President Obama’s probability of winning the Electoral College fell slightly in Friday’s FiveThirtyEight forecast, from 83.9 percent to 82.7 percent on Thursday.
Although Friday’s polls were a bit more equivocal than on some recent days, the shift mainly came because of a decline in the FiveThirtyEight economic index, which continues to have some influence on the forecast.
A government report on Friday showed a decline in real personal income in August, one of the economic data series that the forecast model uses. Furthermore, the July personal income figures, which had initially looked quite positive, were revised downward to show slower income growth.
Real personal income is up 2 percent from one year ago, a below-average number, but one that roughly matches other economic figures. However, it had been flat to negative for much of 2011, and it has increased only at about the growth rate of the population over the whole course of Mr. Obama’s term.
Another economic data series that the model uses, personal consumption expenditures, was also published on Friday and showed sluggish growth in consumer spending after adjusting for inflation.
Numbers from the manufacturing sector have also been poorer lately, and this month’s monthly jobs report was mediocre.
And yet, consumer confidence was up sharply this month across several surveys, reaching a four-month high in The Thomson Reuters/University of Michigan’s monthly measurement of it.
What accounts for the discrepancy? Some of it might be in the timing: the reports on consumption and personal income covered August, whereas the consumer confidence surveys are quicker to update and instead reflect numbers for September. And there have been some recent positives in the economic data, like tentatively better numbers from the housing sector.
Politics undoubtedly also plays a part, however. We often talk about the important role that the economy plays in the election. But Americans perceptions of the economy are also colored by their political views.
Gallup, which tracks numbers on economic confidence daily, found a substantial increase in consumer confidence that was unmistakably timed to the Democratic convention. Most of the increase, however, was caused by Democrats, who now view the economy much more positively. The numbers also improved modestly among independent voters, but there was no improvement among Republicans.
It seems that Mr. Obama and Bill Clinton were able to persuade some voters in their speeches at the convention that there are silver linings in the economy. It is the most persuasive explanation for how subjective perceptions of the economy improved at a time when objective measurements of it had worsened.
But could improved consumer confidence actually translate into improved consumer spending, which could help stimulate the economic recovery? Or do measures of consumer sentiment become unreliable predictors of actual economic behavior once we’re this deep into an election cycle?
Some academic studies of this question find that improved consumer confidence actually may affect consumer behavior — even if the initial impetus for the change was rooted in political factors. Democrats might be going out to dinner a few more times this month — even if Republicans don’t.
On the other hand, Gallup’s tracking of self-reported consumer spending has not seen much of a change since the Democratic convention. Self-reported assessments of spending are not always reliable. But we’ll have to wait several more weeks until September data on retail sales is available, giving us a harder measure of how consumers are actually behaving.
These ambiguous relationships between political attitudes, economic attitudes and actual economic behavior are complicated. It isn’t necessarily so straightforward to say what constitutes the economic “fundamentals,” when those economic variables may be influenced in part by the political attitudes of consumers and businesses, along with fiscal and monetary policy.
However, I am especially wary of using sentiment-based measures like consumer confidence in elections analysis, and forecasting models that have done so have had extremely poor results when applied to real data. If you had an elections model based on horse-race polls and consumer confidence right now, you’d run the risk of double-counting Mr. Obama’s gains, since in this case the consumer confidence figures seemed to have been influenced by electoral politics more than the other way around.
Friday’s Polls
Speaking of silver linings: Friday’s poll numbers were a bit more hopeful for Mitt Romney, at least as compared to his extremely poor numbers over the past week or so. A number of swing state polls congregated at a lead in the low- to mid-single digits for Mr. Obama, but we weren’t seeing as many polls with him ahead in mid- to high-single digit leads.
Mr. Obama’s lead also declined by one point in the RAND Corporation’s national tracking poll, and by two points in the Ipsos online tracking poll, and in a poll published by United Press International. It held steady at six points, however, in the Gallup national tracking poll.
Our “now-cast,” which looks at the polls alone and does not account for the economic data, held steady on Friday, suggesting that it would be premature to make too much out of these shifts. A couple more days’ worth of polls like this, however, and Mr. Romney might be able to make the case that Mr. Obama’s momentum has abated some.
The smart money is that Mr. Obama’s standing in the polls will erode some between now and Nov. 6, especially if we continue to see the mediocre economic data that we did on Friday. It is probably too late in the election cycle for the economic data alone to prevent Mr. Obama from being favored in the race, unless the numbers are poor enough to give voters the sense that the economy has reached some new state of acute crisis. But it could soften the ground a bit for Mr. Romney down the stretch run if he can find some other ways, like the debates, to also make gains.