Abstract
In dynamic mechanism design, researchers typically assume either a seller with no commitment power or one with the absolute power to commit to future threats. This paper introduces the "positive commitment" framework, a realistic middle ground where sellers can legally promise future goods but cannot credibly commit to ceasing negotiations if an offer is rejected. We characterize the Perfect Bayesian Equilibrium of this game, revealing that positive commitment leads to a socially efficient market structure. The equilibrium naturally bifurcates into two channels—real-time spot markets and advance contracts—providing a theoretical foundation for the dual-channel structures observed in industries like online advertising.
Introduction
Traditional models of dynamic sales often oscillate between two extremes of seller power. In the No Commitment scenario, sellers set prices period-by-period; if a buyer refuses, the seller simply tries again, often leaving significant revenue on the table due to the buyer's strategic waiting. Conversely, the Full Commitment model allows a seller to make "take-it-or-leave-it" offers for an entire stream of goods, backed by the threat of never selling to that buyer again if the offer is rejected. While mathematically elegant, the latter is often practically impossible, as sellers are rarely incentivized to ignore willing buyers indefinitely.
This research addresses the gap by proposing Positive Commitment. In this framework, a seller can make binding, legally enforceable promises to deliver future goods (positive commitment) but lacks the ability to commit to not making future offers (lack of negative commitment). This model captures the reality of many modern B2B markets where contracts are enforceable, but the "door is always open" for future business.
The Model and Methodology
The authors model the interaction as a dynamic game between a single seller and a single buyer over $T$ discrete periods. In each period, a new heterogeneous product arrives, and the buyer learns their private valuation for it. These valuations are assumed to be independent and identically distributed (i.i.d.) over time.
To solve this game, the authors characterize the Perfect Bayesian Equilibrium (PBE). Because the standard Revelation Principle—which simplifies mechanism design by assuming buyers truthfully report their types—does not fully apply in settings with limited commitment, the authors develop a modified version. They utilize backwards induction to demonstrate how the buyer’s "outside option" (the utility they expect to receive if they reject current offers and wait for future spot market opportunities) dictates the maximum price the seller can charge for an advance contract.
Key Insights: The Dual-Channel Market
The core discovery of the paper is the emergence of a dual-channel market structure. The equilibrium naturally splits into:
- The Real-Time Channel: Selling the current item at the standard monopoly price ($p^*$).
- The Contract Channel: Selling future items at a discounted "positive commitment" price ($\hat{p}$).
The Pricing Formula
The authors derive a specific price for advance contracts: $\hat{p} = \mu - Z$. Here, $\mu$ represents the expected value of an item, and $Z$ is the utility (information rent) a buyer would expect to extract under standard monopoly pricing in the spot market. Essentially, the buyer will only accept an advance contract if it is cheaper than the utility they could "claw back" by refusing the contract and playing the real-time market.
Results and Social Efficiency
Perhaps the most striking result of the positive commitment framework is the achievement of 100% social efficiency. In standard monopoly settings, efficiency is usually lost because the seller sets high prices that exclude low-value buyers. However, under positive commitment, the seller finds it more profitable to sell every future item in advance at a discount than to risk the buyer waiting for the spot market. Consequently, every item is eventually sold to the buyer, maximizing social welfare.
Revenue Implications
The seller’s total revenue is characterized as $(\mu - Z) \times T$. While this is lower than what could be achieved under full commitment (where the seller can extract more surplus through threats), it is significantly higher than the revenue generated under no commitment. In a uniform distribution example ($[0, 1]$), the spot price is $1/2$ (yielding $1/4$ revenue), while the positive commitment price is $3/8$, representing a substantial boost for the seller.
Conclusion and Future Directions
This research provides a robust theoretical justification for the "fragmented" markets seen in the multi-billion dollar online advertising industry, where premium inventory is often sold via advance contracts while the remainder is liquidated in real-time auctions.
While the current model focuses on a single buyer with i.i.d. valuations, it opens several avenues for future work:
- Multi-Buyer Dynamics: Exploring how competition between multiple buyers affects the pricing of advance contracts.
- Persistent Information: Investigating scenarios where valuations are correlated over time, making the calculation of "information rent" more complex.
- Durable Goods: Applying the positive commitment framework to goods that do not expire, potentially offering a new perspective on the Coase Conjecture.
By bridging the gap between no commitment and full commitment, this paper offers a more nuanced and practically applicable toolset for modern market design.
Summary updated