Addiction to Rewards

51 Pages Posted: 11 Aug 2013

See all articles by Anton Suvorov

Anton Suvorov

National Research University Higher School of Economics

Date Written: June 2013


The paper explores the "hidden costs of rewards" in a dynamic principal-agent framework, in which an informed principal selects in each period a reward for an agent. It shows that rewards are often addictive in that once offered, a contingent reward makes the agent expect it whenever a similar task is faced. This expectation develops because rewards bring bad news: a higher reward is more likely to be offered when the task is difficult or the agent is weak. This, in turn, compels the principal to use rewards over and over again. Furthermore, in a long-term principal-agent relationship there is a two-sided ratchet effect: the principal is concerned about creating addiction for the agent, whereas the agent does not want to appear too enthusiastic. On the principal's side, the ratchet effect implies that there are fewer rewards in a long-term principal-agent relationship than in a situation where the agent faces transient principals while implementing a series of similar tasks. On the agent's side, ratcheting conflicts with a desire to work in order to acquire valuable information about the task attractiveness. If doing the task once allows the agent to fully learn its characteristics, addiction may not develop. In contrast, the principal may first promise a bonus when the task is easy, rather than difficult (or the agent is strong) so as to induce the agent to acquire the good news "by doing". Then, bonuses become redundant thereafter.

Keywords: addiction, rewards, ratchet effect, self-conÂ…fidence, informed principal

JEL Classification: A12, D82, J22, J33

Suggested Citation

Suvorov, Anton, Addiction to Rewards (June 2013). Available at SSRN: or

Anton Suvorov (Contact Author)

National Research University Higher School of Economics ( email )

Shabolovka 26
Moscow, 119049

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Abstract Views
PlumX Metrics