David R. Henderson  

GDP: A Bad Measure of Well-Being

The art of silly norms... Hoxby vs. Dale-Krueger on the ...
Picture this: The U.S. government finally sells the Postal Service. As with other functions moved from the government to the private sector, the privatized post office does what the government did for about half the cost. So, with prices correspondingly lower, people spend roughly half as much as before on mail--which frees them to spend the difference on other desirable things. Because the Postal Service costs over $40 billion a year, the saving is $20 billion. By any reasonable measure, the average person in the U.S. is better off. In fact, the per capita increase in well-being is approximately $20 billion divided by 260 million citizens, or about $80 apiece.

But how does this change show up in gross domestic product? It doesn't. The government's contribution to GDP is measured not by how much value it creates but by how much it costs. So the $40 billion spent by the Postal Service counted as a $40 billion contribution to GDP. Cutting that in half through privatization may shift $20 billion from public to private hands but still adds up--under the conventions of national income accounting--to the same $40 billion. So the net effect on GDP of a $20 billion increase in economic well-being is precisely $0.00.

Obviously this is absurd. Indeed, "gross domestic product is not a good measure of a nation's overall well-being" is what we economics professors tell our students every time we teach a macroeconomics course. (Macroeconomics is the study of employment, inflation, GDP, and economic growth.) But too many macroeconomists promptly forget that basic fact and judge an economy's performance almost solely by the growth of its GDP.

This is from my article, "The Case for Small Government," Fortune, June 26, 1995. I wrote it as a critical response to an earlier Fortune article by Paul Krugman.

Another excerpt:

Krugman asserted that any politician who claims he can raise the economy's growth rate "by as much as three-tenths of a percentage point is naive--or worse." Maybe, but that doesn't mean a politician can't add three-tenths of a percentage point to the growth rate of economic well-being.

Then I went through and showed how it could be done.

Why post this now? Because the point about GDP not being a good measure of economic well-being becomes even stronger when so many valuable services are given away on the Internet. That's the point of an excellent article in the New Yorker, "Gross Domestic Freebie," by James Surowiecki. Excellent, that is, until the last paragraph.

One great excerpt from Surowiecki:

New technologies have always driven out old ones, but it used to be that they would enter the market economy, and thus boost G.D.P.--as when the internal-combustion engine replaced the horse. Digitization is distinctive because much of the value it creates for consumers never becomes part of the economy that G.D.P. measures. That makes the gap between what's actually happening in the economy and what the statistics are measuring wider than ever before.

I also covered some of the same ground in "GDP Fetishism."

HT to Jeff Hummel.

Comments and Sharing

CATEGORIES: Macroeconomics

COMMENTS (13 to date)
Sven writes:

I don't aqree with the postal service example:
If a private company can save half of the costs, it means that they are employing less resources for it. This frees up these resources for other stuff, which will raise GDP.

David R. Henderson writes:

Your second and third sentences are totally correct, up to, but not including, the last 4 words. That was the whole point of the excerpt of my article.

MingoV writes:

Why would anyone consider GDP to be a measure of well-being? GDP only estimates production and services. As noted above, GDP greatly overstates government "production" and services. Because our local, state, and federal governments have expanded more than any other economic sector, GDP has little utility.

Mark V Anderson writes:

I agree that GDP doesn't do a very good job measuring well being, for many reasons. My favorite discrepancies are the non-counting of leisure time and DIY projects. But I think it is used because of a lack of alternatives. Is there a better way to measure well-being?

notsneaky writes:

GDP is not a measure of well-being. It's a measure of (market) production. Of course the abundance of products probably does have something to do with well being, but that's one of them philosophical questions (Zen Buddhism and all that hippy stuff). If there is some goofy econ professors out there who insist on a one-to-one correspondence between the two then shame on them. But that's a straw man.

This is a bit like saying that temperature is not a measure of well being. Well, sure. But it's not suppose to be. It's suppose to just measure production/temperature. What you (or any one else) makes of it is a different matter. My well being is not determined by temperature, but if the temperature is -15 F or 110 F I'm a lot worse of in my well being than if it is 69 F. I think ALL the economists - Austrians, Keynesians, Classicals and whoever else - were able to agree with that. Quit making things harder than they really are.

Andrew_FL writes:

I do like that example by Paul Samuelson in the GDP Fetishism article.

This is a better, more practical example, though.

@notsneaky-Actually, it's boilerplate classic Keynesianism that you'd learn in any introductory Macro course at most major universities, that basically the point of Keynesian policy recommendations is to use government spending and taxing to "correct" fluctuations in "aggregate demand." Aggregate demand is defined as the sum of all expenditures in the economy: its formula is actually identical to, and equivalent to, that for GDP.

Now, certainly at least some Keynesians are aware that GDP is not a measure of well being per se-Samuelson is a good example. But that doesn't mean that Keynesianism is not preoccupied with it to a peculiar degree anyway. Krugman's suggestions on many occasions (fake alien invasion, natural disasters) for how to produce economic well being seem to indicate that if he is aware of this fact, he doesn't think it matters that much and that raising GDP is always good and lowering it always bad, no matter how you do it.

ThomasH writes:

But there is the Rodny Dangerfield question: compared to what? And for what purpose?

Does the fact that GDP underestimates the value of government services because they are not usually priced at the profict maximizing level affect the optimimal size of QE2?

Sam Grove writes:

GDP measures prices, not value.

Bill Woolsey writes:

Think of the measurement of government as an effort to measure its opportunity cost. When real GDP rises, the total amount of private goods and services that could have been produced has risen.

When there is a shift between private goods and government goods, there is no change in GDP. The impact of such shifts on welfare is not captured.

The Keynesian multiplier is a variant of the velocity of money. If newly-created money is distributed by tax cut or transfer payment, it is pretty much the classical quantity theory thought experiment. If it is used to fund added government spending, there is going to be multiplier effect if velocity is greater than one, but the composition of demand and output will shift. The proportion of output made up of government goods and services rises.

In a liquidity trap, most realistic at the zero nominal bound, such that the bonds at least temporarily have a zero yield, it is like new money was created. The multiplier effect is like velocity (related to the transactions demand for money rather than the "speculative" demand for money which would be closer to using money as a store of wealth.) If government spending is funded, spending on private goods will likely rise too.

Now, increased spending on output can only expand real output much when there are surpluses of resources. And that is when prices are above market clearing levels. It is an alternative to a lower level of prices and wages.

If the Keynesian government spending multiplier is 1, and prices and/or wages cannot decrease at all, then private output is left the same and government output rises. If the government goods are worth nothing, this is pointless.

If this were really the situation, then a better way to solve the problem of involuntary unemployment would be work sharing. Share the pain of unemployment? Or, in your "micro" approach, share the joy of added leisure.

I don't think this is a very realistic scenario. A change in the quantity of money using open market operations should be able to fix the problem. Taxes and government spending (and budget deficits and debt) should focus on the sticky issues of welfare rather than total spending on output. It is best handed against a background of a stable growth path of nominal GDP.

Is spending on output "too low" resulting in too little production? GDP gives (imperfect) information. Are real standards of living rising over time. GDP gives (imperfect) information about that.

It is helpful to keep these two things separate.

Price indices similarly have two different roles.

notsneaky writes:


What GDP is and what it measures is independent of Keynsianism, Monetarism, New Classicism etc. Just like how we define temperature is independent of what someone thinks about global warming.

Sure Keynesians want to smooth out the fluctuations in GDP and that may or may not be desirable or feasible. But that in itself is not an argument against GDP as a measure of production (and at the end of the day, economists will always want to measure production)

Unless one is into coming up with convoluted examples, GDP (per capita) going up is generally a good thing. People might want to be careful with pushing this "GDP is not well being" argument. There's nothing anti-statist about it. In fact most of the time you see it being used on the left along the lines of "Cuba might have lower GDP than US but GDP is not a measure of well being anyway, and Cuba has universal literacy so socialism wins!"

As to Surowiecki's point about digitization. Ok. But 1) I wanna see an estimate of how significant (in terms of % of GDP) this is. Which it's not. And 2) the same logic as with the privatization of the post office applies. If digitization drives a traditional publisher out of business, the resources it used, the capital, the labor, etc are still there and you have to account for the fact that they'll be used in alternative production.

But sure there's something to it. It's just like the worn example of a guy mowing his own lawn vs. a guy hiring someone to mow the lawn. But lawn mowing isn't a large chunk of the overall economy and the rises and falls in GDP have very little to do with changes in how lawns are mowed.

Brian writes:

"So, with prices correspondingly lower, people spend roughly half as much as before on mail--which frees them to spend the difference on other desirable things. Because the Postal Service costs over $40 billion a year, the saving is $20 billion. By any reasonable measure, the average person in the U.S. is better off...But how does this change show up in gross domestic product? It doesn't."


I think you are overlooking a crucial point, which is that the economy is not immediately better off by spending half as much on mail service. It really depends on how the freed-up money is used. If the resources are used for productive purposes (i.e. invested), then the economy will grow faster and produce a higher GDP in the future. If the freed-up resources are used for consumption only, then nothing is gained and one might as well spend it on an inefficient postal service.

The same point can be made about the digitalization of the economy. Although these things are often offered for free and therefore make no direct contribution to GDP, their ultimate, indirect effect is to make people more productive, thereby increasing GDP growth. Everything useful eventually makes an impact in the price and value of goods.

So while it's true that GDP is not a direct measure of well-being, it strongly correlates with it. Indeed, other carefully constructed measures of well-being really do no better at predicting reports of subjective well-being than per capita GDP. So GDP is not a bad measure at all.

Brian writes:

Just to supplement my preceding point, here is a blog post on the subject. The take-home message in the work reported in the blog post is

"To summarize our argument, for an economic indicator never intended to assess national well-being, the GDP is surprisingly successful in predicting a population’s subjective well-being. At the same time, the theoretical claim of the social indicators movement about the multi-dimensionality of human concerns is a valid criticism, and conceptually it should be possible to come up with performance measures that embrace this multi-dimensionality better than an economic performance measure alone."

The bottom line is that it's just plain hard to do better than per capita GDP as a measure of subjective well-being.

Erik writes:

If only the government didn't think it was in the business of trying to promote happiness, and instead left it to individuals. Then it would feel it less necessary to collect statistics.

My feeling is that "growth" is mostly an anchoring effect in some sense. If the data were not collected, we wouldn't bother with it either. Sir John Cowperthwaite saved Hong Kong's economy by not generating any economic data for the London government.

The data as such are also to some extent arbitrary, subject to various ideas of what a basket should contain. The data only have absolute meaning in a static society with zero growth, where everything stays the same. Ancient Egypt for example.

Inventing penicillin, aspirin, and safe drinking water each probably did not raise GDP by much, or at all, but prolonged life expectancy by decades.

To a non-economist, the first thought that comes to mind when we hear the word "growth" is "more of the same thing". And why would we want three, four or five cars?

But "growth" also means "improvement". But "improvement" in what sense? To whom? Measured how? The utility function simply can't be computed. I'm not a number, I'm a free man...

Comments for this entry have been closed
Return to top