NO ONE KNOWS OUTSOURCING

Anyone who touts numbers about the extent of outsourcing is blowing smoke. This is my preliminary conclusion drawn after pouring over the BEA's statistics: U.S. International Services: Cross-Border Trade and Sales Through Affiliates, 1986-2002. In particular, I wanted to see what the trade data revealed as to payments to India for service imports.

Look here and you will find a whole lot of nothing. By this I don't mean that there is no offshore outsourcing to India or elsewhere, but that the data available allow no statistical inferences to be drawn. Here's the low down on the data:

The BEA publishes detailed services trade and services sales through affiliates data annually in the October Survey of Current Business.

Total trade in services with India, both receipts and payments, are not reported for the period 1986-2001.

The BEA breaks down the data for unaffiliated trade by country into super-industry level categories: Computer and data processing services, Data base and other information services, R&D and testing, as well as Royalties and license fees. For the most part, payments to India (service imports) in these categories are negligible both in the levels and as a share of total service imports. Payments for computing and data processing services spike up in 1998 from $8 million to $97 million, and then to $132 million in 1999. But in 2001 and 2002, the data show a tapering of these imports. Even so, at its peak, computing and data processing imports only comprise 1% of the total imports for this category. R&D and testing imports have grown to be a sizeable share of the total, from less than 1% in 1995 to 8.1% in 2002.

Only in 1998 does the BEA begin reporting data for unaffiliated Royalties and license fees at a level of detail to show payments/receipts for “General Use Computer Software.” From 1998-2001 payments to India post less than $500,000 each year, and in 2002 the data is suppressed. Royalty payments/receipts between affiliated firms accounts for the vast majority of trade in royalties, however these data are only distinguished by the national origin of the parent and subsidiary, not by the category of service.

The data for affiliated trade in services, it appears, are not directly comparable to the unaffiliated data. (The spreadsheet I put together for you shows payments to foreign MNCs from their majority-owned US affiliates). Whereas the unaffiliated data are classified by the type of service being traded, the affiliated data are classified by the industry of the transacting firm. The industrial classification also means that from 1986-1996 data are classified under the SIC system and 1997 and after data are classified using the NAICS system. For example, under the SIC coding, the most relevant category seems to be “Computer and Data Processing,” but under NAICS the relevant categories include “Software Publishing Industry,” “Data Processing Services,” and “Computer systems design and related services.”

India is not included in the country data, nor do the categories allow us to get at much specificity for India. Outside of Europe and the Western Hemisphere, data are reported under the heading “other countries,” which is further subordinated into Japan and Australia. In other words, the closest we can get to India is all of Eurasia, Africa and the Pacific less Japan and Australia. Also, a lot of the data are suppressed. Even so, payments to the residual of “other countries” seems pretty small relative to the payments to all countries.

What can we learn from this exercise? First, and if nothing else, the data for services trade are pretty poor and sparse. Look at the series for computer and data processing services for India. The fact that the series shows a peak in 2000 and a declining trend thereafter should be the first clue that something screwy is going on and the data is not capturing what is really happening on the ground. Second, these data do not support any conclusions about the current or potential future impact of outsourcing. But once again, even my grandma could tell you that this offshore outsourcing of service work thing is going to be big. See this Washington Post panel of non-crank business academics for more on this topic.

I'll have more to report on this after the Brookings Institution convenes all the usual eggheads to sort out this puzzle.