A Case for Competition in Information Provision

Technical Summary: A Case for Competition in Information Provision

Problem and Motivation
This paper investigates the welfare implications of competition in information provision when sources are biased and can misrepresent facts at a cost. While the "marketplace of ideas" intuition suggests that competition among opposing sources improves decision-making, the authors argue that this outcome is not immediate when claims are neither costless nor fully verifiable. In such environments, a second source can discipline an incumbent but may also introduce its own distortions (e.g., a persuasion rat-race). The central challenge for welfare comparison is equilibrium multiplicity: both monopolistic and competitive market structures admit a continuum of equilibria with vastly different informational properties. Comparing a favorable monopoly equilibrium with an unfavorable competitive one (or vice versa) yields indeterminate welfare rankings. The paper seeks to determine when competition improves the receiver's decisions and whether such improvements translate to higher total welfare.

Methodology and Model
The authors employ a costly communication game with a continuous state space $\theta \in \Theta$ and a receiver choosing between two binary actions, $a^-$ and $a^+$.

• Senders: One or two informed senders observe $\theta$ and report it. Sender 1 is upward-biased (prefers $a^+$), and Sender 2 (in the competitive setting) is downward-biased (prefers $a^-$).
• Costs: Reports are literal but not verifiable. Misreporting incurs a cost $k_j C_j(r_j, \theta)$ that increases with the distance between the report and the truth.
• Equilibrium Multiplicity:
  • Monopoly: A continuum of equilibria exists, indexed by the receiver's skepticism $\lambda$. These range from a least-informative equilibrium (substantial pooling of negative states) to a fully revealing equilibrium.
  • Competition: The authors focus on the "adversarial equilibrium" (characterized in Vaccari, 2023a), where senders mix strategies in an interior interval and report truthfully outside it.

Equilibrium Selection: The Skeptical-Envelope Criterion
To enable a coherent welfare comparison, the paper develops a common belief-based selection criterion, the skeptical-envelope criterion. This criterion applies uniformly to both market structures to select a single outcome in each:

1. Plausible Support: The receiver assigns probability zero to states from which observed reports could not have been sent rationally (within reach and direction).
2. Monotonicity: Higher reports must weakly (and strictly around the threshold) increase the attractiveness of action $a^+$.
3. Local Robustness (The Envelope): For reports occurring with positive probability (atoms), the receiver's belief must be the "least favorable" to the active sender's bias among neighboring interpretations.
   • Under Monopoly: This selects the least-informative equilibrium ($E_0$), where the pooling region is maximal. Standard refinements (like NWBR) fail to eliminate the more informative monopoly equilibria, but the skeptical-envelope condition rules them out by preventing "atomicity premiums" that would sustain higher information levels.
   • Under Competition: The criterion uniquely selects the adversarial equilibrium outcome. Because the report space is two-dimensional and the senders have opposed biases, the envelope conditions coincide, ruling out any equilibrium that does not satisfy the adversarial structure.

Key Results

• Receiver Welfare:

  • Competition does not dominate every monopoly equilibrium. If the receiver is sufficiently skeptical, a monopolist can provide near-full information, outperforming the adversarial equilibrium.
  • However, compared to the selected outcomes (least-informative monopoly vs. adversarial competition), competition improves receiver welfare under symmetric primitives.
  • More generally, competition benefits the receiver when the additional sender faces sufficiently high misreporting costs. As $k_2 \to \infty$, the mistakes caused by the second sender vanish, while its presence continues to constrain the incumbent's persuasion. Thus, for any fixed finite bias of Sender 2, sufficiently high costs make the competitive outcome strictly better for the receiver than the selected monopoly outcome.

• Total Welfare:

  • Improved decision accuracy does not necessarily increase total welfare (the sum of receiver and sender payoffs net of communication costs).
  • The receiver's preferred action ($a^+$ if $\theta \ge 0$) may differ from the action maximizing aggregate surplus. If the monopolist's bias is strong enough, the distortion induced by monopoly (persuading the receiver to choose $a^+$ in some negative states) can generate an aggregate action-payoff gain that exceeds the communication costs.
  • In such cases, the selected monopoly outcome yields higher total welfare than the selected competitive outcome, even though competition improves the receiver's decision accuracy. Conversely, if the receiver's and aggregate interests align on the distorted states, sufficiently disciplined competition improves both welfare criteria.

• Linear-Quadratic (LQ) Analysis:

  • Using an LQ specification, the authors derive comparative statics for the "swing function" (the report by Sender 2 that offsets Sender 1). They show that higher misreporting costs for Sender 2 raise the swing report (making it harder for Sender 1 to manipulate), while higher bias can have non-monotonic effects on the receiver's burden of proof.

Significance and Claims
The paper claims to provide a rigorous framework for comparing information markets by resolving the equilibrium selection problem through a unified belief-based discipline. Its primary contributions are:

1. Methodological: Establishing that standard refinements (like NWBR) are insufficient to select a unique monopoly equilibrium, necessitating a cross-environment selection criterion to make welfare comparisons meaningful.
2. Theoretical: Demonstrating that competition improves receiver welfare when the additional source is sufficiently constrained by misreporting costs, but that this informational gain is not a sufficient condition for social optimality.
3. Policy Implication: The value of competition depends jointly on the sources' incentives, the institutions that constrain manipulation (costs of lying), and the specific welfare criterion (receiver vs. total welfare). Plurality and accountability are complements; adding voices improves decisions only if those voices are sufficiently costly to manipulate, and even then, the outcome may not maximize aggregate surplus.

The authors explicitly limit their scope to a binary decision with two sources observing the same state, noting that extensions to more senders or heterogeneous information are natural directions for future research but not covered in this analysis. They do not claim that competition is universally beneficial or that bias is inherently valuable, but rather delineate the specific conditions under which competition disciplines persuasion and when it may fail to improve social welfare.