Reading The Economist, I stumbled on a passage on which any economist who remembers the national accounting he learned in graduate school or perhaps even in college, would stumble, even if the confusion is pretty common (“Could America’s Economy Escape Recession?” The Economist, dated July 20, 2023):
Although GDP growth has held up well over the past few quarters, its close relative, gross domestic income (GDI), has been anaemic. In theory the two ought to be aligned. GDP tracks all final expenditures in the economy, summing up consumption, investment, government spending and net exports over a specific period.
The characterization of GDP in the last sentence is incorrect or at least misleading. GDP is defined as production (gross domestic product) within a given territory (gross domestic product) of goods used for final consumption or for exports. (Whatever else it also “tracks” is another matter.) By definition, it does not depend on imports and thus on “net exports” (exports minus imports). Whatever the amount of imports, the calculation of GDP is not affected by one cent. I have explained this simple accounting fact in a number of posts and articles, referring to official and independent sources. The best way to approach the question may be to read my last post on the subject and follow the links to other posts and sources. The journalist or editor of the article quoted above is thus mistaken on the very definition of GDP, if it is not just careless writing.
In the same weekly issue of The Economist, another article quickly refers to the correct definition of GDP (“How Much Trouble Is China’s Economy in?” The Economist, dated July 17, 2023):
That is an unconvincing explanation for the weakness of China’s nominal growth, because GDP should count only the value added to a good in China itself, thus excluding the value of imported commodities.
Here, “excluding” correctly implies that imports cannot be added to nor deducted from GDP, because they have no impact at all on the way it is constructed—just like, say, what happens on Sirius has no impact on the way your profits are calculated. This quote contradicts the previous one.
Is The Economist hopelessly confused? Perhaps. Confusion may lead to contradictions. But perhaps I am being too critical. Instead of the venerable magazine being hopelessly confused, the apparent contradiction between the two quotes could mean that at least one writer and one editor—those involved in the second article—know what is the technical definition of GDP and what it implies, what it is designed to represent and not designed to represent.
Putting the last hand on this post, I noticed that the Wall Street Journal is alas well representative of the general confusion about the relationship between imports and the calculation of GDP. Its story of Thursday on the second-quarter growth of GDP is, on this topic, fuzzier but perhaps even more glaringly misleading by using “net trade” instead of “net exports” (“U.S. Economic Growth Accelerates, Defying Slowdown Expectations,” July 27, 2023):
Net trade slightly subtracted from second-quarter growth, reflecting a sluggish global economy.
READER COMMENTS
Jon Murphy
Jul 29 2023 at 9:50am
The confusion goes to the very source, too. This sentence, which is incorrect (or, at the very least, misleading), is in every BEA GDP release:
Pierre Lemieux
Jul 29 2023 at 10:49am
Jon: What you quote is very interesting–and very strange. When BEA bureaucrats are pushed to the wall, they of course agree with me, with any economist who has looked into what GDP is, and with their own methodological handbooks. But they resort to casuistry to reconcile this truth with their misleading statement. Here is a quote from an email I received on June 19, 2015 from a BEA bureaucrat:
A few hours later, the same bureaucrat wrote, no doubt prodded by a higher-up economist:
But she continued to argue that their casuistry, the same as today’s, was correct. See also my EconLog post of October 30, 2017, to which you contributed with one other reader. Most of mankind and Americankind did not seem to care!
Jon Murphy
Jul 29 2023 at 11:18am
Yeah, that bit of wordplay is why I am very very careful when I go over GDP in my Principles classes.
Craig
Jul 29 2023 at 10:06pm
Is this perhaps related in any way to Ireland’s curiously high per capita GDP? I have done some reading that it might be because the national accounting for Ireland ultimately is capturing profits of Ireland based multinationals for profits more or less earned elsewhere in Europe.
Jon Murphy
Jul 29 2023 at 10:20pm
No, that wouldn’t be it. GDP doesn’t capture profits, just production.
Pierre Lemieux
Jul 30 2023 at 11:34am
Craig: That’s a very good question and I don’t agree with Jon on that (it doesn’t happen often!). Profit, as any other value added, comes from production (i.e. it is the corresponding income of production). It is the remuneration of capital for its contribution to production as opposed to the remuneration of labor or land for their respective contributions. So if a company moves its headquarters to Ireland, just like if the single owner of a company moved, the return of capital is earned in Ireland. If Musk moved to Papua New Guinea, the annual return of his capital (say, $10 billion) would be produced in Papua New Ginea. In 2005, several multinational corporations relocated to Ireland to avoid higher taxes elsewhere, giving the country an enormous one‐year 14% GDP boost (26% minus 8%). See my Regulation article. This is one of the distortions created by producers changing their behavior to avoid taxes.
Craig
Jul 30 2023 at 8:53pm
“Profit, as any other value added, comes from production”
However, in this very narrow circumstance the profit might simply be misattributed to Ireland; the result of an accounting trick? I’m not really sure, but from my superficial understanding I am led to believe that the normal GDP accounting might be skewed with respect to Ireland resulting in a very high per capita GDP that might not really reflect reality.
Just to take a quick step back in the day when I was learning about the net trade qualification to the Keynesian CIG(e-i) calculation my then-professor explained that my consumption of imports ‘counted’ and so it had to be backed out to prevent double counting. (I think, that’s an OLD memory). Is that correct?
Now fast forwarding when I saw your post it got me thinking about Ireland and I am thinking that they might be counting something that they likely shouldn’t be and even if its counted elsewhere it might be such that in Ireland’s case this variable, which might not be significant in other countries, really puts a thumb on the scale.
But I digress, in any event I am just curious what both Professors think of Ireland’s GDP and per capita GDP figures. Not that its a conspiracy, but is there something about the economic accounting that, in the case of Ireland, makes the calculation of per capita GDP such that Ireland shows up as much more prosperous than it actually is? (Irrespective of the reason)
I have been to Ireland and I recall it being relatively much poorer than the UK, generally and of course the US and not that it was ridiculously poor, but 114k per capita makes it one of the wealthiest countries in the world today and if that’s the case, well, so be it, good for them, I don’t begrudge them for it, but I do think there might be something askew.
Pierre Lemieux
Jul 31 2023 at 3:45pm
Craig: Your old memory is correct, but I would suspect your professor did not use the misleading expression of “net trade,” because he was precisely explaining to you that only exports are calculated on the expenditure side of GDP.
Why is there an expenditure side of GDP? Because, by definition, GDP is meant to measure production. However, value added is difficult to measure; in fact, it takes a couple of years to get these figures from input-output tables. Because of that, GDP is quarterly or annually measured from its expenditure and income sides, because of course production=expenditures=incomes.
Remember that GDP is, by definition, domestic production. (The previous GNP measured instead production by nationals.) And remember another crucial methodological point: GDP is nothing else than the sum of what is produced domestically, that is, by the residents of the territory in question. (A territory with no residents has or would have no GDP.) For example, I knew a rich guy who moved from the US to Ireland, where he established his residence. From that point on, the return on his capital, now “resident” in Ireland with him (his body), enters the Irish GDP, not the American GDP. Similarly, the production of a foreigner who establishes his residence in the United States is part of American GDP.
It is not a philosophical question: it just depends on what you want to measure, and that is what the conceptors of GDP wanted to measure. (In fact, they first wanted to measure GNP, but then the focus moved, in the early 60s, I think, to GDP.)
Jon Murphy
Jul 31 2023 at 9:08am
Thanks for the clarification to me (and Craig), Pierre. I was thinking the scenerio was the HQ moves to Ireland, but the factory remains in the US (or wherever). In that case, the production would be counted in the GDP of the US and not Ireland, though the profits would show up in GNI, correct?
Pierre Lemieux
Jul 31 2023 at 3:23pm
Jon: The value added by the plants in the US (value of output minus labor and other input costs and minus the cost of capital, which is by hypothesis located in Ireland) is part of US GDP. The key is to always remember this: a country’s GDP is, by definition, the value of production on the territory of this country. (Although the cost of the labor is sometimes presented in the value added by the company, it is not theoretically correct: the cost of labor corresponds to the value produced by the (domestic) workers, which can be measured by their remuneration. I think this is how it is in input-output tables.)
Fazal Majid
Jul 30 2023 at 6:22am
Consider the flagship made-in-China product, the iPhone. Out of the $400-500 bill of materials, most of the value comes from Korea, Japan, the US and Europe (without even considering the value of Apple’s intellectual property), only final assembly, the case and battery are Chinese, accounting for less than $30 or not even 10%. Yet the entire value will be credited to the Chinese GDP as you do not deduct the value of inputs.
Jon Murphy
Jul 30 2023 at 6:32am
Yes, although the production values of the various intermediate steps still show up in their countries’ GDPs, albeit indirectly, as the income generated from selling the intermediate goods are spent.
Pierre Lemieux
Jul 30 2023 at 11:42am
Fazil: I think that the problem you are referring to is that, when the iPhone is imported from China into the US, its import price or value at that stage, which actually includes inputs from a large number of other countries, will be assigned as an import from China, which is not part of American GDP. Apple’s intellectual property income will be assigned to production in the US. And as Jon mentions, if a component of the iPhone assembled in China has been imported from Ruritania, its value will be assigned to the Ruritanian GDP, not the Chinese GDP.
Craig
Jul 30 2023 at 10:23pm
“Apple’s intellectual property income will be assigned to production in the US.”
Is it? I mean, isn’t this where things begin to get murky? What companies like Apple do is they transfer the IP to places like Bermuda and then the Bermuda subsidiary sells the license, no?
https://www.theguardian.com/technology/2019/jan/03/google-tax-haven-bermuda-netherlands
The above article reflects Google’s use of Bermuda
“Firm used Dutch shell company in move known as ‘double Irish, Dutch sandwich’ that cuts its foreign tax bill”
Does this materially impact the GDP calculations?
Pierre Lemieux
Jul 31 2023 at 4:17pm
Craig: You ask, referring to the Google-Shell-Netherlands case (which looks complicated, but I assume it is, in its effects, similar to the transfer-pricing issue):
I think that in principle, it shouldn’t because the residence of Google has not changed. Similarly, transfer pricing for tax purposes does not change where the real profits are made, and the shareholders or their agents must know where that is when they allocate capital. But the extraordinarily voluminous and complex tax rules must complicate the job of the national statisticians (as the job of the tax hunters). So in practice, it may very well create noise in the data the statisticians collect.
If Google moved its residence to Bermuda, of course, it would bring some change in the distribution of the contribution of Google’s capital between the US and Bermuda.
One reason, if not the only reason, tax rules are so complicated is that greedy governments want to make sure that corporations and individuals can’t easily move their capital out of their reach. This is why the US government taxes American residents wherever they move in the world (if their local taxes are not larger than the US taxes). I am surprised that so few national governments do that. Government greed generates high taxes, which necessitate complicated tax rules, which justify intrusive enforcement. Who said that government interventions beget government interventions?
Craig
Jul 31 2023 at 4:27pm
“Who said that government interventions beget government interventions?”
Me and when I bring a cause of action for copyright infringement when you pay the judgment I suppose that will count towards the gross state product of FL, or perhaps TN? 😉
Craig
Jul 31 2023 at 4:55pm
“Similarly, transfer pricing for tax purposes does not change where the real profits are made, and the shareholders or their agents must know where that is when they allocate capital. But the extraordinarily voluminous and complex tax rules must complicate the job of the national statisticians (as the job of the tax hunters). So in practice, it may very well create noise in the data the statisticians collect.”
Let’s do a quick thought experiment here for one second. Let’s assume that the iphone needs 10 widgets which I make for Apple every day in NJ with a widget machine. Assume that I am just a vendor of Google and not owned by Google. Now, the widgets produced in NJ will definitely be produced in NJ and count towards US GDP and NJ gross state product. Now let’s say that I sell that machine to my Bermuda colleague, again still not Google, and I pack up the machine and send it to Bermuda where it is set up still producing widgets for Apple. Ok, well, I don’t think anybody would have difficulty saying that these widgets are now being produced by Bermuda and belong to Bermuda’s GDP calculation.
Go one step further and we move from physical widgets to virtual licenses and now we will presume that I produced a software in NJ needed in the iphone and I sell 10 licenses per day. Still, ok, we will count that for NJ. Likewise I sell that patent/copyright/IP to the Bermuda company and now they are selling the 10 licenses per day. Still, I don’t think we have too much trouble attributing the sale of those licenses from the Bermuda entity as being part of Bermuda GDP because we can still envision two separate companies with some kind of corporate structure selling the licenses and producing those license sales with a sales staff, accounts payable/receivable, etc.
You also noted: “I think that in principle, it shouldn’t because the residence of Google has not changed. ”
Why is that a distinction that makes a difference. We could take my above samples and change them from being vendors of Google to being wholly owned subsiaries of Google and I’d suggest the GDP calculation does not change, no?
But let’s take that additional step and all of those entities are now part of related entites, the Google/Alphabet multinational, and the thing it is, the reason why I now have difficulty attributing this GDP to Bermuda is that Google doesn’t actually employ or do anything at all in Bermuda. NOTHING. Well, not nothing, it receives profits to the tune of billions at PO Box 666 (not a joke, the mark of the devil no less, right?)
With respect to transfer pricing which is a related question, let’s return briefly to the widgets. Let’s now assume the widgets are 1oz of gold which has an objective daily market price established by many buyers and sellers. So if the Bermuda entity sold the 1oz gold widgets for $6k, we would intuitively know that Google was engaged in transfer pricing, but if this is widgets, maybe that’s not so clear or maybe Google doesn’t do it so egregiously, but they still might do it and to the extent they establish the price of the licenses as $X, that’s going to count as the value add for accounting purposes even if its genuinely just straight up tax evasion.
Pierre Lemieux
Jul 31 2023 at 5:28pm
Craig: You write, very humorously, about your frivolous copyright claim, but it does raise the interesting question of whether my payment to you would theoretically be included into GDP. My opinion is that it would not be counted in GDP at all. Remember: GDP measures production. If I have used the copyright you own to produce something myself (a book, a series of articles and posts, and such), my profit will have been included in GDP as a value added (and, on the income side, as part of my [net] income). The $1,000,000 (say) I am condemned to pay you does not correspond to any further production, it is just a transfer. It is exactly as if you had stolen $1,000,000 from me or I had given you $1,000,000. So it is not part of GDP. From now on, of course, it is your production with this copyright capital that will be included in GDP.
This is confirmed by the scant references to copyrights in the BEA’a NIPA Handbook.
Craig
Jul 31 2023 at 6:06pm
“If I have used the copyright you own to produce something myself (a book, a series of articles and posts, and such), my profit will have been included in GDP as a value added (and, on the income side, as part of my [net] income). The $1,000,000 (say) I am condemned to pay you does not correspond to any further production, it is just a transfer.”
But of course if the copyright were a material element in your production and you had done it correctly, you would’ve paid me as you would any other vendor. I agree GDP for the US would not change, but note:
“I suppose that will count towards the gross state product of FL, or perhaps TN?”
There remains the issue of sub-attribution to subnational units!
Bill
Jul 30 2023 at 10:50am
The shoe weight analogy you employ in the Regulation article is brilliant and should eliminate any confusion on this matter.
Pierre Lemieux
Jul 30 2023 at 11:09am
Bill: I agree with you, it’s totally brilliant. I must however confess alas that the idea of this analogy was from Tom Firey, the editor of Regulation. I told Tom at the time that it was brilliant.
David Henderson
Jul 30 2023 at 2:20pm
I read through the whole Regulation article, which is excellent. I hadn’t read it thoroughly since have read it when it came out.
I couldn’t find the “shoe weight analogy” though.
Bill
Jul 30 2023 at 3:59pm
The analogy is referenced here: https://www.econlib.org/imports-as-a-drag-on-the-economy/
Pierre Lemieux
Jul 31 2023 at 4:30pm
David: Thanks, David. Bill (thanks, Bill!) gave you the link to a previous post where indeed I reference my Regulation article on Peter Navarro.
Mark Brady
Jul 30 2023 at 7:19pm
“GDP tracks all final expenditures in the economy, summing up consumption, investment, government spending and net exports over a specific period.”
It tracks final expenditures, viz., consumption, investment, and government spending, each of which has an imported component, and net exports (exports minus imports – the imported component of C, I, G and X).
What’s wrong with that?
Pierre Lemieux
Jul 31 2023 at 4:39pm
Mark: The problem is that “net exports” does not mean “net trade” or the “trade deficit,” which is a constant source of confusion for journalists. Even some economists are confused by that expression. To quote my Regulation article on Peter Navarro (which is the source of the analogy Bill referred to):
If Navarro, with all his genius, is misled by that, imagine who wouldn’t be?
Jim Glass
Jul 31 2023 at 12:34am
Is it? My decades old macro textbook (Robert Gordon) and countless current web pages (see Google) give the formula for GDP as “GDP = C + G + I + NX” [or alternatively for “NX”, “+X -IM”] . Which is just what that last sentence says. They are all wrong? That gives the wrong result?
That’s troubling, as the BEA itself gives: “How is GDP calculated? There is a four-part formula: C + I + G + NX = GDP.” I sure hope it knows what it’s doing!
Looks to me like maybe yet another case of “In theory, theory works, but in practice…” Let’s check this out. As your friendly, if prodded, bureaucrat wrote to you…
So when computing GDP the BEA has these numbers to work with: C, I, G, X, but alas they are “polluted” with imports throughout, which is a practical problem. Happily, it also has the number for imports, IM, which it can use to make a practical corrective subtraction. Thus: “C + I + G +X -IM = GDP”.
If this offends you because “by definition GDP does not depend on imports and thus on ‘net exports’… Whatever the amount of imports, the calculation of GDP is not affected by one cent.” OK fine, in theory.
Now, given the import-polluted nature of C, I, G, and X, in practice what calculation do you use to determine GDP – without subtracting those irrelevant imports?
Jon Murphy
Jul 31 2023 at 6:55am
Jim,
I think you’ve missed Pierre’s point. See his and my exchange above.
Pierre Lemieux
Jul 31 2023 at 5:02pm
Jim: Just in case my exchange with Jon above (and my correspondence with a BEA bureaucrat) did not persuade you, let me just add this. The BEA does not include imports in GDP, and uses the unfortunate concept of “next export” precisely to exclude them. “Unfortunate” because non-economists (and even some economists who haven’t looked into this) easily conclude that the trade deficit subtracts from GDP. The ultimate resource is the BEA’s NIPA Handbook, especially Chapter 2 and, reproduced just below, p. 3 of Chapter 8:
Jim Glass
Jul 31 2023 at 11:08pm
Jim ….The BEA does not include imports in GDP…
Obviously.
and uses the unfortunate concept of “next export” precisely to exclude them.
Of course. Though it is hard for me to see what is “unfortunate” about using a formula that gives the correct — indeed, definitional — answer.
“Unfortunate” because non-economists (and even some economists who haven’t looked into this) easily conclude that the trade deficit subtracts from GDP.
So some people believe GDP is computed as “GDP = GDP – IM”. I get it. But that is not what the formula “C + G + I + X -IM = GDP”, says. The ignorant who believe all kinds of foolishness will always be with us, in all fields, not just economics. Visit a medical forum sometime, or even a physics one (see how such folk react to time dilation and quantum dead-living cats!) But we don’t blame the AMA or APS for that, apart from maybe they should increase their popular education efforts. (Or, oh gosh, consider politics! Or visit the flat earthers!!)
But anyhow, again, if the BEA formula is “unfortunate” it can only be because there is a superior alternative. So I’ll ask again: “in practice what calculation do you use to determine GDP – without subtracting those irrelevant imports?” Which you can suggest the BEA adopt.
Jon Murphy
Aug 1 2023 at 6:33am
Jim-
The problem is not with the formula (as the quotes you highlight show). The problem is with the discussion of the formula. When the BEA says “imports, which are a subtraction to GDP…” many readers (and economists!) take that to mean an increase in imports reduces GDP.
Pierre Lemieux
Aug 1 2023 at 11:24am
Jon: Exactly. And for years (perhaps decades) the BEA has seen its misleading formulation wrongly interpreted and has done nothing about it.
Comments are closed.