Do manufacturers need special treatment?

By CHRISTINA D. ROMER

Christina D. Romer is an economics professor at the University of California, Berkeley, and was the chairwoman of President Obama's Council of Economic Advisers.

EVERYONE seems to be talking about a crisis in manufacturing. Workers, business leaders and politicians lament the decline of this traditionally central part of the American economy. President Obama, in his State of the Union address, singled out manufacturing for special tax breaks and support. Many go further, by urging trade restrictions or direct government investment in promising industries.

A successful argument for a government manufacturing policy has to go beyond the feeling that it's better to produce ''real things'' than services. American consumers value health care and haircuts as much as washing machines and hair dryers. And our earnings from exporting architectural plans for a building in Shanghai are as real as those from exporting cars to Canada.

The economic rationales for a policy aimed specifically at shoring up manufacturing largely fall into three categories. None are completely convincing:

MARKET FAILURES

Government intervention can be justified on efficiency grounds if the free market won't work well. For example, when competition in a market is limited, antitrust laws that prevent monopoly can be helpful.

In manufacturing, the market can malfunction if there are positive externalities across companies. That means that some benefits of a manufacturing plant go to companies other than the one deciding whether to build it. Clusters of manufacturing businesses can be more productive than an individual one. As a result, when an entrepreneur sets up a plant, some of the benefits accrue to other businesses in the area.

This argument could justify government subsidies or tax breaks. But large clustering effects have been hard to find. A study by Professors Glenn Ellison of M.I.T. and Edward Glaeser of Harvard showed that in many industries, businesses were only modestly more clustered than if they were allocated randomly -- suggesting that the benefits, while real, may often be small.

Moreover, the logic of clustering's benefits is likely to apply outside manufacturing. Software development, insurance and entertainment are three service industries where we observe clustering and where positive externalities may be large. Why single out manufacturing for special treatment?

A related argument for subsidizing manufacNike