Académique Documents
Professionnel Documents
Culture Documents
AND DEVELOPMENT
Hassen Guenichi,
Copyright 2014
Introduction
53
54
55
56
Methodology
Unit Root with Structural Breaks: It goes without saying that structural change is
of considerable importance in the analysis of macroeconomic time series. Structural
change occurs in many time series for any number of reasons, including economic
crises, changes in institutional arrangements, policy changes, regime shifts, and war.
An associated problem is the testing of the null hypothesis of structural stability
against the alternative of a one-time structural break. If such structural changes are
present in the data-generating process, but not allowed for in the specification of an
econometric model, results may be biased towards the erroneous non-rejection of
the non-stationary hypothesis.9
Conventionally, dating of the potential break is assumed to be known a priori in
accordance with the underlying asymptotic distribution theory. Test statistics are
then constructed by adding dummy variables representing different intercepts and
slopes, thereby extending the standard Dickey-Fuller procedure.10 However, this
standard approach has been criticized, most notably by L. Christiano, who argued
that data-based procedures typically are used to determine the most likely location
of a break: evidence of an endogeneity or a sample selection problem.11 This
invalidates the distribution theory underlying conventional testing.
In response, a number of studies have developed different methodologies for
endogenizing dates, including E. Zivot and D. Andrews, P. Perron, R. Lumsdaine
and D. Papell, and C. Bai et al.12 Their research has shown that by endogenously
determining the time of structural breaks, bias in the usual unit root tests can be
reduced. P. Perrons 1992 work and his 1997 article coauthored with T. J. Vogelsang
propose a class of test statistics that allows for two different forms of a structural
break, namely, the Additive Outlier (AO) model, which is more relevant for
series exhibiting a sudden change in the mean (the crash model), and the Innovational Outlier (IO) model, which captures changes in a more gradual manner
over time.13
With this in mind, Lumsdaine and Papell (LP) introduced a novel procedure to
capture two structural breaks in a series.14 They found that unit root tests accounting for two structural breaks are more powerful than those that allow for
a single break. In support, D. Ben-David et al. argued that
just as failure to allow one break can cause non-rejection of the unit root null by the
Augmented Dickey-Fuller test, failure to allow for two breaks, if they exist, can cause non15
rejection of the unit root null by the tests which only incorporate one break.
57
c DXti et
i1 i
where DU1t = 1 if t >TB1 and otherwise zero; DU2t = 1 if t > TB2 and otherwise
zero; DT1t = t TB1 if t > TB1 and otherwise zero; and finally DT2t = t TB2 if
t > TB2 and otherwise zero. Two structural breaks are allowed for in both the time
trend and the intercept, which occur at TB1 and TB2. The breaks in the intercept are
shown in equation (1) by DU1t and DU2t, respectively, whereas the slope changes
(or shifts in the trend) are represented by DT1t and DT2t. The optimal lag length (k)
is based on the general-to-specific approach suggested by S. Ng and P. Perron.16
Cointegration Test in the Presence of Structural Breaks: As had been noted as far
back as 1989 by P. Perron, ignoring the issue of potential structural breaks can render
invalid the statistical results not only of unit root tests but of cointegration tests as well.
N. Kunitomo explains that in the presence of a structural change, traditional cointegration tests, which do not allow for this, may produce spurious cointegration.17 In
the present research, therefore, considering the effects of potential structural breaks is
very important, especially because the world economy has been faced with structural
breaks like revolutions and wars in addition to significant policy changes.
P. Saikkonen and H. Lutkepohl (SL) and S. Johansen et al. have proposed a test
for cointegration analysis that allows for possible shifts in the mean of the datagenerating process.18 Because many standard types of data-generating processes
exhibit breaks caused by exogenous events that have occurred during the observation period, they suggest that it is necessary to take into account the level shift in
the series for proper inference regarding the cointegrating rank of the system.
P. Saikkonen and H. Lutkepohl argued that structural breaks can distort standard inference procedures substantially and, hence, it is necessary to make appropriate adjustment if structural shifts are known to have occurred or are suspected.19
The SL test investigates the consequences of structural breaks in a system context
based on the multiple equation frameworks of Johansen-Jeslius, while earlier approaches like A. Gregory et al. considered structural break in a single equation
framework and others did not consider the potential for structural breaks at all.20
According to P. Saikkonen and H. Lutkepohl, an observed n-dimensional time
series Xt = (x1t, . . . , xnt), Xt is the vector of observed variables (t = 1, . . . , T)
which are generated by the following process:21
where DT0t and DU1t are impulse and shift dummies, respectively, and account for
the existence of structural breaks. DT0t is equal to one, when t = T0, and equal to
58
zero otherwise. Step (shift) dummy (DU1t) is equal to one when (t > T1), and is
equal to zero otherwise. The parameters g(i = 1, 2,"), m0, m1, and d are associated
with the deterministic terms. The seasonal dummy variables d1t, d2t, and d3t are not
relevant to this research since our data are yearly. According to SL, the term ut is
an unobservable error process that is assumed to have a VAR (p) representation as
follows:22
ut A1 ut1 . . . Ap utp et t 1; 2
By subtracting ut1 from both sides of equation (3) and rearranging the terms, the
usual error-correction form of equation (3) is given by:
Dut Put1
Xp1
j1
Gj Dutj vt
59
where the reduced rank, r, of the 434 matrix of P equals the number of cointegration vectors in the system and n equals the number of (endogenous) series in
cointegration equation (5). Thus, P can be written as P ab0 , where a and b are
each of the dimension r 3 5 and rank r. The matrix b contains the cointegrating
vectors b, b bt ; . . . ; bt ; while the matrix of the adjustment coefficients a
describes the speed of adjustment of each of the four individual series in Yt to
deviations from the cointegration relationships.
Unit Root Test with Two Structural Breaks: We investigate the stationary status
of the variables using the Lumsdaine and Papell (LP) tests for unit roots in the
presence of two structural breaks.29 Table 1 provides the results of the LP test. The
primary findings of the analysis are as follows: the unit root hypothesis is accepted
for all variables under investigation in each sector. The computed break dates are
60
Sector
Variables
TB1
TB2
LM
Agricultural
Sector
Y
K
L
EU
OP
1973 Q4
1973 Q4
1973 Q2
1973 Q3
1982 Q2
1994 Q2
1991 Q2
1999 Q1
1991Q2
1991 Q2
4.96
5.57
6.55
5.31
4.44
Accepted
Accepted
Accepted
Accepted
Accepted
Industrial
Sector
Y
K
L
EU
OP
1973 Q4
1976 Q3
1973 Q4
1973 Q3
1982 Q2
1991 Q1
1991 Q2
1991Q2
1991 Q1
1991 Q2
5.78
4.52
6.24
6.20
4.44
Accepted
Accepted
Accepted
Accepted
Accepted
Services
Sector
Y
K
L
EU
OP
1980 Q2
1973 Q4
1973 Q4
1973 Q3
1973 Q4
1991 Q2
1991 Q1
1991 Q2
1991 Q3
1991 Q2
4.95
5.15
5.85
5.78
4.44
Accepted
Accepted
Accepted
Accepted
Accepted
Y = sector production in U.S. dollars, K = sector capital stock in U.S. dollars, L = sector
employment number, EU = total energy use in sector in thousand tons of oil equivalent, and OP = world
oil price in U.S. dollars per barrel; R. Lumsdaine and D. H. Papell, Multiple Trend Breaks and the Unit
Root Hypothesis, The Review of Economics and Statistics, vol. 79, no. 4 (1997), pp. 21218.
the most significant from this test, which corresponds closely with the expected
dates associated with the effects of the oil boom in 1973 and the first Gulf War in
1991.
Cointegration Test Results: As explained above, Johansen derived the likelihood ratio (LR) test in order to determine the number of cointegrating relations in
a system of variables by allowing for the presence of potential structural breaks.
We now apply a maximum likelihood approach for testing and determining the
long-run relationship in the model under investigation. As mentioned earlier, in
this procedure Johansen assumed that the break point is known a priori. In table 1,
we determined the time of the break endogenously by the LM procedure. The
empirical results based on this method showed two significant structural breaks in
the model under investigation, which are consistent with the timing of the 1973 oil
shock and the first Gulf War (1991). Therefore, at this stage we include two
dummy variables of regime change in order to take into account the two structural
breaks in the system. Following the Johansen procedure we consider three cases:
an impulse dummy and shift with intercept included, an impulse dummy and shift
with trend and intercept included, and, finally, an impulse dummy and shift with
61
breaks occur in each variable in each sector and that long-run relationships exist in
our predefined model. In the following work, we divide our samples into three
subsamples limited by the break dates and estimate the long-run and the short-run
relationships between the variables. Finally, we try to interpret the causal relationships between variables and determine how each endogenous variable responds over time to a shock to other variables.
Table 2
RESULTS OF THE COINTEGRATION TEST USING THE SAIKKONEN
AND LUTKEPOHL
(SL) APPROACH
Intercept Included (C)
95%
99%
G0
LR
230.04
127.69
54.06
17.33
8.22
Agricultural Sector
0.000
94.4
98.2
0.000
68.5
71.9
0.013
46.5
49.4
0.738
28.4
30.8
0.477
14.1
16.1
105.8
78.5
55.2
35.7
20.4
0
1
2
3
4
0
1
2
3
4
139.76
73.78
30.70
14.24
3.75
Industrial Sector
0.000
94.4
98.2
0.035
68.5
71.9
0.853
46.5
49.4
0.930
28.4
30.8
0.948
14.1
16.1
105.8
78.5
55.2
35.7
20.4
0
1
2
3
4
126.13
72.94
33.76
14.28
4.93
Services Sector
0.000
94.4
98.2
0.048
68.5
71.9
0.690
46.5
49.4
0.910
28.4
30.8
0.844
14.1
16.1
105.8
78.5
55.2
35.7
20.4
G0
LR
0
1
2
3
4
Pval
90%
Pval
90%
95%
99%
232.24
126.47
54.90
18.73
8.08
Agricultural Sector
0.000 107.8 112.9
0.000
79.3
83.6
0.100
54.7
58.4
0.844
33.8
36.8
0.705
16.5
16.4
122.7
92.3
65.8
42.9
23.0
0
1
2
3
4
138.19
92.89
35.64
18.72
5.80
Industrial Sector
0.000 107.8 112.9
0.001
79.3
83.6
0.800
54.7
58.4
0.824
33.8
36.8
0.887
16.5
16.4
112.7
92.3
65.8
42.9
23.0
0
1
2
3
4
125.40
84.07
44.92
20.05
4.23
Services Sector
0.006 107.8 112.9
0.049
79.3
83.6
0.418
54.7
58.4
0.781
33.8
36.8
0.876
16.5
16.4
122.7
92.3
65.8
42.9
23.0
62
VECM Estimation Results in the First RegimeFirst Quarter (Q1) 1960 through
Fourth Quarter (Q4) 1973: As highlighted in table 3, the empirical results of the
Tunisian agricultural sector show that in the long-run relationship in the first regime, physical capital has the most significant effect on agricultural production
Table 3
VECTOR ERROR-CORRECTION MODEL (VECM) ESTIMATION
a
RESULTS IN THE FIRST REGIME
Y(21)
Agricultural Sector
1
b1
K(21)
L(21)
EU(21)
OP(21)
2.34
[5.16]
2.04
[2.06]
1.68
[4.20]
0.81
[4.79]
3.03
4.26
b2
0.77
[3.44]
5.50
[5.15]
0.14
[0.32]
0.50
[2.71]
a1
0.001
[0.17]
0.01
[1.17]
0.002
[2.76]
0.01
[0.68]
0.51
[4.65]
a2
0.03
[3.96]
0.01
[1.66]
0.002
[4.47]
0.002
[0.18]
0.17
[2.24]
Industrial Sector
1
b1
3.26
[2.08]
0.53
[0.61]
0.03
[0.16]
2.18
b2
30.98
[3.67]
11.72
[2.52]
5.38
[5.44]
3.22
a1
0.07
[2.64]
0.17
[2.70]
0.004
[1.83]
0.04
[2.75]
0.22
[1.13]
a2
0.009
[2.05]
0.005
[0.41]
0.0002
[0.51]
0.003
[1.22]
0.17
[4.47]
Services Sector
b1
1
0.83
[0.75]
0.83
[1.08]
0.68
[4.23]
2.37
b2
3.68
[2.20]
2.50
[2.15]
0.46
[1.89]
3.57
a1
0.007
[0.83]
0.03
[2.91]
0.002
[3.24]
0.002
[0.60]
0.51
[4.37]
a2
0.01
[2.15]
0.004
[0.71]
0.002
[5.34]
0.002
[1.16]
0.15
[2.35]
Y = sector production in U.S. dollars, K = sector capital stock in U.S. dollars, L = sector
employment number, EU = total energy use in sector in thousand tons of oil equivalent, OP = world
oil price in U.S. dollars per barrel, and C = drift.
63
F-Statistic
P-Value
6.04746
10.28020
3.05791
5.41766
0.00345
9.6E-05
0.01376
0.00602
0.37142
0.09686
0.25438
1.25674
0.76837
0.15150
0.08900
0.20270
0.48777
0.86683
4.67759
1.31970
0.61672
0.42615
0.02802
0.14729
64
Tunisia are consistent with the view that energy and oil price have a causal impact
on output growth. Our results are also in line with findings by D. Stern, H. Yang,
and S. Lardic and V. Mignon who obtained similar results for other countries.30
Unlike the agricultural sector, the results for the industrial sector tests indicate
that Granger causality is not running between all variables. These results for this
sector are consistent with the neoclassical view that energy is neutral to growth. Our
results are consistent with the findings by U. Erol and E. Yu, E. Yu and J. Choi,
E. Yu and B. K. Hwang, B. S. Cheng, E. Yu et al., and E. Yu and J. C. Jin, which
offered evidence in favor of the neutrality-of-energy hypothesis.31 In the case of the
services sector, the Granger causality is running only in one directionbetween oil
price and output growth (oil price causes output)but is absent between energy use
and services production. These conclusions are consistent with the neoclassical view
that energy use is neutral to growth and are in line with the results of U. Erol and
E. Yu.32 The outcomes are consistent with the research of S. Lardic and V. Mignon,
which suggested that oil prices cause output growth and uncovered an asymmetric
long-run relationship between oil prices and GDP.33
VECM Estimation Results in the Second RegimeFirst Quarter (Q1) 1974 through
Third Quarter (Q3) 1991: In this time period, the oil price coefficients are still significant
for growth in both the agricultural and industrial sectors. One can explain this by the
fact that during this time span Tunisia exported oil and used the revenues for investments projects, which in turn fostered growth. But, the energy use coefficients are
still not significant in terms of the long-run relationships; the energy consumption was
not a limiting factor for the agricultural and industrial sectors during this time frame.
The estimation results for the sector of services suggest that, for the period
19731991, the energy use and oil price were not significant factors for growth
(see table 5). Indeed, for growth only the capital and labor coefficients are significant for a long-run relationship.
For the second period of our sample, the Granger causality tests are reported in
table 6. The results indicate that, like the first regime, for the agricultural sector the
Granger causality runs in both directions, between energy use and agricultural output
growth and is unidirectional between oil price and output growth (oil price causes
output). These results are aligned with findings by K. H. Ghali and M. I. T. El-Sakka
that energy use is not neutral to growth.34 But with the unidirectional relationship
between oil price and agricultural output, we find that oil price has a negative effect
on output growth. The increase in oil price delayed the output growth.
In the case of the industrial sector, the Granger causality is bidirectional between
energy use and output growth and unidirectional between oil price and industrial output.
These results are consistence with the view of non-neutrality of energy to growth. But,
in Tunisia the industrial sector growth is dependent upon fluctuations in world oil prices.
For this period, the services sector output growth is independent of energy and
its price. Indeed, the Granger causality test and VECM estimation fail to find
65
Table 5
VECTOR ERROR-CORRECTION MODEL (VECM) ESTIMATION RESULTS IN THE
a
SECOND REGIME
Y(21)
Agricultural Sector
1
b1
K(21)
L(21)
EU(21)
OP(21)
2.01
[6.41]
0.19
[0.41]
0.27
[1.04]
0.32
[2.62]
1.30
9.99
b2
0.11
[3.11]
1.84
[5.50]
2.35
[2.17]
3.52
[6.75]
a1
1.22
[1.57]
1.38
[2.28]
0.004
[0.22]
2.43
[3.09]
6.64
[3.20]
a2
0.28
[1.65]
0.38
[2.80]
0.0007
[0.17]
0.52
[1.06]
1.42
[3.07]
4.56
[5.65]
0.21
[0.47]
1.05
[2.27]
0.41
[3.10]
3.64
1.94
Industrial Sector
b1
1
b2
0.72
[2.28]
1.65
[3.94]
0.999
[2.20]
0.57
[4.40]
a1
0.36
[2.46]
0.21
[4.03]
0.003
[3.28]
0.14
[2.77]
0.77
[2.49]
a2
0.06
[2.83]
0.05
[1.61]
0.004
[5.13]
0.06
[3.00]
0.31
[1.16]
0.50
[3.01]
1.39
[2.69]
1.54
[2.47]
0.001
[0.02]
6.79
2.69
Services Sector
b1
1
b2
2.09
[0.09]
7.96
[5.32]
6.01
[3.34]
0.396
[2.00]
a1
0.016
[0.09]
0.13
[0.38]
0.009
[1.38]
0.093
[0.67]
2.27
[1.60]
a2
0.042
[0.68]
0.044
[0.37]
0.003
[1.46]
0.064
[1.32]
0.575
[1.16]
Y = sector production in U.S. dollars, K = sector capital stock in U.S. dollars, L = sector
employment number, EU = total energy use in sector in thousand tons of oil equivalent, OP = world
oil price in U.S. dollars per barrel, and C = drift.
a relationship between energy use and output and between oil price and output
growth.
VECM Estimation Results in the Third Regime Third Quarter (Q3) 1991 through
Fourth Quarter (Q4) 2005: The empirical results indicate that in the agricultural
66
Null Hypothesis
Agricultural Sector
Energy does not Granger cause output growth
Output growth does not Granger cause energy
Oil price does not Granger cause output growth
Output growth does not Granger cause oil price
Industrial Sector
Energy does not Granger cause output growth
Output growth does not Granger cause energy
Oil price does not Granger cause output growth
Output growth does not Granger cause oil price
Services Sector
Energy does not Granger cause output growth
Output growth does not Granger cause energy
Oil price does not Granger cause output growth
Output growth does not Granger cause oil price
F-Statistic
P-Value
9.80750
8.33717
7.89687
1.19845
0.00090
0.00202
0.00260
0.32060
2.70741
6.40796
5.64023
1.09865
0.04668
0.25200
0.00186
0.17080
0.00057
3.35635
2.58446
2.19469
0.99943
0.04640
0.08977
0.12649
sector the total energy use (EU) and oil price (OP) coefficients are not statistically
significant. This result shows that these two factors are not a limiting factor on
economic growth in Tunisia. This is due to the governments intervention by subsidizing this sector. In the third long-run relationship, the industrial sector production depends on the total energy use and oil price fluctuations. Indeed, the EU
and OP coefficients are statistically significant (a 1-percent increase in EU and OP
leads, respectively, to an increase of 1.91 percent and a decrease of 0.84 percent in
production). Table 7 provides the VECM estimation results for the third regime.
For the services sector, the VECM estimation results show the dependence of
this sector on energy use, which is the limiting factor to its production growth.
However, the coefficient of oil price in the first cointegration relationship is not
significant. This result can be explained by the fact that the Tunisian services
sector is quite developed and contains many sub-sectors, such as tourism and
communications, which are not dependent on oil price fluctuations.
Table 8 reports the results of the Granger causality tests indicating that for the
agricultural sector Granger causality runs in one direction and only between output
growth and energy use (energy use causes output). Thus, in contrast with the
neoclassical argument that energy is neutral to growth, our results for Tunisia are
consistent with the view that energy has a causal impact on output growth. The
absence of relationships between world oil price and output growth in the case of
the agricultural sector can be explained by the use of government subsidies.35
The results in table 8 indicate that for the industrial sector Granger causality is
bidirectional running between energy use and output growth and is unidirectional
67
Table 7
VECTOR ERROR-CORRECTION MODEL (VECM) ESTIMATION RESULTS
a
IN THE THIRD REGIME
Y(21)
Agricultural Sector
1
b1
K(21)
L(21)
EU(21)
OP(21)
1.44
[4.88]
0.66
[3.88]
1.25
[4.36]
0.30
[1.56]
2.83
3.30
b2
0.12
[1.63]
0.03
[0.44]
1.66
[15.59]
0.04
[1.40]
a1
0.003
[0.11]
0.01
[0.29]
0.006
[3.98]
0.15
[2.85]
0.18
[0.81]
a2
0.19
[2.59]
0.15
[1.54]
0.009
[2.10]
0.53
[3.95]
0.15
[0.26]
Industrial Sector
b1
1
1.62
[5.76]
1.91
[2.35]
0.08
[2.36]
3.71
b2
0.95
[2.11]
0.02
[0.04]
0.19
[3.25]
7.08
a1
0.09
[3.59]
0.13
[3.02]
0.01
[3.38]
0.11
[5.38]
0.34
[4.77]
a2
0.086
[3.50]
0.13
[3.23]
0.003
[1.21]
0.05
[2.95]
0.09
[0.23]
0.19
[3.96]
2.65
[9.17]
2.58
[8.01]
0.08
[0.38]
2.540
1.768
Services Sector
b1
1
b2
1.31
[2.97]
1.34
[21.9]
0.02
[0.31]
0.097
[13.5]
a1
0.004
[0.096]
0.079
[2.11]
0.028
[4.70]
0.08
[3.05]
0.43
[0.50]
a2
0.25
[1.17]
0.65
[3.80]
0.064
[2.34]
0.176
[1.34]
1.50
[0.37]
Y = sector production in U.S. dollars, K = sector capital stock in U.S. dollars, L = sector
employment number, EU = total energy use in sector in thousand tons of oil equivalent, OP = world
oil price in U.S. dollars per barrel, and C = drift.
between oil price and output. These results for this sector and period contradict the
neoclassical view that energy is neutral to growth. For the services sector, however, the Granger causality is running only in one direction between energy use
and output growth (energy use causes output), but it is absent between energy oil
68
Null Hypothesis
Agricultural Sector
Energy does not Granger cause output growth
Output growth does not Granger cause energy
Oil price does not Granger cause output growth
Output growth does not Granger cause oil price
Industrial Sector
Energy does not Granger cause output growth
Output growth does not Granger cause energy
Oil price does not Granger cause output growth
Output growth does not Granger cause oil price
Services Sector
Energy does not Granger cause output growth
Output growth does not Granger cause energy
Oil price does not Granger cause output growth
Output growth does not Granger cause oil price
F-Statistic
P-Value
5.22494
0.15171
0.97049
0.40586
0.00856
0.85962
0.38565
0.66849
3.48777
2.86683
4.67759
1.31970
0.01672
0.02615
0.02802
0.44729
5.15226
0.44611
2.51687
2.61655
0.00303
0.64311
0.09282
0.08493
price and services production. The results show that the services sector growth
depends on energy consumption but not on the world oil price.
Conclusion
The latest advances in econometric theory of structural change and a lack of the
studies examining the relationships between energy and economic growth in developing countries are the main motivations behind this research. In order to
examine the relationships between world oil price and the economic growth
sectors in a developing country (in this case Tunisia), we use the technique of
cointegration in the presence of structural breaks. The predetermined dates of
structural changes divide the sample into regimes that allow us to better evaluate
and understand the relationships among world oil price, energy use, and economic
growth.
The results show that the agricultural and the services sectors are not influenced much by a surge in oil prices. But it is very clear that our results contradict
neoclassical theory that energy is neutral to growth. Indeed, we find that energy
use always has had a great impact on the production of all economic sectors. We
can explain this as the agricultural sector still is supported by the government via
subsidies. In the last decade, the services sector has witnessed significant developed
in part due to the rapid growth in the communications industry as mobile phone
ownership has become widespread among the Tunisian population. Additionally,
69
the tourism subsector has become a major driver of the Tunisian economy, which is
made even more attractive by the cost differential for the tourists coming from high
cost-of-living developed nations and enjoying the relatively lower costs in Tunisia,
where oil prices are maintained by the government.
The industrial sector is influenced by energy use and world oil prices fluctuations,
especially in the last two regimes (first quarter 1974 through third quarter 1991 and
third quarter 1991 through fourth quarter 2005). This is due to the importance of this
sector to the overall Tunisian economy. Indeed, in the first period (first quarter 1960
through fourth quarter 1973), Tunisia was in the first stages of development after
gaining its independence. The industrial sector was nascent and the country had
excess oil production capacity, which is why we find that the production of this sector
was greatly influenced by the use of energy and not by the price of oil. In the last two
time periods, the industrial sector became one of the bases of the Tunisian economy.
Then, particularly in the last regime, Tunisias energy supply and demand balance
became onerousa situation with major implications for the industrial sector as it is
a significant energy consumer. Therefore, it makes sense that a surge in oil prices has
a notable impact on the growth of the industrial sector.
In general, the results reject the neoclassical assumption of the neutrality of
energy to economic growth. Nonetheless, world oil prices do have a significant
effect on the Tunisian industrial sector, although locally the oil prices are maintained by the government. The results of this study are in line with some findings,
but contradict others. In particular, our results are consistent with those of S. Lardic
and V. Mignon, who suggest an asymmetric relationship between oil price and
output growth: the increase in the oil price delayed or had a negative influence on
economic growth.36 Finally, we can better see the effects of world oil prices on the
global Tunisian economy if we add the variable of public expenditure in the model
because many products and sectors are subsidized.
NOTES
1
John Kraft and Arthur Kraft, On the Relationship between Energy and GNP, The Journal of
Energy and Development, vol. 3, no. 2 (spring 1978), pp. 4013; A. T. Akarca and T. V. Long,
On the Relationship between Energy and GNP: A Reexamination, The Journal of Energy and
Development, vol. 5, no. 2 (spring 1980), pp. 32631; E. S. H. Yu and B. K. Hwang, The
Relationship between Energy and GNP: Further Results, Energy Economics, vol. 6, no. 4
(1984), pp. 18690; E. S. H. Yu and J. Choi, The Causal Relationship between Energy and
GNP: An International Comparison, The Journal of Energy and Development, vol. 10, no. 2
(spring 1985), pp. 24972; U. Erol and E. S. H. Yu, On the Causal Relationship between
Energy and Income for Industrialized Countries, The Journal of Energy and Development, vol.
13, no. 1 (autumn 1988), pp. 11322; A. M. M. Masih and R. Masih, Energy Consumption, Real
Income and Temporal Causality: Results from Multi-Country Study Based on Cointegration and
Error-Correction Modeling Techniques, Energy Economics, vol. 18, no. 3 (1996), pp. 16583;
and B. S. Cheng and T. W. Lai, An Investigation of Cointegration and Causality between
Energy Consumption and Economic Activity in Taiwan, Energy Economics, vol. 19, no. 4 (1997),
pp. 43544.
70
2
E. Yu and J. Choi, op. cit.; A. M. M. Masih and R. Masih, op. cit.; and J. Asafu-Adjaye, The
Relationship between Energy Consumption, Energy Prices and Economic Growth: Time Series
Evidence from Asian Developing Countries, Energy Economics, vol. 22, no. 6 (2000), pp. 61525.
3
Frederick L. Joutz, Amany El-Anshasy, and Michael D. Bradley, Oil Prices, Fiscal Policy, and
Venezuelas Economic Growth, Department of Economics, George Washington University,
Washington, D.C., February 28, 2006.
4
S. Lardic and V. Mignon, The Impact of Oil Prices on GDP in European Countries: An
Empirical Investigation Based on Asymmetric Cointegration, Energy Policy, vol. 34, no. 18
(2006), pp. 3910915.
5
Philip A. Olomola and Akintoye V. Adejumo, Oil Price Shock and Macroeconomic Activities
in Nigeria, International Research Journal of Finance and Economics, vol. 3 (2006), pp. 2834.
6
D. I. Stern, Energy Use and Economic Growth in the USA: A Multivariate Approach, and A
Multivariate Cointegration Analysis of the Role of Energy in the US Macroeconomy.
8
R. Lumsdaine and D. H. Papell, Multiple Trend Breaks and the Unit Root Hypothesis, The
Review of Economics and Statistics, vol. 79, no. 4 (1997), pp. 21218; P. Saikkonen and H.
Lutkepohl, Testing for a Unit Root in a Time Series with a Level Shift at Unknown Time,
Econometric Theory, vol. 18, no. 2 (2002), pp. 31348; S. Johansen, R. Mosconi, and B. Nielsen,
Cointegration Analysis in the Presence of Structural Breaks in the Deterministic Trend,
Econometrics Journal, vol. 3, no. 2 (2000), pp. 21649; C. E. Quintos, Sustainability of the Deficit
Process with Structural Shifts, Journal of Business and Economic Statistics, vol. 13, no. 4 (1995),
pp. 40917; and S. Johansen, Determination of Co-integration Rank in the Presence of a Linear
Trend, Oxford Bulletin of Economics and Statistics, vol. 54, no. 3 (1992), pp. 38397.
9
P. Perron, The Great Crash, the Oil Price Shock, and the Unit Root Hypothesis, Econometrica, vol. 57, no. 6 (1989), pp. 1361401, and Further Evidence on Breaking Trend Functions
in Macroeconomic Variables, Journal of Econometrics, vol. 80, no. 2 (1997), pp. 33585, and
S. Leybourne, T. H. Kim, V. Smith, and P. Newbold, Tests for a Change in Persistence against the
Null of Difference-Stationarity, Econometrics Journal, vol. 6, no. 2 (2003), pp. 291311.
10
P. Perron, The Great Crash, the Oil Price Shock, and the Unit Root Hypothesis.
11
L. J. Christiano, Searching for a Break in GNP, Journal of Business and Economic Statistics,
vol. 10, no. 3 (1992), pp. 23749.
12
E. Zivot and D. Andrews, Further Evidence of the Great Crash, the Oil Price Shock and Unit
Root Hypothesis, Journal of Business and Economic Statistics, vol. 10, no. 3 (1992), pp. 25170;
P. Perron, Further Evidence on Breaking Trend Functions in Macroeconomic Variable;
R. Lumsdaine and D. H. Papell, op. cit.; and C. E. Bai, C. Hsieh, and Y. Qian, The Return to
Capital in China, Brooking Papers on Economic Activity, vol. 2 (2006), pp. 61101.
71
13
P. Perron and T. J. Vogelsang, Nonstationarity and Level Shifts with an Application to
Purchasing Power Parity, Journal of Business and Economic Statistics, vol. 10, no. 3 (1992), pp.
30120, and Further Evidence on Breaking Trend Functions in Macroeconomic Variables.
14
15
D. Ben-David, R. Lumsdaine, and D. H. Papell, Unit Root, Postwar Slowdowns and Long-Run
Growth: Evidence from Two Structural Breaks, Empirical Economics, vol. 28, no. 2 (2003), p. 304.
16
S. Ng and P. Perron, Unit Root Test in ARMA Models with Data Dependent Methods for the
Selection of the Truncation Lag, Journal of the American Statistical Association, vol. 90, no. 429
(1995), pp. 26881.
17
P. Saikkonen and H. Lutkepohl, Testing for the Cointegrating Rank of a VAR Process with
an Intercept, Econometric Theory, vol. 16, no. 3 (2000), pp. 373406, Testing for the Cointegrating Rank of a VAR Process with Structural Shifts, Journal of Business and Economic Statistics, vol. 18, no. 4 (2000), pp. 45164, and Trend Adjustment Prior to Testing for the
Cointegration Rank of a VAR Process, Journal of Time Series Analysis, vol. 21, pp. 43556; and
S. Johansen, R. Mosconi, and B. Nielsen, Cointegration Analysis in the Presence of Structural
Breaks in the Deterministic Trend.
19
P. Saikkonen and H. Lutkepohl, Testing for the Cointegrating Rank of a VAR Process with
Structural Shifts, p. 451.
20
A. Gregory, J. Nason, and D. Watt, Testing for Structural Breaks in Cointegrated Relationships, Journal of Econometrics, vol. 71, nos. 12 (1996), pp. 32141.
21
P. Saikkonen and H. Lutkepohl, Testing for the Cointegrating Rank of a VAR Process with
Structural Shifts.
22
Ibid.
23
Ibid.
24
H. Guenichi and S. Benammou, Oil Price, Energy Use and Tunisian Economic Growth: A
Multivariate Cointegration Analysis with Multiple Structural Changes, International Journal of
Business and Management Research, vol. 2, no. 1 (2009), pp. 1751.
25
27
H. Y. Toda and P. C. B. Phillips, Vector Autoregression and Causality, Econometrica, vol.
61, no. 6 (1993), pp. 1367393.
28
72
30
D. I. Stern, Energy Use and Economic Growth in the USA: A Multivariate Approach, and
A Multivariate Cointegration Analysis of the Role of Energy in the US Macroeconomy; H. Y.
Yang, A Note of the Causal Relationship between Energy and GDP in Taiwan, Energy Economics, vol. 22, no. 3 (2000), pp. 30917; and S. Lardic and V. Mignon, op. cit.
31
U. Erol and E. S. H. Yu, On the Causal Relationship between Energy and Income for Industrialized Countries, Time Series Analysis of the Causal Relationships between U.S. Energy
and Employment, Resources and Energy, vol. 9, no. 1 (1987), pp. 7589, and Spectral Analysis
of the Relationship between Energy Consumption, Employment and Business Cycles, Resources
and Energy, vol. 11, no. 4 (1989), pp 395412; E. Yu and J. Choi, op. cit.; E. S. H. Yu and B. K.
Hwang, op. cit.; B. S. Cheng, An Investigation of Cointegration and Causality between Energy
Consumption and Economic Growth, The Journal of Energy and Development, vol. 21, no. 1
(autumn 1995), pp. 7384; E. S. H. Yu, P. C. Y. Chow, and J. Y. Choi, The Relationship between
Energy and Employment: A Reexamination, Energy Systems and Policy, vol. 11 (1988), pp.
28795; and E. S. H. Yu and Jang C. Jin, Cointegration Tests of Energy Consumption, Income,
and Employment, Resources and Energy, vol. 14, no. 3 (1992), 25966.
32
U. Erol and E. S. H. Yu, On the Causal Relationship between Energy and Income for Industrialized Countries, and Time Series Analysis of the Causal Relationships between U.S.
Energy and Employment.
33
34
K. H. Ghali and M. I. T. El-Sakka, Energy Use and Output Growth in Canada: A Multivariate
Cointegration Analysis, Energy Economics, vol. 26, no. 2 (2004), pp. 22538.
35
These include quantities of fuel for farmers, new energy-saving materials, encouraging
farmers by a sum of money or a tax exemption.
36