Recently standard real options models have been extended in various ways. One important extension is the incorporation of Bayesian learning. This new approach encourages active information acquisition and quantifies its impact on decisions.

Description

We examine how several sources of uncertainty impact the investment decision with particular emphasis on learning through time. More specifically, the firm might learn about the true political or business climate by waiting, yet waiting reduces the discounted expected cash flows.

Selected Results

Our results illustrate how learning about policy commitment impacts investment decisions. In order to facilitate this analysis, we incorporate Bayesian learning into real options modelling. This can be understood as a firm’s interpretation of signals from regulatory agencies. Dalby et al. (2018) investigate how investment behaviour is affected by updating a subjective belief on the timing of a subsidy revision. More specifically, how retroactive downward adjustment of a common renewable energy subsidy (fixed feed-in tariffs) impacts the investment decision. Their results indicate that that the effect from a high subsidy level can be significantly mitigated by a perceived unreliable government.

In the same line of work, Hagspiel et al. (2021) study a firm with the option to engage in costly active learning, either through paying an upfront fee or a continuous learning cost. They find that continuous learning results in an incentive to invest sooner in order to avoid further learning cost payments, especially if the learning rate is chosen to be large. Furthermore, their results indicate that clear communication on a future policy attracts investments from firms that are efficient learners. This is because firms that learn more efficiently, adjust their beliefs more rapidly than firms that learn poorly. Hence, efficient laearners invest sooner given that we are in a good economic climate. However, these firms are also more sensitive to bad signals, thus less likely to invest in a bad economic climate.

Firms might also learn by collecting additional data. Ødegaard et al. (2020) investigate how  gathering snow measurements can facilitate learning about future inflow for a hydropower producer. They find that for smaller reservoirs, where the probability of overflow is greater, snow measurements can add considerable value given that the measurements have high accuracy. They conclude that snow measurements might add between 0 and 10% in value for reasonable parameters.

Related Research

• Dalby, P.A.O. , G.R. Gillerhaugen, V. Hagspiel, T. Leth-Olsen, J.J.J. Thijssen (2018). Green Investment under Policy Uncertainty and Bayesian Learning. Energy, Volume 161, 1262-1281.
• Hagspiel, V, R.L.G. Nagy, J.J.J Thijssen (2021). Investment under a disruptive risk with costly Bayesian learning: The optimal choice of the learning rate. Working paper
• Ødegård, H. L., Eidsvik, J., & Fleten, S. E. (2017). Value of information analysis of snow measurements for the scheduling of hydropower production. Energy Systems, 1-19.