Risk Aversion, Information Acquisition, and Technology Adoption

We use a dynamic programming model to study the impacts of risk aversion on information acquisition in technology adoption decisions. In this model, the benefit of the technology is uncertain and, in each period, the decision maker (DM) may adopt the technology, reject the technology, or pay to acquire a signal about the benefit of the technology. The dynamic programming state variables are the DM’s wealth and a probability distribution that describes the DM’s beliefs about the benefit of the technology; these distributions are updated over time using Bayes’ rule. If the signal-generating process satisfies the monotone-likelihood ratio property and the DM is risk neutral, the value functions and policies satisfy natural monotonicity properties: a likelihood-ratio improvement in the distribution on benefits leads to an increase in the value function and moves the DM away from rejection and toward adoption. With risk aversion, the value functions (now representing expected utilities) will be monotonic, but the policies need not be monotonic, even with reasonable utility functions. However, if we assume the DM exhibits decreasing absolute risk aversion and is not “too risk averse,” the policies can be shown to be monotonic. Establishing these structural properties requires the use of some novel proof techniques that may prove useful in other contexts. We also study the impact of changing risk attitudes on the optimal policy.

The online appendix is available at https://doi.org/10.1287/opre.2017.1601.