INVESTIGATING ALTERNATIVE APPROACHES TO HARMONISE MULTI-REGIONAL INPUT–OUTPUT DATA

Abstract

Over recent years a small number of global multi-regional input–output (MRIO) databases were developed to describe the entire global economy at high sector detail. We investigate the differences that arise out of applying different construction procedures for two global MRIO databases: The EXIOBASE database, developed as part of the EU FP6 & 7 programs and the Eora database developed at the University of Sydney. The procedures used in EXIOBASE involve a high degree of interrogation and adjustment throughout the construction of the data set, whilst the Eora MRIO relies on single-step mathematical programming techniques and high-performance computing. We unravel the effect of the different approaches taken to develop the databases by undertaking a number of combinatorial experiments in which we exchange parts of the construction process between the EXIOBASE and Eora build pipelines. We conclude that Eora's highly automated data reconciliation approach produces MRIO databases that are of comparable quality to those constructed with EXIOBASE's multi-step approach. However, the reliability and robustness of the resulting MRIO database largely depend on the level of detail and reliability of the underlying raw data.

Keywords:

• Multi-regional input–output analysis
• Matrix balancing
• Constrained optimisation
• Automation
• Matrix distance

Notes

¹The EXIOBASE database is currently available in two different versions: EXIOBASE v1.0 (developed from the EXIOPOL project) is an MRIO framework for the year 2000, EXIOBASE v2.0 (developed from the CREEA project) is an MRIO framework for the year 2007. EXIOBASE v2.0 is the result of EXIOPOL's follow-up project. The two databases (EXIOBASE v1.0 and EXIOBASE v2.0) differ in size and detail, but mainly cover the same countries and regions. The different structures (particularly the sector classifications) of EXIOBASE v1.0 and EXIOBASE v2.0 are discussed in Section 2.2. EXIOBASE is available under http://www.exiobase.eu/. The database if briefly described in Appendix 1. For more detailed information, see Tukker et~al. 2013a and Tukker 2013

²For more information on Eora, see Appendix 4, Lenzen et~al. 2012; Lenzen et~al. 2013 and Moran 2013.

³A fully assembled IE data set for EXIOBASE v2.0 was not available at the time of this project. The construction of the IE data set that was used during this project is described in Section 2.

⁴A full list of exchange rates is available under http://unstats.un.org/unsd/snaama/dnlList.asp.

⁵NACE is the Statistical classification of economic activities in the European Community. For more information on NACE, see http://epp.eurostat.ec.europa.eu/portal/page/portal/nace_rev2/introduction.

⁶The term SSE denotes the summed square of residuals. The residual is the difference between two corresponding values in each of the compared databases. The term SST denotes the total sum of squares.