DEVELOPING A MULTI-SCALE MULTI-REGION INPUT–OUTPUT MODEL

Abstract

Many efforts have recently been devoted to developing global multi-region input–output (GMRIO) models. Unfortunately, the scales of GMRIO models do not allow them to capture the heterogeneity of regions within a single country. Multi-scale models can provide more comprehensive analyses capable of capturing the interdependencies of the global economy while preserving regional differences. The primary objective of this research is to develop methods for integrating multi-region input–output data sets from multiple spatial scales into multi-scale multi-region input–output (MSMRIO) models. These methods result in models that may have unusual features such as non-square trade coefficient matrices and a mix of industry-by-industry and commodity-by-commodity technical coefficients. To demonstrate the feasibility of MSMRIO modelling, a Canada-centric model was developed. This model includes 47 countries and Canada's 13 subnational regions. A MSMRIO model provides a tool to analyse global issues with a more spatially detailed focus.

Keywords:

• Multiregional input–output tables
• Multi-scale
• Trade coefficients
• Global analysis

Notes

¹$\mathrm{MAE}=(1/N){\sum }_{k=1}^N|A_k-P_k|,$ where N is the number of commodity sectors, k is the commodity sector, A_k is the actual output, and P_k is the predicted output.

²$\mathrm{MADP}={\sum }_{k=1}^N|A_k-P_k|/{\sum }_{k=1}^N|A_k|,$ where all variables are defined as above.

³This is discussed elsewhere (Yamano and Ahmad, 2006; Guo et al., 2009; Zhu et al., 2011) but includes import and export data that do not match, re-exports causing errors in import/export data, valuation differences (c.i.f. versus f.o.b.), assumptions made by national statistical offices and the OECD when converting the data to the 37-by-37 sector classification scheme, and others. On the other hand, Canadian I–O and trade data are very carefully harmonized (see Généreux and Langen, 2002).