📈 economics

Reputation, Risk, and Visibility

This paper demonstrates that increasing the visibility of failure outcomes, such as through Registered Reports, effectively counteracts the tendency of high-reputation experts to play it safe by weakening career concerns and relaxing selection criteria.

Original authors: Georgy Lukyanov

Published 2026-08-04

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Georgy Lukyanov

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Science of Reputation and the "File Drawer" Problem

Imagine a world where scientists, doctors, and engineers are like professional athletes. Their careers depend on how the public and their bosses perceive their talent. If they look like geniuses, they get more funding, better jobs, and fame. If they look like they made mistakes, their reputation takes a hit. This is the world of "reputation economics," a field that studies how the fear of looking bad changes the way experts make decisions.

In this world, experts often face a choice: play it safe with a boring, predictable experiment, or take a huge risk with a bold, new idea that might fail spectacularly. For decades, economists have wondered: Does the pressure to maintain a perfect reputation make experts too scared to try anything new? If a famous scientist tries a risky experiment and fails, does that failure disappear into a "file drawer" (hidden away so no one sees it), or does it stay on the record? The answer to that question changes everything about how we encourage innovation.

The Paper's Big Idea: Making Failures Visible

This paper, written by Georgy Lukyanov, tackles a specific question: Does the way we handle "failures" change whether experts take risks?

The author builds a mathematical model to simulate a scientist's career. In this model, an expert has a "reputation score" (how competent everyone thinks they are). They can choose a Safe path (like a boring, guaranteed experiment) or a Risky path (like a wild, new discovery that might fail). The twist is a new "policy knob" called Visibility.

Think of Visibility as a spotlight.

Low Visibility: If an expert takes a risk and fails, the spotlight turns off. The failure is hidden, shelved, or ignored. Only the successes get the spotlight.
High Visibility: If an expert takes a risk and fails, the spotlight stays on. The failure is recorded, published, and counted just like a success.

The paper asks: What happens to an expert's courage when we turn the Visibility knob up?

The Surprising Findings

The paper reveals two main things, and they might surprise you.

1. Fear of failure isn't the only reason experts play it safe.
Many people assume that experts avoid risks simply because they are terrified of looking stupid. The paper shows that this isn't always true. If the "Safe" option gives no new information (it's just a boring repeat) and failures are hidden just as often as successes, then a career-focused expert does not automatically become conservative just because they care about their reputation. In fact, if their career value grows in a straight line (meaning every bit of success is worth the same), they have no reason to be scared of risk at all. The "caution" we see in the real world often comes from something else: the fact that failures are hidden while successes are celebrated.

2. Making failures visible makes experts braver.
This is the paper's biggest discovery. When the author turns up the "Visibility" of failures (making sure that when a risky experiment fails, it is still seen and counted), something magical happens to high-reputation experts.

Before the reform: A famous scientist with a high reputation is very cautious. They stick to safe, boring projects because they are terrified that a single visible failure will ruin their perfect image.
After the reform: If the policy changes so that failures are always visible (like in a "Registered Report" system where the plan is approved before the experiment starts), that same famous scientist suddenly becomes more willing to take risks.

Why? Because when failures are guaranteed to be seen, the "punishment" for failing is no longer a secret shame; it's just a data point. The paper proves that increasing the visibility of failures actually weakens the fear that holds high-reputation experts back. It relaxes the "selection" process, meaning we see more risky projects from the best scientists, and the success rate of those risky projects might actually go down (because they are trying harder, riskier things).

The "Registered Report" Connection

The paper uses a real-world example called Registered Reports to explain this. In a standard science journal, a scientist runs an experiment. If it works, they publish it. If it fails, they might hide it. This creates a "file drawer" of hidden failures.

In a Registered Report, the scientist writes down their plan before they start. The journal accepts the plan based on its quality, not its results. Then, they run the experiment. If it fails, the journal still publishes it. The paper argues that this simple change—making sure the "failures" survive and are counted—is the magic lever. It flips the incentive. Instead of hiding to protect their reputation, high-reputation scientists feel safe enough to try the risky, groundbreaking work that moves science forward.

The Bottom Line

The paper doesn't just guess; it uses math to prove that if you want experts to take risks, you shouldn't just tell them to "be brave." Instead, you should change the rules so that failing is not a secret. When failures are visible and counted just like successes, the pressure to be perfect actually disappears, and the most famous experts become the most adventurous.

The author provides a clear "map" for testing this in the real world: if you see a field where journals start accepting "Registered Reports," you should see famous scientists suddenly trying more risky projects, and the overall success rate of those risky projects should drop slightly (because they are pushing the boundaries). The paper suggests that this simple change in visibility is a powerful tool to fix the "conservatism" that holds back scientific progress.

Technical Summary: Reputation, Risk, and Visibility

1. Problem Statement

The paper investigates the conditions under which reputational concerns (career concerns) induce experts to adopt conservative strategies, avoiding risky projects in favor of safe ones. While existing literature (e.g., Holmström, 1999; Scharfstein and Stein, 1990) establishes that career concerns can drive herding or caution, this paper isolates a specific, often overlooked policy lever: the visibility of realized outcomes.

The central question is: When does the visibility of failures versus successes alter an expert's incentive to take risks? The author focuses on the "survival rule" $\sigma(y, \pi)$ , which determines the probability that an outcome $y$ (success or failure) remains on the public record given the expert's current reputation $\pi$ . The paper specifically examines how reforms that increase the visibility of failures (such as Registered Reports in academia or clinical trial registries) impact reputational risk-taking.

2. Methodology

The paper employs a minimal, stylized two-period model with the following components:

Agents and Types: An expert of unknown type $\theta \in \{H, L\}$ (competent vs. incompetent) observes a private signal $s$ and recommends an action $a \in \{R, S\}$ (Risky vs. Safe).
Outcomes and Signals:
- Risky Action ( $R$ ): Yields a binary outcome $y \in \{1, 0\}$ (success/failure) with probabilities $p_H$ and $p_L$ ( $p_H > p_L$ ).
- Safe Action ( $S$ ): Yields a baseline outcome with probabilities $r_H$ and $r_L$ . The model allows for the safe action to be informative ( $r_H > r_L$ ) or uninformative ( $r_H = r_L$ ).
Visibility Rule ( $\sigma$ ): After an outcome is realized, it remains on the record with probability $\sigma(y, \pi)$ . This parameter captures whether failures are "shelved" (low visibility) or published/recorded (high visibility).
Objective: The expert is risk-neutral regarding flow payoffs but maximizes the continuation (career) value $V(\pi)$ , where $\pi$ is the public belief that the expert is competent.
Learning Mechanism: The model utilizes Bayesian updating with Monotone Likelihood Ratio Property (MLRP). The expert's strategy is characterized by a cutoff $s^*(\pi)$ in the private signal space.
Analytical Tools: The paper derives a necessary-and-sufficient local sign test for the slope of the risky cutoff ( $ds^*/d\pi$ ). This test relies on the curvature of the career value function $V(\pi)$ and the "jumps" in posterior beliefs weighted by visibility.

3. Key Contributions

The paper makes three primary theoretical contributions:

No-Free-Conservatism Result: It demonstrates that career concerns alone do not generically generate conservatism. If the safe action is uninformative and visibility is symmetric (successes and failures are treated equally), reputational conservatism only arises if the career value function $V(\pi)$ is locally concave. If $V(\pi)$ is linear or convex, there is no reputational incentive to avoid risk.
Visibility Reform Mechanism: It isolates the effect of increasing failure visibility (holding success visibility constant or rising less). The paper proves that increasing the survival rate of failures weakens high-reputation caution. This occurs because making failures visible reduces the "cost" of a failed risky attempt for high-reputation experts, thereby relaxing the selection of risky projects.
General Local Sign Test: The author provides a general condition (Theorem 9) that determines when reputational conservatism arises. This test is agnostic about the curvature of $V(\pi)$ and accommodates informative safe actions. It decomposes the incentives into weighted posterior jumps for both risky and safe branches ( $\Psi(\pi)$ and $\Phi(\pi)$ ).

4. Main Results

Result I: The Baseline (No-Free-Conservatism)

Under the assumption that the safe action is uninformative ( $r_H = r_L$ ) and visibility is symmetric ( $\sigma(1, \pi) = \sigma(0, \pi)$ ):

If $V(\pi)$ is locally linear, the risky cutoff is flat ( $ds^*/d\pi = 0$ ); there is no conservatism.
If $V(\pi)$ is locally convex, reputation encourages risk.
If $V(\pi)$ is locally concave, reputation discourages risk.
Implication: Without asymmetric visibility or informative safe options, "career concerns" do not inherently create a bias toward safety.

Result II: The Visibility Reform

Consider a policy reform that increases the visibility of failures ( $\sigma(0, \pi)$ ) while keeping success visibility fixed (or increasing it less).

Effect on High Reputation: For experts with high $\pi$ , increasing failure visibility strictly increases the slope of the risk premium ( $\Delta'(\pi)$ ).
Outcome: This weakens the caution of high-reputation experts. They become more willing to choose risky projects.
Selection Effect: As high-reputation experts take more risks, the pool of risky projects expands to include marginal projects. Consequently, the conditional success rate of risky projects falls (selection relaxes).

Result III: Identification and Empirical Predictions

The paper maps these theoretical results to testable empirical predictions, specifically in the context of Registered Reports (RR):

P1 (Risk Taking): Following the adoption of RR (which increases failure visibility), the share of risky designs should rise more for high-reputation authors compared to low-reputation authors.
P2 (Selection): Conditional on choosing a risky design, the success rate should fall for high-reputation authors post-reform (as selection relaxes).
P3 (Null Survival): The survival rate of null results should increase, particularly in fields where the "safe" option was previously uninformative.

The paper provides a "staggered adoption" difference-in-differences framework to test these signs without needing to estimate the curvature of $V(\pi)$ .

5. Significance and Claims

The paper claims to shift the policy focus from traditional incentives (bonuses, transfers, committee rules) to the design of outcome visibility.

Policy Relevance: It argues that modest changes to which outcomes persist (e.g., mandating the publication of null results via Registered Reports or clinical trial registries) can correct reputational conservatism without heavy institutional machinery.
Mechanism Isolation: By separating visibility from other incentives, the paper clarifies that transparency reforms can have non-monotonic effects depending on whether they primarily boost the visibility of failures or successes.
Generality: The results are presented as applicable beyond academia, with a specific extension to security-incident disclosure, where regulatory mandates (increasing the floor for failure reporting) can similarly reduce corporate risk aversion.

The author remains modest about the scope, noting that the results are local comparative statics and that the model abstracts from fully dynamic careers and endogenous disclosure strategies, though it provides a framework to analyze these extensions. The core claim is that visibility is a portable policy lever that can flip reputational incentives from caution to experimentation.