Why the Recent Samuelson Article is NOT about Offshore Outsourcing

Arvind Panagariya

            The recent article by Paul Samuelson, “Why Ricardo and Mill Rebut and Confirm Arguments of Mainstream Economists Supporting Globalization” in the Journal of Economic Perspectives (Summer 2004) raises an important question but it is NOT about offshore outsourcing.  While a detailed analysis of the outsourcing phenomenon including its proper definition in the present-day policy context, the associated analytics, and the implications for the overall U.S. income, jobs and wages is provided in the paper “The Muddles over Outsourcing” by Jagdish Bhagwati, Arvind Panagariya and T. N. Srinivasan, forthcoming in the Journal of Economic Perspectives, Fall 2004, here I confine myself to the narrow issue of why the Samuelson paper is not about outsourcing.  Readers interested in details will want to look at the Bhagwati, Panagariya and Srinivasan paper on which this note is partially (though not wholly by any means) is drawn.

            In order to get a proper fix on the problem, note that offshore outsourcing is defined as the purchase of services abroad at arm’s length with the buyer and seller staying in their respective locations.  They include principally the services transmitted electronically through the phone, fax and Internet.  Call centers in Bangalore serving customers in New York and x-rays transmitted from Boston to be read in Bombay are examples of offshore outsourcing through the electronic mediums.  But outsourcing may also take place through more conventional modes of transportation: for example blood samples may be flown from Minneapolis to be analyzed in Manila with reports sent by airmail as well.

The Samuelson Results in a Nutshell

            Let us now turn to the Samuelson article.  In the introduction, the author provides a clear and concise summary of his results and the reader may wish to read it in full.  In the following, I draw upon the summary, directly quoting the crucial paragraphs. 

Samuelson employs the standard Ricardian model, which assumes two countries (called America and China), two goods (called 1 and 2) and one factor of production (called labor).  Because the endowment of labor is taken as fixed in the Ricardian model, any change in the total national income are reflected fully in the change in the real wage.  If the real wage rises, real incomes of all individuals and therefore the nation rise.  Alternatively stated, the wage also represents the per-capita income in the model.  

It is also useful to remember that in a trading equilibrium in the Ricardian model, each country specializes completely in the production of the good in which it has comparative advantage and it exports (setting aside the exceptional case in which one of the economies is gigantic in relation to the other, a possibility not considered by Samuelson and therefore irrelevant for the following discussion).  Of course, under autarky, meaning a situation of no trade, it must produce both goods if it is to consume both goods.

            Samuelson conducts three experiments in this model:

(1)    He starts at autarky and then allows the countries to trade.  Both America and China unambiguously benefit from this opening to trade.  America has a comparative advantage in good 1 and specializes completely in that good and China in good 2.  Nothing controversial arises here.

(2)    Starting at this free-trade equilibrium, Samuelson next introduces a productivity increase in China in the good it exports, good 2.  With more of good 2 produced, its relative price falls.  America can now buy good 2 more cheaply from China, which benefits America.  Nothing controversial arises here either, at least from the American viewpoint.

(3)    Starting once again at the free-trade equilibrium, Samuelson finally introduces a productivity improvement in China in the good it imports and America exports, good 1.  If this productivity improvement is just right to equalize the cost ratios between America and China that gave rise to trade in the first place, all trade is wiped out and America is robbed off all benefits of trade it previously enjoyed.  

This last result, called Act II by Samuelson [with results in (1) and (2) above called Act I(a) and Act I(b) respectively], is the source of the challenge he throws at the economists who advocate outsourcing as being beneficial for America in the overall sense.  Here it is worthwhile to quote the author directly from the summary in the introduction to his paper:

“Act II, however, deals some weighty blows against economists’ oversimple complacencies about globalization. It shifts focus to a new and different kind of Chinese technical innovation. In Act II, China’s progress takes place (by imitation or home ingenuity or . . .) in good 1, in which the United States has previously had a comparative advantage. (High I.Q. secondary school graduates in South Dakota, who had been receiving from my New York Bank wages one-and-a-half times the U.S. minimum wage for handling phone calls about my credit card, have been laid off since 1990; a Bombay outsourcing unit has come to handle my inquiries. Their Bombay wage rate falls far short of South Dakota’s, but in India their wage far exceeds what their uncles and aunts used to earn.) What does Ricardo-Mill arithmetic tell us about realistic U.S. long-run effects from such outsourcings? In Act II, the new Ricardian productivities imply that, this invention abroad that gives to China some of the comparative advantage that had belonged to the United States can induce for the United States permanent lost per capita real income—an Act II loss even equal to all of Act I(a)’s 100 percent gain over autarky. And, mind well, this would not be a short run impact effect. Ceteris paribus it can be a permanent hurt. (“Permanent” means for as long as the postinvention technologies still apply.)

“In Ricardian equilibrium analysis, there is never any longest run unemployment. So it is not that U.S. jobs are ever lost in the long run; it is that the new labor-market clearing real wage has been lowered by this version of dynamic fair free trade. (Does Act II forget about how the United States benefits from cheaper imports? No. There are no such neat net benefits, but rather there are now new net harmful U.S. terms of trade.)”

Why the Phenomenon Analyzed by Samuelson is NOT Outsourcing

            Samuelson’s analytic result in Act II that technical progress in China can wipe out all potential gains for America is not in dispute at all—as I describe below, it has been known to trade economists at least since 1950s when the late Harry Johnson who taught International Trade at the University of Chicago first demonstrated it.  What is in dispute is whether it represents outsourcing.

            Thus consider the example given by Samuelson in the last first of the two paragraphs quoted above (which is incidentally fully in conformity with the definition of offshore outsourcing provided at the beginning of this note): “High I.Q. secondary school graduates in South Dakota, who had been receiving from my New York Bank wages one-and-a-half times the U.S. minimum wage for handling phone calls about my credit card, have been laid off since 1990; a Bombay outsourcing unit has come to handle my inquiries. (Emphasis added)” In the analytic model, the good experiencing productivity change in China is the one exported by the United States to China (i.e., good 1).  But did any high I.Q. secondary school graduates from South Dakota (or elsewhere in America for that matter) handle the phone calls for the customers in China?  Not really.  The calls were made by the Americans and answered by the Americans—no international trade in them took place.  Virtually all activities associated with outsourcing—call center services, x-rays transmitted electronically to be read abroad, transcribing services, accounting services and virtually all back office services—have this property of having been non-traded before the Internet, phone and fax turned them into traded services.  Therefore, the equilibrium at which Samuelson considers the productivity change is simply the wrong one to represent outsourcing:

            At the end of the long quotation above, Samuelson states, “Does Act II forget about how the United States benefits from cheaper imports? No. There are no such neat net benefits, but rather there are now new net harmful U.S. terms of trade.” But one must ask why are there no benefits at all to America from the cheaper imports?  It is precisely because outsourcing has been modeled incorrectly as an export activity in the initial equilibrium.

            In the Bhagwati, Panagariya and Srinivasan paper mentioned above, we (correctly) model outsourcing services as initially non-traded.  Productivity change takes the form of an electronic innovation that turns these non-traded services into traded ones.  The benefits from importing cheaper services now necessarily arise just as many trade economists have stated in their policy writings.  If the country “importing” the outsourcing services is modeled as large, the possibility of a secondary terms of trade change in the previously traded goods (excluding outsourcing which is initially non-traded) would still exist.  These terms of trade may improve or worsen; there is no presumption as in the Samuelson model that they will necessarily worsen.  If they do worsen, the associated loss may or may not wipe out the primary gain from importing cheaper outsourced services (relative to their domestic cost prior to outsourcing).

            One may go further and ask what if we already import outsourced services and China trains more workers that can expand the supply of the services subject to outsourcing?  Does that scenario legitimate Samuelson’s example?  The answer is again in the negative since this is exactly opposite of the Samuelson scenario: China is getting better (through acquisition of skills) at producing the good it exports.  Therefore, the primary effect of the expansion of outsourcing will still be positive.  One must count on a strong negative secondary effect on the terms of trade in the other goods to produce an overall deterioration in the American welfare.

            Up to this point, we have confined attention to the purchase of arm’s length services abroad by America.  But we should not forget that the opening of the electronic medium would also allow America to sell services abroad at arm’s length, for example, legal, medical and educational services as also banking and insurance services provided to customers abroad electronically.  Because America will earn prices that are at least as high as the cost of production of these services, such sales will generate benefits for America in the usual way.  The net effect will be a gain unless there are secondary terms of trade effects that more than offset the primary benefit from the in-sourcing.

The Samuelson Result Applies to a Different Policy Problem

            The Samuelson result, thus, does not capture the outsourcing phenomenon.  But it still captures one specific current fear, the fear that the trading partners of America such as China are getting better at the goods currently exported by the United States.  If China enters in a big way the market for aircraft that it currently imports and America exports, to be sure, that will lower the prices of the Boeing aircraft and will make America worse off.  But this is a different phenomenon than outsourcing.

            As it turns out, the possibility that exogenous changes in productivity or skill accumulation abroad may harm a country, depending on what happens to the prices of that country’s exports, has been well understood in the analytical literature that goes back over half a century. Thus, when the U.S. economy was growing more briskly than the European in the 1950s, Europeans were concerned that US growth injured their standard of living. When Japan was growing rapidly in the 1960s and 1970s, many Americans fearful of Japan becoming the premier world economy, were equally concerned that Japanese growth would harm the U.S.

            The result that all depends on the induced terms of trade change, if any, can be traced back to the literature inspired by the European fears of U.S. productivity growth in the 1950s. In one of several pioneering contributions, Harry Johnson (1954) constructed a two-country, two-good model in which each country specialized entirely in one good.  When the United States economy grew, the production of its export good increased and, provided the import good was not inferior in U.S. consumption, the effect was to increase U.S. exports of its own good, lower the price of U.S. exports and help Europe. Johnson (1955) then generalized the analysis by allowing the production of both goods by each country: this allowed the consumption effect of growth to be offset by the production effect of the growth, so that (consistent with market stability) the terms of trade could either rise or fall, leaving the effect on European welfare, ambiguous.

Whether the change abroad is significant enough, in terms of its net effects on excess demands and supplies of goods at existing terms of trade, and whether it makes sense to worry about sufficiently large “national monopoly power” in international trade such that large terms of trade changes may follow from modest changes in excess demands and supplies, are empirical questions.  The Bhagwati, Panagariya and Srinivasan paper discounts the possibility of significant terms of trade changes following productivity changes and skill accumulation abroad.



Johnson, Harry. 1954.  “Increasing Productivity, Income-price Trends and Trade Balance.” Economic Journal 64: 462-85.

Johnson, Harry.  1955.  “Economic Expansion and International Trade.” Manchester School of Economic and Social Studies 23: 95-112.