I recently presented a pair of papers on the topic of “Incentives for Innovation” at the Finance Department Seminar at Harvard Business School.

In the first paper, entitled “Motivating Innovation”, I model the process of innovation using a class of Bayesian decision models known as bandit problems. Innovation in this setting is the discovery, through experimentation and learning, of actions that are superior to previously known actions. I focus on the tension between the exploration of new untested actions and the exploitation of well known actions. Exploration of new untested actions reveals information about potentially superior actions, but is also likely to waste time with inferior actions. Exploitation of well known actions ensures reasonable payoffs, but may prevent the discovery of superior actions.

I find that the optimal contracts that motivate exploitation and exploration are fundamentally different. Since exploitation is just the repetition of well known actions, the optimal contract that motivates exploitation is similar to standard pay-for-performance contracts used to motivate repeated effort. On the other hand, since with exploration the agent is likely to waste time with inferior actions, the optimal contract that motivates exploration exhibits substantial tolerance (or even reward) for early failures. Moreover, since exploration reveals information that is useful for future decisions, the optimal contract that motivates exploration relies on long-term incentives. Under the optimal exploration contract, an agent that obtains an early failure followed by a success earns more than an agent that obtains an early success followed by a failure. Even an agent that fails twice may earn more than an agent that obtains a success followed by a failure. The institution of tenure, debtor-friendly bankruptcy laws, and golden parachutes are examples of schemes that protect the agent when failure occurs and thereby encourage exploration.

In the second paper, entitled “Is Pay for Performance Detrimental to Innovation“, which was co-written with Florian Ederer, I outline the results of a controlled laboratory experiment which provides evidence that the combination of tolerance for early failure and reward for long-term success is effective in motivating innovation.

In our experiment, subjects control the operations of a lemonade stand for 20 periods. In each period of the experiment, subjects make decisions on how to run the lemonade stand and observe the profits produced by their inputs. Subjects must choose between fine-tuning the product choice decisions given to them by the previous manager (“exploitation”) or choosing a different location and radically altering the product mix to discover a better strategy (“exploration”). To study the impact of different incentive schemes on productivity and innovation, we consider three different treatment groups. Subjects in the first group receive a fixed-wage in each period of the experiment. Subjects in the second treatment group are given a standard pay-for-performance (or profit sharing) contract. Subjects in the third treatment group are allocated a contract that is tailored to motivate exploration. Their compensation is 50% of the profits produced during the last 10 periods of the experiment.

(more…)