trade blocs, currency blocs and the reorientation of …dirwin/docs/blocs.pdf2 b. eichengreen, d.a....

24
Jourdd INTEiBtAMMlAL ECOWOMICS ELSEVIER Journal of International Economics 38 (1995) l-24 Trade blocs, currency blocs and the reorientation of world trade in the 1930s Barry Eichengreen”, Douglas A. Irwinb’* aDepartment of Economics, University of California, Berkeley, CA 94720, USA bGraduate School of Business, University of Chicago, Chicago, IL 60637, USA Received July 1993, revised version received July 1994 Abstract We analyze the impact of commercial and financial policieson the reorientation of trade in the 1930s. We report evidence that commercial policies attenuated prior connections between income growth and trade, and that exchange rate instability marginally discouragedinternational trade. In contrast, the tendency toward re- gionalization commonly ascribedto the formation of trade and currency blocs was already evident to a considerable extent prior to the regional policy initiatives of the 1930s. This reorientation is more properly attributed to ongoinghistoricalforces such as commercial and financial links between countries forged over many years. Key words: Trade blocs; Currency blocs; Regional Trade Arrangements JEL classification: F13; F33; N70 ‘We may characterize the change that occurred as a disintegration of world trade: while previously international settlement took place within a world-wide network of multilateral transactions, there was in the ‘thirties a tendency to achieve settlement either in bilateral exchange between two countries, or within the limited range of countries attached to each other by political or other ties (Folke Hilgerdt, 1942, pp. 90-91).’ * Corresponding author. 0022-1996/95/$09.50 0 1995 Elsevier Science B.V. All rights reserved SSDI 0022-1996(94)001350-O

Upload: others

Post on 27-Jun-2020

27 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

Jourdd INTEiBtAMMlAL ECOWOMICS

ELSEVIER Journal of International Economics 38 (1995) l-24

Trade blocs, currency blocs and the reorientation of world trade in the 1930s

Barry Eichengreen”, Douglas A. Irwinb’* aDepartment of Economics, University of California, Berkeley, CA 94720, USA bGraduate School of Business, University of Chicago, Chicago, IL 60637, USA

Received July 1993, revised version received July 1994

Abstract

We analyze the impact of commercial and financial policies on the reorientation of trade in the 1930s. We report evidence that commercial policies attenuated prior connections between income growth and trade, and that exchange rate instability marginally discouraged international trade. In contrast, the tendency toward re- gionalization commonly ascribed to the formation of trade and currency blocs was already evident to a considerable extent prior to the regional policy initiatives of the 1930s. This reorientation is more properly attributed to ongoing historical forces such as commercial and financial links between countries forged over many years.

Key words: Trade blocs; Currency blocs; Regional Trade Arrangements

JEL classification: F13; F33; N70

‘We may characterize the change that occurred as a disintegration of world trade: while previously international settlement took place within a world-wide network of multilateral transactions, there was in the ‘thirties a tendency to achieve settlement either in bilateral exchange between two countries, or within the limited range of countries attached to each other by political or other ties (Folke Hilgerdt, 1942, pp. 90-91).’

* Corresponding author.

0022-1996/95/$09.50 0 1995 Elsevier Science B.V. All rights reserved SSDI 0022-1996(94)001350-O

Page 2: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24

1. Introduction

The early 1930s witnessed an astonishing implosion of world trade. Between 1929 and 1932 its value in current U.S. dollars fell by 50 percent. Though deflation contributed to the collapse, even at constant prices the volume of trade in 1932 was nearly 30 percent below 1929 levels. As late as 1938, trade volume was still barely 90 percent of 1929 despite the complete recovery of global production of primary products and manufactured goods.]

Superimposed on this decline in volume was a distinct shift in direction. The traditional pattern of multilateral settlements was supplanted by a set of increasingly compartmentalized trade flows, Trade was channelled into self-contained regional, colonial and commercial blocs such as the British Commonwealth, a group of European gold standard countries centered on France, and a Central European trade bloc linked to Germany.

The causes of this unprecedented transformation remain incompletely understood. Most attempts at explanation start with two factors. The first is commercial policies. The post-1929 slump in output and employment intensified protectionism world wide. Dozens of countries raised their tariffs after 1929, the Smoot-Hawley Tariff of 1930 in the United States being only the most prominent of the newly imposed restrictions on commerce. Discriminatory measures were also used to favor trade with certain coun- tries, as tariffs were supplemented by import quotas and licenses to favor particular sources of imports. The Ottawa Agreements of 1932 strengthened tariff preferences within the British Commonwealth. Germany used dis- criminatory trade policies to create a sphere of economic influence. The Reciprocal Trade Agreements Act of 1934 in the United States prompted the negotiation of non-MFN bilateral tariff agreements with selected countries. Thus, commercial policies and trade discrimination may explain not just the contraction of trade but also its compartmentalization.

A second factor affecting trade in the 1930s was international exchange rate policies. The international gold standard splintered into currency blocs: a residual gold bloc, the sterling area, a group of inconvertible currencies tied to the German Reichsmark, and currencies mainly in the Western Hemisphere linked to the dollar. In each bloc, members pegged to an anchor currency and held reserves in the corresponding financial center. Exchange rate stability within the bloc could have encouraged intra-bloc trade while inter-bloc exchange rate variability discouraged trade with the

’ World primary production and industrial production were 7 percent and 11 percent above 1929 levels, respectively. Lamartine Yates (1959, ch. 2) analyzes the implications of different deflators for intertemporal comparisons of trade volumes. Statistics in this paragraph are computed from League of Nations (1939) as are other figures in the remainder of this introduction for which no references are provided.

Page 3: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

B. Eichengreen, D.A. Irwin I J. Int. Econ. 38 (1995) l-24 3

rest of the world. Foreign currency reserves that allowed countries to finance payments imbalances and collateralize borrowing from foreign banks could have further stimulated intra-bloc trade, insofar as these resources could be used more readily to finance trade within the bloc than trade outside it. And countries holding balances in a reserve center could have received favorable market access as a quid pro quo from the reserve- currency country. Whether participation in a currency bloc stimulated intra- bloc trade or discouraged trade with the rest of the world also depended on whether countries used exchange controls and clearing arrangements to stabilize intra-bloc rates (as in the case of the Reichsmark bloc) or supported their currencies through the imposition of ‘exchange-dumping’ duties against nonmembers (as in the case of the Continental gold bloc).*

Because the existing literature only speculates about how trade and currency blocs contributed to the reorientation of trade, we seek in this paper to bring empirical evidence to the analysis of their effects on trade. We estimate gravity models of the pattern of interwar trade, relating the value of bilateral trade flows on indicators of trade and currency blocs along with the standard arguments of the gravity model. We find evidence of significant impacts of both trade and currency policies on the pattern of trade. The standard determinants of bilateral trade flows-incomes, prox- imity and contiguity-had a diminished influence in the 1930s as if commercial policies increasingly overrode their effects. In addition, we find a consistent negative impact of exchange rate instability on the volume of trade.

In comparison with these effects of external trade barriers and exchange rate instability, evidence of regionalization due to the formation of trade and currency blocs is less compelling. Analysis for 1935 and 1938 suggests that members of the British Commonwealth and the Central European trade bloc traded more with one another-and, depending on which estimates are preferred, less with the rest of the world-than would be predicted by their economic characteristics and the average behavior of countries in the sample. But these tendencies are already evident in 1928 before the Commonwealth countries signed the Ottawa Agreements and the German bloc was formed. This suggests that commercial and financial linkages that had developed over previous years contributed importantly to the preval- ence of intra-bloc trade in the 1930s. Still, there is some evidence that the formation of formal trade blocs diverted the members’ trade away from the rest of the world and toward fellow bloc members.

We find that the different currency blocs of the 1930s had different

’ On exchange control and clearing arrangements, see Ellis (1941) and Neal (1979), respectively. Eichengreen (1992) discusses the exchange dumping duties of the gold bloc countries.

Page 4: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

4 B. Eichengreen, D.A. Irwin I J. Znt. Econ. 38 (1995) l-24

implications for trade. Members of the sterling area traded unusually extensively among themselves, due to the trade-promoting effects of intra- bloc exchange rate stability, but also traded unusually extensively with the rest of the world, since the export-promoting effects of their policies of currency depreciation swamped the trade-dampening effects of inter-bloc exchange rate instability. Members of the gold bloc, in contrast, did not trade disproportionately with one another, as if their increasing use of tariffs and quotas neutralized the impact on trade of intra-bloc exchange rate stability, while exchange control countries traded less with one another than their other economic characteristics would predict.

Section 2 of the paper reviews the changes in trading patterns we seek to explain and the associated policy initiatives. Sections 3 and 4 describe our data, methodology and results. Section 5 summarizes the implications.

2. Policy and the pattern of trade

Fig. 1 is one of the most widely reprinted diagrams in all of economics. The ‘Contracting Spiral of World Trade’, first published by the Gsterreichis- then Institutes fiir Konjunkturforschung in 1933 and quickly reproduced by the League of Nations in its Economic Survey for 1932-33, shows the dollar value of world trade spiralling downward from 1929 through 1933. The contraction of trade far exceeded the contemporaneous decline in pro- duction. Real GNP in major industrialized economies fell by 18 percent over this period in which export volume declined 35 percent and import volume dropped 24 percent.3 The contrast grew more pronounced with recovery from the Depression. From 1932 to 1938, GNP rose 29 percent in the industrialized countries, while export volume increased 24 percent and import volume rose a mere 14 percent.

Superimposed on the downward spiral and anemic recovery of trade was a tendency toward regionalization. Between 1928 and 1938 the share of British exports destined for the Commonwealth rose from 44 percent to 50 percent, while the share of British imports from those countries jumped from 30 percent to 42 percent. France’s trade with its colonies and protectorates rose from 12 percent to 27 percent of its imports and from 19 percent to 28 percent of its exports. Germany’s trade with six Southeastern

3 The difference on the import and export sides reflects the improvement in the terms of trade of the industrial countries. Figures in this paragraph are taken from Maddison (1989b, p. 57). Calculations by the Economic Intelligence Service of the League of Nations indicate that the volume of trade fell by 27 percent between 1929 and 1932, a period over which industrial production fell by 30 percent, primary production by a relatively modest 8 percent. This estimate of the change in trade volumes is from League of Nations (1933, p. 212).

Page 5: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

B. Eichengreen, D.A. Irwin I J. Int. Econ. 38 (1995) t-24

OCTOBER

Fig. 1. The contracting spiral of world trade, month by month, January 1929-June 1933 (in millions of U.S. (gold) dollars).

European states (Bulgaria, Greece, Hungary, Romania, Turkey, and Yugoslavia) and Latin America increased from 17 percent to 28 percent of its imports, and from 13 percent to 25 percent of its exports4 Japan’s trade with Korea, Formosa, and Manchuria rose from 14 percent to 39 percent of imports and from 18 percent to 41 percent of exports.

4 The figures in this paragraph are from League of Nations (1939, pp. 34-35). German imports from Balkan and Mediterranean countries alone rose from 11 percent to 20 percent of total German imports between 1929 and 1938. Basch (1941, p. 38).

Page 6: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

6 B. Eichengreen, D.A. Irwin I J. Int. Econ. 38 (1995) l-24

Commercial policies and exchange rate arrangements were the obvious sources of these adaptations. The Ottawa Agreements negotiated between Britain and the Dominions in 1932 aimed to stimulate Commonwealth trade by exchanging tariff preferences. Britain’s goal at Ottawa was to secure market access for her exports. In return for preferential treatment by the Dominions and India, Britain agreed to admit many of their exports duty-free. Gordon (1941, p. 221) reports calculations suggesting that the share of British imports from foreign countries entering duty-free declined from 30 percent prior to Ottawa to 25 percent immediately thereafter. McDougall and Hutt (1954) estimate that the average rate of tariff prefer- ence on Commonwealth imports from the United Kingdom rose from 6 percent in 1929 to lo-11 percent in 1937, and on U.K. imports from the Commonwealth from 2-3 percent to lo-12 percent. Yet authors such as Schlote (1952) and Thorbecke (1960) are skeptical that imperial preference was primarily responsible for the growth of intra-Commonwealth trade; they suggest that the trend was evident earlier, reflecting growing complemen- tarities between the British economy and the Empire as well as diplomatic and political interdependencies developed over the years.

The pattern of Continental European regionalism was more complex. One bloc was comprised of countries of Central and South-Eastern Europe with trade ties to Germany. Austria, Hungary, Poland and Yugoslavia were tied to Germany and to one another by the so-called Schacht Agreements. Germany exchanged access to its market in return for these countries sending it an increasing share of their exports of foodstuffs and raw materials. Germany similarly used its market power and political leverage to induce Greece and Turkey to shift a growing share of their exports toward its bloc.5

Trade among Scandinavia and the Benelux countries increased as they sought to reduce their dependence on Germany. The 01~0 Convention, concluded in December 1930, liberalized these countries’ trade with one another. But the members soon found themselves split between two currency area-the gold bloc and the sterling area-which undermined the effectiveness of their agreement (Condliffe, 1940, p. 307). In June 1932, Holland, Belgium and Luxembourg adopted the Ouchy Convention, lower- ing tariffs and quotas on one another’s exports. This agreement was abandoned, however, in the course of the World Economic Conference of 1933. Italy negotiated bilateral preference agreements with Austria and Hungary (the so-called Brocchi Agreements of 1931-32 and Rome Agree- ments of 1934). Whether U.S. policies led to growing regionalism is difficult to say.

5 How much market power Germany possessed is disputed (see Milward, 1981; Kitson, 1992).

Page 7: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

B. Eichengreen, D.A. Irwin I J. Int. Econ. 38 (199.5) l-24 7

Although the Smoot-Hawley tariff was largely nondiscriminatory, it was rolled back not by a general reduction in import duties but by bilateral agreements reached under the provisions of the Reciprocal Trade Agree- ments Act of 1934. During the following three years the United States concluded trade liberalization agreements with 17 countries, most of which were in the Western HemisphereP In the next three years, six agreements were concluded, half of which involved countries in the Western Hemisphere.7 Although it is tempting to characterize this as leading to the emergence of a U.S.-centered trade bloc, Thorbecke (1960) shows that the share of the trade of Western Hemisphere countries remaining within the region rose significantly only after World War II.

Changes in international monetary and exchange rate arrangements may have also encouraged regionalization. Following Britain’s devaluation in September 1931, countries that had traditionally borrowed and held interna- tional reserves in London pegged their currencies to sterling. Some, like India, were compelled to do so, but others participated voluntarily. This gave rise to the sterling area, a zone of exchange rate stability thought to have encouraged trade among participants (see Meyer, 1952). The sterling area and the British Commonwealth were not the same: Canada, New- foundland, and British Honduras belonged to the Commonwealth but not the sterling area, while Denmark, Norway, Sweden, Portugal, and other non-Commonwealth countries joined the sterling area.

Other countries restricted credit as necessary to defend their gold standard parities and suffered a prolonged slump in production and trade. A few gold standard countries, namely France, Belgium, Switzerland, and the Netherlands, maintained exchange rate stability vis-a-vis one another until the second half of the 1930s which may have also worked to stimulate trade among them. They strengthened their tariffs and import quotas to defend their increasingly overvalued exchange rates, and gold bloc countries applied surtaxes (‘exchange-dumping duties’) against countries devaluing their currencies.’ France and the Netherlands employed quotas as well as

bThe countries were Canada, Guatemala, El Salvador, Honduras, Nicaragua, Costa Rica, Cuba, Haiti, Colombia, Brazil, France, Belgium/Luxembourg, the Netherlands, Sweden, Finland, and Switzerland.

7The agreement with Canada was expanded, while new agreements were concluded with Ecuador, Venezuela, the United Kingdom, Czechoslovakia, and Turkey.

‘As the League of Nations put the point, “Countries with overvalued currencies and consequently high price-levels.. .resorted to increasingly stringent measures aimed at restricting imports and encouraging exports. In the ‘gold bloc’ recourse was mainly had to an extensive system of import quotas and export bounties...” (cited in Woytinsky and Woytinsky, 1955, p. 294). Once these countries abandoned gold convertibility in 1935-36 and allowed their overvalued exchange rates to depreciate, their more draconian trade restrictions could be relaxed. But one gold bloc member, France, suffered continued balance of payments problems and by 1938 had restored tariffs and quotas to t!ie lrv:+: prevailing before 1936.

Page 8: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

8 B. Eichengreen, D.A. Irwin 1 .I. Int. Econ. 38 (1995) 1-24

tariffs, extending quota rights differentially to foreign suppliers as a way of securing market access for their exports.

A third group of countries in Central and Southeastern Europe also maintained their gold standard parities but supported them with exchange controls, not tariffs and quotas. Controls were imposed unilaterally in the summer and fall of 1931, but in November of that year a conference of Danubian countries convened in Prague to consider the negotiation of a multilateral clearing agreement designed to facilitate trade among them. Ultimately, this proposal was not adopted, and the exchange control countries relied on bilateral agreements for clearing (see Anderson, 1946).

Few countries linked their currencies to the dollar as tightly as the members of the sterling area pegged to the pound and the gold bloc currencies were linked to one another. The formal dollar bloc included only the United States, the Philippines, Cuba, and Central America, although countries such as Canada and Argentina followed the dollar at a distance.

Accompanying these changes was a reorientation of trade: the pattern of multilateral settlements was submerged beneath a web of bilateral and regional commercial and monetary arrangements. No consensus exists about the relative importance of trade blocs and currency areas in bringing about these changes. Did discriminatory trade blocs cause the shifts in the pattern of world trade, or can they be dismissed as unimportant? Did the sterling area and other currency arrangements have only a minor impact on the pattern of trade, or were payments arrangements, as others argue, highly influential? We now seek to determine the extent to which trade blocs and currency arrangements were responsible for the changing pattern of trade.

3. Data and specification

Starting with Linneman (1966), a standard framework for investigating the pattern of trade is the gravity model, which relates the value of bilateral flows to national income, population, distance, and contiguityP Most applications specify the model in double-log form, expressing the dependent variable (bilateral trade) and all independent variables in logs, and estimate it by ordinary least squares (OLS). The double-log specification permits coefficients to be interpreted as elasticities but omits country pairs for which trade is zero. This is undesirable insofar as the omitted observations convey information about why low levels of trade are observed.

9 Anderson (1979) and Bergstrand (1985) have shown that the gravity equation can be derived as a reduced form of a broad class of structural models. In applications to contempor- ary data, such equations have proven useful in predicting the pattern of trade and assessing the effects of commercial policies.

Page 9: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

B. Eichengreen, D.A. Irwin I I. Int. Con. 38 (1995) l-24 9

One solution is to forgo the double-log specification for a semi-log form in which the dependent variable (bilateral trade) is expressed in levels, and independent variables in logs. While this allows us to utilize the information contained in the observations for which the observed value of trade is zero, OLS is no longer appropriate since the dependent variable is truncated at zero. The obvious estimator is Tobit (as in Havrylyshyn and Pritchett, 1991), since the analogy between the gravity equation and the standard Tobit model is direct: countries first determine whether to trade, and if they do the value of trade is determined by their characteristics. The semi-log specification is difficult to interpret, however, because the constant elasticity relationship between the independent variables and bilateral trade is lost.”

A third strategy, which we utilize here, preserves the double-log form and yields results similar to Tobit. ‘The dependent variable is expressed as ln(1 + TRADE), where TRADE is the value of bilateral exports and imports. For large values of TRADE, ln(1 + TRADE) = ln(TRADE) and the constant elasticity relationship is preserved; for small values, ln(1 + TRADE) = TRADE, approximating the semi-log Tobit relationship. The equation can be estimated by scaled OLS (or an analogous systems estimator), in which the least squares estimates are multiplied by the reciprocal of the proportion of observations in which TRADE does not equal zero.”

We adopt a specification similar to that used by Frankel et al. (1994):

ln(1 + TRADE,,) = PO + PI ln(GNP, . GNPj)

+ & ln([GNP,IPOP,] . [GNPjIPOPj]) + &DZST,,

+ &CONT + uij,

where TRADE, is the nominal value of bilateral trade between countries i and j, GNP,. GNPi is the product of the two countries’ nominal national incomes, [GNP,IPOP,] . [GNPjIPOPj] is the product of the two countries’ per capita national incomes (also in nominal terms), DZST, is the distance between the economic centers of gravity of the two countries (measured in miles or kilometers), and CONT is a dummy variable for whether the two countries are contiguous.

A problem with this specification is that the disturbance may be correlated

I” One can solve for the local elasticity around the mean by dividing the coefficients by the mean value of the dependent variable, but then the elasticity is necessarily declining with respect to the dependent variable.

” According to Greene (1993, p. 697), “a striking empirical regularity is that the maximum likelihood estimates can often be approximated by dividing the OLS estimates by the proportion of nonlimit observations in the sample.” This is the case in the results we report below.

Page 10: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

10 B. Eichengreen, D.A. Irwin I J. Int. Econ. 38 (1995) l-24

with one of the regressors, thereby biasing the estimates. If a large ujj is drawn, for instance, trade is higher and, by accounting identity, GNP is higher as well (unless trade is always perfectly balanced). Instrumental variables estimation is therefore required, as in Harrigan (1992) and Hummels and Levinsohn (1993), using instruments likely to be correlated with national income but independent of the value of international trade. Harrigan and Hummels and Levinsohn use factor endowments (such as capital stock and labor) as instruments; for reasons of data availability we use arable land and economically active labor force, as described in the data appendix.

The source for trade flows is Folke Hilgerdt’s study The Network of World Trade. The sum of a country’s exports and imports with another country is expressed in millions of U.S. dollars. Distance is the straightline distance (in kilometers) between the economic center of gravity of the respective countries.‘* Nominal national income, sources of which are described in the data appendix, was converted to millions of U.S. dollars using the exchange rates in the Network. We assembled data for 34 countries (561 bilateral flows).

Hilgerdt provides data on bilateral trade for 1928, 1935, and 1938. Estimating equations for the 1920s as well as the 1930s allows us to address the possibility that effects we attribute to the commercial and financial initiatives of the latter decade are in fact long-standing features attributable to other sources. We can test Schlote and Thorbecke’s hypothesis that unusually high levels of intra-Commonwealth trade were already evident before the 1930s for example.

The availability of three cross-sections suggests estimating the three equations as a system using seemingly unrelated regressions (SUR). To the extent that errors are trading-partner-specific (reflecting, for example, characteristics of particular pairs of trading partners that are difficult to capture by their observable economic characteristics), SUR will result in more efficient estimates.

4. Econometric results and interpretation

Table 1 presents estimates of the basic model using various estimation techniques for purposes of comparison. Together, the independent variables explain 50-60 percent of the variation in the value of trade. All the variables display their expected signs. The sum of the coefficients on income and income per capita is larger in 1928 than in 1935-38 regardless of the

I* This follows Linneman (1966). We are indebted to Lant Pritchett for making this data available to us.

Page 11: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

Table

1

Deter

mina

nts

of tra

de

in the

int

erwar

perio

d. ex

clusiv

e of

trade

an

d cu

rrenc

y blo

cs

1928

19

3s

Scale

d Sc

aled

Scale

d Do

uble-

To

bit

Scale

d Sc

aled

Scale

d Do

uble-

To

bit

OLS

IV-O

LS

IV-S

UK

log

IV

OLS

IV-O

LS

IV-S

UR

log

IV

OLS

OLS

1938

Scale

d Sc

aled

Scale

d Do

uble-

To

bit

OLS

IV-O

LS

IV-S

UR

log

IV

OLS

Mean

of

trade

2.0

1 2.0

1 2.0

1 2.4

9 42

.34

1.75

1.75

1.75

2.12

Cons

tant

-4.19

-4.

85

-4.28

4.9

3 -8.

71

-3.12

-3.

50

-3.20

-3.

82

(5.76

) (5.

98)

(5.36

) (6.

85)

(7.93

) (5.

83)

(4.52

) (4.

21)

(5.50

)

Natio

nal

0.91

1 .Ol

1.0

1 0.9

0 1.1

4 0.7

3 0.8

3 0.8

0 0.6

9 inc

omes

(27

.58)

(24.78

) (23

.28)

(20.50

) (17

.62)

(23.34

) (20

.44)

(20.02

) (17

.04)

Per

capit

a 0.3

3 1.1

8 0.8

5 0.8

9 1.3

0 0.1

6 0.8

1 0.6

2 0.6

6 inc

omes

(6.

28)

(20.66

) (17

.74)

(16.30

) (15

.96)

(3.19

) (15

.95)

(14.9i

l) (13

.14)

Dista

nce

--0.78

--0

.73

(12.63

) (11

.26)

Conti

guity

0.7

9 0.7

5 (2.

94)

(2.65

)

561

561

0.75

-0.51

(11

.80)

(8.67

)

0.75

0.47

(2.68

) (2.

02)

561

419

-0.56

-0.

40

(6.34

) (8.

46)

1.38

0.41

(3.75

) (2.

09)

-0.52

(7.

75)

0.51

(1.83

)

-0.58

W

V -0.

35

(5.89

)

n R2

S.E

561

561

561

0.48

0.46

(1.73

) (1.

98)

561

418

0.69

0.65

0.63

0.59

3.081

0.33

0.63

0.58

0.57

0.52

23.52

1.8

5 1.8

5 1.8

5 2.2

2 27

.27

-8.09

-4.

86

-6.39

(7.

99)

(6.W

(7.

64)

-5.31

(6.

49)

0.97

0.77

1.01

(17.00

) (23

.65)

(21.84

) 0.9

1 (20

.37)

1.03

0.16

0.89

0.63

(14.45

) (3.

62)

(18.67

) (15

.72)

--0.37

-0.

53

-0.5

1 (4.

31)

(8.64

) (7.

82)

1.14

0.45

(3.30

) (1.

69)

561

561

0.42

(1.51

)

561

-0.57

(9.

07)

0.38

(1.42

)

561

0.29

0.62

0.59

3.201

1.0

57

2.529

1.6

92

2.611

1.0

34

0.57

1.062

-6.24

-10

.79

(8.53

) (9.

91)

0.86

1.11

(19.73

) (17

.42)

0.73

1.04

(16.59

) (16

.65)

-0.37

-0.

33

(6.79

) (4.

08)

0.30

1.19

(1.37

) (3.

55)

426

561

0.58

0.31

2.582

1.6

42

1.088

1.8

12

2.542

1.0

26

1.038

Notes

: Th

e 34

sa

mple

coun

tries

are

the

Unite

d St

ates,

Cana

da,

Cuba

, Gu

atema

la,

Mexic

o, Br

azil,

India,

the

Ne

therla

nds

Indies

(In

done

sia),

Japa

n/Taiw

an/K

orea,

Austr

ia;

Belgi

um,

Czec

hoslo

vakia

, Fr

ance

, Ge

rman

y, Ita

ly,

the

Nethe

rland

s, Sw

eden

, Sw

itzerl

and,

Bulga

ria,

Denm

ark,

Finlan

d, Gr

eece

. Hu

ngary

, No

rway

, Po

land,

Portu

gal.

Roma

nia,

Spain

, the

So

viet

Union

, Ire

land,

the

Unite

d Ki

ngdo

m,

Austr

alia,

New

Zeala

nd,

and

South

Af

rica.

All

trade

an

d inc

ome

varia

bles

are

expre

ssed

in

millio

ns

of U.

S.

dolla

rs,

using

the

co

nvers

ion

factor

de

scrib

ed

in the

te

xt.

The

depe

nden

t va

riable

for

sc

aled

OLS

and

scale

d SU

R is

ln(

1 +

trade

), wh

ere

trade

=

expo

rts

t im

ports

. Th

e do

uble-

log

OLS

depe

nden

t va

riable

is

In(tra

de),

dropp

ing

obse

rvatio

ns

for

which

tra

de

is ze

ro.

The

depe

nden

t va

riable

for

To

bit

estim

ation

is

trade

=

expo

rts

+ im

ports

. r-s

tatist

ics

arc

in pa

renthe

ses.

R’ of

Tobit

is

the

squa

red

corre

lation

be

twee

n the

ac

tual

and

fitted

va

lue

of the

de

pend

ent

varia

ble.

Page 12: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

12 B. Eichengreen, D.A. Irwin I .I. Int. Econ. 38 (1995) l-24

estimator used, consistent with the fact that trade, after falling along with output and employment during the Depression, failed to rebound fully with the recovery of incomes. This pattern is consistent with the arguments of Woytinsky and Woytinsky (1955), who attribute the reorientation of trade in the 1930s in part to changes in marginal propensities to import. The obvious explanation for these declining marginal propensities is quotas and other binding trade restrictions, the tightening of which prevented imports from rising with incomes in the recovery as quickly as they had fallen in the slump. The effects of distance and contiguity are larger in 1928 than subsequently. This effect seems too large to reflect declining transportation and information costs. Again, our preferred interpretation is that commer- cial and financial policies increasingly dominated the effects of geography.

In Table 1, in the first column for each year we present the scaled OLS results produced by estimation without instrumental variables. There is evidence of structural change between 1928 and 1935 (the Chow test Fc5,,1111) = 11.87) but not between 1935 and 1938 (Fc5, 1111) = 0.42). This finding is robust to the inclusion of dummy variables for trade and currency blocs. The second column of each year presents results from scaled OLS when we instrument the product of national incomes. This raises the coefficient on national incomes and per capita national incomes without noticeably changing the distance and contiguity effects.

The scaled SUR estimates (in the third column for each year) are little different than the scaled OLS results, although the magnitude of the coefficients on per capita income is reduced slightly. The correlation of disturbances across equations is sufficient to reject the hypothesis that the residual covariance matrix is diagonal, indicating that SUR estimation is appropriate.13 This makes sense in that the characteristics of countries not picked up by the other regressors but affecting trade are likely to change slowly over time. This estimation approach will be our preferred method in assessing the impact of trade and currency blocs.

In the last two columns for each year we present results using alternative estimation techniques. In much of the empirical gravity model literature, a simple double-log OLS approach is taken. These estimates, in the fourth column, deliver uniformly smaller coefficients except for a stronger effect of per capita income. The Tobit estimates in the fifth column (also using instrumental variables) most closely resemble the scaled OLS (instrumental variables) results, although the effect of contiguity is more pronounced.

How do these results compare with those for later periods? Using data

2 I3 The cross-equation correlation of disturbances is as follows: r&,, = 0.79, r&,, = 0.71,

r35,38 = 0.71. The likelihood ratio test (x’ distribution) of whether the off-diagonal elements of the variance-covariance matrix are zero (indicating no correlation across years) is 1270.016, decisively rejecting that hypothesis (see Greene, 1993, p. 454).

Page 13: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

B. Eichengreen, D.A. Irwin I .I. lnt. Econ. 38 (199.7) 1-24 13

from 1980 and 1990, Frankel et al. (1994) find that a double-log specification yields uniformly lower coefficients on all of the independent variables than our double-log estimation, although our instrumental variables approach accounts for much of this difference. In results we do not report, a double-log specification without instruments on our interwar data yields similar coefficients on national incomes (around 0.7) to Frankel et al., but lower coefficients on per capita national incomes (around 0.1-0.2 for the interwar period compared with 0.3-0.5 in recent decades). Per capita national incomes are commonly interpreted as a measure of intra-industry trade, and a lower coefficient earlier in the century suggests that intra- industry trade among countries of comparable income levels may not have been as important before World War II as after.14 Frankel et al. also find slightly larger coefficients on distance (ranging from 0.5 to 0.6) compared with interwar estimates (around 0.3-0.6) providing no evidence that declining transportation costs have been an important influence on the pattern of trade. They also find a higher coefficient on contiguity, consistent with the formation of location-based trading patterns rather than colonial- based trading arrangements.

4.1, Trade blocs

Table 2, reporting only scaled OLS and scaled SUR results, introduces variables representing the principal regional arrangements of the 1930s: the British Commonwealth and the German sphere of commercial influence (both trade blocs), and the sterling area, the gold bloc, and the exchange control countries (which we refer to as currency blocs).15 The first dummy variable for each grouping equals unity if both trade partners belong to the same bloc. Its coefficient indicates whether or not these countries tended to trade more with one another as a result. (The exponential of these coefficients yields the impact of these policies on trade in percentage terms.) The second dummy equals unity if one country but not both belong to the same bloc. Its coefficient indicates whether the arrangement diverted trade away from the rest of the world. To test whether the effects we attribute to these variables reflect the commercial and financial initiatives of the 193Os,

I4 This is consistent with Lewis’s (1949) description of pre-World War I and interwar trade as primarily an exchange of manufactured goods and raw materials between industrial and tropical countries.

I5 Exchange controls and clearing arrangements resemble financial restrictions in some respect but commercial policies in others. The International Monetary Fund (in its Annual Report on Exchange Controls and Exchange Restrictions) treats exchange controls separately from trade policies. We follow this convention and discuss exchange controls and clearing arrangements as a form of currency policy. Whether countries adopting them are properly regarded as a bloc are from the data to determine.

Page 14: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

14 B. Eichengreen, D.A. Irwin I J. Int. Econ. 38 (1995) 1-24

Table 2 Determinants of trade in the interwar period: Effect of overlapping trade and currency blocs

Variable 1928 1935 1938

Scaled SUR IV

Scaled SUR IV

Scaled SUR IV

Constant

National incomes

Per capita incomes

Distance

Contiguity

WI in Commonwealth

Commonwealth member

W/in Reichsmark

Reichsmark member

W/in sterling area

Sterling member

W/in gold bloc

Gold bloc member

Wlin exchange control

-4.83 -3.62 -6.72 (5.28) (4.19) (7.41)

1.02 0.81 0.95 (23.03) (20.25) (20.97)

0.85 0.61 0.59 (16.25) (13.67) (13.98)

-0.71 -0.55 -0.51 (9.68) (7.59) (7.20)

0.61 0.28 0.26 (2.19) (1.03) (0.98)

1.31 1.67 1.66 (3.97) (5.22) (5.25)

0.01 -0.13 -0.13 (0.05) (0.87) (0.89)

0.69 0.84 0.89 (2.22) G-4) (3.05)

0.09 -0.31 -0.11 (0.55) (2.01) (0.71)

0.04 -0.17 0.24 (0.20) (0.68) (1.10)

-0.10 0.08 0.29 (0.68) (0.53) (1.93)

0.29 0.53 0.68 (0.73) (1.36) (1.76)

0.56 0.39 0.64 (3.59) (2.55) (4.28)

0.66 0.86 0.67 (1.71) (2.27) (1.90)

Page 15: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

B. Eichengreen, D.A. Irwin I J. Int. Econ. 38 (lW.5) 1-24

Table 2 (continued)

15

Variable 1928 1935 1938

Scaled SUR IV

Scaled SUR IV

Scaled SUR IV

Exchange control member

-0.41 -0.38 -0.29 (2.52) (2.30) (1 SO)

n 561 561 561

H’ 0.67 0.62 0.62

S.E.

Notes: See Table 1.

1.044 0.979 I.006

as opposed to interdependencies already evident in the previous decade, we include the same variables in our equations for 1928.

The findings reported above survive this extension: the basic coefficients retain their expected signs and, other than the occasional per capita income and continguity term, are significant at conventional levels. As before, the effects of distance, contiguity and the sum of national and per capita incomes decline after 1928. The obvious explanation is again in terms of commercial policies, but now commercial policies other than the tariff preferences for which we control, that overrode the effects of income and geography.

The positive coefficient on ‘Within Commonwealth’ suggests unusually heavy trade in 1928 among Commonwealth members (the United Kingdom, Ireland, Canada, India, Australia, New Zealand, and South Africa in our sample). Even though British tariff rates were low prior to the imposition of the 1932 General Tariff, some preferences were already extended to the Dominions, and the latter extended preferences to Britain and one another (Layton and Rist, 1925; Capie, 1983). But it is likely that the prevalence of Commonwealth trade reflected more than preferences alone; commercial and financial linkages that had developed over the years also contributed to these patterns.16

The magnitude and significance of ‘Within Commonwealth’ increases between 1928 and 1935. We can reject the hypothesis that this coefficient is equal across years at the 90 percent level; the x2 test of the nonlinear restriction equating the two coefficients across SUR estimates is 2.78 with

” Hamilton and Winters (1992) similarly find a persistent influence of historical ties in data on U.K. trade with its former colonies and Dominions in the 1980s. de Menil and Maurel (1993) use a gravity equation model to analyze trade for a more limited sample of countries in 1924 and 1926, reaching similar conclusions about the importance of the historical legacy in the case of the successor states of the prewar Austro-Hungarian Empire.

Page 16: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

16 B. Eichengreen, D.A. Irwin I J. Int. Econ. 38 (1995) 1-24

the associated p-level of 0.09. Thus, the Ottawa Agreements of 1932 appear to have reinforced the existing tendency toward intra-Commonwealth trade. The coefficient on ‘Commonwealth Member’, indicating bilateral flows between Commonwealth countries and the rest of the world, is negative, though inconsistently significant, suggesting a weak trade-diverting effect of Britain’s imperial arrangements. While consistent with Condliffe’s (1940) assertion that Commonwealth preferences had a negative impact on the participating countries’ trade with the rest of the world, this effect is already evident (and near1 significant at the 90 percent level in the scaled OLS equation) in 1928. 17 The extent of diversion rises year by year, although only the scaled OLS estimates suggest that it was significant (at the 90 percent level in 1935 and the 95 percent level in 1938).

The coefficients for the Reichsmark bloc are surprisingly similar to those for the Commonwealth. There too a tendency is evident for participating countries (Germany, Austria, Brazil, Bulgaria, Czechoslovakia, Greece, Hungary, and Romania in our sample) to already trade unusually heavily with one another in 1928, a point noted by Ellis (1941, p. 258). There too there is a suggestion of an increase in this propensity between 1928 and 1935-38, although the change in the coefficient is smaller and less significant than in the case of the Commonwealth. Basch (1941) argued that the Reichsmark bloc was not designed to stimulate the volume of intra-bloc trade, only to balance it while enhancing Germany’s self-sufficiency in raw materials; our results are consistent with the first of these conclusions. The scaled SUR estimates suggest that the Reichsmark bloc became strongly trade diverting in the 1930s in contrast to our results for the Common- wealth where there is little evidence of such an effect.18

Compared with the existing literature, then, we are led to a more cautious assessment of the effects of trade blocs in the 1930s. Some of the effects commonly ascribed to them were already evident in the 1920s before the discriminatory policies of the 1930s were adopted, as Schlote and Thorbecke emphasize.

We also investigated the Ouchy Accord of the Benelux countries. Our results for 1928 indicate that these countries already traded heavily with one another prior to the agreement, but the ‘Within Ouchy’ coefficient increases in magnitude in 1935 and again in 1938.r9 Despite the formal failure of the Accord, the extensive political and commercial contacts of which Ouchy was one manifestation evidently led the Benelux countries to trade more extensively with one another than contiguity and propinquity would other-

“According to scaled SUR, there was no trade diversion due to Commonwealth member- ship in that year.

18The x2 test statistic of 4.84 leads us to reject the hypothesis of equality. 19These results are available from the authors upon request.

Page 17: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

B. Eichengreen, D.A. Irwin I J. In!. Econ. 38 (1995) 1-24 17

wise suggest and without concomitant trade diversion. In contrast, the Brocchi-Rome Agreements between Austria, Hungary, and Italy signifi- cantly discouraged trade with the rest of the world, while failing to stimulate the signatories’ trade with one another. Gordon (1941, p. 452) asserts that “the effects of the Rome Agreements. . . were not as favorable as had been anticipated”. Condliffe (1940) provides a similar assessment; our results are consistent with this view.

Finally, we tested for the existence of a Western Hemisphere bloc comprised of the United States and various Latin American countries. Although these countries already traded unusually heavily with one another in 1928, intra-hemisphere trade had become more pronounced by 1935. By 1938, this intra-hemisphere-trade effect is less apparent, perhaps reflecting the more wide-ranging application of the Reciprocal Trade Agreements Act. In no year does there appear to be a diversionary element in Western Hemisphere trade.

4.2. Currency blocs

The remaining coefficients capture the impact on trade of the sterling area, the gold bloc countries, and the Central and Eastern European countries with exchange controls and clearing arrangements. Participation in the sterling area (Australia, Denmark, Finland, India, Ireland, New Zealand, Norway, Portugal, Sweden, and South Africa in our sample) has no discernible impact on trade in any year. In 1938, there is a positive impact of the dummy variable for trade flows between sterling area and non-sterling area countries. The rise of the sterling area’s trade with the rest of the world in the 1930s is commonly ascribed to these countries’ rapid recovery from the Depression, but to the extent that recovery is controlled for by the national income terms, this cannot be the entire explanation. Another possibility is these countries’ reliance on currency devaluation as a recovery-promoting policy. The members of the sterling area all followed the United Kingdom off the gold standard at an early date. Devaluation allowed them to reduce interest rates, helping to stimulate recovery, but also enhanced their international competitiveness, encouraging them to export more than other countries following more conservative international monetary policies (Eichengreen and Sachs, 1985). This last effect may account for the positive coefficient on ‘Sterling Member’.

There is little evidence that gold bloc countries (Belgium, France, the Netherlands, Poland, and Switzerland in our sample) traded disproportion- ately with one another. The only coefficients approaching statistical signifi- cance at standard levels are those for 1938, after the gold bloc was dissolved. As with the sterling area, this could reflect the trade-promoting effects of the simultaneous devaluation of gold bloc currencies, in this case starting in

Page 18: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

18 B. Eichengreen, D.A. Irwin I J. Znt. Econ. 38 (199.5) 1-24

1936. Although the coefficients on ‘Gold Bloc Member’, the variable indicating bilateral flows between gold bloc and non-gold bloc countries, are already positive in 1928, indicating that these countries had a disproportion- ate tendency to trade with other parts of the world even before their currency bloc was formed, they increase in size and statistical significance between 1935 and 1938, consistent with the interpretation of their growing trade in terms of the trade-stimulating effects of the post-1935 depreciations.20

The final pair of coefficients tests for trade creation and diversion on the part of the exchange control countries. While their signs in 1935-38 suggest that these policies encouraged trade among exchange control countries while discouraging trade with the rest of the world, the same tendencies are already evident in 1928, before controls were imposed. The results in Table 2 thus suggest that it is not necessarily appropriate to attribute either effect to the controls themselves. The exchange control countries were a well- defined, isolated group, but were so before the 1930s. In the next subsection we provide an interpretation of the contrast between this finding and the conclusions of qualitative studies.

4.3. Exchange rate variability

Nurkse (1944) argued that exchange rate volatility following the break- down of the international gold standard increased the uncertainty associated with international transactions, thereby discouraging trade. Following Fran- kel et al. (1994), we extend our model to include a measure of exchange rate variability. Our measure is the variance of log(e,le,-,), where e, is the average monthly bilateral nominal exchange rate between two countries over the three prior years, i.e. from 1925 to 1927 in the case of trade in 1928.21

The scaled SUR estimates in Table 3 support Nurkse’s hypothesis mainly for 1928 and, to a much lesser extent, 1938. At the bottom of the table we present the mean value of our exchange rate variability measure: note that variability is most pronounced in the years prior to 1928, when European countries were still undertaking stabilization policies to reduce high inflation rates in the aftermath of World War I.22 The absence of economical forward

*’ Formally, we cannot reject the equality of the coefficient in 1928 and 1935, although the x2 for the change between 1928 and 1938 of 2.43 yields a p-value of 0.119, which is nearly significant at conventional levels.

*’ Gaps in the availability of exchange rate data (which we draw from League of Nations’ publications) reduce the sample size somewhat.

**The European countries that are responsible for the high exchange rate variability in 1925-27 are such countries as Belgium, Denmark, Finland, France, and Italy, which ex- perienced periods of persistent inflation followed by stabilization and periods of deflation prior to pegging their exchange rates.

Page 19: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

B. Eichengreen, D.A. Irwin I .I. Int. Econ. 38 (1995) l-24 19

Table 3 Bilateral trade and exchange rate variability

Variable 1928 1935 1938

Scaled SUR Scaled SUR Scaled SUR

Constant

National incomes

Per capita 1.14 0.79 0.68 incomes (13.92) (13.55) (12.23)

Contiguity

Exchange rate -0.04 -0.03 -0.03 variability (2.39) (1.08) (1.69)

W/in Commonwealth

2.41 (3.35)

1.83 ,190 (5.13) (5.11)

-0.14 -0.01 (0.86) (0.03)

0.88 0.96 (2.69) (2.81)

Commonwealth 0.17 member (0.W

W/in 0.71 Reichsmark (2.45)

Reichsmark -0.01 -0.10 0.01 member (0.08) (0.58) (0.03)

W/in sterling 0.28 -0.20 0.12 area (0.92) (0.84) (0.43)

Sterling member

W/in gold bloc

Gold bloc 0.47 member (2.63)

W/in exchange 0.80 0.23 0.39 control (2.09) (0.49) (0.81)

-5.89 (5.07)

1.33 (17.48)

-0.68 -0.63 -0.58 (7.10) (7.60) (6.96)

0.55 (2.00)

-0.06 (0.76)

0.29 (0.77)

-6.07 -6.81 (5.39) (5.99)

1.03 0.99 (19.38) (17.77)

0.35 0.35 (1.14) (1.11)

0.10 0.25 (0.64) (1.14)

0.21 0.49 (0.55) (1.21)

0.17 0.60 (1.01) (3.51)

Page 20: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

20 B. Eichengreen, D.A. Irwin / J. Int. Econ. 38 (199.5) l-24

Table 3 (continued)

Variable 1928 1935 1938

Scaled SUR Scaled SUR Scaled SUR

Exchange control member

-0.51 -0.62 -0.25 (2.74) (3.49) (1.38)

n 377 435 435

R’ 0.72 0.67 0.64

S.E. 0.921 0.931 0.982

Mean value of exchange rate variability

0.066017 0.000334 0.00454

markets and hedging instruments plausibly accounts for the continued negative effect of variability on trade.

However, the impact of exchange rate variability on trade, while statisti- cally significant in 1928, is economically unimportant compared with other factors. Exchange rate variability reduced bilateral trade-on average-by only $2,500 a year, or by 0.13 percent. The expansion of trade in moving from average exchange rate instability to that enjoyed by countries that were on the gold standard (i.e. from 0.066 to 0.001 in 1928) were at best minor. This is consistent with findings that post-war trade has not been substantially affected by the collapse of the Bretton Woods system and the move to market-determined exchange rates. The magnitude of exchange rate instability was smaller in 1932-34, when no countries were experiencing sustained inflation comparable to that of 1925-27, although it increased in 1935-37 as various countries adopted floating rates. However, the effect on trade, while negative, is statistically indistinguishable from zero in both 1935 and 1938.

Including exchange rate variability alters the coefficients of the trade and currency blocs, since these variables are correlated (partners in a currency bloc having had stable bilateral exchange rates). There is little change in the coefficients on the two sterling area variables between Tables 2 and 3, suggesting that exchange rate stability was not a major source of the area’s impact on trade. Whereas Table 2 indicates that gold bloc countries had a disproportionate tendency to trade with other parts of the world before, during, and after the formation of their bloc, when we control for exchange rate variability the tendency for those Continental European countries that eventually comprised the gold bloc to trade disproportionately with other parts of the world tends to disappear in 1935. The obvious explanation is the

Page 21: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

R. Eichengreen, D.A. Irwin I J. Int. Econ. 38 (199.5) 1-24 21

exchange dumping duties that gold bloc countries levied against imports from countries with depreciated currencies.

After controlling for exchange rate variability, the tendency for the Reichsmark bloc to be trade diverting, noted in Table 2, is weakened. To the extent that these countries traded less with partners outside the Reichsmark bloc than their incomes, distance, and contiguity would predict, this would appear to reflect mainly the effect of exchange rate variability between Reichsmark bloc members and other parts of the world, not merely the diverting effects of the bloc’s commercial policies. Finally, Table 3 indicates a decline between 1928 and 1935 in the size and significance of the coefficients on ‘Within Exchange Control’ which was not evident in Table 2. This supports the belief that the imposition of exchange controls by both trading partners discouraged bilateral trade. That this effect was not apparent in Table 2 is attributable to the offsetting tendency for the stability of bilateral exchange rates between such countries to encourage trade.

5. Conclusions

This paper has reassessed the role of commercial and financial policies in the reorientation of international trade in the 1930s. We report evidence that commercial policies attenuated prior linkages between income growth and trade and that exchange rate instability only modestly discouraged interna- tional transactions. In contrast, the tendency toward regionalization com- monly ascribed to the formation of trade and currency blocs was already evident prior to the regional policy initiatives of the 1930s; to a considerable extent it is attributable to ongoing historical forces such as commercial and financial linkages between countries forged over many years. While there is some evidence that the formation of trade blocs diverted transactions toward fellow bloc members at the expense of trade with the rest of the world, this was only one of several factors at work, as also discussed in Irwin (1993).

We find that the different currency blocs of the 1930s had very different implications for trade. Sterling area countries traded disproportionately among themselves and with the rest of the world. Gold bloc members, in contrast, did not trade disproportionately with one another or with the rest of the world, reflecting their indiscriminate use of tariffs and quotas to prop up increasingly overvalued currencies, which neutralized any stimulus derived from exchange rate stability. Countries applying exchange controls, despite stabilizing their exchange rates, traded less with one another than their economic characteristics would predict, due to the trade-inhibiting effects of those policies.

What are the implications of these findings for discussions of the emerging

Page 22: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

22 B. Eichengreen, D.A. Irwin / J. Int. Econ. 38 (1995) 1-24

trade and currency blocs of the 199Os? They suggest caution for those tempted to attribute to preferential trade and currency arrangements all deviations in trade flows from those predicted by the readily observed characteristics of trading partners. As in the 193Os, those deviations may reflect not just regional policy initiatives but also long-lived historical forces working to encourage disproportionate volumes of trade among certain countries, and perhaps encouraging those countries to cement their relation- ships through the adoption of preferential arrangements. They suggest that trade and currency blocs are not inevitably either trade-creating or trade- diverting. Trade and currency blocs can encourage trade both within the bloc and outside it, as in the case of the sterling area, or discourage trade even within the bloc, as with the countries that linked themselves together in the 1930s through the use of exchange controls and bilateral clearing arrangements.

Acknowledgements

Eichengreen acknowledges the financial support of the Center for Ger- man and European Studies and the Center for International and Develop- ment Economic Research at the University of California at Berkeley. Irwin acknowledges financial support from the James S. Kemper Faculty Founda- tion Research Fund at the University of Chicago’s Graduate School of Business. Much of the research on this paper was completed during Eichengreen’s visit to the Research Department of the International Monetary Fund. For helpful comments, we thank Georges de Menil and two anonymous referees.

Data appendix

This appendix presents the sources used in estimating the gravity equa- tions.

Trade. League of Nations, The Network of World Trade, Geneva, 1942. National income. Austria, Belgium, Bulgaria, Denmark, Finland, Ger-

many, Hungary, Greece, Italy, the Netherlands, Norway, Spain, Sweden, the Soviet Union: B.R. Mitchell, European Historical Statistics, 1750-1975, 2nd revised edn., Facts on File, New York, 1980. France: J.-C. Toutain, ‘Le Produit Interieur Brut de la France de 1789 a 1982’, Economies et Societes, 1987. Switzerland, Czechoslovakia, and Poland: Colin Clark, Conditions of Economic Progress, 3rd edn., Macmillan, London, 1957. Romania: M. C. Kaser and E. A. Radice, The Economic History of Eastern Europe, 1919- 1975, Vol. 1, Economic Performance Between the Two Wars, Clarendon Press, Oxford, 1985. United States: U.S. Bureau of the Census, Historical Statistics of the United States, Colonial Times to 1970, Government Printing

Page 23: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

B. Eichengreen, D.A. Irwin I J. Int. Econ. 38 (1995) l-24 23

Office, Washington, DC, 1975. Canada: M. C. Urquhart and K. A. H. Buckley, Historical Statstics of Canada, Macmillan Press, Toronto, 1965. Ireland and the United Kingdom: B.R. Mitchell, British Historical Statistics, Cambridge University Press, New York, 1988. Australia and New Zealand: N.G. Butlin, ‘Select Comparative Economic Statistics 1900-1940: Australia, and Britain, Canada, Japan, New Zealand’, Source Paper No. 4, Australian National University, December 1984. India and Netherlands Indies: A. Maddison, ‘Dutch Income in and from Indonesia’, Modern Asian Studies 23 (October 1989), 645-670. South Africa: B.R. Mitchell, International Histori- cal Statistics: Africa and Asia, New York University Press, New York, 1982. Japan, Korea, and Taiwan: Ohkawa et al. (1974), Suh Sang-Chul, Growth and Structural Changes in the Korean Economy, 1910-1940, Harvard University Press, Cambridge, MA, 1978, and Sam P.S. Ho, Economic Development of Taiwan, 1860-1970, Yale University Press, New Haven, 1978, respectively. Cuba, Guatemala, Mexico, and Brazil: James W. Wilke, ed., Statistical Abstract of Latin America, UCLA Latin America Center, Los Angeles, 1990. Portugal: N. Valerio, ‘0 Produto National de Portugal entre 1913 e 1947: Uma Primeiero Approximacao’, Revistu de Historiu Economica e Social, No. 11 (Janeir-Junho 1983), 89-102.

Population. Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Italy, Japan, the Netherlands, Norway, Sweden, Switzer- land, the United Kingdom, and the United States: Maddison (1989a). New Zealand: Butlin (1984). Cuba, Guatemala, Mexico, and Brazil: Wilke (1990). India and the Netherlands Indies: Maddison (1989a). Portugal: Valerio (1983). Ireland: Mitchell (1988). The following countries required interpolation between census dates. South Africa: Mitchell (1982). Greece and Spain: Mitchell (1980). Bulgaria, Czechoslovakia, Hungary, Poland, Romania, and the Soviet Union: P.S. Shoup, The Eastern Europe and Soviet Union Data Handbook, (Columbia University Press, New York, 1981).

Distance. Derived from Linneman (1966). Exchange rates. Calculated from the League of Nations Statistical Yeur-

book, various years. Land and labor force (for instrumental variables). International Institute

of Agriculture, International Yearbook of Agriculture Statistics, Rome, various years, p. xi. International Labour Office, Yearbook of Lubour Statistics, Geneva, 1939, p. 5.

References

Anderson, J.E., 1979, A theoretical foundation for the gravity equation, American Economic Review 69, 106-116.

Anderson, P.N., 1946, Bilateral exchange clearing policy (Ejnar Munksgaard, Copenhagen). Basch, A., 1941, The new economic warfare (Columbia University Press, New York). Bergstrand, J.H., 1985, The gravity equation in international trade: Some microeconomic

foundations and empirical evidence, Review of Economics and Statistics 67, 474-481.

Page 24: Trade blocs, currency blocs and the reorientation of …dirwin/docs/Blocs.pdf2 B. Eichengreen, D.A. Irwin / .I. Int. Econ. 38 (1995) l-24 1. Introduction The early 1930s witnessed

24 B. Eichengreen, D.A. Irwin I J. Int. Econ. 38 (1995) 1-24

Capie, F., 1983, Depression and protectionism (Allen & Unwin, London). Condliffe, J.B., 1940, The reconstruction of world trade (Norton, New York). de Menil, G. and M. de Maurel, 1993, Trading with neighbors in turbulent times: Lessons from

the breakup of the Austro-Hungarian Empire in 1919, unpublished manuscript, DELTA. Eichengreen, B., 1991, Relaxing the external constraint: Europe in the 1930s in: G.

Alogoskoufis, L. Papademos and R. Portes, eds., External constraints on macroeconomic policy: The European experience (Cambridge University Press, Cambridge).

Eichengreen, B., 1992, Golden fetters: The gold standard and the Great Depression, 1919- 1939 (Oxford University Press, New York).

Eichengreen, B. and J. Sachs, 198.5, Exchange rates and economic recovery in the 1930s Journal of Economic History 45, 925-946.

Ellis, H.S., 1941, Exchange control in Central Europe (Cambridge University Press, Cam- bridge).

Frankel, J., S.-J. Wei and E. Stein, 1994, APEC and regional economic arrangements in the Pacific, unpublished manuscript, University of California at Berkeley.

Gordon, M.S., 1941, Barriers to world trade (Macmillan, New York). Greene, W.H., 1993, Econometric analysis, 2nd edn. (Macmillan, New York). Hamilton, C. and L.A. Winters, 1992, Opening up trade with Eastern Europe, Economic

Policy 14, 77-117. Harrigan, J., 1992, Openness to trade in manufactures in the OECD, unpublished paper,

University of Pittsburgh. Havrylyshyn, 0. and L. Pritchett, 1991, European trade patterns after the transition, World

Bank Working Paper WPS 748, World Bank, Washington, DC. Hilgerdt, F., 1942, The network of world trade (League of Nations, Geneva). Hummels, D. and J. Levinsohn, 1993, Product differentiation as a source of comparative

advantage?, American Economic Review (Papers & Proceedings) 83, 445-449. Irwin, D.A., 1993, Multilateral and bilateral trade liberalization in the world trading system:

An historical perspective, in: J. de Melo and A. Panagariya, eds., New dimensions in regional integration (Cambridge University Press, New York).

Kitson, M., 1992, The move to autarky: The political economy of Nazi trade policy, Department of Applied Economics, University of Cambridge Working Paper No. 9201.

Lamartine Y.P., 1959, Forty years of foreign trade (Allen & Unwin, London). Layton, W. and C. Rist, 1925, The economic situation of Austria (League of Nations, Geneva). League of Nations, 1933, Economic survey 1932-33 (League of Nations, Geneva). League of Nations, 1939, Survey of world trade, 1938 (League of Nations, Geneva). Lewis, W.A., 1949, Economic survey: 1919-1939 (Allen & Unwin, London). Linneman, H., 1966, An econometric analysis of international trade flows (North-Holland,

Amsterdam). Maddison, A., 1989a, Phases of capitalist development (Oxford University Press, New York). Maddison, A., 1989b, The world economy in the twentieth century (Organization for

Economic Cooperation and Development, Paris). McDougall, D. and R. Hutt, 1954, Imperial preference: A quantitative analysis, Economic

Journal 64, 233-257. Meyer, F.V., 1952, Britain, the sterling area and Europe (Bowes and Bowes, Cambridge). Milward, A., 1981, The Reichsmark bloc and the international economy, in: G. Hirschfel and

L. Kettenacker, eds., Der ‘Fuhrerstaat’: Mythos und Realitat (Ernst Klett, Stuttgard). Neal, L., 1979, The economics and finance of bilateral clearing arrangements, Germany

1934-38, Economic History Review 33, 391-404. Nurkse, R., 1944, International currency experience (League of Nations, Geneva). Schlote, W., 1952, British overseas trade from 1700 to the 1930s (Blackwell, Oxford). Thorbecke, E., 1960, The tendency towards regionalization in international trade (Martinus

Nijhoff, The Hague). Woytinskv, W.S. and E.S. Woytinsky, 1955, World commerce and governments: Trends and