Abstract: Guangzhou is the only city in China which does not limit the non-local cars after the implement of vehicle purchase restriction measures. Because of this particularity, Guangzhou residents choose to lottery, auction or non-local car license plate. Based on the data of car purchase restriction and sampling questionnaires on the options, Guangzhou residents’ choice behavior of car license plate was analyzed. The nested Logit model based on residents’ choice of car license plate was established. The proposed model included the option that people obtained a car license plate in surrounding cities. Parameters of the model were estimated by weighted endogenous sample maximum likelihood(WESML) based on options. The influence of residents’ incomes, the urgency degree of demand, the cognition value of local car plate and other factors on the residents’ decision-making was quantitatively analyzed. The results indicate that：the proposed model and method are superior to the traditional multinomial Logit model and the general maximum likelihood method, which can describe and predict people’s car plate choice behaviors more appropriate.

Key words: traffic engineering, car purchase restriction, lottery, auction, non-local car license plate, sampling based on options〖FGN〗

摘要： 广州市是全国唯一在汽车限购后没有进行外地车限行的城市。由于这一特殊性，广州市居民选择摇号、竞拍或者外地牌。基于限购数据及基于选项的采样问卷调查数据，分析广州市居民车牌选择行为，建立基于居民车牌选择的巢式Logit模型，将广州居民在周边城市上牌作为选项之一，并利用加权及基于选项的最大似然法(WESML)对模型的参数进行估计，定量分析居民收入、需求迫切度、对本地车牌的认知价值等因素对居民决策产生的影响。结果表明：采用的模型和方法要优于传统的多项Logit模型及一般的极大似然法，能够更加贴切的描述和预测居民的选择行为。

关键词: 交通工程, 小客车限购, 摇号, 竞拍, 外地牌, 基于选项采样