Abstract
This paper reconciles a long-standing theoretical divide in durable goods markets by demonstrating that a monopolist's power is strictly limited when facing a finite deadline. While previous theories suggested a monopolist might either be powerless (the Coase Conjecture) or "all-powerful" (the Pacman strategy), this research proves that with discrete buyers and a finite time horizon, durable monopoly profits are at most twice those of a standard static monopolist.
The Durable Good Monopoly Problem
The "Durable Good Monopoly" problem is a classic puzzle in economics. In 1972, Ronald Coase conjectured that a monopolist selling a long-lasting good (like a car or land) would eventually lose all market power. The logic suggests that if consumers know the price will drop later to attract lower-value buyers, they will wait, forcing the monopolist to drop prices immediately to the competitive level.
However, in 1989, Bagnoli et al. presented a counter-intuitive result: if there are a finite number of "atomic" (individual) consumers and an infinite time horizon, the monopolist can extract all consumer surplus using a "Pacman" strategy—systematically selling to buyers one by one at their maximum possible valuations. This paper addresses the gap between these two extremes, asking whether the deadline inherent in real-world markets restores balance to the seller-consumer dynamic.
Methodology: Modeling Discrete Consumers and Finite Horizons
The authors utilize game theory to model the interaction between a single seller and $N$ discrete consumers over $T$ time periods, focusing on finding Subgame Perfect Nash Equilibria (SPNE). Key methodological steps include:
- Backward Induction: Reasoning from the final period $T$, where the seller acts like a standard static monopolist, back to the first period.
- Threat Prices: Defining a "threat price" for each consumer—the price the consumer would be offered in the next period if they refuse to buy now. This concept is central to determining the maximum price a seller can extract in the current period.
- Skimming Property: Characterizing equilibria where high-value consumers buy before low-value ones.
- Recursive Profit Modeling: Developing a recursive relationship to calculate the maximum revenue a seller can obtain across multiple periods.
Key Insights: The Power of the Deadline
The research reveals that the "Pacman" result is driven by the assumption of infinite time rather than the discreteness of consumers. In a finite world, the monopolist's power is significantly curtailed by the deadline. In the final period, the monopolist is forced to set a single "static monopoly price." This creates a floor for consumer expectations that ripples backward through time, preventing the seller from perfectly price-discriminating against every individual.
Furthermore, the authors find that for the "Pacman" strategy to be an equilibrium in a finite setting, consumer valuations must decrease exponentially—a condition that is unlikely in most practical scenarios.
Results and Profit Bounds
The study establishes several key results regarding market power and profit:
- Additive Bound: The authors find a remarkably simple bound: the durable monopoly profit is at most the static monopoly profit plus the static monopoly price ($Π^D \leq Π^M + p_1$).
- Global Profit Bound: The main result proves that durable monopoly profits never exceed twice the static monopoly profits ($Π^D \leq 2 \cdot Π^M$). This bound is shown to be "tight," meaning examples exist that nearly reach this limit but do not exceed it.
- Equilibrium Dynamics: While "non-skimming" equilibria (where low-value buyers purchase before high-value ones) can exist, the authors prove that in a two-period game, these never yield more profit than standard skimming equilibria.
- Incomplete Information: The bounds remain robust even if the monopolist does not know which specific consumer has which valuation, provided the overall distribution is known.
Limitations and Future Work
The current model primarily assumes no discount factor, focusing on the undiscounted case. While the authors suggest discount factors can be integrated, the core proofs remain centered on immediate value. Additionally, the proof regarding the optimality of skimming equilibria is currently limited to a two-period setting; extending this to $T$ periods remains a conjecture. Finally, more complex models involving "hidden" information—where the seller cannot observe the total number of sales each period—represent an open area for future research.
Summary updated