The economics of open-air markets

John List 05 November 2009



Bilateral bargaining has constituted the foundation of markets for centuries, from peasant economies – such as Athen’s Agora, Rome’s Forum, the medieval fairs and markets in England, and the 1,000 year old market in Morocco – to the substantial bazaars and “flea” markets that litter the landscape of developed and developing countries today. Geertz (1978), for example, found that up to two-thirds of the population of the Moroccan town he studied was employed at the local bazaar.

Despite the importance of such markets in shaping economies of yesteryear and allocating goods and services today, little is known about the basic economic principles of such markets. For example, whether the bread and butter of economics – supply and demand curve shape and location – provide accurate predictions of price and quantity realisations in such domains is relatively unknown.

In recent research, I use open-air markets as my experimental laboratory to explore the fundamental underpinnings of such markets (List 2009). My general line of attack is to undertake controlled experiments in these markets where factors at the heart of my conjecture are identifiable and arise endogenously. I then impose the remaining experimental controls to learn something about underlying market behaviour and equilibrating tendencies.

A key result from the field experiments is the strong tendency for exchange prices to approach the prediction of competitive market equilibrium theory. Even under the most severe tests of neoclassical theory (treatments that predict highly asymmetric rents) the expected price and quantity levels are approximated in many market periods. These results suggest that in mature markets very few of the “typical” assumptions, such as Walrasian tâtonnement or centrally occurring open outcry of bids and offers, are necessary to approximate the predicted equilibrium in the field (see also List, 2004).

Yet, such markets are ripe for price manipulation. For instance, in certain cases small numbers of sellers provide homogeneous goods that are jointly purchased from middlemen, barriers to entry exist, and seller communication is continual. Indeed, via interaction with an individual seller in this marketplace, I learned interesting details of just such conspiracies in these markets. Armed with knowledge from my “mole” and using insights gained from months of interaction in the market, I am able to build a bridge between the lab and the field, effectively exploring the behaviour of experimental subjects across sterile and rich settings. This approach has a dual benefit in that it affords an opportunity to marry the vast experimental literature on collusion in laboratory experiments with parallel behaviour in the field.

How effective is collusion?

For more than 100 years, economists and historians serving as expert witnesses, commissioners, and jurists have laboured to assess the “effectiveness” of cartels (Conner, 2005). Throughout the years, those critical of aggressive antitrust policy have embraced the Stiglerian notion that cartels are fragile coalitions, fraught with cheating that will eventually produce outcomes that more closely mirror competitive expectations. For example, when the OPEC cartel began to influence world petroleum prices in the early 1970s, several leading economists predicted its imminent demise. Adelman (1972, p. 71) wrote that “Every cartel has in time been destroyed by one and then some members chiselling and cheating…”

An excellent recent overview of cartel activities is provided by Levenstein and Suslow (2006). They conclude that even though many cartels collapse within a year, the average cartel in their sample lasted between 3.7 and 10 years (see p. 51, Table 1).1 The effect that cartels have on prices is less well understood. Conner's (2005) survey identified hundreds of published social-science studies of private, hard-core cartels that contained 674 observations of long-run overcharges. The primary finding is that the median cartel overcharge for all types of cartels over all time periods is 25%; 18% for domestic cartels, 32% for international cartels, and 28% for all successful cartels (the overcharge rate is calculated by comparing cartel prices to a competitive benchmark).

In these regards the experimental data presented in my study provide some interesting parallels. First, consistent with Stigler (1964), the natural field experimental data suggest that maintaining strict compliance to the collusive agreement is difficult to sustain in a repeated game with secret price cuts and demand uncertainty. In this manner, the data are consonant with the notion that inherent problems associated with maintaining collusive agreements might preclude conspiracies from having considerable influence on prices in similar market structures. Second, however, in those cases where collusion was stable (the lab and framed field treatments), much surplus was lost. Sellers were able to exact large price increases through explicit collusive arrangements and such conspiracies frustrated market efficiency considerably.

These contrasting insights from nearly identical experimental situations highlight a general lesson for policymakers; transference of results across situations (or time, or different industries) is dangerous. Theory teaches us that the determinants of cartel success are rich, and the empirical data herein highlight that sellers are quite sensitive to these factors, even ones that many might consider ancillary. In this spirit, one prominent aspect of the data is that one should take great care when generalising results. Theory and comparative static empirical insights can inform us of general principles, such as when and where to expect collusion, and when to suspect that collusion is having an important influence on pricing and allocation decisions. But, statements on the actual existence, or efficacy, of collusive arrangements are quite difficult to make without actually investigating the industry itself. I conclude in noting that empirical work measuring comparative statics, such as those in this study, can provide a hint about where to look for fire but cannot determine by themselves whether there is an actual fire worth extinguishing.


1 This bimodality of cartel survivorship has been the subject of intense research and our understanding of its causes is just beginning to crystallize. While cheating is certainly an important factor, Levenstein and Suslow (2006, p. 45) argue that “the most frequent causes of cartel failure are entry and bargaining problems.”


Adelman, Morris A. (1972-73), “Is the Oil Shortage Real?Foreign Policy, 71, Winter.

Conner, John M. (2005), “Price-Fixing Overcharges: Legal and Economic Evidence.” American Antitrust Institute Working Paper 04-05.

Levenstein, Margaret C., and Valerie Y. Suslow (2006). “What Determines Cartel Success?Journal of Economic Literature, Vol. XLIV, March, pp. 43–95.

List, John A. (2004), "Testing Neoclassical Competitive Theory in Multi-Lateral Decentralized Markets," Journal of Political Economy (2004), 112(5): 1131-1156.

List, John A. (2009), “The Economics of Open Air Markets”, NBER Working Paper 15420, October.

Geertz, Clifford (1978). “The Bazaar Economy: Information and Search in Peasant Marketing,” American Economic Review, 68(2) May, pp. 28-32.

Stigler, George (1964), “A Theory of Oligopoly,” Journal of Political Economy 72, 44-61.



Topics:  Microeconomic regulation

Tags:  Collusion, bazaar, prices

Professor of Economics, University of Chicago


CEPR Policy Research