
 Start

3086
 Prefix

The empirical results are, in
general, sensitive to the choices of sample period, model specification, form of proxies
for exchange rate volatility, and countries considered (developed versus developing).2
1Several theoretical studies (e.g.,
 Exact

Clark (1973), Baron (1976))
 Suffix

have shown that an increase in
exchange rate uncertainty will have adverse effects on the volume of international trade. Others,
including Franke (1991), Sercu and Vanhulle (1992) have shown that exchange rate uncertainty
may have a positive or ambiguous impact on the volume of international trade flows depending
on aggregate exposure to currency risk (Viaene and de Vries (1992)) and the types of
 (check this in PDF content)

 Start

3251
 Prefix

results are, in
general, sensitive to the choices of sample period, model specification, form of proxies
for exchange rate volatility, and countries considered (developed versus developing).2
1Several theoretical studies (e.g., Clark (1973), Baron (1976)) have shown that an increase in
exchange rate uncertainty will have adverse effects on the volume of international trade. Others,
including
 Exact

Franke (1991), Sercu and Vanhulle (1992)
 Suffix

have shown that exchange rate uncertainty
may have a positive or ambiguous impact on the volume of international trade flows depending
on aggregate exposure to currency risk (Viaene and de Vries (1992)) and the types of shocks to
which the firms are exposed (Barkoulas, Baum and Caglayan (2002)).
 (check this in PDF content)

 Start

3547
 Prefix

Others,
including Franke (1991), Sercu and Vanhulle (1992) have shown that exchange rate uncertainty
may have a positive or ambiguous impact on the volume of international trade flows depending
on aggregate exposure to currency risk (Viaene and de Vries (1992)) and the types of shocks to
which the firms are exposed
 Exact

(Barkoulas, Baum and Caglayan (2002)).
 Suffix

Also see models that study
the impact of exchange rate uncertainty on trade and its welfare costs within a general equilibrium
framework including Obstfeld and Rogoff (2003), Bacchetta and van Wincoop (2000).
2Negative effects of exchange rate uncertainty on trade flows are recently reported by Arize,
Osang and Slottje (2000), Sauer and Bohara (2001) for developing countries, while Gagnon (1993)
f
 (check this in PDF content)

 Start

3731
 Prefix

have a positive or ambiguous impact on the volume of international trade flows depending
on aggregate exposure to currency risk (Viaene and de Vries (1992)) and the types of shocks to
which the firms are exposed (Barkoulas, Baum and Caglayan (2002)). Also see models that study
the impact of exchange rate uncertainty on trade and its welfare costs within a general equilibrium
framework including
 Exact

Obstfeld and Rogoff (2003),
 Suffix

Bacchetta and van Wincoop (2000).
2Negative effects of exchange rate uncertainty on trade flows are recently reported by Arize,
Osang and Slottje (2000), Sauer and Bohara (2001) for developing countries, while Gagnon (1993)
finds insignificant effects for developed countries.
 (check this in PDF content)

 Start

3879
 Prefix

Also see models that study
the impact of exchange rate uncertainty on trade and its welfare costs within a general equilibrium
framework including Obstfeld and Rogoff (2003), Bacchetta and van Wincoop (2000).
2Negative effects of exchange rate uncertainty on trade flows are recently reported by
 Exact

Arize, Osang and Slottje (2000), Sauer and Bohara (2001)
 Suffix

for developing countries, while Gagnon (1993)
finds insignificant effects for developed countries. Baum, Caglayan and Ozkan (2004) report that the
impact of exchange rate volatility on export flows differs in sign and magnitude across the countries
studied.
 (check this in PDF content)

 Start

3967
 Prefix

Also see models that study
the impact of exchange rate uncertainty on trade and its welfare costs within a general equilibrium
framework including Obstfeld and Rogoff (2003), Bacchetta and van Wincoop (2000).
2Negative effects of exchange rate uncertainty on trade flows are recently reported by Arize, Osang and Slottje (2000), Sauer and Bohara (2001) for developing countries, while
 Exact

Gagnon (1993)
 Suffix

finds insignificant effects for developed countries. Baum, Caglayan and Ozkan (2004) report that the
impact of exchange rate volatility on export flows differs in sign and magnitude across the countries
studied.
 (check this in PDF content)

 Start

4033
 Prefix

uncertainty on trade and its welfare costs within a general equilibrium
framework including Obstfeld and Rogoff (2003), Bacchetta and van Wincoop (2000).
2Negative effects of exchange rate uncertainty on trade flows are recently reported by Arize, Osang and Slottje (2000), Sauer and Bohara (2001) for developing countries, while Gagnon (1993) finds insignificant effects for developed countries.
 Exact

Baum, Caglayan and Ozkan (2004)
 Suffix

report that the
impact of exchange rate volatility on export flows differs in sign and magnitude across the countries
studied. We should also note that researchers implementinggravity models (see Frankel and Wei
(1993), Dell’Ariccia (1999), Rose (2000), and Tenreyro (2003) among others) have generally found a
2
More recently Baum et al. (2004) rely on a nonlinear specification rather than linear
a
 (check this in PDF content)

 Start

4259
 Prefix

Baum, Caglayan and Ozkan (2004) report that the
impact of exchange rate volatility on export flows differs in sign and magnitude across the countries
studied. We should also note that researchers implementinggravity models (see
 Exact

Frankel and Wei (1993),
 Suffix

Dell’Ariccia (1999), Rose (2000), and Tenreyro (2003) among others) have generally found a
2
More recently Baum et al. (2004) rely on a nonlinear specification rather than linear
alternatives while integrating the role of foreign income uncertainty in evaluating the
impact of exchange rate uncertainty on bilateral trade flowsbetween several developed countries.
 (check this in PDF content)

 Start

4303
 Prefix

Baum, Caglayan and Ozkan (2004) report that the
impact of exchange rate volatility on export flows differs in sign and magnitude across the countries
studied. We should also note that researchers implementinggravity models (see Frankel and Wei (1993), Dell’Ariccia (1999),
 Exact

Rose (2000), and Tenreyro (2003)
 Suffix

among others) have generally found a
2
More recently Baum et al. (2004) rely on a nonlinear specification rather than linear
alternatives while integrating the role of foreign income uncertainty in evaluating the
impact of exchange rate uncertainty on bilateral trade flowsbetween several developed countries.
 (check this in PDF content)

 Start

4387
 Prefix

Baum, Caglayan and Ozkan (2004) report that the
impact of exchange rate volatility on export flows differs in sign and magnitude across the countries
studied. We should also note that researchers implementinggravity models (see Frankel and Wei (1993), Dell’Ariccia (1999), Rose (2000), and Tenreyro (2003) among others) have generally found a
2
More recently
 Exact

Baum et al. (2004)
 Suffix

rely on a nonlinear specification rather than linear
alternatives while integrating the role of foreign income uncertainty in evaluating the
impact of exchange rate uncertainty on bilateral trade flowsbetween several developed countries.
 (check this in PDF content)

 Start

4703
 Prefix

(2000), and Tenreyro (2003) among others) have generally found a
2
More recently Baum et al. (2004) rely on a nonlinear specification rather than linear
alternatives while integrating the role of foreign income uncertainty in evaluating the
impact of exchange rate uncertainty on bilateral trade flowsbetween several developed countries. Although their findings are mixed, a subsequent analysis by
 Exact

Grier and Smallwood (2007)
 Suffix

using a group of developed and developing countries finds a
significant role in developing countries’ exports for exchange rate uncertainty as well
as a strong role for income uncertainty in most countries.3
In this paper, we present two sets of empirical findings motivated by the theoretical propositions of Barkoulas et al. (2002) that (i) exchange rate uncertainty affects
the volume of trade flo
 (check this in PDF content)

 Start

5037
 Prefix

Although their findings are mixed, a subsequent analysis by Grier and Smallwood (2007) using a group of developed and developing countries finds a
significant role in developing countries’ exports for exchange rate uncertainty as well
as a strong role for income uncertainty in most countries.3
In this paper, we present two sets of empirical findings motivated by the theoretical propositions of
 Exact

Barkoulas et al. (2002)
 Suffix

that (i) exchange rate uncertainty affects
the volume of trade flows and (ii) exchange rate uncertainty affects the variability
of trade flows. Researchers generally motivate the first hypothesis indicating that
exchange rate uncertainty will inevitably depress the volume of international trade
by increasing the riskiness of trading activity.
 (check this in PDF content)

 Start

6565
 Prefix

, Wei,
Sadikov and Zeng (2004) indicate that ‘this negative relationship, however, is not robust to a more
general specification of the equation linking bilateral trade to its determinants that embodies the
recent theoretical advances in a gravity model’ (p. 2).
3One would be tempted to think that the exposure to unforeseenmovements in exchange rates
can be avoided using hedging. However,
 Exact

Wei (1999)
 Suffix

finds no empirical support for the hypothesis
that the availability of hedging instruments reduces the impact of exchange rate volatility on trade.
4For instance, according to Barkoulas et al. (2002), in open economies where the importance of
international trade is sizable, variability of trade flows can significantly impact the variability of the
overall level of economic activity resulting in ‘f
 (check this in PDF content)

 Start

6750
 Prefix

that embodies the
recent theoretical advances in a gravity model’ (p. 2).
3One would be tempted to think that the exposure to unforeseenmovements in exchange rates
can be avoided using hedging. However, Wei (1999) finds no empirical support for the hypothesis
that the availability of hedging instruments reduces the impact of exchange rate volatility on trade.
4For instance, according to
 Exact

Barkoulas et al. (2002),
 Suffix

in open economies where the importance of
international trade is sizable, variability of trade flows can significantly impact the variability of the
overall level of economic activity resulting in ‘financial sector illiquidity, reductions in real output,
and/or heightened inflationary pressures’ (p. 491).
3
volatility is often several times that of aggregate output,as recognized by Engel
and Wang
 (check this in PDF content)

 Start

7155
 Prefix

al. (2002), in open economies where the importance of
international trade is sizable, variability of trade flows can significantly impact the variability of the
overall level of economic activity resulting in ‘financial sector illiquidity, reductions in real output,
and/or heightened inflationary pressures’ (p. 491).
3
volatility is often several times that of aggregate output,as recognized by
 Exact

Engel and Wang (2007) and Zimmermann (1999).
 Suffix

Hence, we consider the effects of exchange rate uncertainty on the volatility of trade flows as animportant factor in the
predictability of aggregate economic activity.
Our investigation concentrates on bilateral trade flows between 13 countries including the US, UK, Canada, Germany, France, Italy, Japan, Finland, Netherlands,
Norway, Spain, Sweden, and Switzerland for the period 1980–1998 on a mo
 (check this in PDF content)

 Start

8889
 Prefix

This observation should not be too surprising as the recent empirical literature has
recorded similar findings. Furthermore, these results are in line with the theoretical
literature. Our second set of findings is new and novel as we provide indirect empirical support to a proposition suggested in
 Exact

Barkoulas et al. (2002).
 Suffix

Specifically,
we show that exchange rate uncertainty has a meaningful empirical impact on the
volatilityof trade flows. We find that 81 out of 143 models tested provide support
for a statistically significant steadystate effect of exchange rate uncertainty on trade
volatility.
 (check this in PDF content)

 Start

10237
 Prefix

and draws
implications for future theoretical and empirical research.
2 Motivation
There is a long list of papers which focus on the effects of exchange rate variability
on trade flows6arguing that exchange rate uncertainty will inevitably depress the
volume of international trade by increasing the riskiness of trading activity. Our twopronged empirical investigation is mainly motivated by
 Exact

Barkoulas et al. (2002)
 Suffix

who
utilize a partial equilibrium model to study the impact of exchange rate movements
on the level and volatility of trade flows. They claim that an analysis considering the
effects of exchange rate uncertainty only on the volume of trade will not be capable of
generating predictions of managers’ optimal behavior.
 (check this in PDF content)

 Start

11077
 Prefix

Notably, few researchers have considered
the second hypothesis despite the fact that the volatility of trade flows relative to that
of aggregate output is sizable: often a factor of two or threetimes GDP volatility,
and as volatile as investment spending.
The model that
 Exact

Barkoulas et al. (2002)
 Suffix

construct assumes that managers’ decisions to export (or import) depend upon both expected returnand risk and suggests
6See notes 1 and 2 above.
5
modeling the impact of exchange rate uncertainty on both thefirst and second moments of trade flows.
 (check this in PDF content)

 Start

13131
 Prefix

Although our hypothesis that uncertaintyin trade flows arising from
exchange rate uncertainty will have serious effects on the macroeconomy is different
from those studies’ emphasis, it would be useful to documentsome of the findings
reported in them. In a recent contribution considering trade flow volatility,
 Exact

Engel and Wang (2007)
 Suffix

lay out a twocountry twosector model to understand international
real business cycles. They state that although countercyclical behavior of net exports
is a well established fact, the literature has neglected thebehavior of imports and
exports which tend to be much more volatile than GDP.
 (check this in PDF content)

 Start

14406
 Prefix

goods to explain the volatility of
trade flows assuming that while switching between home and foreign durable goods is
highly costly in the short run, home and foreign goods are perfect substitutes in the
long run. Their simulation results are generally supportive as their model captures
the volatility of trade flows along with several other features of the data.
The second study is that of
 Exact

Zimmermann (1999).
 Suffix

He, too, provides an international real business cycle model to explain the behavior of components of GDP
and attempts to rationalize trade flow variability. He points out that trade flows are
much more volatile than GDP—as volatile as investment—and that prior research
has not addressed this issue.
 (check this in PDF content)

 Start

15537
 Prefix

In our empirical investigation below, we employ a simple reduced form model
to understand how movements in real exchange rates affect thebehavior of (i) the
level and (ii) the volatility of exports. Although our studycan best be described in
the spirit of
 Exact

Barkoulas et al. (2002),
 Suffix

we do not attempt to capture exchange rate
7Using a model in the spirit of Zimmermann (1999), Engel and Wang show that an increase in
exchange rate volatility can be helpful in explaining the volatility in trade flows.
 (check this in PDF content)

 Start

15636
 Prefix

In our empirical investigation below, we employ a simple reduced form model
to understand how movements in real exchange rates affect thebehavior of (i) the
level and (ii) the volatility of exports. Although our studycan best be described in
the spirit of Barkoulas et al. (2002), we do not attempt to capture exchange rate
7Using a model in the spirit of
 Exact

Zimmermann (1999),
 Suffix

Engel and Wang show that an increase in
exchange rate volatility can be helpful in explaining the volatility in trade flows. But their construct
yields a negative correlation between the levels of exportsand imports, whereas that correlation is
generally positive in the data.
7
uncertainty that may arise from different sources.
 (check this in PDF content)

 Start

18124
 Prefix

To match the monthly frequency of export
data, we must generate a proxy for monthly foreignGDPas the available data is on a
quarterly basis.8Hence, we apply the proportional Denton benchmarking technique
 Exact

(Bloem, Dippelsman and Maehle (2001)) to
 Suffix

the quarterly realGDPseries in order to
produce monthlyGDPestimates. The proportional Denton benchmarking technique
uses the higherfrequency movements of an associated variable—in our case monthly
industrial production—as an interpolator within the quarter, while enforcing the
constraint that the sum of monthlyGDPflows equals the observed quarterly total.
 (check this in PDF content)

 Start

21759
 Prefix

Equation (2) definesthe conditional mean
of the ∆ log real exchange rate (st) as a function of its own lag and ∆ lagged trade
volume as well as a firstorder moving average innovation. The vector of innovations
is defined asut= [ωt,ηt]′. The diagonal elements ofHtare the conditional variances
of ∆ log real exchange rate,σ2xtand ∆ log trade volume,σ2strespectively.
Following
 Exact

Karolyi (1995),
 Suffix

the matrixCis parameterized as lower triangular while
matricesAandBare 2×2 matrices, so that there are eleven estimated parameters
in equation (3). We assume that the errors are jointly conditionally normal with zero
means and conditional variances given by anARMA(1,1)structure as expressed in
10
equation (3).
 (check this in PDF content)

 Start

26084
 Prefix

Given the monthly frequency of the data and the large number of coefficients on highly correlated
regressors to be estimated, we did not find that an unconstrained distributed lag approach produced
usable nor dynamically stable estimates.
12Althoughβ1is the coefficient of a generated regressor
 Exact

(Pagan, 1984; Pagan, 1986),
 Suffix

inference on
the existence of a significantβ1coefficient is not hindered by that issue.
13We note thatσ2
stis a generated regressor, as is the dependent variable in this equation, which
is an augmented autoregression.
 (check this in PDF content)

 Start

27623
 Prefix

4.1 Timeseries properties of the data
In the literature, it has commonly been assumed that trade volume series (real exports) and real exchange rates, in level or log form, are nonstationary (I(1)) processes given evidence from univariate unit root tests. Prior to estimating our system
of equations, we test each series for a unit root using the modified logperiodogram
regression test of
 Exact

Phillips (2007)
 Suffix

as implemented in Baum and Wiggins (2001). We
find that the overwhelming proportion of both log exchange rate and log trade flow
series exhibitI(1) characteristics while the logGDPseries areI(1) except for four
instances.
 (check this in PDF content)

 Start

27657
 Prefix

of the data
In the literature, it has commonly been assumed that trade volume series (real exports) and real exchange rates, in level or log form, are nonstationary (I(1)) processes given evidence from univariate unit root tests. Prior to estimating our system
of equations, we test each series for a unit root using the modified logperiodogram
regression test of Phillips (2007) as implemented in
 Exact

Baum and Wiggins (2001).
 Suffix

We
find that the overwhelming proportion of both log exchange rate and log trade flow
series exhibitI(1) characteristics while the logGDPseries areI(1) except for four
instances. For the remainder of our analysis we keep those series that are clearly
classified asI(1) and drop the remaining series.
 (check this in PDF content)

 Start

36856
 Prefix

relationship in only six models.20The greatest
number of significant effects (10) is registered by Spain, allpositive, followed by the
USand Switzerland (eight each, with seven positive for each country). Given that
the overwhelming majority of the models provide a positive relationship between
exchange rate uncertainty and the volatility of trade flows,considering our findings
in the light of
 Exact

Barkoulas et al. (2002),
 Suffix

we conjecture that either the preponderance of
20At the ten per cent level of significance, we find 90 nonzero coefficients: 82 positive, eight
negative.
17
shocks to the exchange rate process are associated with shocks to the fundamentals
or that fundamental shocks are larger or have a greater impact on the real exchange
rate than other shocks.21
The impact of a one standard deviation increase in
 (check this in PDF content)

 Start

39255
 Prefix

However the complexity of our methodology
leads us to leave this issue for future investigation.
18
employing a bivariateGARCHmethodology. Using these proxies, we investigate the
impact of exchange rate volatility on the mean and the variance of trade flows in
the spirit of
 Exact

Barkoulas et al. (2002).
 Suffix

Our first set of resultssuggest that the impact
of exchange rate volatility on trade flows is indeterminate.Only a small number of
models (30 out of 143) present significant relationships: significant and positive in 23
models and significantly negative in the remaining seven models.
 (check this in PDF content)

 Start

39916
 Prefix

Our second set of findings is new and novel as we investigate the relationship
between exchange rate volatility and that of trade flows. We first show that bilateral
trade volatility is higher than GDP volatility, a stylized fact earlier documented by
 Exact

Engel and Wang (2007) and Zimmermann (1999)
 Suffix

for aggregate trade flows. We then
run a simple dynamic model to see if exchange rate volatilityleads to higher trade
variability. We find that for 81 out of 143 potential models, exchange rate volatility
exhibits a positive steadystate impact on the volatility of trade flows.
 (check this in PDF content)