The US Bureau of Commerce released its GDP figures yesterday. The economy contracted by 4.8% seasonally adjusted, annualized rate. This does not mean that the economy contracted by 4.8%. It means that if the growth of the first quarter were projected over an entire year, then the economy would decrease by 4.8%. Instead, annual growth rate means the economy shrunk by 1.1% in the last quarter, adjusted for seasonal variation.
I have written before about how seasonal adjustments are probably inappropriate in our current economic environment. We know that the seasonal correlations are probably distorted by the current economic shifts, and the large magnitudes of shifts will amplify any inaccuracies. In this case, there is a difference, but it’s not huge: if you look at last table on the BEA report, it gives quarterly GDP without seasonal adjustment. In order to account for the fact that there probably is a significant seasonal component, we compare Q1 2020 with Q1 2019. The economy grew on average 2.3% in 2019, so we would expect that the Q1 2020 GDP level to be $4.72 trillion. Instead, it is $4.63 trillion, a decrease of 1.8% quarterly or 6.9% annualized. This isn’t a huge difference between a 4.8% decrease, but it’s not peanuts, either.
The first lesson from this is that taking any short trend and annualizing it can give some pretty big numbers. Neil Irwin has a good piece about this in the New York Times. The second lesson is that production is a relatively stable variable. -6.9% is historically a very sharp contraction, and it would be a real cause for alarm if this negative growth were to be sustained. However, even in that context, the actual quarterly decline is 1.8%. Even in such a sharp contraction, the overall magnitudes of change are not that large.
The economy didn’t really contract over the entire quarter… really, only March saw a heavy economic contraction as the news relating to Covid-19 led to shutting down large portions of the economy by the end of the month. Assuming GDP in January and February grew at the 2.1% percent that it had grown for the previous 2 quarters, we can estimate that the GDP was at a monthly rate of $1.578 trillion at the end of February. Subtracting the estimated January and February figures from the reported GDP for the quarter gives $1.480 trillion for March, or a $98 bn dropoff, or a -6.2% month to month growth rate. If this trend were to continue, it would imply a -53% annual growth rate!  This is an insanely large figure, and should serve to highlight the dangers of extrapolating a short trend over a long period of time.
Projecting forward, we can see that the confinement has lasted throughout April, and is likely to remain in place to some extent through May, as well. If the 6.2% decrease in March lasts over three months, this would be an overall contraction of 17%, an extreme recession by historical standards. But that is an assumption in itself… the restaurants and bars are already closed, so the impact of that shock has already been felt. If anything, people are starting to adapt to working from home, and a new normal is starting to emerge. People will start finding innovations and efficiencies in this new way of life, and there could even be some potential growth in select areas over the next two months. In that case, if we confine the overall contraction to the current $98 bn decrease in March, it would amount to a 0.5% blip on the overall annual GDP. Moreover, if none of our negative propagation mechanisms kick in, we could see a relatively quick recovery once the epidemic is firmly under control, and we could record positive GDP growth for the year.
How realistic is this scenario? It rests on the assumption that the worst of the shock is over. Let’s test that assumption with the data: we know that certain industries are particularly badly hit: non-essential retail, for example. If the overall contraction in May incorporated that completely, then we could reasonably say the worst is over, but if the retail sales only contracted by single digits, we might assume that the shock hasn’t been fully incorporated into this quarter’s data.
Unfortunately, the specific sector breakdowns are only available in the report with seasonally adjusted data. We can still make approximations, though. “Food services and accommodations” decreased from $866 bn to $793 bn. If we assume that entire decrease was in March, as we did above, this would imply an 26% contraction in the month. If we do similar calculations for other impacted industries, we get seasonally adjusted growth rates of -28% for “recreation services,” -33% for “clothing and footwear,” and -14% for “durable goods.” Keep in mind that the seasonality of these industries is likely larger than overall GDP, so the distortions from calculating seasonally adjusted data might be even larger here. In which case, we might be understating the decline by a large amount. Whether you think these figures capture the entirety of the decline in these industries or whether you think they still have room to fall is a judgement call. I will say that a single month decline in production of any industry of these magnitudes is probably unprecedented, and one shouldn’t discount the resilience of some particular companies in these industries (by moving to online retailers for clothing, for example).
Thus, the figures released yesterday are not what they seem at first glance. The real GDP growth rates are both better and worse than they may originally seem from the report: better because the report magnifies the monthly decline by presenting the growth rate as an annualized figure and worse because they understate the actual contraction in production by using seasonally adjusted figures. The big question is whether this sharp contraction will continue or whether it was a single shock allowing us to start adjusting to our new normal. For this, some potentially scary numbers should be comforting: we see huge declines in particular industries, and if the current headline number incorporates those industry declines, then we might be able to hope that the worst has passed. We would expect the decline to be felt in specific industries, and we do potentially see eye-watering declines in those industries. Whether that means the worst is over economically is up to you.
 Between Q4 2019 and Q1 2020, the economy contracted 5.3% for the quarter, or 20% annualized. However, there was a 3.7% drop in Q1 last year and a 4.0 drop in the previous year, so a significant seasonal drop in Q1 is to be expected. Comparing Q1 2019 to Q1 2020 allows for a more direct comparison, and we incorporate a 2.3% growth rate from the 2019 figure to give an expected GDP figure for Q1 2020.  With growth figures of this magnitude, it is important to use the proper exponential formulas rather than the shortcuts we use with figures closer to zero.  It also shows how quickly a quotient can increase if we decrease the denominator.