Russ Roberts

Coyle Postmortem

EconTalk Extra
by Russ Roberts
Continuing Conversation... Dia... Gavin Andresen on the Present ...

Here's my postmortem on the Coyle episode. Includes my early nominee for worst economics story of the year.

My biggest takeaway from the conversation was realizing the challenge of measuring services, an increasingly important part of the economy over the last century. Not sure what the full implications are--I wish we'd talked about it more.

I meant to ask Diane about the financial sector, particularly challenging part of the service sector to measure. She talks explicitly about the financial sector at some length. Think about the value of what Goldman Sachs does. Or Bank of America. What's the product? I'm not sure how you think about that. The financial sector's "contribution" to GDP has grown dramatically over the last 30 years. What does that really mean? Again, I don't know where to start for thinking about that.

The other topic I enjoyed and learned from was the production boundary. We don't count washing the dishes and we also don't count most of Wikipedia. That part of the conversation really highlighted for me the two different uses of GDP--trying to measure how well the commercial economy is functioning and a measure of our well-being. As more stuff shifts out of commercial activity and into volunteerism/free stuff, the bigger the gap between those two measures. What I should have emphasized is that the issue matters most when the ratio of paid stuff to unpaid stuff is changing. What we're usually looking at is the growth in GDP not the absolute level. If that ratio is changing, the growth rate isn't measuring just the change in the size of the commercial economy.

The latest GDP report came out this week showing very very weak growth--a tenth of a percent, teetering exceedingly close to zero or negative growth. This prompted news reports, see here and here (HT: King Banaian) that health care spending--which surged dramatically--was keeping the economy afloat. From The Guardian:

GDP, the measure of the economy's growth, was disappointing last month, and many economists blamed it on the weather.

But there was one bright spot: people are paying for healthcare, and a lot of it. Republicans may hate hearing this, but it has to be said: you can thank Obamacare.

"Obamacare is boosting consumption," says Ian Shepherdson, chief economist at Pantheon Macroeconomics.

The sharp increase in health care spending can be attributed to the opening of the insurance exchanges, he said, adding that there has been a 9.9% leap in spending on healthcare services. In fact, if it weren't for Obamacare, the US economy would be shrinking right now.

"If healthcare spending had been unchanged, the headline GDP growth number would have been -1.0%."

Uh, no. It doesn't work that way. That last statement is only true if every other component in GDP would have been unchanged if health care hadn't grown by 9.9%. And that's simply not true. Maybe if health care spending had been unchanged, something else would have grown by 9.9%. As somebody responded to me on twitter about this--it implies that an increase in cancer increases GDP. It doesn't. It might increase spending on health care. But unless there's more income, increased spending on health care just means less on something else.

Here's the fallacy. Suppose I want to know your income for the year. I ask you and you tell me you made $50,000 in salary. Another way I can get to that number is to add up everything you spent money on--food, rent, clothing, entertainment, savings and so on. As long as I count everything, I get to the same number, $50,000.

Suppose I find out you spent $5000 on entertainment. It would be very wrong to say that without that spending, your income would only have been $45,000.

The mistake of assuming health care spending adds to GDP is partly confusing accounting with economics. Or arithmetic with economics. It is arithmetically true that your income is equal after the fact to your spending on A+B+C and so on, the different things you can spend money on. But that doesn't mean that changes in A, B, or C change your income.

A Keynesian might respond that in times of slack, increased spending can stimulate the total through the multiplier. I don't really know what that means in a world where everything is connected. The whole idea of government spending increasing GDP in a time of recession relies on the idea that increasing government spending leaves all other spending unchanged. I'm a skeptic about that. But even if you're a Keynesian, it's hard to argue that increasing one component of consumption spending (health care) means the total amount of spending has to go up by the size of the increase.

Comments and Sharing

TWITTER: Follow Russ Roberts @EconTalker

COMMENTS (14 to date)
Ralph Soule writes:

I like this kind of post after you interview someone. I have not listened to the interview yet, but will soon. I really like the podcast. I learn a lot from most of the speaker (a few not so much, especially when covering arcane details of economics).

Becky Hargrove writes:

I enjoyed this discussion, in part because as you indicated, the challenges posed in services terms. Not sure what you mean by the "worst economics story of the year"!

I mostly agree with some production boundaries as they were set up. One's time in work at home consists of a high degree of choice, made possible by labor saving devices. In the past when home responsibilities were not labor of choice, compensation of these duties probably would have made more sense.

Health care spending adds to GDP in the sense that it captures the money being spent, and if the purpose of the Fed is to track monetary flows, the definition works.

However I agree that healthcare spending is accounting in terms of an artificial marketplace. Both healthcare and education need to be on the positive side of the ledger, so that they can contribute to GDP more "honestly".

Jim Clawson writes:

Thanks for adding this as a part of the continuing discussion on these topics. With respect to the candidate for "worst economics story of the year", is it possible that healthcare spending is simply a different animal in terms of how healthcare consumption contributes to perceived growth in GDP?

I couldn't agree more with your fundamental take on consumer consumption of goods and services in terms of their own budget. Namely that increasing spending in one category means a offsetting decrease in spending on other categories; however, when it comes to healthcare, the dollars used for consumption don't belong to the consumer. Those dollars spent either belong to the insurance company or the government.

If the increased healthcare consumption is afforded through private/corporate insurance reimbursements then your overall description of GDP mechanics holds true. The insurance company spends more on fulfilling their obligations to healthcare providers and they, the insurers, engage in less of other activities such as investing. In the case of the government providing reimbursement for increased consumption - especially through increasing the Medicaid roles - isn't it possible that the government is simply using their "magical" powers to conjure money from the ether to subsidize the increased consumption?

In the latter case, where the government is the one paying for the increased consumption of healthcare goods and services, has GDP not legitimately expanded - on paper - as a result of government piping "new money" into the system?

Obviously there are other potentially harmful consequence that come out of this type of spending such as increased deficits, rising inflation etc., but perhaps that spending does "buy" the government GDP numbers that look better in headlines... even if it is all just smoke and mirrors.

Allen Hutson writes:

My first reading in economics was Bastiat, "The Seen and the Unseen," and it is truly incredible that people can continue to make the same mistake over and over again. The Guardian article is just another example of a broken window fallacy, and your refutation mirror's Bastiat's.

Garth writes:
"The financial sector's "contribution" to GDP has grown dramatically over the last 30 years. What does that really mean? Again, I don't know where to start for thinking about that."

I've seen these numbers/graphs before and didn't understand what was driving the numbers. The sector's contributed share of UK GDP is even larger. It couldn't be that lending has become so much more efficient, or that earnings from investment has really contributed so much.

Hopefully you'll further discuss your comment in a future episode.

Amy Willis writes:

@Allen, thanks for reminding us of this terrific Bastiat reading...You can read it online at Econlib!

Michael Moore writes:

"My biggest takeaway from the conversation was realizing the challenge of measuring services, an increasingly important part of the economy over the last century. Not sure what the full implications are..."


Perhaps I'm naive - half of my university macroeconomics course revolved around Smith's "Wealth of Nations" - but my understanding is that trying to measure these services or their value is a fool's or central planner's errand (sorry for the repetition). The market will do a far better job of measuring. Perhaps I missed something.

Virginia A. writes:

How do economists approach "level of uncertainty" indicators concerning economic figures like GDP? It seems important, especially when making comparisons, to know whether a figure should be thought of as (+/- 1 billion) or (+/- 100 trillion).

Also, how do the people who calculate GDP (and the like) know when their calculation is "good enough?" There comes a point where the inherent uncertainty/inaccuracy dwarfs further refinements.

How do economists deal with the temptation to spend time improving precision beyond the point where substantive improvements are possible?

Virginia A. writes:

I appreciated this postmortem, and learned from it.

I've spent too much time thinking about the proverbial(?) straw man character who is a boon to GDP -- the cancer patient who gets in an accident with a chemical tanker on his way to a meeting with his divorce lawyer. I'm glad to know now that such a person does not really increase GDP -- he just causes resources to be spent on cancer treatment, chemical spill cleanup, auto repair, and divorce proceedings instead of other things.

Michael Byrnes writes:

@ Virginia A:

I think in many cases what economists are looking at is the trend. If GDP (or some other indicator) is measured using consistent methodology over time, then what matters is the trend moreso then the absolute level.

John Foster writes:

Michael Byrnes:
I posted a question similar to Virginia's. I question whether GDP accuracy/reproducibility issues can we dismissed with the assumption that the methodology's errors are consistently in the same direction and to the same degree (or better yet, perfectly cancel each other) and thus the trend is all that matters.

Has anyone confirmed that doing the GDP measurement twice by different groups using the same methodology gives the same result? How much room is there for the analysts to throw out "outliers" or "bad" data sets? Is the amount of "adjustment" available to the statisticians greater or less than the amount of inherent error in the measure?

Surely there are innumerable times when a given data set yields figures which the analysts decide "just can't be right" and they go looking for errors to correct. What is their basis for doing so? Data sets which erroneously fall into the "correct" range will be incorporated without question. So what bias is built into the final answer?

Thus I think the question of how much error there is and does it have any consistent bias is relevant.

Add to the above the fact that the agency which publishes the figure has a vested interest in the result.

Add to that the human factor as seen in the recent admissions regarding the (mis-)calculation of the unemployment rate by BLS or Argentina's inflation stats.

With lots of room for error and no way to calibrate the amount of error, isn't there ample room for mischief? And where mischief pays well and is hard to detect, wouldn't you expect a lot of it?

Virginia A. writes:

The Economist's "Big Mac Index" (and the Billion Prices Project) measure something different, of course, and usefulness may be debatable, but I have to remark on the sparkling transparency of the methodology.

It stinks that the indicators that seem to have enough clout to make markets dance are ones that have lots of room for mischief. Or should we say that markets USED to dance to their tune.

Kount von Numbacrunch writes:

I enjoyed the podcast, but I wish you had looked into two questions related to GDP.

1) Let's look at your hypothetical person with a $50K annual income. If his income was unchanged from one year to the next, but he borrowed $10K and spent it, his GDP would show a sharp increase. Governments can do that, and they can even borrow through a bond offering and then buy the bonds with new fiat money.

2) GDP is measured in $ or Euros or what have you, but the unit of measure is not constant. The estimates of the change in the value of the dollar are about as squishy as those of GDP.

I wish you would look into these things in future. The topic might seem too controversial, but maybe we could look at in terms of events in China.

Daniel Barkalow writes:

If you were a contractor, you might decide whether to spend more time on EconTalk or do something that pays depending on your budget. You might actually say, at the end of the year, "I turned down $5000 of contract jobs to do something more fulfilling, because I didn't feel like I needed the money, because I saved $5000 watching YouTube instead of movies." So it's possible, but not definite, that your income would actually go down if your expenses went down, depending on how you make income-related decisions (if you actually make any).

Similarly, the economy has started to have segments where businesses don't make as much stuff as they could make simply because they wouldn't be able to sell it. The original idea of GDP was that production was limited by the fact that production costs (resources, labor) would rise past the (relatively constant) market value of the goods produced at some point, and GDP could therefore measure the total production capacity and how big an effect on supply of consumer goods an additional government military demand would cause. But we're seeing at least some cases where the market value drops past the (relatively constant) production costs instead, such that businesses would be able to make more stuff without a price premium for the extra production if they wanted to.

Of course, while it's definitely wrong to think that GDP is entirely supply-limited, it's also wrong to think that it's entirely demand-limited. Maybe without the increase in medical spending, people would have done something else, or maybe it was too cold outside for anything but medical attention. Maybe healthcare competes for some scarce resource with another supply-limited sector that could have grown as fast or faster. With only the GDP figure, it's impossible to tell.

Comments for this podcast episode have been closed
Return to top