Spatial Pattern,Dynamic Evolution And Spatial Heterogeneity Of Multidimensional Poverty

Multidimensional poverty is the major trend of poverty.It is also a crucial indicator for measuring multidimensional poverty with the strategy of targeted poverty alleviation in Chinese local governments.For years,most scholars at home and abroad mainly study the solo money income poverty,however Sabina Alkire from Oxford University and James Foster at Washington University firstly give a method for measuring multidimensional poverty,that's A-F method based on the theory of Capability Approach of Amartya Sen.Scholars from domestic or overseas apply this method to study the measurement of multidimensional poverty,decomposition in terms of indicators,areas and time,as well as the further study of status continuity and factor leading to multidimensional poverty.While,the previous literature is based on the assumption of homogeneous and balanced area,spatial otherness,another important feature is neglected.We use recently available panel household survey data from China's National Bureau of Statistics in China's Xinjiang Province among 2011,2012,2013 and 2014 year.We also use empirical method to examine spatial distribution and evolution,as well as the relationship between factors resulting in poverty and multidimensional poverty from spatial economic model.Then we give some suggestions about targeted alleviation poverty.Results obtained from empirical analysis show that H is on persisting declination as a whole,the condition of multidimensional poverty is constantly improved.H is the highest in Qizilsu Qirghiz Aptonom Oblasti since 2011,the ranking position in Kashgar Prefecture is increasing,but Hotan Prefecture drop in ranking constantly.Within the 4 years,MPI in Qirghiz Aptonom Oblasti is bigger than other two areas,Hotan Prefecture is less and Kashgar Prefecture is the minimum.(3)From the temporal dynamics,we see a declining trend of H in the most counties,and only parts of counties fall firstly and arise later.(4)Varying from MPI to H,kernel estimation results show that our study area have experienced a good trance of multidimensional poverty.(5)Without taking a consideration of spatial factors,we find that average land variable,average education expenditure variable and population density variable could significantly reduce MPI in fixed effects model.In terms of improving multidimensional poverty,different macroeconomic factors have different effects.In general,average education expenditure variable and population density variable improve MPI.However,after using spatial Durbin model,we find that the geographical concentration of average education expenditure contribute the most to lead to spatial difference of multidimensional poverty,and then population density shrinks the spatial discrepancy.On the contrary,the amount of doctors increases this difference.Average GDP,financial development and the proportion of third industry variable have positive impact in geographical weight,butit is not statistically evident.(6)But in terms of three effect,average education expenditure has a significant overall and indirect effect,as well as the indirect effect of the multidimensional poverty condition overweigh the direct one.Average land variable have a negative total effect to MPI.Average GDP has a significantly negative total effect on multidimensional poverty with indirect effect beyond direct one.The third industry variable,financial development variable and medical system in countryside have a significantly positive total effect on multidimensional poverty,but industrialization rate has a significantly negative indirect and total effect on MPI.