Applications of Real Options To Data-Based Decision Making In Operational Problems

Real options, inherited from financial options, are a useful tool to manage risks, but also have the flexibility to adopt various risk averse attitudes. In the literature, real options are mostly applied to capital budgeting problems. In some fields, like healthcare, real options theory is not widely known, and not many empirical and quantitative studies are based on this. In other fields, like renewable energy, real options theory have been used to generate traded instruments (like weather derivatives) or those that are only exercised but not priced. This dissertation includes three applications of real options arising in different fields. Chapter 2 presents a quantitative strategy to allocate scarce primary care physician work hours into teams in a Patient-Centered Medical Home. The disruptions caused by transferring primary care physicians between teams are priced by real options. The flexibility to reflect various risk averse attitudes is achieved by applying prospect theory. The numerical experiment shows that our allocation strategy creates less disruption to teams that handle large numbers of sicker patients. The model can enhance the quality of service when compared to existing methodologies. In Chapter 3, a model is developed to find a prediction which minimizes the weighted cost of overestimation and underestimation. The costs of overestimation and underestimation are expressed as real options which are priced. The proposed model makes predictions that are competitive with other methods when compared on datasets drawn from three areas: manufacturing, finance, and environment. In Chapter 4, we integrate the model of Chapter 3 with a wind-to-power conversion process to predict the power output from wind speed. The integration process is realized through Ito's Lemma. The predicted power output is the optimal value that minimizes the weighted cost of over and underestimation. Our numerical results show that the proposed integrated model outperforms other benchmarks. This dissertation extends the real options theory to problems where this theory has not been traditionally applied. First, we are studying are day-to-day operational problems, not capital budgeting problems. Second, the real options models (in healthcare, manufacturing, and renewable energy, etc.) are data-driven and empirical. The real options' ability to flexibly reflect various risk averse attitudes is quantitatively demonstrated. Third, the real options in our models are not traded nor exercised. The true advantage of using real options in risk management is realized by its ability to adapt different pricing mechanisms. In conclusion, the exploration in this dissertation reveals some useful insights from the applications of real options to operational decisions.PHDIndustrial & Operations EngineeringUniversity of Michigan, Horace H. Rackham School of Graduate Studieshttps://deepblue.lib.umich.edu/bitstream/2027.42/151533/1/jeffwjx_1.pd