The Winner's Bliss in Common-Value Auctions under Horizontal Differentiation

Technical Summary: The Winner's Bliss in Common-Value Auctions under Horizontal Differentiation

Problem Statement
Standard auction theory posits that in common-value auctions, winning is "bad news" for the winner. The "winner's curse" arises because winning implies that opponents held less optimistic signals, forcing the winner to revise their valuation downward. While recent literature has identified conditions for a "winner's blessing" (e.g., revenue-optimal mechanisms or uncertain market thickness), these require specific environmental features. This paper investigates a distinct channel: horizontal preference heterogeneity. The authors ask whether, in a standard first-price auction with common values but horizontally differentiated preferences, winning can instead convey "good news" about the object's value, creating a phenomenon they term the "winner's bliss."

Methodology
The authors construct a tractable two-bidder, first-price auction model to characterize equilibrium behavior analytically.

• Environment: An object has an unknown state $\theta \in \{A, B\}$ with equal prior probability. Two risk-neutral bidders possess private types $t_i \in \{A, B\}$. A bidder derives utility 1 if the state matches their type ($\theta = t_i$) and 0 otherwise.
• Preference Alignment: The degree of horizontal differentiation is parameterized by $\mu \in [0, 1]$, representing the probability that bidders share the same type ($Pr(t_1 = t_2) = \mu$).
  • $\mu = 0$: Pure horizontal differentiation (opposing preferences).
  • $\mu = 1$: Aligned preferences (standard common-value setting).
• Signals: Bidders observe independent signals $S_i \sim U[0, 1]$. The posterior probability of the state depends on the sum of signals: $Pr(\theta = A | S_i, S_j) = (S_i + S_j)/2$.
• Analysis: The authors derive a symmetric Bayes-Nash equilibrium in monotone bidding strategies and analyze the informational content of winning, expected revenue, and bilateral trade outcomes under varying levels of $\mu$.

Key Contributions and Results

1. The Winner's Bliss Phenomenon
The paper establishes that the informational content of winning depends critically on the degree of preference alignment ($\mu$).

• Mechanism: In a setting with opposing preferences ($\mu < 1/2$), winning reveals that the opponent likely held a signal indicating the state favors the opponent's type. Since the opponent's type is the winner's opposite, this implies the state is actually favorable to the winner.
• Proposition 2: The authors prove that for $\mu < 1/2$, the expected value conditional on winning exceeds the ex-ante expected value ($E[v | s, win] > E[v | s]$), constituting winner's bliss. Conversely, for $\mu > 1/2$, the conventional winner's curse holds. The threshold $\mu = 1/2$ marks the point where the "opposing-type" channel (winning confirms the state) exactly offsets the "same-type" channel (winning implies a weaker opponent signal).

2. Revenue and Information Disclosure
The study analyzes how information disclosure affects seller revenue, challenging the standard linkage principle in specific contexts.

• Revenue Monotonicity: Expected revenue is strictly decreasing in preference alignment ($R(\mu) = 1/2 - \mu/6$). Revenue is maximized when preferences are fully opposed ($\mu=0$) and minimized when aligned ($\mu=1$).
• Disclosure Effect (Proposition 4): The authors analyze a scenario where a platform reveals the state $\theta$ with probability $q$.
  • When preferences are sufficiently opposed ($\mu < 3/4$), revealing the state reduces revenue. This is driven by the "allocation effect" (Board, 2009): revealing the state softens competition because bidders realize they may not value the object.
  • When preferences are sufficiently aligned ($\mu > 3/4$), revealing the state increases revenue, consistent with the linkage principle.
  • Notably, the threshold for the revenue effect ($\mu = 3/4$) differs from the threshold for the winner's bliss effect ($\mu = 1/2$).

3. Advantageous Selection in Bilateral Trade
The model is applied to a bilateral trade setting (buyer type A, seller type B) to examine whether information asymmetry impedes or facilitates exchange.

• Advantageous Selection (Proposition 5): Under horizontal differentiation, the counterparty's willingness to trade acts as "good news." If a buyer (type A) sees a seller (type B) willing to sell, it implies the seller's signal suggests the state is favorable to the buyer.
• Trade Sustainability: Unlike standard adverse selection models (e.g., Milgrom and Stokey, 1982) where trade often collapses, this model sustains exchange. At a price of $p=1/2$, the unique equilibrium yields a trade probability of $4/9$ and a realized surplus efficiency of 88.9% (8/9) relative to the first-best.
• Feedback Loop: The willingness to trade creates a positive feedback loop, allowing trade to persist even under asymmetric information.

Significance and Claims
The paper claims to identify a fundamental channel—horizontal preference heterogeneity—that reverses the standard selection effects in common-value environments.

• Theoretical Shift: It demonstrates that the "winner's bliss" and "advantageous selection" can arise in standard auction and trade settings without requiring complex mechanism design or uncertainty about market thickness.
• Implications: The results suggest that in markets with heterogeneous preferences (e.g., online advertising where advertisers target different demographics), information disclosure may be counterproductive for revenue, and trade may be more robust than traditional adverse selection models predict.
• Scope: The authors note that while their specific parameterization ensures analytical tractability, the intuition extends to general settings. They cite companion experimental work (Nguyen et al., 2023) as providing initial laboratory evidence, but the paper itself focuses on the theoretical characterization of the equilibrium and its comparative statics.