Agriculture Reference
In-Depth Information
and investment that are important in China, which has large skilled and
unsklled labour forces that are ncreasngly moble between sectors.
15
Finally, the sectoral breakdown we have chosen aggregates the 57 sectors
in the database to our more manageable 14, offering the most detail in
agrcultural and marne products.
The most sensitive parameters in determining the trends in self-sufficiency
through tme prove to be the elastctes of demand. Our model employs the
orgnal GTAP CDE (constant dfference of elastctes of substtuton) system.
Its non-homothetcty s a partcular asset for our purpose n that t permts a
range of income elasticities to exist, either side of unity. While this system is
more general than the homothetic ones often used in such models, it is still
restrctve n the wdth of the range compared to stll more general systems.
The CDE system s employed here because of ts parametrc economy.
16
The
ncome elastctes thus emboded n our demand parameters are lsted n
Table A8.2. Because of the restrictiveness of the CDE system the lower bound
for the income elasticity of cereal grains cannot be set below 0.1, despite
evdence suggestng that t s now negatve (Ito et al. 1991; Peterson et al.
1991). As a result, in our analysis the span between the income elasticities of
livestock products and processed foods, which are superior goods, and those
of cereal grans s lkely to be smaller than the truth. One consequence of
ths s that our results probably underestmate the relatve growth n demand
for lvestock products and processed foods and hence they underestmate
the assocated derved demand for cereal feeds and other agrcultural
nputs. Ths offers a second downward bas n our estmates of the cost of
achieving and maintaining agricultural self-sufficiency. Having said this, we
dd recalbrate the CDE parameters so as to mnmse ths downward bas.
17
The span between the ncome elastctes of lvestock products and cereal
grains was increased from 0.8, which was based on 1997 numbers in the
GTAP Version 5 database, to 1.5. The recalibration involved calculating the
new values for the expanson parameter and the substtuton parameter n
the CDE mnmum expendture functon to mply ncome elastctes close
to our target values.
Because the simulations are decade-long projections, it was necessary to
use long-run elastctes of substtuton n product and servce demand. These
are listed in Table A8.3. They are larger than the standard GTAP demand
elasticities, which are designed for simulations over the medium run.
