In this thesis it is shown that the allocation function derived by Boyd et al. (DFI) has close relationship with logit model. The allocation or market share function of both model can be shown to have the same probabilistic structure by simple transformation of random variables.
Simplifying and generalizing these concepts Weibull cost model and Gumble utility model are developed. Gumble utility model seems to be a new explanation for the use of Gumble distribution for logit model.
Finally, a discrete/continuous model by Hanemann is employed to model residential space heating technology selection decision. The parameters are estimated using cross section data from '81 energy census questionaire. The questionaire is designed by Korea Institute of Energy & Resources (KIER) and its results are coded in magnetic tape. The estimation technique used is a kind of weighted least square method termed minimum logit chi-square method.
The experience in parameter estimation seems to imply that future study of this sort must include information on future relative price of briquet and estimated by household as well as current relative prices.