In many applications, agents can more easily influence how data are generated than manipulate the data themselves. For example, students choose which classes to take but cannot modify their grades once assigned, and firms determine how performance indicators are calculated rather than directly altering results. This paper experimentally studies information transmission when data are generated through an unobservable, strategically chosen process. We focus on settings where an informed sender—such as a student or a firm—privately selects a data-generating process (DGP)—a portfolio of classes or an accounting methodology—to shape the beliefs of an uninformed receiver, such as an employer or an investor. Across treatments, we vary which DGPs are feasible and whether some come at a cost. These variations span different levels of information verifiability and allow us to capture, within a unified framework, the core insights of disclosure, cheap-talk, and signaling models. Our findings reveal three main patterns. First, while senders select their DGPs strategically in line with theoretical predictions, receivers often fail to account for this strategic selection. Second, introducing differential costs across feasible DGPs mitigates this DGP selection neglect. Third, while receivers’ biases are robust across environments, their consequences vary: information transmission falters most when the evidence receivers observe is highly sensitive to the sender’s DGP choices. Finally, increasing transparency about the selected DGP—a policy often advocated in practice—does not, in general, improve information transmission.

We conduct an experimental analysis of selective disclosure in communication. In our model, an informed sender aims to influence a receiver by disclosing verifiable evidence that is selected from a larger pool of available evidence. Our experimental design leverages the rich comparative statics predictions of this model, enabling a systematic test of the relative importance of evidence selection versus evidence concealment in communication. Our findings confirm the key qualitative predictions of the theory, suggesting that selection, rather than concealment, is often the dominant distortion in communication. We also identify deviations from the theory: A minority of senders overcommunicate relative to predictions, while some receivers partially neglect the selective nature of the evidence they observe.

This paper studies the effect of mandated disclosure policies when firms can use unnecessary complexity to withhold relevant information from consumers endowed with a private outside option. When the cost that consumers have to pay to solve the complexity is too high, mandated disclosure policies are completely ineffective and firms are de facto allowed to decide which information is disclosed and which one is censored. In contrast, when consumers can afford to pay the cost that enable them to understand the complex information, the effectiveness of the policy depends on the distribution of their outside option, along with the level of the cost. This paper derives simple conditions for consumers' optimal strategy and exploits them to define an algorithm that identifies the optimal disclosure rule under general distributional assumption. These findings can inform regulators about the heterogeneous efficacy of mandated disclosure policies in different contexts.

Economic agents often rely on communication to overcome information asymmetries. When information is verifiable, the unraveling principle predicts full disclosure, as senders should reveal all private information. Yet, experimental evidence shows that with a single relevant dimension, senders frequently withhold unfavorable information, and receivers remain overly optimistic about what is left undisclosed. Many real-world settings, however, involve multiple attributes of interest to receivers. We study how such multi-dimensionality affects disclosure choices and inference. In a controlled laboratory experiment, we show that increasing the number of attributes the sender can disclose from one to two increases the disclosure of unfavorable information. Belief elicitation reveals that this pattern is driven by senders’ expectations: they anticipate greater skepticism about undisclosed dimensions, particularly when disclosing only one. Receivers, in contrast, do not become more skeptical, failing to recognize the strategic link between disclosure choices across attributes.

In many asymmetric information settings, the sender's ability to credibly commit to an information structure is crucial for effective communication. However, the extent to which such commitment is possible often depends on institutional or strategic constraints shaped by the participants themselves. In such cases, commitment power becomes endogenous and must be explicitly modeled. This paper theoretically and experimentally studies a communication game with unverifiable information in which the sender chooses the level of commitment power—that is, the probability they will be unable to revise their information structure after committing to it. We introduce uncertainty over the sender’s incentives and examine how the sender’s optimal commitment level depends on the alignment between their preferences and those of the receiver. Theoretically, we show that as preferences become more aligned, the sender optimally demands less commitment power. This choice requires strategic sophistication, particularly an understanding of the strategic value of commitment in the interaction. Experimentally, however, we find the opposite: senders demand more commitment as alignment increases, suggesting systematic departures from equilibrium reasoning in the use of commitment.

Firms disclose proprietary information to investors through financial statements, shareholder reports, and other public communications that are often excessively complex and technical. This paper develops a theory of the endogenous choice of disclosure complexity that arises from the strategic interaction between firms and heterogeneously sophisticated investors.  Disclosure complexity plays a dual role. Greater complexity increases informativeness, as a firm’s proprietary information—specialized, disaggregated, and multidimensional—often requires a sophisticated and methodologically rigorous presentation to be credibly conveyed. At the same time, higher complexity raises cognitive costs, making disclosures more demanding for investors to process and interpret. The paper presents a two-period model. In the first period, a level of disclosure complexity is implemented, specifying analytical and descriptive dimensions of complexity that disclosure documents must satisfy. In the second period, the firm privately observes its type and decides whether to report transparently or to obfuscate. Even when transparent, a disclosure may remain inconclusive due to the natural constraint of communication—the inherent limits of conveying specialized and disaggregated information—whereas obfuscated disclosures remain inconclusive through deliberate concealment. The analysis characterizes equilibrium disclosure, investor engagement, and welfare under firm- and regulator-determined protocols. A welfare-maximizing regulator selects an under-complex design, while firms may over-complexify to exploit unsophisticated investors. Disclosure complexity thus governs the tension between informativeness and strategic concealment.

This paper studies how behavioral failures in processing strategic information shape both information acquisition and disclosure. We develop a communication model in which cursed receivers systematically underappreciate the strategic link between a sender's message and the underlying state, treating communication as less informative than it truly is. In the model, a continuum of receivers, heterogeneous in their degree of cursedness, decides whether to acquire information from a strategic sender or from an outside source. The sender, who holds verifiable but selectively disclosable information, chooses an omission threshold, revealing favorable states and hiding unfavorable ones. Under full rationality, disclosure fully unravels, and receivers always learn the truth. When receivers are partially cursed, however, omissions are misinterpreted, enabling the sender to extract rents and sustain non-disclosure in equilibrium. Cursedness not only distorts inference but also affects acquisition: cursed receivers may lower their own welfare by relying on inferior outside information sources. We characterize the unique equilibrium and show that more cursed receiver populations or worse outside sources both expand the scope of strategic concealment. The framework provides new behavioral microfoundations for incomplete transparency and motivates laboratory tests on how errors in strategic inference shape the joint production and acquisition of information.