This essay is an edited version of a paper that originally ran in the Journal of Political Economy in October 1993 entitled “George Stigler’s Contribution to the Economic Analysis of Regulation.”
George Stigler changed the way economists analyze government regulation. This enormous legacy is essentially embodied in two articles: his 1962 analysis of electricity rates with Claire Friedland and his 1971 theoretical piece. There were, to be sure, precursors and successors in his own work. But none of them so profoundly affected the course of intellectual inquiry as these two.
The importance of both articles rests more, I believe, with the questions each posed than with the answers. Even by the standards of the day, neither piece evinced the sort of technical sophistication that the profession has come to admire. Neither produced conclusions that were impervious to all serious subsequent challenges. But both produced insights that have altered the course of research in the area to this day.
To understand Stigler’s contribution, it is best to proceed chronologically and to begin with the state of the field around 1962. There is room here only for caricature rather than extensive survey. However, I think it is fair to say that the field derived its main energy from the normative economics of marginal-cost pricing. The proper role of government was to correct the private-market failures that prevented attainment of marginal-cost pricing. Applied economic analysis of regulation was then largely descriptive, and its main tendency was to show how the regulatory institutions constrained departures from marginal-cost pricing.
Monopoly was viewed as the main barrier to marginal-cost pricing, and the “natural monopoly” occupied center stage in the economics of regulation. The prime examples of natural monopoly were supposed to be the public utilities. The main role of utility regulators was held to be prevention of private exploitation of the market power that would inevitably flow from natural-monopoly cost conditions.
And the broad professional consensus was that utility regulation succeeded in this task.
It was understood that regulation did not succeed completely in forcing price down to marginal cost. Political and legal constraints on public subsidization of the difference between marginal and average cost forced the regulators to set something approximating average cost prices. But it was regarded almost as given that without regulatory intervention, utility rates would be substantially higher than they were.
The main problem with this professional consensus was that it had never been subject to empirical verification prior to 1962. In this fact lies the main contribution of the Stigler-Friedland paper: it changed permanently the pervasive tendency of economists to accept without examination the effects of a range of government regulation. It launched an intellectual enterprise that is still in business.
The substance of the Stigler-Friedland article is an analysis of the effect of regulation on electricity rates using a regression estimate of the reduced form
price = f (D, S, regulation)
where (D) is demand shifters (urbanization and income) and (S) is nonregulatory supply shifters (fuel price and share of output from [low-cost] hydro power). The sample is a cross section of states in the early 1920s; regulation is represented by a binary dummy. This setup took advantage of the uneven spread of state regulation of the industry prior to 1920, which left about one-third of the states unregulated. The main result was a statistically weak and small difference (under 5 percent) between prices in regulated and unregulated states. (A similar result was obtained for output.)
The impact of the Stigler-Friedland article on the profession owed as much to this then-startling result as to the methodological innovation of estimating the effect of regulation from an explicit statistical model. Had the result merely confirmed the conventional wisdom, economists might have been less eager to pursue the effects of regulation. It is, in this connection, ironic that the original result is wrong about the magnitude (but not the statistical significance) of the effect of regulation. A miscoding of the dummy variable for regulation (regulated states = +10) and the use of common instead of natural logs caused the estimated effect of regulation to be dramatically understated. If we set the historical record straight by repeating the key regressions on the original data set (provided by Friedland), there are a number of minor discrepancies; but the most important result is the size of the coefficients of the correctly coded dummy variables. They imply that regulation lowered price by about a fourth and thereby caused output to rise by over half.
These specific results were superseded by subsequent work. The more important outcome of the article was that economists began eagerly estimating effects of many kinds of regulation. They also began to confront a number of previously neglected theoretical questions.
One of these questions concerned the prior beliefs that economists should bring to any study of the effects of regulation. If Stigler and Friedland were right about the inadequacy of the view that electricity regulators prevented the depredations of monopoly, what should one expect from this form (and from other forms) of regulation? As is implied by their title (“What Can Regulators Regulate? The Case of Electricity”), Stigler and Friedland grappled with this question and thought they had an answer: that demand and cost functions were sufficiently elastic to preclude a substantial effect of regulation. Perhaps the industry was a natural monopoly in some narrow technical sense. But both customers and capital were sufficiently mobile to leave little pricing discretion for regulators.
Stigler and Friedland did not try to generalize this assertion. However, to Stigler’s colleagues and students of the early 1960s, he did press the case for a basically competitive economy with substantial long-run resource mobility in which regulation was bound to be mainly ineffective. Some of his published work of the period reflects this view. But he gradually abandoned it.
Stigler did this under the accumulating weight of the evidence spawned by the Stigler-Friedland article. Much of this found regulation to be effective, but in ways opposite to what the traditional “public interest” model of regulation would suggest. A clear statement appears in Jordan’s 1972 survey of this first wave of post-Stigler-Friedland empirical studies. He finds that “the essential thrust [of the evidence] has been consistent with implications derived from a producer protection hypothesis once the effects of prior market structure were taken into consideration.” As Jordan saw it, where the prior (i.e., ex-regulation) market structure was monopoly, as in electricity, regulation was found to be ineffective. Where competition would otherwise prevail (his prime examples pertained to transportation), regulation typically lowered output, raised price, and generated monopoly rents. Since effective regulation of a natural monopoly would be inconsistent with the interests of sellers, the single story consistent with the whole pattern of mixed results on the effects of regulation was, according to Jordan, one in which regulators preserved or promoted the interests of sellers rather than consumers.
Jordan’s formulation is symptomatic of a revisionism that, by the early 1970s, was exercising considerable influence over the profession. Within a decade, the benign view of regulation as promoter of the general interest had been mainly abandoned. The ascendant image was of the regulator captured by the regulated.
Although this image was hardly new, the willingness of many economists to embrace it on the basis of mounting evidence was new. As with much else in economics, evidence preceded theory. In this case, the evidence of capture seemed to ask for an explanation of why regulation had come to work in this seemingly perverse way. The answer provided in Stigler’s 1971 article on the theory of regulation stands as his second major contribution to the economics of regulation.
As with the Stigler-Friedland article, it is necessary to distinguish the methodological contribution of the 1971 article from its specific result. The specific result is a theoretical rationale for capture of regulatory agencies by producer interests. In subsequent development of Stigler’s theoretical apparatus, as done by myself in 1976 and by the late Gary S. Becker in 1983, regulators serve a broader constituency than regulated producers. In time, Stigler came to accept these generalizations. So, his capture result needs to be taken in its historical context. In 1971, the wave of professional enthusiasm for evidence of capture was just cresting.
What survives from Stigler’s 1971 article is an integration of the economics of regulation and the economics of politics in which transactions between self-interested suppliers and demanders determine the regulatory outcome. Because of this supplier-demander framework, the body of theory pioneered by Stigler has come to be called the “economic theory of regulation.” In any market, transactions are costly, and his 1971 article is the first serious inquiry into the costs of expressing a politically effective demand to regulators. This yielded an emphasis on the importance of organized interest groups that remains an important part of contemporary analyses of regulation.
The suppliers in Stigler’s theory are unspecified political actors. No distinction is made among legislators, executives, and their regulator-agents. What they have to sell is power: tangibly, power over, say, prices and entry, but ultimately over the wealth of a regulated industry’s buyers and sellers. These two groups compete for access to this power, and the high bidder wins. The currency with which the demanders bid is obviously a bit more complex than the stuff reported in the monetary aggregates. It includes votes delivered in support of politicians, campaign contributions, jobs in the political afterlife, and so forth. The prototypical result of the competition is the triumph of the cohesive producer interest over the diffuse consumer interest. This is manifest in regulatory decisions on prices and entry that transfer rents from consumers to producers. More generally, the political equilibrium in Stigler’s model is one in which cohesive minorities tax diffuse majorities.
To get this result, Stigler drew on several strands of the then-emerging public-choice literature. His basic assumption of political utility maximization was an important ingredient of James Buchanan and Gordon Tullock’s influential 1962 work. In 1957 Anthony Downs had argued that voters would be “rationally ignorant” about most public policies because of weak incentives to acquire information. For this reason, Stigler argued, consumers were an unreliable ally of the rational regulator. Their stake in the regulatory outcome was typically too small to make them informed supporters of proconsumer policies or opponents of anticonsumer policies. By contrast, the producer stake was typically large enough to overcome rational ignorance. But knowledge needs to be translated into power, and in politics this requires organizing to bring pressure to bear on the political process. Here the compact producer groups also have the advantage, because they can more easily overcome the “free-rider” deterrent to collective action emphasized by Mancur Olson in 1965. In short, producers have decisive information and organization cost advantages over consumers.
An important reason for the lasting influence of Stigler’s 1971 article—one recent review calls it “the watershed event” in the literature on the politics of regulation—is primacy. Although the components of Stigler’s theory were not new, the application to regulation was new. And it occurred at an appropriate time. Prior to 1960, economists essentially ignored the political context within which regulation occurs, even as the economic analysis of politics was making important strides. By 1970, economists were seeking a rationale for the seemingly perverse effects their empirical research had uncovered. Stigler’s theory provided it in a form bound to appeal to economists: a model of rational choice. Whatever the subsequent modifications of the theory, the emphasis on regulators as rational actors has remained a durable part of the economics of regulation. So, too, has the emphasis on the role of organized interests in regulatory politics. In this way, the economics of regulation has been decisively changed. The pre-1960, benign view of the effects of regulation has not been entirely superseded. But it is no longer simply assumed. Arguments that regulation enhances efficiency now must show “what’s in it” for the political actors when they move in that direction.
This theoretical transformation came to have consequences not only for the study of regulation but also for economists’ views on public policy. I would credit the line of work begun by Stigler’s theory with a catalytic role in shifting the professional center of gravity toward skepticism about the social utility of regulation. Economists venerate efficiency. When they became convinced that regulation was not in fact primarily efficiency enhancing, their ardor for it cooled. When they became convinced that the primary goal of regulation was not, even in principle, likely to be efficiency maximization, their ardor cooled further. To be sure, regulation does not yet occupy the same place as, say, tariffs. Economists are almost unanimously convinced (without much evidence) that tariffs decrease welfare, though they admit the contrary as a theoretical possibility. Regulation elicits a more diverse reaction, both across types and across economists. But this difference has narrowed considerably, because economists now understand that regulation and tariffs share a common political susceptibility to the influence of organized interests.
Stigler’s theory has also affected the course of empirical research on regulation. His 1971 article contains some direct applications of the theory. Measures of regulatory decisions (e.g., weight limits on trucks) are regressed on measures of the strength or stakes of affected interest groups (farmers’ investments in trucks). However, the main influence of the theory lies in a direction different from such direct applications. It has been to sensitize researchers to look toward interest group pressures for an understanding of the effects of regulation. I can illustrate this by comparing styles of pre- and post-1971 empirical research on regulation. Consider, as an example of the former, the Stigler-Friedland article. The main line of inquiry here is: Did regulation accomplish its stated goal (lower rates)? The Stigler of 1962 armed with his theory of 1971 might have asked instead: Which influential interest groups would the utility commission plausibly serve? And he might then have gone on to inquire about the effects of regulation on, say, workers, large users of electricity, suppliers of fuel and equipment, and so forth. Questions like this are commonplace after 1971.
There is room here only for illustrative examples. One would be the literature on the effects of regulation on wages. The stated goals of ordinary rate-entry regulation are silent about the welfare of workers. But economists came to understand that workers had a concentrated, often organized, interest in regulatory decisions. So they were drawn to investigate, with some success, the connection between regulation and wages. Another example would be the literature on the effects of the convoluted regulation of oil prices following the rise of OPEC. This regulation had minor effects on the prices paid by consumers of refined products, and a pre-1971-type study might have stopped after demonstrating that. Or it might have puzzled over the apparent perversity of the numerous inefficient refineries induced to enter by the regulation. The actual literature focused on the transfer of rents among interest groups (refiners gained and producers lost). Stigler’s theory provides no clear insight about which interest group triumphed here. But it created the intellectual background in which the search for the distribution of rents among these interests becomes a primary focus of research.
George Stigler will be remembered for extending economists’ rational-behavior paradigm to regulation. He argued eloquently in an article from 1982 against our tendency to treat public policy “as a curious mixture of benevolent public interest and unintentional blunders.” In so doing, he sensitized us to the political utility of seemingly inefficient policies. But as Stigler would undoubtedly remind us, inefficiency is just a synonym for our ignorance. When our ignorance can be measured in billions (of dollars of seemingly pointless regulatory expenditures), the challenge to progress in the intellectual enterprise that owes so much to George Stigler becomes clear.
Sam Peltzman is Ralph and Dorothy Keller Distinguished Service Professor Emeritus of Economics at Chicago Booth.