A computationally efficient fixed point approach to dynamic structural demand estimation
| Autor: | Yutec Sun, Masakazu Ishihara |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2019 |
| Předmět: |
Mathematics
Interdisciplinary Applications Economics and Econometrics Mathematical optimization Random coefficients logit MCMC Computer science Economics Monte Carlo method Dynamic Social Sciences Fixed point LOGIT-MODELS symbols.namesake Business & Economics BAYESIAN-ESTIMATION ALGORITHM computer.programming_language Science & Technology Laplace transform Applied Mathematics Estimator Markov chain Monte Carlo Social Sciences Mathematical Methods PERFORMANCE PRACTITIONERS GUIDE DISCRETE-CHOICE MODELS Product (mathematics) Physical Sciences symbols BLP computer Mathematics Mathematical Methods In Social Sciences Nested fixed point Rust (programming language) Panel data PACKAGE |
| Popis: | This paper develops a computationally efficient approach to the estimation of dynamic structural demand with product panel data. The conventional GMM approach relies on two nested fixed point (NFP) algorithms, each developed by Rust (1987) and Berry, Levinsohn, and Pakes (1995). We transform the GMM into a quasi-Bayesian (Laplace type) estimator and develop a new MCMC method that efficiently solves the fixed point problems. Our approach requires no stronger assumptions than the GMM and can thus avoid bias from misspecified models. In Monte Carlo analysis, the new method outperforms both NFP and MPEC, particularly in large-scale estimations. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|