This article was downloaded by: [Universiti Putra Malaysia]

On: 09 November 2011, At: 23:42

Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954
Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH,
UK
The International Trade
Journal
Publication details, including instructions for
authors and subscription information:
http://www.tandfonline.com/loi/uitj20
Bilateral FDI and Canadian
Export Activity
John K. Mullen
a
& Martin Williams
b
a
School of Business, Clarkson University, Potsdam,
New York, USA
b
Department of Economics, Northern Illinois
University, DeKalb, Illinois, USA
Available online: 22 Jun 2011
To cite this article: John K. Mullen & Martin Williams (2011): Bilateral FDI and
Canadian Export Activity, The International Trade Journal, 25:3, 349-371
To link to this article: http://dx.doi.org/10.1080/08853908.2011.581611
PLEASE SCROLL DOWN FOR ARTICLE
Full terms and conditions of use: http://www.tandfonline.com/page/terms-
and-conditions
This article may be used for research, teaching, and private study purposes.
Any substantial or systematic reproduction, redistribution, reselling, loan,
sub-licensing, systematic supply, or distribution in any form to anyone is
expressly forbidden.
The publisher does not give any warranty express or implied or make any
representation that the contents will be complete or accurate or up to
date. The accuracy of any instructions, formulae, and drug doses should be
independently verified with primary sources. The publisher shall not be liable
for any loss, actions, claims, proceedings, demand, or costs or damages
whatsoever or howsoever caused arising directly or indirectly in connection
with or arising out of the use of this material.
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1

Bilateral FDI and Canadian
Export Activity
John K. Mullen
School of Business, Clarkson University, Potsdam,
New York, USA
Martin Williams
Department of Economics, Northern Illinois University,
DeKalb, Illinois, USA
This study examines how Canadian exports to a specic
trading partner are inuenced by outward and inward direct
investment ows to/from that country. A gravity-type empir-
ical model guides a dynamic panel analysis which utilizes
Organization for Economic Cooperation and Development
(OECD) country-level data from 19892007. Besides refuting
the contention that outward foreign direct investment (FDI) dis-
places exports, the ndings also imply a strong role for intra-rm
based export growth in response to inward FDI. The analysis is
enriched by explicitly accounting for the dominant position of the
United States within the context of Canadas overall trade and
investment ows.
This research has been supported in part by an award from
the Canadian Embassys Faculty Research Grant Program,
20042005. The viewpoints presented herein are the authors
alone, and they remain solely responsible for any errors.
Correspondence should be addressed to John K. Mullen, School
of Business, c/o CUSB, Box 5785, Clarkson University, Potsdam,
NY 13699, USA. E-mail: [email protected]
349
THE INTERNATIONAL TRADE JOURNAL, Volume 25, No. 3, JulyAugust 2011
ISSN: 0885-3908 print/1521-0545 online. DOI: 10.1080/08853908.2011.581611
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1

350 THE INTERNATIONAL TRADE JOURNAL
KEYWORDS trade substitution, gravity model, bilateral
FDI, tari-jumping, panel data analysis
* * * * *
I. INTRODUCTION
The integration of the worlds developed economies has been
manifested, in part, by a dramatic rise in foreign direct invest-
ment (FDI) by multinational corporations. At the same time,
the rapid growth in the volume of trade has led to a renewed
interest in the relationship between exports and foreign produc-
tion. Traditional trade models have generally treated exports
and outward FDI as competing ways of serving foreign markets,
implying a degree of substitution between these two activities.
A long-held view is that investment abroad constitutes a tari-
jumping activity that benets multinational rms even as it
displaces export activity. Yet the empirical evidence often demon-
strates a complementary relationship, suggesting the importance
of vertical production in general and intra-rm trade in particular
(e.g., see Head and Ries 2001). The overall relationship between
foreign production and trade remains a complex one, however,
and is likely to vary across countries, industries, and time peri-
ods. Indeed, the trade-investment nexus may be changing as
the worlds economies become increasingly integrated. Identifying
the types of trade and investment policies that are most likely
to aid domestic economic performance requires timely evidence
on the nature of this relationship for specic countries. The
present research oers some recent evidence for a relatively small,
open economy that has historically encouraged both international
trade and investment ows.
Canadas role and performance within the global economy,
as reected by its trade and investment patterns, is interesting
from a number of perspectives. Its experience in the wake of
the North American Free Trade Agreement (NAFTA) potentially
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1

Mullen and Williams: Bilateral FDI . . . 351
holds valuable lessons for similar economies that are contemplat-
ing their level of involvement in regional trading blocs. In that
regard, Canadas heavy reliance on the United States as its major
trading and investment partner is likely to hold implications for
other nations that are in a similar position alongside dominant
trading partners. Further, outward FDI is of vital importance
for countries such as Canada that seek to overcome the limita-
tions of small domestic markets; as such, evidence on the likely
impacts of its direct investment abroad would be illuminating.
For example, any resultant decline in export activity may add
weight to arguments that investment abroad is akin to exporting
jobs. Also, there remain a number of unresolved issues surround-
ing the impacts of increased FDI penetration from the United
States and elsewhere. Concern over the possibility of negative
impacts on domestic industries might prevent policymakers from
adopting strategies designed to encourage inbound foreign invest-
ment. In turn, this could stie potential productivity gains for
domestic rms achieved via knowledge spillovers. Also, the eect
of inward FDI on intra-rm trade and, by extension, the level of
exports remains an unsettled issue, and is a major concern here.
The present research analyzes how Canadian exports are
aected by both outward and inward stocks of FDI to and from
its major trading partners. The empirical methodology exam-
ines this relationship for 16 OECD nations, within a framework
that considers Canadas unique economic and geographic link-
ages to the United States. We focus on exports, rather than total
trade volume, because of their important linkages with domestic
employment and investment levels. Ultimately, the results gen-
erated here may prove valuable to policymakers concerned with
identifying which international economic policies are most likely
to be conducive to domestic economic growth.
The next section of the article reviews the pertinent liter-
ature, and provides relevant background on other issues. The
third section discusses the empirical model, data sources, and
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1

352 THE INTERNATIONAL TRADE JOURNAL
estimation approach. Next, the empirical results are presented
and discussed. A nal section summarizes and oers interpreta-
tions for our ndings.
II. BACKGROUND LITERATURE
Previous research has generally been concerned with the
potential impact of bilateral FDI on production and invest-
ment activities in both host and home countries (Stevens and
Lipsey 1992). The possibility of technology transfers and the
resultant productivity gains accruing to domestic rms from
inward FDI have served to foster policies to attract such invest-
ment. Indeed, some empirical evidence suggests that inward
FDI may stimulate domestic investment rather than displace
it (Borenzstein and deGregorio 1995), perhaps owing to the
increased competitiveness of domestic rms.
Overall, however, the empirical evidence for advanced
economies is mixed, mitigating the claim over whether invest-
ment inows are as benecial as often suggested. Obviously, from
the perspective of the investor (i.e., home) country, an important
issue is whether outward FDI substitutes for export activity. It
is conceivable that outward FDI stimulates exports from certain
industries even as it displaces trade within the industry of the
outbound investment. Yet the existing evidence is unclear about
the overall eect of outward FDI on investment and trade volume
in the home economy. Narrower in scope, our specic concern is
with how a countrys exports to a trading partner are aected
by inward and outward FDI activity with that same country.
Traditional trade theories have been interpreted as imply-
ing that exports and outward direct investment are substitutes.
Firms are viewed as choosing between production either at home
(followed by exports) or abroad (with local sales by their foreign
aliates). This decision is articulated as being driven by consid-
erations such as transportation costs and scale economies. This
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1

Mullen and Williams: Bilateral FDI . . . 353
proximity-concentration tradeo is oered to explain why FDI
may substitute for trade (Brainard 1993).
An additional argument supporting this view is that multi-
nationals locate subsidiaries near dierent markets when plant-
specic xed costs are low (Markusen and Venables 1998).
However, if FDI reects vertical integration whereby rms split
production across dierent countries, then it is conceivable that
such investment may actually stimulate trade. The theoretical
literature oers only weak guidance on the precise relationship
between exports and FDI, especially in view of the role of factors
such as taris, transport costs, the organization of rm activity,
etc. Our understanding of this relationship can be improved by
the provision of appropriate empirical evidence.
The existing literature provides some evidence, at the rm
level, that outward FDI may stimulate trade volume. Lipsey and
Weiss (1984) nd a positive relationship between total exports
of the parent rm and production levels at aliate rms. Sachs
and Shatz (1994) provide evidence of a complementary relation-
ship between FDI and exports for the United States, but this may
largely reect the role of intra-rm trade. Additionally, Clausings
(2000) analysis of multinational rms supports a complementary
relationship between foreign investment activity (as measured
by aliate sales) and intra-rm trade. Finally, Head and Ries
(2001) distinguish between manufacturing vs. distribution ali-
ates of Japanese multinational rms in nding that the degree of
complementarity varies with the importance of intra-rm trade.
Their analysis highlights the role of industry characteristics in
aecting the magnitude of intra-rm trade.
Amiti and Wakelin (2003), based on a sample of 36 developed
and developing nations, nd some evidence of a complementary
relationship between exports and investment liberalization. Yet
their results indicate that this relationship holds only between
partners that are characterized by similar factor endowments and
relatively low trading costs. Eaton and Tamura (1994) attempt
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1

354 THE INTERNATIONAL TRADE JOURNAL
to explain bilateral trade and FDI ows by relying on traditional
gravity model determinants. Their ndings suggest, for both the
United States and Japan, the existence of a large positive rela-
tionship between outward FDI and exports and imports alike. An
OECD study (1998) estimated import and export equations, and
then simulated bilateral trade ows in the absence of FDI for a
sample of 21 countries. These results suggest that FDI generally
has a positive impact on trade levels, but that the magnitudes
of the eect vary dramatically by country. As an example, they
nd that inward FDI stocks increase exports for the U.K., but
not for the United States.
Egger (2001) also reports ndings implying that outward
investment fails to exhibit a signicant (positive or negative)
inuence on exports, based on an analysis of 15 EU members over
the period 19861996. Hejazi and Safarian (2001), using a grav-
ity model linked to the transactions cost theory of multinational
enterprises, conrm a complementary relationship between trade
and FDI for the United States and 51 countries from 1982 to
1994. Again, their study suggests the need to control for sectoral
and/or regional characteristics in order to reect the dierent
motivations underlying FDI.
Unfortunately, the results of existing empirical eorts tend
to vary depending on how investment is measured. Pain, Nigel,
and Wakelin (1998), relying on FDI stocks derived for a sam-
ple of Organization for Economic Cooperation and Development
(OECD) countries, nd that substitution eects were prevalent
over the 197192 period. They suggest that the choice of the
measure of FDI activity (stocks vs. ows) may be pivotal in the
nature of the empirical results. Once again, they caution that
the trade-investment relationship is likely to vary over time and
across countries. Also, many of the economy-wide studies suf-
fer from endogeneity problems because the direction of causality
between trade and investment patterns is not straightforward.
Fortunately, some of these issues may be resolved by improved
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1

Mullen and Williams: Bilateral FDI . . . 355
empirical designs and appropriate statistical techniques. In sum,
the empirical evidence is inconclusive concerning the relationship
between foreign investment and trade in general; more speci-
cally, the impact of inward and outward FDI stocks on a countrys
export activity remains unsettled.
It is clear that a need remains for more recent evidence on
the trade-investment nexus, both for the larger industrialized
countries and for smaller, open economies. In fact, most of the
country-specic evidence is limited to the United States, Japan,
and major European economies. Clearly, this relationship holds
important implications for policymakers in smaller nations that
have historically embraced international trade and investment
ows. Canada is one such country where additional evidence
about the eects of FDI is needed to resolve a number of issues.
A debate continues over whether inward FDI, particularly from
the United States, will harm Canadian industries and cause a
hollowing out of their economic sovereignty.
The eects of outward FDI on exports and domestic employ-
ment remain an unsettled question. Rao, Ashfaq and Legaul
(1994) conclude, based on total elasticity estimates from two
early sub-periods, that exports (and imports) are complemen-
tary to outbound FDI. However, they admit that this result is
tentative, citing a need for more rigorous empirical testing. An
ideal approach should control for other trade determinants within
the framework of a well-designed regression model.
Canadas economy is strongly oriented towards international
investment, with bilateral FDI stocks having approximately
tripled between 1990 and 2007. Yet it has become a net exporter
of capital since the mid-1990s, perhaps owing to its relatively
punitive tax system and general restrictions on inbound foreign
investment. The importance of both stocks of FDI for Canada,
expressed as a percentage of GDP, exceeds the G-7 average
(Holden 2008). Of course, the role of the United States as both a
source and destination for Canadas FDI is paramount. For 2007,
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1

356 THE INTERNATIONAL TRADE JOURNAL
the United States represented 57.6% and 43.9% of its inward
and outward investment stocks, respectively. A similar pattern
is exhibited for Canadas trade ows, with the United States
accounting for 73.5 % and 63.7 % of its exports and imports,
respectively (Foreign Aairs and International Trade Canada
2008). So while it is important to oer new evidence based on
recent data for the Canadian economy, an interesting dimension
arises when considering that the majority of its trade-investment
ows pertain to a dominant partner.
III. METHODOLOGY AND DATA
Empirical Model
The primary objective of this study is to clarify the impact
of outward and inward FDI on Canadian export ows. Our
approach involves estimating statistical relationships that explain
variations in Canadian export activity to major OECD countries
from 1989 to 2007. The empirical technique is an application of
the gravity model, one that is generally consistent with theo-
retical views of trade and multinational corporations (e.g., see
Deardor 1998; Markusen and Venables 1998). Recent empirical
analyses of the trade vs. investment decision often utilize a sim-
ilar approach (e.g., see Amiti and Wakelin 2003; Clausing 2000;
Smith 2001).
In its basic variant, bilateral trade volume between two par-
ties (i, j) depends on individual country characteristics (including
per capita income and population) as well as transportation costs
and exchange rates. These determinants are incorporated into the
following specication:
T
ij
=
0
(GDPC
i
)
1
(POP
i
)
2
(GDPC
j
)
3
(POP
j
)
4
(DIST
ij
)
5
(REXR
ij
)
6
(F
j
)
7

ij
(1)
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1

Mullen and Williams: Bilateral FDI . . . 357
where T
ij
is bilateral trade
1
from i to j, GDPC is per capita
income, POP is population, DIST is geographic distance (as
a proxy for transportation costs), and REXR
ij
is the relative
exchange rate between the trading partners; F
j
represents other
country-specic factors that might impact the volume of trade
ows; is a random disturbance term.
Since our primary concern is with explaining variations in
Canadian export activity, we modify (1) by indexing Canada as
the source country (i). By relying on panel data, we focus on out-
ward trade in the form of exports (EX) from Canada to specic
countries (j) at various points in time (t), obviating the explicit
consideration of Canadas population and per capita income in
the model. After taking logs, the following static specication
includes country xed eects (F
j
) and the stocks of (outward
and inward) Canadian direct investment (FDI) to and from each
country at each point in time:
ln(EX
jt
) =F
j
+
1
lnGDPC
jt
+
2
lnPOP
jt
+
3
lnREXR
jt
+
4
lnINFDI
jt
+
5
lnOUTFDI
jt
+
jt
.
(2)
Besides capturing geographic distance, the xed eects term
should also reect country-specic inuences such as factor
endowments, an aliation with a regional trading bloc, the
degree of tari protection, or trade resistance.
2
We specify a dynamic formulation of the model by including
the lagged value of exports as an explanatory variable. Further,
lagged values for outward and inward FDI are added to capture
1
Some previous empirical work has included aliate sales and even licensing
arrangements (to unaliated rms) as part of the total bilateral exchange between
countries.
2
Anderson and van Wincoop (2003) have derived a theoretic gravity equation
where exports are dependent upon multilateral trade resistance. Fixed eects
estimation may capture such resistance.
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1

358 THE INTERNATIONAL TRADE JOURNAL
the linkages between direct investment and export activities (see
Egger 2001). Relying on the xed eects estimator here may be
problematic because the lagged dependent variable is likely to be
correlated with the error term. A common approach used with
dynamic models is to adopt a rst-dierenced specication as in:
lnEXP
jt
=
1
lnEXP
j,t 1
+
2
lnGPDC
jt
+
3
lnPOP
jt
+
4
lnREXR
jt
+
5
lnINFDI
j,t1
+
6
lnOUTFDI
j,t1
+
jt.
(3)
Although this specication eliminates country-specic eects
that would otherwise be correlated with the regressors, it also is
likely to suer from endogeneity bias. Arellano and Bond (1991)
have popularized a rst-dierenced estimator that corrects for
time-invariant xed eects and endogeneity. Accordingly, GMM
estimation is useful because these techniques utilize instruments
for any endogenous variables. The use of alternative estimators
is discussed alongside empirical ndings in the following section.
Recall that our major focus is on how Canadian export
activity is inuenced by inward (and outward) FDI from (to)
specic trading partners. Consider the lingering question over
whether outward FDI to a specic country displaces exports that
would otherwise be sent to that locale. Yet it is conceivable that
Canadian direct investment to a specic country j (OUTFDI
j
)
may actually stimulate exports to that trading partner if there
were a corresponding increase in vertical production activities
at oshore locales. As such, the sign of the coecient (
6
) for
outward FDI is ambiguous.
The model explicitly examines the hypothesis that inward
FDI from a specic trading partner tends to stimulate exports to
that market as intra-rm trade expands; so a positive coecient
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1

Mullen and Williams: Bilateral FDI . . . 359
for INFDI
j
would conrm this eect. Such a nding may be par-
ticularly relevant for Canada in view of the nature of its economic
integration with the United States. As more U.S. aliates have
established operations in Canada, exports to the United States
have grown. However, empirical evidence is needed to determine
whether or not this is a causal relationship.
Data Sources
Eorts to conduct empirical analyses in this area are often
hampered by the nature and availability of data sources. For
example, FDI stocks at the country-industry level do not typi-
cally exist. The approach herein utilizes available foreign invest-
ment data for the aggregate economy of Canada, as they pertain
to specic trading partners. We employ a time series of FDI
position data for OECD countries to represent the stock of
FDI, generally dened as the book value of assets. As is typically
the case with FDI position data, there is no distinction made
between assets acquired via mergers/acquisitions vs. those gener-
ated through greeneld investments. These data are provided
by the International Direct Investment Statistics Yearbook and
Statistics Canada (most recent two years). Data on exports from
Canada to individual countries are also available from Statistics
Canada. Appendix A provides additional details and sources
for these and other remaining variables. Both the export and
FDI measures are stated in current dollar terms; accordingly,
we convert these values to constant Canadian dollars by relying
on implicit price indexes for exports and investment spending,
respectively (taken from Statistics Canada).
The index of real eective exchange rates is available from
International Financial Statistics published by the International
Monetary Fund. Population data and real GDP per capita
measures are taken from OECD sources. Also, the OECDs
Economic Outlook Database provides information on long-term
interest rates. Finally, data on corporate tax rates are constructed
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1

360 THE INTERNATIONAL TRADE JOURNAL
from international corporate tax rate surveys conducted by the
accounting rm of KPMG, and are available online from the
Institute for Fiscal Studies. To sum, the assembled data rep-
resent Canadas export and bilateral direct investment activity
with 20 major OECD
3
nations from 1989 through 2007. Some
of the 340 country-year observations were deleted due to missing
data for the FDI stock variables, yielding a panel of 16 countries
over 19 years.
IV. EMPIRICAL RESULTS
The empirical approach tests the sensitivity of results to
alternative estimation techniques. Table 1 reports ndings based
on a one-way (country) xed eects (FE) estimator for Equation
(2),
4
utilizing the dummy variable approach. The restricted
F-value for Model I conrms that country-specic attributes
exert a signicant eect on the level of exports shipped from
Canada, apart from the inuence of the other explanatory vari-
ables. Such a nding is not surprising for a variety of reasons.
Note that the distance between nations, an important consider-
ation in gravity models as a proxy for transportation costs, is
time invariant and is therefore treated here as a country xed
eect. Moreover, other country-specic factors are likely to be
operative. These would include an array of cultural, political,
and institutional forces that might inuence the level of exports
to a specic trading partner.
The ndings for the static specication in Model I report
coecient values for most variables that are in accordance with
3
Historical FDI stock data is generally incomplete or non-existent for a num-
ber of OECD countries, especially those that have been in transition to market
economies. Accordingly, the analysis here excludes these OECD countries: Czech
Republic, Finland, Greece, Hungary, Iceland, Poland, Portugal, Slovakia, and
Turkey.
4
Time eects were shown to be unimportant based on ndings from the two-
way FE model, and are not reported here.
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1

Mullen and Williams: Bilateral FDI . . . 361
Table 1
Fixed Eects Estimation [Dependent variable = ln REXP
it
]
a
Model I II III
Constant 6.66835 (1.01) 9.3571 (1.40) 8.80503 (1.35)
ln GDPC
it
0.796856

(6.55) 0.8004

(6.46) 0.760365

(6.26)
ln POP
it
0.752223 (1.29) 0.992885

(1.70) 0.867625 (1.52)

ln EXRATE
it
0.107718 (0.78) 0.083052 (0.60) 0.02584 (0.19)
ln OUTFDI
it
0.035955 (1.56)
ln INFDI
it
0.075236

(2.40)
ln REXP
i,t-1
0.131291

(4.14)
ln OUTFDI
i,t-1
0.025992 (1.19) 0.022884 (1.08)
ln INFDI
i,t-1
0.055432

(1.76) 0.046653 (1.51)

R
2
.976 .975 .977
# observations 304 304 304
F value 96.79

96.73

32.05

a
Numbers in parentheses are absolute values of t-statistics.

,

,

indicate statistically signicance at the 1%, 5%, and 10% levels,
respectively.
hypothesized eects. For example, our results suggest that inward
FDI from a specic country tends to have a stimulating eect on
exports to that nation. Also, there is no evidence that outward
FDI displaces exports to host countries. The coecient for our
exchange rate variable is positive but not statistically signicant;
this result oers no support for the theoretical view that a strong
currency depresses exports.
5
Similar ndings related to exchange
rates are fairly common in earlier work. One explanation holds
that exports may contain high import content (Abeysinghe and
Yeok 1998); this may be especially important as it relates to
re-exports of goods whose inputs have been sourced from the
United States. Also, Clausings (2000) analysis of U.S. exports
fails to produce conclusive evidence on the impact of exchange
rates. Finally, it is possible that this measure may be picking up
5
The choice of how to measure relative exchange rates is critical. We utilize an
index of real eective exchange rates for both countries; thus, EXRATE is dened
as the ratio of the Canadian value for this index relative to that for each specic
country.
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1

362 THE INTERNATIONAL TRADE JOURNAL
exchange rate pass-through, or perhaps is reective of the high
import content of Canadian exports.
In Model II, (one-period) lagged values replace current lev-
els of the FDI variables; the results are quite similar, especially
as they pertain to the FDI variables. The nal FE model
III also includes lagged exports as an explanatory variable.
Once again, we nd general agreement with the hypothesized
eects, and the performance of the FDI variables remains con-
sistent. Of course, these ndings may be suspect because the
lagged endogenous variable is correlated with the error term.
Accordingly, Generalized Method of Moments (GMM) techniques
are well-suited for estimating this type of dynamic model.
Following Arellano and Bond (1991), GMM is often employed
to address estimation concerns related to endogeneity, xed
eects, and/or non-stationarity problems. This estimator, which
is consistent in the absence of serial correlation in the dier-
enced residuals, proceeds by rst-dierencing the data and using
lagged levels as instruments for the (rst-dierenced) endogenous
variables. The basic dierenced GMM estimator is tted to our
dynamic Eq.(3); the ndings are shown in Table 2 and discussed
below.
Results from the two-step dierenced GMM estimator are
reported in column I.
6
Despite potential problems when the num-
ber of time periods is small, these estimates are considered to be
asymptotically ecient.
7
These ndings generally support the
hypothesized eects, as well as the primary results implied by
6
Note that the listed variables have been transformed via rst-dierencing with
the GMM estimator.
7
Results for the single stage estimation are qualitatively similar to those pre-
sented here, but with much larger standard errors; this nding is consistent with
Arellano and Bonds (1991) suggestion that the two-step method may produce
downward biased standard errors when the number of time periods is small. Also, a
diagnostic test (Sargan statistic) rejects the validity of the instruments used in the
single stage estimation.
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1

Mullen and Williams: Bilateral FDI . . . 363
Table 2
Export Demand Equations
$
[Dependent variable = ln REXP
it
]
$$
III IV
I
GMM-DIFF
II
GMM-SYS
GMM-DIFF
(excludes US)
GMM-SYS
(excludes US)
Constant 0.04467

(2.09)
0.007303
(0.82)
.023934
(1.09)
.00744
(0.88)
ln GDPC
it
0.626055

(3.82)
0.03669
(1.39)
0.702833

(4.17)
0.00687
(0.14)
ln POP
it
2.18818
(1.02)
0.118728

(2.36)
0.1763
(0.10)
0.237263

(2.74)
ln EXRATE
it
0.22077
(1.01)
0.237396

(2.79)
0.31012
(1.00)
0.359353

(2.75)
ln REXP
i,t-1
0.13244
(0.79)
0.73901

(11.18)
0.141121
(0.81)
0.547203

(7.08)
ln OUTFDI
i, t-1
0.04842
(0.66)
0.002097
(0.07)
0.2180

(1.72)
0.01667
(0.31)
ln INFDI
i,t1
0.180267

(2.27)
0.143433

(3.19)
0.153468

(2.29)
0.139433

(2.97)
Sargan 6.69
[>1.00]
28.98
[>1.00]
8.21
[>1.00]
27.65
[>1.00]
Sargan-di N.A. 22.29
[>1.00]
N.A. 19.44
[>1.00]
$
Statistical signicance, for coecient values, at the 1%, 5%, and 10% levels
are noted by

,

, and

respectively; numbers in parentheses ( ) are absolute
values of t-statistics. Numbers reported in square brackets [ ] are p-values for
diagnostic tests.
$$
All variables are transformed into rst dierences via the GMM technique;
note that system GMM includes levels equations stacked with those
based on rst dierences alone.
the FE model. Notably, the lagged value of inward FDI contin-
ues to display a positive and statistically signicant impact on
Canadian exports.
However, the performance of two explanatory variables diers
slightly from the FE estimation. Specically, the coecient on the
exchange rate variable is now negativea result consistent with
theoretical expectationsbut remains statistically insignicant.
Also, we now observe that the coecient value for lnOUTFDI
i, t-1
is negative, but once again fails to show statistical signicance.
Thus, the impact of outward FDI on exports remains uncertain.
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1

364 THE INTERNATIONAL TRADE JOURNAL
Finally, the results here now show that lagged exports do not
have a statistically signicant eect on current exports. Recall,
however, that the previous FE estimator may not be appropriate
for the underlying dynamic model since lagged exports represent
an endogenous explanatory variable.
The diagnostic tests performed in conjunction with this
GMM estimator are supportive of our approach. In contrast to
the single-stage results (not reported here), the Sargan test statis-
tic fails to reject the null hypothesis that the instruments are
exogenous. Additionally, an autocorrelation test of order two is
used to test if the rst-dierence residuals exhibit second-order
serial correlation; we fail to reject the null of no serial correlation
in all instances, a nding that corroborates valid instrumentation.
One potential problem with the above estimator is that
lagged levels are often poor instruments for rst-dierences. In
fact, the system GMM estimator (Arellano and Bover 1995) is
commonly used to increase eciency when the data series has a
high degree of persistence. The systems approach adds a levels
equation to the rst-dierenced equation, so that the predeter-
mined and endogenous variables (in levels) are instrumented with
lags of their own rst dierences. We also use two external vari-
ables as instruments; specically, a countrys prevailing long-term
interest rate (LTIR) and its statutory tax rate (STR) are used as
exogenous instruments.
8
For comparison, we apply this system
GMM estimator to the model; results are reported in Column II
(Table 2).
Although some dierences in coecient values and signi-
cance levels emerge from system GMM, the story revealed by
the FDI variables remains largely intact. Once again, inward
FDI has a stimulative eect on Canadian exports. The impact
of outward FDI remains ambiguous however; although there is
8
Note that these variables, though not included as regressors in the model, are
suitable instruments as they are unlikely to be correlated with the error term but
inuence levels of foreign direct investment.
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1

Mullen and Williams: Bilateral FDI . . . 365
a sign change in the coecient value, it remains statistically
insignicant. Finally, the dierence-Sargan test statistic indicates
the validity of the additional instruments used in the systems
approach here.
9
Without regard to which may be the ideal esti-
mator, our ndings demonstrate consistency pertaining to the
inuence of both inward and outward FDI on export activity.
In what follows, we extend the analysis by re-estimating these
models after excluding all U.S. observations from the panel data.
The rationale for doing this is that the magnitude of Canadas
trade and investment ows with the United States may produce a
distorted view of its international economic relationships with its
remaining major trading partners. For comparisons with the full
sample results, Columns III and IV report these ndings based on
the dierenced GMM and system GMM estimators, respectively.
The ndings are generally consistent with those based on the
full sample of 16 country observations. Once again, the diag-
nostic tests conrm the validity of the instruments; specically,
the autocorrelation test statistic and (both) Sargan statistics
fail to reject the null hypothesis of serial independence and
instrument exogeneity respectively. A comparison of Columns
I and II with III and IV shows that the performance of the
explanatory variables remains largely unaltered. Interestingly, we
continue to observe that (country-specic) inward FDI stimulates
export activity. The important implication here for Canada is
that export growth driven by inward FDI, perhaps arising from
intra-rm trade, is not strictly a phenomenon associated with its
dominant trading partner. So the general picture that emerges
is that the relationship between Canadian exports and bilateral
FDI, as it pertains to the U.S. economy, is not markedly dierent
from the broader international pattern. One possible exception is
noteworthy however. While it is far from conclusive that outward
9
Roodman warns that the power of this test is diminished with the use of either
too few or too many instruments, but still advocates its use in distinguishing
the system from the dierence GMM estimator.
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1

366 THE INTERNATIONAL TRADE JOURNAL
FDI displaces exports to a host economy, the results here suggest
that this may be more likely to occur as it pertains to Canadas
relationship with trading partners other than the United States.
Note that, when excluding the U.S. data, we observe a nega-
tive (and statistically signicant in one case) coecient for the
outward stock of FDI. One possible interpretation here is that
outward FDI to the United States may result in a signicant
expansion of intra-rm trade between Canada-based rms and
their U.S. aliates. This nding remains speculative, however.
V. SUMMARY AND IMPLICATIONS
The present research has explored the nature of the rela-
tionship between Canadian export activity and direct foreign
investment. Our analysis relies on existing data sources, includ-
ing OECD direct investment statistics that record stocks of
both inward and outward FDI from a Canadian perspective.
By employing a gravity-type empirical specication, we test for
the short-run impact of inward and outward FDI on exports to
OECD countries. We consider a number of alternative estimators
so that the sensitivity of the results pertaining to key variables
may be examined; in particular, GMM estimation is utilized to
deal with the dynamic nature of the model.
The evidence presented here is interesting, but not provoca-
tive, concerning the performance of variables traditionally used
to explain exports. As expected, per capita GDP generally
demonstrates a positive relationship with exports; on the other
hand, the impact of relative exchange rates on export activity
remains ambiguous. It is conceivable that our exchange rate mea-
sure (dened as the ratio of an index of real eective exchange
rates
10
) may not appropriately capture the pairwise exchange
rate dierentials between Canada and specic trading partners.
10
Recall that real eective exchange rates are constructed by looking at trade-
weighted variations in domestic prices and nominal exchange rates; as such, they
reect changes in each country relative to all of its trading partners.
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1

Mullen and Williams: Bilateral FDI . . . 367
But the primary focus of this study concerns the relation-
ship between bilateral FDI and export activity. The evidence
presented here demonstrates that inward FDI from a specic
source is associated with more exports owing to that country.
This supports the contention that when foreign rms locate in a
host economy they often expect this will expand intra-rm trade,
as is manifested through export growth. It would be much more
dicult, and beyond the scope of this article, to determine what
specic proportion of export growth is attributable to intra-rm
trade per se.
The impact of outward FDI on export activity remains
more ambiguous. For most of the specications examined here,
Canadian FDI into a host economy (OUTFDI) had no signicant
eect on exports to that country. A straightforward interpre-
tation of this nding is that Canadian exports do not appear
to be displaced by direct investment abroad; this implies an
increase in intra-rm trade from Canadian rms to their for-
eign aliates. However, this nding is tempered somewhat once
trade-investment ows with the United States are ignored. Now
we nd some evidence that exports might actually be harmed by
outward FDI. Although speculative, this result implies that tari-
jumping could be part of the motivation behind the increase in
outward FDI by Canadian rms. Yet the evidence here further
implies that this may be less of a motivating factor for direct
investment into the United States, perhaps due to lower taris
via NAFTA provisions.
The overall impacts of FDI remain inconclusive for several
reasons. For one, it is quite possible that outward investment
substitutes for exports in certain industries, even as it stimulates
trade in others. Clearly, an appropriate level of disaggregation of
existing stock or ow data (currently available only for broad
sectors of the Canadian economy) is necessary to yield these
additional insights. More generally, we urge caution in broadly
interpreting our ndings. For example, it is easy to speculate
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1

368 THE INTERNATIONAL TRADE JOURNAL
that outward foreign investment may be slow in enhancing the
international competitiveness of Canadian rms; yet recall that
our analysis examines how outward investment aects exports
to a specic host economy, not overall exports to all trading
partners. Also, our empirical results should not be interpreted as
oering support for the notion that inward FDI stimulates total
exports as a result of generalized productivity spillovers to the
host economy. Nevertheless, the evidence here implies support
for policies to dismantle barriers to both inbound and outbound
foreign investment. Besides increasing the exposure of domes-
tic rms to global competition, such eorts appear unlikely to
shrink the overall level of exports from Canada, especially when
considering the role of the U.S. economy.
The ndings presented here are suggestive of research initia-
tives that may further illuminate how globalization trends aect
the Canadian economy. For example, the role of intra-rm trade
in accounting for export growth holds implications for the open
debate on the hollowing out of Canadian manufacturing. Also,
a fuller exploration of the causal links between bilateral FDI and
both imports and exports may be enlightening.
REFERENCES
Abeysinghe, T., and Yeok, T. L. 1998. Exchange Rate
Appreciation and Export Competitiveness: The Case of
Singapore. Applied Economics 30(1): 5155.
Amiti, M., and Wakelin, K. 2003. Investment Liberalization
and International Trade. Journal of International Economics
61(1): 101126.
Anderson, J. E., and van Wincoop, E. 2003. Gravity with
Gravitas: A Solution to the Border Puzzle. American
Economic Review 93(1): 170192.
Arellano, M., and Bond, S. 1991. Some Tests of Specication for
Panel Data: Monte Carlo Evidence and an Application to
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1

Mullen and Williams: Bilateral FDI . . . 369
Employment Equations. Review of Economic Studies 58(2):
277297.
Borenzstein, E., and de Gregorio, J. 1995. How Does Foreign
Direct Investment Aect Economic Growth? NBER Working
Paper no. 5057. Cambridge, MA: National Bureau of
Economic Research (NBER).
Brainard, S. 1993. A Simple Theory of Multinational
Corporations and Trade with a Trade-o between Proximity
and Concentration. NBER Working Paper 4269. Cambridge,
MA: NBER.
Clausing, K. A. 2000. Does Multinational Activity Displace
Trade? Economic Inquiry 38(2): 190205.
Deardor, A. V. 1998. Determinants of Bilateral Trade: Does
Gravity Work in the Neoclassical World? In J. Frankel (Ed.),
Regionalization of the World Economy, pp. 731. Cambridge,
MA: NBER.
Eaton, J., and Tamura, A. 1994. Bilateralism and Regionalism
in Japanese and U.S. Trade and Direct Foreign Investment
Patterns. Journal of Japanese and International Economics
8(4): 478510.
Egger, P. 2001. European Exports and Outward Foreign
Direct Investment: A Dynamic Panel Data Approach.
Weltwirtschaftliches Archiv 137(3): 427449.
Feinberg, S., and Keane, M. 2001. U.S.-Canada Trade
Liberalization and MNC Production Location. Review of
Economics and Statistics 83(1): 118132.
Foreign Aairs and International Trade Canada. 2008. Canadas
State of Trade and Investment Update. http://www.inter
national.gc.ca/economist (accessed February 15, 2009).
Head, K., and Ries, J. 2001. Overseas Investment and Firm
Exports. Review of International Economics 9(1): 108122.
Hejazi, W., and Safarian, A. E. 2001. The Complementarity
between U.S. Foreign Direct Investment Stock and Trade.
Atlantic Economic Journal 29(4): 420437.
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1

370 THE INTERNATIONAL TRADE JOURNAL
Holden, M. 2008. Overview of Canadian Foreign Direct
Investment. Ottawa, Canada: Parliamentary Information and
Research Service (Library of Parliament).
Lipsey, R. E., and Weiss, M. E. 1984. Foreign Production
and Exports of Individual Firms. Review of Economics and
Statistics 66(2): 304308.
Markusen, J. R., and Venables, A. 1998. Multinational Firms and
the New Trade Theory. Journal of International Economics
46(2): 183203.
Organization for Economic Cooperation and Development
(OECD). 1998. Trade Competition and Foreign Direct
Investment: A New Assessment (DSTI/EAS/IND/SWP
(98)9). Paris: Statistical Working Party of the Industry
Committee.
Pain, N., and Wakelin, K. 1998. Export Performance and the
Role of Foreign Direct Investment. The Manchester School
66(Suppl): 6288.
Rao, S., Ashfaq, A., and Legaul, M. 1994. Canadian Based
Multinationals: An Analysis of Activities and Performance.
In S. Globerman (Ed.), Canadian Based Multinationals,
pp. 63123. Ottawa, Canada: Industry Canada Research.
Roodman, D. 2007. A Short Note on the Theme of Too Many
Instruments. Working Paper No. 125 (August). Washington,
DC: Center for Global Development.
Sachs, J. D., and Shatz, H. D. 1994. Trade and Jobs in U.S.
Manufacturing. Brookings Papers on Economic Activity 1:
169.
Smith, P. J. 2001. How Do Foreign Patent Rights Aect
U.S. Exports, Aliate Sales, and Licenses? Journal of
International Economics 55: 411439.
Stevens, G., and Lipsey, R. E. 1992. Interactions between
Domestic and Foreign Investment. Journal of International
Money and Finance 11(1): 4062.
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1

Mullen and Williams: Bilateral FDI . . . 371
Appendix A
Data Sources and Denitions
Variable Denition Source
REXP
it
Real Exports (in constant Cdn.
$) to country i at time t
Strategis website
(http://www.strategis.ca);
CANSIM Table No.2270001
OUTFDI
it
,
INFDI
it
Outward and Inward stocks of
FDI (in constant Cdn. $)
to/from country i at time t
International Direct
Investment
StatisticsYearbook, 2008,
edition. CANSIM Table No.
3760051, Paris: OECD.
EXRATE
it
Real Exchange Rate Index for
country i at time t
International Monetary Fund,
IFS Online
(http://ifs.apdi.net/imf/)
POP
it
Population of country i at time t http://www.oecd.org
DIST
i
Distance (km) to national capital
city
http://www.eiit.org/
GDPC
it
Real

GDP per capita (in US $) World Economic Outlook

Database, September 2009.
http://www.imf.org/
IR
it
Long-term interest rate in
country i at time t
OECD StatExtracts, derived
from OECD Economic
Outlook, Nov.2008
TAXR
it
Statutory corporate tax rate in
country i at time t
KPMG Corporate Tax Rate
Survey (various issues);
Institute for Fiscal Studies
http://www.ifs.org.uk
Note: Current US $ values converted into constant $ terms via each countrys
GDP deator (also from IMFs World Economic Outlook, Sept. 2008).
D
o
w
n
l
o
a
d
e
d

b
y

[
U
n
i
v
e
r
s
i
t
i

P
u
t
r
a

M
a
l
a
y
s
i
a
]

a
t

2
3
:
4
2

0
9

N
o
v
e
m
b
e
r

2
0
1
1