A Bayesian approach to analyze regional elasticities

INFORMATIVE PRIOR DIFFUSE PRIOR

TABLE 4 - Elasticities conditional on 1989 (Y,t,w/r) with inequality constraints

Moments are quite dissimilar with a diffuse prior. For instance the expected
capital labor substitution elasticity for the Center is double than with a proper prior, but
no general pattern can be easily detected. However, it is very interesting to notice how
the posterior odds in favor of concavity (given by the ratio of accepted samples) varies
with regions. With a proper prior, in North the probability is in line with earlier findings
by Chalfant and Wallace (1992) and is approximately about 0.538, while it declines in
Center (0.397) and further in South (0.357). This is even more striking when we adopt a
diffuse prior. Odds are very against concavity in South (0.043) and Center (0.102),
while in North it is equal to 0.213. We can notice how neoclassical theory deteriorates
in Southern regions. This is coherent with our previous findings since elasticities
densities mostly don’t cover neoclassical space in South. From an economic point of
view, we notice how elasticities are higher in North but still quite low and that
technology in South is very close to a Leontief production function. I would like to
remark that, contrary to Guilkey and Lovell (1981), even with substitution elasticities

15

<<   <   12   13   14   15   16   17   18   >   >>