People Take More Risks When Consequences Come Later

On August 2, 2026, new European Union transparency obligations for certain artificial intelligence systems took effect. The rules concern direct interactions with AI and several forms of AI-generated or manipulated content. They offer a timely example of a familiar decision problem: choices are made today while their consequences materialize later. Research by Emmanuel Kemel of HEC Paris and Corina Paraschiv of Université Paris Cité shows that this delay can influence the risk people are willing to take.

In a computer-based laboratory experiment involving 70 University of Paris undergraduates, participants chose between monetary lotteries with consequences received immediately or one year later. They accepted more risk when the consequences were delayed. The study then separates the source of that increased risk tolerance. It traces the change mainly to how participants weighted probabilities and, in the researchers' interpretation, to greater optimism about the chances of success.

A one-year delay changes risk tolerance

Most economic models of risky choice begin with an immediate sequence: a decision is made, uncertainty is resolved and the consequence follows. Daily life often separates those moments. Voters know an election result before its policies take effect. A job or education program may begin months after an offer. A sanction may follow risky behavior long after the event. Medical decisions can affect health years later.

Earlier experiments had already found that delayed lotteries could increase risk tolerance. Many of them delayed both the moment when uncertainty was resolved and the moment when payment arrived. The two forms of waiting were therefore combined. A decision-maker may respond differently to waiting for the result, waiting for the consequence, or experiencing both delays together.

Kemel and Paraschiv designed their experiment to focus on the materialization of the consequence. The timing of risk resolution remained constant while the timing of payment changed. Their design also controlled for discounting, the tendency to value a future payment differently from an immediate one. The comparison therefore centered on whether the monetary consequence arrived now or one year later.

The odds feel different in the future

Risk tolerance can reflect at least two parts of a choice. One is the value attached to each possible outcome. The other is the weight a decision-maker gives to the stated probabilities. A 20 percent chance, for example, can influence a decision more or less than its numerical value alone would suggest.

To separate these components, the researchers analyzed the choices using rank-dependent utility, a model that allows probabilities to receive different decision weights. The utility participants attached to outcomes was similar whether payment came immediately or after one year. Probability weighting changed.

For delayed consequences, the probability-weighting function was more elevated. Put simply, participants gave greater decision weight to the chances of a favorable result when the payoff lay in the future. The researchers interpret this pattern as greater optimism about the chances of success. The increased risk tolerance came mainly from this shift in probability weighting, while the valuation of monetary outcomes remained broadly stable.

Separating when risk is known from when it is felt

The distinction between risk resolution and consequence materialization is central to the study. In both experimental conditions, the researchers kept the timing of risk resolution constant. The feature that changed was the date at which the monetary consequence materialized. This isolated payment delay from the separate effects of waiting to learn the result.

The design fills a specific gap in earlier work. Previous studies had examined situations where both resolution and payment were delayed, as well as situations where the timing of resolution changed while payment timing stayed fixed. Kemel and Paraschiv isolate the remaining case: payment is delayed while the timing of risk resolution stays the same.

Their results also clarify the source of the behavioral change. A standard expected-utility analysis might attribute higher risk tolerance to a change in the way future outcomes are valued. The broader model used here shows a different pattern. Outcome utility remains similar across the two timings, while the decision weights assigned to probabilities move upward for delayed consequences.

What the finding adds to real decisions

The experiment uses monetary lotteries in a laboratory, and its strongest claim concerns that setting. The sample comprises 70 undergraduates, and the choices involve nonnegative outcomes with one or two possible results. The study provides a controlled test of one timing difference. Wider applications require evidence from other populations, stakes and domains.

Within those boundaries, the finding adds a precise question to risk assessment: when will the consequences be experienced? Two decisions with similar probabilities and payoffs may elicit different attitudes when one consequence is immediate and the other lies a year ahead.

That question is relevant wherever risk and time arrive together. The paper points to voting, deterrence, education, employment and health. Managers also compare alternatives whose consequences unfold on different schedules. The study makes timing a relevant part of how those alternatives are described and compared, alongside the possible outcomes and their stated probabilities.

A choice framed around consequences next month may feel different from the same choice framed around consequences next year, even when the stated probabilities remain unchanged. The experiment identifies that possibility under controlled conditions. Studies with larger stakes and organizational settings would be needed to measure its reach.

A future payoff can change today's choice

Delayed consequences often enter decision analysis through discounting: how much a future outcome is worth today. Kemel and Paraschiv's experiment identifies another route. The value placed on outcomes remained similar across the two conditions, while probability weighting shifted.

A risky choice therefore carries two timing-sensitive questions: how much the future outcome is worth and how likely success feels when the outcome sits at a distance. In this experiment, the second question drove the difference. The wait before consequences arrive changed the odds people felt they were choosing.