Behavioral economics and the Virginia school of political economy: overlaps and complementarities

I hesitate to cite specific works by these prolific scholars as typical of their research—or at least those parts that I am referring to here. The best overviews of their research is provided by Nobel lectures in cases in which a scholar has won that prize, as with Simon (1979) and Kahneman (2002). A good place start with the other two scholars is Thaler (1980) and with Tversky’s joint work with Kahneman: Tversky and Kahneman (1973, 1986).

Again, it with great hesitation that I point to one or two works out of a lifetime of scholarship. For Smith, one should begin with his 2003 Nobel address; for Plott a good place to start is with his 2014 overview of the early days of experimental public choice research.

There are several very lengthy compendia of the works of the Jeremy Bentham and Herbert Spencer. However, a good place to start is with Bentham (1789). Herbert Spencer’s works are also voluminous; I suggest beginning with his first successful book, Spencer (1851) and focus on the sections dealing with the evolution of the physical and social sources of utility.

Information-processing models of the mind had previously played central roles in adaptive models of economics. They are also central to a strand of post-World War II psychological research. See, for example, Kahneman (2002, 2011), Pinker (1999, 2003) and Seligman et al. (2016). Informationally bounded rationality also affected economics and politics, as in Stigler (1961), Downs (1960), and Denzau and North (1994), the latter being the most encompassing of the early discussions by economists. Congleton (1978) is among the first general explorations of informationally bounded choice within the Virginia school.

A useful example of the super-rational approach to political economy is Wittman’s (1995) book on the impossibility of democratic failure.

The Virginia school has its roots in two centers: the Thomas Jefferson Center for the Studies in Political Economy and Social Philosophy at the University of Virginia and the Center for Study of Public Choice at the Virginia Polytechnic Institute and State University and George Mason University (to which it relocated in the early 1980 s).

This essay focuses for the most part on scholars directly associated with these “Centers”, but it bears noting that the Virginia school also includes persons without long affiliations with those centers, such as Mancur Olson, Elinor Ostrom, Dennis Mueller, Charles Goetz, Charles Plott, Geoffrey Brennan, Bruno Frey, Peter Bernholz, Arye Hillman, Randall Holcombe, Gephard Kirchgässner and Michael Munger—many of whom are past visitors to the Center for Study of Public Choice, graduates of its Ph.D. programs, or regular presenters in its seminar series. I am a long-term affiliate—having variously been a student, graduate, research associate and center director during a period spanning more than four decades.

For overviews, see Houser and Stratmann (2012), Plott (2014), Kamm and Schram (2018), or Tyran and Wagner (2018). For a thorough overview of tensions between mainstream rational choice models and experimental results, see Ostrom (1998).

By “natural ignorance” is meant the state of individual knowledge at birth. As mentioned in the text, we are born largely ignorant of all things. Some knowledge is acquired without intent, as for example, that imparted by our parents and stumbled on by accident. Other is consciously sought and acquired. “Rational ignorance” is that part of one’s ignorance that is chosen. That is to say, it exists in cases in which one consciously decides not to pursue a type of information known to exist—such as the Chinese alphabet (for most persons not of Asian descent), quantum mechanics (for those other than professional physicists), and the color of a contemporary rock star’s eyes (for all but the most devoted fans), and so forth. These are cases in which a type or class of information is known to exist, but we (or most of us) choose to remain ignorant.

Such differences would exist even if we were all perfectly rational Bayesians because our direct experiences in life provide only relatively small samples of the choice settings confronted by human beings and the phenomena that we refer to as the “real world”. If our informational systems are not fundamentally Bayesian, no such convergence would necessarily follow from even very large samples.

For applications of the Bayesian approach to economics, see Stoneman (1981); for an early application to public choice, see Congleton (1991). More recently, Bayesian learning has played a key role in machine-learning algorithms; see Taylor and Stone (2009) or Ghavamzadeh et al. (2015). Many economists and game theorists use Baysesian updating as their reference point for rationality and the absence of long-term bias, as for example, in Charness and Chetan (2017).

This expression evidently was the invention of Ralph Waldo Emerson (1849), who argues that one should not worry about consistency, but then concludes that “There will be an agreement in whatever variety of actions, so they be each honest and natural in their hour. For of one will, the actions will be harmonious, however unlike they seem.” The potential irrationality of political decisionmaking was emphasized more recently by Black (1948) and Arrow (1951), who noted that political choices may be inconsistent even if the choices of individual voters are entirely self-consistent. The consistency (or lack thereof) of collective choice captured much of the attention and imagination of public choice theorists for its first four decades.

It should be acknowledged that few mainstream economists actually believe that consumers behave entirely self-consistently or that every individual behaves in a manner that can be regarded as having emerged from fully informed and consistent plans, but many do believe that models based on those assumptions provide useful insights into both individual decisionmaking and political-economic systems.

Downs (1960) also implicitly used natural ignorance to suggest the possibility of biased expectations or complete ignorance about the costs and benefits of government policies. Note that this differed from Stigler’s (1961) theory of the demand for information, which is essentially a demand for sample sizes, from which unbiased estimates would always be possible, albeit with larger error terms associated with smaller samples. Congleton (2001) clarifies this distinction.

My use of the term “clusters” may make it sound as if a sharp break exists among groups of individuals or researchers. Clusters of researchers or research circles are more like colors than atoms. That is to say, they are (improper) subsets along a continuum of research interests and approaches that are sufficiently similar to be identified as a category, rather than completely non-overlapping groups. In Denzau and North’s (1994) terms, members of such categories share mental models.