| According to the Dixit and Pindyck's (1994) theory of investment, oprimal decisions to investment problems that satisfy three requirements--irreversibility, uncertainty in future rewards, and flexibility in timing--need to explicitly consider the cost of exercising the option of making the investment. The Great Lakes levels management problem that I analyze here meets the above three criteria. Construction of a control structure on Lake Erie is the irreversible decision, its future benefits are subject to climate change uncertainty, and the investment decision can be delayed into the future. The International Joint Commission (IJC) conducted a multi-year, multi-million dollar study to analyze this investment decision. They used a simple scenario analysis based procedure where they assumed that the past (NBS) will repeat in the future; they did not explicitly include climate change uncertainty or flexibility in timing in their decision process.;In this dissertation, I explicitly include climate change uncertainty and flexibility in timing into my analytic process. I quantify the expected value of including climate change uncertainty, the expected value of perfect climate change information, and the option value of waiting. I have used a Bayesian Monte Carlo (BMC)-based sequential decision analysis framework for conducting this study. This is the first time that Bayesian Monte Carlo simulation has been combined with sequential decision tree analysis for solving management problems. At each future stage, I update my belief in climate change based on the available observations of Net Basin Supplies. I used the CWRU Great Lakes Hydraulic and Socio-Economic Impact Simulation Model to quantify the net benefits. To assess the importance of climate change compared to other uncertainties, I compare these values; with those associated with uncertainty in shoreline (flooding and erosion) damages. I repeated this analysis under shoreline damage uncertainty. Considering the EVIU's, I infer that for certain scenarios the climate change and shoreline damage uncertainties are of comparable importance. The NPV of the option value of waiting by conducting a three stage decision tree analysis (for acquiring climate change information) was estimated to be ;I also conducted sensitivity analysis with respect to different parameter and decision variable values. Plan 25N was found to be the best plan among Plans 12N, 25N, and 50N. An increase in the number of decision stages increases the total optimal expected net benefits, as expected. The severity of climate change determines whether the net benefits due to global warming in the Great Lakes Basin are positive or negative. In particular, an extreme climate change scenario (MPI) will make a 3-lake plan attractive because of substantial reduction in navigational costs while a mild scenario (UKMO) will make such a plan unattractive. The option value of waiting falls to zero under high discount rates while it can be substantial under low discount rates. Lastly, the BMC simulation procedure was found to be feasible but computationally very demanding. |
