OIL PRICE RISK IN THE EUROZONE: A SECTORAL ANALYSIS

This study investigates how oil price movements impact the main Eurozone industry supersectors returns. We use a multifactor market model in which we incorporate oil price changes as an additional risk factor. In order to account for possible breaks in the relationship, we use the Bai and Perron (1998, 2003) breakpoints identification methodology. We find evidence of the presence of structural instabilities on the relationship between sector stock returns and oil price changes. Different breakpoints are identified, particularly the 2003 Iraq invasion year, the 2008 subprime crisis and the 2012 Euro debt crisis. Moreover, our results prove that stock return sensitivities to oil prices are time varying and sector dependent. Besides, the subprime financial crisis appears to induce a significantly positive effect on the oil-stock market nexus. However, the Euro debt crisis has a mostly negative effect. The other identified breakpoints do not seem to have any significant effect on the oil stock market nexus. Olfa Belhassine (Tunisia), Amira Ben Bouzid (Tunisia) BUSINESS PERSPECTIVES LLC “СPС “Business Perspectives” Hryhorii Skovoroda lane, 10, Sumy, 40022, Ukraine www.businessperspectives.org OIL PRICE RISK IN THE EUROZONE: A SECTORAL ANALYSIS Received on: 10th of May 2017 Accepted on: 6th of September 2017


INTRODUCTION
Oil price is an important economic data that policymakers, investors, speculators, firm managers, risk managers, portfolio managers… watch carefully.Accordingly, empirical investigation of the reaction of equity returns to oil price changes are important for accurately determining asset pricing models and forecasting the return and sensibility of stock markets.In this way, investors would be conscious of the oil price movements' effects on the risk and the value of their portfolios and this is particularly the case of international investors seeking for international diversification benefits.Several studies investigated the oil stock market nexus and confirmed the interaction between stock market and oil market movements ( 2013) that it is also sector dependent.Therefore, the aim of our study is to investigate how oil price movements impact the main Eurozone industry supersectors returns.The sectoral analysis in the case of the Eurozone is of particular interest.In fact, it will be valuable to investors by shedding light on the oil stock market relationship at the sectoral level.
The paper is organized as follows.Section 1 includes a review of relevant literature.Section 2 describes data used and preliminary statistics.In Section 3, we present the methodological framework of the study.Section 4 analyses and discusses the empirical results.Finally, we present some concluding remarks.Hamilton (1983) was the first to demonstrate the existence of a statistically significant correlation between oil shocks and some US recession prior to 1972.As an extension, subsequent studies addressed the issue of the relationship between oil market and stock market.Chen at al. (1986) found no statistical relationship between oil price movements and stock returns behavior.Sadorsky (1999) identified a significant negative short-term relationship between oil price volatility and aggregate S&P500 market index returns from January 1947 to April 1996.

LITERATURE REVIEW
Subsequent studies analyzed the effect of oil price fluctuations on the stock market of different countries.They all reached different conclusions and this was explained essentially by the status of the country (Park and Ratti, 2008…) and the origin of oil shocks (Filis et al., 2011…).More recently, other studies sustained that it is important to examine the oil/stock prices nexus in a sectorwise perspective.A general finding is a positive relationship linking oil market prices and Oil and Gas Sector returns (Faff and Brailsford, 1999 (2011) demonstrated that the non-Oil and Gas sectors are barely sensitive to oil price fluctuation.Elyasiani et al. (2011) showed that US market discriminates the effects of the oil price movements on the basis to whether an industry is an oil-substitute, oil-related, oil-user or financial from 1998 to 2006. Lee et al. (2012) showed that, at the industry level, the relationship between oil price shocks and some sector indices is statistically significant for some countries.
Several recent studies focused on the time varying feature of the oil stock market relationship Fan and Xu (2011) Lee and Zeng (2011) and Moya-Martinez et al. (2014), among others, showed that the oil/stock market relationship is time varying and that the long-run relationship is characterized by the presence of several breakpoints.Moreover, turbulent states and crisis are more likely to influence the oil/stock market relationship and that they can cause a total reversal in the relationship type.Zhu et al. (2014) demonstrated that the relationship between crude oil prices and Asia-Pacific stock market returns is mostly mild.This relationship was positive before the global financial crisis, except in Hong Kong, but it increased significantly as a consequence of the crisis except in the cases of Japan and Singapore.

DATA AND DESCRIPTIVE STATISTICS
The empirical investigation is based on daily sample set, which covers the period from January, 2 nd 2001 to August, 17 th 2015 for a total of 3674 daily observations.Our sample starting date is chosen so that data for the Eurozone and the exchange rate exist and are valid.Last three columns of Table 1 report the Spearmanrank correlation between the variables of interest.
Correlations between supersector equity returns and oil price returns and volatilities are all positive and highly significant, suggesting that for the study period the oil and the Eurozone stock prices have moved in the same direction.As expected, the Oil and Gas and Basic Resources supersectors show the strongest correlation with oil prices (34% and 25% respectively).Besides, correlations between the overall market index and supersector returns are positive and high on average.The low correlation among the independent variables (market return, oil price changes in one hand and market return, SOP in the other hand) indicates that multicollinearity is not a problem.However, SOP and oil returns are highly correlated suggesting that these two variables should not be used as independent variables simultaneously.

EMPIRICAL METHODOLOGY
The relationship between stock market and oil market can be modeled by means of a multifactor market model (Arouri, 2011;Phan et al., 2015...) in which oil market is included as an additional risk factor.
where From a globally diversified industry portfolio, the returns will be mainly influenced by the general Three GARCH-type models are estimated for each supersector, namely: GARCH (1,1) 1 (Bollerslev, 1986), GJR-GARCH (1,1) (Glosten et al., 1993) and EGARCH (1,1) (Nelson, 1991).Then the most appropriate univariate GARCH specification to each series is identified on the basis of Schwarz criterion (SCH), Akaike criterion (AIC) and the Log Likelihood criterion (LogL).The final step consists on re-estimating the best identified GARCH-type model for each supersector but this time accounting for the breakpoints 2 .The same methodology is carried out using the SOP as risk factor instead of the oil index return.

The oil stock market nexus
For each supersector, we further divide the full sample according to the identified breakpoints and then estimate equations ( 4) and ( 5) with the optimal GARCH type residual model for each of the sub-samples.All estimations present a high ² R adjusted and a near to 2 Durbin Watson indi- cating a good fit of the models.
The estimation results of the linear models using OILR and SOP as independent variables (Table 3) show that the scaled oil prices and the change in oil prices have the same effects (sign and significance) on stock market returns.Results also confirm that regardless of the supersector and the subsample estimated, all market coefficients (MR) are statistically significant and positive at the 1% level.
In addition, we can note that there is a time varying and a sector dependent effect of oil price changes.All supersectors are sensitive to oil price changes at least in one sub-period except for Real Estate and Technology supersectors.These sectors appear to not depend on oil price movements because they are not particularly This result is consistent with previous studies.
For the first three sub-periods ending in 2008, our results report a non-significant sensitivity for Foods and Beverages, Health Care and Chemicals indicating that these supersectors are not affected by oil price changes and volatilities.The same results were reported by Nandha and Faff (2008)  The second more sensitive sector to oil prices is Basic Resources.Arouri (2011) argued that the inflation caused by an increase in the price of oil is transmitted to other precious metal markets.This rise of the Basic Material prices will lead to higher profitability of the underlying firms.This result was also reported in Arouri (2011)  The more important negative sensitivity to oil market is recorded for Banks.For this sector, oil price increases has an indirect impact on the profitability of customers leading to a negative impact on volume and profitability of the banking and insurance businesses and other consumer businesses and therefore on the value of those stocks.Thus, diminishing their profits and dropping their stocks prices.Therefore, it is unsurprising to find that the increase of oil prices negatively impacts this supersector sensitivity to oil prices.Our results confirm the previous results found by Nandha and Faff (2008) for Travel and Leisure and Arouri (2011) for Telecommunication, Financials and Utilities.
For the 2008-2012 sub-period, we can note from Table 3 that several sectors' sensitivities to oil shocks switched from significantly negative or non-significant to highly and positively significant after 2008 (Health Care, Personal Household & Goods, Auto and Parts and Chemicals).A positive effect of the subprime crisis is also reported for the Media and Banks supersectors as their sensitivities' to oil price changes turned from highly and negatively significant to negative but non-significant.This positive effect of the subprime crisis was also reported in Broadstock et al, Mollick and Assefa (2013), Zhu et al. (2014) and Tsai (2015).The post subprime crisis is a turnaround period for the world economy.In that period oil prices augmented and restored investors' sentiment which led to an increase in stock market too.
For the after 2012 sub-period, Table 3 reveals that the Euro debt crisis has a negative impact on the oil/stock market nexus.In fact, for 5 Supersectors (Health Care, Personal Household and Goods, Chemicals, Financial Services and Retail) stock market sensitivities' to oil prices become, after 2012, highly negative and significant after being either positive and highly significant or non-significant.Other supersectors' sensitivities (Industrial G&S, Food and Beverages and Auto and Parts) turn to be negative but non-significant after being highly positive and significant.We also notice a drop in the positive and significant sensitivity to oil market for Basic Resources, and the global market index.This latter dropped from 0.25, in the post subprime crisis sub-period, to 0.07 after 2012.Besides, a higher negative and significant sensitivity is recorded for Insurance and Travel & Leisure.
There is evidence that the Euro credit crisis has affected the stock market sensitivity to oil market for several Eurozone supersectors.11 supersectors out of 14 affected by the Euro debt crisis register a sharp negative fall in the stock market sensitivity to oil shocks.Clearly, the Euro debt crisis has a strong negative effect on the oil stock market nexus in the Eurozone.The recession that hit the Eurozone since 2012 added to the sharp decrease in oil prices registered since mid-2014 could explain this negative effect on stock market sensitivities to oil markets.

CONCLUSION
This study investigates how oil price movements impact the main Eurozone industry supersectors returns.Our results show that the oil/stock market relationship is characterized by the presence of structural breakpoints.The number and the breakpoint dates differ among supersectors.The sectoral level analysis demonstrates that sector return sensitivities to oil prices are time varying and sector dependent.Before 2008, our results show strong significant and heterogeneous links between oil price changes and stock markets for most Eurozone supersectors.Besides, we report a positive response of supersectors' sensitivities to oil price changes during the 2008-2012sub-period.However, for the 2012-2015 sub-period, our results point to an overall negative response to oil price changes in some supersectors.
As every research has its own limits the work in this paper could be extended in several ways.It would be of a particular interest to study the asymmetric effect of oil prices as several studies documented that positive and negative oil prices shocks do not equally affect stock markets.
and Arouri et al. (2011) for European sectors.Negative and significant sensitivities are reported for Auto and Parts, Personal Household and Goods, Travel & Leisure, Telecommunications, Banks and Insurance for the entire sub-period and for Utilities before 2005 and Media between 2005 and 2008.

Table 2 .
Multiple breakpoints identification:Bai and Perron (1998, 2003)Test.The results of the procedure developed byBai and Perron (1998, 2003)is reported in this table.The effective sample size is 3674.A maximum of five breaks are allowed and a trimming parameter of 0.15 is used, so each segment has at least 735 observations. change. (AIC) and (SIC), indicate the optimal number of breaks according to Akaike Information Criterion and Schwarz Information Criterion.Break dates are selected based on the sequential procedure.*, **, and ***indicate statistical significance at the 10%, 5% and 1% levels, respectively.
The double maximum tests (UDmax and WDmax) test the null hypothesis of no structural break against an unknown number of breaks.The Sup ( )1 T F + is a sequential test of the null hypothesis of  structural change vs. the alternative hypothesis of a 1 +

Table 3 .
Results of sub-samples estimations with OILR as independent variable

Table 3 (
cont).Results of sub-samples estimations with OILR as independent variable This table presents optimal GARCH-type estimation results for the multifactor linear model (equation 4) using OILR as independent variable.Breaks reports the number of breaks selected by the sequential procedure by Bai and Perron at the 5% significance level.MR is the market sensitivity, OILR is oil price sensitivity.Coefficients of the multifactor linear model (equation 5) using SOP as independent variable are reported with SOP as the scaled oil price.*, **, and *** indicate statistical significance at the 10%, 5% and 1% levels, respectively.
and Moya-Martinez et al. (2014) for Foods and Health Care.Moreover, there is evidence of a positive sensitivity for 4 supersectors (Oil and Gas and Basic Resources for the full sample period and Industrial G&S and Financial Services) from 2003 to 2008.The highest oil sensitivity is observed for the Oil & Gas supersector (0.1768).This result is consistent with theoretical expectation as oil is a direct out-put for this supersector.This finding is in line with previous studies (Mohanty & Nandha, 2011; Moya-Martinez et al., 2014) among others.