A NEW EVIDENCE ABOUT THE SPANISH EXPORT-LED
GROWTH PERIOD (1961-2000)
JACINT BALAGUER and MANUEL CANTAVELLA-JORDÁ
Department of Economics and International Economics Institute, Universitat Jaume I,
12071, Castellón, Spain
Abstract
In this paper, the Spanish export-led growth period is re-examined. For this purpose, it is
taken under consideration the change in the export structure which has been brought about
in the last four decades. The obtained results indicate that not only aggregate exports have a
positive impact on growth but also the allocation of productive resources towards
manufactured and semi-manufactured goods has become a key factor for Spain’s economic
development.
I. INTRODUCTION
A distinctive feature of the Spanish economy has been the rapid and
sustained growth of their exports during the last four decades. As we can see
in Figure 1,1 this phenomenon contrasts with the evolution of exports in the
precedent period in which the economy was characterised by a
protectionism and autarky situation. The process of growth in exports had
its origin in an outward-looking regime known as the Stabilisation and
Liberalisation Plan of 1959. Since then, Spain has carried out a continued
process of liberalising its economy finally leading to the integration of the
country in the European Union in 1986.2 The effects of a new open trade
regime and a policy of promoting exports on domestic real income was the
fundamental concerned dealt in Balaguer and Cantavella-Jordá (2001). In
that previous paper, the export-led growth hypothesis was supported for the
1
last four decades contrasting with the existing evidence for the preceding
period.
Figure 1
If the Spanish exports from the 60s are analysed one can observe a deep
change in its sectoral structure. During that period, several industrial sectors
have succeed in external markets.3 In general, Spain has gone from a
residual exports model, based on the country’s agricultural products surplus,
to a trade model focused on an outward-oriented demand of manufactured
and semi-manufactured products.
Table 1
As it can be seen in Table 1, exports of agricultural goods (mainly citrus
products, wine, vegetable oils and legumes) have gone from representing
half of country’s exports to a little more than the tenth part of total exports.
This gradual lost in the relative weight of traditional products reflects a
significant transformation of the productive specialisation towards semimanufactured goods, capital goods and consumption goods (especially
durable goods).4 Taking into account this evidence, it is interesting to
enquire whether the economic growth has been spurred by a general
increase of exports or by the allocation of resources coming from less
efficient export sectors to those more efficient.5
2
In this paper, the Spanish export-led growth hypothesis is re-examined for
the last four decades. The main objective is to obtain evidence about the
importance that the gradual change in the allocation of productive resources
from primary sectors to other economic activities has accomplished for the
country’s economic development. The movement of resources from one
sector to another has aimed at satisfying the external demand of goods with
greater added value. The empirical approach used in this analysis allows in a
simple way to capture the export growth effect caused by the distribution of
resources among different export sectors.
II. DATA, METHODOLOGY AND EMPIRICAL RESULTS
The empirical analysis uses annual data and the sample period spans from
1961 to 2000. This paper considers three variables: domestic income (Y),
exports (EX) and a variable in relative terms (EF). The latter captures the
allocation of resources towards exports of semi-manufactures, capital goods
and consumption goods (excluding agriculture products) (EF). The EF
variable has been constructed by including exports of consumption goods,
semi-manufactures and capital goods divided all by total exports. The
variables of the model are expressed in real terms. Logarithmic
transformation is used since it provides slope coefficients that can be
interpreted as elasticities. Time series data for each of the three variables are
collected from the Boletín Estadístico del Banco de España.
Table 2
Table 3
3
Before the long-run relationship between income, level of exports and the
structure term can be determined it is important to begin with a univariate
analysis (first step). Testing stationarity of time series leads to the
implementation
of the econometric model using the appropriate
methodology. The stationarity of the series was investigated by employing
the unit root tests developed by Dickey and Fuller (1979, 1981) and Phillips
and Perron (1988). According to the results shown in Table 2, each of the
series appears to be integrated of order one in levels (Engle and Granger,
1987). The second step is to test for cointegration among those three
variables using the Johansen’s (1988) methodology. The lag length of the
level vector autoregression system has been determined by minimising the
Akaike (1969) Information Criterion. Table 3 reports the cointegration
results. It can be observed that a significant and a single cointegrating vector
has been found. This means that income, level of exports and the structure
term for Spain move together in the long run between 1961 and 2000. They
are, therefore, causally related, at least in one direction (Engle and Granger,
1987).
Both MAX and Trace tests suggest that there is a cointegrating
relationship.6 It indicates that aggregate exports and the structure term
positively affect Spanish economic growth over time. These results show
that single-equation estimation for an increase in growth can capture the
long-run relationship. The existing correlation among the three variables is
not spurious.
4
From the estimates one can infer that the presence of multiplier effects on
income are really important. The interpretation of the elasticity of economic
growth with respect to exports should read as follows: a 1 per cent of a
sustained growth rate in total exports would imply an estimated increase of
about 0.22% percent domestic real income in the long run whereas a 1%
increase in the structure term would cause a 0.75% raise in income.
Table 4
In order to examine the robustness of the above results one is referred to
Table 4 which reports the outcomes of the parameter restriction tests for
economic growth, exports and export efficiency. All variables are
significant at 5 per cent significance level. It should be noted, then, that the
combination of those three variables in the model is important for obtaining
a reliable long-run relationship. A parameter restriction test for which the
null hypothesis is that LEX equals LEF indicates that the structure term has
a bigger impact that the overall exports on the Spanish economic growth.
The final step, then, should involve a causality test based on the results
given out of the long-run regression. The standard Granger causality test is
applied. The structure of this equation is the following:
5
n
n
n
i 1
j 1
k 1
DLY   0    i DLYt i    i DLEX t  j     i DLEFt i  RES t 1   t
m
m
n
i 1
j 1
k 1
DLEX   0    i DLEX t i    i DLYt  j     i DLYt i  RES t 1   t
n
n
n
i 1
j 1
k 1
DLEF   0    i DLEFt i   i DLEX t  j     i DLYt i  RES t 1   t
Where t, t and t are uncorrelated disturbances. In order to make valid
inferences on causality all these variables must be stationary. Thus, the first
differences of the three variables (DLY, DLEX and DLEF) and the residuals
(RES) obtained from the cointegrating vector, which contain long-run
information, are included in the Granger causality test structure. Again,
since the results derived from these tests may be sensitive to the selection of
the lag length, the minimum final prediction error suggested by Akaike
(1969) has been used.
Table 5
The results of Granger causality tests are presented in Table 5. The null of
no causality is rejected from LEX to LY and from LEF to LY. This would
confirm the adequacy of the long-run equation and would reinforce not only
the fact that exports cause economic growth but the allocation of resources
from the less efficient export sectors to the more efficient ones would also
have a multiplier effect on Spain’s economic development.
6
III. CONCLUDING REMARKS
The change in the productive specialisation stemming from the absorption
of external markets is, along with exports expansion, a key factor for the
Spanish economic development. This phenomenon indicates that the
resource allocation towards most industrialised export sectors has been
important for its economy.
On the other hand, the obtained estimates reveal that either the export
evolution and the sectoral structure change have originated a multiplier
effect on real income after the open-up period of the Spanish economy.
Moreover, the results show that the productive resources substitution from
the primary sector to the export industry of semi-manufactured and
manufactured goods have brought about a greater effect on economic
growth than the increase in total exports.
Therefore, it can be said that even in circumstances in which the evolution
of exports is kept constant, it is possible that the resource allocation towards
more efficient sectors generated by the external dynamics may cause
significant growth rates. In this sense, not only the export promoting policy
has a fundamental role in the economic growth but also it has to be
supported by a suitable industrial policy.
7
REFERENCES
Akaike, H. (1969) Fitting autoregrressive models for prediction, Annals of
the Institute of Statistical Mathematics 21, 243-247.
Alcaide, J. (1976) Una revisión urgente de la serie de renta nacional en el
siglo XX, in Datos Básicos para la Historia Financiera de España,
1850-1975, Instituto de Estudios Fiscales, Madrid, 1127-1150.
Balaguer, J. and Cantavella-Jordá (2001), Examining the export-led growth
hypothesis for Spain in the last century, Applied Economics Letters,
8, 681-685.
Carreras, A. (1989) La renta y la riqueza, in Estadísticas Históricas de
España, siglos XIX-XX, Fundación Banco Exterior, Madrid, 533588.
Dickey, D. A. and Fuller, W. A. (1979) Distribution of the estimators for
autorregressive time series with a unit root, Journal of the American
Statistical Association, 74, 427-431.
Dickey, D. A. and Fuller, W. A. (1981) Likelihood ratio statistics for
autoregressive time series with a unit root, Econometrica, 49, 10571072.
8
Engle, R. F. and Granger, C. W. J. (1987) Cointegration and error
correction: representation, estimation, and testing, Econometrica, 2,
251-276.
Feder, G. (1982) On exports and economic growth, Journal of Development
Economics, 12, 59-73.
Johansen, S. (1988) Statistical analysis of cointegration vectors, Journal of
Economic Dynamics and Control, 12, 231-54.
Osterwald-Lenum, M. (1992) A note with quantiles of the asymptotic
distribution of the maximum likelihood cointegration rank test
statistics, Oxford Bulletin of Economics and Statistics, 54, 461-472.
Moreno, L. (1997), The determinants of Spanish industrial exports to the
European Union, Applied Economics, 29, 723-732.
Phillips, P. and Perron, P. (1988) Testing for a unit root in time series
regression, Biometrica, 75, 333-346.
Prados de la Escosura, L. (1986) Una serie anual de comercio exterior
español (1821-1913), Revista de Historia Económica, 1, 103-105.
9
Tena, A. (1985) Una reconstrucción de comercio exterior español, 19141935: la rectificación de las estadísticas oficiales, Revista de Historia
Económica, 1, 77-119.
Tena, A. (1989) Comercio Exterior, in Estadísticas Históricas de España
Siglos XIX-XX. Madrid, Fundación Banco Exterior, 329-361.
10
Table 1. Sectoral structure of exports (percentages)
Year
1961
1970
Sector
Agricultural goods*
53.76
34.85
1980
1990
2000
17.86
11.61
11.42
Energy products
5.91
5.49
3.95
4.84
3.62
Consumption goods
11.91
21.38
25.54
25.84
28.05
Semi-manufactures and
capital goods
28.42
38.28
52.65
57.71
56.91
Source: Own calculations from Boletín Estadistico del Banco de España
* Agricultural goods include livestock
11
Table 2. Augmented Dickey-Fuller and Phillips Perron Unit Root Tests (levels and first
differences)
Variables
 (ADF)
(ADF)
(ADF)
 (PP)
(PP)
(PP)
LY
-1.47
-1.79
-0.29
-1.62
-1.83
-0.73
LEX
-1.17
-1.08
1.73
-0.77
-1.82
0.93
LEF
-2.32
-2.19
-1.80
-1.29
-2.27
-1.12
-3.75
-2.81
-2.55
-3.28
-3.73
-3.18
LY
-4.96
-3.90
-2.95
-4.63
-4.21
-3.49
LEX
-8.38
-4.93
-3.88
-8.26
-7.61
-6.47
LEF
Critical
values
-3.19/
-2.60/
-1.62/
-3.19/
-2.60/
-1.62/
(10%, 5%)
-3.53
-2.94
-1.95
-3.53
-2.94
-1.95
ADF stands for Augmented Dickey Fuller ; PP for Phillips Perron;  represents the most
general model with a drift and trend;  is the model with a drift and without trend;  is the
most restrictive model without a drift and without trend .
Numbers in brackets are number of lags used in the ADF test in order to remove serial
correlation in the residuals. When using PP test, numbers in brackets represent the
truncation lag parameter.
Symbol  represents the corresponding variable in first differences.
Tests for unit roots have been carried out on E-VIEWS 3.0.
12
Table 3. Johansen Maximum Likelihood Cointegration Tests (1961-2000)
r: number of
Trace
Critical values
MAX
cointegrating vectors
95% and 90% (MAX)
(null hypothesis)
Critical values
95% and 90% (Trace)
r=0
23.01*
33.26*
20.97/18.60
19.96/17.85
r1
7.23
10.25
14.86/12.07
9.24/7.52
r2
3.01
3.01
3.76/2.68
3.76/2.68
Parameter Estimates (normalised)
Cointegrating vector
Variables
LY
-1
LEX (+)
0.22
LEF (+)
0.75
MAX and Trace are the likelihood ratio statistics for the number of cointegrating vectors.
Estimation has been performed with MICROFIT 3.0.
* Indicates significance at 5 percent level, critical values are based on Osterwald-Lenum
(1992).
In the second panel,Y denotes domestic income, TOUSA, tourism, and q, real effective
exchange rate. In brackets are the expected signs for TOUSA and q. Parameter estimates
express different elasticities.
13
Table 4 Tests of Parameter Restrictions on LY, LEX and LEF
Chi-squared
test statistic
13.27
(Y)
Chi-squared
test statistic
12.63
(LEX)
Chi-squared
test statistic
11.86
(LEF)
The critical value with one degree of freedom at 5 per cent significance level is 3.84
Second Panel H0: LEX=LEF
Chi-squared
test statistic
H1: LEX<LEF
8.91
The critical value with one degree of freedom at 5 per cent significance level is 3.84
14
Table 5. Granger causality tests
Null hypothesis
F-Statistic
Probability
LEX does not Granger cause LY
4.21
0.047
LY does not Granger cause LEX
0.47
0.495
LEF does not Granger cause LY
3.66
0.063
LY does not Granger cause LEF
0.93
0.341
LEF does not Granger cause LEX
1.05
0.311
LEX does not Granger cause LEF
0.12
0.723
Tests for causality have been carried out on E-VIEWS 3.0.
15
19
01
19
04
19
07
19
10
19
13
19
16
19
19
19
22
19
25
19
28
19
31
19
34
19
37
19
40
19
43
19
46
19
49
19
52
19
55
19
58
19
61
19
64
19
67
19
70
19
73
19
76
19
79
19
82
19
85
19
88
19
91
19
94
19
97
6000
5000
4000
3000
2000
1000
0
Figure 1. Exports (discontinuous line) and real income index (continuous line); 1901 base year.
16
1
The domestic income series up to 1953 is obtained from the estimation and review made
by Alcaide (1976). The exports series up to 1913 is drawn from the estimation made by
Prados de la Escosura (1986); from 1914 to 1935 the exports series stems from Tena’s
review (1985). A collection of real domestic income and real exports as well as a lengthy
comment on the construction of those series can be found in Carreras (1989) and Tena
(1989). The more recent data of both series are collected from Contabilidad Nacional de
España (Boletín Estadístico del Banco de España).
2
A very important change of strategy came about in Spanish trade policy with the
Stabilisation and Liberalisation Plan of 1959. This policy was characterised for the
liberalisation of imports leading to the acquisition of the necessary capital goods for the
economy’s productive expansion. Moreover, it was established the golden parity of peseta
at the International Monetary Fund and at the same time the decision of a devaluation from
42 pesetas per dolar to 60 pesetas per dolar was taken. Although the implementation of the
plan was an important starting point, the exports takeoff should be placed between 1967
and beginning of 70s. Three more factors influenced the initial boost: first, a new
devaluation of domestic currency; second, a great expansion of international trade and
third, the gradual improvement and adaptation of export promoting programmes. The
establishment of the former Export Promoting Institute in 1982 (currently, External Trade
Institute) reflects a continued willingness of applying a policy of increasing exports. When
Spain joined the European Union in 1986 its trade policy moved from a cooperation
scenario to an integrated status with the rest of the member countries. The fundamental
feature of this situation has been the lost of sovereignty on the Spanish trade policy.
3
Moreno (1997) deals with the determinants of Spanish exports for different industrial
sectors. His paper shows evidence about the fact that general technological and advertising
effort have a significant effect on the evolution of exports of each one of sectors in the
1978-1989 period.
4
In the beginning of the period the scarce production of cars was oriented to the domestic
market, nevertheless during the last decade the car exports represented almost 60% of total
consumption exports (without including agricultural products).
5
Feder (1982) argues that growth can be boosted not only through the increase of agregate
levels of labour and capital but also through the allocation of resources from the least to the
most efficient sector.
17
6
Trends in the variables and in the DGP were allowed for in estimation, although the
results were not practically sensitive to estimation method.
18