Sector Similarity in Input-Output Networks

Abstract

Input-Output (IO) model is a macroeconomic model describing the inter-sectoral interdependence of economies. It is widely used to analyze environmental impacts from economic activities. The conventional method to build up the IO table is largely based on onerous data collection but simple linear approximation. In order to more accurately construct IO tables and efficiently capture outliers among the dataset, we introduce network theories to investigate the underlying relationships between economic sectors. By probing into similarity between economic sectors, we could conclude correlations and connection patterns between individual economic flows. In this way, even with partial data of one IO table available, it will still be possible to restore the complete map of an IO table by referencing their inherited relationships. The achievement of such prediction will further advance our environmental analysis that based upon IO model via more accurate and up-to-data data. This study focuses on similarity exploration between economic sectors in IO model and constructing a theoretical framework for establishing IO table using network theories of link prediction.