各校計畫成果

活動簡介

The hold-up problem is one of the most fundamental and enduring issues in the field of Behavioral Industrial Organization. Although incomplete contracts have been studied extensively, the literature is lacking empirical, and especially experimental evidence. This paper examines the effects of fixed cancellation payment on the hold-up problem through parameterized modelling and results of a lab experiment. Our experiment results conform to the equilibrium prediction in general: setting the cancellation payment too low can lead to agents being held-up, resulting in inefficiently low investment; setting it sufficiently high can enhance the agent's incentive and solve the hold-up problem, but setting it too high could lead to the agent to invest an inefficiently high amount, i.e. the reverse hold-up problem. Our study has important policy implications that carefully designed cancellation clauses could be harnessed by policymakers and mechanism designers to achieve outcomes that maximize social welfare; Another takeaway from our experiment is the learning effect, which also implies that policymakers could expect a contract regime to become increasingly effective over time.

This paper uses theoretical predictions and experimental evidence to study the effects of cancellation payment in incomplete contracts on the hold-up problem. In particular, we are interested in the situation when setting cancellation payment too low leads to standard hold-up problem, and setting it too high leads to the reverse hold-up problem, a case of particular concern when agents make inefficiently high investments. By setting the cancellation payment at a sufficiently high level, the hold-up problems can get mitigated and socially optimal outcome will be achieved. Our results from the laboratory experiment strongly conform to our theoretical predictions. We use regression analysis and hypothesis testing to show that the observed frequencies of subgame perfect equilibrium play in our 5 treatments support our theoretical predictions; And when the cancellation payment is either too low or too high, we have observed significant efficiency loss due to hold-up and reverse hold-up.

Another takeaway from the experiment is that we observe and then test the learning effects on the participants making equilibrium choices: In general the First-Round effects tend to bias the equilibrium likelihood downwards; while the Last-Round effects tend to go in the opposite direction.

For future research opportunities we are considering an extension study, to design a repeated game setting to examine the effect of reputation building on mitigating the hold-up problem. The present study is based on a static game setting where each pair of participants played a one-shot game (for each treatment); while in real life setting we might observe agents and principals playing repeated games where the principal could build up a reputation of being inclined to renegotiate or not. If this reputation building is known record and credible to the agents, principals might become less inclined to renegotiate thus giving agents more incentives to enter the implicit contract and make sufficient effort (and mitigating the standard hold-up problem). By designing different institutions for the hold-up game, we could look at which one delivers a more efficient outcome.

Distribution of player 2’s choice. It shows that hold-up problem and reverse hold-up problem can be generated by different amount of cancellation cost in the lab.