GTAP Course 7x5 aggregation : Base Model
This aggregation of GTAP uses a 7-good, 5-region aggregation to look at the economic implications of increased South African trade and investment on the rest of Sub-Sahara Africa.  In particular, two types of experiments are considered:

(1) Reduced trade barriers between Europe and South Africa, and within the Southern Africa region -- this research is
reported on the in paper titled: "Global Trade Analysis for Southern Africa" by Rob Davies, originally developed for 
the African Short Course in Global Trade Analysis, held in January of 1998. The PDF file of this paper is included in this version subdirectory. It also involves simulations conducted with several other versions of this data base -- each of them reflecting some kind of a post-FTA situation.

(2) Reduced risk premia and hence increased foreign investment in South Africa. The research behind this version is documented in the paper titled: "Modeling Country Risk and Capital Flows in GTAP" by Gerard Malcolm, GTAP technical paper no. 13, available from the GTAP web site.

******************************************************************************************

The 5 regions in this aggregation are:

 SAFRICA    SOUTH AFRICA
 RESTSAF    REST OF SOUTHERN AFRICA
 RESTSSH    REST OF SUB-SAHARAN AFRICA
 EUNION     EUROPEAN UNION
 RESTWLD    REST OF THE WORLD

The 7 goods are as follows:

 AGRIC      Agriculture
 EXTRACT    Natural Resource, Extractive and related ind.
 FOOD       Food manufacturing
 LITMNFC    Unskilled labor intensive manufactures
 TECHMNFC   Skilled labor intensive manufactures
 HVYMNFC    Capital intensive manufactures
 SVCES      Services

which are aggregated as follows:

AGRIC: Paddy rice, Wheat, Cereal grains nec, Vegetables, fruit, nuts, Oil seeds, Sugar cane, sugar beet, Plant-based fibers, Crops nec, Bovine cattle, sheep and goats, horses, Animal products, Raw milk Wool silk-worm cocoons, Bovine cattle, sheep and goat, horse meat prods,

FOOD: Meat products nec, Vegetable oils and fats, Dairy products, Processed rice, Sugar, Food products nec, Beverages and tobacco products

EXTRACT: Forestry, Fishing, Coal, Oil, Gas, Minerals nec, Petroleum, coal products

LITMNFC: Textiles, Wearing apparel, Leather products, Wood products,

HVYMNFC: Paper products, publishing, Chemical, rubber, plastic products, Mineral products nec, Ferrous metals, Metals nec,

TECHMNFC: Metal products, Motor vehicles and parts, Transport equipment nec, Electronic equipment, Machinery and equipment nec, Manufactures nec

SVCES: Electricity, Gas manufacture, distribution, Water, Construction Trade, transport, Financial, business, recreational services, Public admin and defence, education, health, Dwellings & Svces


EXPERIMENTS RELATING TO ELIMINATION OF TRADE BARRIERS:
(SEFTA1.CMF - SEFTA8.CMF, implemented using  SEFTA1.exp - SEFTA8.exp -- some of which involve other versions
of RunGTAP -- since the initial data base is different).

This is a series of eight experiments designed to address various dimensions of South Africa's trade relations with Europe and the rest of Southern Africa.  The text below describes the issues to be addressed in relation to the South Africa - EU Free Trade Agreement.
Each issue will be addressed by comparing results from different model runs.  The runs will be different in terms of the shocks applied.
In the table below, an 'x' in any cell means that the trade barrier described in that row will be removed in the Command File corresponding to the number at the head of that column.

Issues:
A. What is the effect of including agricultural products in an SA-EU FTA?
=> compare 1 with 2

B. What is the difference between unilateral opening (by EU) and bilateral opening?
=> compare 1 with 3, or 2 with 4

C. What is the effect of removing export subsidies as well as import tariffs?
=> compare 4 with 5.

D. How does the SA-EU agreement interact with a Southern Africa FTA?
=> compare 3 (or 4), 6 and 7 (or 8)

Table describing trade policy experiments:
____________________________________________________________
Trade instrument removed                                                            Experiment #
_____________________________________              _________________
                                                                                        1   2   3   4   5   6   7   8
Non-agricultural import  tariffs from SAFRICA to EU         x   x   x   x   x         x   x
Agricultural import tariffs from SAFRICA to EU                       x       x   x              x
Non-agricultural import tariffs from EU to SAFRICA                   x   x   x          x   x
Agricultural import tariffs from EU to SAFRICA                                x   x              x
Non-agricultural export subsidies from SAFRICA to EU                      x
Agricultural export subsidies from SAFRICA to EU                              x
Non-agricultural export subsidies from EU to SAFRICA                      x
Agricultural export subsidies from EU to SAFRICA                              x
Import tariffs from RESTSAF to SAFRICA                                                  x   x   x
Import tariffs from SAFRICA to RESTSAF                                                  x   x   x
______________________________________________________________


Experiment relating to investment inflows:

In this experiment, the required rate of return by foreign investors in South Africa falls, thereby stimulating an inflow of foreign investment.  With S - I falling, so too, must X - M fall, and export are cut while imports increase.

The reader is referred to technical paper No. 13 on the GTAP web site for detailed motivation of this experiment.

In general, increases in direct capital inflows occur because the investment climate becomes more attractive.  In the case of South Africa in its transition from the apartheid system to a democratic system, this may be because
* negative impacts on the global operations of companies no longer occur as a result of investment in South Africa,
* the risk premium associated with investment in South Africa is lower under a more stable political climate,
* legal restrictions on investment no longer apply, or
* the economic outlook for South Africa is improved.


CLOSURE REQUIREMENTS:
=====================
In order to use cgdslack as a proxy for the risk premium in this way, the model closure must include cgdslack as an exogenous variable (this is the usual GE closure), and the parameter RORDELTA must be equal to 1 in the parameter file (this is already the case in PARAMS.TXT).

1. RISK   	is the regular GE closure, used in the RISK experiment. 

2. RORE	is used for the RORE and ROREBACK experiments: cgdslack(safrica) 
        is 'swapped' with rore(safrica).


Discussion of modelling method (by Gerard Malcolm)
===========================================

We assume that the global bank equalizes expected risk-adjusted rates of return, so that risk-adjusted rates for all regions are equal to some global average.
	RORE(r) / RISK(r) = RORG
where, in accordance with GTAP notational convention, these capitalized variables represent levels, while lower-case variables represent percentage rates of change from initial levels. 
	RORE(r) is a non-risk-adjusted expected rate of return, i.e. it is the expected rate of return in the absence of any default by the borrower.
	RISK(r) represents the ratio of equilibrium returns in region r to the global average rate of return.  For relatively high-risk countries, this ratio will be above 1, and for relatively safe countries below 1.  It is important to note that this variable represents a ratio rather than a certain number of basis points - it is better called a risk ratio' than a risk premium'.
	RORG does not represent a risk-free return but a weighted average of returns around the world.  This formulation differs from the more familiar representation of required rate of return in a country being equal to the risk-free return plus some risk margin.
If we rewrite this as
 	RORE(r) = RORG x RISK(r) 
then by total differentiation and division through by RORE(r) we can obtain
 	rore(r) = rorg + risk(r)
where these variables are percentage changes in their levels equivalents.  This is the analogue of equation (11') in the standard GTAP model in the case where RORDELTA = 1:
 	rore(r) = rorg + cgdslack(r)        

This equation states that the percentage change in the rate of return on investment in region r is equal to the percentage change in the global rate of return plus a disequilibrium factor which is generally exogenous and set at zero in a general equilibrium closure.  Normally, the cgdslack variable is only non-zero when we allow disequilibrium to exist in the market for capital goods.  The main proposition of this paper is that cgdslack can be interpreted to represent a risk premium as defined above, although it was not originally designed for this purpose.  In a general equilibrium closure, cgdslack is unused for any other purpose (being exogenous and unshocked), and therefore we do not disturb any other components of the model by using it in this way.
In the following sections, we continue to refer to cgdslack rather than risk, but interpret it as being equivalent to the percentage change in the variable RISK as defined above. 
Implementation
There are two alternative closures that can be employed.  Closure 1 is the standard GTAP closure, with cgdslack exogenous and rore endogenous for all regions.  This closure can be used under certain conditions described below.  Closure 2 involves swapping' these two variables for the region(s) of interest, so that cgdslack is endogenous for the region/s of interest and exogenous for all other regions, and rore is exogenous for the region/s of interest and endogenous for all others.  No other swap' is required.  In particular, the general equilibrium nature of the closure is preserved, and this means that it is not necessary to endogenise psave.  Which of these two closures is more suitable depends on the experiment to be performed.
In either case, a shock will be imposed on the exogenous variable (cgdslack in Closure 1 or rore in Closure 2).  For an historical simulation experiment (e.g. the South African application described in the following section), information must be available which allows the modeler to calculate an appropriate shock to the exogenous variable.
For Closure 1, it is necessary to have information on the pre- and post-shock values of the risk ratio in the region(s) of interest, or acceptable proxies thereof.  Given that this risk ratio is not a widely-used concept, it may well be the case that such information is unavailable. 
For Closure 2, in order to determine an appropriate shock for rore(r) it is necessary to know what the pre- and post-shock expected rates of return in the region(s) of interest are.  For this closure, it is not necessary to know either the global required rate of return or the risk ratio.  As noted above, the expected rate of return excludes any possibility of default.  A suitable proxy for this variable is more likely to be available (this is case for the South African application, where risk premia on DM-denominated bond issued by the South African government are used).
In a situation where the region of interest r is a small recipient of investment relative to the global total, we can assume that cgdslack(r) and rore(r) are approximately equal, because a change in rore(r) will have little impact on rorg.  In this case, we can use Closure 1 even if we are not able to calculate an appropriate shock for cgdslack.  The results will be approximately equivalent to those obtained from shocking rore(r) under Closure 2.  In the South African experiment, the results obtained under both closures are very similar.
For a hypothetical experiment that does not involve replication of historical events, data availability is no longer the determining factor in which closure should be used.  The modeler may legitimately impose a shock on either variable.  If the modeler wishes to impose a shock on cgdslack, then Closure 1 should be employed.
In summary, Closure 1 can be used if either of the following two conditions are satisfied:
	a shock can be imposed on cgdslack, or
	the region of interest is small.
Otherwise, Closure 2 should be used.  This closure is set up by adjusting the set of exogenous variables specified in the command file.  The swap' can be made for multiple regions.  However, it cannot be made for all regions simultaneously, because this prevents any equilibrating adjustment in rorg from occurring, and no solution is possible.
We wish to characterize the global bank as equalizing risk-adjusted rates of return across regions.  This requires that RORDELTA be set equal to one in the parameter file, regardless of the closure specified.

