The Effect of Natural Gas Price liberalization on Natural Gas Consumption in Residential sector in...

15
315 Journal of Global Economy (ISSN Print-0975-3931, Online -2278-1277), Volume 8 No 4, October-December, 2012 The Effect of Natural Gas Price liberalization on Natural Gas Consumption in Residential sector in Iran Mohammad Babazade 1 , Vahid Ghorbani Pashakolaei 2 and Khalil Ghadimi Dizaj 3 1. Introduction High natural gas consumption in residential sector in Iran is an important issue that Price liberalization and omitting natural gas subsidy is one of the solutions to reduce natural gas consumption in residential sector. Natural gas price in Iranian residential sector is 100 rials per cubic meters while cost Price is 3000 rials per cubic meters in 2010. In this study we applied CPI index to calculate real price. The investigation of nominal and real price of natural gas for the period 1975-2010 shows that in spite of increasing nominal price of natural gas between these years, the real price decreased. Also, results showed that the average growth of natural gas consumption in residential sector was 10%. Because of some facts such as low price of natural gas in Iran, none existence of a suitable substitution, high consumption and its effects on macroeconomic variables it would be necessary to investigate the main factors which affects the natural gas consumption in residential 1 Assistant Professor of Economics, Islamic Azad University Branch Firozkooh, Iran, [email protected] 2 PhD student of petroleum economics,Allame Tabatabaei University, Tehran, Iran, [email protected] 3 Master of pricing, National Iranian Gas Company (NIGC), [email protected] JGE

Transcript of The Effect of Natural Gas Price liberalization on Natural Gas Consumption in Residential sector in...

315 Journal of Global Economy (ISSN Print-0975-3931, Online -2278-1277),

Volume 8 No 4, October-December, 2012

The Effect of Natural Gas Price liberalization on Natural Gas

Consumption in Residential sector in Iran

Mohammad Babazade1, Vahid Ghorbani Pashakolaei

2 and

Khalil Ghadimi

Dizaj3

1. Introduction

High natural gas consumption in residential sector in Iran is an important issue that

Price liberalization and omitting natural gas subsidy is one of the solutions to reduce

natural gas consumption in residential sector.

Natural gas price in Iranian residential sector is 100 rials per cubic meters while cost

Price is 3000 rials per cubic meters in 2010. In this study we applied CPI index to

calculate real price. The investigation of nominal and real price of natural gas for the

period 1975-2010 shows that in spite of increasing nominal price of natural gas

between these years, the real price decreased. Also, results showed that the average

growth of natural gas consumption in residential sector was 10%. Because of some

facts such as low price of natural gas in Iran, none existence of a suitable substitution,

high consumption and its effects on macroeconomic variables it would be necessary to

investigate the main factors which affects the natural gas consumption in residential

1 Assistant Professor of Economics, Islamic Azad University Branch Firozkooh, Iran,

[email protected] 2 PhD student of petroleum economics,Allame Tabatabaei University, Tehran, Iran,

[email protected] 3 Master of pricing, National Iranian Gas Company (NIGC),

[email protected]

JGE

316 Journal of Global Economy (ISSN Print-0975-3931, Online -2278-1277),

Volume 8 No 4, October-December, 2012

sector. Therefore this study aimed to estimate long and short-run natural gas

consumption function in residential sector in Iran.

Our Objectives in this study is:

1) Estimation of long and short- run natural gas consumption functions in residential

sector in Iran;

2) Determination of long-run and short-run price and income elasticities;

3) The impacts of cutting subsidies on residential consumption of natural gas are

studied in two scenarios (in first scenario the liberalization process is considered in 3

years (2011-13) and in second scenario is considered in 5 years (2011-15)).

The present paper is organized as follows: the next (second) section presents literature

review; the third section provides the econometric specification of the model for

residential demand for natural gas and discusses the ARDL cointegration technique.

The forth section presents the data and evaluates the results of the econometric

analysis and examines the withdrawal of subsidies on natural gas. Finally there are

concluding remarks and some recommendations regarding to the results of the study.

2. Literature review Here a few recent researches in this context are reviewed (See Table 1).

Table 1 - Selected empirical results

Results method Country Sources

Income and Price elasticities less than one AIDS Iran Moshiri &

Shahmoradi (2005)

Long run price elasticity = -0.13 Long run Price elasticity = 0.17

STSM Iran Keshavarz &

Mirbagherijam

(2007)

Income and Price elasticities less than one AR Iran Lotfalipor &

Bagheri (2003)

Price elasticity less than one And income elasticity more than one

ECM United

state

Winston T. Lin,

Yueh H. Chen and

Robert

Chatov(1987)

Price elasticity less than one And income elasticity more than one

AIDS Australia

Akmal,

Mohammad, and

David Stern(2001)

317 Journal of Global Economy (ISSN Print-0975-3931, Online -2278-1277),

Volume 8 No 4, October-December, 2012

Short-run & Long run price elasticities less than one

and Total Response to

10% Price increase lead to decrease consumption

between 1.9% and 3.3% in different regions

Dynamic

Model U.S

Joutz and

Trust(2007)

existence of long-run relationships between the

energy

prices before and after the liberalisation, implying

the possibility of substitution among the different

forms of energy

VAR-

cointegration UK

Ferreira, P. Soares,

I. Araujo, M (2004)

that there is a significant relationship between

the fuel demand and fuel subsidy factors

This study recommends a gradually

controlled withdrawal of fuel subsidy at the level it

will be minimally harmful to the economy

Fuel subsidy Nigeria Nwachukwu,

Chike (2011)

Subsidies for electricity, natural gas and coal are

even more pervasive. In virtually all of the countries

studies, the prices of these fuels do not reflect

marginal costs and withdrawing this subsidies is

necessary.

analyses the

Commercial

energy

subsidies

developing

countries Kosmo(2003)

3. Methods

The method in this study is based on this process:

1) Demand Model: Determination of natural gas demand structure in residential

sector

2) ARDL Model: this step we explained research methodology and cointegration

concept

3) Short-run Model: this step we show short-run equation estimation method

4) Stability test: In order to ensure the stability of the long-run parameters of our

econometric specification, we applied the CUSUM and the CUSUMQ tests for the

residuals of the error-correction.

5) Winter’s Method: for determination of amount of dependent variable (natural gas

consumption), we need GDP and price level, price level determined in two scenarios

and GDP forecast with winter’s Method.

3-1. Demand Model

318 Journal of Global Economy (ISSN Print-0975-3931, Online -2278-1277),

Volume 8 No 4, October-December, 2012

A modified residential natural gas demand model in logarithmic form is adopted based

on:

ln𝐶𝑜𝑡 = 𝛼0 + 𝛼1 ln𝑌𝑡 + 𝛼2 ln𝑃𝑡 + 𝛼3 ln𝑃𝑒 +𝜀𝑡 (1)

Where Cot is the residential natural gas consumption (cm3) , Yt is the real gross

domestic product, Pt is the real residential natural gas price (rials/cm3) , Pe is the real

residential electricity price (rials/kwh)

ln is the natural logarithm transformation.

As for the expected signs in Eq. (1), one expects that α1>0 because higher added value

in residential sector should result in greater economic activity and accelerated

purchases of natural gas technology. The coefficient of price level is expected to be

less than zero for usual economic reasons, therefore, α2<0 and α3>0 because the

natural gas and electricity in residential sector is complementary goods.

3-2. ARDL Model

A recent single cointegration approach, known as autoregressive distributed lag

(ARDL).

An ARDL representation of Eq. (1) is formulated as follows pesaran et al. (2001):

(2) lnlnlnlnlnlnln 1161514

0

3

0

2

1

10 itCOtCOtCO

m

i

itCO

m

i

itCO

m

i

itCOCOt PYCOPYCOCO

(3) lnlnlnlnlnlnln 1161514

0

3

1

2

0

10 itPEtPEtPE

m

i

itPE

m

i

itPE

m

i

itPEPEt YCOPYCOPP

(4) lnlnlnlnlnlnln 1161514

0

3

1

2

0

10 itYtYtY

m

i

itY

m

i

itY

m

i

itYYt PCOYPCOYY

It has to be mentioned that the F-statistic obtained by performing the Wald test has a

non-standard distribution, whose asymptotic critical values are provided by Pesaran et

al. (2001).

In Eq.(2), where tCOln is the dependent variable, the null hypothesis of no

cointegration amongst the variables is )0 : ( 6CO5CO4CO0 H against the alternative

hypothesis )0 : ( 6CO5CO4CO1 H . This is denoted as ),( PYCOFCO. In Eq.(4), where

319 Journal of Global Economy (ISSN Print-0975-3931, Online -2278-1277),

Volume 8 No 4, October-December, 2012

tYln is the dependent variable, the null hypothesis is )0 : ( 6Y5Y4Y0 H against

the alternative )0 : ( 6Y5Y4Y1 H . This is denoted as ),( PCOYFY. In Eq.(3), where

tPEln is the dependent variable, the null hypothesis for cointegration is

)0 : ( 6PE5PE4PE0 H against the alternative )0 : ( 6PE5PE4PE1 H . This is

denoted as ),( COYPFPE.

Having identified the existence of a cointegration, general specification for the ARDL

(p1, q2, q3,) model is presented below:

(5) lnlnlnln211

0

3

0

2

1

10

q

i

iti

q

i

iti

P

i

itit PbYbCObbCO

3-3. Short-run Model

We estimated an error correction model associated with the long-run estimates. This is

specified as follows:

(6) EC lnlnlnln 1-t

1

3

1

2

1

10

311

t

q

i

iti

q

i

iti

p

i

itit PYCOCO

Where is the parameter of adjustment speed and 1-tEC denotes the residuals

obtained from the estimated cointegration model of Eq. (1).

3-4. stability test

To test for parameter stability, we follow Pesaran and Pesaran (1997) and estimate the

following error correction model:

∆𝑙𝑛𝐶𝑜𝑡 =∝0+ ∝1𝑖 ∆𝑙𝑛𝐶𝑜𝑡−𝑖 (7)

𝑚

𝑖=1

+ ∝2𝑖 ∆𝑙𝑛𝑌𝑡−𝑖 + ∝3𝑖 ∆𝑙𝑛𝑃𝑡−𝑖 + 𝛾𝐸𝐶𝑀𝑡−1 + 𝜀𝑡

𝑚

𝑖=0

𝑚

𝑖=0

All variables are as previously defined and the error correction term is calculated from

the long-run relationship. After estimating the model the cumulative sum of recursive

320 Journal of Global Economy (ISSN Print-0975-3931, Online -2278-1277),

Volume 8 No 4, October-December, 2012

residuals (CUSUM) and the CUSUM of squares (CUSUMSQ) tests were applied to

test for parameter constancy.

3-4. winter's Method

After evaluation of demand function, we could determine the value of dependence

variable (natural gas consumption). For this purpose we must firstly determine the

value of independence variable and forecast them (GDP and price).

In this section we explain the seasonal GDP forecast process. There is different

forecast method for seasonal data. The most famous seasonal method is moving

average, Single Exponential Smoothing (SES), Double Exponential Smoothing (DES)

and winters' Method. In this paper we used the winter's Method; this method is based

on eq. (8) to (11): Black (1997) and Rubin et al (1989)

Et =∝ Xt

St−L + (1−∝)(Et−1 + Tt−1) (8)

Tt = β Et − Et−1 + 1 − β Tt−L (9)

St = γ Xt − Et + 1 − γ St−L (10)

Ft+k = (Et + KTt)St−L+k (11)

That Et is the smoothing value, Tt related to trend, St seasonal fluctuations and Ft+k

indicated the value of forecast for t+k period. Coefficient indicated the weight of

variables and between 0 and 1, best value for this coefficient determine when the error

of forecast is minimized. Mean Square Deviation (MSD) index is used for estimation

accuracy that definition in eq.12:

MSD = et

2

n (12)

That et equal to Xt-Ft. We suppose that alpha, beta and gamma values are 0.2 and

MSD is 0.004.

321 Journal of Global Economy (ISSN Print-0975-3931, Online -2278-1277),

Volume 8 No 4, October-December, 2012

4. Data and empirical results

The data used in this paper are annual time series spanning the period 1975-2010. The

sources of our data on price of natural gas were obtained from the Iranian national gas

company (INGC) (annual publication). Residential natural gas consumption and GDP

were obtained from the Iranian central bank.

We performed the Augmented Dickey – Fuller (ADF) test to verify the exact order of

integration of the variables. Table 2 below displays the results of ADF tests. Results

show that all variables used in our study are an I(1) in 5% level of critical values

Table 2-ADF tests

Level

p- value

1 st Differences p-

value

Order of

Integration Variable ADF

stat. Variable ADF stat.

LNconsum -2.99 0.28 LNconsum -6.01 0.00 I(1)

LNpriceGas -0.7 0.92 LNpriceGas -5.98 0.00 I(1)

LNpriceEle -0.31 0.18

LNpriceEle -6.22 0.00 I(1)

LNgdp -0.81 0.22 LNgdp -6.03 0.00 I(1)

Note: ADF stands for the Augmented Dickey – Fuller test. All level variables are in logs. ∆ is

the first difference operator.

The F test proposed by Pesaran et al. (2001) can be used to determine whether a long-

run relationship exists. It has to be mentioned that the F-statistic obtained by

performing the Wald test has a non-standard distribution, whose asymptotic critical

values are provided by Pesaran et al. (2001). The calculated F-statistics, together with

the critical values, are reported in Table 3.

While GDP (Y) is the dependent variable no long-run relationship is observed. There

are long-run relationships among variables when natural gas consumption and natural

gas price are the dependent variables at 1% and 10% level, respectively. Overall, the

results of the bounds F test in Table 3 imply that at 10% level, the null hypothesis that

there is no cointegration among the variables in Eq. (2) and Eq. (3) cannot be

accepted, while for Eq. (4), where the calculated F-statistics are less than the critical

values, the null hypothesis is accepted suggesting a long-run relationship between

322 Journal of Global Economy (ISSN Print-0975-3931, Online -2278-1277),

Volume 8 No 4, October-December, 2012

natural gas consumption and price and GDP when natural gas consumption is the

dependent variable.

Taking for granted the existence of a long-run equilibrium, we do estimation by

setting two the maximum lag-length and using the Schwarz Bayesian Criterion (SBC)

to select the lag order of the model. The specification finally selected is an ARDL (1,

0, 0). The derived long-run elasticities and diagnostic tests for the ARDL model are

shown in Table 4. The estimated elasticities display the expected signs which are

negative for the price of natural gas and positive for the GDP, respectively.

Table 3 - Bounds testing for cointegration

Panel

A

Fco(Co/y,p)

=17.25a Fp(P/y,Co)=6.58 Fy(y/Co,P)=1.57

Panel

B

1% 5% 10%

I(0) I(1) I(0) I(1) I(0) I(1)

6.78 8.21 5.45 5.74 4.57 5.6

a. denote statistical significant at 10% level.

Critical values are obtained from Narayan (2005) for 30 observations. K=2 and it is the number of

regressors.

The results indicate that long-run price elasticity is -0.36 meaning that if residential

natural gas price increases by ten percent, the residential natural gas consumption

decreases by 3.6%; also the long-run income elasticity is 0.88. Finally, diagnostic tests

for the underlying ARDL model verify that the residuals are non-serially correlated,

correct functional form, normal, and non-heteroscedastic .

323 Journal of Global Economy (ISSN Print-0975-3931, Online -2278-1277),

Volume 8 No 4, October-December, 2012

Table 4 - Long-run coefficients for the ARDL (1, 0, 0) model

Variable C LnP LnGdp

Coefficient -0.5 -0.36 0.88

t-statistic 1.05 -1.36 1.97**

p-value 0.44 0.51 0.04

Diagnostic tests :

Variable t-statistic p-value

Serial Correlation 0.8 0.32

Functional Form 0.002 0.93

Normality 0.95 0.51

Heteroscedasticity 2.16 0.09

The maximum lag length was set 2.

** indicate 5% levels of significance.

The results of the short-run model (Eq. (6)) and diagnostic statistics for the short-run

ARDL model are presented in Table 5. As expected, all short-run elasticities are lower

in absolute value than those in the long-run model. The short-run price and income

elasticity are -0.12 and 0.62, respectively. The lagged error correction term is

statistically significant with the expected negative sign.

Figs. 1 and 2 below, display the results of CUSUM and CUSUMQ tests, respectively.

In both figures the dotted lines represent the critical upper and lower bounds at the

0.05 level of significance. The visual inspection of Figs. 1 and 2 reveals that there is

no evidence of parameter instability, since the cumulative sum of the residuals and the

cumulative sum of the squared residuals move within the critical bounds.

We used the dynamic ARDL model. For investigating the liberalization process in

first scenario (2011-13): 𝐿𝑁𝑐𝑜𝑛 = −0.051 + 0.15 𝐿𝑁𝐶𝑜𝑛 −1 − 0.12𝐿𝑁𝑝𝑟𝐺𝑎𝑠

+ 0.627𝐿𝑁𝑔𝑑𝑝 (13)

Table 5 - Error-correction representation

a results. lnCO , is the dependent variable

324 Journal of Global Economy (ISSN Print-0975-3931, Online -2278-1277),

Volume 8 No 4, October-December, 2012

Variable C ∆LnP ∆LnGdp 1-tEC

Coefficient -0.05 -0.12 0.627 -0.95

t-statistic 0.356 -0.365 1.95** -5.83***

p-value 0.724 0.71 0.00 0.00

Diagnostic statistics :

R2-adjusted 0.55 Schwarz criterion 39.3049

F-statistic 9.01 Akaike criterion 51.4

DW-statisticb 1.61 RSS 0.02

*** and ** indicate 1% and 5% levels of significance, respectively.

b. DW is the Durbin-Watson statistic and RSS is the residual sum of squares

Fig. 1. CUSUM test

325 Journal of Global Economy (ISSN Print-0975-3931, Online -2278-1277),

Volume 8 No 4, October-December, 2012

Fig. 2. CUSUMQ test

For the liberalization process we investigated monotonous decrease of the natural gas

subsidy (monotonous increase of natural gas price) in three years that indicated in

table5. With use of this prices and GDP value that forecasted with winters' Method for

the first scenario, we could calculate the natural gas consumption for this years. We

also forecast the consumption if subsidy do not withdrawal to compare the

consumption in two different situation. The findings indicate that the annual growth of

natural gas consumption in residential sector has been 9.9 % in the period of 1975-

2010 while this rate will be decreases to 2.7% if subsidy withdrawal during the period

2011 to 2013. These changes indicated in table 6 and Fig3

Table 6- natural gas consumption during the process of subsidy withdrawal in first scenario

Years

Price

(Rials/cubic

meter)

Consumption forecast

if subsidy withdrawal(with

price liberalization)

(million cubic meter)

Consumption forecast

if subsidy does not

withdrawal(without price

liberalization)

(million cubic meter)

2011 834.37 36258.56 44881.61

326 Journal of Global Economy (ISSN Print-0975-3931, Online -2278-1277),

Volume 8 No 4, October-December, 2012

2012 1556.25 36075.51 46321.73

2013 2278.12 37352.13 47676.5

Fig. 3. Residential natural gas Consumption in first scenario (2011-13)

The liberalization process in first and second scenario (2011-15) is the same. We used

the dynamic ARDL model and GDP value. Monotonous increase of natural gas price

indicate that price during 2011 to 2015 will be 593.75, 1075, 1556.25, 2037.5 and

2518.75 rials/ cubic meters respectively. The results indicated that consumption during

2011-15 with consider subsidy withdrawal process will be 37965.73, 38003.13,

39414.1, 41454 and 43931.57 million cubic meter respectively. We also forecast

natural gas consumption without subsidy withdrawal with auto regressive (1) and

results show that consumption will be 44881.61, 46321.7, 47676.5, 48950.9, 50149.9

million cubic meters. The rate of consumption growth will be decreases to 4.2% if

subsidy withdrawal during the period 2011 to 2015 (see table 7 and Fig.4). Therefore,

this policy could reduce the annual growth of natural gas consumption in first and

second scenario by 7.2 % and 5.7% respectively.

0

10000

20000

30000

40000

50000

60000 with price liberalization

without price liberalization

327 Journal of Global Economy (ISSN Print-0975-3931, Online -2278-1277),

Volume 8 No 4, October-December, 2012

Table 7- natural gas consumption during the process of subsidy withdrawal in second scenario

Years

Price

(Rials/cubic

meter)

Consumption forecast

if subsidy withdrawal(with

price liberalization)

(million cubic meter)

Consumption forecast

if subsidy does not

withdrawal(without price

liberalization)

(million cubic meter)

2011 593.75 37965.73 44881.61

2012 1075 38003.13 46321.73

2013 1556.25 39414.1 47676.5

2014 2037.5 41454 48950.9

2015 2518.75 43931.57 50149.96

Fig. 4. Residential natural gas Consumption in second scenario (2011-15)

5. Conclusions

0

10000

20000

30000

40000

50000

60000with price liberalization

without price liberalization

328 Journal of Global Economy (ISSN Print-0975-3931, Online -2278-1277),

Volume 8 No 4, October-December, 2012

This paper examines the effect of natural gas price liberalization on natural gas

consumption in residential sector in Iranian economy. The econometric specification

assumes that the residential demand for natural gas depends on the price of natural gas

and GDP in Iran. To estimate the model, we used ARDL cointegration technique. The

error correction model is consistent with the expectations about the signs of the short-

run parameters and their magnitude which are found lower than their long-run

counterparts.

The results indicates that short-run price and income elasticities are -0.12 and 0.62,

and long-run price and income elasticities are -0.36 and 0.88, respectively. Low price

elasticity leads to low efficiency of price policy because when the prices are too low

the responsiveness of demand with respect to price could be low as well. Hence it is

possible that in much higher range of prices the consumer response to price changes

could increase too, withdrawing subsidies on natural gas is one of the solutions to both

increasing the efficiency of price policy and reducing natural gas consumption.

Withdrawing subsidies on natural gas leads to reduction of annual growth of natural

gas consumption in first scenario (2011-13) and second scenario (2011-15) by 7.2 %

and 5.7% respectively. Also this sudden increase in natural gas price probability could

have significant negative side effects such as welfare loss or fuel poverty which

should be considered.

References 1- Akmal, M. Stern, D. (2001) ‘the structure of Australian residential energy demand’, Centre

for Resource and Environmental Studies, Ecological Economics Journal, No.0101. pp. 1-35.

2- Black, K. (1997) ‘Business statistics, contemporary decision making’, 2nd Ed. West

Publishing Company

3- Engle, R. Granger, C. 1987 ‘Cointegration and error correction representation: estimation

and testing’, Econometrica Journal, 55. pp. 251–276.

4- Ferreira, P. Soares, I. Araujo, M. (2005) ‘Liberalization, consumption heterogeneity and the

dynamics of energy prices’, Energy Policy Journal, vol.33, pp. 2244–2255.

5- Halicioglu, F. (2007) ‘Residential electricity demand dynamics in Turkey’, Energy

E c o n o m i c s J o u r n a l , 2 9 p p . 1 9 9 - 2 1 0 .

6- Joutz, F. Trost, R.P. (2007) ‘An economic analysis of consumer response to natural gas

Prices’,American Gas Association

329 Journal of Global Economy (ISSN Print-0975-3931, Online -2278-1277),

Volume 8 No 4, October-December, 2012

7- Keshavarz, Gh. Mirbagherijam, M. (2007) ‘the survey of natural gas demand (residential

and commercial) in Iran’, pajoheshhaie eghtesadie Iran Journal, No. 32, pp 137-160.

8- Kosmo, M. (2003) ‘Commercial energy subsidies in developing countries Opportunity for

reform’, Energy Policy Journal, Vol. 17(3), pp. 244–253.

9- Levin, R. I. D. S. Rubin, J. P. Stinson, & JR. E. S. Garoner, (1989) ‘Quantitative approaches

to management’, 7th Ed. McGraw-Hill Editions.

10- Lotfipor, M. Bagheri, A. (2003) ‘The estimation of residential natural gas demand in

Tehran’, pajoheshhaie eghtesadie Iran Journal. No. 16, pp 133-151.

11- Moshiri, S. Shahmoradi, A. (2005) ‘the estimation of natural gas and electricity residential

demand in Iran’, tahghighate eghtesadi Journal. No.72. pp 305-335

12- Noferesti, M. (1999) ‘Integration and cointegration in econometrics’, Tehran. Rasa

Institution. pp 90-102.

13- Nwachukwu, M.U. Harold C. (2011) ‘Fuel subsidy in Nigeria: Fact or fallacy’, Energy

Journal, Vol.36. pp. 2796-2801.

14- Pesaran, M.H. Pesaran, B. (1997) ‘Working with Microfit 4.0: Interactive Econometric

Analysis’, Oxford University Press. Oxford

15- Pesaran, M.H. Shin, Y. Smith, R.J. (2001) ‘Bounds testing approaches to the analysis of

level relationships’, Applied Econometrics Journal, 16 (3), pp. 289–326.

16- The information and data management sector, (2010) Gas contract management affairs.

Iranian national gas company (INGC)

17- Winston, T. Lin Y. Chen, H. Chatov, R. (2002) ‘The demand for natural gas, electricity and

heating oil in the United States’, Resources and Energy Journal. Vol. 9(3), October, P. 233-

258.