At his blog, Apricitas Economics, Joseph Politano has written two pieces about economic data not adding up. The data in question are on two of the most important economic topics: employment and output.
The Bureau of Labor Statistics uses two different surveys to measure employment. One is the establishment survey, which is answered by businesses, and the other is the household survey, which is answered by workers. We get the total number of jobs from the establishment survey, and we get the unemployment rate from the household survey.
In theory, those surveys are two different ways of getting at the same information. If a survey asks you about your job or asks your employer about your job, both will say that your job exists and it is occupied, which is one of the top things we care about when considering the labor market. There is also information that’s unique to each survey, which is why both are needed, but the overlap is considerable.
They never tell exactly the same story because they use different methodologies, and the scale of data collection means that discrepancies will be inevitable. But Politano writes that right now, the discrepancy is much larger than normal:
Those two reports, however, are in near total disagreement over the state of the US labor market over the last four months. If you believe the household survey, then 168,000 fewer people have a job than in March. If you believe the establishment survey, then businesses added more than 1.6 million jobs since March. It’s the difference between a strong boom and total stagnation.
There are known factors that contribute to the discrepancy, but even when adjusting for them, Politano writes, there’s still a gap of around 1 million jobs:
Once you account for workers taking on second jobs, leaving self-employment, exiting the agricultural sector, and some other conceptual differences between the two surveys, the household survey would estimate only a 722,000 increase in nonfarm payrolls since March—leaving a 958,000 gap still unexplained. In fact, adjusting the household survey to establishment survey concepts estimates a nonfarm payroll level 1.1 million below official establishment survey data.
Politano speculates that increased immigration and the unreliability of other jobs-related data could explain some of the rest of the discrepancy, but it’s still not entirely clear what the problem is or what it means for the economy.
In another post, Politano looks at output data and finds similarly wide gaps between measures that should be similar. He writes, “Official statistics for GDP and GDI have now completely diverged — and the gap between the two stands at nearly 1 trillion dollars.”
Gross domestic product and gross domestic income should be exactly identical. GDP looks at output from the consumption side, and GDI looks at output from the income side, but in economic theory, that’s the same thing. It’s an accounting identity in macroeconomics (Y = I). Just intuitively, it makes some sense: Everything you buy is income to someone else, so whether we add up all the money people spend or we add up all the money people make, we should get the same number.
In the real world, of course, this doesn’t work, and GDP and GDI are never actually identical. But, similar to the employment data, the gap is much larger than normal right now. Politano writes:
If you believe real GDP then the economy has been shrinking for half a year. If you believe real GDI then the economy is still growing (albeit at a slower-than-normal rate). On the flip side, if you believe nominal GDP then total spending has slowed down to a level consistent with ~5% long-run inflation, but if you believe nominal GDI then incomes are growing at a level more consistent with ~8% long-run inflation.
Again, Politano considers a range of possible explanations, all of which seem plausible. He points to underestimating investment and manufacturing output, overestimating corporate profits and aggregate wages, and miscounting trade as possibilities to help explain the discrepancy. But these explanations are still not entirely satisfying, and Politano writes that “it might take years before more comprehensive data and revisions bring GDP and GDI together.”
These gaps aren’t necessarily huge problems. As Politano writes, “Even with the massive GDP-GDI gap, both measures are essentially saying the same thing from a bird’s eye view: nominal growth is extremely high, real growth is relatively low or negative, and inflation is still high.” There’s no evidence that any of this is being tinkered with for political gain by anyone, and each side of the political debate can point to totally legitimate government data to back up its arguments.
It’s always hard to answer the question, “How’s the economy?” because there’s just so much to consider, and there will always be some people who are doing well and others who are doing less well at any given time. But now, government statistics that measure the same economic concepts are providing divergent answers to a much greater degree than in the past. We’re truly in uncharted waters.