Tuesday, August 4, 2020

How To Measure Quarterly Changes In GDP Can Make A Big Difference

We have had dramatic headlines and commentary in recent days since the BEA issued its initial estimate of quarterly changes in GDP, which they do not officially measure on an shorter time period. This is a measure of the average GDP in one quarter compared to the average GDP in the next quarter.  Looking at Q1 of this year and Q2 of this year, they reported the largest quarterly decline ever recorded, -32.9% on an annualized rate, about -9.5% on a quarterly rate. This is a sharper decline than seen either for any pair of quarters in the Great Depression or the immediate post-WW II demobilization, much less such later events as the Great Recession of 2007-09.  Of course this generated big headlines and much breathless commentary, including quite a few commentators who did not get it that the -32.9% number was the annualized rate rather than the quarterly rate, so we had quite a few of them foolishly talking about how the economy had "fallen by a third."  Yikes.

As I have noted in previous posts here, the economy has been doing a lot better than a lot of reporting and forecasts have let on, even as it is certainly slowing down now.  But if rather than looking at quarterly averages, which are heavily influenced by the fact that the economy did not "fall off a cliff" until mid-March, about 5/6 of the way through Q1 so that most of Q1 was at the high pre-fall level, one looks at where the economy was at the end of Q1 and compare it to where it was at the end of Q2, one gets a dramatically different story.  Over on Econbrowser Menzie Chinn has produced a figure from IHS Markit that estimates these shorter period changes. According to them the US GDP at the end of March (end of Q1) was at about $17.6 trillion annual amount while at the end of June it was at about $18.0 trillion, about 2.2% higher on a quarterly rate, which is about 9% higher on an annualized rate.  Needless to say, there is a dramatic contrast between -32.9% and +9.0%, but I have seen zero media commentary on this.

Indeed, growth has continued through July, although probably at a slower rate than in June, with a slower growth of retail sales of only 3.2% suggesting it has indeed slowed down quite a bit.  But in fact the economy has been growing for probably more than three months, something although "worst ever" stories seem to have ignored, and indeed the figure Menzie showed that went through May and June but not July sure looked a lot like a V, if just a bit flatter going up than down.

For anybody for whom this is just unacceptable to think about and you must think all is bad, well, unemployment remains high and indeed apparently the unemployed numbers have started going up again.  The economy may be ahead of where it was at the end of March, but it is still pretty far from where it was in February, and it is definitely slowing down, with its fate clearly closely tied to what happens with the coronavirus pandemic, which is not easily forecast..

Barkley Rosser


Not Trampis said...

Barkley the 9.5% drop is the annual rate or what Menzie says is the Yr/Yr change.

IMHO this is the figure we should be looking at not the annualised one.

rosserjb@jmu.edu said...


The -9.5% is the quarterly rate, not the "annual" nor the annualized rate. The latter is -32.9%. The dramatic looking contrast is between that -32.9% and the annualized of what happened between end of Q1 and end of Q2, which is +9.0% about, which is pretty sharp. And it is easy to say that both of these are the "annualized GDP growth rate between the first and second quarters." No wonder thare has been so much confusion and running around in circles about these numbers and characterizations.

rosserjb@jmu.edu said...

Mate, :-) (to Not Trampis)

2slugbaits said...

In this case the -9.5% number is unhelpfully ubiquitous and leading to some confusion. The annualized GDP level fell by 9.5% between 2020Q1 and 2020Q2. That 9.5% drop works out to a 32.9% drop if the economy continued to fall at the same rate for the next three quarters. But by sheer coincidence it's also the case that the GDP level in 2020Q2 was 9.5% lower than it was a year ago in 2019Q2. I think the best way to think of the 32.9% value is as a kind of instantaneous rate of change. The problem is that the BEA is attempting to measure an instantaneous rate of change by taking an average of activity across 3 months, and 3 months is a long way from instantaneous. I do think that whole discussion got a lot more attention than it deserved, and my additional comment here only adds to that undeserved attention. There is no single best way to express changes in GDP. It really depends on the question you're trying to answer. Sometimes you're interested in the instantaneous rate of change expressed as an annualized rate. Sometimes you're more interested in filtering out the quarterly noise by using 4 quarter averages. Sometimes you're interested in comparing where you are now relative to where you were one, two, three or more quarters ago...a kind of Reaganesque question asking if you're better off today than you were four quarters ago.

rosserjb@jmu.edu said...

What really sticks out is that most of the time these distinctions between one measure and another do not matter. When things are going along more or less "normally" these competing measures do not differ all that much (aside from an annualized number being different from one measured for a shorter period). It is a sign of how extreme and uneven and unforecastable the current situation has been that suddenly makes it that using one measure suddenly produces something highly different from another measure. And this is exacerbated by the extreme differences in sectoral patterns, with some sectors booming while others are totally collapsing.

2slugbaits said...

The "S" part of the SAAR calculation (i.e., the seasonality adjustment) is probably compromised as well. This might be a time when we want to consider non-seasonally adjusted GDP levels as well as seasonally adjusted.

Not Trampis said...


The ANNUAL rate qtr on qtr this year to last year fell 9.5%.

SAAR is every other country is seasonally adjusted ANNUAL rate.

Me thibks we have a difficulty in language.

everybody other's SAAR is more meaningful than the USA's which I might say your article is helpful in understanding as usual. I should never forget my MATE Sluggsy either.

your one too Barkes

rosserjb@jmu.edu said...

NT, mate,

Well, I think seasonal adjustment, which is a mysterious ad hoc procedure, although there well-known methods for doing it, if not universally agreed upon, is really only useful for looking at the annualized version of a sub-annual rate. There is no need to seasonally adjust the change fro Q2 2019 to Q2 2020. It is only important if one is looking at from something like what has been at issue here, Q1 2020 to Q2 2020, when indeed one might want to correct for seasonal peculiarities. But, as has been noted elsewhere, how the US BEA does it when making their quarterly estimates of GDP changes is not publicly known, although I think they describe roughly how they do it in some materials.

But I have sympathty 2slugs's main point. Seasonal adjustment can cover up useful information. There is a case for not doing it, at least some of the time, and in these wild times where the data noise is so loud as to be a deafening thunder, simply looking at the unadjusted numbers might really be more useful.

Kara Whitaker said...

Please share as you read okay you may save a soul today, natural herbs
are really great, I was cured from HSV by a doctor called DR Okiti from West Africa, now I believe that natural herbs and roots has their own way of working, based on medical report I was told that HSV has no cure even when I saw a post about Dr Okiti how he cured many people with his herbal medicine I doubted it at first but just decided to give it a try because I was so desperately trying to get rid of my problems not knowing it was going to be the end of this deadly virus in my body, please if you have anyone with this same virus or diseases like this contact Dr Okiti via email drokitiherbalhome100gmail.com or whatsapp +234 705 067 0365. I'm a living testimony of his herbal medicine and don't forget to share as you read okay, please save a soul today.