![Sample our Environment & Sustainability journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5935/TOC-Subject-Sample-2021-Environment-Sustainability.jpg"})
ABSTRACT
Global economic analysis requires consistent and balanced data, which necessitates the reconciliation of datasets from both national and international sources. In the case of the Global Trade Analysis Project Data Base, datasets supplied by international sources are considered preferable to national input–output (I–O) tables. As a result, the national I–O data can experience significant adjustments during the reconciliation process due to differences between the national and international datasets. The purpose of this paper is to examine the extent to which national I–O data change during reconciliation. The results demonstrate that the I–O data are altered by the construction process, particularly from the reconciliation of the national I–O data to the international trade and energy datasets. Closer examination reveals potential issues with both the trade and energy datasets, as well as the national I–O data – illustrating the challenges associated with reconciling data from multiple sources.
Acknowledgements
The authors would like to thank Patrick Jomini for his encouragement in producing this paper, and Peter Minor for his comments and suggestions and for sharing the Cyprus example, which was undertaken by ImpactECON as part of a project for the Ministry of Energy, Commerce, Industry and Tourism, Cyprus and the European Commission. We would also like to thank the two anonymous reviewers for their very helpful comments and suggestions.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
2 Protection data are a critical component of the GTAP Data Base and include tariffs, agricultural domestic support and export subsidies.
5 Other global datasets, namely WIOD (Timmer, Citation2012), EXIOBASE (Tukker, Poliakov et al., Citation2009) and EORA (Lenzen, Moran, Kanemoto and Geschke, Citation2013), use fewer international datasets and give the national I–O data preference. These datasets are generally used for I–O analysis and do not contain protection data (i.e. tariffs). Like GTAP, these datasets must also deal with inconsistencies between the various datasets.
6 For more information on the construction process see Harslett (Citation2013) and Aguiar et al. (Citation2016).
7 National Central Banks and other institutions, such as the International Food and Policy Research Institute, have also developed I–O tables or Social Accounting Matrices.
8 See Huff et al. (Citation2000) and https://www.gtap.agecon.purdue.edu/databases/contribute/default.asp.
9 Other issues are outlined in the Appendix I.
10 If the original I–O table does not satisfy this requirement then it is up to the contributor to aggregate or split sectors/commodities prior to contributing the I–O table to the CGTA, so that there is a one-to-one or many-to-one mapping between the 57 GTAP sectors and the sectors in the contributed I–O table.
11 Table A1 in Appendix I provides summary statistics, including the number of sectors (g) in the contributed I–O tables. Later this information is used to establish if national I–O tables with missing sectors change more than those with all 57 GTAP sectors.
12 The list of all countries is provided at https://www.gtap.agecon.purdue.edu/databases/regions.asp?Version=8.311.
13 Coal (GTAP sector, coa), oil (oil), gas (gas), petroleum and coal products (p_c), gas distribution (gdt) and electricity (ely).
14 This is a GTAP requirement, rather than an SNA one.
15 Re-exports are removed from both imports and exports of the intermediate economy and a compensating change is made to the bilateral trade of the two partner countries. A re-export margin is then added to the exports of the intermediate economy to the ultimate importing country to take account of any service provided by the intermediate country. The following presentation provides an overview of the reconciliation of the trade data, including the treatment of re-exports: https://www.gtap.agecon.purdue.edu/events/Short_Courses/webcourse/modules/lectures/Trade_Data/Presenter/index.htm.
16 FIT was originally developed as part of the SALTER project at the Australian Industry Commission (James and McDougall Citation1993), and has since been extended for constructing the GTAP Data Base.
17 Composite regions are regions for which there are no contributed tables (identified by prefix ‘x’). An I–O table for a composite region is constructed by matching each country within the composite region with a country for which a national I–O table exists. This matching is done by identifying the country closest in terms of per capita GDP and within the same geographical region. This country's I–O table then acts as a proxy for the missing country.
18 Labor is also disaggregated into skilled and unskilled labor, although this does not result in any changes to the share of labor in total value-added.
19 The use of averages allows for comparison of the sales and costs shares by use and input. Using the summation of entropy scores would overstate the importance of the sales share difference since there are more inputs (117, as opposed to 61 uses) to sum over.
20 Note that these tables were obtained after agricultural targeting and hence they underestimate the potential changes that occur to agricultural commodities’ cost shares in those countries that undergo agricultural targeting.
21 The high entropy values on land in energy products reflect the fact that, when land payments are contributed, they are often included as an input in energy production. These values are removed by the CGTA. Instead, land payments are incorporated into the cost structures of agriculture, and natural resources into the energy commodities.
22 This is because the GTAP Data Base targets GDP. With GDP targeted and the trade balance determined by the reconciled bilateral trade data, absorption must adjust.
23 Note that change in stocks has been removed from the ranking, because the changes (in change in stocks) are all −100% (because they are removed). This does not remove the impact of removing change in stocks, since removing stocks will impact other use shares.
24 The analysis was done using Stata (www.stata.com).
25 We also find this explicitly in a regression where we use the number of agricultural sectors, instead of the number of total sectors, as an independent variable; we do observe a positive effect from the number of agricultural sectors on entropy.
26 Improvements have been made in this area in recent versions, with the CGTA strongly discouraging contributors from disaggregating agriculture based on inadequate information; preferring instead to disaggregate at the CGTA using FAO data.
27 Since our econometric analysis does not include older versions of these I–O tables, we are unable to test if I–O tables have improved over time.
28 Note, however, that total export and import figures reported by the IMF and the World Bank do not always match and are not globally reconciled, hence in some ways this merely raises further issues.