The CRUSOE Suite

Includes executable versions of three standard CGE models which all work with 346 databases following the same format. 1997 databases corresponding to 78 different regions of the world are supplied in each of 10-sector, 39-sector or 57-sector aggregations -- these have been adapted from version 5.4 [Interim_Release_5.4_Aug03] of the GTAP database. As well, 2004 databases for 112 regions of the world are supplied in at 10-sector level, from version 7.1 [R7.1_17May10] of the GTAP database. A Windows program, CRUSOE45, is supplied which enables you to create similar databases with more or different sectors -- providing you have access to the GTAP database (either version 4, 5, 6 or 7).

Prerequisite: basic GEMPACK knowledge is assumed.

Update October 2011: More detail about factor taxes (see end of this page)

Update May 2011: Programs and data updated.

Update March 2008: The CRUSOE45 program has been amended to work with the more modern CHM help file format. Output file now contains a new summary producer prices IO table.

Update September 2004: The CRUSOE45 program has been amended to work with [pre-release] version 6 of the GTAP database.


CRUSOE - a simplified version of ORANI-G
MINIMAL - an even simpler model
Countries and Sectors of CRUSOE
Using the CRUSOE45 windows program to create more CRUSOE databases
Downloading and Installing CRUSOE
Further Background Information

1 Introduction

The ORANI-G CGE model is based on the Australian ORANI model described in the book by Dixon, Parmenter, Sutton and Vincent. ORANI-G is used as the basis of an annual course conducted by CoPS, which simultaneously introduces participants to the theory of general equilibrium models and to the techniques of solving them with GEMPACK. A monograph is available which describes and explains the model. There is a web page devoted to the model.

ORANI-G has proved an excellent starting point for the construction of other single-country models. Various researchers have adapted it to a number of countries including Denmark, Philippines, Chinas, Korea, Indonesia, Sri Lanka, Malaysia, Brazil, Vietnam, Pakistan, South Africa, Fiji and Thailand.

The CRUSOE project extends the usefulness of ORANI-G by providing for it a series of databases covering 78 countries or world regions. These databases are drawn from Version 5.4 [Interim_Release_5.4_Aug03] of the GTAP database. The same TABLO-generated model works with any of the 78 databases. Also, a Windows program, CRUSOE45, is supplied which enables you to create similar databases with more or different sectors -- providing you have access to the GTAP database.

The GTAP model must apply the same model structure and level of sectoral detail to all regions. This exerts pressure towards a rather simple, 'lowest common denominator' data base. For a single-country CGE model,a more comprehensive database could usually be compiled -- given time, effort and skill. The CRUSOE databases could be useful when these resources are unavailable, or as a makeshift which would allow preliminary modelling efforts to proceed while a larger database is being constructed.

Some features of ORANI-G are not supported by data derived from GTAP. A simplified version of ORANI-G, called CRUSOE, is provided, which omits these features. The CRUSOE databases work for both ORANI-G and CRUSOE models, as well as for MINIMAL, a still simpler model devised for instructional purposes.

Another interesting possibility is to use either the CRUSOE or the ORANI-G model to examine the effects of the same shock on a variety of different regions. There are few other AGE models for which such a variety of directly comparable databases are available.

IMPORTANT: Not all the GTAP data for each region is fully updated each time a new GTAP database is released. Before relying too heavily on a national database extracted from the GTAP data, you should find out (from Purdue) how old is the original input-output table that GTAP used for your region. Please read these cautionary notes about what the GTAP Data Base is NOT !

2 CRUSOE - a simplified version of ORANI-G

Although the GTAP model boasts unexcelled richness of detail in its treatment of world trade flows and distortions, ORANI-G offers a more elaborate treatment of individual countries. Thus, some features of ORANI-G are not supported by data derived from GTAP:

The CRUSOE databases accommodate the data needs of ORANI-G either by providing appropriately dimensioned matrices of zeroes or by applying default rules to expand the GTAP database. See below, Further Background Information. An alternative approach is to simplify the theory of ORANI-G so that only data included in the GTAP system are required. Following this approach, the CRUSOE suite also includes a stripped-down version of the ORANI-G model, named CRUSOE. The CRUSOE model:

The result is a simpler, leaner model than ORANI-G which may well be more suitable as a starting point for constructing a new model. The disadvantage is that it is not so well documented. However, since the notation used by CRUSOE is the same as that used by ORANI-G, the majority of the ORANI-G documentation (supplied in Microsoft Word format) applies to CRUSOE as well, although some sections will no longer be relevant.

Any of the 78 regional databases may be used interchangeably with either the ORANI-G or CRUSOE models. Not all of the data items on the databases will be used by every model. In addition, each database includes summary tables of costs and sales for each sector, both as flows and as percentages of sectoral output. These tables are included for convenience only: they are not required by either model.

3 MINIMAL - an even simpler model

The Crusoe databases can also be used by the MINIMAL model which been developed for teaching purposes. See: here for more about MINIMAL. Compared to ORANI-G or CRUSOE, MINIMAL makes the following simplifications:

The version of MINIMAL that you can download below has been modified for use with the CRUSOE database: it is slightly different from the standard version of MINIMAL. The chief change is the addition of runtime sets, so that it can be used with databases containing different numbers of sectors. Two of the headers [1PTX, 0TAR] have different names. And, the convention about sign of the export demand elasticity is different. These differences will become relevant if you want to use CRUSOE data with the standard MINIMAL.

4 Countries and Sectors of CRUSOE

Version 5.4 of the GTAP database contains 57 sectors and 78 regions. These are listed at headers OSEC and OREG in each data file. For completeness, a CRUSOE data file has been created for every GTAP region, even though some of them are composite regions which would form a strange basis for a single country model.

The older Version 5.4 [1997] of the GTAP database is now in the public domain. Since GTAP derives revenue from selling data, it is not possible to freely distribute extracts of their recent data which distinguish more than 10 commodities. According, the 112 single-region CRUSOE databases from GTAP 7.1 (2004), which originally contained 57 sectors, have been aggregated to 10 sectors. The mapping used for sectoral aggregation is shown at header MAPS in each data file.

If you purchased the GTAP database, you could use the GTAPAGG and CRUSOE45 Windows programs (described below) to create similar databases with more or different sectors. To purchase the GTAP database, or to get more information about the GTAP sectoral and regional classification, contact GTAP.

5 Using the CRUSOE45 windows program to create more CRUSOE databases

If you have purchased any of versions 4 to 8 of the GTAP database, you can use CRUSOE45 to create more single-country databases in the Crusoe format. It assumes that you have used the GTAP program GTAPAgg (which is supplied with the GTAP database) to create a ZIP archive containing the following 3 files:

The above 3 files must follow the standard GTAP formats. These formats changed slightly between versions 4 and 5 of the GTAP data base. CRUSOE45 accepts either format.

The RunGTAP command File..Version Archive..Create Zip can also be used to create a ZIP archive containing the above files.

Suppose you wanted to create an ORANI-G database for Poland. You would first use GTAPAGG to create a GTAP database which (a) distinguished Poland as a separate region, and (b) distinguished the sectors you wanted. GTAPAgg would store this database in a ZIP. Then you would use CRUSOE45 to open the ZIP, select Poland, and create a HAR (header array) file which could be used by any of the ORANI-G, CRUSOE or MINIMAL models.

Note: If you do NOT have any GEMPACK licence, you should aggregate the GTAP sectors to 40 or fewer -- then you can run the MINIMAL and CRUSOE models without any licence. For ORANI-G, you might need to aggregate the sectors even more.

Download CRUSOE.EXE (5 Mb) and save it in a temporary directory. Then run CRUSOE.EXE to install the CRUSOE45 program.

6 Downloading and Installing CRUSOE

The versions of ORANI-G and MINIMAL that you can download below have been modified for use with the CRUSOE database. The chief change is the addition of runtime sets, so that they can be used with databases containing different numbers of sectors. ORANI-G comes in two flavours, ORANIG98 and the more modern version, ORANIG01.

Create a CRUSOE subdirectory on your hard disk, and download and unzip some or all of the following files there. The first download contains source (TAB), executable (EXE, AXS and AXT) and DOC files for each of the 4 models [ORANIG01, ORANIG98, CRUSOE and MINIMAL]. The other downloads contain numerous sample databases which work with any of the four models.

CrModels.ZIP (13.4 MB) source and EXE files and documentation for all four models.
crdata5_10.ZIP (806 KB) 10-sector data files for each of 78 GTAP 5.4 regions.
crdata5_39.ZIP (4.8 MB) 39-sector data files for each of 78 GTAP 5.4 regions.
crdata5_57.ZIP (9 MB) 57-sector data files for each of 78 GTAP 5.4 regions.
crdata7_10.ZIP (1.2 MB) 10-sector data files for each of 112 GTAP 7.1 regions.

The 10-sector files may be useful for teaching, while the 39-sector versions are the largest usable without a GEMPACK licence. The original sectors and the mapping used for sectoral aggregation are shown at headers OSEC and MAPS in each data file.

CrModels.ZIP contains 3 long MsWord documents describing ORANIG98, ORANIG01 and MINIMAL models. There is no separate document for the CRUSOE model but most of the ORANIG98 document will apply.

If you do not have GEMPACK, you will also need to obtain ViewHAR and ViewSOL, programs for viewing the binary data and solution files used by GEMPACK. See GEMPACK helper programs.

6.1 Running the models

The 4 CMF files which are supplied (in CrModels.ZIP) all refer to a single-country data file called INPUT.HAR. Before running a model you will have to create such a file. For example, if you wished to conduct experiments for the USA, you should issue the DOS command:


You can run ORANIG01.EXE by typing:


After a short time, the solution file ORANIG01.SL4 should be produced, which you could view using ViewSOL.

The supplied CMF files compute the effects of 1% increase in real wages in a short-run closure. They have been prepared in such a way that the same numerical results are computed for CRUSOE and both ORANI-G models. That is, running ORANIG98 with ORANIG98.CMF yields the same results as does running CRUSOE5 with CRUSOE5.CMF. Results from MINIMAL are likely to be similar but not identical to those from the other models.

Each simulation also produces an updated (post-simulation) file suffixed UPD. These UPD files can be used as the input for subsequent simulations: they contain only the data items actually used by the model. Thus if you wanted to use the MINIMAL model and wished to remove data that was used only by ORANI-G from your input data file, you could edit the CMF file to set the shock value to zero. The resulting UPD file would be much smaller than the original input file, but all the remaining numbers would be identical to the original.

7 Further Background Information

This section gives a few details of how the GTAP database was converted to ORANI-G format. The principal change, of course, was that for each country, all exports and imports, which are distinguished by destination and source in GTAP, were aggregated into one. Trade taxes were aggregated in the same way. Each country's expenditure on shipping services was treated like an ordinary export.

The investment demand column in the CRUSOE database was formed from the inputs to GTAP's capital goods sector. ORANI-G requires a investment demand column for each industry: this was formed by splitting the single investment column in proportion to capital rentals. The same procedure was used to derive the industry-specific investment totals used by CRUSOE.

The margins matrices and inventory column needed by ORANI-G are simply matrices filled with zeroes. The MAKE matrix is diagonal. None of these is required by CRUSOE or MINIMAL.

The GTAP primary factors "NatRes" (aka NatlRes) and "Land" were combined to become the ORANI-G primary factor "Land".

Where possible, elasticities were derived from GTAP. For households, Frisch and LES expenditure elasticities were derived from budget shares and CDE parameters. The domestic/import Armington elasticities from GTAP were used in the same way for ORANI-G. The import/import Armington elasticities from GTAP were used, in ORANI-G, as a proxy for export demand elasticities.

In GTAP, industry outputs are subject to production tax, and also to user-specific usage taxes. The former became, for ORANI-G, the 'other cost tickets' while the latter became the commodity tax matrices. So for ORANI-G, the production taxes enter into the value of output, whilst for GTAP they do not. For the most part, this difference has little economic significance.

The MINIMAL model has a greatly simplified treatment of commodity taxes. Its database contains only 2 vectors of indirect tax revenues - one for domestic commodities, and one for imported. An implication is that, in MINIMAL, commodity tax rates are not user-specific (as they are in GTAP, ORANI-G and CRUSOE5). Thus, information is lost in converting to the MINIMAL format. Also, the GTAP primary factors "Capital", "NatRes" and "Land" were combined to become the MINIMAL primary factor "Capital".

The ORANI-G, CRUSOE and MINIMAL models all define GDP from the income side as

GDP = Factor Income + Indirect Taxes

and none of these models have any treatment of direct taxes. Nor do they allow for industry-specific taxes on particular primary factors.

The GTAP database can incorporate income taxes, and does allow for industry-specific taxes on particular primary factors. The GTAP model does not draw a distinction between direct and indirect taxes: it defines GDP from the income side as

GDP = Net Factor Income + Net (Direct+Indirect) Taxes + Depreciation

One might therefore ask, how do, say, industry-specific taxes on land appearing in the GTAP data get converted to ORANI-G form? Are they Direct or Indirect taxes?

The GTAP relation between personal incomes and firms' factor costs may be depicted as:

    EVOA  Payments to owners of factors
+ INCTAX  Income Tax (PTAX)
     VFM  Factors at market prices
  + ETAX  Industry-specific factor tax
    EVFA  What industries pay for factors (tax-inclusive)

The factor costs appearing in the ORANI-G, CRUSOE and MINIMAL data correspond to factor prices faced by firms, and are equivalent to the GTAP EVFA concept. They include,for example, GTAP's industry-specific taxes on land, and GTAP's factor income taxes. Hence, all land taxes are implicitly treated for ORANI-G purposes as Direct. Capital costs include depreciation. Several extra headers are included to help understand the connection between GTAP and ORANI-G concepts:

ETAX  Tax on use of endowment good i by industry j = EVFA - VFM
FACR  Factor income report
EMAC  GDP Expenditure Aggregates
TMAC  Indirect Tax Aggregates
IMAC  GDP Income Aggregates

The first two of these headers could be used to extend to ORANI-G GTAP's additional detail about primary factor taxation. However, the "additional detail" may be incomplete or crude, especially in older GTAP databases. Please read Chapter 13 of the GTAP documentation: Income and Factor Taxes (McDougall and Hagemeyer) for further information.

The data conversion was done using 2 TABLO programs. The source code for these, in files GDATA4.TAB (for GTAP 4 format data) and GDATA5.TAB (for GTAP 5 or 6 format data), are included in the CRUSOE package.

Complaints and suggestions to

Go back to
Links to other models and programs by Mark Horridge
CoPS Home Page (bottom of this page has "Search COPS site")