【摘要】：傳統(tǒng)的面板數(shù)據(jù)是從均值角度進(jìn)行研究,但這會受經(jīng)典假設(shè)條件的約束.而考慮面板數(shù)據(jù)的分位回歸模型,可以更加全面地描述響應(yīng)變量條件分布的全貌.文章引入自適應(yīng)懲罰函數(shù)構(gòu)造了自適應(yīng)懲罰的分位回歸面板數(shù)據(jù)方法,并證明所提出的估計量具有大樣本性質(zhì).蒙特卡洛模擬結(jié)果顯示該方法相對于均值回歸更具優(yōu)勢,是處理面板數(shù)據(jù)的有效手段.文章最后對我國居民交通通訊消費進(jìn)行案例分析,得到了有利于決策的參考信息.
[Abstract]:The traditional panel data is studied from the mean point of view, but it is constrained by the classical assumptions. The quantile regression model considering panel data can more fully describe the conditional distribution of response variables. In this paper, an adaptive penalty function is introduced to construct a quantile regression panel data method for adaptive penalty, and it is proved that the proposed estimator has a large sample property. Monte Carlo simulation results show that this method is superior to mean regression and is an effective method to process panel data. Finally, the paper analyzes the case of communication consumption of residents in our country, and obtains the reference information which is beneficial to the decision-making.
【作者單位】： 中國人民大學(xué)應(yīng)用統(tǒng)計科學(xué)研究中心中國人民大學(xué)統(tǒng)計學(xué)院;蘭州財經(jīng)大學(xué)統(tǒng)計學(xué)院;新疆財經(jīng)大學(xué)統(tǒng)計與信息學(xué)院;
【基金】：中國人民大學(xué)科學(xué)研究基金(中央高�；究蒲袠I(yè)務(wù)費專項資金資助)項目成果(15XNL008)資助課題
【分類號】：O212.1
，

本文編號：2455263