
On his blog yesterday, Harvard economics professor Greg Mankiw stated that the idea of excluding government spending from GDP is “amazingly stupid.” Unfortunately, he didn’t say why. It would be nice to know why.
I’m not sure if Greg is aware of this: a Harvard University professor who won the Nobel Prize in economics in 1971, Simon Kuznets thought it was a good idea. As I wrote in my biography of Kuznets in David R. Henderson, ed., The Concise Encyclopedia of Economics:
Simon Kuznets is best known for his studies of national income and its components. Prior to World War I, measures of GNP were rough guesses, at best. No government agency collected data to compute GNP, and no private economic researcher did so systematically, either. Kuznets changed all that. With work that began in the 1930s and stretched over decades, Kuznets computed national income back to 1869. He broke it down by industry, by final product, and by use.
I added:
Although Kuznets was not the first economist to try this, his work was so comprehensive and meticulous that it set the standard in the field. His work was funded by the nonprofit National Bureau of Economic Research, which had been started in 1920. Kuznets later helped the U.S. Department of Commerce to standardize the measurement of GNP. In the late 1940s, however, he broke with the Commerce Department over its refusal to use GNP as a measure of economic well-being. He had wanted the department to measure the value of unpaid housework because this is an important component of production. The department refused, and still does.
Now, it’s true that what earned Kuznets the Nobel Prize was not his work on GNP but his work on economic growth. But clearly he was an expert on GNP and, by extension, GDP.
So why did Kuznets want to exclude government spending from GNP? Here’s how Onur Ozgode, put it in “The Invention of Economic Growth: The Forgotten Origins of Gross Domestic Product in American Institutionalist Economics,” Promarket, October 31, 2021:
Operationalization of national income statistics as a macroeconomic interface required a final and yet controversial modification in Kuznets’s GNP: the inclusion of the state as an economic sector in national income estimates. Kuznets excluded the state in these calculations because he believed public services were an input for private production. Since businesses factored in the taxes they paid for these services in pricing their products, their inclusion would cause double counting. Yet, an emerging group of economists, calling themselves “Keynesians,” argued against Kuznets, pointing to the inconvenient fact that his justification did not account for a state that ran large deficits. Because they believed deficit spending was the only effective policy for economic recovery, policymakers needed a measure of the state’s contribution to economic activity to fine tune economic stimulus.
Who’s right: Kuznets or Mankiw? I don’t know. Whose idea is “amazingly stupid”? Neither Kuznets’s nor Mankiw’s. Whose idea is just stupid? Neither again.
The accompanying picture is of Simon Kuznets.
READER COMMENTS
John R. Samborski
Mar 4 2025 at 8:09am
Mr. Henderson,
Can’t you show a picture of Simon Kuznets that at least shows his whole face? Thanks.
Rick Samborski
David Henderson
Mar 4 2025 at 9:40am
Click on the link to my bio of him and you’ll see it.
John Hall
Mar 4 2025 at 8:10am
It all depends on what you’re trying to measure. GDP less government spending can have value as an indicator if you want to focus on private output. But, if you calculate GDP with government spending, it is trivial to subtract out government spending (on a nominal basis, a little more complicated for real, the BEA provides a real estimate for GDP less government spending, trade, and inventories).
David Henderson
Mar 4 2025 at 9:41am
True.
Craig
Mar 4 2025 at 8:13am
Back in the day I had a suspicion that government spending, to the extent it included transfer payments, might be double counting things? If I spend a $1 that goes to C. Ok….if I pay a $1 in Social Security tax it goes government who l, when disbursing that dollar that $1 goes to G, but now the retiree spends it and doesn’t that dollar get counted in C?
David Henderson
Mar 4 2025 at 9:42am
Government spending in GDP does NOT include transfer payments. It’s government expenditures on goods and services: military hardware, payments to government workers, red tape, etc.
Tyler Watts
Mar 4 2025 at 9:44am
The “G” in Y = C + I + G + NX only includes government _purchases_ of goods and services (e.g. building roads and bridges, building inventories of munitions, paying government employees). Transfer payments like welfare or Social Security benefits are excluded from the “government” part of GDP.
Tyler Watts
Mar 4 2025 at 9:47am
Dr. Henderson was quicker on the draw! 😁
Craig
Mar 4 2025 at 9:50am
Thanks! I probably learned and forgot that, its been a while since I was in Econ 101 im the 1900s
David Henderson
Mar 4 2025 at 12:32pm
Just call me Quick Draw McGraw. 🙂
Craig
Mar 4 2025 at 1:16pm
For an encore, some winning stock pics please? Thanks!
David Henderson
Mar 4 2025 at 2:41pm
Vanguard Total Market Index, for the long term.
Jon Murphy
Mar 4 2025 at 8:47am
Thanks for posting this. It revealed a major error in my understanding: I thought Kuznets wanted to include government spending and others wanted to exclude it. In fact, I said so quite confidently to Michael Gibberson on Twitter just yesterday. Time to go correct some mistakes.
David Henderson
Mar 4 2025 at 9:42am
You’re welcome.
Andrew_FL
Mar 4 2025 at 2:45pm
Technically Kuznet’s proposal was narrower than this. Kuznets proposed excluding war time expenditures on war related output from national income, conceptually. During the war he had in mind, this would indeed have excluded the bulk of government purchases from GNP, but it would not do so today.
The issue I have with the idea of deducting all or some portion of government purchases from GDP is that it breaks equivalence with GDI. Regardless of how one views the value the government gets for what it pays, the income received buy the sellers of product to the government is as much income as any other source would be. It seems difficult to justify excluding it.
David Henderson
Mar 5 2025 at 2:02pm
Thanks, Andrew.
I looked around and couldn’t find a cite for your statement. I’m not challenging your claim. I am hoping that you have a cite.
Andrew_FL
Mar 7 2025 at 1:03pm
I think Kuznets mentions and puts forth his alternative aggregates in a few different papers but I believe the one I have in mind is National Product in Wartime
David Henderson
Mar 7 2025 at 2:09pm
Thanks. I’ll check it out.
steve
Mar 4 2025 at 4:50pm
You have always, meaning since I discovered it, been able to find GDP and GDP with the federal share excluded on FRED ie it has always been available and easy to find if you wanted. There certainly hasn’t been any effort to hide the numbers. Seems to me that both numbers would be good to know. However, just like there are multiple numbers you can use to look at unemployment or inflation we have traditionally used GDP which i think works well enough for most discussions.
Steve
Monte
Mar 4 2025 at 8:06pm
Yes. Mankiw is not without his own critics, but at least they provide hooks to hang their beef on:
What Is Our Children Learning: Or, Greg Mankiw and the Terrible, Horrible, No-Good, Very Bad Textbook (now with a single-page goodness!)
A Response to Greg Mankiw – Part 1
Did Greg Mankiw really just brandish his $170 textbook as evidence of the benefits of unfettered trade?
Occupy Greg Maniw!
Notwithstanding its obvious drawbacks, wouldn’t excluding G from the GDP formula provide a clearer picture of how the private sector is performing without the distorting effects of government spending? And if we see over time that GDP growth scales closer to private sector output, might we conclude that less government intervention in the economy is necessary?
Craig Richardson
Mar 5 2025 at 9:55am
David,
Nice piece and timely. One twist on this debate about how to measure the economy is how authoritarian regimes can use GDP statistics to twist the truth around the supposed health of their economy and gain financial grants and loans as a result from other countries.
After I wrote a book on Zimbabwe’s collapse in 2004 and many articles hence due to the collapse of property rights in the early 2000s after expropriation of commercial farmland, I was puzzled by the country’s economic resurgence between 2009-2011. Its real GDP was growing at 7.3%, ranking it one of the world’s fastest growing countries.
The puzzle was this- property rights were still demolished, so how could the a market work so well without them?
I wrote an article for the Cato Institute, “Zimbabwe: Why is one of the world’s least free economies growing so fast?” https://www.cato.org/policy-analysis/zimbabwe-why-one-worlds-least-free-economies-growing-so-fast
It turned out that Zimbabwe’s spectacular turnaround from minus 18% the year before was due to several things. It dollarized, but that was a necessary but not sufficient condition.
In fact, the government juiced the economy, with deficits rising from $124 million in 2008 to $583 million by 2011. But how could they finance this, if there was no appetite for Zim government bonds?
Answer: the IMF and the Chinese government gave the Zimbabwe government hundreds of millions of dollars in “hardship grants” and loans. Zimbabwe’s economy was like an athlete who relies on steroids to build muscle mass.
I found that 65% of all GDP growth during this time was due to government deficit spending.
Between 2008-2009, nominal government (G) grew by 345%. The private sector grew by in nominal terms by 13%.
My point in all this, is that for developing countries, the shorthand for a nation’s health has long been GDP. It’s been *assumed* that government spending is fairly stable, so that PDP (private domestic product) is highly correlated.
In the case of Zimbabwe at least, the lesson is that foreign aid can be given all too quickly to corrupt regimes using GDP as a shorthand version of “success,” without asking some more penetrating questions about the health of the private sector.
David Henderson
Mar 5 2025 at 2:03pm
Thanks, Craig. Interesting story.
It implicitly backs up what Andrew said in his comment above. The “G” is really misleading if it changes a lot in a relatively short time.