Regulating a Monopolist with Unknown Demand: Costly Public Funds and the Value of Private Information

Abstract

In this paper, we analyze the optimal regulation policy when the regulated firm has better information concerning the market demand than the regulator. We show that introducing a cost on public funds into the Planner's objective function does not lead to qualitative results similar to those obtained by introducing distributional considerations. In particular we show that under constant marginal cost the full information policy is not implementable and that the optimal regulatory policy results in informational rents. The social value of private information and the firm's informational rents are both increasing functions of the cost of the public funds.

Regulation of non-marketed outputs and substitutable inputs

“…6 Upwards distorted observable input levels coincide with upwards distorted prices for the inefficient type as shown in Laffont and Tirole (1993). They also agree with the results in a setting with unknown cost and demand functions as long as shadow costs of public funding are considered (Aguirre and Beitia (2004)). Noticeably, the case of prices below marginal costs, as found in Lewis and Sappington (1988b) and Armstrong (1999), is mainly always possible, while bunching of two types may be unavoidable in case of a very asymmetric distribution of costs or very flat isoquants.…”

“…In our paper, unlike Lewis and Sappington (1988b) and Armstrong (1999), we consider shadow costs of public funding instead of distributional welfare preferences. Despite technical differences, this is largely in line with the analysis of Aguirre and Beitia (2004). 4 However, in contrast to all these papers, we solve the two-dimensional adverse selection problem for a non-marketed good environment and a production process that involves two substitutable inputs with an uncertain isoquant and input factor costs.…”

“…It is well known that in a simple setting with two types of the firm, the efficient type is incentivized via a contract with first best (price) levels along with some positive rent, while the inefficient type's contract includes prices below the first best and no rent (e.g., Laffont and Tirole (1993)). This analysis has been extended to represent multiple dimensions of information asymmetry in terms of adverse selection, e.g., by Lewis and Sappington (1988b), Dana (1993), Armstrong (1999) or Aguirre and Beitia (2004). While Dana (1993) analyzes a multi-product environment, Lewis and Sappington (1988b), Armstrong (1999) and Aguirre and Beitia (2004) consider two-dimensional adverse selection with only one screening variable.…”

“…This analysis has been extended to represent multiple dimensions of information asymmetry in terms of adverse selection, e.g., by Lewis and Sappington (1988b), Dana (1993), Armstrong (1999) or Aguirre and Beitia (2004). While Dana (1993) analyzes a multi-product environment, Lewis and Sappington (1988b), Armstrong (1999) and Aguirre and Beitia (2004) consider two-dimensional adverse selection with only one screening variable. Specifically, the latter three derive optimal regulation strategies in a marketed-good environment (in the sense of Caillaud et al (1988)) with unknown cost and demand functions.…”

“…While the efficient type is always set to first best input levels, the other contracts' (observable) input levels are distorted upwards. 6 Separation of at least three types is 4 Aguirre and Beitia (2004) show the difference between shadow costs of public funding and distributional welfare preferences based on a model with continuous probability distribution, while we assume a discrete distribution. 5 Noticeably, with the (discrete) two-dimensional adverse selection problem, our problem setting is technically closest to the model discussed by Armstrong (1999).…”