Economics Letters 55 (1997) 235240
Exports and economic growth: Evidence from 19th Century Europe 1
John Thornton
a b
a,b,
OECD, Economics Department, 2 rue Andre-Pascal, Paris 75775 Cedex 16, France Institut d Etudes Politiques de Paris, 27 rue Saint-Guillaume, Paris 75337, France
Received 8 October 1996; revised 9 January 1997; accepted 30 January 1997
Abstract Cointegration and Granger-causality tests on data for six European countries from around the middle of the 19th Century to 1913 show that exports and GNP were cointegrated and Granger-causally related with export-led growth in Italy, Norway and Sweden. 1997 Elsevier Science S.A. Keywords: 19th Century Europe; Exports; Economic growth; Cointegration; Granger causality JEL classication: F43; O11
1. Introduction Exports are generally reckoned to contribute positively to economic growth through such means as facilitating the exploitation of economies of scale (Helpman and Krugman, 1985), relieving the foreign exchange constraint (McKinnon, 1964), enhancing efciency through increased competition (Bhagwati and Srinivasan, 1979; Krueger, 1980), and promoting the diffusion of technical knowledge (Grossman and Helpman, 1991). A large number of empirical studies in this area have focused on the causal relationship between exports and economic growth; 2 these studies, which have reported mixed results, almost without exception have related to developments in real GNP (or real GDP) and real exports since the 1950s. In contrast, this paper investigates the link between exports and economic growth in six European countries from around the mid-19th Century until the eve of World War 1 (WW1). The period and the countries are of particular interest in several respects. It was a prosperous and peaceful era characterized by relatively high rates of real GNP growth and large
*Tel.: (33-1) 45248831; fax: (33-1) 45249050; e-mail: john.thornton@oecd.org 1 The views expressed in this paper are those of the author and should not be attributed to the institutions with which he is afliated. 2 A selection of the relevant papers includes Ahmed and Kwan (1991); Chow (1987); Hsiao (1987); Jung and Marshall (1985); Kwan and Cotsomotis (1991); Thornton (1996); Xu (1996). 0165-1765 / 97 / $17.00 1997 Elsevier Science S.A. All rights reserved. PII S0165-1765( 97 )00074-8
236
J. Thornton / Economics Letters 55 (1997) 235 240
international ows of goods and services, capital and labor (migration) with Europe accounting for about one-third of world output and trade in the second half of the 19th Century. International trade was conducted in the context of stable institutions and market freedoms, with much of the world having moved to xed exchanged rates by adopting the gold standard and European countries generally pursuing liberal trade policies.3 In addition, the government sector of the economy was relatively small, with developments in goods and labor markets largely left to market forces. Maddison (1995) summarizes the period as one in which the world ...relied on simple rules and protection of property rights. There were no international organizations like the OECD, IMF, BIS and GATT to manage a world system, and no equivalent of the World Bank, UN agencies, or bilateral aid donors to direct capital ows in the light of developmentalist objectives (p. 65). In short, the period from the mid-19th Century to WW1 provides a relatively unencumbered policy setting and xed institutional framework to test the relationship between exports and economic growth.
2. Data, methodology and results Data for the study are taken from the compilation of European historical statistics in Mitchell (1975). Annual data for real GNP and real exports (nominal exports deated by the implicit GNP deator) from this source are available from around the mid-19th Century for Denmark, Germany, Italy, Norway, Sweden and the United Kingdom; 4 all series are expressed in local currency. Exports in relation to GNP increased over the sample period in each country: from 26% to 28.1% in Denmark, 17.3% to 19.3% in Germany, 5.5% to 9.7% in Italy, 11.3% to 21.2% in Norway, 10.1% to 19.8% in Sweden, and from 10.1% to 19.8% in the United Kingdom. The methodology employed to investigate the relationship between exports and economic growth follows three steps. The rst step is to test the order of integration of the natural logarithm of the levels of real exports (REX) and real GNP (RGNP) over the sample periods. Thus, Table 1 gives the results of standard Phillips-Perron and augmented Dickey-Fuller (ADF) test statistics, the latter calculated to test for the presence of unit roots under the alternative hypothesis that the time series in question is stationary around a xed time trend. Each of the series appears to be integrated of order I(1) in the level form but I(0) in rst differences (Engle and Granger, 1987). Given the unit root results, the second step is to test for cointegration between real exports and real GNP in each country using the Johansen (1988) maximum likelihood approach. The Akaike Information Criterion was used to select the number of lags required in each case (Enders, 1995). The cointegration test results are presented in panel (a) of Table 2. A single signicant cointegrating vector was identied using either the maximum eigenvalue or trace statistic in ve of the six cases. On the basis of the results, I conclude that real GNP and real exports are cointegrated, and, therefore, causally related, in each country with the exception of Sweden. The third step is to carry out a standard Granger causality test augmented with an appropriate error-correction term derived from the cointegration relationship found in ve of the six countries. In addition, I include the average tariff rate in each country to control for the impact of trade liberalization on the short-run dynamics of the GNP-exports relation.5 Following Bohara and
3 4
A good discussion of economic trends in the period can be found in Maddison (1995). Precise sample periods for each country are provided in Table 1. 5 I am grateful to an anonymous referee for the suggestion to control for the effects of trade liberalization.
J. Thornton / Economics Letters 55 (1997) 235 240
237
Table 1 Unit root tests: real GNP and real exports Variable Denmark REX DDenmark REX Denmark GNP DDenmark GNP Germany REX DGermany REX Germany GNP DGermany GNP Italy REX DItaly REX Italy GNP DItaly GNP Norway REX DNorway REX Norway GNP DNorway GNP Sweden REX DSweden REX Sweden GNP DSweden GNP United Kingdom REX DUnited Kingdom REX United Kingdom GNP DUnited Kingdom GNP Period 18701913 ADF Statistic 22.8895 26.5854 ** 23.5636 * 24.3614 ** 20.6415 25.1593 ** 23.0915 25.0133 22.3509 27.8357 ** 20.5307 26.3629 ** 22.8709 26.9801 ** 1.0721 24.0338 * 22.6192 26.7630 ** 21.0850 1 23.3802 23.0592 26.3096 ** 22.2295 26.1133 * Phillips-Perron statistic 2.0821 27.0199 ** 0.8256 25.4906 ** 2.6796 27.7822 ** 21.0526 28.0452 21.1648 212.2253 ** 6.1434 ** 211.3494 ** 20.2545 29.6031 ** 21.2629 24.7235 ** 20.9108 26.8491 21.3727 210.0354 ** 22.7035 26.3915 ** 21.6677 27.4567 **
n
18801913
18611913
18651913
18611913
18501913
Notes: The ADF test is based on the regression: Dxt 5 a0 1 a1 T 1 a2 xt 21 1 S mi Dx t 2i 1 t where D is the difference i 51 operator, T is a linear time trend, and is a stationary random error. Sufcient lags were included to eliminate serial correlation. ** , * and 1 indicate signicance at the 1%, 5% and 10% levels, respectively.
Kaempfer (1991); Gardener and Kimbrough (1989); Thornton and Molyneux (1995), the ratio of total government revenue from import duties to total imports is used to proxy the average tariff rate.6 On this measure, the average tariff rate fell from the start to the end of their respective sample periods by 80% in the United Kingdom, 56% in Denmark, and about 35% in Norway and Sweden, and remained broadly unchanged in Germany and Italy (although with considerable variation over the sample period in each case).7 An appropriate formulation of a Granger-type test of causality in this regard is:
n i 51 n i 51 n i 51
DRGNPt 5 a0 1 S b1 DRGNPt 2 i 1 S fi DREX t 2 i 1 S Fi DATAR t 2 i 1 d EC t 2 i 1 t
n I51 n I 51 n i51
(1) (2)
DREX t 5 u0 1 S si DREX t 2 i 1 S di DRGNPt 2 i 1 S Fi DATAR t 2 i 1 j ECt 2i 1 mt
6
This is not an ideal measure of the average tariff rate since the ratio is sensitive to changes in the composition of imports as well as to changes in tariff rates. A more appropriate ratio would be that of revenue from the tariff on goods that could be produced domestically to the total of such imports. For each country, unit root tests indicate that on this measure the average tariff also was I(1) in level form and I(0) in rst difference. 7 The relevant data are also in Mitchell (1975).
238
J. Thornton / Economics Letters 55 (1997) 235 240
Table 2 Cointegration and Granger causality test results (a) Cointegration test results 1 Null hypothesis Denmark r50 r#1 Germany r50 r#1 Italy r50 r#1 Norway r50 r#1 Sweden r50 r#1 United Kingdom r50 r#1 (b) Granger-causality test results Causality DREXDRGNP DRGNPDREX DREXDRGNP DRGNPDREX DREXDRGNP DRGNPDREX DREXDRGNP DRGNPDREX DREXDRGNP DRGNPDREX DREXDRGNP DRGNPDREX
l 5% max 17.19 * 0.03 15.77 * 0.00 41.93 * 11.81 20.05 * 1.77 8.47 0.44 36.38 * 15.81
F-statistic
lmax 14.07 3.76 14.07 3.76 21.07 14.90 14.07 3.76 14.07 3.76 27.07 20.97
TRACE 17.22 0.03 15.77 * 0.00 59.45 * 17.52 21.83 * 1.77 8.92 0.44 60.01 23.63
TRACE 5% 15.41 3.76 15.41 3.76 31.52 29.68 15.41 3.76 15.41 3.76 47.21 29.68 t-statistic
Denmark Germany Italy Norway Sweden United Kingdom
1
f 12.8175 ** 1.5945 0.4447 3.0231 * 4.5595 * 0.0821
d 5.6988 ** 0.0225 0.2485 0.1250 2.0633 0.6380
F 0.7806 1 2.7414 20.2671 1.9230 2.4408 2.0243 0.1177 1.6032 0.2894 4.8928 * 0.1214 0.4861
EC t 21 3.8349 * 2.5099 * 3.0376 * 23.9767 * 0.6262 2.6850 * 22.4196 *
lmax is the maximal eigenvalue test statistic for at most r cointegrating vectors against the alternative of r 11 cointegrating vectors; TRACE is the stochastic matrix trace test for at most r cointegrating vectors. ** , * and 1 indicate signicance at the 1%, 5% and 10% levels, respectively.
where D is the difference operator, RGNP and REX are in natural logarithms, ATAR is the natural logarithm of the average tariff rate, t and mt are zero-mean, serially uncorrelated random error terms, and EC t 2i is the error-correction mechanism. In Eq. (1) causality implies DREX Granger-causing DRGNP, provided that some fi is not zero. Similarly, in Eq. (2) DRGNP is Granger-causing DREX if some di is not zero. Granger-causality is indicated in two ways. First, F-statistics are calculated under the null hypothesis that in Eqs. (1) and (2) all the coefcients of fi , di and Fi 50. Second, according to Granger (1988), independent variables cause the dependent variable if the error correction term in Eqs. (1) and (2) is statistically signicant. The F-statistics and the t-statistics on the error-correction term are reported in panel (b) of Table 2. Since the results from Granger-causality tests are sensitive to the selection of lag length, I follow the relatively common practice of presenting results from equations using the minimum nal prediction error suggested by Akaike (1969) to determine the appropriate lag length. Note that no error-
J. Thornton / Economics Letters 55 (1997) 235 240
239
correction term is included in the case of Sweden because of the lack of a cointegrating relationship between real exports and real GNP. The results show that developments in real GNP and exports in the 60 years or so prior to WW1 were closely related in the short-run as well as the long-run. They indicate unidirectional Granger-causality from real exports to real GNP in Italy, Norway, and Sweden; unidirectional causality from real GNP to exports in the case of the United Kingdom, and bi-directional causality between real exports and GNP in Denmark and Germany.
3. Conclusions This paper tested the relationship between exports and economic growth using data on real exports and real GNP for six European countries from around the mid-19th Century to WW1. The approach applied a test of cointegration as a pre-test strategy for Granger tests of causality between the two variables, whilst trying to control for trade liberalization. For each country, real exports and real GNP were I(0) in their rst differences; with the exception of Sweden, real exports and real GNP were cointegrated. Granger-causality was uni-directional, running from exports to GNP in Italy, Norway and Sweden; uni-directional from real GNP to exports in the United Kingdom, and bi-directional in Denmark and Germany. Thus, the expansion of exports was an integral part of the economic growth process in Europe in the second half of the 19th Century, and a leading contributor to growth in some countries.
References
Ahmed, J., Kwan, A.C.C., 1991. Causality between exports and economic growth. Economics Letters 37, 243248. Akaike, H., 1969. Fitting autoregressive models for prediction. Annals of the Institute of Statistical Mathematics 21, 243247. Bhagwati, J. and T. Srinivasan, 1979. Trade policy and development, in: R. Dornbusch and J. Frenkel, eds., International economic policy: Theory and evidence (Johns Hopkins University Press, Baltimore) pp. 1-35. Bohara, A.K., Kaempfer, W.H., 1991. A test of tariff endogeneity in the United States. American Economic Review 81, 952960. Chow, P.C.Y., 1987. Causality between export growth and industrial performance: Evidence from NICs. Journal of Development Economics 26, 5563. Enders, W., 1995. Applied Economic Time Series (John Wiley and Sons, New York). Engle, R.F., Granger, C.W.J., 1987. Cointegration and error correction: Representation, estimation and testing. Econometrica 55, 251276. Granger, C.W.J., 1988. Some recent developments in the concept of causality. Journal of Econometrics 39, 199211. Gardener, G.W., Kimbrough, K.P., 1989. The behavior of US tariff rates. American Economic Review 79, 211218. Grossman, G.M. and E. Helpman, 1991. Innovation and growth in the global economy (MIT Press, Cambridge). Helpman, E. and P. Krugman, 1985. Market structure and foreign trade (MIT Press, Cambridge). Hsiao, M.C.W., 1987. Tests of causality and exogeneity between exports and economic growth: The case of the Asian NICs. Journal of Economic Development 12, 143159. Johansen, S., 1988. Statistical analysis of cointegration vectors. Journal of Economic Dynamics and Control 12, 231254. Jung, W.S., Marshall, P.J., 1985. Exports, growth and causality in developing countries. Journal of Development Economics 18, 112.
240
J. Thornton / Economics Letters 55 (1997) 235 240
Krueger, A., 1980. Trade policy as an input to development. American Economic Review 70, 188292. Kwan, A.C.C., Cotsomotis, J., 1991. Economic growth and the export sector: China 19521985. International Economic Review 5, 105117. Maddison, A., 1995. Monitoring the World economy 18201992 (OECD Development Centre Studies, Paris). McKinnon, R., 1964. Foreign exchange constraint in economic development and efcient aid allocation. Economic Journal 74, 388409. Mitchell, B.R., 1975. European historical statistics (Macmillan Press, London). Thornton, J., 1996. Cointegration, causality and export-led growth in Mexico, 18951992. Economics Letters 50, 413416. Thornton, J., Molyneux, P., 1995. Macroeconomic determinants of tariff policy in a developing economy: Costa Rica, 196392. Journal of Post Keynesian Economics 17, 636644. Xu, X., 1996. On the causality between export growth and GDP growth: An empirical reinvestigation. Review of International Economics 4, 172184.
