收入－消費分佈結構的非參數回歸估計：成都市居民戶抽樣調查的案例分析

微觀計量經濟研究中，居民收入－消費分佈結構的研究通常採用參數模型進行實證分析，所得分析結果往往與所設定的模型和條件緊密相關。為了避免設定的模型和條件與實際經濟現象產生較大的偏差，我們可以在資料較多的情況下，利用非參數回歸估計技術來考察收入與消費的相依關係和變化趨勢。本文將首先討論收入－消費模型的非參數回歸估計，研究引入非參數估計模型的必要性和適應性，並列出基本統計性質。由於恩格爾係數為觀察居民消費結構的重要指標，我們也將討論恩格爾曲線的非參數估計問題。利用成都市居民家庭住戶“十五”期間某年的抽樣調查資料，我們由案例資料分析，考察城市居民抽樣住戶的收入－消費基本結構。案例分析將集中討論抽樣資料中極端值（或異常值）對分析結果推斷的影響；利用殘差分佈的估計，研究抽樣資料穩定性對模型建模的影響。

關鍵字

非參數回歸估計；抽樣調查；居民戶；分佈結構

並列摘要

In micro-econometrics, the study on household income and consumption structure is usually studied by parametric regression models to make empirical analysis, whose results are related with the specification on the models. In order to avoid the large difference between the specification and the real economic world, we may use the nonparametric regression estimation to study the relations and trends between income and consumption as long as we have enough samples. We firstly introduce nonlinear parametric model to study the income and consumption structure. Based on the model, we discuss the necessity and adaptability of the nonparametric regression models, and give the basic statistical characteristics. Moreover, we will study the nonparametric estimation of Engel curve because Engel coefficient is an important index about household consumption structure. By using the samples of the survey data in Chengdu in some year among the 10th five year plan, we investigate the basic income and consumption structure. We will mainly study the effect of the extreme values (or outliers). By estimating the distribution of regression residual, we study the impact of the stability of the sampling data on modeling.

並列關鍵字

Nonparametric regression estimation ； sampling survey ； the residents in Chengdu ； the structure of income and consumption distribution

國際替代計量

收入－消費分佈結構的非參數回歸估計：成都市居民戶抽樣調查的案例分析

全文下載

主題瀏覽