for re-election, other factors being equal. Below that threshold, then he would be an underdog.

The actual numbers have come in just slightly ahead of the 150,000-job benchmark, it turns out, suggesting that Mr. Obama might be favored to win re-election, but is not a lock to do so.

And that’s pretty much what we’re seeing in the polls. There is not much time for the polls to change, and if they are right, Mr. Obama will win the Electoral College. But Mr. Obama’s advantage is marginal enough that Mitt Romney could win if the polls miss a couple of points high on Mr. Obama in the swing states.

What this jobs report perhaps does do is remove Mr. Romney’s best opportunity to shift the polls in his direction before the election. We’re very likely to wake up on Tuesday knowing that the polls have Mr. Obama as the Electoral College favorite. We’ll have to wait until Tuesday night (or longer) to know whether the polls have it right.

But if they do, and Mr. Obama wins narrowly, the outcome will be broadly in line with what the jobs numbers predict.