摘要
经典的计量经济模型常以正态分布作为假定分布,但这种假定往往与现实不符。在此情况下,如果依然按照正态性假设进行参数估计,则会产生一定的偏差。文章使用28种分布来拟合城镇居民家庭消费支出数据,经过比较得出广义贝塔Ⅱ型分布(GB2)最为有效;基于GB2分布重新计算得出居民边际消费倾向约为0.58,而基于经典线性回归分析得出的边际消费倾向约为0.53,这意味着如果不考虑居民消费支出服从的分布形态,则会低估城镇居民边际消费倾向约10%。
The classical econometric model usually takes normal distribution as the assumed distribution, but this kind of assumption is often inconsistent with the reality. In this case, if the parameter estimation is still carried out according to the assumption of normality, certain deviation will occur. This paper uses 28 kinds of distribution to fit the data of urban household consumption expenditure. After comparison, the generalized beta Ⅱ distribution(GB2) is the most effective. Through the recalculation based on the GB2 distribution, the marginal propensity to consume is about 0.58, while the marginal propensity to consume based on the classical linear regression analysis is about 0.53, which means that the marginal propensity to consume of urban residents will be underestimated by about 10% if the distribution pattern of consumer expenditure is not considered.
作者
王超
Wang Chao(School of Mathematics and Statistics,Anyang Normal University,Anyang Henan 455000,China)
出处
《统计与决策》
CSSCI
北大核心
2022年第5期28-31,共4页
Statistics & Decision
基金
国家社会科学基金资助项目(19BTJ051)。
关键词
消费支出
非正态分布
广义贝塔Ⅱ型分布
边际消费倾向
consumption expenditure
non-normal distribution
generalized betaⅡdistribution(GB2)
marginal propensity to consume
作者简介
王超(1979—),男,河南安阳人,博士,副教授,研究方向:统计理论与方法应用。
