Posted on Wednesday, December 8, 2010
A recent op-ed in the Wall Street Journal declared: "The fact is that for every job outsourced to Bangalore, nearly two jobs are created in Buffalo and other American cities." This article, and others like it, asserts that while outsourcing may send some jobs overseas it simultaneously creates more jobs and that the net is positive job creation.
Often these articles have data that compares companies that outsource with companies that don't. Their data shows that the outsourcers grow more rapidly than the non-outsourcers and add lots of jobs in the United States. This data is then interpreted to mean that outsourcing is creating more jobs in the U.S. than are lost through outsourcing.
While this sounds impressive, it is probably wise to take a hard look at the data before attempting to reduce unemployment in the U.S. by advocating more outsourcing. So let's check out a small hypothetical example.
We will see that, depending on the statistics you choose to look at, you can conclude that outsourcing makes companies prosperous and productive and creates jobs in the U.S., or alternatively that outsourcing destroys jobs in the U.S. and adds to the trade deficit.
Consider a hypothetical industry with 20 companies that make the same products and compete with each other. Each company has 10,000 jobs in the United States and these jobs split into two types: those that are can be outsourced (we will call these the O-type jobs), and those that cannot be outsourced (the NO-type jobs). Let's assume that in a typical company one-third of the jobs are O-type and two-thirds are NO type.
As outsourcing becomes possible, 10 of the companies decide to outsource and 10 decide not to. Let's call them the O-companies and the NO-companies.
As the outsourced work can be done for half the cost of the same work in the United States, the O-companies quickly gain a substantial cost advantage and take over the industry from the NO-companies. Once this process is complete we have only 10 companies left: All are O-companies. Each O-company has doubled its market share and therefore needs 20,000 workers to get the work done. Since these are O-type companies the 20,000 jobs are split: 13,333 of these workers are in the United States and 6,666 overseas.
At the industry level it is clear that the industry has gone from 20 companies of 10,000 each, a total of 200,000 U.S. jobs, to 10 companies with 13,333, U.S. jobs. This makes a total of 133,333 U.S. jobs. The result: 66,666 jobs have left the U.S.
However if we now do a study that is not at the industry level but rather looks at the individual companies, as was done in the study referred to in the Wall Street Journal article, the picture is quite different.
Each surviving company in the industry has doubled its output and each now has 13,333 employees in the United States. This is 3,333 more jobs than each company had before outsourcing began. Each company has not only replaced the 3,333 jobs it originally outsourced, but also has added 3,333 new jobs.
Comparing the new industry with the old would also show that U.S. productivity in the industry has gone up since outsourcing became the norm. The new smaller U.S. work force is now producing the original amount of product.
Even if you measure productivity more carefully, using value added in the United States per U.S. employee, that measure goes up too. The value added in the United States is the total value the product is sold for, less the value of any purchased input. In our case the purchased input is the product from the O-jobs now being done overseas. That input now costs half of the one-third formerly paid for those O-jobs in the U.S. or one-sixth. So the value added in the U.S. is five-sixth of the total. A U.S. workforce of two-thirds its original size is producing five-sixth of the value add. Value add per capita has gone up. But remarkably enough, despite the productivity numbers, all these U.S. workers are actually doing just the same thing they were doing before.
At the industry level it is clear that jobs have disappeared. It is also true that the industry has gone from having no impact on trade to becoming an import industry, one that adds to the current trade deficit. But if you look at the companies themselves, they are growing, they are prosperous, they are productive, and they have added U.S. jobs. We might easily conclude that outsourcing creates growth and jobs in the United States.
This hypothetical example illustrates the perils of dealing with statistics without a clear view of the processes that generate them.
Or to put it in another way, we have the well-known quote, which Mark Twain attributes to Disraeli: "There are three kinds of lies: lies, damned lies and statistics."
Ralph Gomory, Research Prof. NYU, Pres. Emeritus, Alfred P Sloan Foundation, Former IBM SVP Science-Tech