离散选择模型的基本原理及其发展演进评介.pdfVIP

离散选择模型的基本原理及其发展演进评介 # 151 # 离散选择模型的基本原理 及其发展演进评介 聂 冲 贾生华 ( 浙江大学管理学院) = 离散选择模型的研究真正兴起于 19 世纪50 年代末, 于微观计量经 济学的范畴该模型能够对个体和家庭行为进行经验性的统计分析, 因而在经济学 和其他社会科学中得到广泛的应用本文从离散选择模型的基本性质及效用最大化 的理论背景出发, 指出logit 模型虽然使用的是最早并且最为广泛的离散选择模型, 但是其存在着三大局限性: 不能表示随机口味的变化暗含成比例的替代形式和不 能处理不可观测因素在不同期间相关的情形GEV ( 含嵌套logit) probit 和混合 logit 模型等其他的离散选择模型, 很大程度上都是为了避免这些限制而产生并发 展起来的 离散选择 logit 局限性 F06114 A Research on The Theoretical Basis and Evolution of Discrete Choice Models Abstract : The boom of discrete choice models , hich belong to the area of microeconometrics, ere really beginning from the end of 1950s1T he models can be used to analyse the behaviors of private and family experientially, and they are idely applied to economics and other social sciences1 T his paper begins ith the theoretical basis and utility maximization, then demonstrates that there are three limitations of logit: it can not represent random taste variation; it implies propor- tional substitution across alternatives and cannot handle situations here unob- served factors are correlated over time, though logit is by far the easiest and most idely used discrete choice model1 Other models, such as GEV ( including nested logit) , probit and mixed logit, have arisen largely to avoid these limitations1 Key words: Discrete Choice Model; Logit; Limitation 2000 10 11 , 2000 # 152 # 5数量经济技术经济研究6 2005 年第11 期 # (James Heckman) # (Daniel McFadden) , , James Heckman , Zvi Griliches, L1 L1 Thurstone, Jacob Marschak, Ducan Luce, Danny Kahneman, Amos T versky, Moshe BenO Akiva, Charles Manski K