Gross Distorted Product?

[RichardR]

Everyone knows that GDP figures are showing growth in the (US) economy, although this is not yet reflected in any significant jobs growth. The usual explanation for this is that the improvements in productivity mean GDP can grow without adding workers, and/or that the jobs growth is just around the corner. However, an article in this week's Economist (http://economist.com/finance/displayStory.cfm?story_id=2577317) citing Jan Hatzius, an economist at Goldman Sachs (you'll need a subscription to read it), suggests a different explanation – the GDP may be overstated.

GDP attempts to measure the value of goods and services produced. There are several ways to get to this figure and all are open to some degree of error. The usual method is to start by adding up all domestic spending, government spending and investment by companies. To that figure you then add exports and subtract imports (including imports of intermediate inputs). The reason for this should be obvious: if your $ value of goods consumed includes imports, you have to take out these imports to find out the $ value of goods produced here. (And vice-versa for exports.) Hatzius thinks the imports figure may be understated due to undercounting imports of intermediate inputs of goods and services produced abroad by American firms that have outsourced jobs to cheaper countries.

Hatzius offers two pieces of evidence that GDP is overstated. Firstly, he calculated GDP using a different basis - summing incomes, instead of expenditure. This shows that GDP has grown by only 2.8% per annum since the end of 2001, 0.8 percentage points less than the expenditure measure.

Secondly he notes a divergence of the growth rates of GDP in the goods (consumer spending) sector compared with that of industrial production. The two figures used to track each other closely but in the past two years, a wide gap has opened:

http://economist.com/images/20040410/CFN629.gif

In the year to the fourth quarter, industrial production rose by only 1.4%, while goods-sector GDP surged by 8.0%. The Economist suggests that industrial-production figures are likely to be the more reliable of the two, because they come directly from industry reports.

Why does Hatzius think this difference is due to understating outsourced imports? When American firms outsource call-centre and information-technology-support jobs to India and other Asian countries, the result should be higher imports of services, yet official statistics do not show such an increase. America's recorded imports of software services from India are also much smaller than India's reported exports of such services to America.

IMO, correcting the GDP for this error (is that's what it proves to be), is necessary for the index to tell us useful information. What it appears to tell us is that outsourcing has so far been more serious than we may have thought.

[shanek]

Originally posted by RichardR
Hatzius offers two pieces of evidence that GDP is overstated. Firstly, he calculated GDP using a different basis - summing incomes, instead of expenditure. This shows that GDP has grown by only 2.8% per annum since the end of 2001, 0.8 percentage points less than the expenditure measure.

Um, that's called National Income, and it's a distinct measurement separate from GDP. Using National Income is invalid in this context. Also, there's always a lag between growth in GDP and growth in National Income. So the effect he mentions is hardly unexpected.

Secondly he notes a divergence of the growth rates of GDP in the goods (consumer spending) sector compared with that of industrial production. The two figures used to track each other closely but in the past two years, a wide gap has opened:

http://economist.com/images/20040410/CFN629.gif

In the year to the fourth quarter, industrial production rose by only 1.4%, while goods-sector GDP surged by 8.0%. The Economist suggests that industrial-production figures are likely to be the more reliable of the two, because they come directly from industry reports.


Not all of the consumer spending component of GDP is used to create goods. There are services as well, and services do not count as industrial production. Although we should expect a high degree of correlation between the two (and there is), the widening gap can easily be explained by the economy becoming more service-oriented, with more of our goods being produced overseas.

When American firms outsource call-centre and information-technology-support jobs to India and other Asian countries, the result should be higher imports of services, yet official statistics do not show such an increase.

Not necessarily. It could be reflected in a decrease in exports as well, or an increase in foreign investment, or a combination of these.

What it appears to tell us is that outsourcing has so far been more serious than we may have thought.

No, what it tells us is that you can come to any conclusion you want as long as you misunderstand enough of the statistics.

[RichardR]

Originally posted by shanek
Um, that's called National Income, and it's a distinct measurement separate from GDP. Using National Income is invalid in this context. Also, there's always a lag between growth in GDP and growth in National Income. So the effect he mentions is hardly unexpected. Do you have a source for that definition?

According to this site (http://www.bized.ac.uk/stafsup/options/notes/econ210.htm#Heading123), "National Income" includes the two methods quoted by the Economist for measuring the GDP. The "Income Method" of calculating GDP appears to be what you are calling "National Income":

National Income Accounts

The relationships explained in the sections above form the basis of national income accounting. The aim of national income accounting is to place a money value on this year's output. There are three methods of calculation.

Income Method

The income method adds together the total value of all incomes that have been earned in the relevant time period. These may include income from employment, income from self employment, profits, surpluses of public (government) corporations and rent. Note that only incomes earned from supplying a factor service are counted. Transfer payments are ignored.

Expenditure Method

The government adds up all the money spent in buying this year's output. This will be the total of Consumption, investment, government expenditure and net exports (exports - imports). This ignores:
• Indirect taxes and subsidies included in the selling price.
• Spending on second-hand goods.

(Third one snipped)

Originally posted by shanek
Not all of the consumer spending component of GDP is used to create goods. There are services as well, and services do not count as industrial production. Although we should expect a high degree of correlation between the two (and there is), the widening gap can easily be explained by the economy becoming more service-oriented, with more of our goods being produced overseas. Easily explained? They correlated closely until recently. In the year to the fourth quarter, industrial production rose by only 1.4%, while goods-sector GDP surged by 8.0% - that's quite a difference all of a sudden. There may be different explanations but I don't think you can easily explain it away.

Originally posted by shanek
Not necessarily. It could be reflected in a decrease in exports as well, or an increase in foreign investment, or a combination of these. But you ignored this:

America's recorded imports of software services from India are also much smaller than India's reported exports of such services to America. What else could explain that?

Originally posted by shanek
No, what it tells us is that you can come to any conclusion you want as long as you misunderstand enough of the statistics. Quite.

[shanek]

Originally posted by RichardR
Do you have a source for that definition?

I'm using the definition used by the US Census Bureau (www.census.gov). They're a good source for all of these kinds of statistics.

According to this site (http://www.bized.ac.uk/stafsup/options/notes/econ210.htm#Heading123), "National Income" includes the two methods quoted by the Economist for measuring the GDP. The "Income Method" of calculating GDP appears to be what you are calling "National Income":

It's using pretty much the same definition. Read that site carefully: National Income DOES NOT count Transfer Payments, which are included in the Consumer Spending portion of GDP when they're spend (or in the Investments portion when they're invested, but that's unlikely). It's also limiting income to that "earned from supplying a factor service," meaning that things such as capital gains aren't counted either; again, if these gains are reinvested they're counted in the Investment portion of GDP and if they're spent in the Consumer Spending portion of GDP.

Also, more subtly, GDP is supposed to consider the value of the end product. In other words, GDP counts the price of (say) a newspaper, but not the price of the ink, paper, and other materials that go into making the newspaper; if it did, then those materials would get counted twice since their cost is already figured in to the value of the newspaper. But National Income includes the income of everyone, regardless of whether or not what they're working on is an end product. This is fine, since their labor will all be figured into the final price of the good or service, but it does show a very subtle distinction. If people suddenly consider a good or service more valuable, then its price will go up, but the incomes of all of the people who work to produce that good or service may not go up proportionally. Remember, we're not counting things like corporate profits or capital gains in National Income.

Easily explained? They correlated closely until recently.

And they still do, just to a wider range. But pretty much one still goes up as the other one does.

But you ignored this:

America's recorded imports of software services from India are also much smaller than India's reported exports of such services to America. What else could explain that?

I didn't ignore it; it's just irrelevant to my point.

[RichardR]

Originally posted by shanek
It's using pretty much the same definition. Read that site carefully: National Income DOES NOT count Transfer Payments, which are included in the Consumer Spending portion of GDP when they're spend (or in the Investments portion when they're invested, but that's unlikely). It's also limiting income to that "earned from supplying a factor service," meaning that things such as capital gains aren't counted either; again, if these gains are reinvested they're counted in the Investment portion of GDP and if they're spent in the Consumer Spending portion of GDP. OK, so National Income excludes certain things that are included in GDP (and vice versa). But all the Economist article said was:

"another estimate, found by summing incomes—which in theory should be the same—says that GDP has grown at an annual rate of only 2.8%"

The Economist doesn't say that Hatzius used National Income – you did. How do you know that Hatzius' calculation excluded transfer payments, for example? Do you know he didn't make the calculations and any necessary adjustments correctly?

Originally posted by shanek
Also, more subtly, GDP is supposed to consider the value of the end product. In other words, GDP counts the price of (say) a newspaper, but not the price of the ink, paper, and other materials that go into making the newspaper; if it did, then those materials would get counted twice since their cost is already figured in to the value of the newspaper. But National Income includes the income of everyone, regardless of whether or not what they're working on is an end product. This is fine, since their labor will all be figured into the final price of the good or service, but it does show a very subtle distinction. If people suddenly consider a good or service more valuable, then its price will go up, but the incomes of all of the people who work to produce that good or service may not go up proportionally. Remember, we're not counting things like corporate profits or capital gains in National Income. Good point. It also works in reverse though, if people regard something as being worth less.

Originally posted by shanek
And they still do, just to a wider range. But pretty much one still goes up as the other one does. Isn't "a wider range" a different way of saying "they are not correlating"? If not, what would be your definition of "not correlating"?

Originally posted by shanek
I didn't ignore it; it's just irrelevant to my point. But not to mine. :D So I think it still stands as evidence supporting Hatzius' hypothesis.

Don't you think it's worth considering and investigating in more detail?