Black Gold’s Price Plunge: Are Conventional and Interest Free Islamic Banks Equally Vulnerable?

Transcription

1. Black Gold ’s Price Plunge: Are Conventional and Interest Free Islamic Banks Equally Vulnerable? Ghulame Rubbaniy College of Business, Zayed University, PO Box 144534, Khalifa City, Abu Dhabi, UAE Ghulame.Rubbaniy@zu.ac.ae, Phone: +97 2 599 348 Osama El-Temtamy Bissett School of Business, Mount Royal University Calgary, Alberta, Canada otemtamy@mtroyal.ca Walayet A. Khan Schroeder Family School of Business Administration, University of Evansville, 1800 Lincoln Avenue Evansville, IN 47722, wk3@evansville.edu, Phone: 001-812-488-2869 Abida Perveen Department of Management Sciences Comsats University Islamabad, Sahiwal Campus, Pakistan abida.saima@cuisahiwal.edu.pk, Phone: +92 40 4305001 Keywords: Islamic Banks, Conventional Banks, Credit Risk, Insolvency Risk, Oil Price Plunge, Loan Loss Reserves, Withdrawal Risk JEL Codes: G21, G32, G33 _________________________________________________________________ 1
2. Black Gold ’s Price Plunge: Are Conventional and Interest Free “Islamic” Banks Equally Vulnerable? Abstract We investigate the effects of oil price declines on credit and insolvency risks for the banking industry within specific banking specializations (conventional, interest free “Islamic”, and conventional banks with Islamic windows), from 2000 through 2016, at both the aggregate and country levels in the Gulf Cooperation Council (GCC). Our findings show that falling oil prices increase credit risk for the banking industry, particularly for banks operating in Kuwait, Qatar, Saudi Arabia and the United Arab Emirates. However, Islamic banks’ credit risk was not affected. Utilizing both accounting-based and marked-based proxies for the insolvency risk, our analysis shows that oil price plunges do not increase insolvency risk of the banking industry regardless of bank specializations. We argue that government bailout packages contributed to these findings. Our findings will be of interest to many stakeholders including regulators who rely on empirical evidence to develop sound banking polices in similar oil dependant economies. 2
3. 1 . Introduction: Oil, or black gold, fuels global economic growth, and its changing prices have a significant impact on world economies. Mussa (2000), for instance, estimates that an increase of US $5 per barrel1 reduces global economic growth by 0.3% by the following year. In the backdrop of the oil price plunge of 2015, analysts and policy makers examined the consequences of the falling oil prices for oil exporting economies such as the GCC, Russia and Iran. They found the greater than 50% decline in oil price from its peak affected not only the oil companies but also the banks operating in oil dependent economies such as the GCC. The adverse impact of negative oil price shocks increases these banks’ vulnerability to credit and insolvency risks in two ways: Firstly, falling oil prices decrease governmental revenue while increasing the fiscal deficit, and in such a situation, governments seek banks to finance their fiscal budgets. In addition, a decline in oil prices decreases oil and gas sector revenue, depletes the value of oil company assets, and increases toxic debt for banks extending loans to these oil companies. The banking sector plays a critical role in any globally integrated economy, therefore any adverse shock to this sector can quickly produce systemic pressures for other sectors of the economy, generating crisis (for instance, the 2007-2008 global financial crisis). The impact of declining oil prices on the banking sector’s risk has largely been ignored in existing literature; for the most part, researchers simply report a significant influence of oil prices on stock market performance in both developing and developed countries (Alquist & Guénette, 2014; Basher & Sadorsky, 2006; Chen, 2014). We attempt to fill this 1 To simplify the currency risk, we convert all local currencies into US dollars at the exchange rate of the last day of every year of our sampled data. 3
4. gap by investigating the impact of oil price plunges on bank risk , particularly the credit and solvency of those operating in the GCC, and whether the impact varies across bank specializations. The GCC is one of the world’s main oil exporters, maintains the largest oil reserve (41% of global oil), possesses a significant share of the international oil market and 65% of OPEC oil; thus, any adverse shocks to oil prices most certainly would weaken economies in this region. For example, in October 2014, IMF chief Christine Lagarde reported that a decline of $25/barrel in oil price would reduce the GDP of GCC economies by 7% and continuing lower prices would increase the fiscal deficit of Gulf States. Gulf News (December 29, 2015)2 reported that prolonged lower oil prices would reduce government revenues and put high pressure on private sector financial institutions to finance investment projects. Figure 1 shows that oil prices remained at lower levels until the year 1999 but increased to $138/barrel in June 2008, resulting in negative consequences for global economies and stock markets (Arouri & Rault, 2012; Fayyad & Daly, 2011; Hamilton, 2009; Jones & Kaul, 1996; Papapetrou, 2001). However, oil prices plummeting to $37/barrel in December 2015 adversely affected oil dependent countries, particularly in the GCC region, whose main export was oil. A persistent oil price decrease can damage these oilexporting economies. Figure 1 illustrates the tightening financial conditions within these economies and plots the crude oil prices against oil and gas sector stock indices of World, GCC, US, Chinese, 2 http://gulfnews.com/business/economy/gulf-economies-embark-on-reforms-as-oil-prices-plunge-1.1645441 4
5. and emerging markets . Overall, a strong positive correlation exists between oil prices and oil and gas sector indices (See Figure 1), where falling (increasing) oil prices accompany these markets’ plummeting (surging) oil and gas sector indices. Figure 1: Log Crude oil prices and DataStream oil and gas sector indices from 2006-2016 1.1. Interest free “Islamic Banking”: This historically strong correlation between oil prices and the stock indices of oil and gas sector at international level may prompt a possible threat for financial institutions having high exposures to the energy sector (for instance worldwide outstanding loans of banks towards energy sector were around US$3 trillion in the year 2014). This high concentration of credit portfolios to the energy sector is likely to increase the credit portfolios risk for the banks in the GCC region where banks’ net incomes are driven by energy market developments (Husain et al., 2015). 5
6. There are few studies documenting the relationship between oil prices and the risks banks face , particularly in a setting where loan portfolios are concentrated towards the energy sector. In addition, almost no study exists which compares how the oil price plunge and banks’ risks vary across interest free or Islamic and conventional banks in the GCC region. Most Islamic banks operating in Islamic countries face a unique set of risks. These risks are unique to the interest free banking model used mainly by these banks that offer pure Islamic banking products. In addition, these risks are partially borne by conventional commercial banks that offer interest free banking products through their Islamic window branches. Some of these interest free Islamic products include but not limited to Mudarabah, Murabaha, Ijara, Musharaka, and Sukkuk. See appendix 1 for a bereif description of these products. Based on the interest free financial products discussed above it is clear that Islamic Banks have no autonomy over the projects they finance, nor can they demand collateral from customers in case of profit and loss sharing (PLS) Mudaraba accounts (Errico & Farahbaksh, 1998). Khan and Ahmed (2001) explain the withdrawal risk3 associated with PLS investment account holders (IAH) in both conventional and Islamic banks. This withdrawal risk compels Islamic banks to pay competitive returns to investment account holders irrespective of banks’ actual performance, which in turn forces shareholders to raise additional equity capital. Osoba (2003) suggests that despite withdrawal risk the depositors in Islamic banks remain loyal because of unavailability of other options on religious grounds. Farooq and Zaheer (2015) show that Islamic banks can perform 3 It is the risk of withdrawal of deposits from Islamic banks and then investing in conventional banks for higher returns. 6
7. relatively better than conventional banks during panics , as they are less prone to deposit withdrawals and less likely to cut lending during the down-market times. This loyalty factor helps to mitigate the adverse impact of withdrawal risk and reduces the probability of bank defaults. However, complexity of Islamic loans, moral hazard4 associated with PLS contracts, lack of collateral, and limited penalties on defaults increase these banks’ vulnerability to credit risks (Elnahass, Izzeldin, & Abdelsalam, 2014; Gropp & Vesala, 2004; Hamza & Saadaoui, 2013). While Islamic banks have long-term relationships with depositors allowing them to withstand higher losses, they simultaneously face operational limitations such as prohibition of interest, speculation or Gharar and gambling which, unlike conventional banks, reduces their credit portfolio optimization. Thus, they are more exposed to added insolvency risks as their credit is concentrated in the specific sectors and not well-diversified (Abedifar, Molyneux, & Tarazi, 2013). Since Islamic banks face different risks than conventional banks, we suspect that the impact of oil price plunges across conventional and interest free banking to vary as well. Given that the GCC region controls a significant share of the oil market and possess a unique concentration of energy sector loan portfolios, falling oil prices can destabilize the entire GCC region’s financial system. Therefore, insights about the relation between negative oil price shocks and banks’ vulnerability to these risks become more important in the GCC region. A financial stability report (2014) published by Central Bank of Oman 5, for example, documents that in their macro-financial stress testing analysis, oil price plunges lead to a decline in real GDP, which in turn negatively influences the banking Moral hazard problem is associated with manager’s unnecessary risk taking at the cost of depositors, as the investors are ultimate risk bearers. 5 http://www.cbo-oman.org/FinancialStability/CBOFSR20May2014.pdf 4 7
8. sector ’s default rates and capital adequacy ratio. Assuming risk characteristic may vary across conventional and Islamic banks operating in GCC countries, we formulate the following research questions to fill the above gap: (1) Do oil price plunges affect credit and insolvency risks of the banks operating in the GCC countries? (2) Does banks’ vulnerability to credit and insolvency risks vary cross-sectionally and across conventional and Islamic banks? We contribute to banking and energy finance literature in several ways. We are the first to examine the effect of falling oil prices on the credit and insolvency risk of banks both in the GCC region and at the individual country level. Second, existing financial market literature (Arouri & Rault, 2012; Effiong, 2014; Mussa, 2000) use crude oil prices, oil price shocks, oil price risk, and oil price increases as proxies of the oil price shocks. Contrary to mainstream literature, we follow Mork (1989) and use falling oil prices as a proxy of the oil price shocks, which is an extension of Hamilton (1983), who documents that considering oil price changes on both sides does not help to estimate the effect of oil price shocks. Finally, we investigate the effect of negative oil price shocks on bank risk across specializations, as Islamic and conventional banks employ different business models, and we believe that banks’ risk profiles differ across conventional and Islamic banks. Overall, we find that declines in oil prices significantly increase the credit risk of the banking sector in the GCC region, as well as commercial banks with Islamic windows. Aditionally, our country-level analysis shows that Kuwait, Qatar, Saudi Arabia and United Arab Emirates are particularly more prone to elevated credit risk under these conditions. However, using both accounting-based and market-based proxies of insolvency risk, we do not find evidence that negative oil price shocks increase the insolvency risk of the Gulf 8
9. banking industry . We argue that bailout packages presented by the wealth funds to the GCC banks is a probable reason for counter-intuitive results. The rest of this paper follows as: Section 2 explains the research design and methodology. Section 3 presents the empirical results and finally Section 4 concludes. 2. Research Design and Methodology: 2.1. The Data: The envisaged annual data for this study comes from three sources: Bankscope, Datastream and World Bank database. We extracted bank-specific data from Bankscope. Records on Historical crude oil (WTI) prices in USD/barrel, share prices, number of shares, short-term debt, long-term debt, stock volatility, and 3-months United States treasury bills annual data6 are from DataStream. Finally, macro-economic variables are from the World Bank database. Our sample includes nearly every Islamic bank, conventional bank, and commercial banks with Islamic windows operating in Gulf countries. We chose Gulf countries because of two reasons: (1) their dependence on oil revenues, as any negative oil price shock is likely to affect the financial sector and economic activities in these countries; (2) about one-third of the sampled banks are Islamic, which is a significant proportion to compare with the rest of the sampled banks. To capture the changes in economic and geopolitical conditions, we used the 16-year time span from 2000-2016, which is also marked with significant regional growth for Islamic and conventional financial institutions as well as for high oil price volatility. To 6 We used US Treasury bills because currencies in Gulf countries are pegged with the US dollar. 9
10. deal with outliers in computation of insolvency risk (Z-score) variable, we excluded extreme values of Z-score above 75th quintile. To make our results comparable with existing literature, we dropped insurance companies, investment and trust corporations, central banks, discontinued and merged banks, investment banks and duplicated banks from our data. We also corrected for banks specializations by accessing the websites of relevant banks, as the specialization of many of the banks is not correct in the Bankscope database. Our final sample includes unbalanced panel data of 291 observations for 21 commercial banks, 203 observations for 21 Islamic banks and 278 observations for 20 conventional banks with Islamic windows. Our total banks’ distribution includes the following observation data: 451 bank year observations from Bahrain, 351 from Kuwait, 155 from Oman, 165 from Qatar, 239 from Saudi Arabia and 464 from United Arab Emirates. Panel A of Table 1 shows the descriptive statistics of the variables used in this study. The average of our main explanatory variable, negative oil price shock (oil price decrease OPD) and (lagged oil price decrease L.OPD), is -0.13% and -0.14% with a standard deviation of 0.26% and 0.28% respectively. The average values of OPD and L.OPD clearly show the dominance of negative values of OPD’s computed observations, which has a negative skewness value of -2.36 and -2.19, and a lower value of standard deviation indicates limited spread of the OPD and L.OPD around the mean value. Following Abedifar et al. (2013), we use two proxies of credit risk: loan loss reserves ratio and non-performing loans ratio. The loan loss reserves ratio has an average value of 7.63%, a standard deviation of 11.62%, and a kurtosis value of 21.74%. The nonperforming loan ratio has a bit higher average value of 8.35% with a standard deviation 10
11. of 12 .68%, indicating slightly more variation in this proxy of credit risk. We measure insolvency risk using three alternative measures: Z-score, distance-to-default, and probability of default. Z-score shows a mean (median) value of 27.29 (40.70), suggesting that banks in the GCC region are not likely to bankrupt, as their Z-scores are very high. Our distance-to-default (DD) measure shows a mean (median) value of 3.71% (3.44%), indicating a lower insolvency risk for banks operating in the GCC region, as average distance-to-default is higher in these banks. Our probability of default as a measure of insolvency risk reports a mean value of 0.01% with standard deviation of 0.03%, demonstrating a low average probability of default for the banks in the GCC region. Following literature, we also control for bank specific characteristics, for instance bank size (Hughes, Mester, & Moon, 2001; Kane, 2010), loan growth (Dell'Ariccia & Marquez, 2006), total asset growth, cost inefficiency (Kwan & Eisenbeis, 1997), capital asset ratio (Kwan & Eisenbeis, 1997), and loan earning asset ratio. The controlled country specific factors include GDP per capita and GDP (Abedifar et al., 2013). We sort our data into sub-samples based on bank specializations. Table 1, Panel B depicts the descriptive statistics. For Islamic banks, the loan loss reserves ratio (as the proxy of credit risk) is 4.13% with a standard deviation of 12.58% and a positive skewness of 2.67, as compared to commercial banks, which yield a higher loan loss reserves ratio of 5.67% but lower standard deviation of 6.96%. The average loan loss reserves ratio for commercial banks with Islamic windows is 4.94%. These figures suggest that, due to the unique characteristics of Islamic banks (e.g. interest free operations), their default loan ratio is less than that of conventional banks because Islamic banks offer deposits on a PLS basis and do not charge interest on loans. 11
12. We also use non-performing loans ratio as the proxy of credit risk . Non-performing loan ratio averages for Islamic banks, commercial banks, and commercial banks with Islamic windows are 5.50%, 6.48%, and 5.22% respectively. These statistics support the view that non-performing loans are higher in commercial banks, as these banks advance loans without transferring losses to borrowers. For robustness purposes, we use three measures, Z-score, distance-to-default and probability of default, to estimate insolvency risk. Descriptive statistics in Panel B of Table 1 show that mean values of Z-score of Islamic banks, commercial banks, and commercial banks with Islamic windows are 28.74, 37.91 and 31.48 respectively. These values show that all three types of banks are unlikely to be insolvent, as Z-score values are very high. The distance-to-default measure of insolvency risk has an average of 3.99 for Islamic banks, 4.03 for commercial banks, and 3.42 for commercial banks with Islamic windows. These statistics show that insolvency risk is most (distance to default is lower) in commercial banks with Islamic windows and least in commercial banks. Finally, probability of default averages 0.01 across all bank specializations. 12
13. Table 1 : Descriptive Statistics Panel A: Descriptive Statistics of All Banks Variable Oil Price Decrease Oil Price Decrease Lag Loan Loss Reserves Non-Performing Loans Ratio Insolvency Risk-Z-score Ratio Insolvency Risk-DD Insolvency Risk-PD Bank Size Capital Asset Ratio Loan Earning Assets Loan Ratio Growth Total Assets Growth Cost Inefficiency Ratio LN GDP Per Capita GDP Per Capita Growth N Mean Median Skewness Kurtosis Std. Dev. 1825 1566 1407 1177 961 576 576 1812 1800 1625 1537 1713 1718 1808 1800 -0.13 -0.14 7.63 8.35 27.29 3.71 0.01 7.76 29.67 0.55 18.56 17.21 48.35 10.38 -0.58 0.00 0.00 7.59 9.23 40.70 3.44 0.00 9.26 38.56 0.74 25.52 24.10 51.28 10.78 2.41 -2.36 -2.19 3.96 3.72 0.65 1.62 7.54 -0.27 1.39 -0.61 6.50 7.69 8.78 0.17 -0.21 8.30 7.28 21.74 20.07 2.40 8.51 81.15 2.69 3.92 2.35 76.73 115.30 107.40 1.52 4.56 0.26 0.28 11.62 12.68 20.49 1.70 0.03 1.99 25.38 0.27 57.83 39.41 53.54 0.46 5.03 Panel B: Descriptive Statistics per Bank Specialization Islamic Banks Commercial Banks Variables Obs. Mean SD Mean SD Oil Price Decrease 203 -0.14 0.27 -2.21 7.49 291 -0.12 0.24 -2.39 Oil Price Decrease Lag 174 -0.15 0.28 -1.98 6.33 259 -0.12 0.25 Loan Loss Reserves Ratio 185 4.13 4.48 2.67 12.58 279 5.67 4.45 Non-Performing Loans 164 5.50 6.69 Ratio Insolvency Risk-ZScore 1.86 6.34 273 6.48 7.05 103 28.74 21.56 0.59 2.10 137 37.91 Insolvency Risk-DD 96 3.99 1.99 2.11 11.08 148 Insolvency Risk-PD 96 0.01 0.02 4.02 21.24 Bank Size 201 8.23 1.24 Capital Asset Ratio 201 20.51 16.83 Loan Earning Assets Ratio 200 0.63 0.20 Loan Growth 183 37.72 Total Assets Growth 186 26.29 Cost Inefficiency Ratio 194 53.49 LN GDP Per Capita 199 10.41 GDP Per Capita Growth 198 -0.39 Skewness Kurtosis Obs. Commercial Banks with Islamic Windows Obs. Mean SD 8.70 278 -0.13 0.26 -2.48 8.75 -2.26 7.85 254 -0.13 0.27 -2.34 7.94 1.82 6.96 270 4.94 4.26 2.71 12.50 2.56 10.96 261 5.22 5.16 1.95 7.04 18.95 -0.00 2.21 161 31.48 20.08 0.58 2.19 4.03 1.69 1.12 4.66 130 3.42 1.59 1.11 3.97 148 0.01 0.01 4.15 23.36 130 0.01 0.03 4.74 33.78 2.82 286 8.76 1.53 -0.57 3.49 273 8.91 1.35 -0.23 12.76 286 17.34 16.00 20.09 273 14.76 5.80 2.24 -0.75 3.58 285 0.61 0.19 3.62 270 0.66 0.14 85.76 5.02 35.92 276 16.41 28.24 23.54 265 15.88 18.30 39.95 3.34 18.10 277 13.49 24.64 56.04 268 16.96 40.26 35.16 5.23 47.93 282 34.12 12.94 43.22 272 39.01 14.32 0.47 0.19 1.59 286 10.47 0.44 -0.12 1.72 273 10.33 0.50 0.30 1.41 4.82 -0.97 4.10 285 -0.25 5.21 -0.16 4.42 272 -1.01 4.84 -0.71 3.50 -0.27 3.01 Skewnes Kurtosis s 4.18 -0.77 3.73 5.89 4.35 Skewnes Kurtosis s -0.54 1.12 2.02 13.19 2.91 5.75 12.5 187.98 6 3.15 The mean and standard deviation are in percentage form and skewness and kurtosis are in levels. Source: Self-construction from Bankscope and Datastream databases 13 18.41
14. Table 2 presents a pairwise correlation matrix and does not indicate any major issue of multicollinearity among independent variables . Most variables exhibit small correlation values. Moreover, the mean variance inflation factor (VIF) is also less than 10, which rejects the existence of multicollinearity in independent variables. Table 2: Correlation Matrix LLR Variables ZSCOR E DD PD L.OPD FSIZE CAR LEAR LGRO W AGRO W CINEF 1 LLR NPL ZSCORE DD PD L.OPD FSIZE CAR LEAR LGROW AGROW CINEF LNGDPC GDPCG NPL LNG DP C 0.86* 1 -0.24* -0.22* 1 -0.11* -0.11* 0.35* 1 0.07 0.09* -0.15* -0.41* -0.06* -0.08* -0.02 0.00 0.06 1 -0.38* -0.45* 0.23* 0.17* -0.15* 0.09* 1 0.23* 0.22* -0.14* -0.05 0.04 -0.02 -0.68* 1 -0.43* -0.38* 0.28* -0.10* -0.06 0.00 0.31* -0.38* 1 -0.12* -0.17* -0.00 -0.07 0.01 0.05 -0.03 0.02 0.02 1 -0.21* -0.25* 0.06 -0.04 0.00 0.05* -0.00 0.04 -0.01 0.43* 1 0.25* 0.30* -0.18* -0.01 0.02 -0.02 -0.21* 0.18* -0.14* -0.02 -0.12* 1 -0.06* -0.13* 0.00 -0.17* 0.02 0.01 0.05* -0.08* -0.00 0.06* 0.13* -0.10* 1 0.00 0.00 0.15* 0.10* -0.14* 0.04 -0.02 0.04* -0.05* -0.005 0.00 -0.07* -0.3 GDPCG 1 Pairwise correlation matrix in which * indicates significance at 5% level. In addition, our mean variance inflation factor (VIF) is less than 10 that also rejects the multicollinearity in independent variables. Source: Self-construction from Bankscope and Datastream databases 2.2. Research Methodology 2.2.1. Hypothesis Falling oil prices in energy markets reduce the revenues of oil-dependent governments and the profitability of oil and gas exploration companies unable to reduce oil extraction cost. As a result, in the financial sector, banks with loan portfolios concentrated in the oil 14 1
15. and gas industry are particularly hard-hit . Any reduction in oil prices is likely to adversely impact their earnings through lower interest income, reduced savings, and increased loan losses (Husain et al., 2015), thus increasing their credit risk. Kinda, Mlachila, and Ouedraogo (2016) support the view that negative shocks in commodity prices adversely affect the financial system. These shocks increase the nonperforming loans, reduce bank profits, and may cause potential banking crises by increasing the banking sector’s credit risks. The effects are particularly prominent in countries without proper governance mechanisms and those lacking a diversified export base. For instance, prolonged diminished oil prices may increase the number of nonperforming loans for GCC banks due to concentrated exposure to the oil and gas sector. In addition, since information asymmetry and moral hazards are more likely to occur in Islamic banks than conventional banks augmented with the additional risks unique to the interest free produces offered by Islamic banks we believe that the impact of an oil price plunge on credit risk may vary across the banks operating in the GCC region. Based on these premises, we develop the following hypotheses: H1: While oil price plunges generally increase banks’ credit risks, Islamic banks are more prone to this risk than conventional banks. A persistent fall in oil prices is likely to affect the revenue of both the oil companies and the governments operating in the GCC, which in turn is likely to decrease banks loan recovery and increase loan losses. The persistent decline in oil prices changes the outlook for hydrocarbon exporters, and, because of lower liquidity within the banking system, causes an increase in interbank rates in GCC countries. Further, credit rating agencies lower the ratings of oil exporting countries (e.g. Bahrain, Kazakhstan, Oman, 15
16. and Saudi Arabia). Banks in oil exporting countries mainly finance the government deficits; for example, in 2015, Saudi Arabia withdrew $106.7 billions from the Saudi Arabian Monetary Agency. On the other hand, commercial banks mainly finance the fiscal deficit in the United Arab Emirates. An adverse outlook for the hydrocarbon exporters may increase the default risk for banks (Sommer & Sommer, 2016). Based on these premises, we further hypothesize that: H2: Oil price plunges increase banks’ insolvency risk, and the magnitude of this effect varies across bank specializations and countries. 2.2.2. Empirical Model: We use panel regressions to test our above stated hypotheses. Our dependent variable is credit risk, while our main explanatory variable is negative oil price shocks. Our panel regression model is an extension of Abedifar et al. (2013), and incorporates oil price decrease as a main explanatory variable. The oil price decrease proxy is proposed by Hamilton (1983), and the negative oil price shock is computed as follows: ∆ log(