基于多元线性回归的每股收益影响因素分析——以安徽省为例

Analysis of Influencing Factors of Earnings Per Share Based on Multiple Linear Regression ——A Case Stufy of Anhui Province

针对安徽省制造业上市公司每股收益的实证研究,将基于杜邦原理分析得到的若干影响每股收益的因素作为回归模型中的解释变量,同时均值处理收集得到的2014~2016年数据,代入Eviews软件进行参数估计。并对模型进行多重共线性,异方差和自相关检验,根据检验结果对模型不断修正完善。最终运用逐步回归法和加权最小二乘法得到关于每股收益影响因素的多元线性回归式,并提出关于提高上市企业每股收益的有益建议。

Focusing on the empirical study of the listed company’s earnings per share in Anhui Province,the paper takes the factors that affect the earnings per share as the explanatory variables in the regression model,based on the theoretical analysis of the DuPont principle,and simultaneously substitutes collected data from 2014 to 2016 which is processed with mean method into Eviews software for parameter estimation.The multiple co-linearity,different variance,autocorrelation tests are performed on the model which is continuously revised and refined according to the test results.Finally,multiple linear regressions on the factors affecting earnings per share is obtained by using stepwise regression method and the weighted least square method,and some useful suggestions are proposed for improving the earnings per share of listed companies.