|
Abstract:
|
Logit models and multinomial models have been widely used to model the individual's utility and preference. Based on individuals' microeconomic data, their behaviors, choices, and demands can be well estimated. When individual's data are not available, logit models and multinomial models can still be estimated based on aggregated data of multiple individuals at the same time point. However, in reality, economic data are often collected at mixed frequencies, i.e. some high-frequency whereas others low-frequency. We proposed estimation methods for logit models and multinomial models modeling high-frequency behaviors whereas the individual's behavior data are at low-frequency though the characteristic data of individuals are at high-frequency. Simulation studies have been conducted to valid our estimation methods. A real data application has been shown to illustrate our method.
|
