Vehicle Brand Selection: Multinomial Logit Model

We develop the model of multiple choice of McFadden (1974) that find the determinants of consumers for choosing several alternative brands of automobiles: HyunDai, Korean GM, Kia, and SsangYong, etc. We performed four types of discrete dependent variable models. Estimation of binary probit results shows, as the number of members in the family, the choice for GM increase. Discrete LS with breaks and discrete threshold regression shows that the marginal effect on choice of specific automobile is affected by the attributes of alternatives such as the price of car and the engine displacement(capacity of engine, CC). We also estimated the semiparametric model of maximum score. In the case multiple choice estimatied by multinomial logit model, the results lead to the guess that higher family income increase the choice of Kia and (Korean) GM. Using NaSTab data in Korea, the estimation of conditional logit was difficult since we have choice-specific data. These results implicate that characteristics and attributes both affect the choice of car brand of Korean consumers.