David R. Henderson  

Great Stagnation or Lousy Data?

Nick Rowe vs. Scott Sumner... Creative Destruction: Sumner E...

Price controls really do matter.

Any claim we make based on aggregate data is only as good as those data. Brink Lindsey has pointed out that, contrary to Tyler Cowen, the so-called Great Stagnation that Tyler writes about is not so much a stagnation as a reversion to a lower long-term growth trend. Byran Caplan posts about it here.

Lindsey shows that the annual growth rate of real per capita GDP fell from 2.5 percent in the period from 1950 to 1973 to a still-healthy 1.9 percent in the period from 1973 to 2007. But something happened between 1971 and 1975 that virtually every economist doing economics then knows about but that few of them and few of the younger economists ever take into account: Nixon's price controls. (The only economist I can think of who did calculations as if price controls had existed is Alan Reynolds.) Taking price controls into account, growth from 1950 to 1973 doesn't look as good as Lindsey's numbers and growth after doesn't look as bad. In other words, the difference becomes smaller.

On August 15, 1971, Nixon froze all prices for 90 days. Then he put the economy through various phases of price controls and, early in 1973, started relaxing price controls. The removal of price controls occurred from early 1973 to sometime in 1975.

So what? Here's what. When prices are repressed, the recorded inflation rate is below the true inflation rate. That means that a measure of real GDP during the period of price control will overstate real GDP growth. When prices are decontrolled, the recorded inflation rate is above the true inflation rate. A measure of real GDP during decontrol will understate real GDP growth.

One way, then, to get a better measure of growth is to skip the period from 1971 to 1975. Measure real GDP per capita from 1950 to 1971 (the price controls were imposed late in the year) and then measure it from 1975 (as I recall, prices, except for oil and gasoline, were pretty much decontrolled by early 1975) to 2007. Doing that, we get the following for growth rates of real GDP per capita:

1950 to 1971: 2.28 percent.

1975 to 2007: 2.09 percent.

Notice how much the difference narrowed from Lindsey's numbers. This cuts the measured difference by a hefty two thirds.

Now, you might say that I'm cheating because I've left out 1973-75 recession. I have 3 responses:
1. Precisely because almost the whole recession occurred during price decontrol, the size of the recession was overstated. Real GDP growth looked worse than it was.
2. When you measure growth over various eras, you usually go peak to peak or trough to trough so you can leave out the short-term effects of recessions. So taking 1950 to 1971 and 1975 to 2007 is, even aside from my price control point, a more legitimate comparison.
3. Touche. So let's keep those years but estimate 1973 real GDP four percentage points lower and then see what we get. Why 4 percentage points lower? On the assumption that the price controls distorted prices by 2 percentage points a year for two years.

If we then do the calculation, we get a narrowing, but it narrows less. We get the following annual growth in real GDP per capita:

1950 to 1973: 2.29 percent.

1973 to 2007: 2.0 percent.

Notice that we cut Lindsey's measure by half.

Comments and Sharing

COMMENTS (12 to date)
volatility bounded writes:

Talk of growth in GDP and talk of household income is highly misleading. The globalists always try to change the subject from typical workers income to GDP, GDP per capital, household income, corporate profits, etc. Hah.

The real debate is over growth in wages of typical workers. The link below has lots of pretty graphs illustrating this point.


Henderson knows this. So does Caplan. They just don't want to talk about. Like politicians, ducking tough questions.

OneEyedMan writes:

I know of one recent attempt to account for price controls:
Oil, Automobiles, and the U.S. Economy: How Much have Things Really Changed?
by Valerie Ramey and Daniel Vine

This paper re-examines whether the impact of oil shocks on the aggregate economy, and on the motor vehicle industry in particular, has changed over time. We find remarkable stability in the response of aggregate real variables to oil shocks once we account for the additional cost of shortages and rationing during the 1970s. To understand why the response of aggregate real variables has not changed, we focus on the motor vehicle industry, because it is considered to be the most important channel through which oil shocks affect the economy. We find that, contrary to common perceptions, the share of motor vehicles in the goods-producing sector of the economy has shown little decline over time. Moreover, within the motor vehicle industry, the recent oil shocks had similar effects on segment shifts and capacity utilization as the shocks during the 1970s.

Tyler Cowen writes:

The book is about the *median* income data, growth there has been much slower...including post- wage and price controls. You are responding to a claim which isn't even in the book.

fundamentalist writes:

Very nice! Thanks!

Keith E writes:

Here is a hypothesis.

One needs to look at population growth rates in addition to per capita GDP growth. Because, the faster the population grows the higher the savings rate needed to secure a given growth rate. As the population growth rate declines, the less savings required.

1950 into the 1970's was a period of falling birth rates for the US. The fall in savings rate lagged, creating a temporary boost to per capita growth rates.

However, government grew during this time and by the 1970s a combination of lower individual savings rates plus government spending (and borrowing) lowered the nation's effective savings and investment rate, bringing the per capita growth rate back down to high population growth rate times.

Bob Murphy writes:

Good post David (though I haven't read Tyler's book, so maybe he is right that this bounces off him harmlessly).

This is also why the growth in official real GDP figures during World War II is bogus. Taking a cue from Bob Higgs' research, I spell out the argument in the middle of this article.

jsalvatier writes:

Normally I think of you as overly ideological and not very interesting, so it means something when I say, this is a really interesting fact. Good on you for bringing it to us.

David R. Henderson writes:

Thanks. I'll check it out.
@Tyler Cowen,
Sounds as if there's not a good substitute for reading the ebook. The links I'll go to, sometimes, to save $4. :-)
@Bob Murphy,
Thanks. I also did the same for those data, drawing on Higgs, for my post-WWII study. See: http://mercatus.org/publication/us-postwar-miracle

David R. Henderson writes:


David R. Henderson writes:

On my response to Tyler. I should have said "lengths," not "links." That was an unintentional pun.

Kevin Dick writes:

I haven't read the book yet either, though it's on my to-do list. However, I thought of a possible explanation to why growth might appear to slow. I'll throw it out there and probably someone will tell me that Tyler addresses it or that it's wrong for some other reason.

I think the 70s mark the point at which significant classes of consumer goods started to see increasing returns to scale / decreasing marginal costs. For example, Tyler's book in 1975 would have had a marginal cost of several dollars (printing, distribution, retailing). Today, it's a few cents. Same with music, movies, video games, business productivity tools, person-to-person communications, etc.

It seems to me that when all your output faces increasing marginal costs, increased output is a good measure of, well, goodness. But as more of it has decreasing marginal cost, not so much.

B.B. writes:

I agree that price controls distorted economic data. But the distortions would have been temporary.

Suppose prices and real GDP were accurately measured in mid-1971, just before price controls. Suppose that after decontrol and the adjustment was complete, prices and real GDP were accurately measured again, say, in 1976.

Then we can and should include the late-1971 to 1975 period in all our calculations, as long as don't make any time period from late-1971 through 1975 as the start or end period of a comparison.

My approach does not omit the years, but just includes them inside the range of calculation.

Comments for this entry have been closed
Return to top