The Economist is a serious magazine engaged in rational discourse, whatever you think of its political philosophy. Exceptions happen. One can be found in the opening section (“The World this Week”) of the current issue (May 25, 2019). It is all the more surprising as it relates to a simple and narrow topic. We read:
There was some head-scratching this week, as data showed Japan’s economy growing by 2.1% in the first quarter at an annualised rate, defying expectations of a slight contraction. Most of the growth was explained by a huge drop in imports. Because they fell at a faster rate than exports, gdp rose.
As I and other economists, including Scott Wolla of the St. Louis Fed, have explained, a change in imports cannot explain measured GDP growth, for the simple reason that imports do not enter into GDP at all. GDP, as its name (gross domestic product) indicates, measures gross domestic production only. At some stage in the practical task of measuring GDP, national statisticians do subtract imports because they were already included in the raw data they work from (in consumption expenditures, for example). The final result is, at it must be by definition of GDP, that imports did not affect the calculation of GDP or of GDP growth at all.
Once we understand what GDP is and how it is measured (more about this in the links above), the argument is a simple accounting and logical one: whether M represents imports or any other variable, it is always true that +M-M=0.
One way to see the main point is the following. Suppose “the country,” that is, all individuals and their intermediaries, exported all the GDP they produced and used the proceeds to import everything they consumed. We could not say that imports explain why GDP is zero. GDP would not be zero: it would still be equal to domestic production, which would happen to have been all exported and exchanged for imports.
Thus, the drop in imports in no way explains the higher growth of Japanese GDP.
READER COMMENTS
Mark Z
May 28 2019 at 4:53pm
Playing devil’s advocate, can we be sure that it didn’t happen that: a decline in imports led to an increase in demand for (and subsequently supply of) domestic production of some goods that were formerly imported.
The (imo more plausible) alternative of course is that an increase in domestic production led to substitution of domestically produced goods for imports, the increase in GDP therefore causing the decline in imports, rather than the other way around. But one would then expect to see increases in GDP followed by (at least in the short run) a decline in imports, right?
Pierre Lemieux
May 28 2019 at 6:31pm
Mark: Let’s suppose you are right. The measured (ex post) GDP figure cannot explain why you are right, because imports are not part of GDP. If you have not already done so (I suppose you have, but repeating the reference may be useful to other readers), it’s worth reading Scott Wolla’s September 2018 Page One Economics short article.
Aleksander
May 29 2019 at 3:05am
This makes it sound like a decline in imports absolutely could influence the calculated GDP, if not the “real” GDP of the economy. Fewer imports means that the number to be subtracted from the “raw data” is smaller; and unless the raw data and the import estimates are of equal precision, this would indeed impact the final result, and make it look like import numbers influenced the GDP.
Pierre Lemieux
May 29 2019 at 11:47am
The two numbers are the same M. M has been added and the statisticians subtract M. You will have more details if you follow the links in my post.
Benjamin Cole
May 29 2019 at 6:40am
Well, maybe.
Of course, if Country A buys less smartphones from Country B, then money formerly exported is then spent on domestic producers in Country A, boosting GDP. Only if there is zero slack in the Country A economy would this not apply.
So, when Japan imported less, the money not exported instead circulated domestically, boosting GDP.
The US seems to even yet have plenty of slack. The unemployment rate gives only a squishy feel to the amount of labor on the sidelines. Seems like more and more people are re-entering the labor force (let us hope attracted by higher wages).Capacity utilization rates are still low-ish by historical standards. I know of no industry straining under demand (except perhaps housing in regions where new housing cannot be built, or infrastructure in regions where new infrastructure cannot be built)
Anything that tightens up US labor markets is welcome. We can shoot for a couple generations of tight labor markets, or roll-out the red carpet at the White House for AOC.
Pierre Lemieux
May 29 2019 at 12:09pm
My response would be the same as I gave to Mark Z above—although I do not agree with the (Keynesian?) theory you adumbrate including the implicit idea that there is no PPF and that the economy can wander anywhere in goods space.
Jon Murphy
May 30 2019 at 9:06am
Not necessarily. As Pierre said, it assumes resources are not scarce, or the country is operating inside the PPF for some reason, or costless resource allocation.
Imports are not homogenous goods and they are not necessarily substitutes for domestic goods. If, for example, Cell phone importation is limited, it does not immediately follow (even assuming slack) that the limiting country’s GDP will be boosted. If resources are diverted from more productive uses (which, by definition and experience, they are), then yes domestic cell phone production will be boosted, but at the expense of something else. This is true even if there is slack. Hell, trade restrictions made the Great Depression worse, and that was a period of extreme slack.
Do not make the mistake of confusing an accounting identity with an economic theory. Someone recently wrote the following metaphor (I can’t remember who, but I think it was here):
A= W + S, which just means that the total number of Animals (A) on a given island is comprised of the number of Wolves (W) and Sheep (S). It does not logically follow that increasing the number of Wolves necessarily increases the number of Animals on the island.
Jon Murphy
May 30 2019 at 9:07am
Oops, Mike Davis beat me to it below with Nick Rowe’s blogpost.
mike davis
May 29 2019 at 8:48am
Accounting identities may be good way of summarizing information but they don’t explain anything. This point was made in a beautiful blog post from Nick Rowe
https://worthwhile.typepad.com/worthwhile_canadian_initi/nick-rowe/
Pierre Lemieux
May 29 2019 at 11:56am
I agree with you and Nick Rowe. A fortiori, one cannot use an accounting identity to show that both sides of the identity will change in one way or another because some variable absent from the identity has changed.
Michael Rulle
May 29 2019 at 9:04am
Excellent point. People confuse accounting with economics all the time——it is understandable, but still wrong. My favorite is the idea that we have a Social Security Trust Fund. Of course we do not in any economic sense. We have a de facto unified cash in and cash out budget mechanism, borrowing included. The SSTF is useful however, as it does track how much has been diverted from payroll taxes to budgetary expenditures unrelated to Social Security.
R. Miller
May 29 2019 at 5:16pm
You’re absolutely correct about the SSTF. Both the treasuries in the trust fund and the liabilities incurred by social security represent promises to pay future beneficiaries from future revenues. What the trust fund does is: 1) make explicit that SS is backed by the full faith and credit of the US, 2) accounts for these unfunded liabilities, and 3) contained in 2 is a proper accounting of US debt. The trust fund certainly is not a lock box of money saved for current taxpayers for their future benefits.
I for one think the trust fund should invest in corporate securities or mutual funds for better yield, inflation protection, and political mischief.
Mark Bahner
May 30 2019 at 12:06pm
I think using the “full faith and credit of the U.S.” phrase makes SS “Trust Fund” bonds appear to be like other U.S. government bonds that are salable in the open market, but SS “Trust Fund” bonds aren’t salable in the open market.
If the federal government refuses to honor Treasury bonds salable on the open market, then the credit worthiness of the federal government is harmed, which harms everyone in the U.S. If the federal government refuses to honor the bonds in the SS “Trust Fund” only the recipients of Social Security are harmed. That’s a huge difference that the “full faith and credit of the U.S.” phrase inappropriately blurs.
Benjamin Cole
May 30 2019 at 8:53am
As I contemplate this post, and many others in the “right wing” or “libertarian” circles, I see the “free market” types (with whom I identify in many regards) dancing closer graveside, where the Trumpers can give a shove.
The right-wingers are ever the apologists for income stratification, huge and chronic trade deficits, open borders for immigrants, an entangling and huge military and so on.
Yes, there may be valid reasons for some of these stances. But making these the front-burner topics, and never addressing such issues from the perspective of the huge (and working) employee class is a mistake.
Take the corporate tax cuts. Yes, a good idea. Trump also doubled the standard deduction, a better idea. Why did not the right wing call for a tripling of the standard deduction, and hold off on corporate tax cuts for a year or two? Working is just as vital as investing, and the world seems completely saturated with capital as it is.
Why all the crying for tax cuts, but never on onerous payroll taxes?
In other words, the right-wing should position itself as not for welfare, but certainly and absolutely and powerfully for eliminating taxes on wages to the greatest extent possible. Say, no taxes on the first $100k in wages. That would be a winning position.
Why is it “right-wing” to want lower taxes on capital but go mute on wages? (Even Ronald Reagan said it is bad to tax wages more heavily than capital).
Everyone knows why upper-income people want lots of cheap imported labor. But if that it the plank of the right-wing, you will lose.
In other words, the right-wing should constantly press for measures that tighten labor markets, by increasing aggregate demand or open avenues for self-employment, such as the decriminalization of push-cart vending, and the elimination of property zoning.
Making sure workers keep all their wages and can find affordable market-rate housing in free and open property markets strikes me as worthy goals.
The intellectual right-wing is pickled in formaldehyde.
Trump owns the GOP now.
Jon Murphy
May 30 2019 at 9:08am
It’s not clear to me what this rambling post has to do with anything.
Benjamin Cole
May 30 2019 at 7:59pm
OK, I was guilty of rambling.
But just remember, the unorganised disheveled Trump campaign completely flattened the GOP establishment in 2016.
The Establishment GOP talks the way you talk.
You know what else? I do not know why my spellcheck has gone British.
art andreassen
May 30 2019 at 12:25pm
Sir: Your example of a country exporting all its production and importing all its demand only results in a net export value of zero in the accounts. The other demand components will show a value equal to the imports they purchase so GDP will equal the value of exports. This is a perfect example of the problem with unsdjusted accounts. In your case each component would be equal tits value of imports, exports would equal production and to imports so they would total, before netting out imports, twice domestic production. A decline in imports impacts GDP depending on whether the decline is a result of a decline in demand or if it is a shift of demand from imports to domestic production.
You point out the problem of the accounts concerning imports. The GDP value is net of all imports while the individual components include imports in their values (even the export component contains imports). That is the cause of the common misconception that PCE is 70% of GDP. Imports are 15% of PCE so to be consistent with the GDP value this 15% must be removed before the calculation. Until a few years ago the BEA correctly did not include a table showing the distribution of the components GDP but oddly all of a sudden it has appeared without but adjusting for the import content.
Unfortunately removing the imports from each demand component is not straightforward but can be approximated using input/output analysis. See: “Induced Consumption: Its Impact on Gross Domestic Product and Employment”, Chentrens and Andreassen, 17th Federal Forecasters Conference, 2009. In a paper: “The U.S. Content of ‘Made in China'”, Hale and Hobjin, FRBSC Economic Letter, August 8, 2011, fig. 1, shows the import content of PCE to be 16%.
art andreassen
May 30 2019 at 10:17pm
Sir: Your example of a country exporting all its production for imports would result in imports netting that of the exports while having all its demand components satisfied by imports so GDP=M=X. This is a perfect example of why the accounts are misunderstood. You say imports are not in GDP but your example shows that is not true. If imports declined then all of exports, i.e., production, would not be spent and so would be purchased domestically and GDP would increase. In the real world a decline in imports can increase or have no impact on GDP depending on whether the drop is due to a shift in demand from imports to domestic production or due to a fall in demand.
You recognize that imports are embedded in demand and assume they have been adjusted out of GDP by national statisticians, if you can find out where and how please inform. Imports are netted only out of the GDP value but not from the demand components, in reality the use of net exports actually nets imports from exports and thus from GDP. All individual components contain imports. This is why saying PCE is 70% of GDP is incorrect, imports are 15% of PCE which should be so adjusted before the ratio is calculated. Until a few years ago the BEA correctly did not include a distribution of GDP and its components in the SCB but for some incomprehensible reason it started publishing a distribution not adjusted for imports.
Unfortunately removing imports from the individual components is not straightforward but can be approximated using input/output analysis. See: “Induced Consumption: Its Impact on GDP and Employment” , Chentrens and Andreassen, 17th Federal Forecasters Conference, 2009. A paper: “The U.S. Content of ‘Made in China'”, Hale and Hobjin, FRBSF; Economic Letter, August 8, 2011, fig. 1. presents an historical series of the import content of PCE.
Jon Murphy
May 31 2019 at 7:32am
No. GDP would stay the same. In the case above, instead of GDP = X, GDP would equal C. Merely shifting from X to C does not change GDP at all.
A shift from imports to domestic consumption would not necessarily increase GDP. Resources are scarce. An increase in domestic production in one industry, all else held equal, necessarily means less production in another industry.
That’s simple. We have data on imports. We just subtract them out from the overall production number. That’s why imports are included in at all despite the fact they have nothing to do with GDP. The St. Louis Fed article linked to in the post has more details.
No. See any textbook or the St. Louis Fed article.
Pierre Lemieux
Jun 1 2019 at 12:01am
What Jon Murphy said. And as he said, the first Scott Wolla article is well worth reading. I would just emphasize that, in order to buy something (say, domestic production or imports), you must first have produced something to trade in (or borrowed it from somebody else who has produced it, or been gifted it). Remember that GDP is production, and it’s only because, by definition, production=expenditures=income that we can measure production as expenditures or income; it does not mean that you can spend something you have not produced (or borrowed from somebody who has produced it, or been gifted it).
art andreassen
May 31 2019 at 9:43am
Sir: We are dealing with accounting balances here. You would agree that the production of exports adds to domestic product and imports subtract from domestic product, don’t you? You do realize that using net exports as a demand component to sum to get GDP actually removes imports from exports, so imports do impact GDP. Where do the imports go, do they sit outside Oakland or Staten Island? No, they satisfy demand and are embedded in the demand components. Of course imports can be removed from shipments but that does not mean they are removed from the purchases that compose the demand components. We are dealing with the national Accounts and how the are actually created not how you imagine they are created.
Jon Murphy
May 31 2019 at 9:50am
No, I don’t. By definition of the accounting identity, it is not true that imports subtract from domestic production.
art andreassen
May 31 2019 at 10:44am
Jon: I go by PCE+INV+GVT+X-M=GDP where exports add to GDP and imports subtract from GDP . Which do you go by?
Jon Murphy
May 31 2019 at 12:51pm
Same. C+I+G+NX. In either case, imports, by definition, do not affect GDP. This is in any textbook. Or the links in this post. Or the BEA’s website.
art andreassen
May 31 2019 at 1:18pm
Jon: You’re missing the main point, your NX is arrived at by subtracting imports from exports and adding the remainder as the change in GDP. If imports had no affect on GDP they would not be included in the equation at all but both imports and exports do impact GDP separately. All the components of GDP have imports embedded in them, using the net export figure only cancels out the addition that exports make to GDP.
Jon Murphy
May 31 2019 at 2:02pm
Correct. That’s why they are not. The C, I, and G components all have imports in them. Since we cannot easily separate out imports from those components (all the data is aggregated), we simply subtract out imports from the total equation to correct for that. Otherwise, we are counting GDP too high.
Again, this is fairly standard stuff. Check out the links in this blog post or the BEA’s website.
Jon Murphy
May 31 2019 at 2:12pm
See page 8 of this BEA primer on GDP. It goes over in detail how GDP is calculated: https://www.bea.gov/sites/default/files/methodologies/nipa_primer.pdf
art andreassen
May 31 2019 at 7:40pm
Guys: Quoting from the article : ” imports do not enter into the GDP at all”. Now you are telling me they do which is what my point has been all the time. Even you admit hey are embedded in all the demand components. Because BEA can’t measure and remove them from the components it subtracts them in total from exports to remove them from the GDP total but they are in GDP. No one thinks that all imports are exported. So imports do impact demand components and thus their change has impacts GDP.
Jon Murphy
Jun 1 2019 at 8:39am
Art-
Again, no. Imports do not factor into GDP at all. They are not included in GDP. +M-M=0. Basic algebra.
This is a handout I created for my classes explaining the common mistake you’re making that Imports affect GDP.
Pierre Lemieux
Jun 1 2019 at 12:16am
Art: +M-M=0, that is, imports are not included in GDP. [0=zero. The font used on the blog for zero strangely resembles the letter of the alphabet, which is why I used code in my post.]
You write:
This is not true. Instead, you must write: using the net export figure only cancels out the addition that IMPORTS WOULD OTHERWISE AND FALSELY MAKE TO MEASURED GDP.
As Jon emphasize, you must read the Wolla article.
art andreassen
Jun 1 2019 at 12:36pm
Pierre: Again, subtracting imports from only the export component removes imports from the the calculated total of the demand components (GDP) but it does not remove imports from any individual demand component except of course exports (which is entirely eliminated). So yes, imports impact GDP since imports impact each individual component even exports. You still haven’t responded to my point that your example of the entire production going for imports would result in GDP=M=X and thus imports would supply every dollar of every demand component. If that is true then any change in imports would change GDP. That was the source of my confusing reference to net exports.
Pierre Lemieux
Jun 1 2019 at 8:54pm
Art: The individuals in the economy have produced Y (=GDP). They exchange Y for what they want (individually in a free country, collectively in a non-free society). What they want can be what other individuals in the domestic economy have produced (Y) or what foreigners have produced (you have imports M in the second case). Whatever the respective proportions of Y and M, Y doesn’t change (as an accounting matter): it is simply exchanged, after being produced, with different individuals.
Perhaps it will help if you think of Y in real terms, as a set of things—say, 1 million barrels of oil, 15 million cars, and 10,000 billion bushels of corn.
Todd Kreider
Jun 1 2019 at 2:12am
I haven’t read the “serious” Economist in almost ten years, and I assure you that I haven’t missed a thing.
Comments are closed.