The Limitations of GDP

I see that Brookings is finally coming around to the realization that gross domestic product is a pretty crude measure of economic well-being:

In a new working paper, Karen Dynan of Harvard University and the Peterson Institute for International Economics and the Hutchins Center’s Louise Sheiner conclude that changes in real Gross Domestic Product (GDP) do a reasonable job in capturing changes in a nation’s economic well-being with one important exception. They argue that the exclusion of non-market activities that increase economic well-being merits more attention, particularly given the growing importance of such activities.

They cite several areas where measurement falls short of the conceptual ideal. First, the national accounts may mismeasure nominal GDP arising from the digital economy and the operation of multinational corporations. Second, deflators used to separate GDP into nominal GDP and real GDP may produce a biased measure of inflation. For goods and services that do not change in quality over time, current deflator methods work reasonably well. For new goods and services, or goods and services that are changing in quality, current methods may not capture consumer surplus well.

I actually think it’s easier than that to recognize the limitations of GDP. A simple thought experiment will do. What if the entirety of the increase of GDP resulted from increases in the value of a single stock? That’s not too big a stretch. That’s actually happened and the company was Microsoft. Did that result in the United States being better? There’s simply no way to tell—that requires you to do what economists call “comparing utility functions”.

I would submit that the rise of importance of GDP in policy considerations was largely a post-war phenomenon that took place during a period of American history that was very different from today when incomes were much more equal than they are and an assumption of relative equality is built into the notion. I would further submit that the rise in importance was accelerated in the 1970s when econometrics came to rule the roost in economics. There’s nothing like a pseudo-empirical measure like GDP to capture the imagination of people who aren’t particularly good mathematics and who would like to appear empirical without actually being empirical.

7 comments… add one
  • Guarneri

    GDP may have a number of shortcomings but I’m not sure there is a better statistic for what it was originally intended to do, or intended to do today. (Originally intended to measure the success of the war output effort and today just measure goods and services output)

    I don’t quite understand where you are coming from on the stock value. Stock trades aren’t counted in GDP, just the commissions.

  • There have been months when all of the increase in GDP could be attributed to the sale of stock by a few of the top Microsoft stockholders. I wrote about it at the time.

    The problem with GDP as a metric is that in the past it could be used reasonably as a proxy for economic wellbeing. Nowadays not so much. Brookings’s explanation of why it is a lousy proxy is that it doesn’t capture everything but I think there’s a simpler reason. Because of the changes in income distribution GDP is no longer as much of a proxy for economic wellbeing as it used to be.

    Another very simple thought experiment. Imagine that everybody in the country lost $1,000 except for Warren Buffett who gained $330 billion. Is the country better off, worse off, or no difference? Technically, I suppose the answer is no difference but in reality the country would be worse off. How about if Warren Buffett picked up $330 billion and everybody else lost $1?

    Maybe an even better example would be if you took $1,000 from everybody and buried it.

  • CuriousOnlooker

    I thought GDP was the “monetary value” of all goods and service produced over a certain period of time. It is not changes in the net worth of the country or of a group of individual.

    Arguably that’s the classic weakness of GDP; i.e. a natural disaster destroys goods; the monetary value to replace it all is 10 billion. The GDP will go up by 10 billion even through no one is actually richer (the capital stock did not increase).

  • Sales count including sales of stock.

  • Guarneri

    Sorry to be hard headed. Trades of stock are simply not a component of GDP accounting. That’s definitional.

    Of the two types of stock sales, the closest you can come to your issue is an IPO – if the proceeds are used to purchase capital. An economist would say savings were used to make a capital investment/business investment/goods and services. But that’s a 1 time conversion of savings to capital.

    Post IPO trades are just a reshuffling of the chairs. Outside of GDP, and equivalently priced purchase and sales in the capital (savings) market. That we can leave to the sellers of stocks and Mr Buffet to reach an acceptable trade, with no effect on goods and services.

    I’m not sure GDP was ever intended to formally be a measure of economic well being. As I said earlier, it really came into vogue during WWII as a way to measure goods and services output and the efficacy of the war effort. But to the degree it measures goods and services output, and those require labor as a factor, it does correlate with employment, certainly a factor in economic well being.

    I thought you might be commenting on black market transactions. A fault for sure, but one that I doubt changes dramatically over reasonable periods of time. Further, since we focus primarily on percentage changes it is a minor error.

    Since thought experiments are in vogue. The “2% Kid” gave us, well, 2%. Let’s suppose the current economy throttles down and settles in at 3%. The last time I looked that’s a 50% greater growth rate. And if you run that out over 5-10-20 years its a huge difference in economic well being, reflected in the statistic. Flawed as it might be against a perfect statistic straw man, it is still the best we have.

  • CuriousOnlooker

    Maybe the government measures things differently. But a sale of stock would not increase GDP except for the commission, which counts as a service in facilitating the sale.

    That’s how they count home sales; existing home sales do not generate much GDP since its just the commission, new home sales do.

  • Guarneri

    That’s correct, curious. And the underlying principal is to not double count. I don’t know if Dave was just trying to juxtapose moves in the value of stock market capitalization with moves in goods and services production or what. But the notion that stock sales count in GDP is simply wrong.

Leave a Comment