统计研究
統計研究
통계연구
Statistical Research
2014年
4期
102~107
,共null页
二元面板 截面相关 条件Logit Copula函数
二元麵闆 截麵相關 條件Logit Copula函數
이원면판 절면상관 조건Logit Copula함수
Binary Choice Panel Data; Cross-Sectional Dependence; Conditional Logit; Copula Function ~：~ t ~, I~ o ~ ~＇~＇, ~ ~ EM ~ ~ （ McMillen,
基于双变量FGM类函数的Copula方法可以简单有效估计带截面相关的固定效应二元面板数据模型.为了能够构造出Copula分布函数,我们在经典的条件Logit的估计之后再做一次极大似然估计,得到异质性的估计,这样有了参数和异质性就能得到残差从而估计得到相关系数,进而构造Copula分布函数.蒙特卡罗模拟发现基于Copula得到的参数估计量在收敛速度上要明显好于经典的条件Logit估计量.
基于雙變量FGM類函數的Copula方法可以簡單有效估計帶截麵相關的固定效應二元麵闆數據模型.為瞭能夠構造齣Copula分佈函數,我們在經典的條件Logit的估計之後再做一次極大似然估計,得到異質性的估計,這樣有瞭參數和異質性就能得到殘差從而估計得到相關繫數,進而構造Copula分佈函數.矇特卡囉模擬髮現基于Copula得到的參數估計量在收斂速度上要明顯好于經典的條件Logit估計量.
기우쌍변량FGM류함수적Copula방법가이간단유효고계대절면상관적고정효응이원면판수거모형.위료능구구조출Copula분포함수,아문재경전적조건Logit적고계지후재주일차겁대사연고계,득도이질성적고계,저양유료삼수화이질성취능득도잔차종이고계득도상관계수,진이구조Copula분포함수.몽특잡라모의발현기우Copula득도적삼수고계량재수렴속도상요명현호우경전적조건Logit고계량.
This paper proposes a method to estimate binary-choice panel models with fixed effect and cross-sectional dependence based on FGM family bivariates copula function. To construct copula-based likelihood functions, we estimate a common intercept in place of the heterogeneities after estimating the interested parameters with conditional Logit models. With these results, we can compute the general residuals and the correlation coefficients, and thus construct the copula- based likelihood functions. Monte Carlo simulation shows copula-based estimators have better convergence rate than classical conditional Logit estimators.
