There are two (traditional) ways to measure the economy. The first you hear about all the time. It's gross domestic product, or the sum of everything the economy produces in a given year. The second you don't hear about too often. It's gross domestic income, or the sum of what everyone earns in a year.
In theory, the two measure the same thing. In reality, they don't, as they're based on different samples of data.
And a funny thing is happening lately. Measured by GDP, the economy is growing, but just slowly. Yet measured by GDI, economic growth looks far stronger.
The Commerce Department announced yesterday that GDP grew 3% last quarter, and 1.7% for all of 2011. That's weak, especially coming off a deep recession, when you'd expect to see a strong rebound.
But GDI shows the economy grew 4.4% last quarter and 2.1% last year. The latter is still low, but the big uptick in last quarter's growth is impressive. It gives weight to the argument that the economy is really, truly getting better. It also supports the theory that the recent spurt in jobs growth was real and not a statistical blip -- something relatively meager GDP growth cannot do.
Lest you think GDI routinely overstates growth, it doesn't. It actually shows the recession in 2008 and 2009 was deeper and more severe than GDP made it out to be:
Source: Bureau of Economic Analysis.
Why is this important? I asked Justin Wolfers, a Wharton economics professor who's been vocal about the disparity between GDP and GDI, whether GDI should be favored over GDP when gauging the economy's growth. He responded:
Yes. The evidence compiled by [Fed economist] Jeremy Nalewaik is that in real time, GDI is more correlated with other indicators of the business cycle. Moreover, the early GDI [reading] predicts the direction and size of future GDP revisions. Thus, I prefer GDI to GDP on purely statistical grounds -- it provides a sharper picture both of where we are, and where we are going.
So then, ignore GDP? Not necessarily. Here's Wolfers again:
There are three hypotheses you could consider: 1. Put all your weight on GDP; 2. Put 50/50 weight on GDI and GDP; 3. Put all your weight on GDI. The data are clear that #1 is a bad idea. Just how much weight to put on GDI is an open question. 50/50 isn't a bad idea (this is what Nalewaik modestly recommends), and there's a chance that you might do even better still by putting a bit more weight on GDI.
Splitting GDP with GDI 50/50 would show the economy grew by 3.7% last quarter -- fairly strong, all things considered.
There's another part of the economy where official statistics paint divergent views: the job market.
The Bureau of Labor Statistics calculates employment two ways. First is the payroll survey, which collects data from a sample of businesses. When you see headlines like, "The economy created X jobs last month," that's from the payroll survey. The second is the household survey, which asks actual households about their employment status. It's where the unemployment rate comes from.
The payroll survey shows the economy added 1.2 million jobs over the past six months. That's pretty good! But the household survey shows it was even better, adding 2.1 million jobs. That's great!
But which should you pay attention to?
The household survey is far smaller than the payroll survey, making it more susceptible to noisy swings. Yet like GDI, its bullishness might be telling us something. Wolfers explained:
Over longer time periods, the statistical noise in the household survey becomes less important, relative to the underlying signal. This is important, because there are real concerns about the payrolls survey, which relies a lot on knowing the universe of firms out there, while the household survey requires knowing a lot about the universe of households. Truth is, we know a lot about the universe of households, and far less about the universe of firms. Thus the payrolls survey is supplemented by a statistical "birth-death" model which can go wrong -- perhaps especially at turning points. ...
Also, note that over recent months the early payrolls numbers have been revised upward by about 40,000 per month. This is another reason to believe that we may actually be understating current employment growth -- the latest good numbers may turn out upon revision to really in fact be very good, and perhaps even terrific.
I think the takeaway here is not that the economy is certainly stronger than it appears, though that may be true.
Rather, the disparity in these statistics highlights their imperfection. Maybe GDP is right -- or maybe GDI is. It's hard to tell. It's the same with various employment figures. None offers precision, and that has to be kept in mind before drawing precise conclusions, regardless of whether you're bullish or bearish.
"It's important to be aware of that imperfection," Wolfers admitted. "And it makes sense to use the least imperfect statistics. But the truth is, no single statistic perfectly captures the state of the business cycle, which is why we tend to look at so many."
"The big takeaway," he finished, "we should make policy with a full understanding of the enormous uncertainties we face."
Grain, meet salt, in other words.
Check back every Tuesday and Friday for Morgan Housel's columns on finance and economics.