In 2008 two Princeton Economists, Faruk Gul and Wolfgang Pesendorfer, published an increasingly influential methodological statement, "The Case for Mindless Economics" (hereafter "GP08"). Professors Gul and Pesendorfer publish regularly together and they also happen to be among the tightly-knit group of core-gate-keepers in the economics profession. So, for example, if you look at the submission guidelines of Theoretical Economics [TE], co-edited by F. Gul, you can read: "If you have previously submitted your paper to Econometrica, you have the option of requesting that the referees' reports and covering letters and the editor's decision letter be transferred to the coeditor assigned to handle your paper at TE." Of course, until very recently Pesendorfer was one of the co-editors at Econometrica. (It would be impolite, of course, to view these journals as rent-seeking instruments, but how else to interpret economically this policy: "a paper judged to be unlikely to be acceptable by a second round will be rejected, either without consultation with referees or in response to referee reports. In either case, the submission fee will not be refunded.") Econometrica does have an important "conflict of interest policy," but that does not prevent group-think. Either way, we can safely treat GP08 as a proxy for (recent) establishment views in economics.
The main and (almost) only target of GP08 is what they call "neuro-economics," which they conflate with (experimental) research on the brain. (They also frequently use the term "philosophy" to refer to an enterprise completely irrelevant to "economics" now and always.) Gp08 systematically ignores experimental research conducted by, say, economists (e.g. Vernon Smith and his various collaborators) that also focus on what GP08 calls "economic data." This is important to keep in mind when we evaluate the main thesis of GP08, which is that economics is mainly about rational choice theory (and its natural extension). The thesis is offered as a descriptive account of "common practice" among economists (1), although we also learn that given the economic "evidence" available to economists this approach has also rightly earned a "central role in economics." (43-44) Here's a statement of the main thesis:
While Gp08 does not mention the falling costs of computing and the increasing availability of data, this description of the "standard approach" fits well the recent trend toward data-mining (as practiced by economists)--note that all kinds of possible correlations available in economic data are systematically neglected in this approach. Before I evaluate the claims of PG08 and the arguments offered for them critically, it is worth noting how much of contemporary economics this approach leaves out: agent-based modelling/simulations, pure theory, and experimental economics do not have a seat at the table. Approaches that use methods alongside these -- e.g., institutional, historical, even biological analysis -- also have a dubious status.
Now the main strategy of GP08 is to insist (ad nausea) that claims about the brain are simply irrelevant to economics: "Neuroscience evidence cannot refute economic models because the latter make no assumptions and draw no conclusions about the physiology of the brain. Conversely, brain science cannot revolutionize economics because the latter has no vehicle for addressing the concerns of economics." (1-2) As the economist Bryan Caplan points out GP08 systematically conflates brain with mind as well as neuroscience with psychology. If we grant the arguments of GP08, all that has been ruled out is the relevance to economics of brain&neuroscience not alternative sciences of mind/psychology (and Kaplan offers reasons for not even granting this much).
One peculiar consequence of the strategy of GP08 is the retreat from hegemony. Gone are the days of economic imperialism. Economics is now presented as a modest intellectual enterprise with (a) a fairly limited domain (a method that is only appropriate to particular data), so that economists and psychologists simply "have different objectives and address different empirical evidence." (2; see also 9 and especially 25) It is striking that these economists see no gains-from-trade in the intellectual division of labor. (I have ridiculed this stance as specialization without trade.) Moreover, (b) the reformulated standard approach has remarkably limited social policy objectives:
Greater psychological realism is not an appropriate modeling criterion for economics and therapeutic social activism is not its goal. Welfare analysis helps economists understand how things are by comparing the existing situation to how things might have been in a plausible alternative institutional setting; welfare theory is not a blueprint for a social movement. (44)
Now, on (b), first, not-so-long ago even proponents of the new welfare economics (Lange, Arrow, Samuelson, etc.) had far more ambitious aspirations. Second, until very recently widely admired economists were rather fond of comparing their policy advice to the way medical types treat their patients (for example, recall when Milton Friedman and his followers advocated "shock therapy" as a "medicine" for various economic/social ills).
While I share the distrust of paternalist social engineers exhibited by GP08, we should not ignore how their "therapeutic" is a bit of a weasel-word. Because the standard approach is still allowed to diagnose "problems and anomalies" in institutional regimes that are not "efficient" (3; see also "market failures" (13)). For example, "Normative statements (farm subsidies are inefficient) are used to define new positive questions (what makes farm subsidies persist?) that lead to better models of the underlying institution." (3o) It's not clear what according to GP08 prevents respectable economists from calling for more efficient institutions. After all, they happily claim that "farm subsidies could be eliminated and farmers could be compensated in a way that would increase the economic welfare of all US households." (29)
Of course, GP08 pretends as if their "positive" analysis assumes trivial claims about the nature of revealed preference. Yet, their approach takes the institutional framework and rules as given. So, the preferences that are revealed only track "given...constraints (available options)." (6; this leaves aside the fact that some preferences may never be revealed for "supply" reasons within the market.) This makes the standard approach remarkably friendly to existing institutions. Again, as a fellow-anti-revolutionary traveler I should be happy with this stance, but I am a philosopher not somebody claiming to be a "scientist." (40)
Finally, GP08 repeatedly asserts that the standard approach is "successful." (16-17 & 29, 35, 40) But we are never told what this empirical success amounts to. We can share Gul & Pesendorfer's skepticism about the relevance of FMRI scans to economics (41), while still noting that the main thesis of GP08 unabashedly advocates what one might charitably describe as curve-fitting. Evidently curve-fitting is supposed to lead to good predictions, but surprisingly enough no examples are offered of this. (Of course, even minor rule-changes or institutional movement will cause havoc for such predictions, so one wonders what they might have in mind.) As it happens, the paper they use to explain (13) the standard approach and its virtues, “Temporal Resolution of Uncertainty and Dynamic Choice Theory,” by Kreps and Porteus (1978), is extremely popular in mathematical finance theory (17). A field that undoubtedly has had huge impact on society, but not exactly an empirical success story. GP08 suffers from bad timing. Is there any evidence our Princeton professors have learned anything from experience?
Okay, that's enough, for now.
Recent Comments