Weekly Philo of Economics: the architect of neo-Liberalism (Harberger) and his theoretical critic (Khan)

Let's distinguish among three uses of "neo-Liberalism:"

1. The imposition of markets absent rule by discussion (imperfectly approximated in Liberal Democracies)--often, by way of a dictator, foreign aid institutions [IMF, World Bank, etc.], foreign experts [e.g., Post-Communist Russia]); this is generally associated with 'shock therapy' of some sort and often occurs in times of 'crisis' or 'emergency.'
2. The use of market-friendly-rhetoric within democracies to justify "privatizing" government services; the privatized entities become private (quasi-) monopolies managed by a rent-seeking, credentialed, managerial class.
3. The tendency by the financial services industry to 'privatize gain and socialize risk' at the expense of their much poorer fellow tax-payers. 

Now (2) and (3) are deplorable, but only (1) primarily involves social engineering by experts.

Of course, in practice (1-3) can be blended in all kinds of ways: often the 'privatized' industries come with various implicit government guarantees, so then an instance of (2) turns into an instance akin to (3).  There may be more uses of "neo-Liberalism" that I am overlooking, of course! But I want to distinguish between these three varities of neo-Liberalism from instances of what is known as "economic imperialism." Economic imperialism occurs when the methods and doctrines of economics are extended into domains not previously associated with economics. Gary Becker's work on the family is a paradigmatic instance of economic imperialism. (As I have pointed out "economic imperialism" predates Becker; it was probably first diagnosed by the sociologist Talcott Parsons a few generations before Becker.)

Neo-Liberalism and economic imperialism are both commonly associated with so-called "Chicago economics." But distinct aspects of economics have been brought to bear to support the varieties of neo-Liberalism. In my previous blogging I have focused quite a bit on how aspects of finance theory paved the way for (3). Below I focus on the way the economics profession became/is implicated in (1). A key figure is the "Chicago"-economist, Arnold Harberger. Harberger's entanglement with neo-Liberalism is not merely academic; he was the key mentor to the Chilean (and other Latin American) "Chicago Boys" prior to and at the time of the dictatorships of the 1970s. The understandable focus on Milton Friedman by critics of "Chicago" has made informed discussion of these matters more complicated. As I have tried to explain, there are genuine technical differences in the ways Friedman and Harberger practice economics. (This is not meant to exonerate Friedman's role(s) in (1-3) nor his attempts at defending the Chicago Boys [see my piece].)

Harberger matters to (3) because he is one of the key figures who rhetorically helped turn applied welfare economics into a "professional consensus" generating device in so-called project evaluation in which the economist-as-expert is no different from a professional engineer or a doctor. All the bold/italicized terms figure prominently in the first few pages of Harberger's classic 1971 article. The consensus that Harberger aims for is in the service of "applied welfare economics," that is to "increase, to society's general benefit, the influence on public policy of good economic analysis." (1971: 786; I would argue that is no surprise that Harberger echoes Thomas Kuhn then popular at Chicago.) From a longer historical perspective one can say that Harberger attempts to return applied economics to its roots in civil engineering (still present in French and Latin countries). Harberger is a paradigmatic instance of what the economist, MA Khan, calls (following Oakeshott) a "theoretician" (recall here). Keep that in mind.

One nice thing about mathematical technology is that it can be turned into an inferential device that with a given input turns into determinate answer. It is, thus, eminently suitable to generate consensus. So, Harberger proposes to privilege a particular mathematical technique to be used by economist-experts in evaluating public policy (associated with establishing the so-called "consumer surplus"). 

First, part of Harberger's argument is that the proposed techniques are no worse (and maybe even better) than existing aggregation-technologies used by public policy economists. For example, "national income" is routinely treated as "measure" of aggregate welfare. (787) Harberger realizes that both measures are imperfect, of course.  But what makes both measures attractive is that they facilitate public policy advice. Harberger is disarmingly frank about this: "they can readily be used to define a set of policties that characterizes a full optimum. This entails no more than introducing taxes, subsidies, or other policies." (795) Here Harberger is copying a  move that (future Nobel laureate) Koopmans executed in 1949 on behalf of econometrics: economists can become privileged policy advisors if they adopt inferential technologies that allow them to speak in one voice.

Second, Harberger grants that the technique he advocates is silent on capturing distribution effects from changes in policy. This has two non-trivial consequences: (a) it makes the approach he advocates extremely attractive to circumstances of the sort intended to be captured by neo-Liberalism (1) above. So, the technocratic expert can claim to offer consensus policy evaluation and advice in the context when particular distribution effects can be safely ignored (or suppressed). Harberger may not have foreseen this in 1971, but he defended the approach through the next few decades while his students were advising dictators (and others) in Latin America. Moreover, (b) while consumer surplus approaches are not data-free, they yield the directionality of their prescriptions with relatively limited empirical research. I learned this by accident when I learned how in the 1950s Harberger's approach displaced other (more data-rich) research approaches that were used within economics (at Chicago, in fact). So, it is not just a purported consensus-generating technology, it is also cheap to use and yields its results fairly quicky. (It is"efficient.") Again, this makes it especially attractive in circumstances of crisis/emergency. 

Third, Harberger's approach presupposes that policy-makers are made alert to the limitations of the proposed approach; according to Harberger economists "must have the modesty and honesty not to claim for our profession more than we are particularly qualified to deliver." (786) But he never stops to reflect on the fact that in most policy environments economists may face incentives that make them less likely to be fully forthcoming about what is left out of their adopted framework. This wouldn't matter so much if he had explored the opportunity costs of adopting his consensus project. In particular, he never even considers that a portfolio of alternative techniques that economists could bring to the policy-environment might, in fact, provide better guidance in policy context. Let's stipulate on his behalf that the the techniques of the (alternative) portfolio are individually less efficient than the consumer-surplus approach, but it might well be that such an alternative portfolio might lead to more informed, even wiser public utterances in policy context (and may be less subject to abuse during a crisis/emergency.)

Fourth, Harberger likes to present the consumer-surplus approach as akin to civil engineering. Economists have grown extremely fond of the comparison. But engineers are very good at both empirically exploring possible risk-factors in implementing their suggestions and building rather huge margins of error into these. This is, in part, a consequence because they recognize not just the limits to their knowledge, but also that their mathematical techniques cannot capture all the relevant factors at play. Harberger's focus on generating consensus, in fact, is deliberately designed to suppress (uncertainty induced) disciplinary variance and the exploration of risk-factors.  

Fifth, so far I have pretended that Harberger's approach can deliver the consensus it promises. But in a wide-ranging 1992 lecture, Khan shows that  this can't be so.  (In the first instance, Khan does so by working with models of the sort advocated by Harberger.) Now, in addition to the very technical criticisms lodged by Khan, what Khan's analysis reveals is that the postulates underlying the technique adopted by Harberger are extremely sensitive to decisions by the modeler on what empirical data to incorporate. (This will be especially obvious in contexts where important economic activity is not captured adequately by official data.) So, in context where Harberger's approach does result in consensus one must question to what degree this is the product of the model itself, or either due to very limited data or due to shared values by the modelers (and their policy-clients).

Khan puts the issue delicately: "Economic  theorizing represents a double movement -- from reality to model to reality. Each such movement represents a compromise and it makes little sense to me to search for the compromise" (557; emphasis in original). One reason why Harberger is a paradigmatic technician in Khan's terminology (recall above), is that the judgments involved in facilitating this "double-movement" are not made transparent either in the technique deployed nor in the way the result of the double-movement are presented to policy-clients.

Moreover, Khan's analysis reveals that the technique advocated by Harberger are not robust in the real world. In his comments on Khan's paper, Rodgers puts the point nicely: "the fundamental problem is that in  order to approach the real world you need to introduce complexity." (564) Given that we're dealing with a policy oriented technique, this lack of robustness ought to be fatal.

Finally, Khan's criticism hasn't been fatal to Harberger's approach. To understand this consequence, one could do worse than asking cui bono? As I have been arguing (here, here, here) incentives matter in economics in the absence of extremely robust countervailing theoretical virtues. This will remain true long after neo-Liberalism's moment has passed.