## Wednesday, August 06, 2008

### GDP & The Terms Of Trade Issue

After I posted a comment under a post by James Hamilton, it seems that the issue over terms of trade and what should be considered real GDP has finally started. James Hamilton has commented the issue here, , anonymous blogger "knzn" commented the issue here and Brian Wesbury commented the issue here. Hamilton's fellow econbrowser blogger Menzie Chinn commented the issue here, but he really doesn't argue for or against it and instead simply explained the difference between the production deflator and the purchases deflator.

I start with James Hamilton. He use a "Crusoe"-island approach, which I kind of like.

"In 2007, Islandia produced 500 coconuts, which residents sold to themselves for \$1 each, and imported 1 barrel of oil, which cost \$100. The oil was paid for by borrowing from foreigners.

Ready to calculate GDP? We want the dollar value of domestically produced final goods and services, to wit, the dollar value of the 500 coconuts. Got the answer? Very good. Nominal GDP in Islandia for 2007 was \$500. If you wanted to describe that in real terms, you'd call it 500 coconuts. You don't count the oil in either nominal or real GDP because Islandia didn't produce any oil.

Here are the numbers for 2008. We grew 510 coconuts, sold them for \$1.01 each, and still imported 1 barrel of oil, paying \$125 for it. So nominal GDP was \$515.10 (a 3% increase) and real GDP was 510 coconuts (a 2% increase). The change in the implicit GDP deflator would be the change in the ratio of nominal GDP to real GDP, namely, +1%.

But wait a minute, Islandia's pundits decry. How can your crummy accounting claim that inflation was only 1%? Last year we bought 500 coconuts and 1 barrel of oil for \$600, but this year if we tried to buy the same thing it would cost us \$630. The inflation rate, they tell you, is obviously 5%, not 1%. You must have intentionally cooked the books, they charge, just to make the economy appear better than it is!"

However, as much as I like the approach, his conclusion is wrong. As he wrote, what we want is the value of domestic production. But what value? Value is not intrinsic in the things we produce, value is value in terms of the things we want. The value of a worker's earnings does not lie in how many things he produce, but in how much he can buy for it. Saying that terms of trade is irrelevant for the value of domestic production is like saying that the relation between wage and consumer price levels is irrelevant for real weekly earnings of workers.

The logic behind the GDP deflator approach would thus imply that if a company executive tells workers that they're going to have to accept a 90% wage cut even as their cost of living is unchanged, they shouldn't mind because the workers will continue to work as hard and produce as many goods as before for the company and since their personal GDP deflator falls 90%, the value of their wages are unchanged (even though their nominal wage is down 90% and their cost of living unchanged)! Similarly, a company executive could tell shareholders not to worry about the 50% decline in revenues since that decline was entirely the result of reduced prices on the goods they sell, and since the company's GDP deflator is down 50%, revenues aren't really down at all!

Or similarly, supermarkets could raise prices by 50% and tell customers that they shouldn't mind paying more because their wages (their personal GDP deflator) haven't been raised! And suppliers to a company could similarly tell the company that they shouldn't object to paying more to them as long as they abstain from raising the price they charge their customers! Or in the case of Islandia, those foreign oil producers could tell them that even though the world price is \$125, they shouldn't mind paying them \$200 since that won't affect the GDP deflator and thus not reduce real GDP!

As for knzn, he takes a different approach and says that he does not care about the value of production. Well, he is free to care or not to care about whatever he wants, but
the vast majority of people do care about the value of their income and that is why people laugh at Phil Gramm's "mental recession"-remarks (See here for a clip of how people spontaneously laugh when Obama tells them about those remarks) that contradict their daily experience even as the official GDP statistics support Gramm. Economists who want people to think that GDP isn't something laughable better care.

He continues by saying "Personally, I care a lot more about the people that don’t have jobs, or potentially won’t have jobs in the future." But first of all, why then look at GDP numbers in the first place? Why not only look at employment numbers? And secondly, the employment numbers are far more consistent with the terms of trade adjusted GDP number as employment has been falling (and would have fallen a lot more if not for the flawed "Birth/Death" model).

He then says that he cares more about productivity (which he defines in unadjusted terms) because that is supposedly more stable then terms of trade movements. But even setting aside the fact that GDP is supposed to measure the value of production regardless of how stable or not certain aspects are, it is not true that productivity is particularly stable. If you go to the Bureau of Labor statistics web site, you see that their measure of productivity (non-farm business output per hour) moves up and down in a quite erratic way. In 2003 for example, productivity at an annualized rate rose by 5.6% in the second quarter and 10.3% in the third quarter only to fall by 0.4% in the fourth quarter.

As for Brian Wesbury, who delivered the by far worst GDP forecast of all economists(3%, even as he assumed a 3% increase in the GDP deflator!), he agrees that 1.1% is a far too low inflation figure and notes like I did that if you deflate the nominal GDP increase with the domestic purchases deflator, real GDP declined. However, he says that would be misleading and says that if you include import prices to calculate real GDP you must add imports to GDP, a sum he says increased more than 5%. But that does not follow. What the domestic purchases price index include is first of all usually not directly import prices, but the prices of various forms of domestic demand, like private consumption, government consumption and business investments, and that is usually not the same even for imported goods, as the retail price of for example a sweater made in China includes the margins of the American retailer (and often also wholesaler). That means that including imports would double-count it. However, as discussed above, the value of domestic production is determined (partly) by how much imported goods that domestic production can buy.

Jim Driscoll said...

You may have done better in explaining it to a rube like me by sticking to coconuts and oil. Sorry :-)

EcnomPic Data has an example using coconuts, oil, and fertilizer which I found (mostly) understandable in explaining the relationship of inputs and goods produced:

Econompic

It points out the basic problem in the coconut argument, which is that very little is produced without input from an outside source.

I'd love to hear your comments on that... Is that essentially what you're getting at?

7:27 PM
stefankarlsson said...

I did mention coconut and oil, but I think most people better understand it by using real-life examples involving workers and firms.

As for the post in the link, it does not appear to make the same argument as I do, as it only focus on nondurable goods in the CPI and GDP deflator. Although this discrepancy can in part probably be traced to the terms of trade factor I focus on, it also involves the tendency for the PCE deflator to increase slower than the CPI and the fact that the GDP deflator also contains investment goods (which the CPI like the PCE deflator does not include).

10:42 PM
newson said...

"money, bank credit, and economic cycles" by jesus huerta de soto.

he explains at length why gdp is such an unsatisfactory measure, and why the national accounts are so misleading (page 305 onwards). i found his explanation more comprehensible that reisman et al.

7:32 AM
stefankarlsson said...

Newson: no I haven't read it. Does it differ from our previous discussion of this issue?

7:37 AM
jake said...

This is the coconut / fertilizer argument that Jim mentioned earlier. I'd love to hear your thoughts..

A better example would have been coconuts and a fertilizer that is essential to grow coconuts. In this case had fertilizer doubled in price, I have feeling that \$515.10 worth of coconuts would not have gotten you the 510 initially assumed by the "Island's BEA" in the first round of its Real GDP calculation, but rather only 470 as the higher price of fertilizer increased the price of the final good, hurting demand. Thus, rather than an increase in GDP from 500 to 510 coconuts, there was a recession to 470 (same nominal GDP in both cases).

1:46 PM
newson said...

well i thoroughly recommend the whole book; i feel it's got the most convincing case for 100% reserve banking i've yet seen presented. i really found the legal chapters great.

the pdf is on mises, and i think the gdp issue is much more clearly argued that in reisman's work. i know we've had this dialogue before, but i can't help but viewing the measure of gdp with great suspicion, and see recognition of its worth as a big victory for keynesianism.

if you're interested, the relevant part on the mises pdf is page cccxxxv (335/906).

5:37 PM
stefankarlsson said...

Jake: if the fertilizer is imported, then this price increase would not have affected real GDP according to the standard methodology.

10:18 PM
Cahya said...

Nice to see your good blog.

7:38 PM
knzn said...

"the vast majority of people do care about the value of their income"

"...why then look at GDP numbers in the first place? Why not only look at employment numbers?"

Because GDP numbers can be useful in forecasting labor demand

"...the employment numbers are far more consistent with the terms of trade adjusted GDP number as employment has been falling"

I don't see what it means to say "consistent with" in this context. You could take two statistics that have nothing to do with each other and are growing at the same rate and say they are "consistent." What we should be interested in is whether there's a causal link, and I don't see what the link is between employment and terms-of-trade adjusted GDP

"...it is not true that productivity is particularly stable"

Actually it is true, if you're comparing productivity to the terms of trade. For one thing, productivity always rises over any significant period of time, while terms of trade go all over the place. The question is how much information current growth rates give you about future growth rates. In the case of terms of trade, I would say, zero at best (more likely there is information, but it is in the other direction, "contrary information," if you will). In the case of productivity it is greater than zero.

4:00 AM
stefankarlsson said...

Well, the aggregate satisfaction with their economic situation should be related to aggregate real income, which means that they care on a aggregate level.

"I don't see what it means to say "consistent with" in this context. You could take two statistics that have nothing to do with each other and are growing at the same rate and say they are "consistent."

Consistent with means that they correlate better. Go to the BLS web site and compare how employment changed during Q1 2007 when unadjusted GDP rose just 0.1%, but terms of trade improved, with employment during Q2 2008 when unadjusted GDP rose 1.9%, but terms of trade significantly deteriorated.

"What we should be interested in is whether there's a causal link, and I don't see what the link is between employment and terms-of-trade adjusted GDP"

You don't? Well, let me help you out on that one. If higher import prices reduce domestic purchasing power, this means that Americans will have less money to buy domestic goods, meaning fewer jobs.

"Actually it is true, if you're comparing productivity to the terms of trade. For one thing, productivity always rises over any significant period of time, while terms of trade go all over the place. The question is how much information current growth rates give you about future growth rates. In the case of terms of trade, I would say, zero at best (more likely there is information, but it is in the other direction, "contrary information," if you will). In the case of productivity it is greater than zero."

Just what basis do you have for these assertions? You can't just make assertions about empirical data without providing evidence.
I provided a concrete example of how extremely volatile and erratic productivity is, and if you go to the BLS web page you can see that the numbers are similar for other years.

10:29 AM
ScottB said...

To me, you are confusing two different concepts. Real GDP measures the quantity of output in a time period. Purchasing power, as measured by real incomes, is different, because it depends in part on the price of imports. You seem to be saying that if purchasing power declines due to terms of trade, then real GDP should also decline. But it seems to me to be perfectly reasonable to have increasing real production with a decrease in purchasing power (due to stagnant wages and a jump in import prices) at the same time.

I am surprised as anybody by the low deflator and high real GDP. But in looking at the details, it makes sense to me.

3:25 AM
stefankarlsson said...

Scott, GDP is supposed to measure the value of output. And the value of output is the value it has to the producers (the inhabitants of a country). GDP separated from value is a completely meaningless and irrelevant concept.

8:45 AM