Pollution and Expenditures in a Penalized Vector Spatial Autoregressive Time Series Model with Data-Driven Networks

Andree, Bo Pieter Johannes ; Spencer, Phoebe ; Chamorro, Andres ; Wang, Dieter ; Azari, Sardar Feredun ; Dogo, Harun

World Bank, Washington, DC

关键词: ENVIRONMENT;    POLLUTION;    POVERTY;    PENALIZED INFERENCE;    SPATIAL ANALYSIS;   

DOI  :  10.1596/1813-9450-8757
RP-ID  :  WPS8757

学科分类：社会科学、人文和艺术（综合）

来源: World Bank Open Knowledge Repository

【 摘 要 】

This paper introduces a Spatial VectorAutoregressive Moving Average (SVARMA) model in whichmultiple cross-sectional time series are modeled asmultivariate, possibly fat-tailed, spatial autoregressiveARMA processes. The estimation requires specifying thecross-sectional spillover channels through spatial weightsmatrices. the paper explores a kernel method to estimate thenetwork topology based on similarities in the data. Itdiscusses the model and estimation, focusing on a penalizedMaximum Likelihood criterion. The empirical performance ofthe estimator is explored in a simulation study. The modelis used to study a spatial time series of pollution andhousehold expenditure data in Indonesia. The analysis findsthat the new model improves in terms of implied density, andbetter neutralizes residual correlations than the VARMA,using fewer parameters. The results suggest that growth inhousehold expenditures precedes pollution reduction,particularly after the expenditures of poorer householdsincrease; that increasing pollution is followed by reducedgrowth in expenditures, particularly reducing the growth ofpoorer households; and that there are significant spilloversfrom bottom-up growth in expenditures. The paper does notfind evidence for top-down growth spillovers. Feedbackbetween the identified mechanisms may contribute topollution-poverty traps and the results imply that pollutiondamages are economically significant.