Colin Camerer on Strategic Management

12 May 2007 at 10:24 am Nicolai Foss 4 comments

| Nicolai Foss |

Strategic management researchers are, as a rule, practically oriented folks who typically do not have much patience with lofty debates in the theory of science. Say the word “ontology” and you will have eyes rolling in the audience (yes, I have tried it!).

I am currently working on a chapter on methodological/philosophy-of-science discussions in strategic management for Giovanni Battista Dagnino’s forthcoming Handbook of Research on Competitive Strategy and, given the above characterization, I have actually been surprised by the number of published papers on meta-theoretical issues in strategic management.

One of the best, most succinct, and, without any doubt whatsoever, most provocative papers on this is a more-than-two-decades old piece (1985), “Redirecting Research in Business Policy and Strategy,” written by a then-very-young Colin Camerer. Today Camerer is of course best known as an advocate of behavioral economics, but he contributed a handful of excellent methodological strategy papers in the mid-1980s (including an award-winning piece with Keith Weigelt). Here is the abstract of the 1985 paper:

Because business policy research has primarily been a series of inductive generalizations of case studies, theories have been typically ambiguous and untested, and have not progressed swiftly. Deductive theorizing, by contrast, yields clear, often non-obvious conclusions that can be debated effectively and generalized slowly; so realism of current models can be sacrificed for progress towards realistic future models. Deductive theorizing, with more attention to a game-theoretic definition of equilibrium and to recent ideas from economics, should be one new direction for policy research. Of course, these deductive models will inevitably draw their inspiration from the richness of careful observation and exhaustive checklist-making that are the hallmarks of induction. Specific avenues for new research are described, and the importance of teaching non-obvious theories is defended.

The abstract doesn’t really reveal it, but even today the paper is so provocative and uncompromising that some parts of it have to be read to be believed. In other words, a great piece. One thing that is worth pondering is why the “Camerer program” of relying much more on deductive, game-theoretic reasoning has met with such limited success in terms of the number of practicioners. Only a handful of strategy people such as Rich Makadok, O&M guest blogger Steven Postrel, Michael Ryall, Pankaj Ghemawat, Adam Brandenburger, and Peter Zemsky adhere to the program in their research practice, but some of them are more properly described as working within economics than within strategic management (in terms of where they publish; this also goes for those who publish in the Journal of Economics and Management Strategy). It seems that to the extent that strategic management has become more rigorous, it has been in the dimension of empirical analysis rather than formal methods (see this). Will this change?

Entry filed under: - Foss -, Recommended Reading, Strategic Management.

Private Provision of Public Goods, Mario Puzo Edition Design for the Bottom of the Pyramid

4 Comments Add your own

1. Joe Mahoney | 12 May 2007 at 11:52 am

Nicolai,

The entire debate on induction vs deduction is like the debate about whether it is best to hop on your left leg or your right leg to move forward. With the possible exception of an intellectual decathalon athlete with one leg who may come along, I am confident to say that the most progess will be made by those using both.

Further, as one mathematician once noted “Theorems Come First and Axioms Follow”. That is, almost all published papers are False in the sense that they are written deductively but that is NOT how the (inductive and deductive) process of writing a paper is typically done. The published paper is a Reconstructed logic (and NOT the logic-in-use). That the process of research SHOULD BE the reconstructed logic as Camerer (1985) maintained is dubious.

Finally, the exercise in game theory is to put the rabbit in the hat within the model and then pull the rabbit out of the hat. Doctoral students today may be taught as many as twenty different equilibrium concepts in their Ph.D. Economics courses. (Although they can get by with Dominant Strategy, Nash, and Perfect Equilbrium Concepts). All of this work in game theory might lead one to the conclusion of Nash — Ogden, and not John — “progress [in game theory] was OK once, but it has gone on for far too long.” How useful are all the game theory refinements (beyond the three concepts named above) to Strategic Management and Strategic Managers? The likely answer is: Not Very Useful.
2. John Mathews | 16 May 2007 at 4:35 am

Hi Nicolai and Joe

Sorry to have been absent from these fascinating debates until now.

From a philosophy of science perspective, I think the key issue is not so much induction vs. deduction, since the framework of strategic management if taken into the disequilibrium domain will have few testable propositions of the kind generated in game theory. But there is one appealing and overwhelming parallel between science and business, and that is that entrepreneurial dynamics have an immediate interpretation as the test of a (business or market) conjecture.

Several such conjectures are formed in the “mind” of the firm, or more immediately, in the mind of the entrepreneur. The conjecture can cover a huge array of possibilities. It might be a conjecture about the possibility of blending a technology of interactivity with a technology of digital TV reception, to generate a particular way of delivering an interactive digital TV service. It might be a proposition that lowering the price of a service by a mobile phone company will increase its marketshare. Whatever the conjecture, it is made in conditions that Knight described as uncertainty, that is, in conditions that cannot be made the subject of an insurance policy. In a free enterprise system, there is a way to resolve such uncertainty, and that is to put the conjecture to the test.(I think Knight was the first to state this clearly.)

Now the test of a conjecture in this sense, like the testing of a scientific conjecture, with which it is perfectly analogous, carries some important implications. The first is that the entity conducting the test has the resources needed to do so. As a professor, I might entertain the conjecture that a word-processing program that writes books automatically, without the need for thought, might have a large potential market. But unless I am prepared to go into the market and secure the resources needed to test such a conjecture, it remains a speculation. (This is a caveat that I would attach to the otherwise insightful ‘knowledge theory of the firm’.) Likewise a scientific conjecture can only be tested in a laboratory, with all the required lab equipment, trained lab technicians, and intellectual capital needed to make the test of a conjecture credible.

The second implication is that there is a reasonable means available of deciding whether the test is successful or not. In the scientific field, this involves canons of good practice, plausibility and concordance with the observed facts (in the Popperian sense, of refuting a conjecture). In the business field the criteria are no less ruthless, but less transparent: the firm conducting the test of the conjecture must eventually make a profit, or the conjecture is deemed “false.” Harper’s 1996 book is a terrific engagement with these issues.
3. spostrel | 17 May 2007 at 6:22 pm

As I sweat through a very painful revision of what started out as an inductive, exploratory paper, I find the notion that the strategy field is friendly to inductive approaches blackly humorous. A certain proportion of scholars in the field appear to be actively disgusted, even outraged, by accounts of exploratory discovery. Finding things you weren’t looking for in the first place (and then building theory about them) apparently is equivalent to chewing with your mouth open.

On the other hand, we see ideas like the RBV, which have a strong deductive flavor, take hold fairly strongly. The field’s aversion (now fading) has been to rigor and formality, not deduction per se.

As far as the multiplicity of equilbrium concepts in game theory goes, that is not an important source of indeterminacy in most modeling situations. The correct concept for a given application is usually pretty intuitive, which is why the literature in that area mostly consists of plausible generalizations (refinements) refuted by specific counterexamples (e.g. subgame perfection vs. the centipede game) where the generalization just “feels” wrong.

The indeterminacies to worry about are timing, information sets, strategy spaces, etc., where more than one plausible set of assumptions often exists. Of course, using no formal game theory and thus sweeping these matters under the rug is by no means an improvement. But it makes people feel better.
4. Peter Klein | 17 May 2007 at 9:42 pm

Fast Company published a piece a couple of years ago on the use of game theory in business decision making:

http://www.fastcompany.com/magazine/91/debunk.html

Apparently none of the experts questioned — neither David Levine, nor Hal Varian, nor Preston McAfee — could cite a single example of an actual firm using game theory to make a business decision, with the exception of bidding at spectrum auctions.