Professional Documents
Culture Documents
Economics
ELSEVIER
Abstract
In this paper robust elasticity estimates of coal demand for China are derived using annual
data 1953-92. In so doing, we illustrate the use of a powerful yet practically convenient and
recently developed modelling procedure devised by Stock and Watson (known as Dynamic
OLS (DOLS)), who provide evidence, based on Monte Carlo simulations, of this estimator
being superior in small samples compared to a number of alternative estimators, as well as
being able not only to accommodate higher orders of integration but also to account for
possible simultaneity within regressors of a potential demand system. Furthermore, cointegration and error-correction methods are employed to derive short-run price and income
elasticities. Estimated results are quite robust not only in terms of statistical prowess but
also in terms of economic intuition and indicate that, over the long run, both price and
income elasticities are close to unity. While short-run price and income elasticities are less
(in absolute value) than their long-run counterparts, there seems to be some divergence in
short-run parameters from a subsample analysis. Overall, results seem to imply that for
China, coal consumption should remain relatively constant as future modernization strategies for economic development are pursued. In addition, the study has clear methodological
implications for estimating the long- and short-run elasticities in a demand function in
general, and in a wide variety of fields in future applied research.
* Corresponding author. E-mail: a-masih@adfa.oz.au
0140-9883/96/$15.00 Cc) 10% Elsevier Science B.V. All rights reserved
P I I S 0 1 4 0 - 9 8 8 3 ( 9 6 ) 00016-3
316
1. Introduction
There have now been several studies devoted to the understanding of the
relationship between macroeconomic variables on the one hand and resource
related variables on the other (see Harvie and Hoa, (1993)). Some have attempted
to study the macroeeonomic effects arising from a variety of resource related
shocks (examples include Pindyck (1980) on effects of the first oil price shock,
Gregory (1976) for the Australian resource boom, Eastwood and Venables (1982)
on reaction from resource hikes in open economies). Others have concentrated on
the specification or modelling approach of energy demand in deriving elasticities
(see Zilberfarb and Adams (1981) for income elasticity estimates of energy consumption using cross-country data for developing countries; Dunkerley (1982) on
estimating energy demand for a selection of developing countries; Hoa (1992) for a
multivariate simultaneous approach in modelling Thai energy demand; Bentzen
and Engsted (1993) in estimating short- and long-run Danish energy demand
elasticities; Engsted and Bentzen (1993) on a rational expectationist approach to
Danish energy demand, and Bentzen (1994) in analysis of Danish gasoline demand).
Related to this is a wide body of literature investigating the causal relationship
between energy consumption and economic growth (see Kraft and Kraft (1978) on
the seminal study finding evidence in support of unidirectional causality from
income to energy for the US; Akarca and Long (1980) who contested these
findings; Abosedra and Baghestani (1991) who re-opened the debate with findings
in support of Kraft's and Kraft's initial study; Stern (1993) who uses an ordinary
vector-autoregression (VAR) with variables in first-difference form on US data
(1947-90) finding evidence against the hypothesis that gross energy Granger causes
GDP, and Masih and Masih (1996c) in an investigation of bivariate causality
between energy consumption and economic growth for six Asian LDCs).
Modelling energy and economic related variables, however, is often associated
with problems of a practical nature which have serious consequences for inference
making based on the estimates. First, apart from the standard problems of
departures from classical assumptions of the regression framework, reduced form
equations typically suffer from endogeneity of regressors leading to simultaneity
among the variables in the system. Secondly, the variables used in such studies are
usually very aggregated and exhibit non-stationarity (see Nelson and Plosser (1982)
and Perron (1988) on evidence of non-stationarity in several macro-aggregates) but
this has been remedied by applications of cointegration techniques (Engle and
Granger, 1987) and developments in error-correction modelling. Thirdly, the small
sample nature of data-sets being employed in energy economics (due mainly to
317
lOne of the problems associated with employing cointegration techniques is that its power is vitally
dependent upon the actual length of time that the sample period spans. This has been supported in
Monte Carlo studies conducted by Shiller and Perron (1985) and Hakkio and Rush (1991). For purposes
of infcrence making, however, standard tests of hypothesis in regression models are usually associated
more favourably with data on a high frequency, i.e. large sample and asymptotically distributed.
However, as is often the case with long-spanning data, particularly in aggregated variables used ill
energy analysis, only annual data offering a small number of obsevvations can be relied upon to be
available on a consistent basis.
318
these variables over a long time horizon to which the techniques employed in this
paper adequately cover. To this extent, this study helps to fill this void.
The paper is organized in the following manner. Section 2 contains a brief
overview of China's coal endowment, consumption and contribution in economic
growth; a brief and intuitive account of the statistical methodology employed is
provided in Section 3 prior to application of methods and a discussion of results in
Section 4. Some broad policy implications, conclusions and a summary of contributions this study had for future research are made in Section 5.
Among all the alternative types of China's resource endowment, coal dominates
by far; so much so that in 1989, it accounted for about 74% of its aggregate energy
production. The coal reserves lie both to the north and southern provinces (Tang
and La Croix, 1993), though its quality is superior in the northern reserves.
According to Tang and La Croix (1993), there are three key features which
determine the structure of China's total energy and coal potential: coal is predominant in China's energy consumption, accounting for almost all (95%) of the
country's primary energy consumption during the 1950s; before declining over the
next two decades and rising again in the 1980s due to too heavy dependence on oil;
while coal endowments have helped in making China a self-sufficient energy
consumer, it has never been a serious candidate for a potential export earner; even
though aggregate energy consumption in China is high in world rankings, it is very
low in terms of per capita consumption (only one-third of the world average).
The model used in this analysis is dictated by the typical formulation postulated
by economic theory for aggregate energy demand functions, in its double-log form
as~
logC t = c~ + / 3 ( l o g P t) + 6(logYt) + v~
(1)
319
1.5
11.5
1.4
11
1.3
10.5
lO
9.5
Co~ C o n ~ m p t m ~ e f t
, ,',
,'1
1.2
1.1
0.9 0
,"''-
Scale
~ ~ . - - "
~l~o~P~e~lgn(~e)
0.8 n"
0.7
8.5
0.6
I I I I I [ I [ : 1 : : l : : I I I I I I I I I I I I I I I I I I I I I I
Fig. l.
0.5
AggregatecoalconsumptionandrealcoalpriceinChina:1953-92.
6.5
11.5
5.5
11
10.5
4.5
lO
" ,~ / . ' ~
9.5
3.5 !
2.5 ~
1.5
8.5
0.5
Fig. 2.
Aggregate
coal consumption
in C h i n a :
1953-92.
The data used in this analysis are annual ranging from 1953 to 1992, and were
obtained from historical annual statistics reported in recent issues of the China
Statistical Yearbook. Plots of aggregate coal consumption compared to real price of
coal and real income are presented in Figs. 1 and 2.
320
3See Masih and Masih (1995a, 1996b) for an investigation of the dynamics of economic activity within a
multivariate cointegrated system. Bivariate Granger causality tests using cointegration techniques have
also been undertaken in Masih and Masih (1995c).
4A treatment of the sequential steps involved in applying the PP tests appear in Taylor (1993). Basically,
Perron (1988) shows that if a time series in trend stationary and if no account is made of this in
implementing the testing procedure, this may lead to high probabilities of making a type 11 error. While
the precise form of the assumptions (with regard to distributional properties of error terms, etc) is
contained in Perron (1988), the following sequence is suggested:
(i) Apply Z(t~,), Z ( ~ 2) and Z ( ~ 3) respectively and if the unit root hypothesis is rejected we should
halt the procedure here.
(ii) If the unit root hypothesis cannot be rejected then the greatest power may be obtained by
estimating equations associated with the Phillips-Perron transformations of the relevant t- and
F-statistics, Z(t~,) and Z(qb~). Due to the fact that these two tests are not invariant to the constant
term, this is only valid if the drift term (/x*) used in test equations applied in (i)was zero. In this
respect these two tests should only be used if Z(q~ 2) cannot be rejected.
321
r/(u) = ( 1 / T 2) E S ~ / - ~
where
t=l
S t = ~_,v i , t =
1.... T
(2)
i-I
5This guideline in considering the stochastic properties of univariate time series is also used in an
empirical analysis containing error-correction modellingby Mehra (1994).
322
Ax, =
r, axt_
+ nx,_
(3)
i=1
where X t is a column vector of the m variables, F and H represent coefficient
matrices, A is a difference operator, k denotes the lag length, and 6 is a constant.
If II has zero rank, no stationary linear combination can be identified. In other
words, the variables in X t are non-cointegrated. If the rank r of H is greater than
zero, however, there will exist r possible stationary linear combinations and H may
be decomposed into two matrices a and /3 (each m r) such that H = a/3'. In
this representation b contains the coefficients of the r distinct cointegrating
vectors that r e n d e r / 3 ' X t stationary, even though X t is itself non-stationary, and a
contains the speed-of-adjustment coefficients for the equation, v
3.2.3. Estimation o f long-run equilibria: Stock-Watson dynamic O L S
Stock-Watson (1993) Dynamic OLS (DOLS): B = [c, a,/3]', X = [1, Pt, Yt]
6While applications of the JJ procedure have been quite popular in a multivariate context, results
arrived from JJ statistics in bivariate studies have also been shown to be more robust than those arrived
adopting the Engle Granger approach (see, by example, Masih and Masih, 1994, 1995b).
7With respect to both the Engle-Granger and JJ approach, it is important to acknowledge that should
non-cointegration not be rejected at conventional significance levels, it is possible that the residual term
may display fractional behaviour and still be mean-reverting implyingfractional cointegration. For such
an approach see Masih, R. and Masih, A. (1995a).
C, = B ' X t +
323
j=K
~_, ~jAp,_j +
j=-J
~
j=
A/AYt_ / + ~,
K
In estimating the long-run parameters of the demand function, we adopt the DOLS
procedure which basically involves regressing any 1(1) variables on other I(1)
variables, any I(0) variables and leads and lags of the first differences of any I(1)
variables. However, since an investigation of the short-run dynamics are also of
interest in our analysis and important for several other factors of modelling, the
standard VECM formulation described in the later section will also be employed in
facilitating our inferences regarding the short run.
3.2.4. Short-run dynamics: vector error-correction modelling (VECM)
Engle and Granger (1987) demonstrated that once a number of variables are
found to be cointegrated, there always exists a corresponding error-correction
representation which implies that changes in the dependent variable are a function
of the level of disequilibrium in the cointegrating relationship (captured by the
error correction term) as well as changes in other explanatory variable(s).
Assume that we are dealing with non-stationary processes and consider a model
appearing in first differences in a VECM form given by (2). Then by the JJ method
we may deduce the number of cointegrating vectors (using both the trace and
maximum eigenvalue tests) and the /3' vector estimates. The long-run constraints
implied by the estimated cointegrating vectors are imposed by way of constraints in
the simple VAR model (with variables appearing in first-differences) via lagged
error-correction terms.
Model 2
General vector error-correction specification with B = [c, o~,/3 ]', X = [1, P,, Y~]
j=k
j=l
xc, = E 4,j xc _j + E
j=l
j=m
jAe, j +
j=0
q- ~
ajar, j
j=0
~ i ( C t - i - B ' X t _ I ) + 6.t
i=1
Having described what constitutes our VECM, we should place the contributions
of this model in an intuitive context. When the variables are cointegrated, then in
the short term, deviations from this long-term equilibrium will feed back on the
changes in the dependent variable in order to force the movement towards the
long-term equilibrium. If the dependent variable (i.e. say, coal consumption) is
driven directly by this long-term equilibrium error, then it is responding to this
feedback. If not, it is responding only to short-term shocks to the stochastic
environment. The significance tests of the 'differenced' explanatory variables give
us an indication of the 'short-term' effects, whereas the 'long-term' causal relationship is implied through the significance or otherwise of the 't' test(s) of the lagged
324
325
Table 1
Kwiatkowski-Phillips-Schmidt Shin (KPSS) tests for mcan stationarity and Phillips Perron (PP) tests
for unit roots a
KPSS
Phillips Perron
Variable
7( Ix)
0.804**
0.659**
1.017"*
- 1.745
-0.637
-0.394
(-4.018)
( 8.614)
(-4.460)
-2.864
-2.747
-2.087
( 4.025)
( 8.915)
(-4.491)
Yt
0.915"*
0.902**
1.148"*
- 1.741
- 0.868
-0.849
(-3.578)
( - 5.497)
(-4.879)
- 1.818
- 1.913
-1.900
( - 3.898)
( - 5.401)
(-4.809)
0.799**
0.857**
0.983**
-0.216
-0.299
- 1.383
(-4.234)
(-6.878)
( 3.419)
-2.718
-2.191
- 1.926
( 4.073)
(-7.921)
( - 3.945)
Pt
Z(t )
Z(t ,~ )
aThe KPSS test statistic tests the null hypothesis that the variable in question is mean stationary (the
5% critical value for r/(/x) is provided in Kwiatkowski et al. (1992, p. 166, Table 1)). Figures presented in
parentheses for PP tests refer to the adjacent test carried out on the variable in first-differenced (A)
form. Z(t,) is the Phillips-Perron (PP) test allowing for a drift term, whereas Z(t,,) is the PP test
allowing for a drift and a deterministic trend. The null hypothesis is that the variable under consideration contains a unit root in its autoregressive representation. The 5% critical values for a sample size of
50 pertaining to Z(t,) and Z(t~,) are -3.45 and -2.89, respectively. ** indicates rejection of the null
at the 5% level of significance.
9For a more detailed discussion of the JJ procedure than that presented in this paper, see Cuthbertson
et al. (1992), and a wide-ranging survey by Muscatelli and Hum (1992).
326
Table 2
J o h a n s e n and Juselius's tests for multiple c o i n t e g r a t i n g vectors u
Vector
H~
Hypotheses
H~
Test statistics
Max E i g e n v a l u e ( A )
Trace
[c,, g,Y~]
Panel B: S u b - S a m p l e (1953-1969)
[C,, P,, Y,]
r = 0
r _< 1
r _< 2
r > 0
r > 1
r
3
27.657**
11.204
3.182
42.043**
14.386
3.182
r = 0
r < 1
r < 2
r > 0
r > 1
r = 3
38.567**
7.182
0.878
46.627**
8.059
0.878
r
0
r < 1
r _< 2
r > 0
r > 1
r = 3
21.225"*
9.286
2.509
33.017"*
11.795
2.508
[C,,Pt,Y~]
stochastic trends underlying this set of variables associated with the demand
system.
327
Table 3
S t o c k - W a t s o n dynamic OLS long-run parameter estimates of coal demand a
S t o c k - W a t s o n dynamic OLS (DOLS): B = [c, a , / 3 ] ' , X = [1, P~, Y~]
j=l
j=l
C, = B'X, + ~
j=
~7,AP,_i + ~
1
j=
hiaY~_ i + ;,
1
Variable
Const
6.6179* **
(0.1711)
-0.9914"*
(0.4956)
1.0743* * *
(0.1113)
0.8741'
(0.5216)
-0.4159'**
P~
Y~
A PI
AY~
(0.1184)
APt+ 1
A Pt I
AYt 1
AYt+ L
0.3166
(0.4094)
0.6193
(0.3509)
0.2760***
(0.0986)
0.7225**
(0.3137)
0.0933
0.9455
328
Table 4
Vector error correction models of coal demand a
General vector error-correction specification with B = [c, a , / 3 ]', X = [1, P,, Y~]
j=k
j=!
j. + Y'. ~/,e,
Variable
Full sample
Constant
6.6179* * *
(0.1711)
0.5722***
(0.1178)
- 0.2125 "~
(0.1344)
-0.4243***
(0.1508)
-0.1928"*
(0.1012)
1.1875* * *
(0.1327)
- 0.5989"**
(0.1908)
0.1871"**
(0.0630)
-0.8296***
0.5886***
ACr l
APr
Ap~ i
APt 2
A y~
A Yt- t
'~t I
Price elasticity
Income elasticity
Adjustment
Length (yrs)
5.345
Statistic/test
SSR
1.3651
RZ-adjusted
0.8105
Serial correlation
LMI[ X2(I)]
2.431
LM2[ X 2(2)]
1.759
LM4[ X 2(4)]
1.479
Functional form
R E S E T [F(m,n)] 0.410
Heteroscedasticity
Het [ XZ(I)]
1.482
A R C H [ X2(1 )]
0.059
Normality
J a r q u e - B e r a [ X2(21].998
j=m
1.2923"* *
(0.1801)
- 0.5578"*
(0.2827)
-0.2122"*
(0.1026)
-0.8221"
0.7346***
4.7116
Summary of diagnostics
B'X,
,) + ~,
Second subsamplc
0.0211"
(0.0127)
0.6340
(0.1813)
- 0./)363
(0.0490)
-0.1415"
(0.0775)
0.0836
10.0821)
0.7048* * *
(0.1 t33)
- 0.6479* **
(0.1631 )
-0.3957***
(0.1135)
-0.2558*
0.0569***
2.5272
0.0897
0.8723
0.0038
0.7057
2.238
0.554
0.257
2.161
2.234
1.179
0.4672
0.992
0.965
0.494
0.384
0.042
2.524
1.241
~'The lag structure was specified using a general to specific approach. SSR refers to the sum of squared
residuals. Distributional properties of diagnostics are respectively: LM(I), LM(2) and LM(4) tests for
the null of 1st, 2nd and 4th order serial correlation amongst the residuals; Het: a test based on
regression of squared residuals on a constant and squares of the fitted values; A R C H : a test for
first-order autoregressive conditional heteroskedastic effects; RESET: Ramsey's REgression Specification E r r o r / ' - t e s t with (m, n) degrees of freedom; and the J a r q u e - B e r a X2(2) L M test for normality of
residuals. *** ** and * indicate significance at the 1, 5 and 10% levels.
329
that the estimated model is stable over the sample period. This result tends to
suggest that the impact of any structural change over the entire sample period,
although could be evident, at least in terms of model stability, does not appear to
be significant.
There are three noteworthy results these estimates have to offer. The first is in
terms of the importance of the real price of coal in explaining coal demand, in
terms of expected sign and statistical significance. The picture these results portray
seems to be that all variables influence demand in the short term. Prices are quite
important and this result seems to hold regardless of whether income variables are
present or not. The second interesting result is through the statistical significance
and magnitude of the error-correction term. This term (since it appears in its
levels) indicates that real prices and income do, as a component of the long-term
cointegrating relationship through the lagged error-correction term, jointly influence demand over the long term. Thirdly, most intuitively, the error-correction
term is significant with an adjustment coefficient of -0.1871, indicating that in the
case we are off the long-run demand curve, overall demand adjusts to its long-run
equilibrium level with about 18.7% of the adjustment taking place within the first
year (for a detailed interpretation of the significance of the error-correction term
in linear time-series relationships in the presence of model instability, see R.
Masih, (1996)). In other words, given there is a displacement off the long-run
demand curve, the system will take just over five consecutive years to restore a new
equilibrium. In terms of actual time, this is quite slow indicating that prices are
quite sluggish to adjust.
This also implies that coal consumption bears the brunt of any shock to the
demand system. The sign of the ECT coefficient also indicates that changes in the
demand adjust in an opposite direction to the previous period's deviation from
equilibrium. Had the error-correction term not been significant this would imply
that in this estimated model, any short-run adjustment to long-term equilibrium is
primarily through the other variables in the system and not through the channel of
coal consumption.
Finally, as one of our base goals of this analysis, the error-correction model
estimates provide a quantitative assessment of the short-run price and income
elasticity of demand. Given that we are modelling a simple linear relationship with
coal consumption as the dependent variable, the ECM provides a price elasticity
estimate of about -0.8296 for the whole sample period. This result implies that,
although at least in the short run, aggregate demand for coal should decline if the
average price per ton of coal increases, but the decline in quantity demanded is less
than proportionate. Income elasticity estimates are not unlike the short-run price
elasticity estimate, being significantly less than unity in absolute value, although
associated with the expected sign.
In order to assess to what extent there may have been a shift in the short-run
response of coal consumption to changes in prices and income, VECMs were also
specified and estimates over two subsamples incorporating the years 1953-69 in
one sample and 1970-92 in the other. This subsample analysis will also provide us
330
with further insights into whether there was any significant change in the adjustment parameter.
The sub-sample results are presented alongside the full-sample estimates in
columns 3 and 4 of Table 4. Once again, most coefficients seem to be significant
and most joint lags of price and income are associated with expected signs. A
battery of diagnostic checks does not tend to reveal any violations from standard
error assumptions or model specification. Comparing both subsamples, results for
the initial period seem to be strikingly similar to the estimates derived for the
model over the full sample period, with short-run price and income elasticities of
-0.8221 and 0.7346 respectively. The adjustment coefficient is - 0 . 2 1 1 2 indicating
that it takes about 4.7 years for demand to revert back to a new equilibrium
following an exogenous shock. These results, however, are quite in contrast to
those arrived at after estimating a VECM over the second subsample which
provides a short-run price elasticity of -0.2558 and a short-run income elasticity of
close to zero. Furthermore, the adjustment coefficient is larger (in absolute value)
in the latter sample than in the earlier period, implying a quicker adjustment
length of approximately 2.5 years.
331
displacement from the long-run demand curve, seemed to be slightly quicker over
the 1970-92 period than the 1953-69 period.
From this empirical analysis we may draw the following conclusions which may
be of some assistance to economic policy-makers and China's economic and
resource-use plans. First, the evidence of long-run elasticities being close to unity
implies that aggregate intensity of coal consumption should remain constant as
future economic development takes place in China. Secondly, the low short-run
elasticities over the latter half of the sample period could be justified by the fact
that the rate of overall energy consumption during the 1980s, in particular, was
much lower than the rapid use of energy to foster China's economic development,
experienced during the 1950s and 1960s. Third, although there appears to be a
divergence of short-run elasticities over the two subsample periods, the overall
relationship was found to be quite robust as well as stable.
Given that coal dominates China's share of natural resource endowment and the
availability of adequate published price data, this analysis was limited to estimating
parameters involving a particular component of China's energy endowment. Moreover, a simple linear relationship was specified dictated by very fundamental
theoretical postulates. Time-series data permitting, an interesting exercise would
be to conduct similar analyses for deriving elasticities of China's other three major
conventional energy resources: oil, natural gas and hydropower. This would provide
a much broader and general overview of energy demand behaviour in China and
also contribute valuable insights into the effectiveness of China's energy price
reforms, especially in the light of its recent modernization strategies. More importantly, such analysis would also hold implications for Chinese total energy consumption and conservation policies in the prospective future.
Acknowledgements
Authors would like especially thank: Jim Stock, Soren Johansen, Jesus Gonzalo,
F. Gerard Adams, Katarina Juselius, Deane Terrell, Sean Holly, Warwick
McKibbin and Adrian Pagan for their very helpful comments and discussions; and
Sam Jegatheswaran who provided most able research assistance. The second
author would also like to acknowledge financial support provided through a
University of New South Wales Special Research Grant. Any remaining errors
a n d / o r omissions are attributable solely to the authors who also equally share
responsibility to the paper.
References
Abosedra, S. and H. Baghestani, 1991, New evidence on the causal relationship between United States
energy consumption and gross national product, Journal of Energy and Development,14 285-292.
Akarca, A.T. and T.V. Long, 1980, On the relationship between energy and GNP: a re-examination,
Journal of Energy and Development,4 326-331.
Bentzen, J., 1994,An empirical analysis of gasoline demand in Denmark using cointegration techniques,
Energy Economics, 16 139-143.
332
Bentzen, J. and T. Engsted, 1993, Short and long run elasticities in energy demand: a cointegration
approach, Energy Economics, 15 9-16.
Campbell, J.Y. and P. Perron, 1991, Pitfalls and opportunities: what macroeconomists should know
about unit roots, in O.J. Blanchard and S. Fischer (eds.), NBER Macroeconomics Annual 1991, MIT
Press, Cambridge, MA.
Cuthbertson, K., S.G. Hall and M.P. Taylor, 1992, Applicd Econometric Techniques, Simon and
Schuster, University of Michigan Press, Ann Arbor, MN.
Dickey, D. and W. Fuller, 1979, Distribution of the estimators for autoregressive time-series with a unit
root, Journal of the American Statistical Association, 74 427-431.
Dickey, D. and W. Fuller, 1981, Likelihood ratio statistics for autoregressive time series with a unit root,
Econometrica, 49 1057 1072.
Dickcy, D. and S. Pantula, 1987, Determining the order of differencing in autoregressivc processes,
Journal of Business and Economics Statistics, 5 455-461.
DeJong, D.N., J.C. Nankervis, N.E. Savin and C.H. Whiteman, 1992, Integration versus trend stationarity in time series, Econometrica, 60 423 433.
DeVany, A. and W.D. Walls, 1993, Pipeline access and market integration in the natural gas industry:
evidence from cointegration tests, Energy Journal, 14 1-19.
Dunkerley, J., 1982, Estimation of energy demand: the developing countries, Energy Journal, 3 79 99.
Eastwood, R.K. and A.J. Venables, 1982, The macro-economic implications of a resource discovery in
an open economy, Energy Journal, 92 285-299.
Engle, R.F. and C.W.J. Granger, 1987, Cointegration and error correction: representation, estimation,
and testing, Econometrica, 55 251 276.
Engsted, T. and J. Bentzen, 1993, Expectatidns, adjustment costs, and energy demand, Resources and
Energy Economics, 15 371 385.
Granger, C.W.J., 1986, Developments in the study of cointegrated economic variables, Oxford Bulletin
of Economics and Statistics, 48 213-228.
Granger, C.W.J., 1988, Some recent developments in a concept of causality, Journal of Econometrics, 39
199 211.
Government of China, China Statistical Yearbook (Annual, various issues).
Gregory, R.G., 1976, Some implications of the growth of the mineral sector, Australian Journal of
Agricultural Economics, 20 1-91.
Hakkio, C.S. and M. Rush, 1991, Cointegration: how short is the long run'? Journal of International
Money and Finance, 10 571-581.
Hareic, C. and T.V. Hoa, 1993, Long-term relationships of major macro-variables in a resource-related
economic model of Australia: a cointegration analysis, Energy Economics, 15 257-261.
Hendry, D.F., 1986, Econometric modelling with cointegrated variables: an overview, Oxford Bulletin of
Economics and Statistics, 48 201 212.
Hoa, T.V., 1992, A multi-equation model of energy consumption in Thailand, lntcrnational Journal of
Energy Research, 16 381 385.
Johansen, S., 1988, Statistical analysis of cointegration vectors, Journal of Economic Dynamics and
Control, 12 231-254.
Johanscn, S., 1991, Estimation and hypothesis testing of cointegrating vectors in Gaussian autoregression models, Econometrica, 59 t551-1580.
Johansen, S. and K. Juselius, 1990, Maximum likelihood estimation and inference on cointegration with
applications to money demand, Oxford Bulletin of Economics and Statistics, 52 169 210.
Kwiatkowski, D., P.C.B. Phillips, P. Schmidt and Y. Shin, 1992, Testing the null hypothesis of
stationarity against the alternative of a unit root: how sure are we that economic time series have a
unit root?, Journal of Econometrics, 54 159-178.
Kraft, J. and A. Kraft, 1978, On the relationship between energy and GNP, Journal of Energy and
Development, 3 401-403.
Masih, A.M.M. and R. Masih, 1994, On the robustness of cointegration tests of the market efficiency
hypothesis: evidence from six european foreign exchange markets, Economia lnternazionale, 47
160 180.
Masih, A.M.M. and R. Masih, 1995a, Temporal causality and the dymamic interactions among
333
macroeconomic activity within a multivariate cointegrated system: evidence from Singapore and
Korea, Weltwirtschaftliches Archiv, 131.
Masih, A.M.M. and R. Masih, 1995b, Investigating the robustness of tests of the market efficiency
hypothesis: contributions from cointegration techniques on the Canadian floating dollar, Applied
Financial Economics, 5.
Masih, A.M.M. and R. Masih, 1996a, Docs money cause prices or the other way around'? Some
multi-country econometric evidence including error-correction modelling from Southeast Asia,
Journal of Economic Studies (forthcoming).
Masih, A.M.M. and R. Masih, 1996b, Bivariatc and multivariate tests of money-price causality:
evidence from a small developing economy, Journal of International Development (Formerly:
Manchcster Papers on Development) (forthcoming).
Masih, A.M.M. and R. Masih, 1996c, Energy consumption, real income and temporal causality: results
from a multi-country study based on cointcgration and error-correction modelling techniques,
Energy Economics (forthcoming).
Masih, R., 1906, Analysing unemployment, vacancies and the role of long-term unemployment in the
Bcveridge curve using a cointegration/vector error-correction modelling approach, Rivista Internazionalc di Scienze Economiche e Commerciali (International Review of Economics and Business)
(forthcoming).
Masih, R. and A.M.M. Masih 1995a, A fractional cointegration approach to empirical tests of PPP: new
evidence and methodological implications from all application to the Taiwan/US dollar relationship,
Wcltwirtschaftlichcs Archly, 131.
Masih, R. and A.M.M. Masih, 1995b, Macroeconomic activity dynamics and Granger causality: new
evidence from a small developing economy based on a vector error-correction modelling analysis,
Economic Modelling, 12.
Mehra, Y.P., 1994, An error-correction model of tile long-term bond rate, Federal Reserve Bank of
Richmond Economic Quarterly, 80 49-68.
Muscatelli, V.A. and S. Hum, 1992, Cointcgration and dynamic time series models, Journal of
Economic Surveys, 6 1-43,
Nelson, C.R. and C. Plosser, 1982, Trends and random walks ill macroeconomic time series, Journal of
Monetary Economics, 10 129-162.
Newey, W.K. and K.D. West, 1987, A simple, positive semi-definite hcteroscedasticity and autocorrelation consistent covariance matrix, Econometrica, 55 703-708,
Owcn, A.O. and P.N. Neal, 1989, China's potential as an energy exporter, Energy Policy, 17 485 500.
Pcrron, P. 1088, Trends and random walks in macroeconomic time series, Journal of Economic
Dynamics and Control, 12 297 332.
Phillips, P.C.B. 1087, Time series regressions with a unit root, Econometrica, 55 277 301.
Phillips, P.C.B. and B.E. Hansen, 1990, Statistical infercnce in instrumental variables regression with
1(1) processes, Review of Economic Studies, 57 99-125.
Phillips, P.C.B. and M. Loretan, 1991, Estimating long-run equilibria, Review of Economic Studies, 58
407-436.
Phillips, P.C.B. and P. Perron, 1988, Testing for a unit root in time series regression, Biometrika, 75
335-346.
Pindyck, R.S., 1980, Energy demand in the developing countries, in R.S. Pyndick (cd.), The Structure of
World Energy Demand, MIT Press, Cambridge, MA.
Saikkonen, P., 1991, Asymptotically efficient estimation of cointegrating rcgressions, Econometrica
Theory, 7 1-21.
Shiller, R. and P. Pcrron, 1985, Testing the random walk hypothesis: power versus frequency of
observation, Economics Letters, 18 381-386.
Stall, V., 1988, Energy in China's Modernization: Advances and Limitations, Sharpe Armonk, NY.
Stern, D.I., 1993, Energy and growth in the USA: a multivariate approach, Energy Economics, 15
137-150.
Stock, J.H. and M. Watson, 1988, Testing for common trends, Journal of the Amcrican Statistical
Association, 83 1097-1107.
334
Stock, J.H. and M. Watson, 1993, A simple estimator of cointegrating vectors in higher Order integrated
systems, Econometrica, 61 783-820.
Tang, C. and La Croix, 1903, Energy consumption and economic activity in China, Energy Journal, 14
21-36.
Taylor, M.P., 1993, Modelling the demand for UK broad money, 1871-1913, Review of Economics and
Statistics, 75 112-117.
Tegene, A. and F. Kuchler, 1994, Evaluating forecasting models of farmland prices, International
Journal of Forecasting, 10 65 80.
Thomas, R.L., 1993, Introductory Econometrics: Theory and Applications, Longman, London.
Zilberfarb, B.Z. and F.G. Adams, 1981, The energy-GDP relationship in developing countries, Energy
Economics, 3 244-248.