Pitfalls in GDP Accounting
By Robert P. Murphy
The single most important measure in discussions of national economies is Gross Domestic Product (GDP). Most commentators take it for granted that real GDP per capita is effectively the same thing as “standard of living” and that the growth rate in real GDP is a good measure of the effects of economic policies.
Although the concept of GDP dominates economic and policy discourse, it comes with serious pitfalls. Good textbooks stress some of these dangers, but some additional problems are rarely discussed. In this article, I review some of the standard warnings and then elaborate on some of the subtler ones. Although GDP is a useful concept, we must take care to avoid what David R. Henderson has dubbed “GDP fetishism.”1 The GDP concept can both overstate and understate the well-being of individuals, and even highly competent economists often commit mistakes when using it.
GDP: Definition and Calculation
One formal definition of GDP is “the market value of all the final goods produced in the entire country in the course of a year.”2 The “market value” component is important because economists use money prices to solve the problem of aggregation. For example, if one household spends, say, $20,000 on a new car, while another household spends $100 attending a baseball game, then the new car counts as much as 200 baseball games in the GDP calculation.
The purpose of restricting GDP to final goods (and services) is to avoid double counting. For example, suppose that a farmer sells a quantity of wheat for $100; a miller takes the wheat and, after refining, sells flour for $120; a baker takes the flour and, after baking, sells loaves of bread for $150; and, finally, a retailer packages the loaves and sells them in the grocery store for $200. Since the purpose of this operation is to provide food to consumers, the proper measurement of output is “$200 worth of bread.” We can calculate this either as the $200 that the consumers spent on final consumption or as the “value added” at each stage of production ($100 + $20 + $30 + $50). But we don’t add up the total revenues at each stage because that would overstate how much output was being newly produced in a given period.
When estimating GDP with real-world data, economists use two approaches. One is to add up the total income earned by the various factors of production. The other is to add up the total expenditures on final goods and services that are produced within the country. These two approaches are formally equivalent because in a given transaction, the seller’s income equals the buyer’s expenditure.
In practice, it is often easier to obtain expenditure data, which is why most people associate GDP with the popular formula: GDP = C + I + G + (X-M), which stands for (private) Consumption plus (private) Investment plus Government consumption and investment, plus Net Exports (exports minus imports).
Now that we’ve reviewed the basic concept and formula, we can discuss several of the pitfalls in taking the GDP concept too seriously—Henderson’s “GDP fetishism.”
Standard Warnings About the GDP Concept
Textbooks warn readers that the GDP concept has drawbacks. For example, official estimates cannot include black market activities. More generally, any output that is not bought on a market will not be included. The classic, though historically dated, example of this is a man marrying his housekeeper, who ends up performing the same tasks after marriage. In this example, the marriage would result in a drop in official GDP, even though the woman’s work remains the same. A more modern example is the typically unpriced services that billions of people receive from the Internet—an enormous amount of production that does not show up in the official GDP statistics.
It’s important to remember that GDP is a gross measure of output (not a net measure), meaning that the GDP concept fails to account for the “using up” of valuable resources that may have been necessary in order to generate the measured flow of output. Textbooks often give the example that standard GDP calculations do not include the harmful impacts on environmental quality that may result from economic processes. Consider, for example, producing 1,000 cars in two ways that are alike in every respect, except that one produces air pollution. There’s a potentially big difference in human welfare, yet standard GDP calculations would score them the same.
Similarly—although it is not stressed as much as the environmental aspect—suppose that one method of producing 1,000 cars draws down on the fixed stock of iron ore located in a country’s mines, while another method relies on only renewable resources. Again, standard GDP calculations would score the two methods as equivalent, ignoring the “deductions” from the wealth of the country in the form of mineral resources.
Government Spending and Leisure
In his Econlib article, Henderson provides two other examples of problems with GDP. The first is that government spending is treated as equivalent to private spending. Henderson’s example is the budget of the Transportation Security Administration (TSA). It makes sense to say that if consumers spend $6 billion on cars, then this is “$6 billion worth of production.” But if the TSA spends $6 billion hiring employees to scan and/or pat down airline travelers, does that really represent “$6 billion worth” of services to Americans?
A second problem Henderson cites is the omission of leisure from the GDP calculation. As an extreme example, suppose that workers are suddenly twice as productive and choose to spend only half as much time at work while producing the same amount of output. The GDP calculation would show no change from one year to the next, though in reality, our hypothetical people would be much better off.
We can go the other way and see how GDP could overstate well-being because it omits leisure. Suppose that a hurricane destroys much of the housing in a small country. The following year, the residents work twice as many hours to consume the usual flow of goods and also rebuild their houses. In this scenario, measured GDP might double, but leisure is much less.3
Confusing Accounting with Causality
People often use the GDP formula to erroneously derive conclusions about economic causation. For example, in the wake of the financial crisis of 2008, some proponents of “stimulus” spending argued that a boost in government spending on infrastructure would obviously raise GDP because the textbooks tell us that GDP = C + I + G + (X-M). If the government increases G, according to this argument, GDP obviously must increase, as an increase on the right-hand side of the equation “had to” be balanced by a comparable increase on the left-hand side.
However, the textbook formula does not mean that (say) a $100 billion increase in G must go along with a $100 billion increase in GDP. For all we know, a $100 billion increase in G might cause a $40 billion drop in private consumption (C) and a $60 billion drop in private investment (I). In this case, GDP would remain unaffected, and the private sector would shrink to perfectly offset the growth in government. The textbook GDP formula is consistent with both outcomes, so the accounting tautology, by itself, tells us nothing about the impact of an increase in government expenditures on the economy.
Paul Krugman also seems to have misused the GDP formula when he argued that free trade deals cannot increase total demand because an increase in exports is counterbalanced by an increase in imports.4 Yet Krugman’s claim rests on an implicit theory that may be false.
Here’s a simple counterexample: Suppose that a small island nation is heavily dependent on international trade. It originally has a GDP of $100 billion, consisting of C = $74 billion, I = $10bn, G = $15bn, X = $100bn, and M=$99bn. The financial press might report that this economy rests 74 percent on consumption and that “net exports contribute only 1 percent to output.” However, a war breaks out, and a rival power places a naval blockade around the island, and, thus, its people soon run out of crude oil and beef. Economic privation ensues. Businesses refrain from investment, while households and the government adopt extreme austerity measures. The next year, the government reports: C = $15bn, I = $0bn, G = $5bn, X = $0, M = $0. With a new GDP of $20bn, the economy has crashed by 80 percent.
“Thus, we have yet another example where the GDP formula has led even a Nobel laureate on trade to make a simplistic mistake.”
Now, if he were to be consistent with his argument about trade deals having no net effect on total demand, Krugman might look at my hypothetical island example and claim that the change in international trade reduced GDP by only $1 billion; the collapse of exports was almost perfectly counterbalanced by a collapse of imports. Yet it is clear in this example that the blockade is what destroyed the economy and that restoring international trade is the path to recovery. Thus, we have yet another example where the GDP formula has led even a Nobel laureate on trade to make a simplistic mistake.
Using the GDP Formula to Erroneously Infer Economic Importance
It is also typical for commentators to use the GDP formula to assess the relative importance of its various components, even though this may be misleading. For example, consumption represents about 70 percent of GDP in the United States, leading many pundits to fret when consumers do not spend as much as expected during the holiday rush. These pundits imply that when households save more, it threatens economic growth, and that if only we had Christmas every month, there would be no unemployment problem. However, such thinking relies on a theory of how the economy works, and not simply on the GDP statistics. Indeed, according to economic models that assume markets tend to clear—that is, that quantities demanded and supplied are equal—a large increase in household saving will lead to lower consumption, but also will foster higher investment spending. In these models, a drop in consumption spending does not threaten employment and paves the way for faster long-run growth because of the increased investment.
Because of the potentially misleading size of consumption in the standard GDP formula, Mark Skousen has lauded the Bureau of Economic Analysis’s recent decision to report estimates of “Gross Output,” which include the expenditures by businesses at each stage of production.5 Returning to our simple bread example from the beginning of this article, in addition to the $200 spent by consumers on loaves of bread, the Gross Output figure would include the miller’s $100 investment in wheat, the baker’s $120 investment in flour, and the retailer’s $150 investment in the wholesale bread.
To reiterate our earlier discussion, the conventional GDP approach doesn’t include these expenditures by the businesses at each stage in the bread industry because they represent spending on intermediate goods, not final goods. There is a reason for this decision: we want to avoid double counting, so, for example, we don’t want to count the $120 spent by the baker on flour if we’re already counting the $200 spent by consumers on the final loaves of bread. However, the danger here is that the GDP concept gives an exaggerated importance of consumer spending in the overall economy. In particular, even though the $120 spent by the baker on flour doesn’t add anything to the GDP calculation, it is certainly economically critical. If all of the bakers suddenly decided to refrain from reinvesting their revenues in buying more flour, then the output of bread would pretty quickly come to a screeching halt. No matter how much money the final consumers were willing to offer the grocery store owner, she couldn’t sell them any bread if the bakers had previously stopped the gross reinvestment of their proceeds.
Arbitrary Time Frame: New versus Gross Investment
Our discussion of the bread industry revolved around intermediate versus final purchases. If a baker spends $120 buying flour that is used up during the year, then that expenditure does not count in GDP. On the other hand, if the baker spends $10,000 buying a brand new oven, then that is classified as a final good and does contribute $10,000 to that year’s GDP.
However, this distinction is somewhat arbitrary. Suppose that, instead, we calculated GDP over the entire lifespan of a new oven rather than over one year. In that case, the $10,000 spent on the oven in the beginning of the period would be economically equivalent to the $120 spent on the flour; all of these resources would be “used up” in the production of final loaves of bread for consumers. Therefore, the GDP calculation is sensitive to the time period chosen, even though this shouldn’t be relevant to economic well-being.6
Although the concept of GDP is useful, it has achieved unwarranted importance in modern discussions of the economy and government policy. Economists should do a better job of warning the public of the many drawbacks to its use.
David R. Henderson, “GDP Fetishism,” Library of Economics and Liberty [Econlib] featured article, March 1, 2010.
This particular wording is quoted in David R. Henderson’s Econlib article, and comes from Paul Heyne, Peter Boettke, and David Prychitko, The Economic Way of Thinking, 10th edition, p. 352.
Note that these observations supplement the discussion of the “broken window fallacy” made famous by Frederic Bastiat; I am pointing out that even if a disaster genuinely led to (correctly calculated) total production in the following period, it still would not indicate improved well-being, because GDP does not include the “production” of leisure.
Paul Krugman, “Things free trade doesn’t do,”New York Times blog post, December 10, 2009. For my full critique see Robert P. Murphy, “Krugman Falls Into the Keynesian Accounting Trap,” Mises Daily article, December 21, 2009.
For a summary of Skousen’s critique of the standard GDP formula and why he thinks his measure of Gross Output is superior, see Mark Skousen, “Gross Output (GO) a Major Discovery as a Top Line in National Income Accounting,” October 12, 2016.
Some readers might think that capital depreciation takes this into account in conventional GDP calculations, but this is not correct. Depreciation is a component in the income approach to GDP estimation, but it does not appear in the expenditure approach. In particular, the expenditure approach would include the $10,000 market value of the oven when it was purchased as part of that year’s GDP, but the flow of bread output would not be reduced (for GDP calculations) by a depreciation charge on the oven in subsequent years.