Early Friday, the Bureau of Economic Analysis released its third look at economic growth in the first quarter. The results were disappointing: BEA now estimates that Q1 growth was only 2.7%, down from the prior estimate of 3.0%. A key reason: consumer spending was weaker than previously thought.
As I noted in May, the monthly release of GDP data is inevitably followed by commentators claiming that “consumer spending makes up 70% of the U.S. economy” (see, for example, here). Unfortunately, that isn’t right. Consumer spending appears to be about 70% of the economy based on a seemingly obvious calculation (consumer spending divided by GDP), but that ignores the way that macroeconomic accounting handles imports. For reasons detailed in my earlier post, careful analysis suggests that the actual ratio is about 60%.
One reason the 70% error is so common is that doing the correct calculation requires a great deal of work; for example, you need to estimate the fraction of consumer purchases that come from imports. If we want commentators to start using the right figure, we need an easier way to get the idea across using the information reported in the headline GDP release.
Here’s one idea: Compare consumer spending to a measure of overall demand. To do so, we start with the usual macroeconomic identity:
GDP = C + I + G + X – M,
which says that GDP equals Consumer spending, Investment, Government spending, and eXports minus iMports (which are subtracted to avoid double-counting). Looking at this identity, you see that C, I, G, and X can be viewed as measures of demand from consumers, businesses, governments, and overseas markets, while M is a measure of supply from overseas producers.
To get a more reasonable measure of the importance of consumer spending, we can calculate what share of “overall demand” (C + I + G + X) comes from consumers. As shown in the chart, that measure (in red) has been roughly 60% for decades. The usual, misleading measure of consumer spending’s importance (in blue), however, has been up around 70% over the past decade, but used to be lower back when imports were smaller.
The C / (C + I + G + X) measure of consumer spending’s importance is hardly exact. For example, it doesn’t consider how much consumer spending actually comes from imports. However, it’s the simplest measure I could think of that comes close to the right answer. But maybe readers have an even better idea?
P.S. Thanks to Cornelia Strawser for helpful discussion of this measurement challenge.
Shouldn’t your denominator be Gross Domestic Purchases or
GDP less exports plus imports?
The other question that occurs to me is how to treat residential investment. In the national accounts it is treated as investment because it has a long live and provides a stream of services over time.
But it is also consumption and maybe should be added to the personal consumption data.
It is consumption rather than investment because it is not a tool that is used to produce other goods as capital equipment does.
It is like the bromide from the 1990s that the Japanese had beautiful factories and miserable homes while the US had beautiful homes and miserable factories.
Rather than arguing about what the right number is, it might help to think first about what’s really being shown. The issue is arising because we’re comparing different things in the ratio: Consumer spend is a measure of demand, GDP is a measure of supply. Since we have a significant trade gap, total supply and demand differ, and that’s what causes the issue.
So what you’re arguing for is making this more “apples-to-apples”, asking the question of what percentage of total demand is created by consumers? This is valid, but is perhaps a different question.
We are conditioned to think about production as being the true basis for an economy, because demand cannot support the economy in the longterm without supply (short term priming the pump only helps overcome “irrational” behavior and leads to inflation in the absence of goods). Thus, production IS the economy and IS the appropriate denominator. The difference between the production based economy and the demand based estimate should (theoretically) go away in the long term. So 70% is the right number to use (in my opinion) because it gives a more accurate view of what percentage of our production (long term capacity to support ourselves) goes to consumer spending.
This seems to me more helpful for policy discussions. Let’s take an extreme example, and assume that C=100% of GDP, and I+G = 50%, balanced by imports. Would we prefer to raise a red flag and say that we are spending everything that we produce (leaving nothing for investment in ourselves), or conceal it by saying that 67% of our spend is consumer spend?
Hi DSM — Thanks for your thoughts. I would summarize as follows: “Wait a minute Donald, the right answer always depends on what question you are asking. If the question is how much are we spending on consumption relative to our income, then 70% is a pretty good number.”
Good point. My argument applies to a different question, namely how much does consumption affect short-term (and long-term, but this usually comes up in the short-term) movements in aggregate demand. For that question, I would argue that 60% is about right for the U.S. economy today. And 67% would be the right answer for your helpful example (assuming imports are used proportionally by the C, I, and G sectors).
But if the question is about the extent to which we are living within our means — or not — then 70% is the right figure for the U.S. and 100% is right in your example.
Thanks,
–Donald
Rather than asking them to calculate C+I+G, why not say GDP+Imports? They already have one and the value of the other ought to be a fairly easy look-up.
Mr. Marron: To correctly adjust the distribution of GDP components for the influence of imports requires a little more than just removing the total value of imports from total GDP and reestimating the distribution of the remaining components to this new total. To compensate for imports those sold to each individual demand component must be removed at an industry level before a new distribution is calculated.
To accomplish this requires first the allocation of imports sold directly to each component by industry followed by the allocation of imports used as production inputs to the purchases of each component. Approximately one half of the annual total of industry outputs is sold directly to demand while the other half is sold as inputs into the production process; the same is true for imports. Not only does the total value of each demand component (including imports) vary from year to year but so does the industry distribution of these. The only way to account for the annual variations by total and by industry is with input/output analysis.
Hi Art — Yes, that’s exactly the problem. To get the precise answer you need to do the kind of manipulations you described in your paper. But that’s not a plausible solution for most users of these data. Unless BEA does these calculations somewhere deep in the release tables (?), we need an approximation. C / (C + I + G + X) or, as D. Watson say, C / (Y – M) strikes me as the easiest candidate for many users. Thanks.
Mr. Marron: Thanks for your quick response to my reply. I agree that adjusting the final demand components to reflect the portion supplied by imports is neither easy nor straight forward. However BEA makes available annual I/O tables for the years 1998 to 2008 along with certain earlier years. Since this data is available moderately sized organizations could make the adjustment on their own. I would think that this adjustment is of such interest that a little pressure on BEA might induce them to think outside the box and do it themselves and make it available to the public along with the GDP accounts.