Profitable collusion on costs: a spatial model (Q826655)

The authors present a model where profitable collusion on transport costs under delivered pricing arises for inelastic as well as elastic demand where -- in the latter case -- increased transport cost comes at the cost of reduced sales. The authors observe that \begin{itemize} \item the most profitable collusion happens when firms locate at the market peripheries, and it does the most damage to social welfare; \item the stability of collusion increases as the firm locations move toward each other but that there won't be collusion beyond a certain proximity; \item collusion on transport cost is strictly more stable than that on final price. \end{itemize} In the model, two firms engage in business in four steps. First, they decide whether or not to enter into a cartel: they choose collusion on transport cost only if the cartel is both more profitable than competing and the cartel is stable. Then firms choose locations: the first best (most profitable) location is selected for a cartel or price competition, depending on the outcome of the first step; the locations are assumed fixed, including when determining stability. In the third step, the firms set transport costs: the most profitable collusive unit transport cost or that associated with competition. Finally, the firms set price, identified with Bertrand spatial price discrimination. Backward induction is used to solve the model. The firms can choose location in the interval \([0,1]\) where customers are uniformly distributed. Transportation cost is linear in the distance, and costs per unit distance can be chosen from a bounded interval \([t_{\min},t_{\max}]\). Each consumer has either inelastic demand with reservation price which is sufficiently large to make it profitable to serve all customers, or -- in the elastic case -- demand is linear at each location. It is obtained that in absence of collusion, each firm adopts the minimum per unit distance transport cost \(t_{\min}\); the optimal firm locations are \(1/4\) and \(3/4\), which minimize total transport cost to serve the market; spatial price discrimination generates optimal social welfare. If the firms decide to enter into cartel, profit maximization leads to choosing the maximum permitted value \(t_{\max}\) of the unit transport cost, and the optimal firm locations are \(0\) and \(1\). It is also shown that the cartel is stable if and only if the discount factor exceeds \(t_{\max}/(t_{\min}+2t_{\max})\), and the cartel with collusion on transport cost is more stable -- that is, it can admit lower market discount rates -- than that for collusion on price; thus there exists a range of market discount rates in which a price cartel will not exist but a cost cartel would. The authors also analyze the case where firms are allowed to participate in more stable cartels even when they are less profitable than the optimal cartel locations at the endpoints, and determine optimal firm locations for the most stable cartel. The authors argue that the transport cost cartel could be less detectable by authorities, as under the cost cartel authorities would continue to observe market prices that reflect the rival's delivered cost and so it would present as typical Bertrand competition. The result that cost cartels are associated with firm locations toward the peripheries of the market could be applied in practice in the sense that extremely disparate locations either in physical or attribute space would hint to anti-trust authorities the existence of an agreement to increase transport costs/differentiation.