Abstract: This paper develops a model of serial experimentation where an entrepreneur sequentially experiments with different ideas until one of them succeeds. The entrepreneur starts with one idea, and new ideas arrive over time, but switching from the current idea to a new idea is irreversible and requires funding by an investor. A more creative entrepreneur both generates ideas faster and creates more profits when an idea succeeds. Since creativity is not directly observable, investors infer creativity from an entrepreneur’s timing of switching, creating an incentive for the entrepreneur to distort their behavior to influence these perceptions. We show that equilibrium distortions depend not only on market conditions but also on the technological environment, in particular, the abundance of creativity in the underlying technology. When creativity is abundant and new ideas are easy to generate, both types distort toward faster switching, much as in standard signaling models. When creativity is scarce, however, the more creative type distorts toward slower switching while the less creative type distorts toward faster switching, causing the two types to move in opposite directions. Founder-specificity provides another technological dimension that influences the direction of distortion. When projects are founder-specific, switching generates an abandonment risk: although more creative entrepreneurs produce more profitable ideas, they also switch more frequently, increasing the chance that projects are left unfinished. This creates a trade-off that may lead investors to prefer less creative entrepreneurs. We show that investors’ preference for creativity under observable creativity is preserved when creativity becomes unobservable, and we characterize the resulting equilibrium distortions when creativity is either preferred or not in this environment.

Abstract: This paper examines the strategic role of transparency in a 'winner-takes-all' innovation race, where two competing firms race to achieve three consecutive breakthroughs in order to obtain the innovation. At each stage of the race, firms decide on their level of transparency, which affects the likelihood of their breakthroughs leaking to competitors. Our model reveals that full transparency can emerge as an equilibrium strategy during the early stages of the race, but also that transparency must decline as technological leads extend. If transparency enhances productivity, then a firm’s openness gradually decreases as its lead grows, while if transparency solely affects information leakage, the firm’s strategy becomes ``bang-bang'' — either fully transparent or fully opaque. These findings offer insights into real-world dynamics such as the evolving transparency strategies observed in the race for artificial general intelligence, where early transparency has given way to increasing secrecy.

Abstract: This paper studies a dynamic principal-agent problem in which effort is observable, but failure after high effort is especially informative about the agent's underlying suitability. Because poor outcomes following visible effort lead observers to update more sharply against the agent, the threat of replacement can weaken rather than strengthen incentives. I show that replacement risk causes the agent to start shirking earlier, that is, at a higher reputation level than in the benchmark without replacement concerns. I then study how the principal should respond when the quality of the replacement option is uncertain. When the principal can commit to a deadline but cannot credibly disclose information about the realized alternative, deadline commitment can dominate no commitment when replacement is sufficiently costly for the agent. When commitment is infeasible but disclosure is possible, the optimal disclosure policy depends on the agent's sophistication: selective disclosure is optimal with a naive agent, whereas full disclosure is optimal with a sophisticated agent. More broadly, the paper shows that observable effort does not eliminate informational distortions; rather, when failure after high effort is especially revealing, it can generate a new form of moral hazard, namely signal jamming through shirking.

Abstract: As a companion paper to Kim and Tomaino (2025), this project develops a screening model in which investors design contract menus—consisting of pivot flexibility and profit shares—to screen entrepreneurs by their creativity. More creative entrepreneurs value flexibility and therefore select contracts that grant greater discretion to switch to new ideas in exchange for lower profit shares, while less creative types prefer contracts offering higher shares but tighter constraints. The model characterizes the optimal stationary menu and examines how technological conditions, such as the creative composition of new entrepreneurial inflows, shape contract design.

Abstract: This project examines how differences in information quality between public and private funders affect entrepreneurial experimentation. Public agencies have to rely on noisier performance signals, which can reduce the perceived cost of failure and encourage bolder innovation, while private investors monitor more closely and may discourage risk-taking. Using U.S. patent data linked with public grants records, the project studies when informational noise becomes beneficial rather than costly.