I’m going through old copies of the Journal of Law and Economics, the first journal to which I subscribed, before sending them off to a young woman who is majoring in economics and thinking about going into a graduate program. I came across an article by my UCLA industrial organization professor, Sam Peltzman. I took his two-quarter IO sequence in my first year at UCLA, which was his last year before he moved to the University of Chicago.

The article I came across is his “The Gains and Losses from Industrial Concentration,” JLE, October 1977. I had marked it up a a good bit. In this article Peltzman asks the question “[I]f [industry] concentration and profitability are indeed related, what market process produces the relationship?” That’s a pretty important question.

He then states:

The traditional answer has been that high concentration facilitates collusion and super marginal-cost pricing, for which some profitability measure is a proxy. Unfortunately, this answer does not logically follow from the usual evidence, so its acceptance by economists and practitioners of antitrust policy is little more than an act of faith.

After going through the data, he does a nice back-of-the-envelope calculation showing that breaking up concentrated industries would hurt the firms, by raising costs, and consumers, by raising prices.

He writes:

To get at the magnitude of the risks facing an anticoncentration policy, we can focus on industries which have a four-firm concentration ratio greater than .5. The average concentration ratio in this sector is around .7, and the typical member spent something over 70 cents per dollar of output for payroll and raw materials. Now imagine that through a divestiture action the concentration ratio for such an industry is reduced to .5. Given our empirical results, this action could raise unit costs on the order of 20 percent, which in turn would raise price by 10 to 15 percent. Assuming unit elastic demand, the lower figure would impose a cost on consumers of around 9.6 cents per dollar’s worth of output, of which 9.1 cents would be a transfer to producers. Resource costs would increase by around 12.7 cents per dollar of output, so producers would lose 3.6 cents per dollar, and the total loss would be just over 13 cents. [DRH explanation of the math: producers gain 9.1 cents per dollar and lose 12.7 cents per dollar, leaving a net loss of 12.6 minus 9.1, or 3.6 cents per dollar. Consumers would lose 9.6 cents per dollar and so the overall loss is 9.6 cents plus 3.6 cents, which is 13.2 cents.] Since this concentrated sector currently accounts for around one-fourth or 250 billion dollars of manufacturing sales, any extensive deconcentration program would risk imposing losses which are many times greater than the typical estimates of the benefits such a policy might have been thought to produce.

Sam was always so good at these back of the envelope calculations. I remember being in Washington briefly in the summer of 1974 and finding out about a conference at the American Enterprise Institute at which Sam spoke. He presented a similar kind of calculation on an issue involving the pharmaceutical market. An economist from, I think, the Labor Department named Alex Maurizi was the discussant. In his discussion, he challenged some of Sam’s numbers. But I pointed out, in Q&A, that Maurizi’s alternative numbers would make Sam’s point all the stronger. This back-of-the-envelope style was one of the best things I learned from Sam.