We all know that we are going through one of the worst economic downturns in US history. But how bad exactly is the downturn? The recently released Gross Domestic Product data for the second quarter of 2020 paint a very grim picture, with the headline number suggesting that the economy contracted by -32.9%.
GDP is a measure of all economic activity that takes place in a quarter or a year. Was there really one-third less activity in the second quarter compared with the first quarter? No there was not. The actual number is about a 7% decline. I’ll explain more how I came up with that number, but let me stress this is still a very bad number. It’s the worst we have on record, possibly the worst in US history, probably even worse than any one quarter of the Great Depression (if we had directly comparable data). Still, a number like -32.9% is not a very helpful number in the current context.
Interpreting economic data is challenging during the current economic crisis. My intent is not to downplay the harm, but to give it proper context. For example, I have previously written that the unemployment rate understates how much pain there is in the labor market right now. In contrast, the recently released GDP data overstate the economic pain.
To understand why the number overstates the economic pain, we need to understand two adjustments that the Bureau of Economic Analysis makes to the GDP data. These are changes they have always made, not changes specific to the current crisis.
First, the data is “annualized.” Essentially an annualized number tells us that if the current growth rate continued for a full year, that’s how much the economy would contract for the full year. In normal times it’s a useful number because we normally think of GDP in annual terms. But at this point in the current crisis it is probably more useful to think in terms of one quarter at a time. To convert the -32.9% figure to a quarterly figure, we don’t simply divide by 4, as might seem logical, because there is a cumulative effect. Once we do the proper conversion, the quarterly GDP decline in the second quarter was about -9.5%, rather than -32.9% (here’s a more detailed explanation of the calculation for a 12-month annualized number).
Second, the data is “seasonally adjusted.” Seasonal adjustments are useful during normal times, essentially to tell us whether the current quarter GDP data is better or worse than a “normal” quarter. For example, we know that the first quarter data is usually much lower than the prior fourth quarter. The fourth quarter is the holiday shopping season, so there is much more economic activity. It wouldn’t be very helpful every year for the BEA to tell the country that GDP declined from the fourth to the first quarter: everyone expects that. The seasonal adjustment tells us whether the change was better or worse than a normal quarter-to-quarter change. Similarly, there is usually more activity in the second than the first quarter, so even if economic activity stayed the same from the first to the second quarter this year, BEA would call that a decline. That makes sense in normal times, but I don’t think it does right now.
Once we make both the changes, putting the number on a quarterly basis and removing the seasonal adjustment, we arrive at the -7.0% figure I mentioned earlier. This figure gives us a much better picture of how much the economy contracted from the first to the second quarter. If we perform the same calculation and start in the fourth quarter of 2019, we see that GDP has shrunk by about 12.3%. That’s still an immense amount of economic pain in two quarters, but it’s much smaller than if you simply added together the “headline” first and second quarter numbers, which makes it appear the economy has shrunk by almost 40%.
All the data I am using can be found in the official BEA release of the data, but you need to scroll all the way down to the last page and look at Appendix B, and do a simple calculation yourself.
One final note: these numbers will be revised. BEA considers this release to be an “advanced” number, and it will be revised in August and September, though the revisions are usually not large (but who knows in these uncertain times). In late October, just days before the Presidential election, we will see the first BEA estimates of GDP for the third quarter of 2020.
Since we are only one month into the quarter now, no one can say what the numbers will look like for the third quarter of 2020, but there is a good chance they will be record-breaking on the upside this time. Once again, it will be important to contextualize those numbers rather than just looking at the headline number.