GTAP 9 Data Base

Overview
Sector Listing
Region Listing
Documentation
Data Issues
License Agreement
Price List
Order Form
Data Base Archives
FlexAgg2
Technical Support

Data and Utilities

Overview
Free Data & Utilities

Africa Data Base

Overview
Order Form

Contribute Data

How to Contribute?
I-O Table Submission

GTAP Data Bases: GTAP 7 Data Base Issues

If you have encountered an issue with the GTAP 7 Data Base which you would like addressed, please email a detailed account to the Data Team.


IO Tables
Sales Dispositions of Sugar Crops and Raw Milk
Martin Banse reported some strange uses of sugar cane and beet and raw milk. We had looked into this and found:
  • the main source of the problem is the contributed tables;
  • in some cases, sub-optimal disaggregation technique in house may be a contributing factor;
  • some of the strange uses may be valid (for example, use of sugar cane by the chemicals industry in Brazil).
A response to this has been prepared. Subsequent to this note, some improvements have been made to the IO table of India in-house and for Mexico and EU member countries by the contributors.

Proposed Action: To be fixed in GTAP version 8 Data Base cycle.


Low Share of Capital in the Oil Industry in ‘rest of Western Asia’ (XWS)
Review of a pre-release of GTAP 7 Data Base showed that the share of capital earnings in oil industry costs in ‘rest of Western Asia’ was about a sixth of the normal level. The reason is that main primary region proxy for the composite XWS is Cyprus, and the Cyprus table has a low capital share.

Proposed action: none.


Low Share of Capital Industry in the Electricity Industry in Russia
Truong Truong reported an extraordinarily low share of capital earnings in the cost of electricity production in Russia (about 3%), about one tenth of the normal level (about 30%) in a pre-release of GTAP 7 Data Base. Possible reasons include implicit subsidy of electricity industry capital, or implicit subsidy of its energy inputs. A new table was contributed for Russia later on and now this share has increased to above 8% in the final GTAP 7 Data Base.

Proposed action: none.


More Issues with PCR and PDR
David Laborde pointed out some issues with the usage patterns across the regions in the GTAP sectors PDR and PCR. First there are cases where pcr is mainly autoconsuming PCR, instead of consuming PDR. It is an issue, since the pcr sector is disconnected from the agricultural sectors and the land market. But since the land for rice is quite different from other land types, we can expect that CGE implications of this data issue are rather limited. Robert McDougall, in his response, noted that this is not an issue if it arises from the contributed table, as auto-consumption can be netted out, unless there are commodity taxes that act like production taxes. However, if it results from problems in I-O table construction, it's likely that those problems have resulted in other and more substantive defects.

There are other important issues in the usage across many regions for PCR. Given the possibilities of joint production, usage issues in PDR are not so important, but those in PCR are. Some of these issues originate from the contributed IO tables, while others arise from the issues in the database construction procedure that involves agricultural sectoral disaggregation.

Proposed Action: Revise and improve the agricultural disaggregation module in the databse construction process and always aggregate PCR and PDR in the contributed tables and then do the disaggregation in-house, for the GTAP v8 Data Base cycle. The same procedure needs to be applied for all the tables, including those used to construct the representative IO tables.

Our usual approach to severe cost structure problems in the contributed table is first (A) to raise them with the contributor; if that is not productive, then (B) to aggregate the problem sectors with selected other sectors in the contributed table. Then the disaggregation program applies independent data to disaggregate them. Unfortunately, plan B would not work so well here, because of the disaggregation program limitation described above.

We have in the design stage a new approach to I-O table disaggregation that would remove that limitation, and address other known issues. It involves using the "representative table" as a source of cost shares for non-agricultural inputs into rice growing and rice processing. It does unfortunately entail a thorough reorganization of the existing program; I hope nevertheless we can implement it for release 8.


Myanmar I-O Table
David Laborde found that there is no value added in several sectors in Myanmar I-O table.

Proposed Action: We are currently looking for external reviewers/new contributors to revise this table, but unless a new contributor steps forward to help us fix this table it will be removed.


Unusual Sales Shares
Betina Dimaranan found several issues related to unusual sales shares in different I-O tables, particularly focusing on the disposition of several agricultural products as inputs to production of other sectors. Here is this summary report.

Proposed Action: Terrie Walmsley is working with the contributors to improve the IO tables in the future releases. Currently, our IO table check programs consider only the cost-shares for entropy comparisons, while we need to consider the sales shares to identify and fix the problems similar to the ones noted by Betina Dimaranan.


Protection Data
Land Payments Data
Wally Tyner observed that land payments (called direct payments in USA and paiement compensatoire in the EU) are a part of OECD Other Transfer Payments (OTP), which are allocated across factors and sectors based on their shares to preserve the powers. However, these land payments need to be allocated separately to land alone, not across the factors as is being done now. However, Hsin Huang opines that doing this would involve some judgment on the economic impact of particular programs within OTP, which is not advisable for preserving the neutrality of the data base.

Status: This cannot be addressed in the standard data base as it involves policy judgments and is more of a research issue. However, interested users may make use of this data on detailed payment categories to modify GTAP tax/subsidies data according to economic policy realities. OECD statisticians have created this Excel pivot table containing all the PSE data by program. In the first sub-sheet, the data for EU and US (and other countries not expanded) for the category in question OTP, which is the category of payments based on non-current production parameters, with no obligation to produce.


Merchandise Tariffs and Travelers' Expenditures
The paddy rice investigation also brought up a point previously made by David Laborde, that we should more consistently respect the distinction between cross-border trade and travelers' expenditures in data handling; in particular, that we should apply the merchandise tariffs and the IEA energy trade volumes to cross-border trade only, not to total trade. This also is discussed in the paddy rice note.


US Cotton Subsidies
Gaspar Frontini (European Commission) noted some discrepancies in the computation of US cotton subsidies in GTAP 7 Data Base. This has led to further in-house investigation with the finding that given the current level of aggregation at which the OECD domestic support data is being contributed, it is not possible to arrive at an accurate figure for cotton subsidies, but only for the subsidies in the GTAP sectors PFB, V_F and OCR put together. Users are recommended to perform Altertax simulations to adjust any such rates, by reviewing papers such as Anderson and Valenzuela (2007) that modify the GTAP Data Base to incorporate this difference in domestic support by increasing the power of support to 1.4 as against 1.1 in GTAP Data Base Version 6 Data Base.

Proposed Action: For GTAP 7.1 Data Base, this has been fixed as the OECD give us the break-up among these three sectors for domestic support. This will also be taken care of in the future releases.


EU Agricultural Export Subsidies in GTAP 7 Data Base
The following link provides a note that summarizes the discussion among GTAP staff and data contributors before deciding the source to be used to compute EU agricultural export subsidies.

Some Discussions on EU Agricultural Export Subsidies in GTAP 7 Data Base


Trade Data
US Wheat Exports
David Laborde pointed out that GTAP Data Base shows US exports in wheat above USD 6.5 billions, while the actual figure for 2004, according to USITC, USDA and FAO is around USD 5.1 billions. Therefore, the exposure of US farmers to world market could be strongly overestimated in GTAP 7 Data Base.

Proposed Action: Robert McDougall and Mark Gehlhar will look into this issue.


Other Issues
Classification of Rice
In connection with a discussion on GTAP-L, Robert McDougall found that we wrongly classify husked rice as paddy rather than processed. A note provides further detail. Given the need to remap the trade and tariff data, revision may be a matter for a future release.

Proposed Action: To be fixed in a a future release.


Macro Data
During the inclusion of updated EU IO tables, discrepancies between macro data from EUROSTAT and GTAP data were discovered and demanded an investigation about the reason for these differences.

Our primary source for macroeconomic data comes from the World Bank. The World Bank relies on National Income Account (NIA) data from the OECD for high-income countries. Besides the expected discrepancies due to different harmonization procedures of the OECD and EUROSTAT, we were able to confirm that OECD NIA data has been updated and therefore, we updated GTAP’s macro data sets.

During this investigation, we were also able to fill in some actual data where before we relied on estimations. In version 7, we lacked private and government consumption (C and G respectively) for Belgium, Cyprus, Luxembourg, and Malta and relied on estimates. Apart from these, Hungary, Greece, and Bulgaria are countries for which the updated macro data significantly differs from previous release.

Proposed Action: In future, we expect to substitute these estimates with actual data.