Analysis And Forecast Of Resident Consumption Structure Based On Compositional Data

Economic growth has led to an increase in residents' income and consumption expenditure.Consumption structure is continuously developing towards high level and high quality.In recent years,consumption has been a primary driving force for economic growth.Meanwhile,as a mirror of the local economic situation,the level of consumption structure is in proportion to the local economy.This thesis takes the urban residents of Shanxi Province as the main line,focusing on the consumption structure,studies their consumption status,and explores their consumption structure,and their consumption status,thereby promoting the economic development of Shanxi Province.In addition,the research on the consumption structure will be conducted based on the theoretical knowledge of compositional data.Specifically,it combines the consumption structure of the research object with the theory of compositional data,and makes a comprehensive analysis.(1)This thesis describes and analyzes the consumption structure of urban residents in Shanxi Province.The data from 2003 to 2017 are visualized to describe the trend change.And it is compared with the consumption structure of urban residents in China.Results show that the changing trend of the eight consumption structures of Shanxi urban residents remains basically the same with the national trend.In detail,the proportion of food,clothing,household goods,culture,education,and entertainment items has declined,while that of consumer items,such as housing,transportation and communications,and health care,is on the rise.(2)It analyzes two influencing factors of consumption structure,residents' disposable income and population age structure.A grey relational analysis is used to study the influence of income on consumption structure.Considering the multicollinearity of the variables,this thesis selects the partial least square regression method to eliminate the redundant dimensions due to the fact that population age structure and consumption structure are compositional data.To improve the interpretation performance of the model,symmetrical logratio is carried on variables,so that the transformed compositional data components are correspond to the original ones,respectively.Results show that the consumption structure of Shanxi urban residents is constantly optimized with the increase of income,and also changes with the change of population age structure.The increase in income,the decline in the proportion of children and the rise in the proportion of the elderly,stimulate the consumption demand of developmental and enjoyment information,and reduce that of basic survival needs.(3)The future consumption structure is forecasted.Most scholars directly use the absolute value of consumer expenditures to predict the trend.However,this paper applies the prediction model of compositional data,where the Aitchison distance of compositional data is used as the prediction evaluation index.Two single prediction models,regression model and grey model,are adopted respectively.By comparing the advantages and disadvantages of the single models,this paper combines the two models by taking the prediction results of the regression model as the input of the grey model.The prediction results show that the combination prediction model has a better performance.