Banking is like manufacturing explosives. However carefully you manage it, there are inevitable blow-ups. But if you keep production units small and sufficiently separated, you can minimize the overall damage that any eventual accident will cause.

Ravi Jagannathan, a finance professor at Northwestern University, coined that metaphor in a recent journal article. As someone who earns a living trying to come up with catchy metaphors and analogies to explain economics to ordinary people, I can only bow in honor.

The important point he and his collaborator, John Boyd, from the University of Minnesota, make is that if the failure of a very large bank or other financial institution can threaten the stability of the economy as a whole - as the actual or feared failures of Bear Stearns, Lehman Brothers, AIG, Merrill Lynch and others shook the U.S. economy in 2008 - then we should ensure that few, if any, institutions ever get that large.

It may cost us something through regulation to keep financial institutions smaller than they might be without it. There may be economies of scale and scope in banking and other financial services. That is, the per-unit cost of such services may fall as financial companies get bigger or as they broaden the range of products they offer. Keep banks from growing in size or offering more products and you raise the cost of their services to households and businesses.

This economies-of-scale-and-scope argument was the rationale for tearing down the regulatory walls we erected separating ordinary commercial banking from investment banking and insurance in response to the 1929 market crash. The same argument also impelled episodic merger frenzies among banks and other financial firms over the past 20 years.

There are economies of scale in manufacturing explosives, too. But that industry learned early on that the probable losses from catastrophic accidents in large facilities more than outweigh such scale economies. So explosives factories usually contain multiple small parallel production lines, sufficiently spaced so that an explosion in one will not propagate to others.

In the case of banking, the actual cost efficiencies from megabanks turned out to be small if not negative. Large banks often were unwieldy. Customers and shareholders suffered alike. Often, only the CEOs who engineered the mergers came out ahead.

Exactly how Boyd and Jagannathan would limit bank size is beyond the scope of this column. But their illustration of why it is necessary is, in itself, a useful contribution to reform.

Economist Edward Lotterman teaches and writes in St. Paul, Minn. Write him at ed@edlotterman.com.