Juan-Pablo Montero and Juan Ignacio Guzmán∗ May 11, 2006
Abstract Following the structure of many commodity markets, we consider a few large ﬁrms and a competitive fringe of many small suppliers choosing quantities in an inﬁnite-horizon setting subject to demand shocks. We show that a collusive agreement among the large ﬁrmsmay not only bring an output contraction but also an output expansion (relative to the non-collusive output level). The latter occurs during booms and is due to the strategic substitutability of quantities (we will never observe an output-expanding collusion in a price-setting game). We also ﬁnd that the time at which maximal collusion is most diﬃcult to sustain can be either at booms orrecessions.
In Table 1 we reproduce Orris C. Herﬁndahl’s Table 3 (1959, p. 115) with a summary of the evolution of the so-called international copper cartel that consisted of the ﬁve largest ﬁrms and operated during the four years preceding the Second World War. Herﬁndahl argues that the cartel was successful in restricting output during the periods of low demand (denoted as Quotastatus and associated to lower spot prices in the London Metal Exchange) but failed to extend
Montero (email@example.com) is Associate Professor of Economics at the Catholic Univertity of Chile (PUC) and Research Associate at the MIT Center for Energy and Environmental Policy Research; Guzmán (firstname.lastname@example.org) is PhD candidate at PUC. We thank Claudio Agostini, Jeremy Bulow, Joe Harrington,Bill Hogan, Jason Lepore, Salvador Valdés, Felipe Zurita and seminars participants for helpful discussions and comments. This work was completed while Montero was visiting the Harvard’s Kennedy School of Government under a Repsol YPF-Harvard KSG Research Fellowship. Montero also thanks Fondecyt (Grant # 1051008) for ﬁnancial support.
such restrictions to the periods of high demand whenthe cartel and non-cartel ﬁrms returned to their non-collusive output levels.1 Herﬁndahl’s description appears consistent with some existing collusion theories; in particular, with Rotemberg and Saloner’s (1986) prediction for the evolution of a cartel under conditions of demand ﬂuctuations in that collusive ﬁrms have more diﬃculties in sustaining collusion during booms (i.e., periods of highdemand).2 We advance a diﬀerent behavioral hypothesis in this paper. We posit that the large output expansions undertaken by cartel members during the two booms (Jan.—Nov. 1937 and Oct.—Dec. 1938) may not necessarily reﬂect a return to the non-collusive (i.e., Nash-Cournot) equilibrium but rather a continuation with the collusive agreement in the form of a coordinated output expansion of cartel membersabove their Nash-Cournot levels.3 The objective of this paper is to explore the conditions under which a collusive agreement, if sustained, can take an output-expanding form at least during some part of the business cycle and discuss its welfare implications. Although we do not run any empirical test, we will see that the international copper cartel of 1935-39 as well as many of today’s commoditymarkets appear to be good candidates in which such collusive characterization may apply. There are basically two reasons for that. First, in these markets a ﬁrm’ strategic variable is its level of production while prices are cleared, say, in a metal exchange. Second, collusive eﬀorts, if any, are likely to be carried out by a fraction of the industry (typically, the largest ﬁrms) leaving, forincentive compatibility reasons, an important fraction of the industry (consisting mostly of a large number of small ﬁrms) outside the collusive agreement but nevertheless enjoying any eventual price increase brought forward by the collusive agreement (we will often refer to the group of non-cartel ﬁrms as competitive fringe and to the group of potential cartel ﬁrms as strategic or large ﬁrms).4 It...