AN ANALYSIS OF UK IMPORTS USING MULTIVARIATE COINTEGRATION†
| Author | Jeremy Smith,Nigel Sedgley |
| Date | 01 May 1994 |
| Published date | 01 May 1994 |
| DOI | http://doi.org/10.1111/j.1468-0084.1994.mp56002002.x |
OXFORD BULLETIN OF ECONOMICS AND STATISTICS, 56, 2(1994)
0305-9049
AN ANALYSIS OF UK IMPORTS USING
MULTI VARIAI 'E COINTEGRATIONt
Nigel Sedgley and Jeremy Smith
I. INTRODUCTION
Imports of manufactures are crucial to the overall balance of payments
position. In 1989, the current account deficit was £27bn, the largest ever, and
imports of manufactures totalled £96bn. Since 1970 a deterioration in the
UK's price competitiveness and relative efficiency has led to the UK share of
world manufacturing production, and the manufacturing share of total UK
employment, both declining. In fact, the import share of total demand for
manufactures in the UK has been rising since 1970, although it appears to
have levelled off in the last couple of years (see Figure 1). According to
Cuthbertson (1985) this long-term trend is due to a combination of increased
international specialization in production and an endemic weakness of the
UK manufacturing sector.
A familiar implication of the UK's poor trade performance is that it
imposes an external constraint on growth, as future domestic demand growth
will lead to a further deterioration of the trade account. This tendency for
imports to increase rapidly on the upswing of the cycle (see Figure 2) may
lead to further balance of payments or exchange rate crises. The quantitative
significance of this problem depends in part on the marginal propensity to
import.
This paper examines the existing import equations used by the HM
Treasury (HMT) and National Institute of Economic and Social Research
(NIESR) which were estimated in the late 1980's and compares these models
with one estimated using the Johansen (1988) multivariate cointegration
technique.
Section II looks at the recent history of the modelling of imports, in the
NIESR and HMT. Section 111 briefly introduces and discusses the data used
f The authors would like to thank Keith Church, Keith Cowling, Ken Wallis and an
anonymous referee for helpful suggestions. This paper is based on the University of Warwick
Msc dissertation of the first author. The views expressed in the paper are those of the authors
and are not necessarily those of Robert Fleming Securities.
Figures from the OECD suggest that between the periods 1960-67 and 1980-90 world
imports as a percentage of GDP rose from 12.1 percent to 18.7 percent.
© Basil Blackwell Ltd. 1994. Published by Blackwell Publishers, 108 Cowley Road, Oxford 0X4 IJF,
UK & 238 Main Street, Cambridge, MA 02142. USA.
136 BULLETIN
in this study. Section IV uses the Johansen procedure to estimate an imports
equation and tests for parameter restrictions, and, in particular, the hypo-
thesis of a unit long-run income elasticity of imports. Section V simulates the
model obtained in Section IV, and concluding remarks are offered in Section
VI.
II. COMPARATIVE MODEL PROPERTIES
Historically, studies of British import behaviour have discovered that either
the income elasticity of demand for imports is considerably greater than unity
and/or imports reflect a disconcertingly large positive time trend (see, for
example, Hotson and Gardiner (1983)). Import functions are traditionally
estimated in log-linear form and assume that demand depends on an activity
variable and relative prices; a time trend is often included to pick up auto-
nomous (non-price) effects (see Wilkinson (1992) for a recent example of
such a log-linear model estimated for Australia).2 The equation is written as,
ln(M) 6 + 6 ln(Z)+ 62 ln(RP,)+ 63 ln(S() + 64 ln(X,) (1)
where,
M real imports (aggregate or some sector)3
Z = activity variable (expenditure, activity, demand)
RP= measure of relative prices (imports to home produced)
S = Trend measure (often time)
X= Other variables (including a capacity utilization variable).
Total UK demand for a category of goods, such as manufacturing goods,
must be met either from imports or domestic output. Hence, there are three
relevant variables: Total Demand, demand for domestic manufactures, and
demand for imports. If two of them are modelled, the third will be a residual.
In the late 1980's the NIESR proposed an alternative functional form
based on Deaton and Muellbauer's (1980) Almost-Ideal-Demand-System
(AIDS). For a given total demand function log-linearity in the import demand
function implies a different, non-linear functional form for the demand for
domestic output. There is, however, no theoretical reason for treating these
two demands as different in character, and it would seem natural to choose
the same functional form for imports as domestic output, as below:4
(M/Z), = a0 + a1 ln(Z)+ a2 ln(RP)+ a3 ln(S,)+ a4 ln(X() (2)
(Y/Z)1= ß0 + ß1 ln(Z,)+ ß2 ln(RP1)+ ß3 ln(S,) +ß4 ln(X,) (3)
2 Hibble (1990) argues that as much as 99.4 percent of US trade is accounted for by multi-
national companies, in which case any analysis of imports ought to take into account the
decisions made about the location of production by these companies.
3Even concentrating on some sector of imports, such as manufacturing imports, involves a
high degree of aggregation across a heterogenous set of products: capital goods, intermediate
goods and consumer goods.
4i type of specification has also been used for specifying a manufacturing export
equation, see Anderton (1992).
© Basil Blackwell Ltd. 1994.
To continue reading
Request your trial