Page 14 - Azerbaijan State University of Economics

THE JOURNAL OF ECONOMIC SCIENCES: THEORY AND PRACTICE, V.72, # 2, 2015, pp. 4-23

are concerned, our results suggest that an exchange rate devaluation would trigger an
improvement in the balance of trade for such spheres as heavy metals, organic and
inorganic chemicals, and the agricultural industries. These sectors are composites,
i.e. consisting of a number of interconnected smaller industries. The interested
reader can return to Table 3 to revise which industry loads stronger on Factors 6 and
8. For policy purposes, it is enough to refer to the two composite factors to convey a
clear and concise message on which sectors of the domestic economy would most
positively respond to a potential exchange rate devaluation.
We complete the representation of results with the CUSUM and CUSUMSQ tests
for parameter stability in Figures 1 and 2. To preserve space, we report the illustrations
for the case of Factor 6 (“Heavy Metals and Inorganic Chemicals)” only. Remaining
graphs are available upon request. Note that policy implications for this particular factor
are even stronger, since the short-run (J-curve) and long-run dynamic (M-L condition)
are found to be recursively stable.
3. Conclusion and Advice for Future Research
This paper has introduced an innovative way on how to empirically investigate
the J-curve hypothesis and the Marshall-Lerner condition. Exploratory factor analysis
has been employed for the first time in this stream literature, and the posterior regression
analysis establishes the trade balance-exchange rate nexus. We have applied the
procedure to the bilateral industry-level trade between the United States and China. We
have successfully compressed a large dataset of 59 industries into just 9 composite
common factors. All assessment tests pass the necessary requirements. ARDL
regressions reveal that for 3 of the 9 factors an exchange rate devaluation triggers a
positive long-run response in the factorial balance of trade. Moreover, the J-curve effect
is observed in the short run. The superiority of this paper‟s approach is that final results
carry intuitive and clear policy implications, are not excessively spacious, and account
for any underlying across-industry commonalities.
Future studies have great room for maneuvering and improvement on this paper‟s
methodology. First, a factor-augmented approach could be applied to a variety of
regions and time periods, in order to improve interpretation of the existing industry-level
studies. Second, this paper has focused primarily on the eigenvalue-based rule of factor
extraction. Future attempts can experiment with confirmative factor analysis, i.e.
imposing a concrete amount of factors. Third, the factor-augmented approach is useful
for testing the workability of an existing hypothesis. For example, we theorize that there

14