Pricing and Capacity Contracts in Service Supply Chains

Services, like manufactured products, are often jointly provided by multiple complementary providers that form a supply chain. A good example is the provision of communication services over the Internet, where traffic typically travels over the networks of multiple interconnected Internet Service Providers (ISPs). This decentralized structure leads to complex competitive relationships among providers. In this paper, we study the interaction between (i) market structure, (ii) pricing structure and (iii) market outcome in settings where a series of capacity-constrained providers offer complementary services to delay-sensitive customers over a congestible network. We first consider what we call a Resource Pricing (RP) structure whereby each provider sets prices and the capacity level for its own resources. We find for the base case of cross-traffic only, that providers charge the same price in equilibrium, regardless of their relative capacity costs, and get the same revenue. The total equilibrium price is increasing -- and the quantity, capacity, aggregate industry revenue and consumer surplus are decreasing -- in the number of providers. These distortions under multiple providers are the result of multiple marginalizations, as observed in complementary monopoly settings, but are significantly amplified by congestion effects. The RP structure with both cross-traffic and local-traffic and two providers reveals interesting interaction effects between the cross-traffic market and the local monopoly markets. The monopoly chooses not to price discriminate between local- and cross-traffic. In the case of two providers, each typically does price discriminate between local- and cross-traffic. The total cross-traffic price is typically - but not always - higher than the sum of the local traffic charges, and higher than its marginal delay externality. The extent of price discrimination under two providers depends on the relative magnitude of local-traffic to cross-traffic delay-sensitivity and (naturally) on the demand elasticity parameter. Under fixed capacity, the lower cross-traffic subsidizes the local market by freeing up capacity, compared to the monopoly solution. In these cases, the demand rate is higher in local markets, leading to higher consumer surplus and revenue, while the opposite holds true in the cross-traffic market. Under variable capacity, each provider charges the monopoly price for the local-traffic, but a higher cross-traffic price than the monopoly, due to double-marginalization. However, the ``double mark-up'' of cross-traffic price relative to the monopoly case is reduced, compared to the case in which there is no local traffic. Numerical examples suggest that the percentage profit and welfare loss due to double-marginalization can be significantly reduced by these local-to-cross traffic interaction effects, and that these effects increase in significance the more delay-sensitive the local-traffic relative to the cross-traffic. We next consider Profit Sharing (PS) structures. In contrast to the RP structure, each provider is solely responsible for setting the retail prices for the traffic-markets whose transmissions originate on her network resource. Providers agree to share the profits (aggregate revenues and capacity costs) according to some predetermined fraction. We show that such a PS structure results in the monopoly solution. We analyze Peering Contracts, which are currently quite prevalent among major ISPs. We show that in the presence of congestion effects, peering contracts generally lead to sub-optimal industry performance, compared to the monopoly solution, and that they generally perform well only in the absence of congestion effects. We analyze the bargaining issues pertaining to the design of this PS scheme in our setting with multiple traffic markets and congestion-induced externalities among markets.