We conducted a meta-analysis to explain the systematic variations found in estimated trade
effects of technical measures using both data sampling and methodology differences. Although it
is impossible to control for all the differences among the studies, we controlled for the
determinants that are most likely to matter, based on theoretical findings as well as important conjectures found in the previous empirical literature.
Analyses of agriculture and food industries lead to estimates of trade effects of technical
measures, which are less likely to be positive. Trade flows in these sectors tend to be more
impeded by technical measures than do trade flows in other sectors. Further, we find systematic
impeding effect of SPS regulations on agricultural exports sourced from developing countries
and going to high-income countries. Both robust regression and MNL approaches sustain this
important finding which suggests that SPS regulations are trade barriers rather than catalysts in
the set of studies analyzed here. We find that models that control for the “multilateral resistance”
terms using country-pair dummies are more likely to yield positive and significant estimates of
trade effects of technical measures than those that do not control for multilateral resistance.
Similarly, the former studies are less likely to yield negative significant trade effects than are the
latter.
The evidence of the three technical measure proxies is mixed. The three proxies tend to
have a positive effect on the estimates of trade effects of technical measures. No strong evidence
shows that the three different forms of technical measure proxies (count, frequency, dummy)
would lead to systematically different trade effects in the robust regression, however, the MNL
results strongly suggest that studies based on a count proxy yield estimates that are more likely to
be positive and much less likely to be negative. These two effects are the largest in magnitude for
the count proxy. The results on proxies, although convoluted, are consistent with ruling out a
negative influence of these proxies on the estimated trade effects of technical measures. The
aggregation level of the trade data could also affect the estimated trade effects, and the more
disaggregated data tend to provide more positive significant estimated trade effects of technical
measures relative to the conditional sample mean of t-values. These effects were found in the robust regression results but could not be confirmed with the MNL approach because of lack of
statistical significance.
In the future one could pool our dataset with studies analyzing multilateral, harmonized,
and reciprocal technical measures and incorporate technical measure estimates associated with
these standards. These standards have a different function with much potential to exhibit trade-expanding ability and with ambiguous effects on cost of production.
We conducted a meta-analysis to explain the systematic variations found in estimated trade
effects of technical measures using both data sampling and methodology differences. Although it
is impossible to control for all the differences among the studies, we controlled for the
determinants that are most likely to matter, based on theoretical findings as well as important conjectures found in the previous empirical literature.
Analyses of agriculture and food industries lead to estimates of trade effects of technical
measures, which are less likely to be positive. Trade flows in these sectors tend to be more
impeded by technical measures than do trade flows in other sectors. Further, we find systematic
impeding effect of SPS regulations on agricultural exports sourced from developing countries
and going to high-income countries. Both robust regression and MNL approaches sustain this
important finding which suggests that SPS regulations are trade barriers rather than catalysts in
the set of studies analyzed here. We find that models that control for the “multilateral resistance”
terms using country-pair dummies are more likely to yield positive and significant estimates of
trade effects of technical measures than those that do not control for multilateral resistance.
Similarly, the former studies are less likely to yield negative significant trade effects than are the
latter.
The evidence of the three technical measure proxies is mixed. The three proxies tend to
have a positive effect on the estimates of trade effects of technical measures. No strong evidence
shows that the three different forms of technical measure proxies (count, frequency, dummy)
would lead to systematically different trade effects in the robust regression, however, the MNL
results strongly suggest that studies based on a count proxy yield estimates that are more likely to
be positive and much less likely to be negative. These two effects are the largest in magnitude for
the count proxy. The results on proxies, although convoluted, are consistent with ruling out a
negative influence of these proxies on the estimated trade effects of technical measures. The
aggregation level of the trade data could also affect the estimated trade effects, and the more
disaggregated data tend to provide more positive significant estimated trade effects of technical
measures relative to the conditional sample mean of t-values. These effects were found in the robust regression results but could not be confirmed with the MNL approach because of lack of
statistical significance.
In the future one could pool our dataset with studies analyzing multilateral, harmonized,
and reciprocal technical measures and incorporate technical measure estimates associated with
these standards. These standards have a different function with much potential to exhibit trade-expanding ability and with ambiguous effects on cost of production.
การแปล กรุณารอสักครู่..