Re: offshoring & phantom GDP
On Jun 10, 2007, at 10:30 PM, Tim Francis-Wright wrote:
Doug Henwood wrote:
Business Week - June 18, 2007
Michael Mandel “The real cost of offshoring”
U.S. data show that moving jobs overseas hasn’t hurt the economy. Here’s why those stats are wrong
That makes two recent Business Week cover stories that seem to indicate questioning of traditional thinking
Goldman Sachs minimizes the BW story in today’s economic commentary.
Doug
Goldman Sachs US Economic Research US Daily: Is GDP Growth a Phantom? (Smyth)
June 11, 2007
*This week’s Business Week cover story argues that “offshoring” has =
led to an overstatement of real GDP in recent years. The problem is a =
quirk in the calculation of the import price index, which leads to an =
insufficient deflation of nominal output.
*The mechanism seems real enough, but the effect is a modest one. The =
Business Week estimate of a $66 billion impact equates to only about =
half of 1 percent ofGDP. Moreover, this number is probably an upper =
bound because some of the underlying assumptions about the =
composition of trade appear to be on the aggressive side. The =
Business Week estimates about the potential size of the drag on =
manufacturing productivity also seem overly pessimistic.
*Another, similar, channel through which GDP could be overstated is =
through undercounting of services and intermediate goods imports. =
Together these factors may help to explain why the real goods =
component ofGDP has consistently been growing faster than industrial =
production in recent years.
This week’s Business Week cover story identified a potential problem =
with the national accounts that might have led to overestimation =
ofGDP growth in recent years. The issue stems from interactions =
between the way in which imports are deflated and the rapid increases =
in imports, partially due to offshoring and outsourcing, over the =
last several years. Underestimation of real GDP can impact our =
estimation of productivity increases for the economy as a whole, the =
size of the offshoring effect, and the amount of international trade.
Specifically, the effect comes from the deflator for import prices =
not keeping pace with rapid increases in how many types of new goods =
are entering the country. The Bureau of Labor Statistics (BLS) keeps =
track of the changes in price as the same good enters the country =
over time. Normally, this works fine. However, when a good that has =
not been imported before enters the country for the first time, the =
BLS faces a problem. There is no way to calculate the price change =
for that good; hence they just assume its price moved by the same =
amount as overall import prices.
This method of introducing new goods can interact with a rapid =
increase in offshoring to cause distortions to output measures. =
Consider the case of a good that was made in the United States, but =
now is being made overseas. Presumably, for a good to be offshored, =
the price must be less than it previously being produced for in =
America. Thus, there should be a discrete price drop as the US =
production of the good ceases and the cheaper foreign production =
takes over. But, as discussed above, this is not captured in the =
import price index if the good was not imported before as the index =
does not include the actual price drop but only the change in overall =
import prices.
Since GDP is computed looking at the total value of goods sold in the =
US, with the effect of imports backed out, an understatement of the =
drop in import prices can lead to an overstatement in real GDP. Put =
differently, when the transition to foreign production occurs, the =
real value of imports looks smaller than it really is as more of the =
nominal value of imports is assumed to be price. This smaller value =
then does not back enough out of GDP to account for how much is =
really imported. Instead, with the incorrect adjustment, real GDP now =
looks higher than it should be.
We think the effect identified by Business Week is real. Looking =
through the descriptions of the construction of data, it appears that =
the chain of events does occur as Business Week posits. As always =
with us economists, there are some caveats. The main one is that the =
effect only occurs when a good enters the imported goods sample, any =
goods that are already being imported are not subject to these =
concerns. Further, the BLS appears to attempt to link new goods to =
slightly different ones that might have already been imported, a =
similar process to what occurs in the CPI. However, constraints =
(budget, manpower, complexity and so on) limit their ability to do so.
That said, we think the estimated size of the effect, $66 billion in =
2006 dollars, is on the high side. We make this judgment because =
their analysis: (1) attributes the entire increase in imports from =
low-cost countries to new varieties of goods, (2) assumes a quite =
large price advantage for the newly imported goods over the domestic =
good they are replacing, and (3) doesn’t allow for any substitution =
between countries (e.g. for goods formerly produced in Korea now =
being produced in China). As the article acknowledges, the =
assumptions made in trying to assess the magnitude are by necessity =
heroic. This necessity stems from the fact that the data needed to =
answer the question do not exist, leaving very rough back-of-the- =
envelope calculations as the only route. Still, because of the =
reasons detailed above, we view the $66 billion as roughly an upper =
bound on the effect.
Even that upper bound, compared to the economy as a whole, is a large =
though not earth shattering impact. The $66 billion that GDP might be =
overstated corresponds to half a percent of total US GDP-certainly =
nothing to sneeze at but not something that turns our view of the =
world completely on its head. Put another way, it’s about 3% of the =
increase in real GDP since 2003, the time frame over which the effect =
was calculated, or less than 0.2 percentage points per year on the =
growth rate. Other factors are much more important overall than this.
One of the main implications the article draws is that, if the output =
effect is mainly concentrated in manufacturing, this sector’s =
productivity growth could be substantially overstated. However, =
manufacturing productivity is calculated using industrial production, =
not GDP. This is important because industrial production is a direct =
output measure and is therefore not vulnerable to import price =
mismeasurement. Over the past several years, manufacturing =
productivity has been growing at a 3=BD% annual rate, faster than most =
measures for the economy as a whole.
Still, over the last several years, the real goods component of GDP =
has consistently been growing faster than industrial production. =
Since the beginning of 2003, the real goods component of GDP has =
grown at nearly a 5% annualized pace-a bit less than twice as fast as =
industrial production has been growing. As real goods GDP is =
calculated indirectly as total domestic spending on goods plus =
exports minus imports of such items, it is subject to the problems =
stemming from incorrect import deflation in the Business Week article =
while the industrial production data is not. Therefore, faster growth =
of real goods GDP than industrial production is consistent with the =
goods GDP measure being overstated.
The import price deflator effect is probably part of the story, but =
it is not the only possible channel for outsourcing and offshoring to =
have an effect. Another source comes from not fully measuring imports =
of services that enter into US production. The essential story is =
that a company offshores part of their production, perhaps a call =
center or some form of intermediate production overseas. Then these =
imports of intermediate goods and services (especially services since =
these tend to be poorly measured) are undercounted. For instance, in =
2005 (the most recent data available) the Bureau of Economic Activity =
[sic - DH; s/b Analysis] puts “computer information services” imports =
from India at $400 million. This figure seems small both relative to =
anecdotal evidence and the fact that one single large Indian software =
firm had 2005 revenues of over $1 billion. This mismeasurement of =
imports is similar to what occurred with the deflator problem as the =
undercount of imports leads to an overestimation of GDP. (For more =
details on this process see “Has Outsourcing Inflated the GDP =
Numbers?” US Economics Analyst, 04/13) However, unlike the import =
deflator story, which should affect only real GDP, this would also =
affect nominal GDP.
Seamus Smyth