The Impact of International Oil Price on Asian
Natural Gas Premium Based on Dynamic
Autoregressive Model
Presented By: Xinlei Yang
School of Business Administration,
China University of Petroleum (Beijing)
Agenda
•
•
•
•
Introdution
Methods
Results
Conclusion
Introduction
• Compared with North America and Europe, the premium on natural import gas
price has been higher in Asia-Pacific region for a long period. Natural gas InterRegional price spread has been mainly under the influence of oil price.
• The oil price has effect on Natural gas Inter-Regional price spread via influence
on natural gas pricing(mainly LNG pricing method, linked to oil price) and
energy substitutes.
Methods
• Use the dynamic autoregressive model (VAR) to analyze oil price and natural gas -Asia Europe region price spread(AERS), Asia-North American natural gas region price
spread (ANARS) Interaction relationship by using monthly data from 2007 to 2015.
• Then, using interval time VAR model to test the asymmetry of oil price to Natural gas
Inter-Regional price spread in different periods, and to test the endogeneity of natural
gas Inter-Regional price spread.
Unit Root test：ADF test
Table 1: Results of Unit Root Test
variable
T value
1%critical value 5%critical value 10%critical value Stable or not
brent
-2.216485
-3.493129
-2.888932
-2.581453
No
AERS
-2.067649
-3.493129
-2.888932
-2.581453
No
ANARS
-1.464163
-3.492523
-2.888669
-2.581313
No
▽brent
-5.924884
-3.493129
-2.888932
-2.581453
Stable
▽AERS
-7.930008
-3.493129
-2.888932
-2.581453
Stable
▽ANARS
-9.23309
-3.493129
-2.888932
-2.581453
Stable
Brent indicates the monthly price of Brent, AERS is the monthly data of the spread between
natural gas prices and natural gas prices in Japan, ANARS is the monthly data of the spread between
natural gas prices and US natural gas prices; ▽brent, ▽AERS and ▽ANARS are the
corresponding differential sequence.
Johansen cointegration test
Table 2: Co-integration test results table
The original
hypothesis: the
number of
cointegration
Maximum eigenvalue statistical
Trace critical value
Trace
threshold
Maximum
statistics
eigenvalue
5%
1%
5%
1%
equations
0
36.65211
29.68
35.65
27.27875
20.97
25.52
1
9.373364
15.41
20.04
7.727261
14.07
18.63
2
1.646104
3.76
6.65
1.646104
3.76
6.65
Granger test
Table 3 results of Grainger causality test
original hypothesis
Lag order
Observed value
F statistics Significance level
AERS is not brent Granger Cause
1.61445
0.164
brent is not AERS Granger Cause
2.93086
0.0168
0.2949
0.9146
brent is not ANARS Granger Cause
3.98048
0.0026
ANARS is not AERS Granger Cause
3.83631
0.0033
AERS is not ANARS Granger Cause
1.33426
0.2568
ANARS is not brent Granger Cause
5
103
So brent is AERS and ANARS Granger Causes, while ANARS is AERS Granger cause.
Var Model and Stability Test
Figure 1: AR root chart:
Impulse Response Function (IRF)
Figure 2: Asian-American natural gas regional price spreads on the oil price impulse
response map
Figure 3: Asian-American natural gas regional spreads on the Asian-European
natural gas regional impulse response map
Figure 4: Asian-European natural gas regional price spreads on oil price impulse
response map
Figure 5: Asian-European natural gas regional spreads on Asian-American natural gas
regional spread impulse response map
AERS Variance Decomposition
Table 4: AERS variance decomposition table
Period
S.E.
BRENT
AERS
ANARS
1
0.932602
0.098674
99.90133
0
2
1.441526
4.362284
94.66389
0.97383
3
1.842211
9.304645
89.50352
1.191833
4
2.148537
12.21537
86.32117
1.463457
5
2.370697
13.44583
84.59015
1.964022
6
2.526605
13.63487
83.63325
2.731877
7
2.635678
13.29282
82.94905
3.758131
8
2.714289
12.76919
82.23793
4.992881
9
2.774045
12.26214
81.38146
6.356399
10
2.822036
11.84863
80.39126
7.76011
ANARS fluctuations mainly from BRENT and AERS, from their contribution
less than the sum contribution of BRENT and AERS.
ANARS Variance Decomposition
Table 5: ANARS variance table
Period
S.E.
BRENT
AERS
ANARS
1
0.898826
4.963825
37.6762
57.35998
2
1.282914
2.960236
36.71237
60.32739
3
1.566217
5.718416
36.95767
57.32391
4
1.812699
10.16248
36.65959
53.17793
5
2.034699
14.89095
35.797
49.31205
6
2.234711
19.29878
34.61856
46.08265
7
2.414053
23.19745
33.30944
43.4931
8
2.574424
26.58934
31.97604
41.43461
9
2.717963
29.54582
30.67429
39.77989
10
2.846989
32.1518
29.43157
38.41663
Compared with the variance of AERS, ANARS can be found to be more
susceptible to oil price and other regional prices than AERS.
Endogenous test
Table 6: impulse response value of regional natural gas price spread
Maximum response value
2007-2011
AERS
0.711493
ANARS
0.7543
variance
0.00045811
2012-2015
0.36506
0.520339
0.00602789
There is a certain endogeneity of regional natural gas price spread,
and the influence of oil price on natural gas price spread is the same.
Results
• The fluctuation of oil prices has a significant effect on the Asian-American
natural gas regional spreads and the Asian-European natural gas regional
spreads, and the AERS and ANARS will also affect each other.
• ANARS and AERS are different in response to oil price fluctuations, while the
interaction effect between ANARS and AERS is asymmetric. The contribution
of oil price fluctuations to AERS and ANARS volatility was 11.85% and
32.15%. The contribution of AERS to ANARS was 29.43%, while ANARS
was 7.76% for AERS.
• The ANARS is more intense reflection of oil prices
Conclusions
• In the oil price at a higher price or a stable recovery period, consider strengthening
trade with North American natural gas.
• Strengthen development and utilization of domestic and foreign resources and
enhance resource guarantee capacity.
• Establish a Gas Trading Center to improve bargaining power.
Thanks!