First, I would like to thank Jason Czarnezki for inviting me to contribute to the ELS space as a guest-blogger this week. Following up the excellent series of posts by Richard Lempert on the ELS movement is no easy task. I have neither sufficient experience as a practicing social scientist nor in the legal academy to supply any meaningful insights on the ELS movement's placement in historical context. But I do hope to start with some thoughts about what the contributions of empiricism to antitrust might tell us about both the promise and limits of ELS (can I substitute empirical law and economics and ELS interchangeably in this context without loss of generality and without offending anybody?).
After all, antitrust represents a best case scenario in terms of susceptibility to empirical influence. The conditions for impact are ideal. Professor Lempert’s last post noted that antitrust was somewhat immune to the legal academy’s “relative inattention to social science.” Why? Well, let me start with what I hope to be two relatively non-controversial and highly correlated assertions. The first is that antitrust is uniquely suited to benefit from economic analysis, especially antitrust empiricism. The second is that antitrust doctrine has embraced economic analysis and empiricism. This was not always so. But there is no real dispute that antitrust welfare economics has won the day. Modern antitrust doctrine directly incorporates economic analysis: will a merger increase prices? Will the conduct at issue produce an anticompetitive effect? What is the impact of the conduct on consumer welfare?
What then, does a body of law that expressly incorporates economic thinking, and particularly empirical learning about business practices, look like? Are there lessons to be learned for other areas of law attuned to empirical learning or is antitrust so uniquely situated that the ELS/antitrust experience should be disregarded for this purpose? While I don’t have conclusive answers to all of these questions, I suspect that the successes and failures of empiricsm and antitrust over time can tell us something about both the potential and limits of ELS more generally.
It is useful to start with a brief and stylized history of the last 60 years of empirical industrial organization ("IO"). In the 1950s and 60s, empirical IO (led by Joe Bain and his students) focused on Census industry-level data and went looking for a relationship between industry concentration and profitability. Researchers would use OLS to estimate the following equation (where H is a measure of industry concentration):
Profits = B0 + B1*
H + (other stuff) + έ
B1 > 0 was taken to represent a positive relationship between concentration and prices or profits and set the basis for a good deal of antitrust thinking regarding concentration policy. This literature established the Structure-Performance-Conduct (SCP) paradigm in antitrust. The obvious endogeneity that arises here is that the error soaks up unobserved industry level characteristics that may be correlated with H. Other critiques, i.e. Fisher and McGowan (1983), focused on problems with the use of accounting profit data. But I want to focus a bit on the evolution of quantitative analysis in empirical IO because it makes a nice point about the rate at which doctrine can reflect a new empirical consensus.
The 1970’s saw the “
Profits = B0 + B1*
Hi + B2* Sif + (other stuff) + έif
The punchline? The relationship between concentration and profits largely disappeared while the B2 was typically positive and significant. The resolution of this quantitative debate might well be considered the empirical IO’s finest hour. Real and important policy debates were resolved because of this literature. The success of this initial stage in the history of empirical IO is ucontroverted, leading to Bill Baxter’s 1982 Horizontal Merger Guidelines, which set the foundation for a coherent economic approach to mergers.
One of the lessons of the
One of the most
promising developments for antitrust in modern economic analysis is the New
Institutional Economics. In
general terms, this body of work seeks to extend and enrich understanding of
the microanalytic details of business behavior and the industry settings that
shape firm conduct. The most
impressive recent competition policy work I have seen reflects the NIE's
teachings about the appropriate approach to antitrust analysis. Much of the
FTC's best work follows the tenets of New Institutional Economics and reflects
careful, fact-based analyses that properly account for institutions and all
the relevant theories, not just market structure and market power theories.
To date, TCE's most
important contribution to antitrust is its ability to provide much better
explanations for most vertical relationships than market power theories. This
contribution has come not so much from theory (although many theoretical papers
apply TCE to vertical relationships) as from careful empirical work. Some
papers use econometric analysis to estimate the importance of broad categories
of transaction costs to the nature of vertical relationships across companies
or industries.
The second stage involved empirical work in its broadest
sense: detailed case studies, qualitative analysis, and quantitative
analysis. To pick just one example, Ben
Klein’s important work on vertical restraints ranging from vertical integration
and block-booking to resale price maintenance, tying, and exclusive dealing
(and now, slotting allowances). Again,
the impact of antitrust law is unmistakable. The legal approach to vertical restraints is much improved relative to
its state in the 1960s and 70s. Of
course, there are debates amongst economists with respect to certain decisions
and the question of whether a specific restraint was anticompetitive. But again, few would question that the
court’s approach to vertical restraints has not improved dramatically as the
result of what I have described as “second stage” empirical IO.
The late 1980s and 1990s represent a third stage in modern IO. During this era, IO was dominated by game theoretic models. Much has been said in the legal academy critiquing this literature for its lack of testable implications and, of course, because the mathematics make it quite difficult for the median legal academic to comprehend, much less use as an input into his own research. Despite these critiques, which I largely agree with, the appeal of this literature is easy to understand. One can easily understand this literature as a response to: (1) intuition that the tools of neoclassical economics are found lacking in markets characterized by few players while game theory incorporates the sorts of strategic interactions one might think important in those markets, and (2) the desire to formalize and notions previously in the literature. This literature, however, has had very little real influence on antitrust law and policy. Muris again diagnoses the problem correctly:
I do not claim that
theoretical, mathematically-oriented economics lacks any value. Improving
theory strengthens any discipline, and important theoretical developments in
economics often come through mathematical modeling. Nonetheless, I am struck
that, despite the enormous recent volume of theoretical, highly mathematical
I.O literature, its effect on antitrust policy and law has been quite small. The fundamental
reason for this modest influence is that, although empirical work overturned
the SCP paradigm, too much of modern I.O. theory adopts the SCP approach by
making market structure the only important market feature in the model. Put
differently, although the SCP debate reveals that there is no systematic
relationship between market structure and the competitiveness of the market,
much of modern I.O. theory allows only market structure and assumed
market power to be important determinants of the competition.
The focus on theory over the past few decades has
left
antitrust practitioners and academics unsatisfied. The highly
formalized theory did not produce
readily testable implications, nor did it lend itself to available
data. Some of this is changing. Detailed micro-level data are more
widely
available. Empirical IO is developing
new techniques. Perhaps the new focus on
empirical IO after the game theory revolution represents a return to
empiricism
and a “fourth stage.” While modern IO
has made undeniable contributions to antitrust policy, especially in
the area
of merger simulation, recent contributions fall obviously short of
those made
by the previous two stages. Of course,
diminishing marginal returns explains some of this. The
Harold
Demsetz once famously announced at a lecture at
Antitrust economics is at its weakest when addressing broad issues of normative policy and at its strongest when explaining particular pricing and marketing practices, an example of which is Aaron Director’s insight into tie-in sales. The strength of economics in specific applications derives from our willingness to accept that the monopoly model is or is not applicable to a particular situation. Nonetheless, on broad issues, such as the identification of excessive degrees of monopoly and the problem of monopolization, economics does not yet offer a coherent message to antitrusters.
I believe this point maintains its importance today. It might even be seen as a call to arms for future
empirical IO economists. There are
opportunities abound. Economists might
do well to use antitrust problems as a research guide. For example, very little is known about the
empirical consequences of exclusive dealing and exclusionary distribution
contracts that are the cause of much antitrust litigation. Simulation models for mergers between
suppliers generally assume a very passive role for the retailer when in reality
retailers have considerable influence over product placement decisions and
relaxing these assumptions might provide more accurate predictions of the
effects of real world mergers. Empirical
evaluations of vertical restraints at the source of recent antitrust
litigation, such as bundled rebates, slotting allowances, and more, are also in
limited supply.
But I’ve
taken you through this lengthy historical journey for a reason: to
evaluate whether the experience of empiricism in antitrust might tell
us anything useful about nurturing this endeavor in other fields so
that it might experience the same successes and avoid similar
pitfalls. If you’ve hung around this long, I might
as well go ahead and share some thoughts (I hesitate to call them
answers) on those issues that I raised at the beginning of the post.
More importantly, I wonder what ELS scholars
in other fields think about the development of the relationship between
empiricism and advances in legal doctrine? But to respond with some
initial and tentative thoughts to my own set of questions:
What
then, does a body of law that expressly incorporates economic thinking, and
particularly empirical learning about business practices, look like?
Interestingly, and perhaps
surprisingly, antitrust doctrine still involves economic debates at what appear
to be the most basic levels, e.g. what is market power (and specifically, does
post-contractual opportunism involve the exercise of such power, and until
recently, does a patent confer antitrust market power?). But it is difficult to conclude that the
developments of antitrust law in response to empirical antitrust scholarship
have been anything but positive for consumer welfare and efficiency. To the extent that one accepts either of
these measures as the appropriate maximand for antitrust policy, empirical
antitrust scholarship has been a success.
Are
there lessons to be learned for other areas of law attuned to empirical
learning or is antitrust so uniquely situated that the ELS/antitrust experience
should be disregarded for this purpose?
It is certainly difficult to generalize the antitrust experience to other areas of law. Not only because of unique features of antitrust law, but also because of the intervening game theory revolution in industrial organization (and other fields). The focus on pure IO theory deprived antitrust of a steady source of empirical research for at least one “generation” of scholarship. This lack of attention means that empirical antitrust questions are in abundant supply. Novel business practices are frequently subjected to antitrust scrutiny when applied by dominant firms. The recent focus on pure theory only increases the value of empirical scholarship because antitrust policymakers must assess which of the possible market outcomes (a set expanded by the theory literature) is most plausible. Modern IO economists are now focusing in increased numbers on improving empirical techniques and contributing to our empirical knowledge of markets. Only time will tell whether the next generation of empirical IO scholarship proves relevant to antitrust law and policy.
While I do not find comparisons between antitrust and other fields inherently invalid, I am quite happy to concede that I know very little about the relationship between ELS and other fields of law, and even less about how this relationship has changed over time in specific fields. With these caveats in mind, I suspect that the disconnect between IO economists and antitrust policymakers was a function of the popularity of game-theoretic approaches and the desire to formalize that has swept through economics (for better or worse) and largely independent of the needs of antitrust policymakers. My own sense is that this type of disconnect is now rare in other fields. Perhaps I’m wrong. But if I’m right, it suggests that a key factor in ELS influencing the law in a positive way is that lines of dialogue between policymakers (lawyers) and empiricists (economists in this case) be kept open. In antitrust, agencies like the FTC play an important role not only in doing research, but helping to set research agendas for economists (not to mention obtaining data). Perhaps the newly developing trend of housing PhD programs in law schools is another way to foster this type of communication and a supply of economists interested in problems that are, well, interesting to lawyers and policymakers. These are all tentative thoughts, and I welcome your reactions and experiences.
My next post will be: "The Father of Empirical Law and Economics," a bit of an ode to an economist I take to be a central, but unappreciated (well, as much as one can be with a Nobel), figure in empirical law and economics.
Comments