Having settled on the Armington specification for international
trade, it remains to determine the size of the so-called ‘‘Armington
elasticities’’ of substitution amongst imports from competing
sources of supply. For this, we rely on the econometric study of
trade elasticities published by Hertel et al. (2007) who use
variation in bilateral trade and transport costs in order to estimate
these elasticities. These point estimates and the associated
standard errors are a direct input to our study. In addition, we
conduct a SSA with respect to four sets of key parameters that
govern land mobility, substitution in production, consumption and
trade in the model. The SSA varies each parameter over the range:
+/ 30% of the baseline value. We assume the variation takes the
symmetric triangular distribution. This permits us to place
confidence intervals on the resulting production (Fig. 2) and
welfare impacts (Table B1) to reflect the uncertainty inherited from
the behavioral parameters in the model. The error bars show
reasonably small deviations from the point estimates, suggesting
that our findings are robust.