Causal Analysis of Oil Prices and Macroeconomic Performance: Evidence from India

Abstract

The oil price affects every corner of economic activity throughout the world. Economies of all oil-producing and oil-consuming countries are considerably influenced by the volatility in the crude oil prices in the international market. Although a number of studies have investigated on the causation of oil prices on macroeconomic determinants, but in the context of India, not many studies can be traced in the literature, despite the fact that India ranks among the top 10 largest oil-consuming countries. The goal of the article is to shed light on the impact of volatility in international crude oil prices on macroeconomic indicators of Indian economy. For the purpose of this study, West Texas Intermediate Crude Oil Prices are taken as the benchmark, that is, independent variable. The set of dependent variables include selected macroeconomic indicators, that is, balances of payments, inflation, exchange rate, index of industrial production, stock market returns, etc. The empirical analysis that covers data series of 10 financial years from April 2000 to March 2010 is done by using Unit Root Test and Granger Causality Test in Vector Autoregressive (VAR) framework. The findings of the study indicate that inflation in India has bidirectional causal relationship with international crude oil prices.

Keywords

Causality crude oil India macroeconomy vector autoregressive

Introduction

In the modern economic world, oil is the leading source of commercial energy and solely accounting for around 40 per cent of today’s world energy mix. Oil, or black gold, being an internationally traded commodity is affected by the laws of demand and supply, international politics, economic recessions, and so on. There are several causes which made crude oil prices to be volatile, and the volatility of crude oil prices in turn has affected many related sectors and stock market indices. The impact of oil price fluctuation is expected to be dissimilar in oil-importing and oil-exporting countries, as an increase in oil prices is considered as good news in oil-exporting countries and bad news in oil-importing countries; the reverse is expected when the oil price decreases. The economic impact of oil price changes is an issue that continues to attract considerable attention of economists, researchers, policy-makers, etc., especially at the time when oil prices have continued to rise over the past three years.

India ranks among the top 10 largest oil-consuming countries. The Organization of the Petroleum Exporting Countries (OPEC) has identified China and India as their main buyers of oil in Asia for several years to come. Oil accounts for about 30 per cent of India’s total energy consumption and the country’s total oil consumption is about 2.2 million barrels per day. India imports about 70 per cent of its total oil consumption and makes no exports. It faces a large supply deficit, as domestic oil production is unlikely to keep pace with demand. India’s domestic oil production in 2010 was only 0.8 million barrels per day. The balance recoverable reserve in the same year was about 733 million tonnes, of which 394 million tonnes was offshore and the onshore reserve was 339 million tonnes. In 2010, India had a total of only 2.1 million barrels refining capacity per day.

There are three crude oil varieties which are commonly traded on exchanges and used as benchmarks, namely, Brent (benchmark in European region), West Texas Intermediate (WTI) (benchmark in US region) and Dubai (benchmark in Gulf region). In this article, researchers have considered WTI as the benchmark of crude oil, as WTI crude oil is of very high quality. Its API (American Petroleum Institute) gravity is 39.6 degree which makes it a ‘light’ crude oil, and it contains only about 0.24 per cent of Sulphur which makes it a ‘sweet’ crude oil.

This article is organised into five sections. The first section discusses the research framework and begins with the introduction to crude oil and its importance for humankind. The second section reviews the studies and research work carried out on forecasting of crude oil prices and its relation with the macroeconomy of oil-exporting and oil-importing countries. The third section includes research objectives and the related research methodology for achieving the objectives followed by a brief description of database and the econometric techniques applied. The fourth section of the article presents analysis of data using the Augmented Dickey–Fuller Unit Root Test and Granger Causality Test. The fifth section summarises and concludes the whole study on the basis of the analysis conducted.

Literature Review

Over the past 20 years, a number of scholars have explored relationships between oil price shocks and the macroeconomic performance of various economies. Different methods of analysis have yielded diverse results, sometimes sharply, and sometimes modestly different. This section offers review of objectives, methodologies and findings of selected researches conducted in India and abroad.

Nian (2009) applied Box-Jenkins and Generalised Autoregressive Conditional Heteroscedasticity (GARCH) approach for forecasting crude oil prices. The author concluded that based on history and patterns, ARIMA (1, 2, 1) model is able to produce accurate forecast in crude oil prices. He, however, mentioned that the GARCH (1, 1) is the better model for daily crude oil prices because of its ability to capture the volatility by the non-constant of conditional variance. Hou and Suardi (2011) used parametric GARCH models to characterise crude oil price volatility. Based on crude oil price data for two markets, Brent and West Texas Intermediate, he concluded that out-of-sample, volatility forecast of the non-parametric GARCH model yields superior performance compared to an extensive class of parametric GARCH models.

Blanchardy and Gal (2007) characterised the macroeconomic performance of a set of industrialised economies in the aftermath of oil price shocks of 1970s and of the last decade. Focusing on the differences across episodes, the authors developed four hypotheses, namely, good luck, smaller share of oil in production, more flexible labour markets and improvements in monetary policy for examining the effects of the recent increase in the oil price on inflation and the economic activity. The study concluded that there is significant effect of increase in oil prices on inflation and the economic activity of the countries under consideration. Sill (2007) stated that oil-price increases may lead to significant slowdowns in economic growth as five episodes of the last US recessions were preceded by significant increases in the oil prices. The author, in this article, examined how run-ups in the oil price can affect output growth and inflation. He described the historical evidence on the relationship between oil prices, economic growth and inflation, and also discussed the channels by which increase in oil price might affect the economy of a nation.

LeBlanc and Chinn (2004) estimated the effects of oil price changes on inflation for the US, UK, France, Germany and Japan using an augmented Phillips curve framework and the statistical estimates. The authors found that increase in oil price had a modest effect only on inflation in US, Japan and Europe. Cheng, Su and Tzou (2009) investigated the out-of-sample value-at-risk (VaR) forecasts in gold markets by considering both oil volatilities and the flexible model construction by using the combined BHK (Brenner, Harjes & Kroner, 1996) and power GARCH (PGARCH) models to consider not only the effect of spot prices but also the endogenised power term. The empirical results indicated that PGARCH-HV model with its flexibility in power term for data transformation and the high volatility of crude oil is the best model for VAR forecasting.

Imarhiagbe (2010) analysed the impact of oil prices on stock prices of selected major oil-producing and oil-consuming countries (namely, Mexico, Russia, Saudi Arabia, India, China and the US). He considered nominal exchange rate as additional determinant and applied Vector Autoregression. The empirical results supported unit root in all variables (except Saudi Arabia and the US). The variance decomposition and impulse response tests applied in the study confirmed existence of oil prices and exchange rates as factors influencing stock prices in all selected countries. Gileva (2010) investigated the dynamics of oil prices and their vulnerabilities through application of different econometrical tools, for example, principal component analysis for finding out the fundamental factors contributing to this process, and ARCH, GARCH for forecasting the crude oil prices.

As, there is insufficient number of researches carried on effects of oil prices and macroeconomic performance with respect to India, the present research seeks to establish causal relationship between oil prices and macroeconomic performance.

Research Objectives and Methodology

This article intends to analyse comprehensively the causation of international crude oil prices with special reference to India. Accordingly, the perceived objectives of the study are as follows:

to examine the stationarity in the Indian macroeconomic indicators and international crude oil prices, and

to analyse the causal relationship between international crude oil prices and selected Indian macroeconomic indicators.

Data Description

The analysis includes Wholesale Price Index, Index of Industrial Production, Gold price in domestic market, Balance of Payments, Gross Domestic Product, Stock Exchange Returns (calculated by taking base of S&P CNX Nifty index of National Stock Exchange of India), and Foreign Exchange Rate as the macroeconomic indicators of India and West Texas Intermediate (WTI) spot prices as the benchmark of international crude oil prices. The study considers data for 10 financial years spanning from April 2000 to March 2010. The brief description of selected indicators with their sources of collection is presented in the Table 1.

Table 1.

List of Variables

Variable Name	Symbol	Data Source
Crude Oil Spot Price (WTI dollars per barrel)	CRO	www.eia.org
Wholesale Price Index (monthly avg., base: 2004–2005 = 100)	WPI	www.rbi.org
Index of Industrial Production (base: 1993–1994)	IIP	www.rbi.org
Gold Price in Domestic Market (₹ per 10 grammes)	GLD	www.rbi.org
Balance of Payments (monthly in US$ million)	BOP	www.rbi.org
Gross Domestic Product (thousand crore ₹)	GDP	www.rbi.org
Stock Market Returns (log. returns of S&P CNX NIFTY)	SMR	www.nseindia.com
Foreign Exchange Rate (₹ per unit of US$)	FER	www.rbi.org

Source: Authors’ own experiences.

Analytical Tools

The following econometric tools (available in STATA IC 10 software) have been applied for empirical investigation.

Unit Root Test

Before using the time series data for further investigation, its stationarity must be tested. To test the stationarity, Augmented Dickey–Fuller (ADF) test for Unit Root ‘t’ test has been applied. The model form of ADF test is as follows:

\begin{array}{l} Δ y_{t} = α + β_{y_{t - 1}} + δ t + ς_{1} Δ y_{t - 1} \\ + ς_{2} Δ y_{t - 2} + ... + ς_{k} Δ y_{t - k} + ε_{t} \end{array}

(1)

Here, k is the number of lags; and y_tis the time series data under consideration.

The test is based on the null hypothesis that the variables under consideration contain a unit root, and alternative hypothesis is that the variables are generated by a stationary process.

Granger Causality Test

The researchers have applied Granger Causality Test (proposed by Granger, 1969) in a Bivariate VAR Framework. According to Granger, if there exists causal relationship between variables, they can be used to predict behaviour of each other. In general, a time series X is said to be Granger cause of another time series Y, if using past values of X improves the prediction of current values of Y. In other words, if changes in X precede changes in Y, X will said to be ‘Granger Cause’ of changes in Y.

It is important to mention here that Granger Causality Test is valid on the assumption that variables are stationary. This test is helpful in providing insights into the short-run relationship. The mathematical expression of Granger Causality Test is as follows:

Δ X_{t} = α + \sum_{i = 1}^{p} β_{i} Δ X_{t - 1} + \sum_{j = 1}^{q} γ_{j} Δ Y_{t - 1} + δ_{1} Z_{t - 1} + u_{t}

(2)

Δ Y_{t} = α + \sum_{i = 1}^{p} β_{i} Δ Y_{t - i} + \sum_{j = 1}^{q} γ_{j} Δ X_{t - j} + δ_{2} Z_{t - 1} + v_{t}

(3)

Testing causal directions among the variables of interest in the Granger sense involves use of F-tests of the joint significance to test whether lagged information on a variable Y provides any statistically significant information about a variable X in the presence of lagged X. If not, then ‘Y does not Granger-cause X’.

Vector Autoregression

Vector Autoregression models are used to forecast and analyse causal relationship among economic time series variables. The use of VARs for causal inferences is known as structural modelling. Mathematically, in a Vector Autoregression (VAR) model, each of the endogenous variables is explained by its lagged or past values, and the lagged values of other endogenous variables in the model. Thus, a Bivariate Vector Autoregressive model for X_t and Y_t can be formulated as follows: X_t(4)A₀A₁X_t_–1A₂X_t_–2A_nX_t_–nA₁Y_t_–1A₂Y_t_–2A_nY_t_–nU_t Y_t(5)A₀A₁X_t_–1A₂X_t_–2A_nX_t_–nA₁Y_t_–1A₂Y_t_–2A_nY_t_–nU_t

In the above equations, A₀is a vector of constant terms, A_nis the matrices of constants to be estimated, U_t is a vector of residuals (assumed to be white noise), and n is the lag length.

Results and Discussion

This section presents results derived from Descriptive Statistics, Augmented Dickey–Fuller Unit Root, Johansen Cointegration and Granger Causality Test.

Summary Statistics

Basic descriptive information about the selected variables is summarised in Table 2. Number of observations for all the variables is 120 as the study covers the monthly observations for 10 years.

Table 2.

Results of Descriptives

	Obs.	Min.	Max.	Mean	SD	Skewness (SE)	Kurtosis (SE)
CRO	120	19.39	133.88	52.34	25.86	1.02 (0.221)	0.80 (0.438)
WPI	120	81.7	136.3	104.47	15.71	0.34 (0.221)	–1.09 (0.438)
IIP	120	154.9	350.4	221.58	49.38	0.44 (0.221)	–0.87 (0.438)
GLD	120	4,267.17	17,159.42	8,103.17	3,693.37	0.97 (0.221)	–0.22 (0.438)
BOP	120	–87,193	118,479	24,812.85	36,185.04	0.21 (0.221)	2.20 (0.438)
GDP	120	450,265	1,716,765	887,691	341,945.2	0.67 (0.221)	–0.57 (0.438)
SMR	120	0.76	1.19	1.01	0.06	–0.44 (0.221)	0.09 (0.438)
FER	120	39.32	51.2	45.58	2.58	–0.64 (0.221)	0.97 (0.438)

Source: Authors’ own calculations.

Results of descriptive statistics show that over the period from 2000–2001 to 2009–2010, the WTI spot prices, which are the indicators of crude oil prices in international market varied between 19.39 and 133.88 dollars per barrel. The relative range for the values of other variables is highest in case of BOP and lowest for SMR. The values of mean and standard deviation shown in the table indicate clearly that SMR and FER are the most consistent variables, and GLD, BOP and GDP are the most inconsistent variables as they have maximum coefficient of variation. Table 2 also depicts positive skewness for all the variables except SMR and FER, which indicates that the large tail of the distribution lies towards the higher values of the variables under study. Kurtosis, the degree of peakedness in a curve of the frequency distribution is the highest for BOP and the lowest for SMR.

Lag Order Selection

To identify proper lag order for conducting the causality analysis in Bi-variate Autoregressive framework, the researchers applied several minimum value-based criteria. The results of Likelihood Ratio (LR) test are contained in Table 3.

Table 3.

VAR Lag Order Selection Criteria

Lag	LR	FPE	AIC	SIC	HQIC
0		5.00	88.94	89.01	89.13
1	1,775	4.00	74.74	75.43*	76.45*
2	136.77	3.80	74.66	75.97	77.89
3	203.34	2.00	74.01	75.94	78.76
4	140.64*	2.00*	73.90*	76.45	80.17

Source: Authors’ own calculations.

Notes: *Indicates lag order selected by the criterion.

LR = Likelihood ratio.

FPE = Final prediction error.

AIC = Akaike information criterion.

SIC = Schwarz information criterion.

HQIC = Hannan-Quinn information criterion.

On the basis of the results of selected criteria, a model with four lags order is selected as the Final Prediction Error (FPE), Akaike Information Criterion (AIC), and the Likelihood Ratio (LR) suggest lags order of four for the model. However, Schwarz Information Criterion (SIC) and Hannan-Quinn Information Criterion (HQIC) have chosen a model with two lags.

Augmented Dickey–Fuller Unit Root Test

Before applying Granger Causality Test on the selected variables, a formal test is required to confirm time series properties. For this purpose researchers applied Augmented Dickey–Fuller (ADF) test of unit root. The lag length selected for applying test is four based on the FPE and AIC. In this test, the null hypothesis is that a variable contains a unit root, and alternative hypothesis is that the variables are generated by a stationary process.

The results of ADF test contained in Table 4 clearly indicate that ‘t’ value for all the variables is less than critical value. Hence, the null hypotheses are rejected at 1 per cent level of significance, and thus, it can be said that all the variables under study are generated by a stationary process. The tests for integration of order zero I(0) are carried out on the level of the variables, and the tests for integration of order I(1) are carried out on the first difference of the variables. The test results show that all the variables except SMR have a unit root, indicating that their own levels are non-stationary.

Table 4.

Results of Augmented Dickey–Fuller Unit Root Test

Variable	Order of Integration	Z(t)	p-value
CRO	I (0)	–1.728	0.4169
DCRO	I (1)	–5.122	0.0000
WPI	I (0)	0.917	0.9933
DWPI	I (1)	–5.688	0.0000
IIP	I (0)	0.618	0.9881
DIIP	I (1)	–6.090	0.0000
GLD	I (0)	1.294	0.9966
DGLD	I (1)	–5.023	0.0000
BOP	I (0)	–2.813	0.0565
DBOP	I (1)	–5.543	0.0000
GDP	I (0)	1.244	0.9963
DGDP	I (1)	–4.881	0.0000
SMR	I (0)	–4.117	0.0009
FER	I (0)	–2.454	0.1270
DFER	I (1)	–2.922	0.0428

Source: Authors’ own calculations.

Note: 1%, 5% and 10% critical values are –3.505, –2.889 and –2.579, respectively.

Granger Causality Test

In a Bi-variate VAR framework, the results of Granger Causality Test are highly sensitive to the order of lags in the autoregressive process as this test is employed to investigate whether the past of one time series improves the forecast of the present and future of another time series. The results of Granger Causality between international crude oil prices and macroeconomic determinants of Indian economy are presented in Table 5.

Table 5.

Results of Granger Causality Test (Lag Order: 3)

Null Hypothesis	F-Statistics	p-Value	Ho Rejected/Failed to Reject
CRO does not Granger Cause WPI	8.09606	0.0000	Ho rejected
WPI does not Granger Cause CRO	3.47193	0.0104	Ho rejected
CRO does not Granger Cause IIP	1.03015	0.3952	Ho failed to reject
IIP does not Granger Cause CRO	1.95659	0.1064	Ho failed to reject
CRO does not Granger Cause GLD	1.91223	0.1137	Ho failed to reject
GLD does not Granger Cause CRO	2.06423	0.0906	Ho failed to reject
CRO does not Granger Cause BOP	1.70682	0.1538	Ho failed to reject
BOP does not Granger Cause CRO	2.49260	0.0473	Ho rejected
CRO does not Granger Cause GDP	1.67026	0.1622	Ho failed to reject
GDP does not Granger Cause CRO	0.99449	0.4139	Ho failed to reject
CRO does not Granger Cause SMR	1.72749	0.1492	Ho failed to reject
SMR does not Granger Cause CRO	1.77081	0.1401	Ho failed to reject
CRO does not Granger Cause FRE	0.97568	0.4241	Ho failed to reject
FRE does not Granger Cause CRO	2.82200	0.0285	Ho rejected

Source: Authors’ own calculations.

Notes: (a) Significant at 5% confidence level.

(b) FPE and AIC were used to determine the appropriate lag lengths.

The results show (Table 5, Appendix 1) that the international crude oil prices have unidirectional causal relationship with BOP and FRE. Only WPI comprehends bidirectional causality with the international crude oil prices. Results also indicate that there is no causation impact of crude oil prices variation in international market on India’s IIP, GLD, GDP and SMR.

Conclusion

In a nutshell, WTI spot prices have bidirectional causal relationship with inflation in India. It indicates that the past data of WTI spot prices can forecast the present and future trend of inflation and industrial production in India and vice versa. Further, inflation indices in India are found to be highly related with international crude oil prices. On the basis of results extracted by employing Granger Causality Test, it can be concluded that the position of forex reserves and balance of payments have causation impact on international crude oil prices. Macroeconomic indicators such as index of industrial production, gold prices, gross domestic product and stock market returns are of no use for predicting the international crude oil prices. These findings of the study have implications for academicians, economic researchers, investors, policy-makers and the financial institutions.

Footnotes

Causal Analysis of Oil Prices and Macroeconomic Performance: Evidence from India

Table 1A.1.

Data Sheet

Year	Month	CRO (WTI $/Barrel)	WPI	IIP	GLD	BOP	GDP	FER	NIFTY	NRETURNS
2000–2001	Apr	25.72	81.7	156.5	4,460.00	–4,445	467,806.7	43.64	1,469.03	0.914
	May	28.79	81.8	160	4,473.60	–4,445	467,806.7	43.98	1,312.65	0.893548804
	Jun	31.82	82.3	154.9	4,521.15	–4,445	467,806.7	44.69	1,451.74	1.105961223
	Jul	29.7	82.5	156.4	4,529.20	–1,882	450,265	44.77	1,445.26	0.995536391
	Aug	31.26	82.6	157.5	4,516.25	–1,882	450,265	45.68	1,350.94	0.934738386
	Sep	33.88	83.37	158.6	4,518.13	–1,882	450,265	45.88	1,371.27	1.015048781
	Oct	33.11	85.1	157.2	4,537.61	1,9438	539,007.2	46.34	1,201.60	0.876267985
	Nov	34.42	85.3	163.1	4,482.69	1,9438	539,007.2	46.78	1,240.59	1.032448402
	Dec	28.44	84.9	171.9	4,540.80	1,9438	539,007.2	46.75	1,291.43	1.040980501
	Jan	29.59	85.4	170.2	4,465.58	14,532	540,560.6	46.54	1,316.96	1.019768783
	Feb	29.61	85.5	166	4,369.57	14,532	540,560.6	46.51	1,371.91	1.041724882
	Mar	27.25	85.7	177.1	4,268.60	14,532	540,560.6	46.62	1,214.47	0.885240285
2001–2002	Apr	27.49	86.2	160.4	4,267.17	6,857	507,586.9	46.78	1,116.41	0.919
	May	28.63	86.4	162.5	4,440.58	6,857	507,586.9	46.92	1,159.44	1.038543188
	Jun	27.6	86.6	159	4,400.20	6,857	507,586.9	47	1,107.15	0.954900642
	Jul	26.43	86.8	160.4	4,379.58	2,285	492,430.9	47.14	1,077.98	0.973653073
	Aug	27.37	87.1	162.2	4,448.64	2,285	492,430.9	47.12	1,069.01	0.991678881
	Sep	26.2	87.2	161.7	4,631.30	2,285	492,430.9	47.64	949.43	0.888139494
	Oct	22.17	87.5	162.2	4,687.60	17,387	590,940.1	48.02	953.92	1.004729153
	Nov	19.64	87.4	167	4,603.96	17,387	590,940.1	47.99	1,031.62	1.081453371
	Dec	19.39	87.2	177.1	4,576.46	17,387	590,940.1	47.92	1,075.87	1.042893701
	Jan	19.72	86.7	176.9	4,692.69	30,064	585,906.1	48.33	1,087.20	1.010531012
	Feb	20.72	86.6	170.3	4,901.04	30,064	585,906.1	48.69	1,138.17	1.046881898
	Mar	24.53	87.2	184.2	4,920.21	30,064	585,906.1	48.73	1,159.33	1.018591247
2002–2003	Apr	26.18	87.5	167	5,040.63	8,137	548,716.1	48.91	1,120.74	0.966713533
	May	27.04	87.7	169.2	5,225.80	8,137	548,716.1	49	1,079.80	0.963470564
	Jun	25.52	88.7	166.2	5,313.20	8,137	548,716.1	48.96	1,065.90	0.987127246
	Jul	26.97	89.3	171.8	5,187.50	23,943	538,777.8	48.76	1,034.70	0.970728961
	Aug	28.39	90.1	172.2	5,129.20	23,943	538,777.8	48.58	977.60	0.944814922
	Sep	29.66	90.2	171.8	5,239.05	23,943	538,777.8	48.44	987.12	1.009738134
	Oct	28.84	90.2	173.6	5,210.40	29,349	622,899	48.37	955.12	0.967582462
	Nov	26.35	90.4	173.9	5,242.92	29,349	622,899	48.25	992.26	1.038885166
	Dec	29.46	90.1	188	5,443.60	29,349	622,899	48.14	1,074.25	1.082629553
	Jan	32.95	90.4	188.8	5,752.04	20,608	696,336.5	47.93	1,073.48	0.999283221
	Feb	35.83	91.3	182.2	5,771.30	20,608	696,336.5	47.73	1,055.84	0.983567463
	Mar	33.51	92.5	195	5,432.71	20,608	696,336.5	47.64	1,016.38	0.962626913
2003–2004	Apr	28.17	93.3	174	5,191.60	25,711	602,671.6	47.37	965.08	0.949526752
	May	28.11	93.4	180	5,560.77	25,711	602,671.6	47.03	963.20	0.998051975
	Jun	30.66	93.5	177.3	5,509.40	25,711	602,671.6	46.63	1,068.59	1.109416528
	Jul	30.76	93.4	183.1	5,362.50	39,526	602,950.8	46.22	1,150.01	1.076193863
	Aug	31.57	93.6	182.1	5,462.17	39,526	602,950.8	45.93	1,261.13	1.096625247
	Sep	28.31	94.6	184.7	5,718.40	39,526	602,950.8	45.85	1,369.03	1.08555819
	Oct	30.34	94.9	184.4	5,695.19	33,187	716,158.4	45.39	1,506.10	1.100121984
	Nov	31.11	95.3	188.2	5,850.00	33,187	716,158.4	45.52	1,580.02	1.049080406
	Dec	32.13	95.3	202	6,094.07	33,187	716,158.4	45.59	1,740.06	1.101289857
	Jan	34.31	96.3	203.9	6,179.04	45,569	712,144.4	45.45	1,906.00	1.095364528
	Feb	34.69	96.9	197.3	6,017.73	45,569	712,144.4	45.27	1,848.67	0.969921301
	Mar	36.74	96.9	210.7	5,986.48	45,569	712,144.4	45.14	1,779.63	0.962654233
2004–2005	Apr	36.75	97.5	189.5	5,915.80	33,986	681,580	43.93	1,848.45	1.03867096
	May	40.28	98.0	192.3	5,736.46	33,986	681,580	45.25	1,640.20	0.88733804
	Jun	38.03	98.3	190.3	5,862.12	33,986	681,580	45.5	1,506.12	0.918253871
	Jul	40.78	99.2	198.7	6,059.42	–2,926	692,804	46.04	1,568.08	1.04113882
	Aug	44.9	100.6	197.8	6,125.80	–2,926	692,804	46.34	1,615.30	1.03011326
	Sep	45.94	100.5	202.8	6,170.43	–2,926	692,804	46.09	1,691.96	1.047458676
	Oct	53.28	100.7	204	6,360.83	29,648	788,437	45.78	1,794.98	1.060887964
	Nov	48.47	101.5	202.7	6,550.83	29,648	788,437	45.12	1,873.94	1.043989348
	Dec	43.15	100.5	220	6,444.62	29,648	788,437	43.97	2,021.94	1.078977982
	Jan	46.84	101.0	219.2	6,148.60	55,199	808,644	43.75	1,977.83	0.978184318
	Feb	48.15	101.1	208.9	6,107.50	55,199	808,644	43.67	2,067.39	1.04528195
	Mar	54.19	101.4	231.4	6,262.20	55,199	808,644	43.69	2,096.23	1.013949956
2005–2006	Apr	52.98	102.7	204.9	6,150.58	5,437	776,554	43.74	1,987.10	0.947939873
	May	49.83	102.5	213	6,030.38	5,437	776,554	43.48	2,002.28	1.007639273
	Jun	56.35	102.9	213.6	6,134.23	5,437	776,554	43.58	2,134.29	1.06592984
	Jul	59	104.0	208.1	6,058.26	22,964	782,661	43.53	2,236.70	1.04798317
	Aug	64.99	104.1	212.9	6,249.00	22,964	782,661	43.62	2,357.56	1.054034962
	Sep	65.59	104.9	217.4	6,534.60	22,964	782,661	43.91	2,511.70	1.065381157
	Oct	62.26	105.4	223.9	6,873.80	–21,209	901,830	44.81	2,486.78	0.990078433
	Nov	58.32	105.5	214.8	7,167.50	–21,209	901,830	45.72	2,574.66	1.035338872
	Dec	59.41	104.9	232.5	7,585.93	–21,209	901,830	45.64	2,772.61	1.076883938
	Jan	65.49	105.4	237.9	7,925.00	58,704	928,577	44.39	2,892.68	1.043305766
	Feb	61.63	105.6	227.3	8,038.04	58,704	928,577	44.32	3,019.48	1.043834783
	Mar	62.69	105.7	251.9	8,059.40	58,704	928,577	44.48	3,235.78	1.071634851
2006–2007	Apr	69.44	107.8	225.2	8,984.77	29,006	898,726	44.94	3,494.06	1.079820012
	May	70.84	108.7	237.9	9,969.39	29,006	898,726	45.4	3,437.41	0.98378677
	Jun	70.95	109.9	234.4	8,951.92	29,006	898,726	46.05	2,914.91	0.84799602
	Jul	74.41	110.8	235.5	9,559.00	10,526	914,663	46.45	3,092.11	1.060790899
	Aug	73.04	111.5	234.8	9,468.93	10,526	914,663	46.53	3,305.58	1.069037001
	Sep	63.8	112.2	243.5	8,998.20	10,526	914,663	46.11	3,492.13	1.056434877
	Oct	58.89	112.7	234	8,694.86	33,761	1,051,612	45.46	3,649.43	1.045044142
	Nov	59.08	112.6	248.8	9,139.92	33,761	1,051,612	44.85	3,868.61	1.060058694
	Dec	61.96	112.2	263.7	9,133.13	33,761	1,051,612	44.63	3,910.18	1.010745462
	Jan	54.51	112.4	265.5	9,069.17	90,341	1,087,239	44.33	4,037.06	1.032448634
	Feb	59.28	112.6	252.2	9,545.00	90,341	1,087,239	44.15	4,083.74	1.01156287
	Mar	60.44	112.8	289.1	9,369.60	90,341	1,087,239	44.02	3,731.13	0.91365513
2007–2008	Apr	63.98	114.5	250.7	9,321.09	46,183	1,055,935	42.14	3,947.28	1.057931511
	May	63.46	114.7	263.1	8,878.13	46,183	1,055,935	40.78	4,184.39	1.060069212
	Jun	67.49	114.8	255.3	8,707.42	46,183	1,055,935	40.77	4,222.17	1.009028795
	Jul	74.12	115.7	255	8,741.35	118,479	1,054,972	40.41	4,474.18	1.059687317
	Aug	72.36	116.0	260.3	8,835.58	118,479	1,054,972	40.81	4,301.36	0.961373928
	Sep	79.92	116.0	260.5	9,310.96	118,479	1,054,972	40.36	4,659.92	1.083359682
	Oct	85.8	116.3	262.6	9,690.96	105,515	1,212,568	39.32	5,456.62	1.1709686
	Nov	94.77	116.8	261	10,340.38	105,515	1,212,568	39.43	5,748.58	1.05350565
	Dec	91.69	116.7	284.7	10,311.00	105,515	1,212,568	39.43	5,963.57	1.037398801
	Jan	92.97	117.5	281.9	11,290.96	99,512	1,257,947	39.37	5,756.35	0.965252357
	Feb	95.39	119.0	276.2	11,887.83	99,512	1,257,947	39.73	5,201.56	0.903621218
	Mar	105.45	121.5	304.9	12,631.74	99,512	1,257,947	40.35	4,769.50	0.916936458
2008–2009	Apr	112.58	123.5	266.3	11,810.42	9,310	1,240,482	40.02	4,901.91	1.02776182
	May	125.4	124.1	274.6	12,143.00	9,310	1,240,482	42.12	5,028.66	1.025857268
	Jun	133.88	127.3	269.2	12,353.00	9,310	1,240,482	42.82	4,463.79	0.887669876
	Jul	133.37	128.6	271.3	13,027.96	–20,725	1,257,146	42.83	4,124.60	0.924013002
	Aug	116.67	128.9	264.7	11,860.64	–20,725	1,257,146	42.93	4,417.12	1.070920817
	Sep	104.11	128.5	276.2	12,220.28	–20,725	1,257,146	45.56	4,206.69	0.952360362
	Oct	76.61	128.7	262.9	12,690.87	–87,193	1,392,298	48.65	3,210.22	0.76312255
	Nov	57.31	126.9	267.6	12,142.92	–87,193	1,392,298	48.99	2,834.79	0.883051629
	Dec	41.12	124.5	284	12,922.60	–87,193	1,392,298	48.63	2,895.80	1.021521876
	Jan	41.71	124.4	284.8	13,472.80	1,493	1,392,161	48.83	2,854.36	0.985689619
	Feb	39.09	123.3	276.8	14,800.22	1,493	1,392,161	49.26	2,819.21	0.987685506
	Mar	47.94	123.5	305.9	15,232.20	1,493	1,392,161	51.2	2,802.27	0.993991224
2009–2010	Apr	49.65	125.0	269.5	14,474.57	561	1,368,799	50.06	3,359.83	1.198967266
	May	59.03	125.9	280.7	14,620.83	561	1,368,799	48.53	3,957.96	1.178023888
	Jun	69.64	126.8	290.2	14,638.85	561	1,368,799	47.77	4,436.37	1.120872874
	Jul	64.15	128.2	290.8	14,720.37	45,600	1,419,468	48.47	4,343.10	0.978976055
	Aug	71.05	129.6	292.8	14,952.08	45,600	1,419,468	48.33	4,571.11	1.052499367
	Sep	69.41	130.3	302	15,722.61	45,600	1,419,468	48.43	4,859.31	1.063048144
	Oct	75.72	131.0	289.7	15,882.39	8,243	1,610,613	46.72	4,994.11	1.027740564
	Nov	77.99	132.9	299.8	17,056.80	8,243	1,610,613	46.56	4,953.54	0.99187643
	Dec	74.47	133.4	334.3	17,159.42	8,243	1,610,613	46.63	5,099.74	1.029514246
	Jan	78.33	135.2	331.2	16,705.80	9,833	1,716,765	45.96	5,156.22	1.011075074
	Feb	76.39	135.2	317.6	16,525.43	9,833	1,716,765	46.32	4,839.57	0.938588734
	Mar	81.2	136.3	350.4	16,613.96	9,833	1,716,765	45.49	5,178.15	1.069960761

Source: Authors’ own calculations.

References

Blanchardy

O.J.

Jordi

Gal

(2007). The macroeconomic effects of oil price shocks: Why are the 2000s so different from the 1970s? (NBER Working Papers 13368). US: National Bureau of Economic Research.

Brenner

R.J.

Harjes

R.H.

Kroner

K.F.

(1996). Another look at models of the short-term interest rate. The Journal of Financial and Quantitative Analysis, 31, 85–107.

Cheng

W.H.

J.B.

Tzou

Y.P.

(2009). Value-at-risk forecasts in gold market under oil shocks. Middle Eastern Finance and Economics, (4), ISSN: 1450-2889.

Gileva

(2010). Econometrics of crude oil markets. Thesis submitted to the Department of Economics, University of Paris.

Granger

C.J.

(1969). Investigating causal relations by econometric models and cross spectral methods. Econometrica, 37(3), August, 424–438.

Hou

A.J.

Suardi

(2012). A nonparametric GARCH model of crude oil price return volatility. Energy Economics, 34(2), 618–626.

Imarhiagbe

Samuel

(2010). Impact of oil prices on stock markets: Empirical evidence from selected major oil producing and consuming countries. Global Journal of Finance and Banking, 4(4), 15–31.

LeBlanc

Michael

Chinn

M.D.

(2004). Do high oil prices presage inflation? The evidence from G-5 countries (UC Santa Cruz Economics Working Paper No. 561; SCCIE Working Paper No. 04-04). Retrieved 27 February 2012, from http://ssrn.com/abstract=509262

Nian

L.C.

(2009). Application of ARIMA and GARCH models in forecasting crude oil prices. A dissertation submitted to the Faculty of Science, University of Teknologi, Malaysia.

10.

Sill

Keith

(2007). The macroeconomics of oil shocks. Business Review, (Q1), 21–31.