Monday, May 13, 2013

In the Church of Path Dependence

The First Commandment is not, Brevity is the Soul of Wit, but more like Never Give an Inch. As one of its high priests, Paul David, has made all too obvious over the years in his attempts to recover some dignity after having his claim of market failure in his famous 1985 paper Clio and the Economics of QWERTY demolished by Stan Liebowitz and Steve Margolis in several papers over the years.

Let's visit a very small part of one such attempt from the year 2000, as David attempts to deny he actually said (about the QWERTY typewriter keyboard) that marketplace competition resulted in standardization on the wrong system (emphasis in David's paper);

Actually, it is within the context of static general equilibrium analysis that economists developed the concept of ‘market failure’ – namely, that the Pareto optimality of allocations arrived at via atomistically competitive markets is not guaranteed except under a stringent set of convexity conditions on production and preference sets; and, further, it requires the existence of markets for all extant and contingent commodities. One may or may not accept the usefulness for pragmatic policy purposes of defining ‘market failure’ in a way that takes those conditions as a reference ideal. Analytically, however, it remains a total non sequitur to assert that the essence of path dependence – a property defined for analyses of dynamical and stochastic processes–consists in asserting propositions regarding the possibility of ‘market failure’ that were proved first in the context of purely static and deterministic models. 
Quite the contrary proposition holds: under full convexity conditions a non tatonnement general equilibrium process can be shown to converge in a strictly path dependent manner on one among the continuum of valid ‘core’ solutions which satisfy the criterion of Pareto optimality (see Fisher 1983, and David 1997b). This should be sufficient to expose the logical error of claiming that the essential difference between models of path dependence and standard neoclassical analysis must be the former's insistence on the presence of ‘market failure’. 
The reader is spared the rest of the double talk by David that follows in this paper, as the above should be enough to at least induce the suspicion that Professor David is hoping no one actually reads what he has to say. Instead, let's see what Douglas Puffert, who was Professor David's Stanford research assistant back when the Clio paper was written, had to say recently (2010) when he had the opportunity to write an article on path dependence for the Economic History Association's encyclopedia (which is, at least, unlike David's ramblings, readable); failure per se has never been the primary concern of proponents of the importance of path dependence. Even when proponents have highlighted inefficiency as one possible consequence of path dependence, this inefficiency is often the result of imperfect foresight rather than of market failure. Market failure is, however, the primary concern of Liebowitz and Margolis. This difference in perspective is one reason that the arguments of proponents and opponents have often failed to meet head on....
Market failure?  Perish the thought.

Unfortunately for both David and Puffert, Brian Arthur (the other high priest of Path Dependence) is still around to contradict them, as he did just last year in his comments on Neil Kay's Rerun the tape of history and QWERTY always wins;

QWERTY, as a standard—or better as an example of what the market has served us up in the long evolution of one particular technology—has become in economics a focal point, a rallying point for a larger issue: whether the market can lock us into an inferior standard. And this itself is part of a still larger issue: whether the free markets of capitalist economies can drive us into inferior outcomes.
..... The correct question is whether economic markets can lock in to inferior outcomes.


  1. In any event, the technology of typing has moved on. Free (or certainly cheap) software is available to reassign the keys on most computer keyboards. There is no collective market choice requiring any individual to stick with QUERTY.

    If any one does continue with querty, it must be because they see no reliable, useful return from learning and using another system.

    So, how bad could querty be compared with any alternatives? If QUERTY is say 3% worse than the most efficient (discovered) arrangement, then how is that a market failure?

    Certainly, government can't claim to do better. One could only dream of a world where government choices and implementations were only 3% less than optimal, whatever "optimal" is.

  2. Everything you say is true, Andrew, but it is amazing how some of the best minds--or at least some who think they're among the best minds--in economics, can't get that simple idea through their thick heads.

    We're well beyond the point where they are deserving of any professional, scholarly, or even man-in-the-street indulgence. Consider this from Brad DeLong, back in 2005 on EH.Net;

    I also found myself mystified by what I take to be Liebowitz and Margolies's principal argument. As I understood it, they say that: --the evidence that a Dvorak keyboard is better than Qwerty is suspect. --in fact, we don't know that the Dvorak keyboard is any better than Qwerty. --hence the market worked.
    But I think that this argument deconstructs itself. I believe that Liebowitz and Margolies are correct in their claim that we don't know which keyboard would be better. But that doesn't mean that the market "worked." By now--more than 100 years after the invention of the keyboard as we know it--any sensible and efficient social resource planning and allocation mechanism for establishing standards would have collected the information we need to know what the most efficient keyboard layout is, and whether we should switch.
    Our--market--social resource planning and allocation mechanism has not done so. Therefore it is inefficient: there is an important piece of information that we should have--the relative efficiency of keyboard layouts--that we don't. The market has let us down.
    In my view, L&M have trapped themselves. The more they undermine the case for the superiority of Dvorak, the more they undermine any claim that the market gathers and processes information about what standards would be efficient. The more they claim that the market is an efficient mechanism for gathering and processing information about standards, the more trustworthy does the information we do have that suggests the superiority of Dvorak appear to become.
    So I find myself thinking that Liebowitz and Margolies have deconstructed themselves: they can sustain their case that we don't know that Dvorak is more efficient only be conceding the broader point that the market is pretty lousy at providing incentives for people to evaluate the relative utility of different standards.

    Deirdre McCloskey had some fun at DeLong's expense over that then, but I don't think it shook his faith any.