Mobile Edition
Print
Get the Mag
Weekly UpdatesIntroduction
The Joint Executive Committee (JEC) of the eastern trunk line railroads (1879-1887) is one of the most thoroughly studied cartels in the entire economic literature, because it is one of the few cartels with a large trove of data for scholars to use in testing theories of collusive behavior. (1) The cartel attempted to control both prices and market shares of eastbound freight (and later passengers) from America's breadbasket to the major eastern export cities.
Whether or not the JEC represents an example of successful collusion is a matter of some dispute. Clearly, contracts broke down on several occasions, though in general the railroads would end a competitive reversion with a new contract that the JEC members agreed to enforce. Although it seems that in some periods the railroads did earn monopoly profits, the weight of scholarly opinion suggests that the cartel can at best be considered a qualified success (Ellison 1994; Grossman 2004).
Exactly why the JEC failed to maintain collusion successfully is not entirely understood, although some factors have been identified. For example, entries by new rail companies clearly impacted contract arrangements (Ulen 1978), and demand shocks seem to have led to breakdowns (Ellison 1994).
Yet it is unclear whether these events themselves caused collusive contracts to fail or whether there were inherent problems within the JEC that events like demand shocks to the general economy merely exposed. In fact one aspect of JEC firms that was at least potentially problematic to maintaining collusion has generally not been explored by scholars. That is, the large capital indebtedness of all the companies in the cartel. Railroads were among the most capital intensive enterprises in the entire American economy and all of them had significant yearly debt service charges--typically at fixed rates.
This paper suggests that such debt might well have played a part in the struggles of the JEC to manage collusion. We will argue that the capital charges and the consequences of not meeting them might well have hastened, if not guaranteed, defection strategies on the part of the cartel members.
We show this first in the next section by comparing the JEC to a contemporaneous cartel: the railroad express. As Grossman (2004) argues, the two cartels differed in a number of ways, but perhaps the most striking was that the members of the express cartel were all net creditors, receiving each period a positive cash flow from the debt they were holding. We suggest that their creditor position gave them room for less confrontational strategies than those adopted by the JEC's membership. Indeed, the express used strategies that JEC could not afford. In the third section, we show why from the standpoint of theory alone we would expect successful collusion within the JEC to have been unlikely, but more likely in a cartel such as the express. A standard model of collusion is adapted to illustrate this result. Concluding remarks follow.
The JEC and the EXPRESS
Debt was an ongoing fact of life for all railroads and the benchmark interest rate for the bond market was the rate given to the highest quality railroad bonds not government securities (Grossman 2000). In 1884, the Lake Erie and Western Railroad, a party to the JEC, reported outstanding debt of over $85 million, and had net fixed debt charges for that year of over $6 million. (2) In 1885, the New York, West Shore & Buffalo RR, a subsidiary of the JEC's New York Central assumed $50 million in new debt as part of a recapitalization plan. The Baltimore & Ohio, another JEC company, in the same year consolidated the debt of just one of its branches by assuming a $10 million mortgage. (3) All railroads, in fact, were net debtors and their obligations did lead many into default (including the Erie) at various times in their history.
Consider how this might have affected decision making among the cartel's members. A heavily indebted firm observes falling prices in the market and does not know if this is due to defection by another cartel firm or to a demand shock. The former might require a trigger strategy reversion while the second a coordinated cut in output to raise prices while maintaining market shares. If information is incomplete, with the need for revenue paramount, the firm will likely increase output immediately before it can determine the cause of the downturn. Since every firm is in the same position, the result would be that everyone appears to defect: all increase output and prices fall to competitive levels regardless of the origins of the initial decline in prices. Most analyses of the JEC do show in fact that contracts broke down during demand shocks (Ellison 1994).
At the time the JEC was attempting with difficulty to form binding collusive agreements, the five leading railroad express firms had created a cartel that was at the time finishing its second decade under essentially the same cartel contract (Grossman 1996). Interestingly, it was in approximately the same kind of business as the railroads. The express companies were movers of freight, although the express specialized in carrying financial instruments and small packages rather than the bulk freight the rails carried. However the express had little capital since it leased rail cars from the railroads.
Grossman (2004) notes a number of differences between the cartels that might account for the variation in success rates. But one that is especially striking is the fact that all express firms were net creditors. By the 1880s, all five of the leading express firms had accumulated large portfolios composed primarily of railroad debt securities. (4) American Express and Adams, the two largest express companies had portfolios worth over $20 million (Grossman 2000). Of course the express companies like the JEC railroads faced the same general economic currents. But unlike the railroads, no express firm went into receivership during the entire life of the cartel, which extended from the late 1860s into the early years of the twentieth century.
What might this have meant with respect to strategic behavior? Consider again the same scenario as the one facing JEC firms: a decline in prices that could be either a defection or a demand shock. But an express company, with a base of revenue coming from the debt that it owns, can wait in period one to learn the causes of the fall in prices and if it is due to a general demand shock it will cut output so that collusive shares can be maintained. And in fact in the history of the express, no downturn in the economy led to widespread defection from cartel agreements (Grossman 1996).
Oligopolistic Games and Net Credit Position
The very different responses to demand shocks would be predicted by a small modification in a standard model of oligopolistic supergames often used to illustrate cartel strategy (Friedman 1971; Shapiro 1989). The basic model assumes that the players are in an industry of n firms, (n = 1, 2 ....) and they choose whether or not to collude as a quantity-setting cartel. Initially, we impose no debt constraints on the firms; they will choose to collude so long as the value of collusion exceeds the value of defection, where [q.sub.i](t) = output of firm i in period t, and [[pi].sub.i] [[q.sub.i](t), [q.sub.2](t), ..., [q.sub.i](t), ... [q.sub.n] (t)] = profit of firm i in period t. That is, profit for firm i depends on the output of the other firms in the industry.
The payoff to firm i for entire game is the present value (Vi) of profit:
[V.sub.i] = [[infinity].summation over (t=1)] [[delta].sup.t-1] [[pi].sub.i] [[q.sub.l] (t), [q.sub.2] (t), ..., ([q.sub.i] (t), ..., [q.sub.n](t)] (1)
where 0 [less than or equal to] [delta] 1 / 1 + r [less than or equal to] 1 is the discount factor.
When firms collude, the one-period profit for firm i is [[pi].sup.*.sub.i] , while [[pi].sup.c.sub.i] denotes the one-period profit for firm i in the event the cartel breaks down and all firms revert to a Cournot game. A grim trigger strategy is assumed; all set the collusive output each period as long as all other firms have done the same, but all revert to the Cournot output if any one firm deviates. Let [[pi].sup.r.sub.i] denote firm i's one-period profit if it optimally deviates. Thus, the payoff to firm i for the entire game if it deviates is:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (2)
the sum of the profit today and the discounted value of Cournot profits in every period thereafter.
In contrast, the payoff from continuing to cooperate is:
[V.sup.*.sub.i] = [[pi].sup.*.sub.i] / 1 - [delta] (3)
which is the present value of collusive profits forever. Of course, firm i will not deviate if:
[V.sup.*.sub.i] [greater than or equal to] [V.sup.r.sub.i], or (4)
[[pi].sup.*.sub.i] / 1 = [delta] [greater than or equal to] [[pi].sup.r.sub.i] + [delta] [[pi].sup.c.sub.i] / 1 - [delta], or : (5)
[delta] [greater than or equal to] [[pi].sup.r.sub.i] - [[pi].sup.*.sub.i] / [[pi].sup.r.sub.i] - [[pi].sup.c.sub.i] = [bar.[delta]] (6)
This shows that if the discount factor is sufficiently great, collusion will be sustained. The value of the discount factor is the ratio of the gain today of reneging and the loss tomorrow of reversion back to Coumot (Shapiro 1989). Now we modify the model to include uncertainty and variation in payoffs to debtors and creditors. For simplicity, we consider two collusive duopolies in which there is uncertainty in a one-period game. The first consists of firms that are net creditors and, the second, net debtors. For the second group, if output of either firm falls substantially, revenue is insufficient to meet fixed charges, and bankruptcy is assumed to occur. Thus, the primary difference between the two cartels is the single-period penalty associated with not defecting. This penalty is not captured in (1)-(6), but can be illustrated in the following example.
FEED