The Agglomeration of Exporters by Destination
Andrew J. Casseya
a
School of Economic Sciences, Washington State University, 101 Hulbert Hall, P.O. Box 646210, Pullman, WA
99164, USA. Phone: 509-335-8334. Fax: 509-335-1173. Email: [email protected].
Katherine N. Schmeiserb,∗
b
Department of Economics, Mt. Holyoke College, 117 Skinner Hall; 50 College St, South Hadley, MA 01075, USA.
Phone: 612-910-2538. Email: [email protected].
Abstract
Precise characterization of informational trade barriers is neither well documented nor understood,
and thus remain a considerable hindrance to development. Using spatial econometrics on Russian
customs data, we document that destination-specific export spillovers exist for developing countries,
extending a result that was only known for developed countries. This result suggests behavior
responding to a destination barrier. To account for this fact, we build on a monopolistic competition
model of trade by postulating an externality in the international transaction of goods. We test the
model’s prediction on region- and state-level exports using Russian and U.S. data. Our model
accounts for up to 40% more of the variation than in gravity-type models without agglomeration.
This finding has important development implications in that export policy that considers current
trade partners may be more effective than policy that focuses only on the exporting country’s
industries.
Keywords: agglomeration, distribution, exports, development, location, Russia
JEL: D23, F12, L29
∗
Corresponding author
Preprint submitted to .
September 21, 2011
1
1. Introduction
2
One of the unsolved problems in economic development is the lack of knowledge about the
3
size and nature of barriers to trade, and how these barriers affect growth and income. Great
4
theoretical and empirical strides have been made by country-level and firm-level studies, yet the
5
precise characterization of barriers to trade continues to elude researchers. A better understanding
6
of the size and nature of barriers to trade may lead to better trade-focused development policy.
7
Using spatial econometrics on Russian customs data, we document that firms exporting to
8
the same destination are located near each other. This agglomeration is beyond the clustering of
9
exporters around domestic ports (Glejser, Jacquemin, and Petit 1980; Nuadé and Matthee 2007)
10
or gross domestic product (GDP). We document that destination-specific export spillovers exist
11
in a developing country such as Russia. Previous empirical studies first established that export
12
spillovers exist and are destination specific, but the data are from developed countries such as
13
the United States (Lovely, Rosenthal, and Sharma 2005), France (Koenig 2009), and Denmark
14
(Choquette and Meinen 2011). We also document destination-specific exporter spillovers using
15
spatial econometrics instead of a logit estimator. Therefore our results confirm the previous work
16
using a different method, and extend it to developing countries.
17
Additionally, we develop a theory to account for this destination-specific agglomeration in a
18
trade setting. We assume there are economies of scale in the transaction cost of international trade
19
at the firm level. In principle, destination-specific export spillovers could work through the fixed
20
or variable costs of trade. Econometric results by Choquette and Meinen (2011) provide evidence
21
for export spillovers working through variable costs as well as fixed costs. Furthermore, though
22
Koenig, Mayneris, and Poncet (2010) do not find robust support for spillovers working through
23
the variable cost, they do find evidence of this when the total number of exporters is small. Our
24
Russian data shows that the number of exporters selling to the same country from the same region
25
in Russia is often small, and thus are more likely to have the destination-specific export spillovers
26
working through variable costs. Notwithstanding these results, both Choquette and Meinen and
27
Koenig et al. measure agglomeration exclusively as the ratio of similar exporting firms to total
28
firms within some geographic area. With this measure, they find that destination-specific export
2
29
spillovers work primarily through the fixed cost. We try alternative measures of agglomeration and
30
find that variables of group activity such as the aggregate region-country weight of shipments work
31
more through the variable costs of trade.
32
The Russian customs data show that, by far, most shipments are small in weight. As long as
33
shipping prices to exporters include a fixed cost per container (buying the container) or per desti-
34
nation (permit to unload) and are based on weight (Micco and Pérez 2002), then there are shipping
35
cost savings in grouping small exports together on a boat or plane heading to the same destination.
36
Alternatively, aggregate export weight may be thought of as the result of an informational spillover
37
about foreign market access. The exact nature of the externality is not important for our model or
38
results. However, we do offer some initial evidence in distinguishing between the two. We test our
39
theory using region-level data from Russia and state-level data from the United States.
40
We find that using the region-level export data from Russia, our model with agglomeration
41
functioning through the variables costs of trade accounts for 10% more of the variation in export
42
share than the benchmark model. Using U.S. state export data, our model accounts for 40% more
43
of the variation. Therefore, we have uncovered empirical evidence of a barrier to trade large enough
44
for firms to respond to it by agglomerating and we provide theoretic justification for it.
45
In our model, an exporter may skip over a destination that has more appealing demand in order
46
to achieve decreased transactions costs in selling to a less preferred destination. All else the same,
47
this creates variation in the destination choices of otherwise identical firms in different regions.
48
Lawless (2009) documents such skipping of “preferred” destinations at the national-level, but her
49
evidence is inconsistent with the theory developed in Eaton, Kortum, and Kramarz (Forthcoming).
50
Our theory better accounts for the geographic pattern of exporting by creating a regional hierarchy
51
of export destinations that differs from the national hierarchy.
52
That exporter clustering by destination exists for both developed and developing suggests firms
53
are responding to destination-specific transaction costs of trade. This finding offers insight into
54
improvement of trade-focused development policy in that the policy should perhaps respect the
55
destination of exports in addition to the exporting country’s industrial base and comparative ad-
56
vantage.
3
57
The transaction-level data that motivates our theory are from Russia. We choose to study
58
Russian exporters because Russia is a developing country that is geographically large and diverse.
59
It borders fourteen countries (with China and Ukraine being major trading partners), has four
60
coasts (Arctic Ocean, Pacific Ocean, Baltic Sea, and Black Sea) and diverse economic subregions.
61
Finally, the laws in Russia limit the movement of firms regionally. (See Appendix A for a more
62
detailed description). This allows us to model a firm’s export destination decision without also
63
modeling its location decision within Russia. Furthermore, though Russia has a unique history
64
that may affect development (relationship with former Soviet countries, corruption, and remnants
65
of central planning to name a few), Cassey and Schmeiser (2011) show that the Russian data
66
qualitatively and quantitatively match firm-level data sets from Colombia and France. Therefore
67
the development lessons from this work apply more generally than Russia alone.
68
The findings of Cassey and Schmeiser (2011) notwithstanding, because we are aware that Rus-
69
sian data may be different from other countries’ data, we test the macro-level predictions of our
70
theory on the World Institute for Strategic Economic Research’s (WISER) Origin of Movement
71
U.S. state export data (OM). These data contain export information for all 50 U.S. states and the
72
same 175 countries that we use with the Russian data. Furthermore, our U.S. data are a panel
73
whereas the Russian customs data are a cross-section. Having the time dimension allows us to
74
conduct robustness tests on the U.S. data that we cannot do with the Russian data.
75
2. Facts on the location & agglomeration of Russian exporters
76
In this section, we show that 1) exporters are clustered regionally given output and the location
77
of ports and 2) exporters shipping to the same destination are further clustered. The purpose of
78
this section is to establish these two agglomeration facts using a spatial econometrics method. Our
79
empirical method differs from that used by Koenig (2009) and Choquette and Meinen (2011), who
80
use a logit estimator to see if the odds that a firm exports to a particular country depends on the
81
ratio of other firms in the same area also exporting to that country to total firms, and also from
82
that of Lovely et al. (2005) who put an industry concentration variable on the left hand side of
83
a linear model with the difficulty of reaching an export destination on the right hand side. Our
4
84
interest differs from the previous work in that we use data from a developing country instead of
85
a developed country. Therefore, one way of interpreting the purpose of this section is establishing
86
the generality of the above-mentioned agglomeration facts across countries in various stages of
87
development. Because of the uniqueness of our data, we begin by describing them.
88
2.1. The Russian data
89
To document the existence of destination-specific agglomeration of exporters, we use 2003 data
90
from the Russian External Economic Activities (REEA) set reported by Russian customs. The
91
REEA data are shipment-level customs data of uniquely identified firms, providing value of goods
92
shipped (in 2003 USD), weight of shipment, exporter location within Russia, and destination coun-
93
try. See Cassey and Schmeiser (2011) for complete description of these data, as well as descriptive
94
statistics and stylized facts.
95
We only consider manufacturing exports because firms that exported natural resources were
96
necessarily located in the region endowed with those resources (Bradshaw 2008). In addition, we
97
only use manufacturing data so that we can later have a direct comparison with the OM data for
98
the United States.1
99
From information on the customs form, we use data from firms who are making the export
100
decision—who may or may not be the firm that manufactured the final good. We do this because
101
we want to use the location of the exporter and not the producer (if they are different). Therefore,
102
we include manufacturers, distributors, and wholesalers who are coded as the principal agent in
103
interest. We do not include freight-forwarding firms because these firms could bias results if they
104
are located somewhere other than the true exporter’s location. Also, we eliminate observations in
105
which the “originating firm” is located outside of Russia since we believe they indicate re-exports
106
rather than true Russian exports, as well as to be consistent with the U.S. data.
107
To check for consistencies in the data, we refer to Cassey and Schmeiser (2011) who compare
108
the REEA data with product data from the United Nations and find close agreement. They
1
Cassey (2009) shows that for the United States, manufacturing data are the only data that can be reliably used
for the state of production of the export instead of the state where the export began its journey abroad, whereas
agriculture and mining data are not reliable.
5
109
also compare the REEA data to statistics obtained with firm-level data from Colombia (Eaton,
110
Eslava, Kugler, and Tybout 2008) and France (Eaton et al. Forthcoming). The export patterns
111
are qualitatively and quantitatively similar. We are therefore confident that our data are not
112
greatly influencing or biasing our interpretation or generality of results, in particular towards any
113
geographic region.
114
We combine the REEA data with information about the the gross domestic products (GDP) of
115
each region using Russia: All Regions Trade & Investment Guide (CTEC Publishing 2004, 2006).
116
(We do not have data on the number or size of domestic firms.) We also use the World Economic
117
Outlook Database (IMF 2006) for GDP information for the 175 countries in our sample. Finally,
118
we calculate the great circle distance from the capital of each Russian region to the capital of each
119
country in the world.
120
Our data have two large drawbacks that limit the forthcoming empirical analysis. First, though
121
we have transaction-level data on shipments that we can match to unique firms, we are not able to
122
pinpoint exporter location by exact physical address. The best we can do is assign each exporter
123
to one of Russia’s 89 federal regions.2 This limitation forces us to conduct clustering tests on
124
the region-level (and therefore to later perform econometric tests of the model at the region-level
125
rather than the firm-level). See Appendix A for a detailed description of the political organization
126
of Russia and also for a table listing the regions. Second, we only have the customs data and
127
thus we are not able to match Russian firms with their characteristics. We do neither know the
128
employment, age, or total sales of the firms in our data nor do we know anything about firms that
129
are not exporting. Therefore we cannot compare the clustering of Russian manufacturing exporters
130
to that of the Russian manufacturing industry or subsectors therein, nor systematically exclude
131
other sources of firm clustering such as natural, labor, or infrastructural advantages. Given this
132
limitation, we instead show that Russian firms exporting to the same country are non randomly
133
located within Russian regions, controlling for variables such as distance, importer GDP, and
134
preferences of the importer for particular goods. This kind of clustering cannot be fully explained
2
Russia’s federal regions are somewhat similar in geographic scope to U.S. states. The number of regions has decreased
since 2003 because of mergers. It is common to study Russia at this level of geographic disaggregation. See Broadman
and Recanatini (2001) for example.
6
135
without destination-specific export spillovers.
136
2.2. Russian exporters and exports
137
First, consider the number of exporting firms by region. Figure 1 shows that the regional
138
concentration of Russian manufacturing exporters is diverse. The majority of exporters are located
139
either around Moscow and St. Petersburg or the Kazakhstan/Mongolia/China border. Nine regions
140
did not have any reported exporting firms in 2003. Figure 1 shows that in Russia, as elsewhere,
141
many exporting firms are located near major ports. However, the figure also shows that there are
142
regions with many exporting firms that are not located near a major port. A map of Russia with
143
major rivers and railroads is in Appendix A.
144
The top destinations for Russian exports in 2003 were, in order: China, United States, United
145
Kingdom, Japan, Ukraine, Kazakhstan, Turkey, Netherlands, Germany, and Iran. Of these, only
146
Ukraine and Kazakstan were part of the Soviet Union. Additionally, Russia does not currently
147
have a Warsaw Pact nation as a top destination.3 Because we do not see evidence of favored trade
148
relations with former Soviet or Eastern bloc countries, we do not treat them any differently than
149
Western countries.
150
Federal laws in Russia severely limit the ability of Russian firms to change region (Botolf 2003).
151
Because of this, firms choose how much to export to each country in the world, taking their location
152
as given. Furthermore, because the Soviet Union did not trade with western countries, we believe
153
it is likely that the Soviet central planners chose firm locations for reasons other than the expansion
154
of trade (see Huber, Nagaev, and Wörgötter (1997) and Bradshaw (2008)). Other evidence that
155
international trade was not important in Soviet economic planning is that the large Pacific port of
156
Vladivostok, home of the Soviet Pacific Fleet, was closed to foreign vessels until 1991.
157
2.3. The geography of Russian exporters
158
To show that exporters are clustered, we first use the customs data to calculate the number of
159
firms in each region that export anywhere in the world. Figure 1 shows this as well as the location
3
Soviet trade was with countries of similar ideology such as Yugoslavia and Poland.
7
8
Figure 1: Number of exporting firms, by region. Cut-offs correspond to quintiles so that there are an equal number of regions in each category. See Appendix
A for a map of Russia with the regions labeled and railways indicated.
160
of ports, bodies of water, and neighboring countries, without modification for population or GDP.
161
Cut-offs for the groups are determined so that each group has the same number of regions.
162
Next, we weight the exporter count observations by regional GDP. We compare each region’s
163
exporter count to the overall mean. Figure 2 highlights regions with a concentration of exporters
164
far from the mean. Darker colors indicate regions whose exporter count, weighted by GDP, is much
165
larger than the mean. As seen in the figure, there exist regions in Russia that have statistically
166
significantly more exporters than expected based on regional GDP. Furthermore, many of these
167
export-concentrated regions do not have a major port. Figure 2 also shows there are regions
168
(weighted by GDP) with less exporting firms than the mean. These regions, indicated with the
169
lightest shades, are not nearly as far below the mean as the regions much larger than the mean.
170
We use this information to calculate the Moran’s I statistic for spatial autocorrelation. The
171
expectation of the statistic is −1/(89 − 1) under a GDP-weighted random distribution of exporters
172
to Russia’s 89 regions. With our data, the Moran’s I statistic for spatial autocorrelation is 0.16∗ (z−
173
score = 6.37) indicating that we reject spatial (GDP-weighted) randomness with 99% confidence.
174
Additionally, the clustering is not based on regions that have major ports. In Appendix C, we
175
conduct two other tests to confirm this result.
176
To see what the clustering is based on, consider a specific example: the location of Russian
177
firms that export to the United States and the location of Russian firms that export to Canada.
178
We choose these two countries because they have similar economies (though the United States is
179
much larger), distance from Russia, and consume similar goods (at the 2 digit level, six of the
180
top ten imported industries from Russia are the same). The Moran’s I for the location of Russian
181
exporters shipping to the United States is 0.04∗∗∗ (z = 1.93) and it is 0.03∗∗∗ (z = 1.68) for Canada.
182
Thus there is clustering for exporters to the United States and Canada with 90% confidence. To
183
see that it is not the same regions (and therefore the same firms within a region) that specialize in
184
exporting to both countries, see figure 3.
185
Figure 3 shows the deviation from the mean number of firms that export to the United States
186
and the mean number of firms that export to Canada. Notice that there are more regions in Russia
187
that have exporters shipping goods to the United States than Canada, but that is not surprising
9
ALL EXPORTERS
Figure 2: Deviations from the average number of exporting firms, by region, weighted by GDP. Moran’s I (89) = 0.16∗
(z − score = 6.37).
188
given the difference in GDP and that the United States is the second largest receiving country of
189
Russian exports. The second fact is that the regions where exporters shipping to the United States
190
are most concentrated (shaded black in figure 3 and indicating more than 2.5 standard deviations
191
from the mean) are not the same regions as those in which exporters to Canada are concentrated.
192
Therefore, Russian exporters to the United States are clustered in a different pattern than the
193
Russian exporters to Canada.
194
To support this claim, we perform a bilateral Moran’s I statistic for these two countries. The
195
statistic differences the number of exporters in each region i shipping to country c from the average
196
number of exporters also shipping to c and compares it to the difference between the number of
197
exporters in another region j shipping to country d from the average also shipping to d. Because
198
of its construction with respect to the other country d, the bivariate Moran’s I is not symmetric.
199
A statistic different from E(I) suggests a spatial pattern (clustering if positive) between exporters
10
UNITED STATES
CANADA
Figure 3: Deviations from the average number of exporting firms, by region, weighted by GDP. For the Unites States,
Moran’s I (89) = 0.04∗∗∗ (z = 1.93) and for Canda, Moran’s I (89) = 0.03∗∗∗ (z = 1.68).
11
Canada
China
Germany
Great Britain
Japan
Poland
United States
Ukraine
Table 1: Moran’s I for Location of Firms Exporting to Select Countries
Canada
China
Germany
Great
Japan
Poland United Ukraine
Britain
States
.0345∗ −.0012
.0243
.0313
.0300
.0196
.0194
.0340
.0089
.0023
−.0448∗ −.0458∗ −.0147 −.0450∗ −.0375∗ −.0408∗
.0249 −.0472∗
.0324∗
.0610∗∗ .0248
.0547∗
.0406
.0637∗∗
∗
∗∗
∗∗∗
∗∗
∗∗
.0215 −.5160
.0677
.0886
.0676
.0277
.0783
.0916∗∗∗
∗
∗∗
.0221 −.0164
.0319
.0533
.0227
.0337
.0593
.0677∗∗
∗
∗
.0146 −.0514
.0581
.0370
.0446
.0115
.0450
.0658∗∗
.0189 −.0367∗
.0413
.0571∗ .0525∗∗ .0307
.0369∗
.0751∗∗
∗
∗
∗∗
∗∗
∗∗
.0258 −.0426
.0593
.0801
.0626
.0520
.0773
.0220
Source: Cassey (Forthcoming)
Notes: Observations on the diagonal are the Moran’s I for the location of exporters shipping
to that country. Off-diagonals are the bivariate Moran’s I representing the degree of spatial
correlation between the exports to the “row” countries and exports to the “column” countries. Observations have been normalized with respect to regional GDP. Bold observations
are the bivariate statistic where both countries have a statistically significant univariate Moran’s I.
∗
,
∗∗
, and
∗∗∗
indicate p-values less than .1, .05, and .01.
200
shipping to country c and those shipping to country d. The Moran’s I for Canada–U.S. is 0.0194 and
201
for U.S.–Canada it is 0.0189. Neither of these statistics is different from E(I) indicating that the
202
geographic location of exporters shipping to country c is statistically different from those shipping
203
to country d.
204
This pattern of exporter agglomeration holds when we expand the sample. Consider table 1,
205
taken from Cassey (Forthcoming). Observations on the diagonal are starred if there is statistically
206
significant clustering, indicating that Russian exporters are clustered around the destinations of
207
Canada, Germany, Great Britain, and the United States. The lack of stars on the off-diagonal
208
indicates the clustering pattern differs to these two countries if there are stars on the diagonal. (Stars
209
on the off-diagonal indicate the spatial autocorrelation of Russian exporters is the same for the two
210
countries.) Therefore, the pattern of clustering is not the same for seven of twelve possibilities (in
211
bold in table 1). One advantage of using the Moran’s I to find exporter agglomeration is that is is
212
easier to identify for which specific countries there is clustering. Our evidence contrasts somewhat
213
with that in Lovely et al. (2005) in that we find clustering around relatively open countries rather
214
than relatively closed countries that are difficult to export to. Perhaps this is a difference between
215
a developing country such as Russia and a developed country such as the United States or perhaps
216
this is due to Russia’s relationship with countries otherwise considered relatively closed.
12
217
To verify that this clustering by destination of exported shipments holds more generally, we
218
consider the location of the destination relative to Russian firms using the gravity equation. Though
219
the gravity equation takes several forms, we use the version derived by Anderson and van Wincoop
220
(2003). In this formulation, the exponent on exporter GDP, Yi , and importer GDP, Yj , must be
221
one, so we divide both sides by exporter and importer GDP. Aggregate region-country exports are
222
Xij and great circle distance are Dij . As recommended by Feenstra (2004, p. 161), we insert an
223
export-specific set of binary variables, Si , and an import-specific set of binary variables, Tj , to
224
account for the unobservable multilateral resistance terms. Using Ordinary Least Squares (OLS),
225
we estimate,
log
Xij
Yi Yj
= − 7.72
(1.033)∗∗∗
− 1.32 log Dij +
N = 2985,
(.110)∗∗∗
R̂2
89
X
κi Si +
i=2
175
X
δj Tj + εij
(1)
j=2
= 0.58, RM SE = 2.03
226
where ∗∗∗ on the robust standard errors indicates the corresponding coefficient is significantly differ-
227
ent from zero at with 99% confidence. The regression results suggest that the overall explanatory
228
power of the gravity equation to account for Russian regional exports is low compared to the
229
R2 ≥ 0.7 common in country-level data sets that often do not use binary variables to increase the
230
goodness of fit.4
231
We consider the relatively low adjusted R2 from using subnational data (compared to the R2
232
on national data) in a gravity equation as evidence that something is missing from this equation
233
that applies to the regional level, but not at the geographic scope of the national level. Given that
234
many theories of trade predict a gravity equation-like relationship—Anderson (1979) and Chaney
235
(2008) for example—we have documented a weakness in existing theory applied to subnational
236
data. See Appendix D for the estimation of gravity nonlinearly to address the fact that only 19%
4
The estimates for a “naive” gravity equation are
log Xij
=
− 0.84 +
(0.580)
0.56
(0.043)∗∗∗
log Yi +
0.29
(0.020)∗∗∗
log Yj −
N = 2985, Rˆ2 = 0.15, RM SE = 2.45
13
1.17
(0.065)∗∗∗
log Dij + εij
237
238
239
of our possible observations are nonzero.
To check the robustness of our results, we use a similar method to Koenig (2009) and Choquette
and Meinen (2011) on our Russian data. We estimate the following logistic regression:
Eij
=
4.576 + 0.078 log Mij − 1.456 log Dij − 0.108 log Yi + 0.252 log Yj + εij
(.069)∗∗∗
(.004)∗∗∗
(.007)∗∗∗
(.004)∗∗∗
(.003)∗∗∗
(2)
N = 3, 979, 007
240
where Eij = 1 if a firm in region i exports to country j and Mij is the number of firms in region i
241
exporting to country j. These findings support our results using the Moran’s I, that agglomeration
242
exists and is destination-specific.
243
Using customs data from Russia, we have shown that Russian exporters are clustered but that
244
these clusters differ depending on the country receiving the exports. Next, we develop a theory to
245
account for this fact.
246
3. A theory of exporter agglomeration based on shipment destination
247
We develop a theory of trade where exporters agglomerate with other exporters shipping to the
248
same location due to international transaction costs. Our model is a monopolistic competition trade
249
model in the spirit of Melitz (2003) and Chaney (2008) but modified to incorporate agglomeration
250
as in Ottaviano, Tabuchi, and Thisse (2002) and Fujita and Thisse (1996).
251
The export agglomeration by destination we documented could be achieved either as a reduction
252
in the fixed costs of exporting or the variable costs of exporting. We base our firm-level theory on
253
the agglomeration being achieved through the variable cost because of evidence by Choquette and
254
Meinen (2011) of this mechanism. Koenig et al. (2010) finds non robust evidence of agglomeration
255
occurring through the variable cost as well, in particular when the number of firms in the area
256
exporting to the same country is small, as it often is in Russia. We therefore choose to explore this
257
avenue, while acknowledging that it is not the only path to take.
14
3.1. Motivation and evidence
258
259
Our theory may take a few forms in the real world. The first version is a traditional treatment
260
of an externality. As in Head, Ries, and Swenson (1995), agglomeration occurs within a region
261
because of knowledge spillovers that are limited in geography. In our case, the knowledge that is
262
spilled is about exporting to a specific country (because of good contacts, translation, or effective
263
bribes paid as a percent of shipment value, for example) regardless of the exporter’s product.
264
A second story is that firms benefit from economies of scale in transportation and shipments.
265
Because of containerization in rail, ship, and air transportation, there is a fixed size to the shipment.5
266
As documented in Besedeš and Prusa (2006), most shipments are small.
267
This is true in our Russian data as well. Table 2 shows the quartiles for shipments by weight.
268
As the table shows, 50% of shipments weighed less than 185 kg (407 lbs) and 25% or less than 14
269
kg (31 lbs). Thus most shipments are small, with a few huge shipments.
Table 2: Weight of Transaction-level Shipments, by Quartile
Quartile
(%)
Weight
(kg)
25
14
50
185
75
4716
100 1,597,020,500
N
115,133
270
Because most shipments are small, exporters cannot fill a standardized container or a ship on
271
their own. Hence if firm A is exporting to China, firm B can add its goods to the shipment. Either
272
the transportation company would be willing to sell its remaining room to firm B below average
273
cost or firm A would be willing to sell its extra container room. Thus both firm A and B pay a
274
lower transportation cost than if they shipped alone. Micco and Pérez (2002) find evidence of this
275
by estimating that a doubling of trade volume weight from a particular port reduces transportation
276
costs on that route by 3–4%. They write: “In general, even though most of these economies of
5
Korinek and Sourdin (2010) show that shipping companies quote transportation rates in cost per container.
15
277
scale are at the vessel level, in practice they are related to the total volume of trade between two
278
regions.”
279
Regardless of which of these motivating tales is the most reasonable, the modeling of them is the
280
same. The particular stories above should not be taken too seriously. For example, we know that
281
some countries require that imports in containers only come from one firm. But we think these two
282
stories exemplify the spirit of why we posit an externality in transaction costs. Nonetheless, though
283
not a focus of our paper, we do an initial test on the two competing mechanisms in section 4.4.
284
3.2. The model
285
Our model is a variation of Melitz (2003) as expanded by Chaney (2008). There are N non-
286
symmetric destination countries spread spatially across the world. We do not model the character-
287
istics of countries other than that each is a passive consumer of Russian goods and each differ in size
288
and location. The representative consumer in each country j has standard taste-for-variety utility
289
on the consumption of an aggregate good composed of the varieties of Russia goods available in
290
that county. Countries differ exogenously in their income endowment Yj and their spatial location.
291
They differ endogenously in the availability of Russian goods, their aggregate price index, and their
292
aggregate consumption.
293
There is a continuum of Russian exporters distributed spatially throughout Russia. These
294
exporters differ in manufacturing productivity in addition to their location in space. Firm produc-
295
tivity, ϕ, is drawn from a Pareto distribution and cannot be changed. Each firm, identified by their
296
productivity and the region where they are located, produces a unique good.
297
There is a per-unit cost of production wi that is common to all firms in a region, but differs
298
across regions. (We believe this is plausible given the relative lack of labor mobility within Russia.)
299
Because of the rigidness of Russian law, we assume that firms are not free to change location.
300
Finally, there is a fixed cost, fij , associated with exporting from each region i in Russia to each
301
country j in the world.
302
Given their productivity, the cost of production, and the fixed cost to export, a firm in region
303
i maximizes profits in each market j it sells to by choosing the price, pij (ϕ), and quantity, xij (ϕ),
16
in that market:
304
πij (ϕ) = pij (ϕ)xij (ϕ) − wi
xij (ϕ)
τij (Aij , ·) + fij .
ϕ
(3)
305
Notice there is a region-destination variable transportation cost, τij (Aij , ·). This cost is not
306
a constant, but a function of the agglomeration term Aij , among other variables such as physical
307
distance which will be specified later and are now represented with a dot. Turning τij into a function
308
of other exporter activities based on the activities of the entire region is the primary theoretical
309
contribution of this work. For technical purposes, the form of τij (Aij ) can be very general, though
310
we think it economically reasonable to suppose τij (Aij ) > 0, τij0 (Aij ) < 0, and τij00 (Aij ) > 0. If the
311
solution to (3) is not positive, then the firm does not export to j.
312
3.3. Equilibrium
Our equilibrium concept uses Bertrand competition between the monopolistic competitors.6
Bertrand competition gives the price for each firm’s output in each destination:
pij (ϕ) =
σ wi
× τij (Aij , ·)
σ−1 ϕ
313
where σ > 1 is the elasticity of substitution in the utility function. Unlike the constant markup
314
common in Melitz-style models, the per-unit price charged by each firm in a destination also depends
315
on our agglomeration term Aij such that the greater the agglomeration, the lower the transaction
316
cost, and thus the lower the price.
317
From the zero profit condition of Bertrand competition, there exists a threshold productivity
318
for each region exporting to each destination. The equilibrium aggregate price index for country
319
j is given by the prices of firms from each region that are above the productivity threshold and
320
export to country j. Finally, equilibrium aggregate exports to country j from region i are:
Xij = Yi Yj
6
wi τij (Aij , ·)
θj
−γ
(wi fij )
γ
1− (σ−1)
λ
Results are the same under Cournot competition because there is a continuum of goods.
17
(4)
321
where θj may be interpreted as a weighted trade barrier as in Chaney (2008) or multilateral resis-
322
tance as in Anderson and van Wincoop (2003), γ > σ−1 is the parameter on the Pareto distribution
323
of firm productivities, and λ is a constant.
324
325
326
327
328
Our agglomeration effect Aij shows up in equation (4) in two places. It shows up directly in
P
1−γ/(σ−1) [w τ (A , ·)]−γ −1/γ .
τij (Aij , ·) but also indirectly in θj =
i ij
ij
i Yi (wi fij )
4. The benchmark gravity equation and our model
We derived our theoretical counterpart to gravity in equation (4). We take this prediction to
the Russian data at the region-level. To do this, we convert (4) into reduced form by taking logs,
γ
log Xij = log Yi Yj − γ log τij (Aij , ·) + 1 − γ −
σ−1
329
log wi + γ log θj + 1 −
γ
σ−1
log fij + log λ.
We rewrite to get,
log
330
Xij
= α + β1 log τij (Aij , ·) + β2 log wi + β3 log θj + εij
Yi Yj
where α = log λ, β1 = −γ, β2 = 1 − γ −
γ
σ−1 ,
β3 = γ, and εij = (1 −
(5)
γ
σ−1 ) log fij .
331
In order to estimate (5), we assume a functional form for how the agglomeration term, Aij ,
332
relates to bilateral trade costs, τij . We assume trade costs are a function of physical distance, Dij ,
333
and the agglomeration term, but nothing else. Often the gravity literature uses variables such as
334
common language, colonial history, location of overseas trade offices, and exchange rates as barriers
335
to trade. But these variables are not nearly as important in our data because there is not much
336
variation across regions within Russia. Therefore, while acknowledging some of these variables may
337
be of second-order importance, we assume:
τij (Aij , ·) = Dij × A−η
ij .
338
339
(6)
This technology’s modeling nests the gravity model by setting η = 0. Later, we test if η = 0.
We take (5) to our Russian data. We perform our experiments at the region-level rather than
18
340
the firm-level to document the increased explanatory power that the agglomeration term provides
341
for empirical trade flows. By doing so, we do not directly distinguish whether the agglomeration
342
works through the variable or fixed cost at the firm-level, since both would be realized as an increase
343
in aggregate exports. Also, we cannot directly distinguish what the correct agglomeration term is.
344
We consider two possibilities. The first is the number of exporters, similar to the agglomeration
345
measure used by Koenig (2009) and Choquette and Meinen (2011). The second agglomeration
346
measure we consider is aggregate weight of exports from each region to each country.
347
4.1. Agglomeration as measured by the number of exporters
348
We have data on Xij , Yi , Yj , Dij , and the number of exporters Mij . We do not have reliable data
349
on wi and fij , and therefore cannot calculate θj . We use a set of region-specific binary variables
350
Si to account for wi and any other unobserved unilateral region features. We use a set of country-
351
specific binary variables Tj to account for θj and any other unobserved unilateral country features.
352
We let the bilateral fixed cost be the error term.
log
Xij
Yi Yj
= − 7.133 − 1.327 log Dij + .032 log Mij +
(.910)∗∗∗
(.111)∗∗∗
(.048)
89
X
i=2
κi Si +
175
X
δj Tj + εij
(7)
j=2
N = 2985, R̂2 = 0.58, RM SE = 2.03
353
The results indicate that the number of firms in region i exporting to country j is not statistically
354
different from zero in accounting for aggregate exports. This variable brings no additional explana-
355
tory power over standard gravity, which is reflected in the estimates being nearly identical to that
356
in (1) and identical adjusted R2 .
357
4.2. Agglomeration as measured by aggregate weight
358
As an alternative to measuring agglomeration by the number of exporters, we consider another
359
way of measuring agglomeration, aggregate weight. At the industry or product level, it must be
360
the case that an increase in shipment weight will cause an increase in shipment value. However,
361
this mechanical relationship between weight and exports is less strict across industries. That is,
19
362
the value of a shipment of computers and steel is not as strongly related to the combined weight
363
as either one or the other of these products. We choose to aggregate our data and focus on all
364
manufacturing industries at once and thus we are less concerned about a purely mechanical effect
365
driving our results than otherwise.7
366
Let Wij be the aggregate weight of exports shipped from region i to country j. First consider
367
a logistic regression similar to (2) except replacing the number of exporting firms with aggregate
368
weight:
Eij
=
0.567 + 0.230 log Wij − 0.764 log Dij − 0.329 log Yi + 0.076 log Yj + εij
(.074)∗∗
(.002)∗∗∗
(.008)∗∗∗
(.004)∗∗∗
(.003)∗∗∗
(8)
N = 3, 979, 007
369
Thus the estimate on aggregate weight is statistically significant and larger than the estimate on the
370
number of exporters from (2). Because of this evidence, we continue onto our preferred specification
371
of the gravity equation, which includes the agglomeration term as measured by aggregate weight.
log
Xij
Yi Yj
= − 14.787 − .813 log Dij + .253 log Wij +
(.986)∗∗∗
N = 2985,
(.101)∗∗∗
R̂2
(.014)∗∗∗
89
X
i=2
κi Si +
175
X
δj Tj + εij
(9)
j=2
= 0.64, RM SE = 1.88
372
Compare (9) to the original gravity equation (1). In addition to our agglomeration term being
373
significant at the 99% level, the adjusted R2 improves from .58 to .64. Therefore our model with
374
an agglomeration term based on the aggregate weight of shipping costs accounts for 10.2% more
375
variation in export share than (1).
376
Because of our choice of how weight enters the transportation cost function (6), our model
377
predicts that the coefficient of physical distance Dij is equal to −γ and the coefficient on aggregate
378
export weight Wij is equal to γη. From these two predictions, we calculate η = 0.310, which is
379
economically significant. We conduct a nonlinear Wald test of whether η = 0. We reject that η = 0
7
Koenig et al. (2010) does not find evidence of exporting clustering by industry, though they do find “spillovers on the
decision to start exporting are stronger when specific, by product and destination.” But since we focus on aggregate
exports, we do not report the results when industry is included.
20
380
with 99% confidence (F (1, 2755) = 46.35), so η is statistically significant as well. This suggests
381
our choice of technology is reasonable. Note that when we use the number of exporters as our
382
agglomeration term, η = 0.024 and we cannot reject that η = 0.
383
Another endorsement of our model comes from our estimate of γ. We cannot reject that γ = 1
384
with 99% confidence which agrees with Axtell (2001) who reports that γ is near one for many
385
countries. Furthermore, notice that the estimate for γ in (1) without the agglomeration term is
386
larger than Axtell.
387
Using our estimate on γ and Chaney’s estimate for
γ
σ−1 ,
we calculate σ = 1.40. This is consistent
388
with the requirement that γ > σ − 1 and that σ > 1. At first, σ seems too low for an estimate
389
of the elasticity of substitution, since it implies very large markups in price over marginal cost.
390
Additionally our σ is low compared to the 3 to 8 range estimated by Hummels (2001), but Hummel’s
391
estimates are within a product category whereas ours is for all manufactured tradeables. Crozet
392
and Koenig (2010) estimates an average across trade-weighted industries of 2.25. Furthermore, as
393
argued by Rauch (1999), if there are informational costs to trade in the data that do not appear
394
as physical distance as in equation (6), then the estimate for σ will be low. When we use the
395
estimates from (7), σ = 0.34 indicating that goods are complements. We find this further evidence
396
that measuring agglomeration by aggregate weight is appropriate.
397
Though we derived our reduced form aggregate equation from firm-level theory, there is a
398
potential problem with the estimates in (7) due to the mechanical relationship between aggregate
399
shipment weight and export value. Though the relationship between export weight and value across
400
industries is far less strong than within industries, there still exists the possibility of an endogeneity
401
problem. To address this endogeneity concern, our approach is to follow Koenig (2009) and use
402
aggregate export weight from the previous year, 2002. Unfortunately, we only have the 2003 cross-
403
section of Russian customs data. Thus in order to proceed, we switch to U.S. state-level export
404
data for which we have a panel.
405
4.3. Application to U.S. state exports & robustness
406
Though we established the facts about agglomeration of exporters around the destination of
407
their shipments and motived our model using transaction-level customs data from Russia, there
21
408
is nothing in the reduced form equation (5) that requires customs data. Therefore, we take our
409
model to another regional export data set. The data set is the Origin of Movement export data for
410
U.S. states. We do this to check that our previous results are applicable to a wider set of countries
411
than Russia and to address the possible endogenity due to using aggregate shipment weight on the
412
right-hand side.
413
A detailed description of the OM data is in Cassey (2009). Because of his findings on data
414
quality, we limit our observations to manufactured exports only. We also restrict our sample to
415
2003 to match the Russian data. The same sample of countries is used except that Russia as a
416
destination replaces the United States. Also there are 22 observations which have positive export
417
sales but a weight of zero. We do not use these observations. Beginning with the benchmark gravity
418
equation in (1) and using OLS, we find:
log
Xij
Yi Yj
=
2.21 − 2.05 log Dij +
(.137)∗∗∗
(1.26)
N = 7061,
419
420
421
R̂2
50
X
κi Si +
i=2
175
X
δj Tj + εij
(10)
j=2
= 0.54, RM SE = 1.28
The estimated coefficient on distance is much larger than for the Russian data, and the constant
is not statistically significant at the 1% level. For us, the relevant statistic is R̂2 = .54.
Applying the U.S. data to (5) yields:
log
Xij
Yi Yj
= −10.09 − .953 log Dij + .433 log Wij +
(.950)∗∗∗
(.102)∗∗∗
N = 7061,
R̂2
(.006)∗∗∗
50
X
i=2
κi Si +
175
X
δj Tj + εij
(11)
j=2
= 0.75, RM SE = 0.93
422
The R̂2 increases from .54 in the benchmark to .75 in our model, an increase of 40.2%. (The results
423
for the data pooled over 1997 to 2008 are essentially the same.) We estimate η = 0.454∗∗∗ . As
424
with the Russian data, we can confidently reject that η = 0 (F (1, 6835) = 81.87). Furthermore, the
425
estimated coefficients in equation (11) are similar to those using the Russian data in (9) whereas
426
the same is not true for the benchmark. Another encouraging result is that our estimate for γ is
427
again not statistically different from one, as before. Furthermore, our point estimate is close to
22
428
the 0.99 to 1.10 range for U.S. data reported in Axtell (2001) whereas the benchmark is too high.
429
Finally, with the U.S. data, we estimate σ = 1.5.
430
One concern with the results so far is that they are driven by the mechanical relationship
431
between export value and weight. Because our data is across manufacturing industries, we do
432
not think this relationship is strong. Nevertheless, we repeat the estimation using aggregate weight
433
lagged one year to 2002. This severs any mechanical relationship since past aggregate weight cannot
434
affect current export value in any way other than a reduction of the transaction costs of trade. We
435
lose some observations compared to 2003 because those state–country pairs did not trade in 2002.
log
Xij
Yi Yj
= − 4.11
(1.177)∗∗∗
− 1.50 ∗ log Dij + .243 log lag Wij +
(.124)
(.010)∗∗∗
50
X
i=2
κi Si +
175
X
δj Tj + εij (12)
j=2
N = 6617, R̂2 = 0.60, RM SE = 1.15
436
In this case, we find that our model accounts for 12% more variation in the data than the bench-
437
mark. The estimate on the aggregate weight term is statistically significant, though smaller than
438
previously. Therefore, we find that the agglomeration term is important even if it is lagged one
439
year.
440
Applying the OM state export data to our model strengthens the results achieved with the
441
Russian data since the U.S. data was not used in the spatial econometrics that guided the theory.
442
4.4. Explorations on clustering mechanism
443
Now that we have shown that agglomeration as measured by aggregate weight is a statistically
444
and economically significant variable in the gravity equation of trade, we do exploratory analysis to
445
determine which mechanism the agglomeration term works through. Our model cannot distinguish
446
agglomeration occurring because of an informational spillover about how to export to a specific
447
destination or economies of scale in shipping. However, we expect that savings from consolidat-
448
ing shipments to the same country would be more prevalent for small firms than big firms, who
449
presumably have their own distribution networks and can fill a container by themselves.
450
We re-estimate (9) for separate subsamples based on firm size. We split firms into size based
23
451
on total exports to the world.
Small firms: log
Xij
Yi Yj
= − 22.03 − 0.55 log Dij + .325 log Wij +
(1.737)∗∗∗
N = 851,
Xij
Medium firms: log
Yi Yj
(.150)∗∗∗
R̂2
(.036)∗∗∗
κi Si +
175
X
i=2
j=2
89
X
175
X
δj Tj + εij
= 0.42, RM SE = 2.43
= − 21.40 − .744 log Dij + .400 log Wij +
(1.628)∗∗∗
89
X
(.120)∗∗∗
(.030)∗∗∗
κi Si +
i=2
δj Tj + εij
j=2
N = 1075, R̂2 = 0.45, RM SE = 2.40
Xij
large firms: log
Yi Yj
= − 25.10 − .717 log Dij + .601 log Wij +
(.777)∗∗∗
(.075)∗∗∗
(.019)∗∗∗
89
X
i=2
κi Si +
175
X
δj Tj + εij
j=2
N = 2624, R̂2 = 0.53, RM SE = 2.54
452
These estimates show that the effect of the agglomeration term as measured by aggregate weight
453
is greater for large firms than medium or small firms, though the coefficient is still statistically
454
and economically significant for all sizes. This evidence suggests that destination specific export
455
agglomeration is working though an informational spillover rather than consolidation of shipments.
456
5. Conclusion
457
The theoretical and empirical international trade literature typically studies flows at the country-
458
level. However more recent papers such as Eaton et al. (Forthcoming); Bernard and Jensen (2004);
459
Tybout (2003); and Arkolakis and Muendler (2010) study firm-level export decisions. All of these
460
articles study the production and technology characteristics of the firm and their export decision—
461
what to export, how much to export, and to which countries. The findings from this work indicate
462
there are barriers to exporting whose precise characterization continues to elude researchers.
463
The work in Koenig (2009), Koenig et al. (2010), and Choquette and Meinen (2011) shows
464
export spillovers are important for firm export decisions in developed countries if those firms are
465
exporting to the same destination. We use the physical location of exporting establishments in
466
Russia to see if there is exporter agglomeration by destination in a developing country. Using the
467
Moran’s I statistic from spatial econometrics, we find there is agglomeration of exporters, not only
24
468
in general and around ports, but also in terms of the destination to which exports are shipped.
469
This evidence is supported by a logistic regression.
470
We believe this clustering of exporters by destination indicates firm behavior confronting economies
471
of scale in the transaction cost of trade to a specific destination. We posit that this agglomeration
472
of exporters by destination is an externality from how-to-export information spillovers or economies
473
of scale in shipping costs. We measure this agglomeration two ways: by the number of firms ex-
474
porting and by the aggregate weight of exported shipments. We find that the agglomeration term
475
as measured by aggregate weight increases the explanatory power of the gravity model of trade.
476
Using aggregate weight, our model’s prediction is verified with the Russian data and outside
477
sources, and our model accounts for more of the variation in export share than the benchmark
478
Anderson and van Wincoop (2003) or Chaney (2008) gravity-type model. We then test our model
479
on U.S. state export data and are able to account for 40% more variation in export share than the
480
benchmark.
481
We perform exploratory tests to find evidence through which mechanism the externality op-
482
erates. We find that the agglomeration term is more important for large firms than small firms,
483
indicating that destination-specific knowledge on how to export is important.
484
The documenting of agglomeration by destination and the confirmation of our model takes a step
485
toward understanding the size and nature of the international barriers to trade. Our combination
486
of localization techniques and externality modeling from industrial organization with the tools from
487
firm-level international trade models yields new empirical and theoretical insights into the nature
488
and size of barriers to international trade, and the behavior of firms to overcome these barriers.
489
The implications for trade-focused development is that policy perhaps should consider where extent
490
exports are operating when promoting exports rather than just particular industries or countries.
491
Acknowledgements
492
The authors thank Thomas Holmes, Yelena Tuzova, Anton Cheremukhin, and seminar participants
493
at Vasser College, Washington State University, Kansas State University, University of Scranton,
494
University of Richmond, and the New York Economics Association annual meeting. Cassey thanks
25
495
Qianqian Wang and Pavan Dhanireddy for research assistance, and Jeremy Sage for help with
496
ArcGIS. Portions of this research supported by the Agricultural Research Center Project #0540 at
497
Washington State University.
498
Appendix A. The political organization of Russia
499
In 2003, Russia had 89 federal regions. These regions were separated into 49 oblasts (provinces),
500
21 republics, 6 krais (territories), 10 okrugs (autonomous districts), 1 autonomous oblast, and 2
501
federal cities (Moscow and St. Petersberg). The divisions are based on ethnic groups and history.
502
Each region has equal representation in the Federal Assembly, though there are differences in
503
autonomy. However these differences are minor (and decreasing over time) because each region
504
is attached to one of eight federal districts administered by an envoy of Russia’s President. The
505
districts are designed to oversee regional compliance with federal laws.
506
Most taxes are set federally. In 2003, regions could set corporate property tax, gambling business
507
tax, and transport tax within bounds established federally (CTEC 2006, p. 955). Exports of goods
508
from Russia are subject to a value-added tax of 0% throughout Russia (p. 960), except on some
509
exports of oil and natural gas. Export customs duty rates are set federally as well (p. 961).
510
Appendix B. The location of products and ports
511
512
We show that the location of Russian exporters can neither be accounted for by the location of
export commodities nor the location of borders and ports.
513
Russia’s major manufacturing industries are: all forms of machine building from rolling mills
514
to high-performance aircraft and space vehicles; defense industries including radar, missile produc-
515
tion, and advanced electronic components, shipbuilding; road and rail transportation equipment;
516
communications equipment; agricultural machinery, tractors, and construction equipment; elec-
517
tric power generating and transmitting equipment; medical and scientific instruments; consumer
518
durables, textiles, foodstuffs, and handicrafts.
519
Russia’s largest export commodities are: petroleum products, wood and wood products, metals,
520
chemicals, and a wide variety of civilian and military manufactures. Products such as petroleum,
26
Marmara Deniz
Kolpos Kassandras
Arabian Sea
Gulf of Oman
27
8
_
^
16
52
_
^
77
7
Republic (21)
Region (48)
Territory (7)
City (1)
Autonomous Province (10)
48
70
40
35
58
5
12
79
33
28
Autonomous Region (1)
Region Type
83
56
62 82
50
46
15 45
57
80
72
66
65
21 37
84
42
87
68
78
34
54
32
53
49
3
18
74
86
29
31
76
39
17
73
81
20
10
2
13
63
4
88
30
59
44
36
Philippine Sea
65
64
14
Figure A.1: Russia’s Regions in 2003. The City of Moscow Region is included in the map only as the star and not listed separately.
Rail
Capital
43
85 71
27 69
24
22
6
51 11
1
38 61
25
67
60
41 55 75 47
15 9
23
26
Ü
521
wood, metals, and chemicals need to be processed to become manufactured goods. Therefore it
522
is possible that the location of exports is the same as the location of the refineries and processing
523
plants for these goods.
524
Figure 1 shows that though many exporters are located in the same region as a major port, there
525
are regions that are not close to a major port in the top quintile of the number of exporting firms.
526
The largest ports are Azov (Rostov, Black Sea), Kaliningrad (Kaliningrad, Baltic Sea), Nakhodka
527
and Vostochny (Primorsky, Sea of Japan, largest), Novorossiysk (Krasnordar, Black Sea, largest),
528
Primorsk (Leningrad, Baltic Sea, largest, lots of oil), and Saint Petersburg (Saint Petersburg, Baltic
529
Sea). Cassey (2010) documents that access to water is an important predictor of U.S. state exports.
530
Appendix C. Exports in the context of output
531
If external geography played no role in the location of exporters, the pattern of region-country
532
aggregate exports (Xij ) would be identical to the pattern of regional production (Yi ) with some
533
randomness. In this case, the destination of Russian sales would not matter for the location of
534
Russian exporters. To show that this does not hold, we conduct two simple tests. The first test is
535
to run the following regression:
log Xij
= − 5.40 +
(0.386)
0.44
(0.046)∗∗∗
log Yi + εij
(C.1)
N = 2985, R2 = 0.03, RM SE = 2.62
536
The index i runs over the set of Russia’s regions and the index j runs over the set of countries
∗∗∗
537
in the world. Standard errors are robust and
indicates the estimated coefficient is significantly
538
different from zero with 99% confidence. The extremely low explanatory power as given by the R2
539
indicates that the pattern of production in Russia does not account for the pattern of exports in
540
the data.
The second test is to compare the ratio of exports from each Russian region to the ratio of GDP
from each region to Russia. We modify the familiar concept of location quotient to put Russian
28
exports by region into the context of overall economic activity,
Xi
LQi =
X
Yi
.
Y
541
If the LQ is greater than 1 then the region’s export share to the world is larger than it’s share
542
of GDP. This indicates the region is relatively specialized in exporting in comparison to overall
543
economic activity. The strength of this approach is that it controls for the geography of economic
544
activity within Russia,
545
The average LQi is 1.454 with standard deviation 0.489. But though the average LQi is not
546
significantly different from one, the regions with the largest LQi are several times the standard
547
deviation. Figure C.1 shows the histogram of LQi . Though most region’s LQi is around 1, there
548
are many that are in the right tail of figure C.1. Thus there are regions in Russia that are heavily
549
concentrated in exporting compared to overall output and regions that do not export anything at
550
all.
551
Because we only consider manufacturing activity, we repeat the proceeding exercise by scaling
552
the regional GDP data by the share that is from industry. The results (not reported) are not
553
qualitatively different from before. Therefore, we have documented that Russian exporters are not
554
simply located in the geographic pattern of industrial economic activity.
555
Appendix D. Robustness
556
One feature of the Russian export data is the high frequency of zeros. There are 89 regions in
557
Russia and we have 2003 GDP data (IMF 2006) and distance for 175 countries. Therefore, there
558
are 15,575 possible observations. Of these, only 2991 have positive exports, or 19%. (We also lose
559
some observations because of missing regional GDP data from Russia.) Santos Silva and Tenreyro
560
(2006) show that (2) with OLS can bias estimates when there are many zeros in the data. We
29
40
30
Percent
20
10
0
0
10
20
LQ_exports_GDP
Figure C.1: LQ Histogram
30
30
40
561
follow them and use the poisson quasi-maximum likelihood estimator.
Xij
Yi Yj
Xij
Yi Yj
−1.04 ∗
= exp
∗
(0.150)
−20.06 × log Dij
+
(1.189)
89
X
i=2
κi Si +
175
X
δj Tj
× εij .
j=2
N = 15400, AIC = 526.00
89
175
−.804 ∗
.116 ∗
X
X
(0.119)
(0.027)
∗
= exp −12.695 × log Dij
× log Wij
+
κi Si +
δj Tj × εij .
(1.231)
N = 2985,
i=2
j=2
AIC = 456.00
562
Though the estimates for the regression with the agglomeration term change quantitatively, the
563
are the same qualitatively. The agglomeration term is economically and statistically significant.
564
Importantly, the model with the agglomeration term has a lower Akaike Information Criterion
565
than the model without the term, indicating the model with agglomeration is preferred. However,
566
region-country observations of zero exports also have zero weight for those exports. Santos Silva
567
and Tenreyro do not consider the model where zeros occur on both the left and right hand side of
568
the regression.
569
References
570
, 2004, 2006. Russia: All Regions Trade & Investment Guide. CTEC Publishing LLC, Moscow.
571
Anderson, J. E., 1979. A theoretical foundation for the gravity equation. American Economic
572
573
574
575
576
Review 69 (1), 106–116.
Anderson, J. E., van Wincoop, E., 2003. Gravity with gravitas: A solution to the border puzzle.
American Economic Review 93 (1), 170–192.
Arkolakis, C., Muendler, M.-A., 2010. The extensive margin of exporting goods: A firm level
analysis, NBER working paper no. 16641.
577
Axtell, R. L., 2001. Zipf distribution of U.S. firm sizes. Science 293 (5536), 1818–1820.
578
Bernard, A. B., Jensen, J. B., 2004. Why some firms export. Review of Economics and Statistics
579
86 (2), 561–569.
31
580
581
582
583
584
585
586
587
588
589
590
591
592
Besedeš, T., Prusa, T. J., 2006. Product differentiation and duration of US import trade. Journal
of International Economics 70 (2), 339–358.
Botolf, P., 2003. Divergence and dispersion in the Russian economy. Europe Asia Studies 55 (8),
1165–1185.
Bradshaw, M., 2008. The geography of Russia’s new political economy. New Political Economy
13 (2), 193–201.
Broadman, H. G., Recanatini, F., 2001. Where has all the foreign investment gone in Russia?, world
Bank Policy Research working paper no. 2640.
Cassey, A., Schmeiser, K. N. L., 2011. Multilateral export decompositions, Washington State University and Mt. Holyoke College.
Cassey, A. J., 2009. State export data: Origin of movement vs. origin of production. Journal of
Economic and Social Measurement 34 (4), 241–268.
Cassey, A. J., 2010. State export behavior and barriers, School of Economic Sciences, Washington
593
State University working paper no. 2010-14.
594
URL
595
behavior_july2007.pdf
596
597
598
599
600
http://www.econ.umn.edu/~cassey/Appendix/State_export_behavior/state_
Cassey, A. J., Forthcoming. An application of the Ricardian trade model with trade costs. Applied
Economics Letters.
Chaney, T., 2008. Distorted gravity: The intensive and extensive margins of international trade.
American Economic Review 98 (4), 1707–1721.
Choquette, E., Meinen, P., 2011. Export spillovers and the extensive and intensive margins of
601
trade, unpublished.
602
URL
603
Submissions2011/S-F-22.pdf
http://www.etnpconferences.net/sea/sea2011/PaperSubmissions/
32
604
605
Crozet, M., Koenig, P., 2010. Structural gravity equatons with instensive and extensive margins.
Canadian Journal of Economics 43 (1), 41–62.
606
Eaton, J., Eslava, M., Kugler, M., Tybout, J. E., 2008. Export growth in Colombia: Firm-level
607
evidence. In: Helpman, E., Marin, D., Verdier, T. (Eds.), The Organization of Firms in a Global
608
Economy. Harvard University Press, Camdridge, pp. 231–272.
609
610
Eaton, J., Kortum, S., Kramarz, F., Forthcoming. An anatomy of international trade: Evidence
from French firms. Econometrica.
611
Feenstra, R. C., 2004. Advanced International Trade. Princeton University Press, Princeton.
612
Fujita, M., Thisse, J.-F., 1996. Economics of agglomeration. Journal of the Japanese and Interna-
613
614
615
tional Economies 10 (4), 339–378.
Glejser, H., Jacquemin, A., Petit, J., 1980. Exports in an imperfect competition framework: An
analysis of 1,446 exporters. Quarterly Journal of Economics 94 (3), 507–524.
616
Head, K., Ries, J., Swenson, D., 1995. Agglomeration benefits and location choice: Evidence from
617
Japanese manufacturing investments in the United States. Journal of International Economics
618
38 (3–4), 223–247.
619
620
621
Huber, P., Nagaev, S., Wörgötter, A., 1997. The changes in location of Russia industry in early
transition 1987–1993, Institute for Advanced Studies, Vienna, East European Series no. 42.
Hummels, D., 2001. Toward a geography of trade costs, Center for Global Trade Analysis, Purdue
622
University GTAP working paper no. 1162.
623
URL http://www.mgmt.purdue.edu/faculty/hummelsd/research/toward/TGTC.pdf
624
IMF, 2006. World Economic Outlook Database. International Monetary Fund, Washington, DC,
625
http://www.imf.org/external/pubs/ft/weo/2006/01/data/index.htm (accessed December
626
15, 2006).
627
628
Koenig, P., 2009. Agglomeration and the export decisions of French firms. Journal of Urban Economics 66 (3), 186–195.
33
629
630
631
632
633
634
635
636
637
638
639
Koenig, P., Mayneris, F., Poncet, S., 2010. Local export spillovers in France. European Economic
Review 54 (4), 622–641.
Korinek, J., Sourdin, P., 2010. Martime transport costs and their impact on trade. Applied Economic Perspectives and Policy 32 (3), 417–435.
Lawless, M., 2009. Firm export dynamics and the geography of trade. Journal of International
Economics 77, 245–254.
Lovely, M. E., Rosenthal, S. S., Sharma, S., 2005. Information, agglomeration, and the headquarters
of U.S. exporters. Regional Science and Urban Economics 35 (2), 167–191.
Melitz, M. J., 2003. The impact of trade on intra-industry reallocations and aggregate industry
productivity. Econometrica 71 (6), 1695–1725.
Micco, A., Pérez, N., 2002. Determinants of maritime transport costs, Inter-American Development
640
Bank working paper no. 441.
641
URL http://www.iadb.org/res/publications/pubfiles/pubWP-441.pdf
642
Nuadé, W., Matthee, M., 2007. The geographical location of manufacturing exporters in South
643
Africa, United Nations University–World Institute for Development Economics Research working
644
paper no. 2007/09.
645
646
647
648
649
650
651
652
Ottaviano, G., Tabuchi, T., Thisse, J.-F., 2002. Agglomeration and trade revisited. International
Economic Review 43 (2), 409–436.
Rauch, J., 1999. Networks versus markets in international trade. Journal of International Economics
48 (1), 7–35.
Santos Silva, J. M. C., Tenreyro, S., 2006. The log of gravity. Review of Economics and Statistics
88 (4), 641–658.
Tybout, J. R., 2003. Plant- and firm-level evidence on “new” trade theories. In: Choi, E. K.,
Harrigan, J. (Eds.), Handbook of International Trade. Basil-Blackwell.
34
653
WISER, Various years. Origin of Movement State Export Data. World Institute for Strategic Eco-
654
nomic Research, Holyoke, MA, http://www.wisertrade.org (initially accessed Nov. 15, 2005).
35
Table A.1: Geographic Regions
Region
Type
Adygeya
Agin-Buryat
Altai
Altai
Amur
Republic
Autonomous Okrug
Krai
Republic
Oblast
Arkhangelsk
Astrakhan
Bashkortostan
Belgorod
Bryansk
Oblast
Oblast
Republic
Oblast
Oblast
Buryat
Chechen
Chelyabinsk
Chita
Chukotsky
GDP
(millions)
Region
Type
Mordovia
Moscow
Moscow
Murmansk
Nenets
Republic
Federal City
Oblast
Oblast
Autonomous Okrug
3735.1
1884
54851.7
2766.4
1686.2
Nizhny Novgorod
North Ossetia-Alania
Novgorod
Novosibirsk
Omsk
Oblast
Republic
Oblast
Oblast
Oblast
7720.1
726.8
1356.1
5797.2
4363
Republic
Republic
Oblast
Oblast
Autonomous Okrug
1685.9
346.4
7995.8
1859.4
11325.8
Orenburg
Oryol
Penza
Perm
Permyakia
Oblast
Oblast
Oblast
Oblast
Autonomous Okrug
4345.8
1594.1
1706.1
8058.3
4305.3
Chuvash
Dagestan
Evenkiysky
Ingushetia
Irkutsk
Republic
Republic
Autonomous Okrug
Republic
Oblast
1741.8
2326.3
141.8
167.4
5999
Primorsky
Pskov
Rostov
Ryazan
Saint Petersburg
Krai
Oblast
Oblast
Oblast
Federal City
Ivanovo
Jewish
Kabardino-Balkar
Kaliningrad
Kalmykia
Oblast
Autonomous Oblast
Republic
Oblast
Republic
1250.9
298.3
938.3
1783.5
4770.8
Sakha
Sakhalin
Samara
Saratov
Smolensk
Republic
Oblast
Oblast
Oblast
Oblast
Kaluga
Kamchatka
Karachayevo-Cherkessia
Karelia
Kemerovo
Oblast
Krai
Republic
Republic
Oblast
1852.9
1016.8
412.5
1668.1
5948.7
Stavropol
Sverdlovsk
Taimyrsky (Dolgano-Nenetsky)
Tambov
Tatarstan
Krai
Oblast
Autonomous Okrug
Oblast
Republic
Khabarovsk
Khakassia
Khanty-Mansi
Kirov
Komi
Krai
Republic
Autonomous Okrug
Oblast
Republic
4253.7
997.4
26409.8
2165.5
3941.4
Tomsk
Tula
Tuva
Tver
Tyumen
Oblast
Oblast
Republic
Oblast
Oblast
3600.3
2683.8
166.6
2573
2145.2
Koryaksky
Kostroma
Krasnodar
Krasnoyarsk
Kurgan
Autonomous Okrug
Oblast
Krai
Krai
Oblast
150.6
1144.2
9573.7
9548.8
1386.9
Udmurtia
Ulyanovsk
Ust-Ordynsky Buryatsky
Vladimir
Volgograd
Republic
Oblast
Autonomous Okrug
Oblast
Oblast
3393.1
2025
145.4
2326.3
4770.8
Kursk
Leningrad
Lipetsk
Magadan
Mari El
Oblast
Oblast
Oblast
Oblast
Republic
2066.6
4595.1
3406.5
797.6
853.2
Vologda
Voronezh
Yamalo-Nenets
Yaroslavl
Oblast
Oblast
Autonomous Okrug
Oblast
3962.7
3646.4
11325.8
3626.9
352.8
77.8
3130.4
269.6
1901.5
GDP
(millions)
1281.7
84742.3
15517.4
2834.3
876.1
1052.1
9707.2
6366.7
2301.8
15122.6
4526.4
–
9541
4557.5
1794.7
3823.1
11133.6
36269
1765.6
11075
Notes: In 2003, Russia had 49 oblasts (provinces), 21 republics, 6 krais (territories), 10 okrugs (autonomous districts), 1
autonomous oblast, and 2 federal cities (Moscow and St. Petersberg).
36