Point/Counterpoint: Please Don't Regulate My Internalities
April 9, 2015 02:23 PM
Brian F. Mannix & Susan E. Dudley
First appeared online May 2015
Click here for the online article
Employing the Kaldor-Hicks criterion (Boardman, Greenberg, Vining, & Weimer, 2011, 32) benefit-cost analysis poses the question: “Would the people who support a particular policy be willing to pay an amount of money sufficient to induce all of those people who oppose that policy to change their votes and thereby achieve unanimous consent?” The compensation is hypothetical, of course, as is the voting; but this thought-experiment provides a framework in which to weigh the net welfare effects of government decisions on real people. In the context of a standard economic analysis of externalities, the hypothetical question—could the winners compensate the losers—makes eminent sense. In the context of ill-defined “internalities,” however, its meaning is far less clear. Are people supposed to be compensating themselves? Why would that induce them to change their votes? And how is it possible that, after all this hypothetical compensation is done, they appear to be many billions of dollars better off than before they paid themselves to do what they did not want to do in the first place?
Drawing on the behavioral literature, Allcott & Sunstein attempt to build a model that relaxes the standard assumption of consumer rationality, so government agencies can design policies that are more optimal in the real world. But the exercise demonstrates just how challenging this is. The authors’ model assumes a consistent, self-defeating “bias” in consumers’ (but not regulators’) decision making. Agencies are somehow able to quantify this bias (presumably by observing rare but revealing inconsistencies in otherwise consistent consumer misbehavior) and to discern consumers’ “true” preferences. Then, rather than tell consumers where they are wrong, the government corrects their errors by offering subsidies, thus inducing them to make optimal decisions in spite of themselves. The authors do not mention who pays the subsidy, nor how their welfare might be affected.
Note that, before the government intervenes, there is no market failure in the Allcott & Sunstein model—only the presumed internal failure of consumer decisionmaking. By using subsidies to distort prices away from true marginal cost, the government is giving false information to consumers, and creating a market failure where none existed. The magic in the model is that the government-manufactured market failure exactly offsets the presumed internal consumer failure, leading to a new equilibrium where resources are allocated optimally!
Of course, such a system would inevitably make unbiased, rational consumers worse off. The model only works because it assumes there are no rational consumers, somehow giving the government a monopoly on rationality. Not only that, but this government-by-perfectly-tuned-subsidies appears to be immune to the rent-seeking and other distortions that we would ordinarily expect to affect the incentives of government actors. The EPA’s Renewable Fuel Standards (RFS) subsidize the production of ethanol from corn, for example, despite the fact that this is harmful to water quality and air quality, and imposes substantial (and regressive) costs on consumers of food as well as gasoline. Yet wealthy ethanol interests will spend millions of dollars over the next year, trying to ensure that no presidential candidate can get through Iowa without committing to support this appalling program (Prentice, 2015). It is unrealistic to propose subsidies without taking into account the perverse incentives that accompany them.
Economic models tend to be formalized and stylized, so we should not be too quick to dismiss the Alcott-Sunstein model; it is interesting and instructive. But it is not a plausible replacement for the standard model. It is undoubtedly true that people are not always rational and do not always behave in their own best interest, but we generally assume that such behavior does not become economically important because markets tolerate and correct for this. Consumers who consistently display intransitive preferences, for example, can be turned into “money pumps”—causing them either to learn rationality, or to cease being consumers. Hence a model that depends upon rational consumers can make a fair claim to being robust; a model that is overly dependent upon rational government—or even just on consistently irrational consumers—can make no such claim (Viscusi & Gayer, 2015).
In arguing for a theory of correctable internalities, Allcott and Sunstein point to research that does not really require such a theory. Yes, consumers make better decisions when they are well-informed, when transaction costs are low, and when costs are not “shrouded.” All of these examples are consistent with a traditional analysis of rational consumer behavior. Similarly, in examining the rationale for CAFE standards, they are correct in asserting that “private net benefits could arise only in the presence of some non-externality distortion.” But that distortion could be due to an easily corrected lack of information, rather than some internal defect in consumers that can only be corrected by force. And the most plausible explanation for the astonishing private benefits in this case is that the regulatory analysis is wrong, rather than that consumers are. Indeed, the authors recognize “a discrepancy between the engineering model results, which suggest very large internalities [leading to suboptimal vehicle purchase decisions], and … empirical estimates, which do not.”
In the face of such conflicting evidence, the safest path is for the government to stick to correcting market failures, and not to start creating them as correctives for some internal defect that is assumed only to affect private actors. Regardless of how smart they are, and how much information they may assemble, regulators can never know more than a minuscule fraction of the information that is known to the multitude—particularly the internal knowledge and preferences that make each of us unique, as producers, as consumers, and as autonomous individuals (Hayek, 1946).
We are not naïve. We recognize that even in a well-functioning democracy, there will be countless government policies adopted that could never pass a benefit-cost test. Sometimes there may be good overriding reasons for adopting them, but sometimes they are simply policy mistakes. In either case, we ask that our economic analyses be transparent, honest, and objective about the effect on human welfare. In the Kaldor-Hicks calculus, neither the government, nor corporations, nor any other artificial person gets a vote; only real humans experience welfare changes. Neither does the government get to put its thumb on the scale. Any truthful analysis of benefits and costs will tell us what consumers think, not what the regulator thinks consumers should think. We do not allow the government to change the results of elections because of some theory of irrational and biased voters; neither should we allow it to distort consumers’ revealed preferences in an economic analysis.
Conventional benefit-cost analysis can be used to optimize policies that correct market failures, thereby improving the welfare of rational and autonomous consumers. Paternalistic benefit-cost analysis, as espoused by Allcott & Sunstein, can be used to justify regulations that instead create market failures, thereby—if you believe the heroic assumptions—improving the welfare of irrational consumers. By systematically distorting prices and giving consumers inaccurate information, the rational, omniscient, benevolent regulator can correct our internalities and steer the economy to a more optimal set of decisions, even if we mere mortals are unable to see it. Which set of assumptions sounds more realistic? Never mind that; which world would you prefer to live in?
BRIAN F. MANNIX is a Research Professor in the George Washington University Regulatory Studies Center, 805 21st Street, N.W., Suite 612, Washington, DC 20052 (e-mail: email@example.com).
SUSAN E. DUDLEY is Director of the GW Regulatory Studies Center and a Distinguished Professor of Practice in the George Washington University Trachtenberg School of Public Policy and Public Administration, 805 21st Street, N.W., Suite 612, Washington, DC 20052 (e-mail: firstname.lastname@example.org).
Boardman, A. E., Greenberg, D. H., Vining, A. R. & Weimer, D. L. (2011). Cost-benefit Analysis: Concepts and Practices (4th ed.). Upper Saddle River, NJ: Pearson Education.
Hayek, F. A. (1945). The Use of Knowledge in Society. American Economic Review, XXXV, 4, 519–530. American Economic Association.
Prentice, C. (2015). “Iowa biofuels industry will spend millions promoting RFS”–Governor. Reuters Market News, March 29, 2015. Found at:
Viscusi, W. K. & Gayer, T. (2015). Behavioral Public Choice: The Behavioral Paradox of Government Policy. Vanderbilt Law and Economics Working Paper Number 15-2. Found at: http://ssrn.com/abstract=2559408.