The real peso–dollar rate and US–Mexico industry trade: an asymmetric analysis
| Date | 01 September 2018 |
| DOI | http://doi.org/10.1111/sjpe.12175 |
| Author | Mohsen Bahmani‐Oskooee,Hanafiah Harvey,Scott W. Hegerty |
| Published date | 01 September 2018 |
THE REAL PESO–DOLLAR RATE AND
US–MEXICO INDUSTRY TRADE: AN
ASYMMETRIC ANALYSIS
Mohsen Bahmani-Oskooee*, Hanafiah Harvey** and Scott W. Hegerty***
ABSTRACT
In an attempt to improve upon previous analyses and find further evidence for
exchange rate theories such as the ‘J-curve’, numerous studies have introduced
novel econometric approaches that might help uncover significant results through
disaggregation and nonlinearity. This study applies the nonlinear cointegration
method of Shin et al. (2014) to US–Mexican trade balances in 91 individual
industries. While the linear model yields support for the ‘J-curve’ effect in 16
industries, the nonlinear model raises this number to 29 which includes the two
largest industries that engage in 35% of the trade between two countries. Fur-
thermore, while the short-run asymmetric effects of exchange rate changes were
discovered in almost all industries, short-run effects translated to significant
long-run asymmetric effects in 52 industries.
II
NTRODUCTION
While it is well-know n theoretically tha t the depreciation of it s currency
should result in an im provement of a countr y’s trade balance –even after
a period of adjustment duri ng which trade temporari ly deteriorates –such
a clear result is often no t shown in the literature . As is discussed in
reviews by Bahmani -Oskooee and Ratha (2 004) and Bahmani-Os kooee and
Hegerty (2010), emp irical studies of the so- called ‘J-curve’ phe nomenon
often find statistical ly insignificant connec tions between the exchan ge rate
and trade balance, reg ardless of the method o f analysis. Over time ,
improvements in methodology, including disaggregating trade by partner
and by industry, an d applying new approa ches to time-serie s analysis, have
attempted to uncover these ‘missing’ res ults. Nonetheless, c lear long-run
improvements in trade fol lowing a devaluation (of ten accompanied by
short-run decline s), are difficult to demo nstrate. As the two afor ementioned
*University of Wisconsin-Milwaukee
**Penn State University-Mont Alto
***Northeastern Illinois University
Valuable comments of two anonymous referees as well as those of the editor are greatly
appreciated. Remaining errors, however, are our own.
Scottish Journal of Political Economy, DOI: 10.1111/sjpe.12175, Vol. 65, No. 4, September 2018
©2018 Scottish Economic Society.
350
literature review s note, numerous empir ical analyses have b een performed
for a wide range of coun try pairs and individ ual industries, wit h the same
mixed results. It is th ought that a ‘true’ rela tion might be obscure d by
movements in opposi te directions that ca ncel each other out and a re
obscured by aggrega tion.
Over the past decade, cointegration analysis has increasingly been employed
in this type of study. Since results using this approach are still often insignifi-
cant as well, the most recent innovation has been to acknowledge that cur-
rency appreciations and depreciations might themselves have asymmetric
effects, and that they must be separated in the estimation process. This can be
done by the nonlinear approach to the Autoregressive Distributed Lag
(ARDL) methodology of Pesaran et al. (2001), which has become a work-
horse in this branch of the literature. This approach has often failed to
uncover significant effects, however.
Recent research has att empted to uncover this re lationship via nonline ar
methods. As Fedoseeva ( 2016) notes, the detrim ental impact of currency
appreciations can be ab sorbed via exporters’ m arkup strategies; dep recia-
tions are less affected, leading to an asymmetric effect between the two direc-
tions of movement. The m ain reason for asymmetri c effects, as suggeste d by
Bahmani-Oskooee and F ariditavana (2016) is ch ange in traders’ expecta tion
and reaction to currency depre ciation vs. appreciatio n. Furthermore, Bus-
siere (2013) has shown t hat due to price rigidit y, import and export price s
respond to exchange rate chang es asymmetrically. If impo rt and export
prices respond to excha nge rate changes asymmet rically, clearly trade vol-
ume and eventually tr ade balance would follow the suit and respond asym-
metrically too.
The NARDL approach of Shin et al. (2014) incorporates positive and
negative movements as se parate variables, pe rhaps allowing for more s ignifi-
cant results. The sam e ‘J-curve’ effect, firs t modeled by Magee (1973) , and
improved upon by Bahman i-Oskooee (1985) and Ros e and Yellen (1989),
can still be estimated to eval uate the impact of currency mov ements. But
this new method can also show whet her appreciations or deprec iations have
a relatively stronger i mpact. Symmetric affect s only exist if the two exchang e
rate variables have th e same sign and are of the sa me magnitude; otherw ise
they are shown to be asymmet ric. This type of study, c ould then isolate
which industries or cou ntries are susceptibl e to currency movement s in a cer-
tain direction, and m ight be useful for policy makers or business peop le in
these areas.
This study examines the case of US–Mexico bilateral trade in 91 industries,
over the period from 2002 to 2016 during which monthly data are available.
While the North American Free Trade Agreement (NAFTA) has significantly
expanded trade since its 1994 implementation, it has been politically con-
tentious even longer. The 1992 US Presidential election was influenced by the
entry of eccentric businessman H. Ross Perot, whose relatively high third-
party vote share (but gain of zero electoral votes) is often credited in assisting
US–MEXICO INDUSTRY TRADE 351
Scottish Journal of Political Economy
©2018 Scottish Economic Society
the victory of Democrat William J. Clinton. Twenty-four years later, business-
man Donald J. Trump won a narrow victory over Hillary Clinton on a plat-
form that included strict immigration restrictions and a renegotiation of the
NAFTA treaty. The peso, which had been depreciating due to declining oil
prices and political conditions in Mexico, began to fall further. It is expected
that this depreciation will continue, particularly as leftist politician Andres
Manuel Lopez Obrador continues to lead in the polls ahead of the 2018 Mexi-
can Presidential election.
Figure 1 shows the (log) trade balances for the four largest industries. Mex-
ican vehicle exports have increased in recent years, while fuels and oils have
declined. Electric Machinery peaked before the 2008 crisis (probably due to
increased foreign competition, coupled with the recession). Figure 2 shows
(log) industrial production, as well as exchange rate movements. The peso has
steadily weakened in the latter part of the sample; we expect this to be
reflected in an increase in the US trade deficit with Mexico, even after control-
ling for income. This, however, may not be the case for all industries, particu-
larly those with the pricing power to avert a decline. These industry trade
flows must, therefore, be tested empirically, particularly in light of recent
developments.
A previous application of the linear ARDL approach, by Bahmani-Oskooee
and Hegerty (2011), examined 102 industries from 1962 to 2004 using annual
data and found that only 24 showed the expected long-run response to a peso
depreciation. Bahmani-Oskooee et al. (2016) examine Mexican trade, but only
at the bilateral (trade partner) level. We re-examine recent US–Mexico indus-
try trade here to assess whether the NARDL approach which incorporates
asymmetry analysis will show a stronger sensitivity to exchange rate move-
ments. Overall, we find that 15 of 91 industries exhibit asymmetric improve-
ment after a dollar depreciation, but that possible asymmetries are worthy of
further investigation.
Figure 1. (Log) trade balances for four large industries.
Note: 84, nuclear reactors, boilers, machinery, etc.; parts; 85, electric machinery, etc.; sound
equip; Tv equip; pts; 87, vehicles, except railway or tramway; and parts, etc.; 27, mineral
fuel, oil, etc.; Bitumin Subst; mineral wax.
352 M. BAHMANI-OSKOOEE, H. HARVEY AND S. W. HEGERTY
Scottish Journal of Political Economy
©2018 Scottish Economic Society
To continue reading
Request your trial