Scott Sumner  

Why don't we see big NGDP crashes any longer?

Obama Pulls a Cheney... Kudos to Brad DeLong...

Before WWII, the government did not measure nominal GDP. But economists Robert Gordon and Nathan Balke put together estimates of nominal GNP, for the period after 1875. (For the US, changes in NGDP and NGNP are highly correlated.)

We used to occasionally see big drops in NGNP; here are some examples:

1893:1 to 1894:2 down 20%
1907:3 to 1908:1 down 14%
1920:2 to 1921:1 down 29%
1929:3 to 1933:1 down 53%
1937:3 to 1938:1 down 14%

During the "Great Recession", NGDP fell by 3% from mid-2008 to mid-2009. Europe had a similar decline, and then almost no decline at all in the "double dip" recession that began in 2011, which was quite serious. Conversely, small declines in NGDP seem to cause more harm than in the old days, perhaps because wages are stickier.

Here I'd like to discuss the puzzling fact that we no longer see big drops in NGDP. One explanation is that Fed policy now prevents this from occurring. And I think that's the right answer. But it's also a bid strange, because modern central banks are so widely seen as being "out of ammo". Admittedly this is less true of the Fed in 2016, as rates have started rising. But in Europe and Japan, rates are lower than at any times during the Great Depression (although T-bill yields did become a tiny bit negative during the late 1930s.)

If you truly believed that central banks were out of ammo, then you'd presumably be puzzled that NGDP did not occasionally plunge by double digits. All it would take is an increase in the real demand for base money, an occurrence that is not at all unlikely when base money is a close substitute with Treasury debt. So why don't we see that anymore?

In my view we don't see deep NGDP declines because investors correctly understand that central banks won't allow them. This creates an expectation that next year's NGDP won't be very different from this year's NGDP. And these expectations help stabilize current AD. In contrast, under the gold standard, the price level was close to a random walk.

Why don't others see things this way? Perhaps because central banks have an asymmetric reaction function. They will work hard to prevent catastrophic drops in NGDP, doing "whatever it takes" but won't push aggressively to achieve more rapid growth, once the economy has moved to a "satisfactory" growth path. So they seem impotent.

I was reminded of this a few days ago, when Mario Draghi caused the euro to soar and European stock prices to plunge, with suggestion that further rate cuts were unlikely. Previously he had a "whatever it takes" can-do attitude, and I think his complacency at the press conference shocked the markets. Draghi will work hard to prevent a eurozone depression, but won't take controversial steps to promote a faster recovery, if the current recovery is seen as adequate.

Comments and Sharing

COMMENTS (8 to date)
Kevin Erdmann writes:

Good points.

But, I have been wondering about this in terms of public vs. private or discretionary vs. non-discretionary.

There is a lot more spending that is either routed through public institutions, or mandated. And, things like health insurance, which don't react cyclically without some difficult dislocations.

Housing is sort of funny here, because it also doesn't react to cyclical shocks very well, but most of it is imputed rent payments, that are a part of NGDP that doesn't even involve a cash transaction, except for the mortgage payment, which is also not something that can adjust.

So, it seems to me that there is a smaller and smaller portion of the economy that can contract when demand shocks hit. This means that it takes a smaller demand shock to create dislocations. If supply-side inflation or spending growth is built into health, or education, or housing, or general mandates and public dispensations, then the cyclically vulnerable part of the economy may even begin to see dislocations before NGDP growth even gets close to zero.

Private investment and PCE on nondurable goods took big dives in the recession.

I think this is something important to consider regarding attempts at Keynesian stabilization. This problem makes the scale of potential mitigating factors higher. It might actually be a problem that public spending, health insurance, etc. are so stable during contractions. Could making them countercyclical actual worsen the cyclical pressures on the cyclically vulnerable parts of the economy?

Scott Sumner writes:

Kevin, Good points, but it's important to distinguish between nominal and real GDP. Even if some sectors are stable in real terms, you can have quite large fluctuations in NGDP if the demand for money rises, and the supply doesn't keep up.

Andrew_FL writes:

Based on many things I've read, I think Balke & Gordon probably overstate Pre-Fed volatility, but I'm unaware of where to find alternate quarterly data. I do know, however, where to find alternative annual data. So to compare apples to apples I first took annual averages of their NGNP data, and then I identified the big drops: 1882-1885, 1892-1894, 1907-1908, and why not let's throw in some that occurred on the Fed's watch, too: 1920-1921, 1929-1933, and 1937-38. These correspond to all your big drops, plus one you missed. There were a couple of single year drops which were not as big, also.

Anyway, I compare the percentage drops in Balke & Gordon NGNP to Johnston & Williamson NGDP for the same periods.

1882-1885 -10.97% vs -5.08%
1892-1894 -11.46% vs -13.53%
1907-1908 -8.82% vs -10.99%
1920-1921 -24.06% vs -16.73%
1929-1933 -45.96% vs -45.32%
1937-1938 -6.43% vs -6.02%

I have to admit, I'm surprised, if anything Balke & Gordon NGNP is less volatile than Johnston & Williamson NGDP. However, 1882-1885 is a striking exception, where J&W estimate only about half as much of a drop in NGDP as B&G do NGNP.

I think it's somewhat misleading to say that the Fed not allowing any actual drops in NGDP to occur is a result of the Fed stabilizing NGDP. To say that it's due to them stabilizing NGDP, rather than simply having volatile but high NGDP growth, you'd need to show they no longer allow large departures from the growth trend. I think you could show that (albeit for a much shorter, more recent period), but saying "NGDP doesn't have big drops anymore" is not enough.

Effem writes:

"Conversely, small declines in NGDP seem to cause more harm than in the old days"

Amazing how little we understand the system sometimes.

My theory: People/businesses solve for a desired level of wealth volatility. If you reduce the volatility of the business cycle (as Fed & fiscal policy have done) then leverage will be added to the system to restore the desired level of wealth volatility. Wealth then becomes more sensitive to any given change in the economy. Large wealth swings are able to overwhelm fed/fiscal policy (e.g., GFC) thus creating business cycle volatility again.

Sure, the Fed could do an even better job reducing business cycle volatility. But we'll just add more leverage making the above true again.

Scott Sumner writes:

Andrew, Good point about the growth trends, but I'd make the same claim if we switched to that criterion. For instance, the growth trend prior to the Great Recession was about 5%, not much higher than under the gold standard.

Jose Romeu Robazzi writes:

Isn't the economy taken as a whole more robust? We have more sectors, more diversification, globalization, etc. Does not that play a role in more NGDP stability? Aren't we giving too much credit to the monetary authority here?

Scott Sumner writes:

Jose, No, those points relate to RGDP, not NGDP.

NGDP is not about "the economy", it's about money.

Jeff writes:

While I don't disagree with your general point, I suspect that the reason is more likely due to a reduction in the structural imperfections that cause a nominal shock to have real effects, such as an increase in labor mobility, decrease in asymmetric information, etc.

Comments for this entry have been closed
Return to top