RBP
07-27-2012, 09:39 PM
Let me begin with an important reminder. Unless you have a particular interest in economic or political forecasting, the major upshot of Friday’s gross domestic product report, which estimated that the economy grew at an annualized rate of 1.5 percent in the second quarter, is simply that the American economy is still not doing very well.
What economists call “trend” or average growth is supposed to be closer to 3 percent. Moreover, after many past recessions, we’ve had above-trend growth in the year or so after the recession to help make up for the lost productivity in the economy.
We didn’t have that this time around. After the latest revisions to the data, only two quarters since the recession officially ended came in above that 3 percent threshold; all the others were below it.
So, there remains an output gap of about $800 billion dollars. That’s the difference between what economists think annual G.D.P. would be if the economy were at its full productive capacity, and what it actually is.
But we concern ourselves with the more narrow question of election forecasting at this blog. On that front, I agree with other analysts who suggest that the G.D.P. numbers don’t contain all that much new information.
In the case of how our forecasting model is designed, this is true — it does not use the backward-looking G.D.P. numbers.
The reason for this is simple: G.D.P. is reported on a quarterly basis. That’s a bit slower than we’d like, when there are other good data series that have a shorter lead time.
Moreover, the initial estimates of G.D.P. are often very noisy. It’s possible that Friday’s numbers will eventually be revised to show that growth was actually negative, meaning that the economy was potentially in recession. It’s also possible that they’ll eventually be revised upward to show 3 percent growth or better. It’s very common to see quarterly G.D.P. numbers revised by a full percentage point in one direction or the other, and reasonably common for the adjustments to be larger than that.
Instead, the bulk of our economic index (60 percent of the weight) is assigned to a series of four monthly indicators that correlate reasonably well with G.D.P. over the long run, and that economists use to help date recessions and recoveries.
These numbers are noisy too, of course — some more so than others. (The series called personal income, for instance, tends to get especially substantial revisions, while the jobs numbers are a bit more stable.)
But if you look at what they say in combination with one another, you can usually get a reasonably good estimate of G.D.P. a couple of months before the figure is released.
That was the case here. The reported figures of 1.5 percent growth in the second quarter, and 2 percent growth in the first quarter, are consistent with what the economic index had already “priced in” to its expectations.
The chart below shows the normalized figures for the four G.D.P. proxies: payroll jobs, industrial production, personal consumption and personal income. The normalization process translates these numbers so that they are on the same scale as one another, and the same scale as quarterly G.D.P. reports.
http://i.imgur.com/aapDa.jpg
The four indicators tell different stories about the economy. Income growth has been very poor, for example, while industrial production has been pretty encouraging.
For better or for worse, this is a pretty normal state of affairs. Economic data is really noisy, and different series often diverge quite a bit, especially before revisions.
Still, if you average the four numbers together, you get an index of 1.7 percent growth. That almost exactly matches the G.D.P. growth rate during the previous six months, which is the period that our model is mostly concerned about.
Growth at this rate would ordinarily make a president’s re-election prospects extremely tenuous: probably about 50-50, according to our model and others.
The reason our economic index sees Mr. Obama as a very modest favorite for re-election is because it also considers inflation, which is assigned 15 percent of the weight. And inflation has been very low.
Imagine the counterfactual: that on top of all the other problems in the economy, we also had, say 6 percent inflation. Then gas might cost $4.75 a gallon, and you’d notice the change in prices pretty frequently when you went to the grocery store. Under those circumstances, our model would have Mr. Obama as an underdog.
Of course, the G.D.P. numbers and other economic statistics are pretty tenuous on their own. If things get much worse than this in the next three-and-a-half months before the election, our economic index will start to see Mr. Obama as an underdog even with the low inflation rate.
What economists call “trend” or average growth is supposed to be closer to 3 percent. Moreover, after many past recessions, we’ve had above-trend growth in the year or so after the recession to help make up for the lost productivity in the economy.
We didn’t have that this time around. After the latest revisions to the data, only two quarters since the recession officially ended came in above that 3 percent threshold; all the others were below it.
So, there remains an output gap of about $800 billion dollars. That’s the difference between what economists think annual G.D.P. would be if the economy were at its full productive capacity, and what it actually is.
But we concern ourselves with the more narrow question of election forecasting at this blog. On that front, I agree with other analysts who suggest that the G.D.P. numbers don’t contain all that much new information.
In the case of how our forecasting model is designed, this is true — it does not use the backward-looking G.D.P. numbers.
The reason for this is simple: G.D.P. is reported on a quarterly basis. That’s a bit slower than we’d like, when there are other good data series that have a shorter lead time.
Moreover, the initial estimates of G.D.P. are often very noisy. It’s possible that Friday’s numbers will eventually be revised to show that growth was actually negative, meaning that the economy was potentially in recession. It’s also possible that they’ll eventually be revised upward to show 3 percent growth or better. It’s very common to see quarterly G.D.P. numbers revised by a full percentage point in one direction or the other, and reasonably common for the adjustments to be larger than that.
Instead, the bulk of our economic index (60 percent of the weight) is assigned to a series of four monthly indicators that correlate reasonably well with G.D.P. over the long run, and that economists use to help date recessions and recoveries.
These numbers are noisy too, of course — some more so than others. (The series called personal income, for instance, tends to get especially substantial revisions, while the jobs numbers are a bit more stable.)
But if you look at what they say in combination with one another, you can usually get a reasonably good estimate of G.D.P. a couple of months before the figure is released.
That was the case here. The reported figures of 1.5 percent growth in the second quarter, and 2 percent growth in the first quarter, are consistent with what the economic index had already “priced in” to its expectations.
The chart below shows the normalized figures for the four G.D.P. proxies: payroll jobs, industrial production, personal consumption and personal income. The normalization process translates these numbers so that they are on the same scale as one another, and the same scale as quarterly G.D.P. reports.
http://i.imgur.com/aapDa.jpg
The four indicators tell different stories about the economy. Income growth has been very poor, for example, while industrial production has been pretty encouraging.
For better or for worse, this is a pretty normal state of affairs. Economic data is really noisy, and different series often diverge quite a bit, especially before revisions.
Still, if you average the four numbers together, you get an index of 1.7 percent growth. That almost exactly matches the G.D.P. growth rate during the previous six months, which is the period that our model is mostly concerned about.
Growth at this rate would ordinarily make a president’s re-election prospects extremely tenuous: probably about 50-50, according to our model and others.
The reason our economic index sees Mr. Obama as a very modest favorite for re-election is because it also considers inflation, which is assigned 15 percent of the weight. And inflation has been very low.
Imagine the counterfactual: that on top of all the other problems in the economy, we also had, say 6 percent inflation. Then gas might cost $4.75 a gallon, and you’d notice the change in prices pretty frequently when you went to the grocery store. Under those circumstances, our model would have Mr. Obama as an underdog.
Of course, the G.D.P. numbers and other economic statistics are pretty tenuous on their own. If things get much worse than this in the next three-and-a-half months before the election, our economic index will start to see Mr. Obama as an underdog even with the low inflation rate.