The factor proportions version of neoclassical trade theory has been recently applied to
empirical investigations. There are two main types of empirical investigations reported: studies that
address the specialization using typical measures of it, whatever sectors in which specialization
occurs (e.g., Anderson and Marcouiller, 2002; and Svaleryd and Vlachos, 2005), and studies that
address determinants of specialization within each sector in the economy (Harrigan, 1997; and
Gopinath and Upadhyay, 2002). This study follows the last line of research.
The above-mentioned studies, regarding the relationship between production factors
and specialization, have considered this relationship within groups of countries. The first
reference (Harrigan, 1997) studied a group of Organization for Economic Cooperation and
Development (OECD) countries, and the second reference (Gopinath and Upadhyay, 2002) studied one
sample of OECD countries and a `developing' countries ad hoc
sample. While Harrigan's study used a seemingly unrelated regressions procedure, the study by Gopinath and Upadhyay used a
random effects approach. However, the estimation approaches in both the studies failed to
consider unobservable country-specific characteristics that may be correlated with other regressors.
While Harrigan did not implement a fixed effects or instrumental variable approach, Gopinath
and Upadhyay seemed to fail to potentially correlate country-specific effects with right-hand
side factors and productivity. These studies faced obvious data constraints. Therefore, we intend
to improve both the estimation method—a corrected fixed effects approach developed by
Bruno (2005a and 2005b)—and the data coverage—a broader sample of developing
countries. Unfortunately, we have also faced some data constraints. |
