Estimation of Nested Logit and Threshold Regression Model for Household Motorcycle Selection
Byung‐Woo Kim
Abstract
Byung‐Woo Kim
Abstract
We develop the model of binary choice model that finds the determinants of households for choosing the brands of motorcycle: DaeLim, Hyosung, and Suzuki, etc. Rercent development in discrete dependent variable model makes it possible to estimate unobserved threshold affecting consumers’ marginal effect of income or wealth on choice of brands. We performed the estimation of several types of qualitative dependent variable models. Estimation of binary probit results shows, as a mean of family transportation vehicle for the choice of motorcycle shows that it is not an inferior good considering the systematic selection of sample. But, discrete threshold regression shows that the domestically produced motorcycle(e.g. DaeLim brand) become an inferior good as the consumer(household) have relatively high income. In the case multiple choice estimated by multinomial logit model, the results lead to the guess that higher family income decrease the choice of motorcycle, and then increase the preference for cars. This multiple choice model is used widely in the field of Marcov Switching, and EM algorithm in ML, etc.
A significance statement is not available in the OpenAlex record.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
We develop the model of binary choice model that finds the determinants of households for choosing the brands of motorcycle: DaeLim, Hyosung, and Suzuki, etc. Rercent development in discrete dependent variable model makes it possible to estimate unobserved threshold affecting consumers’ marginal effect of income or wealth on choice of brands. We performed the estimation of several types of qualitative dependent variable models. Estimation of binary probit results shows, as a mean of family transportation vehicle for the choice of motorcycle shows that it is not an inferior good considering the systematic selection of sample. But, discrete threshold regression shows that the domestically produced motorcycle(e.g. DaeLim brand) become an inferior good as the consumer(household) have relatively high income. In the case multiple choice estimated by multinomial logit model, the results lead to the guess that higher family income decrease the choice of motorcycle, and then increase the preference for cars. This multiple choice model is used widely in the field of Marcov Switching, and EM algorithm in ML, etc.
Key concepts: Multinomial logistic regression, Discrete choice, Econometrics, Multinomial probit, Probit, Ordered probit, Selection (genetic algorithm), Mixed logit