New MIRAGE version, incorporating NTBs for goods and services #

This new version incorporates a number of improvements progressively implemented since 2008 (but not all of them; some will be added later). It is still in progress, so you are advised to refer to this page regularly if you decide to use the new version, as new features will be added (this sentence will be removed when the version is final). The current version is pretty stable, future evolutions will only concern new features like the modelling of tariff rate quotas.

Simplifications #

In most cases, TRQ rents are considered to be captured by importers. This assumption has become the standard feature in Mirage. As Mirage only considers regional agents, it implies that no explicit transfer has to be modelled.
Unskilled labour is now fully mobile across sectors. While a single worker is definitely not fully mobile across sectors (architect can be pretty bad cardiac surgeons), the assumption of a global full mobility is reasonable. Shocks usually considered rarely imply the displacement of more than 1% of the labour force from one sector to the other in one year, very often much less than that. This remains far below the natural replacement of the labour force as a consequence of annual entry and exit (retirement). Furthermore, the rigidity that was assumed in earlier versions had poor dynamic properties.
Some special treatment for developing countries regarding imperfect competition calibration has been removed.
The computation of results has been simplified. A single ‘Indicateurs.gms’ file has been created, that summarises the information contained earlier in ‘Indicateurs.gms’, ‘Variations.gms’ and ‘Sorties.gms’; ‘Result_ini.gms’ and Result_var.gms’ have been replaced by a single ‘Res.gms’ file.

NTBs #

Non-tariff barriers have been introduced in the model. They are read in the ‘sets.gms’ program, and aggregated using a trade-weighting scheme to the level of the Mirage aggregation. The corresponding Trade.gms file has to be generated again (or downloaded again using the link provided above) each time a new GTAP database is released.

NTB_Ser is based on gravity equations, and available for all services sectors and all countries.
NTB_Se2 provides information on barriers to mode 3 trade. It is based on a two-step methodology. First a score is computed based on actual observed barriers. Then the impact of this score on the profitability of firms operating in the importing country is used to translate this score into an ad-valorem equivalent. The methodology used in this second step is pretty questionable. Therefore, it is advised to use NTB_Se2 only when modelling barriers to mode 3 is required.
NTB_Goods is based on Kee, Nicita & Olarreaga estimates at the HS6 level. These estimates are aggregated up to the GTAP level using a trade weighting scheme. This process transforms the original 2-dimension dataset into a 3-dimension dataset. The KNO dataset is rich but not complete. In particular, full GTAP sectors can be missing for all EU15 countries. This implies that no NTB will be proposed when Europe is disaggregated, and NTBs to the 10 new 2004 member countries are used for the whole EU in some significant sectors like motor vehicles, when the EU is considered as a single region.
NTB_Good2 is a variation on the KNO database. A border effect technique is used to differentiate EU exporters from non-EU exporters on the EU market. The KNO protection level is assumed to be an average level of protection. However, this technique does not reveal a very large difference between both groups of exporters.

In most cases, NTBs are modelled as an iceberg trade cost: when a good is produced for a foreign country, it takes more production factors (capital, labour, natural resources, etc) and more intermediate consumption to produce it. NTBs in services can also be modelled as an export tax, when the barrier is mostly a rent increasing barrier that favors companies which are allowed to export. The choice of specification is made in the Spreadsheet interface (Outdated).

Trade facilitation #

Trade costs associated to time have been incorporated in the model as iceberg costs. Data come from a database provided by Minor and Tsigas. If you decide to use them, please consider the following answer from Peter Minor when I asked him which paper to quote:

<blockquote> Dear Yvan,

The following three articles should be the complete set of references. The underlying methodology for estimating the value of time in trade is found in the first two articles. The last article provides the methodology for porting the values of time found in the 2007 article into GTAP.

Hummels, D., 2007. Calculating Tariff Equivalents for Time in Trade. United States Agency for International Development.
Hummels, D. 2001. Time as a Trade Barrier. [ http://www.mgmt.purdue.edu/faculty/hummelsd/research/time3b.pdf Unpublished manuscript].
Minor, P., and M. Tsigas, 2008. Impacts of Better Trade Facilitation in Developing Countries: Analysis with a New GTAP Database for the Value of Time in Trade. 11th Annual Conference on Global Economic Analysis, Helsinki, Finland.

Best, Peter </blockquote>

Three kinds of times are considered: time to accomplish customs procedures, time at the port, and transportation time. These three categories are considered both to import and export products. The valuation of time vary with sectors.

In practice, data are introduced in the model in the ‘sets.gms’ file.

Welfare decomposition #

The introduction of iceberg trade costs led to a new source of welfare improvement.

The reduction of import iceberg costs improves terms of trade for the importing country. This improvement is already accounted in the terms-of-trade component of the welfare decomposition. The reduction of export iceberg costs also improves welfare (and GDP). As this contribution was not accounted in the decomposition of welfare gains, a new component has been introduced, called ’trade-cost gains (exporter)’.