Knowledge Problem

Lynne Kiesling

On Thursday President-elect Obama named law professor Cass Sunstein to head the Office of Information and Regulatory Affairs, an executive-branch office with the mission of analyzing and coordinating federal regulation. Most recently, Sunstein is known for his work with Richard Thaler on “choice architecture” and behavioral public policy, including their book Nudge.

Others have already opined about Sunstein’s appointment, including Saturday’s editorial from the Wall Street Journal and this Forbes column from Glenn Reynolds. Matt Welch helpfully aggregates several commentaries on Sunstein’s appointment in his Reason post. The consensus seems to be that Sunstein’s approach to regulation is not premised upon bureaucratic imposition and control, that his primary focus is to improve markets and competition, not to stifle them through government regulation, and that much of this is likely to be irrelevant because OIRA is a politically impotent office, so his efforts will have little effect.

Most observers will be watching to see what form Sunstein’s focus on “choice architecture” and behavioral public policy will take when and if it gets implemented. The premise of this behavioral approach to what I (and Elinor Ostrom) would call institutional design is Sunstein’s and Thaler’s so-called “libertarian paternalism”. Sunstein and Thaler draw their normative policy implications from behavioral economics research that suggests that real-world decision-making does not always result in individuals making the decisions that theoretical models suggest would be optimal, or rational. In his review of Nudge, Will Wilkinson cites some examples:

If a cafeteria puts its key lime pie a bit out of the way, fewer people will succumb to delicious temptation. If employees are not required to fill out confusing paperwork to enroll in a savings plan, more of them will enroll.

One area where this behavioral issue comes up in electricity instutional design is the design of fixed-price default service contracts, and whether the default contract should be opt-in or opt-out. Those who argue it should be opt-in base their argument on consumer inertia and status quo bias — if the heretofore-regulated incumbent is the default service provider and the default contract is opt-out, innate consumer inertia would predispose them to stay on the default contract, even if competing suppliers could now enter the market and offer services that would likely be appealing to those customers. Thus by reinforcing existing inertia and status quo bias, the opt-out default contract reduces competitive entry into new retail electricity markets, and stifles retail competition. This opt-in/opt-out design dimension is relevant for health care plans, retirement plans, any sort of contractual choice situation with a status quo option.

In some ways this approach is a refreshing change from regulation that is informed by traditional, neoclassical economic modeling of individual choice, which is devoid of any formal attention to the real cognitive constraints that individuals face daily when bombarded with information and making choices in the face of budget constraints, information constraints, and (to use the Austrian phrase for it) radical ignorance — some of what individuals need to know to make the decisions that they would in our formal theoretical models is simply unknowable. Note here that the ideas I am using originate in the work of Herb Simon on bounded rationality and the use of heuristics in individual decision-making, and of Michael Polanyi on tacit knowledge and the extent to which we know stuff that we don’t know we know, and our brains subconsciously make choices constantly, which enables us to function in our information-rich environment without being overwhelmed. This behavioral approach to institutional design takes into account two important aspects of reality:

1. Individuals do face cognitive limitations on their decision-making, and have evolved adaptation mechanisms to deal with them.

2. To live together in civil society, we do have to engage in collective action, which requires institutional design.

In other words, take people as they really are, and when we have to engage in collective action, design institutions that take into account those cognitive realities. Yes. But here’s where my skepticism about the Sunstein-Thaler libertarian paternalism kicks in. Much of what is important in public policy is shaping incentives and future behavior. If we take Simon and Polanyi, and Hayek, seriously, then it’s crucial to acknowledge that a lot of what shapes incentives and future behavior is not just unknown — it’s unknowable. We don’t know our own preferences completely — we discover them through market processes, and through the process of making decisions. We don’t know what future choice sets will look like, let alone what our preferences over those choices will be. These things are simply unknowable, and therefore designing institutions to affect those unknowable incentives and outcomes is almost always likely to end up in failure. This approach to choice architecture cannot overcome the knowledge problem. So I think the best that we can hope for from such a so-called “libertarian paternalism” approach to regulation is small, local changes in design parameters that are well-known and relatively non-controversial.

Another aspect of the challenge to this approach to institutional design is indeed the elitist-paternalist critique: how will the regulator know in every situation that Choice B actually would make the individual better off than Choice A? What makes you think that you know what is better for me than I do? Seriously? Again, the unknowability of preferences until presented with a real choice situation, even to the person possessing them, makes it highly unlikely that any third party can know that one outcome would be better for a person than another outcome. This is old territory that has been covered elsewhere, including this April post from Will Wilkinson, so I will stop here, except for one observation: just because we face cognitive limitations to decision-making, that does not necessarily mean that the ultimate outcomes of decision-making are inefficient, or irrational, or “anomalies”, to use one of Thaler’s favorite linguistic tropes. Again, taking Simon and Polanyi and Hayek seriously, we devise conscious and subconscious adaptations to these cognitive realities, we use technology to help us filter and to make decisions that better reflect our preferences as we perceive them. Who are you to say that the outcome of that process is irrational or inferior?

Another question to ask when thinking about the process of institutional design and behavioral economics: why would these “Nudge” rules be any less prone to political manipulation, lobbying, and rent-seeking than more traditional top-down institutional design? The electricity default contract is an example; the traditional “consumer advocate” community generally uses its political and advocacy power to argue that opt-in default would expose consumers to too much unwanted price volatility and too much effort to avoid that price volatility, so the default should be everyone going on a fixed-price contract unless they want something else. “Nudge”-based institutions are still institutions of collective action, and thus are still going to have the same incentives for manipulation, lobbying, and rent seeking.

Finally, while I do think that the process of institutional design, including regulatory institutional design, can benefit from thinking about these choice architecture questions, I do think the “libertarian paternalism” nomenclature does a disservice to the language. “Nudge” applied to regulatory policy is not strictly libertarian, because regulation is still ultimately grounded in coercion. The “choice architecture” language does a much better job of communicating what this idea is about — incorporating behavioral and cognitive reality into institutional design. At least the “Nudge” approach to regulatory policy is more respectful of individual autonomy and choice than the likely bureaucratic alternatives.

In devising OIRA policy I’d like to hear Sunstein invoke another of his former Chicago colleagues, Ronald Coase, and state that in promoting and facilitating open, transparent markets, the most important role of economic regulatory policy is to reduce the transaction costs that prevent private parties from engaging in mutually-beneficial exchange. That means using OIRA to evaluate the entry barriers and other transaction costs that federal regulation can create. OIRA does cover other areas of regulatory policy where the ideas of transaction costs are not as strictly relevant, but thinking in terms of reciprocal benefit, reducing transaction costs, and the alignment of, for example, environmental and economic incentives through the reduction of transaction costs and better-defined property rights transcends economic regulation.