The name is absent

Table 3. Price, Cross-price, Pork, and Exchange Rate Index Elasticities for U.S. Soybean
Exports

Ordinary Least Squares Estimation                                                       ----------Elasticities^b..........

-.32       .13       1.14       -.68

Two-stage Least Squares Estimation

-.10      -.45       1.10       -.64

• Average share of the U.S. export market, 1983-85 (Appendix).

^b Calculated at the sample means (Appendix).

^c Elasticities times market share, computed from unrounded data.

" Implausiblesign.

^e Mexico, Portugal, Israel, Switzerland, Canada, Norway, Greece, Indonesia, and Egypt.

¹ Nineteen countries that imported most of U.S. soybeans (Appendix).

markets (Table 3). The elasticities were summed
two ways: first by totaling all that had the
expected sign, and then by including the
implausibly-signed estimates, which changed
the price, cross-price, and exchange rate index
elasticity estimates by 22, 37, and 60 percent,
respectively.

Elasticity estimates from the six-equation
estimation may contain elements of aggrega-
tion bias from the EC and ROW equations.
Within the six country markets aggregated for
the EC equation, one might expect similarly
sloped expansion paths at various levels of pork
production because these EC countries are
geographic and economic neighbors and have
similar standards of living. Consequently, one
may not expect serious aggregation bias effects
in the EC equation. The ROW equation, how-
ever, contains nine diverse countries which span
continents and range from developed to devel-
oping economies (countries in Appendix). Ex-
pansion paths at various levels of pork produc-
tion could not be expected to be as similar across
the ROW countries as in the EC. Hence, one
would expect greater effects of aggregation
bias in the ROW equation than in the EC equa-
tion. However, in the six-equation weighted-
market-share elasticity estimation (Table 3),
the ROW market share is only 19 percent, and
the ROW elasticities are weighted accordingly.

RESULTS

Price and cross-price elasticity estimates from
the SUR six-equation estimation are closer to
those from the OLS single-equation estimation
(which probably contains aggregation bias but
did not reveal evidence of simultaneous equa-
tion bias) than to the 2SLS estimates (Table 3).
The 2SLS estimation, normally used to correct
for simultaneous equation bias (assumed or
otherwise), also probably contains aggregation
bias. If total exports influence the U.S. price,
conventional econometric procedures would
suggest the 2SLS estimates are better than the
OLS estimates. However, in this case, the 2SLS
estimation appears to introduce distortions in
the price and cross-price elasticity estimates
that exceed those that may be attributed to
aggregation bias.

Our OLS and six-equation deflated soybean
price elasticity estimates of -0.32 and -0.25 are

<<   <   3   4   5   6   7   8   9   >   >>