1 | P a g e
A SVARX Analysis of Oil Price Shocks and Subsidy Policy in Indonesia
EBM877A20
Andin Nurita Sofiana
Supervisor: dr. J.P.A.M. (Jan) Jacobs June 2014
Abstract
There is a close relationship between oil price and subsidy spending since decision in subsidy policy depends on the fluctuation of oil prices. This study explores the relationship between oil price shocks and macroeconomic variables including fiscal factors in Indonesia during 1990 – 2013. The use of Structural Vector Autoregression Model with exogenous variables (SVARX) creates the possibility to capture dynamic interactions between estimated variables. Impulse response analysis shows that economic growth represented by real GDP responds positively to oil price shocks in the short run. Negative response of economic growth to oil price shocks appears after six quarters. Furthermore, fiscal and monetary authorities respond to oil price shocks by increasing government subsidy and interest rate. In this case, response from government by protecting Indonesian economy from oil price shocks through fiscal and monetary policy could be effective in the short run. On the other hand, government spending responds positively to oil subsidy shocks. While inflation rate needs time lag in order to respond oil subsidy shocks and responds positively after second quarter, real GDP (in percentage change) responds directly and positively to oil subsidy shocks. It could mean that subsidy policy temporarily affects economic growth although it should be paid using high government expenditure.
2 | P a g e
1. Introduction
1.1. Motivation
As economies become more inter-connected in this globalization era, shocks that occur in one country can easily transfer to other countries, creating economic instability in a short period of time. One of external shocks, oil prices volatility, has a significant effect on the stability of open economy countries. Oil price shocks have been regarded one of the many reasons for global economic slowdown especially for oil importing countries (Hamilton 1983), including Indonesia. Therefore, Alom (2011) suggests oil price should be considered for policy making. Oil price shocks enforce government to intervene the economy through fiscal and monetary policies. Bernanke, Gertler, and Watson (1997) uncover that oil price shocks which are responded by tighter monetary policy lead to economic downturns. On the other hand, Pieschacon (2009) argues fiscal policy is a very important transmission mechanism, as it determines the degree of exposure of domestic variables to an external shock especially oil price shocks.
Indonesia, currently struggling to meet oil production targets in the short term, is sensitive to the impact of external changes especially from oil price fluctuation. The increase in global oil prices forces government to augment oil subsidy. However, oil price subsidy has become controversial since the existence of subsidy lies between protection and distortion. The rationale of subsidy is to unable low-income people to cope with high oil prices. On the other hand, subsidy policy gives incentives for middle and high-income people to increase their consumption level, which eventually leads to higher inflation. This dilemma is an attractive subject to be further discussed.
1.2. Objectives
3 | P a g e
growth (GDP) and inflation rate to oil subsidy spending. This study employs Structural Vector Autoregression with exogenous variables (SVARX) in order to capture the dynamic interaction between variables and to treat some variables as exogenous factor in the model. The study utilizes quarterly data of global oil prices, oil subsidy spending, government expenditure, interest rate, inflation, and real Gross Domestic Product from 1990 until 2013.
While some studies mainly explore about the link between oil prices and monetary policy, this study focuses on the relationship between oil prices and fiscal factor. In the impulse response analysis, there is also response of macroeconomic variables to oil subsidy policy.
According to the findings, there is positive response from real GDP (in percentage change) to oil price shocks at first quarter then it reaches negative point after six quarters. In addition economic growth positively responds to oil subsidy policy. Oil price shock seems distorting economic growth after six quarters. On the other hand, while creating government expenditure, oil subsidy supports economic growth. In this case, subsidy policy temporarily and positively affects economic growth.
1.3. Background
4 | P a g e
Figure 1 Indonesia Oil Production and Consumption (thousand barrels/ day) Source: US Energy Information and Administration
5 | P a g e
Figure 2 Fuel Economic Cost and Subsidized Price (Indonesian Rupiah) Source: Centre for Strategic and International Studies
According to Figure 3, the largest proportion of subsidy in government expenditure is oil subsidy. The amount spent on energy subsidies (fuel subsidy and electricity subsidy) is substantially larger than other public spending such as education, environmental protection, health, and housing. Oil subsidy varies between 10% and 28% of the government budget between 2001 and 2010. Oil subsidies encourage overconsumption of fuel, delay the adoption of energy-efficient technologies, and crowd out high-priority public spending, including spending on physical infrastructure, education, health and social protection. Moreover, oil price fluctuations leading to oil subsidies influences not only government spending but also other macroeconomic variables such as interest rate, inflation, and economic growth.
6 | P a g e
The remainder of the study is organized as follows. The second section provides a theoretical framework and Indonesia economic performance related to the topic. The third section reviews the methodology. The data provides in the fourth section. Results and analysis are described in the fifth section. Conclusions will close the presentation.
2. Literature Review
It is important to explore how macroeconomic variables including fiscal variables respond to oil price shocks. Bernanke (2006) states that the demand side is affected by oil price shock because of the decrease in consumer spending. Oil price movements may also have an impact on inflation. Increase in oil price creates lower productivity and output. Furthermore, it could lead to a higher consumer price and eventually, creates inflation. There is empirical evidence showing increasing fuel prices by $0.25 per liter results, on average, in an increase in the cost of living (i.e., of the consumer price index, CPI) of around 6 percent (Anand et al, 2013).
The rising oil price pushes the monetary authority to increase interest rate. Kilian (2009) argues there are two main channels of transmission of oil price shocks to monetary policies. One is the increased cost of producing domestic output (which is akin to an adverse aggregate supply shock); the other is reduced purchasing power of domestic households (which is akin to an adverse aggregate demand shock). The latter channel of transmission may be explained by increased precautionary savings and by the increased operating costs of energy-using durables.
7 | P a g e
generates significant fiscal saving. On the other hand, subsidy reforms lower household real incomes of all income groups. This explanation could illustrate the influences of subsidy which are increasing household real incomes (protection) and lowering fiscal saving (distortion).
In addition, there are a huge number of studies analyzing the relationship between oil prices and economic growth. Rodriquez and Sánchez (2005), studying the impact of oil prices on real economic activity in the main industrialized OECD countries, find evidence of a non-linear impact of oil prices on real GDP. It means that the effects of an increase in oil prices on real GDP growth are found to differ substantially from those of an oil price decrease. Kilian and Vigfusson (2013) forecasted US real GDP using oil price. They employed the Hamilton model and also find that there is non-linear predictive relationship between oil prices and real GDP. Moreover, according to Kilian and Vigfusson (2011), there are asymmetric responses in real output because of oil price shocks, which are in line with economic theory. The asymmetric responses are seen in location effect and the uncertainty effect. Both effects imply that an unexpected increase in the real price of oil will cause a negative response of real output that is larger in absolute terms than the positive response of real output to an unexpected decline in the real price of oil of the same magnitude.
8 | P a g e
Figure 4 explains the effects of oil price shocks on macroeconomic variables. Theoretically the oil price shock shifts Short Run Aggregate Supply (SRAS) up, causing output and employment to fall. In absence of further price shocks, prices will fall over time and economy moves back toward full employment.
It is also essential to analyze the response of economic growth to oil subsidy. Oil subsidy depresses economic growth through several channels (IMF, 2013). First, subsidy discourages investment in the energy sector. Low and subsidized prices for energy can result in lower profits or outright losses for producers, making producers difficult to expand energy production. Second, subsidy crowds out growth-enhancing public spending since some countries spend more on energy subsidies than on public health and education. Third, subsidy diminishes the competitiveness of the private sector over the longer term.
According to the relation between variables explained in previous part, the transmission mechanism of oil price shock can be illustrated in figure 5. Oil price shocks which are exogenously given could increase inflation. Both Indonesia authorities, fiscal and monetary authorities, respond higher inflation by increasing interest rate and increasing subsidy spending. Responses from the authorities will eventually influence economic growth. Nevertheless the variables are possible to have causal relationship which can be analyzed with Granger Causality tests.
9 | P a g e
3. Methodology
Methodology of SVARX estimation is started by SVARX specification. Checking the stationarity of the data is the second step, continued by determining the lag length and, causality test. Finally, response of estimated variables can be analyzed through impulse response function and forecast error variances decomposition.
3.1 SVAR Specification
The standard structural VAR representation is
(1) With [ ] [ ]
is the lagged of endogenous variables in matrix form. represents the vector of serially and mutually uncorrelated structural innovations. The standard SVAR can be modified in order to include exogenous variables which can be represented as
(2)
This study includes fiscal factor, crisis dummy, and dummy for break time based on ZA-tests, the model can be rewritten as
(3)
10 | P a g e
where is the global oil price represented by West Texas intermediate price. is the interest rate, Interest rate is used to analyze response of monetary authority to oil price shock. is the inflation rate, is the percentage change in fiscal spendings, is subsidy spending and, is the percentage change of real GDP, representing economic growth. is additional variable which is dummy variable of economic crisis in Asia and is treated as exogenous. are dummy variables in order to capture structural break for each variable.
Structural Vector Autoregression which included exogenous variables (SVARX) model is employed in order to capture the dynamic interaction between variables in the model. SVARX model is often used in macroeconomic analysis. The SVARX approach has the advantage in modeling dynamic behavior of economic variable and forecasting. In addition the generalized IRF, as a part of SVARX estimation, captures the dynamic responses of a variable of interest to innovations in the variable itself and in other variables. IRF helps to understand the response of fiscal policy to the shock of oil price and the response of economic growth to the shock of oil subsidy. Moreover, the analysis in forecast error decomposition helps to explain the percentage of the error in the forecast of fiscal policy which is attributed to the oil prices shock after certain period. Another essential element in SVARX is identification problem. As it has already mentioned in introduction, global oil prices are treated as exogenous since it is assumed that it cannot be intervened by Indonesian government. The restriction imposed in the model is recursive restriction as shown in equation (5).
11 | P a g e
Where , , , , , and are the structural disturbances, which are, oil
price, inflation rate, interest rate, subsidy spending, government spending and real GDP. , , , , , are the residuals.
3.2 Unit root test
This estimation tests whether a time series variable is non-stationary using an autoregressive model. A well-known test that is valid in large samples is the Dickey– Fuller (DF) test. The test is extended in 1981 for use with other autoregressive models, the resulting test is called the Augmented Dickey–Fuller test or the ADF test.
Zivot and Andrews (1992) introduce unit root test assuming that shocks can be treated as exogenous events. They follow Perron’s characterization of the form of structural break and come up with three models to test for a unit root: (1) model A, which permits a one-time change in the level of the series (intercept); (2) model B, which allows for a one-time change in the slope of the trend function, and (3) model C, which combines one-time changes in the level and the slope of the trend function of the series. In order to test for a unit root against the alternative of a one-time structural break, Zivot and Andrews use the following regression equations:
∑ (Model A)
∑ (Model B) ∑ (Model C)
Where is an indicator dummy variable for mean shift occurring of each possible break date (TB) while is corresponding rend shift variable.
and
12 | P a g e
The null hypothesis in all three models is which implies that the series ) contains a unit root with a drift that excludes any structural break, while the alternative hypothesis implies that the series is a trend-stationary process with a one-time break occurring at an unknown point in time.
3.3 Lag Selection
A critical element in the specification of SVARX is the determination of the lag length. The importance of lag length determination is demonstrated by Braun and Mittnik (1993) who show that estimates of a SVARX whose lag length differs from the true lag length produces inconsistent impulse response functions and variance decompositions. Technically, the lag selection criteria is usually considered include Akaike’s information criterion (AIC), Schwarz’s information criterion (SIC), Phillips’ posterior information criterion (PIC), and Keating’s (1995)’s application of the AIC and SIC criterion (KAIC and KSIC).
3.4 Granger Causality Test
Granger causality test is a statistical hypothesis test for determining whether one time series is useful in forecasting another. Ordinarily, regressions reflect "mere" correlations, but Clive Granger, who won a Nobel Prize in Economics, argued that a certain set of tests reveal something about causality.
The concept of Granger-causality which has been introduced by Granger (1969) is defined in terms of predictability and exploits the direction of the flow of time to achieve a causal ordering of associated variables.
3.5 Stability Test
13 | P a g e
polynomial are inside the unit circle. If it is satisfied, so the defined VAR model is stable.
3.6 Impulse Response Function (IRF) and FEVD (Forecast Error Variances Decomposition)
An impulse response function (IRF) measures the time profile of the effect of shocks at a given point in time on the (expected) future values of variables in a dynamical system (Pesaran and Shin, 1998). In other words, The IRF gives the nth-period response when
the system is shocked by a one-standard-deviation shock. On the other hand, forecast error variances decomposition is the proportion of nth-step ahead forecast error variance
of variable which is accounted for by the innovations in certain variable in the dynamic system. The idea of variance decomposition is to decompose the total variance of a time series into the percentages attributable to each structural shock.
3.7 Diagnostic Tests of the Residuals
Diagnostic tests are conducted to ensure that model does not lead to misspecification which also causes autocorrelation in the residual. There are several estimations in order to check the specification of the model. First, Portmanteau Autocorrelation Tests (Box-Pierce-Ljung-Box Q statistics) for residual correlation analyze existence of autocorrelation up to chosen lag. Second, LM Tests estimate existence of autocorrelation up to certain lag. Third, Normality Tests is the multivariate version of Jarque Bera tests. It compares skewness and kurtosis to those from a normal distribution.
4. Data
14 | P a g e
tend to fluctuate with increasing trend from 1990 to 2013. On the other hand, interest rate and inflation fluctuate in the certain range except in 1998 – 1999. It reflects the response of these variables to the Asian economic crisis.
15 | P a g e
Figure 6 Data Illustration
Source: Ministry of Finance of Indonesia, Indonesian Statistic Bureau and US Energy Information and Administration
Oil prices used in this analysis is nominal oil prices. Consumer and government are considered to respond nominal oil price rather than real oil prices because they are more visible (Kilian and Vigfusson, 2011). Furthermore, global oil price is chosen since Indonesian Crude oil Price (ICP) has the same trend so it can represent oil price variable. It is shown in Figure 7 which illustrates ICP and WTI trend from first quarter in 2005 to first quarter in 2013. The Indonesian government periodically sets the value of Indonesian crude oil price (ICP) as determination of economic oil cost in Indonesia. ICP is the average of crude oil price from all registered oil plants in Indonesia, which are about 52 plants in 2013. In addition, this study employs WTI rather than ICP as measure of oil price since oil price shocks traditionally have been associated with events in the global oil market (Kilian and Vigfusson, 2011).
16 | P a g e
Figure 7 ICP and WTI (USD/barrel)
Source: Indonesian Ministry of Energy and Mineral Resource and US Energy Information and Administration
Dummy variable of economic crisis in Asia covers 1998 and 1999. This estimation follows Hallward-Driemeier and Rijkers (2011) who use 1998 and 1999 as crisis year dummy for Indonesia. This variable is treated as exogenous variable in VAR estimation. Interest rate is used to analyze response of monetary authority to oil price shock. Oil subsidy spending is government policy used to responds oil price shocks. Government spending variable represents fiscal authority to oil price shocks and subsidy shocks. Real GDP represents economic growth. The period of research is from 1990:1 to 2013:4. This period is chosen in order to capture the influence of Syria crisis which leads to higher global oil price and also higher oil subsidy in Indonesia.
5. Analysis
5.1. Unit Root Test
In SVARX estimation, it is important to analyze the stationarity of data. There are two unit root tests conducted in this study, namely the Augmented Dickey–Fuller (ADF) test and Zivot-Andrews Test. Table 1 shows the ADF test results of all variables. LG_SA is log of government expenditure (seasonally adjusted), LGDP_SA is real GDP
17 | P a g e
(seasonally adjusted), INF is inflation rate, R is interest rate, WTI is global oil price, and SUB is oil subsidy spending. According to the ADF test, inflation and interest rate are stationary in level degree. Moreover, all of variables are stationary in the first difference.
Variables ADF t-statistic Lag McKinnon Critical Value Conclusion
1% 5% 10%
LG_SA 1.166299 0 -3.5007 -2.8922 -2.5832 Non – stationary LGDP_SA -0.143308 1 -3.5014 -2.8925 -2.5834 Non – stationary
INF -5.323537 2 -3.5022 -2.8929 -2.5835 Stationary
R -3.708154 1 -3.5014 -2.8925 -2.5834 Stationary
WTI -0.736858 2 -3.5022 -2.8929 -2.5835 Non – stationary
SUB 1.692875 10 -3.5092 -2.8959 -2.5852 Non – stationary
DLG_SA -10.37823 0 -3.5014 -2.8925 -2.5834 Stationary DLGDP_SA -7.638200 0 -3.5014 -2.8925 -2.5834 Stationary DINF -7.355116 2 -3.5030 -2.8932 -2.5837 Stationary DR -7.125761 0 -3.5014 -2.8925 -2.5834 Stationary DWTI -9.228033 1 -3.5022 -2.8929 -2.5835 Stationary SUB -3.967968 9 -3.5093 -2.8959 -2.5852 Stationary
Table 1 Unit Roots Test using ADF
18 | P a g e Variables ZA intercept Break Time ZA trend Break Time ZA Intercept and trend Break Time
LG_SA I(1) 1996Q1 I(1) 1998Q2 I(1) 1996Q1
LGDP_SA I(0) 1998Q1 I(1) 1998Q2 I(0) 1998Q1
INF I(0) 1997Q4 I(0) 1998Q2 I(0) 1999Q3
R I(0) 1997Q4 I(0) 1998Q3 I(0) 1999Q3
WTI I(1) 2008Q4 I(0) 1999Q2 I(0) 2004Q2
SUB I(1) 2010Q1 I(0) 2009Q4 I(0) 2008Q4
Table 2 Unit Roots Test using Zivot-Andrews (structural break) 5.2. Choosing Lag Length
As explained in section 3.2, there are some lag criteria which can be used to choose lag in SVARX estimation. In this case, lag chosen by Schwarz Information Criterion (SC) and Hannan-Quinn Information Criterion (HQ) are lag 3.
Lag LogL LR FPE AIC SC HQ
0 -1598.046 NA 9.55e+08 37.70220 38.89264 38.18156 1 -1246.389 598.2212 681455.4 30.44572 32.65654 31.33595 2 -1082.242 256.5975 36880.97 27.49981 30.73100 28.80092 3 -996.1226 122.7448 12336.22 26.34765 30.59921* 28.05962* 4 -953.3558 55.05612 11625.55 26.19209 31.46403 28.31494 5 -913.8458 45.41376 12459.51 26.11140 32.40372 28.64512 6 -855.6801 58.83431 9354.505 25.60184 32.91453 28.54644 7 -796.6299 51.58412* 7604.383* 25.07195 33.40502 28.42742 8 -748.5700 35.35437 9152.575 24.79471* 34.14816 28.56106
19 | P a g e
5.3. Granger Causality
According to Granger Causality test as illustrated in Table 4, Causality are appeared in the relationship between oil price and oil subsidy spending, between inflation and interest rate and, between inflation and economic growth (α = 15%). All of the variables jointly do Granger Cause economic growth. Table 4 provides information about one-way and two-way relationship between variables estimated. However, Granger Causality is not in line with reality. As a small open economy country, Indonesia cannot influence global oil prices. Therefore, restriction applied in the estimation is based on mechanism transmission illustrated in Figure 5.
Dependent variable: WTI (Global Oil Prices)
Excluded Prob. Explanation
SUB 0.0254 SUB does Granger Cause WTI
DLG 0.0605 DLG does Granger Cause WTI
DLG does Granger Cause WTI
All 0.0061 All estimated variables do Granger Cause WTI
Dependent variable: INF (Inflation Rate)
Excluded Prob. Explanation
R 0.0002 R does Granger Cause INF
LGDP_SA 0.0229 LGDP_SA does Granger Cause INF DLG does Granger Cause WTI
All 0.0022 All estimated variables do Granger Cause INF
Dependent variable: R (Interest Rate)
Excluded Prob. Explanation
INF 0.0000 INF does Granger Cause R
20 | P a g e
All 0.0000 All estimated variables do Granger Cause R
Dependent variable: SUB (Oil Subsidy Spending)
Excluded Prob. Explanation
WTI 0.0107 WTI does Granger Cause SUB
INF 0.1003 INF does Granger Cause SUB
R 0.1044 R does Granger Cause SUB
LGDP_SA 0.0210 LGDP_SA does Granger Cause SUB
All 0.0213 All estimated variables do Granger Cause SUB
Dependent variable: DLG (Government expenditure in percentage change)
Excluded Prob. Explanation
All 0.7408 All estimated variables do DLG Dependent variable: LGDP_SA (GDP Real)
Excluded Prob. Explanation
INF 0.0005 INF does Granger Cause LGDP_SA
All 0.0012 All estimated variables do LGDP_SA
Table 4 Granger Causality Results 5.4. Stability Test
21 | P a g e
Figure 8 AR Root Graph 5.5. SVAR
According to the model and matrix that already mentioned in the section 3, the restriction imposed in this SVARX is recursive restriction. In addition, I have included dummy for crisis and dummies for break time in order to capture structural change in each variable. SVARX is modified version of VAR which includes some exogenous variables. However it has already known that either VAR or SVARX estimation cannot be directly interpreted. Therefore, impulse response function and forecast error decomposition of variances are essential in SVARX analysis.
5.6. IRF and FEVD
The generalized IRF captures the dynamic responses of a variable of interest to innovations in the variable itself and in other variables. In the context of this study, the IRF analysis reveals the reaction of the macroeconomics variables to oil prices. In the impulse response figures, the horizontal axis shows the
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
22 | P a g e
timeframe, where one period represents one quarter. The vertical axis in Figure 9 shows the percentage change in oil price itself, inflation rate, interest rate, subsidy spending, government expenditure and GDP growth due to 1 percent change in oil prices shocks. Growth of GDP responds positively to a positive WTI shocks at first period and reaches negative point after six quarters. Furthermore subsidy spending responds positively to oil price shock during four quarters. Inflation and interest rate also responds positively to oil prices in the first period. The increased of oil subsidy spending and interest rate could be a response from fiscal and monetary authority because of increased of oil price shocks and inflation rate.
Figure 10 presents responses of all variables to subsidy spending shock. The vertical axis shows the change in all variables due to subsidy spending shocks expressed as a percentage. Government expenditure (in percentage change) responds negatively to subsidy spending shocks at first quarter, then it increases until it reaches peak in second quarter. It has persistent decrease after that quarter. Inflation does not have significant respond in the first quarter but negatively responds in the second quarter and it has increasing trend after second quarter. In addition, an unexpected shock in subsidy spending is associated with persistent increase in economic growth.
23 | P a g e
Figure 9 Responses of all variables to oil price shocks
-4 0 4 8
1 2 3 4 5 6 7 8 9 10
Response of WTI to WTI
-.6 -.4 -.2 .0 .2 .4 .6 1 2 3 4 5 6 7 8 9 10
Response of INF to WTI
-1.0 -0.5 0.0 0.5 1.0 1.5 1 2 3 4 5 6 7 8 9 10 Response of R to WTI -15,000 -10,000 -5,000 0 5,000 10,000 1 2 3 4 5 6 7 8 9 10
Response of SUB to WTI
-.03 -.02 -.01 .00 .01 .02 .03 1 2 3 4 5 6 7 8 9 10 Response of DLG to WTI -.006 -.004 -.002 .000 .002 .004 .006 1 2 3 4 5 6 7 8 9 10
24 | P a g e
Figure 10 Responses of all variables to subsidy spending shocks
-3 -2 -1 0 1 2 3 4 1 2 3 4 5 6 7 8 9 10
Response of WTI to SUB
-.6 -.4 -.2 .0 .2 .4 1 2 3 4 5 6 7 8 9 10
Response of INF to SUB
-1.0 -0.5 0.0 0.5 1.0 1 2 3 4 5 6 7 8 9 10 Response of R to SUB -4,000 0 4,000 8,000 12,000 16,000 1 2 3 4 5 6 7 8 9 10
Response of SUB to SUB
-.06 -.04 -.02 .00 .02 .04 1 2 3 4 5 6 7 8 9 10 Response of DLG to SUB -.004 -.002 .000 .002 .004 .006 .008 1 2 3 4 5 6 7 8 9 10
25 | P a g e
Lags Q-Stat Prob. Adj Q-Stat Prob. df
1 23.43843 NA* 23.69599 NA* NA*
2 61.79885 NA* 62.90887 NA* NA*
3 90.93448 NA* 93.02660 NA* NA*
4 135.6963 0.0000 139.8231 0.0000 36 5 170.2465 0.0000 176.3589 0.0000 72 6 220.6353 0.0000 230.2632 0.0000 108 7 265.6941 0.0000 279.0327 0.0000 144
Table 5 Residual Test
According to the result above, there is indication that autocorrelation exists after lag 4. It means there is no serial correlation at lag 3.
6. Conclusion
The objective of this paper is to investigate the impact of oil price shock and subsidy shock on the macroeconomic indicators. The macroeconomic variables investigated include inflation rate, interest rate, government expenditure, and economic growth. According to the ADF test, all of the variables are stationary in the first difference. In addition, Zivot-Andrews test which includes structural break indicates that all variables except growth of government expenditure are stationary in level degree. Since there are several dummies and variables should be included as exogenous variables, then the analysis is estimated by SVARX model.
26 | P a g e
government expenditure responds positively to subsidy spending shocks during two periods. In addition, an unexpected shock in subsidy spending is associated with persistent increase in economic growth. It means in the short run subsidy protects economic welfare which is represented by real GDP although it could distort the economy by raising inflation and government spending.
27 | P a g e
References
Alom, Fardous. 2011. “Economic Effects of Oil and Food Price Shocks in Asia and
Pacific Countries: An Application of SVAR Model.” Paper presented at the 2011 NZARES Conference.
Alshahrani, Saad A., and Ali J. Alsadiq. 2014. “Economic Growth and Government
Spending in Saudi Arabia: An Empirical Investigation.” IMF Working Paper WP/14/3.
Anand, Rahul, David Coady, Adil Mohommad, Vimal Thakoor, and James P. Walsh.
2013. “The Fiscal and Welfare Impacts of Reforming Fuel Subsidies in India.” IMF Working Paper WP/13/128.
Bernanke. Ben S, Mark Gertler, Mark Watson, Christopher A. Sims and Benjamin M. Friedman. 1997. “Systematic Monetary Policy and the Effects of Oil Price Shocks.”
Brookings Papers on Economic Activity 1:91-157.
Braun, Phillip A, and Stefan Mittnik. 1993. “Misspecifications in Vector
Autoregressions and Their Effects on Impulse Responses and Variance Decompositions.” Journal of Econometrics 59: 319-41.
Chung, Rae Kwon. 2012. “Fuel Subsidy Reform in Indonesia.” UN ESCAP.
CSIS Team. 2011. “Penyesuaian Subsidi BBM, Pilihan Rasional Penyelamatan
Ekonomi..” Centre for Strategic and International Studies.
Gottschalk, Jan. 2001. “An Introduction into the SVAR Methodology: Identification,
Interpretation and Limitations of SVAR models.” Kiel Working Paper No. 1072.
Granger, C. W. J. 1969. “Investigating Causal Relations by Econometric Models and
Cross-spectral Methods.” Econometrica 37 (3): 424–438.
Hamilton, James D. 1983. “Oil and the Macroeconomy since World War II.” Journal of
28 | P a g e
Hamilton, James D. 2000. “What is an Oil Shock?” NBER Working Paper 7755.
Hatemi-J, Abdulnasser. 2002. “Multivariate Tests for Autocorrelation in the Stable and
Unstable VAR.” Paper was presented at the 2002 Conference for Econometric Modelling in Pretoria, 3-5 July.
Holton, Christopher J. 2012. “What are the effects of fossil-fuel subsidies on growth, the
environment, and inequality?” University of Nottingham.
International Monetary Fund. 2013. “Energy subsidy reform: lessons and
implications.”
Johansen, Søren. 1988. “Statistical Analysis of Cointegration Vectors.” Journal of
Economic Dynamics and Control 12 (2–3): 231–254.
Kilian, Lutz. 2008. “Not All Oil Price Shocks Are Alike: Disentangling Demand and
Supply Shocks in the Crude Oil Market.” University of Michigan and CEPR.
Kilian, Lutz. 2009. “Oil Price Shocks, Monetary Policy and Stagflation.” University of
Michigan and CEPR.
Kilian, Lutz. 2010. “Oil Price Volatility: Origins and Effects.” WTO Staff Working
Paper, No. ERSD-2010-02.
Kilian, Lutz, and Robert J. Vigfusson. 2011. “Nonlinearities in the Oil Price-Output
Relationship.” Macroeconomic Dynamics, Cambridge University Press, vol. 15(S3):337-363.
Kilian, Lutz, and Robert J. Vigfusson. 2013. “Do Oil Prices Help Forecast U. S. Real
GDP? The Role of Nonlinearities and Asymmetries.” Journal of Business & Economic Statistics, Taylor & Francis Journals, vol. 31(1):78-93.
Koval, Igor Y. 2007. “Petroleum and Peso.” The University of Texas at El Paso. UMI
29 | P a g e
Oriakhi. D. E., and Iyoha Daniel Osaze. 2013. “Oil Price Volatility and its
Consequences on the Growth of the Nigerian Economy: An Examination (1970-2010).” Asian Economic and Financial Review. 3(5):683-702.
Ozcicek, Omer, and W. Douglas McMillin., 1999. “Lag length Selection in Vector
Autoregressive Models: Symmetric and Asymmetric Lags.” Applied Economics, 31, 517–524.
Pesaran, M. Hashem and Shin Yongcheol. 1998. “Generalized Impulse Response
Analysis in Linear Multivariate Models.” Economics Letters, Elsevier, 58(1): 17-29.
Pieschacon, Anamaria A. 2009. “Oil Booms and Their Impact through Fiscal Policy.”
Stanford University.
Jiménez-Rodríguez, Rebeca, and Sanchez, Marcelo. 2005. “Oil Price Shocks and Real
GDP growth: empirical evidence for some OECD countries.” Working Paper Series 0362, European Central Bank.
Shahbazi, Kiumars. 2013. “Investigating the Granger Causality Relationship between
the Consumption of Different Oil Products and Economic Growth in Iran.” International Journal of Accounting and Economics Studies, 1(3):74-83.
Waheed, Muhammad, Tasneem Alam, and Saghir Pervaiz Ghauri. 2007. “Structural
30 | P a g e
APPENDIX A
Plots of Estimated timimg of structural break by Zivot-Andrews procedure
Appendix A provides Zivot-Andrew Breakpoints graph representing structural break of the data. The vertical axis shows t-ratio, whereas the horizontal axis describes the estimation period.
1. Global Oil Price
32 | P a g e
5. Log of Government Expenditure in first difference
33 | P a g e
APPENDIX B
Forecast Error Variances Decomposition
Appendix B provides forecast error decomposition of variance. The vertical axis shows the contribution rate to each variable from all estimated variables, whereas the horizontal axis represents the lag periods of impact from innovation (unit: quarters).
Figure B.1
The contribution rate to global oil prices from all estimated variables
Figure B.2
The contribution rate to inflation rate from all estimated variables
Figure B. 3
The contribution rate to interest rate from all estimated variables
Figure B.4
34 | P a g e
Figure B.5
The contribution rate to government expenditure (in percentage change) from all estimated
variables
Figure B.6
