The Study Of The Influence Factors Of The Residents’ Consumption Confidence And The Improvement Of The Index

Based on 105479 survey(2010Q1-2015Q4)samples from 23 cities in the west coast strait economic zones,this paper established an ordered Probit model to explain the underlying causes of Consumer Confidence Index.This paper find that the GDP per capita of the region is the most important factor affecting the confidence index of residents.Besides,the respondents’ level of education also affects their confidence to invest in real estate,which is a key indicator of the index system.Meanwhile,the regression between the indicators shows that that the variables are not completely independent.Such as the expectation of the economy has positive relationship with the real estate purchase confidence and family economic income;the purchase of real estate and durable goods change with the same direction,but with the opposite direction with the expenditure of tourism;the inverstment proportion increases as the inverstment expense increases.Then,in order to evaluate the effectiveness of the confidence indexes of the ability to forecast the economy.I used the analyzes the volatility analysis,the correlation analysis and Grange causality test to test the secondary index of the CCI of the GDP.The result shows that the indexes have different predictability:the confidence of consumption does granger cause the GDP,it can be used as an economic indicator;but the confidence of tourism and GDP does not exist Causal relationship,can not explain the trend of the economy;the confidence of finance is lagged the changes of GDP,which is the reason for the former.The above analysis proves the deficiency of the index in the prediction of economy,so this two aspects needed to be improved:1.The collinearity between the indexes;2.The differences of the ability to forcast.On the one side,I picked up the data of 2010-2013 as the traing set,and then used the principal component analysis(PCA),on the other side,I construct the quantitative and qualitative loss functions.Constrained regression and genetic algorithm are used to optimize the weights with the minimum of the function.Finally,using the rest of the data,by comparing the two methods,the optimal combination of indicators and weights is obtained,and the questionnaire is adjusted to make it more representative in oder to increase its ability to forecast the economic trend of the economy of the west coast strait economic zones.According to the index obtained by the questionnaire can be more accurate forecast of economic trends,so the government and enterprises can make a better plan.The most important thing is that we try to find an effective and unified method of index selection and its weight for the first time.