Oil Price Shocks and Sectoral Stocks Behaviour in Nigeria: How Relevant is Asymmetry and Structural Breaks?


 In this paper, we model the relationship between oil price and stock returns for selected sectors in Nigeria using monthly data from January 2007 to December 2016. We employ both the Linear (Symmetric) ARDL by Pesaran et al. (2001) and Nonlinear (Asymmetric) ARDL by Shin et al. (2014) and we also account for structural breaks using the Bai and Perron (2003) test that allows for multiple structural changes in regression models. Our results indicate that the strength of this relationship varies across sectors, albeit asymmetric and breaks. We identify two structural breaks that occur in 2008 and 2010/2011 which coincidentally correspond to the global financial crisis and the Arab spring (Libyan shut-downs), respectively.Moreover, we observe strong supportfor asymmetry and structural breaks for some sectorsin the reaction of sector returns to movement in oil prices.These findings are robust and insensitive when considering different oil proxy.While further extensions can be pursued, the consideration of asymmetric effects as well as structural breaks should not be jettisoned when modelling this nexus.JEL codes: C22; C51; G12; Q43


Introduction
Despite the recent global move to switch from the major use of crude oil to other energy forms, especially renewables, its use and price have not ceased to be relevant in driving the economies of most countries. Since the empirical discovery that oil price changes matter in the driving of economic activities following the various demand and supply shocks that hit the world crude oil market in certain years (such as the 1970s and early 1990), assessment of oil price link with certain macroeconomic indicators appears unending in the economic literature. The reason for this continuous quest is not farfetched. Firstly, crude oil appears to have direct or indirect significant influence on almost all sectors of the economy (especially in oil-dependent countries), such that any instability in its price poses severe to the economy. This has warranted the need to often discover other plausible indicators apart from GDP that can be affected by oil price changes. Such indicators include inflation, interest rate, agricultural prices, stock prices, bank performances, etc. (see Cologni and Manera, 2008;Darby, 1982;Abbott, et al., 2018;Lee and Lee, 2018, inter alia). Secondly, mixed reports herald the literature concerning oil-macroeconomic indicators nexus, following different oil price shocks.
For instance, studies like Jones and Kaul (1996) and Darby (1982) discover negative relationship,  support positive relationship, while Blanchard and Gali (2007) and Davis and Haltiwanger (2011) find neutral effect.
On this basis, oil-stock nexus has been followed by a significantly large focus. Various transmission channels have been put forth concerning the pass-through of oil price into stock market behaviour. The traditional mechanism is that oil price can directly drive stock prices through its influence on the future cash flows, or indirectly through the rate of interest upon which most future cash flows are discounted (Jones and Kaul, 1996).
Thus, production cost rises, and subsequently, firms' earnings and stock prices are impacted. Secondly, coming from the angle that inverse relationship exists between stock prices and discount rate, increased oil price raises nominal interest rate through increased expected price level. Thus, stock prices collapse at the increase in interest rate. This is the interest rate channel (see Rafailidis andKatrakilidis, 2014 andRatti, 2009).
From another perspective, Fowowe (2013) explains that the stand of the firm as an oil user or producer matters in determining whether oil price shocks will affect stock prices through its initial influence on cash flows. Stock returns of oil consuming firms are expected to be adversely affected by a rise in oil price, and otherwise for oil-producing firms (Mohanty et al., 2011). This is because, for oil consumers, high oil price raises inflationary expectations and nominal interest rate. This further increase cost, penalizes investments, reduces firms' earnings and dividends, and thus causes adverse effect on stock prices. For oil producers, oil price increase is a gain as it boosts earnings and dividends. The resultant outcome is increase in stock prices.
Meanwhile, it is likely on the background, but unrevealed, understanding that the effect of oil price changes on stock prices depends on the status of a firm as an oil consumer or producer that Kilian and Park (2007) also suggest that the source of oil price shock (either demand-or supply-side) is also critical in the response (either positive or negative) of stock prices. This is justified by the fact that demand-and supply-side shocks are mainly associated with oil consumers and producers respectively. As disclosed in their analyses, the observed adverse response of stock returns of the US to oil price is due to demand shocks caused by the lack of surety of sufficient supply of future crude oil, while the supply shock which was as a result of the unexpected global expansion led to positive impact on stock returns. This conclusion partly aligns with the findings of Mohanty et al. (2011) that oil prices impose negative influence on stock returns for oil consumers, like the USin this case.
Apart from the issue surrounding the lack of consensus, another interesting puzzle arose among the studies that conclude significant effect, and that is whether the response of stock prices to oil price shocks is symmetric or asymmetric. The need to determine this spur from how large or small the risks or gains of asymmetric tendencies in the response of stock prices to oil price shocks may be to investors. Empirically, while Reboredo and Ugolini (2016) and Reboredo and River-Castro (2014) suggest symmetric effect, Bouri et al. (2016) and, Soucek and Todorova (2013) support asymmetric effect.
However,  lend support to both sides.
Against this background, we revisit for Nigeria, the possible relationship-either significant or not, and if significant, either symmetric or asymmetric-between oil price shocks and stock prices. Examining this nexus for Nigeria is particularly of utmost importance for certain reasons. The Nigerian Stock Exchange (NSE) has recorded a phenomenal growth lately, as high economic performance is driven by the rise in initial public offering (Fowowe, 2013). With this her outstanding economic activities, she has maintained a giant stand in the African continent. Hence, since the stock market's response to oil price shocks may undoubtedly affect her real economy, other African countries are consequently at the mercy of such effect (Gil-Alana and Yaya, 2014). Therefore, we make certain additions to the stock of existing knowledge. Firstly, majority of the studies carried out for Nigeria considered the aggregate stock market through the use of the All Share Index as proxy for stock prices (see studies like Fowowe, 2013;Gil-Alana andYaya, 2014 andAdaramola, 2011). To the best we know, the comprehensiveness of this study ranks it highest among the few notable studies in Nigeria that consider sectoral analyses, as it is the first to capture as high as ten individual sectoral stock prices in a single study. This does not only enable clear-cut policy by investors having understood the heterogenous nature of each sector, but also leads to optimization of cross-sector investment decisions and possible gains.
Secondly, we take a departure from the common methodology engaged by most studies. Rather than the GARCH-type and SVAR models often adopted (see Gil-Alana and Yaya, 2014;Arouri et al., 2011;Zheng and Su, 2017;Filis et al., 2017, inter-alia), we consider the non-linear Autoregressive Distributed Lag (NARDL) proposed by Shin et al. (2014) which proves its superiority by accounting for both short-and long-run asymmetries in each sectoral analysis. This is achieved by the positive and negative partial sum decompositions of the oil price. The NARDL approach has some intrinsic worth as it allows modelling the cointegration relation that could exist between the endogenous and exogenous variables and more especially testing both the linear and nonlinear cointegration. These aforementioned merits of the NARDL approach may also be valid for nonlinear threshold Vector Error Correction Models (VECM) or smooth transition models; however, these models may suffer from the convergence problem due to the proliferation of the number of parameters which is unlike the NARDL model. In all, unlike other error correction models where the order of integration of the considered time series should be the same, the NARDL model relaxes this restriction and permits combining data series having different integration orders (see inter alia, Shin et al. 2014).Meanwhile, the short-and long-run symmetric models will also be estimated in order to test if asymmetry matters. This also appears to be the first notable study to consider this approach for oil price-stock nexus from the perspective of sectoral consideration in Nigeria.
Foreshadowing our main results, the strength of the relationship between oil and stock varies across sectors. However, these responses are positive across board, largely stronger in the long-run albeit asymmetry. On the other hand, only two sectors are significantly affected by structural breaks, each sharing the symmetric and asymmetric models. Their results are largely positive and stronger in the long-run too, except the negative impacts of the breaks in their respective periods. While further extensions can be pursued, the consideration of asymmetric effects as well as structural breaks should not be discarded when modelling this nexus.
We present the remainder of this paper as follows. Second 2 opens the discoveries of past studies, Section 3 gives the methodology and data description. In Section 4, we render the presentation and discussion of results including the robustness checks, and Section 5 highlights the policy implications and concludes the paper.

Review of Relevant Literature
The stock market being a very important sector in the analysis of economic activities has attracted great attention coupled with huge empirical research in examining the relationship between oil price and stock markets (Ghosh and Kanjilal, 2016). Overtime, the major interest in this study has basically been on global and national scales respectively. While studies like Jones and Kaul (1996), Lean and Badeeb (2017), Bastianin et al. (2017), Broadstock and Filis (2014), Dutta (2017), Ftiti (2015, Gupta An innovative study by Jones and Kaul (1996), using quarterly data for Canada, Japan, the UK, and the US between the period 1947 and 1991 find that oil price has a negative effect on aggregate stock returns. Lean and Badeed (2017) uses the Non-linear Autoregressive Distributed Lag (NARDL) to study the asymmetric impact of oil price on Islamic sectoral stocks using monthly data between January 1996 until June 2016.
The study shows that there are weak linkages between the changes in oil price and the Islamic composite index. However, the responses of the Islamic real sectors indices associated with negative changes in the oil price are higher than those associated with positive changes in oil price in the long-run. In a bid to find evidence in the relationship between oil price shocks and stock market returns in the United States and China, Broadstock and Filis (2014) adopts monthly data ranging from January 1995 to July 2013, using a Structural Vector Autoregressive (VAR) model. The study finds that the effects of oil price shocks differ widely across industrial sectors as well as the fact that China is seemingly more resilient to oil price shocks than the US.
Interestingly, Bastianin et al. (2017) focuses on the impacts of oil price shocks on stock market volatility in the G7 countries using a Vector Autoregressive (VAR) model, finding that stock market volatility does not respond to oil supply shocks. On the contrary, demand shocks impact significantly on the volatility of the G7 stock markets. Dutta et al. (2017) examines the impact of oil price uncertainty on Middle East and African stock markets using a GARCH-jump model. The study records that stock returns are sensitive to the fluctuations in the implied oil volatility index and that timevarying jumps do exist in the stock returns. Gupta (2016) analyses oil price shocks, competition, and oil & gas stock returns, using a comprehensive firm-level monthly data from 70 countries spanning from 1983 to 2014. The study uses panel Ordinary Least Squares and finds that oil price shocks positively impact firm-level returns. Khalifa et al. (2017) models the causes and consequences of energy price shocks on petroleumbased stock markets in the Asian countries using daily data from July 18th, 2006 to July 30th, 2015. The study uses Spillover Asymmetric Multiplicative Error Model (SAMEM) and finds that recent global financial crises and the geopolitical instability have a significant impact on the selected petroleum-based stock markets.  revisits the oil price and stock market nexus using a Nonlinear Panel ARDL approach and finds that stock prices of both oil exporting and oil importing groups respond asymmetrically to changes in oil price. Silvapulle et al. (2017) models the non-parametric relationship between crude oil and stock market prices in net oil importing countries and find that stock market fundamentals play a significant role in determining the oil-stock price relationship. Using Copula models, Kayalar et al. (2017) model the impact of crude oil prices on financial market with daily data ranging from January 10, 2005 to April 6, 2016. The study finds that most oil exporting countries show higher oil price dependency, whereas, emerging oil importing markets are less vulnerable to price fluctuations.
Using wavelet approach, Ftiti et al. (2015) studies the oil price and stock market comovement in the G7 countries, during the period 1998 to 2013. The study finds that stock markets are more sensitive to oil shocks originating from demand shocks. Huang et al. (2017) examines the simultaneous response of ten selected sectors to oil price shocks between January 2000 to January 2016 at a daily frequency. The study finds that each sector may lead or lag behind other sectors in different frequencies to move with an oil price shock, but transportation, utilities and consumer discretionary are sectors have higher probability to lag behind other sectors, while materials and telecommunications are the sectors with higher possibility to lead other sectors.
In a bid to reveal the Jump dynamics in the relationship between oil prices and the stock market, Fowowe (2013)  More recently, Benkraiem et al. (2018) examines the insights into the US stock market reactions to energy price shocks using the recently developed Quantile Autoregressive Distributed Lags (QARDL) model by Cho et al. (2015). The study finds a negative longand short-run relationship between WTI crude oil and Henry Hub natural gas prices on the one side and S&P 500 stock prices on the other side, only for medium and high quantiles.
The results from the previous studies are intuitive, however, most of the studies done in Nigeria have not empirically examined the effect of (i) oil price shock on sectoral stock returns with structural breaks, (ii) asymmetry effect of oil price shocks on sectoral stock returns, which is a major thrust of this study. By this, the policy relevance of this research work in Nigeria is carried out. Hence, this study is an unassuming attempt in this regard to fill the gap.

Data description
This paper covers the oil price (OP), interest rate (INT) and returns of ten (10)  Data on oil price is obtained from the US Energy Information Administration (EIA) website (http://www.eia.doe.gov), whereas data on the US CPI is obtained from the Organisation for Economic Co-operation and Development (OECD) statistical database.

Methodology
The motivation for this paper is in two folds. It employs the recently developed NARDL model of Shin et al. (2014) to examine the short run and long-run asymmetrical effects of oil price shocks on sectoral stocks in Nigeria. Also, we modify the Shin et al. (2014) to account for structural breaks in the model as there appears to be evidence of some notable shifts in the series (see, Figure 1). Disregarding these breaks when they exist may bias regression results (see, inter alia, Salisu and Fasanya 2013). The NARDL model is an asymmetric expansion of the linear ARDL model of Pesaran et al. (2001), which is a single cointegration and error correction approach. The NARDL share some similarities with other models such as the nonlinear threshold Vector Error Correction Models (VECM) or smooth transition models but these models may suffer from the convergence problem due to the proliferation of the number of parameters. The other advantages of using the NARDL approach are well documented in the works of Van Hoang et al (2016) and Nusair (2016).
For the purpose of robustness, however, we consider both the linear (symmetric) ARDL and non-linear (asymmetric) ARDL (with and without breaks). Since the NARDL model is a non-linear expansion of the symmetric ARDL model of Pesaran et al. (2001), it is useful to start by presenting the symmetric ARDL model. The Linear ARDL without structural breaks can be written as: where is the stock returns of various sectors (i), and the logarithm of real oil price. Wald test (F statistic) by imposing restrictions on the long-run estimated coefficients of one period lagged level of op, INT and sr i to be equal to zero, that is, the null hypothesis of no cointegration states that 0 : 1 = 2 = 3 = 0, is tested against the alternative hypothesis of 0 : 1 ≠ 2 ≠ 3 ≠ 0. Then the calculated F-statistic is compared to the tabulated critical value (Pesaran et al, 2001 However, to restore equilibrium immediately may not be possible because of the speed of adjustment. This could be caused by the lags and adjustment process used to capture changes in any of the factors affecting oil price or sectoral stock returns overtime. Hence, the error correction model can be used to capture the speed of adjustment which defined as in the model (2) expressed below: We extend the model in equations (1) and (2) to include endogenous structural breaks.
The model is then specified below: Following the previous definition of the earlier variables, the breaks are captured in equation (3) (5) breaks in the regression model and is therefore considered a more general framework for detecting multiple structural changes in linear models. We also test for the existence of long run relationship in the presence of structural breaks using the ARDL test. In essence, we are also able to determine long run and short estimates for the real oil price -sectoral stock returns in the presence of structural breaks. In addition, the results obtained are compared with those from equation (1) to see if accounting for breaks in the regression is necessary.
Subsequently, the Wald test is used to test for the joint significance of structural breaks in equation (3). That is, we test . The rejection of the null hypothesis implies that the breaks are important and should be included in the model, hence, suggesting the adoption of equation (3) while the non-rejection implies that structural breaks do not matter in the symmetric case.
To examine the role of asymmetries in the model, cointegrating NARDL model of Shin et al. (2014) that accommodates the potential short-and long-run asymmetries would be of great interest. In actual fact, this model uses the decomposition of the independent variable into its positive oil price changes and negative oil price changes. This is premised on the fact that economic agents may respond differently to positive and negative changes in oil price. The decomposed oil price (op) Δ + and negative Δ − partial sums for increases and decreases such as: Given the definitions in (4)-(5), Shin et al. (2014) show that the linear ARDL model (1) can be modified to account for asymmetries to produce the following nonlinear ARDL model: Equation (6) can be rewritten to include error correction term as: Introducing structural breaks into the NARDL framework, we extend equation (6) to include the relevant break dummies.
The definitions of the parameters still follow the sequence of the earlier models. We also conduct structural break test to ascertain the significance of including the breaks in the NARDL model. Besides, we also use the F-distributed Bound test to confirm the presence of long run relationship and the Wald test was equally used to verify the role of asymmetry in the presence of structural breaks.

Preliminary Results
Following the specification of the relevant models in the early section, this section entails the presentation of their empirical results and discussion. The properties of the series are firstly described through certain statistical tools and graphical illustrations, followed by the tests of stationarity. We stretch the unit root test in two folds. The first set (Augmented Dickey-fuller (ADF) and Philip-Perron (PP) test) do not consider structural breaks while the second (Perron-Vogesland test) accounts for this possibility along the time path of the variables. Finally, either the symmetric or asymmetric ARDL models or both are analysed depending on the conclusion of the Wald test. We further characterize the joint movement between oil price and stock returns by plotting the latter against the former, as provided in Figure 1. Except for the stock returns of the agriculture sector, there seems to be a positive correlation between oil price and the returns of all other sectors; it seems especially stronger for the conglomerate, construction and consumer goods and services sectors. The interaction just appears to be mixed along the way for other sectors, thus confirming the tendency of the difference in the reaction of the stock markets of different sectors to shocks in oil price. Put differently, due to the peculiar characteristics of each sector, the extent to which oil price induces their returns may vary. With this outcome being a symmetric assessment, whether or not each sector's stock returns asymmetrically respond to oil price cannot be uncovered by the graphical analysis. The next section reveals this.
Lastly under the preliminary analyses, we subject all the series to the test of stationarity following the standard practice for time series data. Two sets of unit root are employed.
Included under the first are Augmented Dickey-Fuller (ADF) and Phillip-Perron (PP) unit root tests which do not account for structural breaks along the time path of the variables. The second type accounts for structural breaks, and considered under this category is the Perron-Vogesland (PV) unit root test. Presented in Table 2,and following the null hypothesis of both test types that series is not stationary, their results reveal mixed integration order. Specifically, oil price exhibits non-stationarity until it is subjected to first difference (i.e. I(1)) while all the stock returns observe stationarity in their level forms (i.e. I(0)). The fractional integration of the variables justifies our decision to engage the Autoregressive Distributed Lag (ARDL) model for proper estimation. Meanwhile, the break date for each variable is also reported in the same Table 2.

Oil Price-Sector Returns (without Structural Breaks)
Conventional for time series data within the framework of the ARDL model is to determine whether long run relationship is evident between the variables under consideration using the ARDL Bounds cointegration test. Given the null hypothesis as the absence of cointegration, the value of the F-test being greater than the upper bound value of the Bounds test at the maximum of 10% significance level (lower part of Tables 4 and 8), there is evidence of long run relationship between oil price and each of the sectoral stock returns. Also, in order to determine if asymmetries matter in the oil-stock nexus of each sector, we carry out the Wald test with the results shown in Table 3. It is discovered that there is need to account for asymmetries in only six sectors which are agriculture, consumer goods, conglomerate, financial services, health and general services. Therefore, the asymmetric results are presented for the six sectors while the symmetric results are provided for others. Table 4 presents the results without the inclusion of structural breaks. For the symmetric case, oil price appears not to be a significant factor driving the stock returns of both natural resources and oil and gas sectors in both long run and short run, while it really matters for both construction and industrial sectors in both periods. Turning to the asymmetric results, both positive and negative oil price changes influence the behaviour of the stock market, as all the sectoral stock returns are except the general service sector. Meanwhile, this significant influence cuts across both the short-and long-run. This reveals the strength of asymmetries in relation to oil-stock nexus of most of the sectors of Nigerian economy.
Also, two things are further discovered from the results. Largely, the effects are positive for the significant sectors, thus indicating that higher oil price is a plus for existing or potential investors in these sectors. In other words, as oil price increases, a positive effect is laid on the stock returns of the investors. This positive relationship between oil price and the stock market is confirmed in the literature by prominent studies including , Apergis andMiller (2009), Bouri et al. (2016), and Kilian and Park (2007). The reason for this is not far-fetched. Nigeria is an oil exporting country and this status makes her a net gainer of the increase in oil price as her revenue increases. Over time, this tends to stimulate the activities of the domestic economy through a higher investment in both financial and physical assets .
The second noticeable fact about the results relates to the magnitudes of the effects across both short-and long-run periods. The impact of oil price on the sectoral stock returns is stronger in the long run, irrespective of whether the model is symmetric or otherwise. Of the seven sectors whose stock returns are significantly impacted by oil price in both short-and long-run, only financial services sector has its short-run estimates impact higher than the long-run's. We can conclude from this that, although economic activities are immediately aroused following increased inflow of revenue from increasing oil price, the sustained investments in both financial and physical assets tend to yield higher returns in the long-run. In fact, it is a general knowledge that most long-term investments tend to give higher returns than the short-term investments.
Again, it is interesting to note that for all the significant asymmetric models, increase in positive oil price changes increases stock returns more than the increase in negative changes in oil price does, except for the agricultural sector when the reverse holds.
Expectedly, interest rate is a strong determinant of the stock returns of all the sectors in the short-run, while its significance is mixed in the long run.

Oil Price-Sector Returns (with Structural Breaks)
Having discovered that not accounting for structural breaks along the time path of variables that operate on high frequency can produce bias results, this study further accounts for the significance of certain structural factors that could have altered the oilstock relationship. Under this section, we first determine the break dates following the approach of Bai-Perron (2003) for the asymmetric model. The results are presented in Table 5. We find that only the financial, health, industrial and natural resources sectors have breaks. The break dates are respectively reported in Tables 5, and they centre around the global financial crisis of 2008 and the positive oil price shock of 2010/2011.We therefore carry out the Wald test to determine if asymmetry really matters having considered structural breaks. As shown in Table 6, only the financial services sector supports asymmetric relationship between oil price and stock returns among the four sectors. Furthermore, since the three other sectors go for symmetric relationship, we also use the Bai-Perron (2003) test in a linear version to determine if they observed structural breaks within the years under consideration in this study.
Only natural resources sector has a break in December, 2008 (see Table 7).
Therefore, since the Wald test proves that only the asymmetric model of the financial services sector matters (see Table 6), and the symmetric Bai-Perron (2003) test reveals that among the three other sectors that support symmetry, only the model for natural resources has significant break (see Table 7), we report the results of the oil price-stock returns relationship of the two sectors only, having made empirical provision for structural breaks. As seen in Table 8, oil price is still a significant determinant of stock returns even when structural breaks have been modeled in the regression analysis. Like above, the symmetric and asymmetric impacts of oil price is still positive for natural resources and financial services sectors respectively, but it is also greater in the long run for both cases. The positive effect is also greater for positive oil price changes in both short-and long-run for the financial services sector that supports asymmetry.
Furthermore, we discover that the structural breaks associated with the two sectors prove significant on the stock returns of each one at different periods. For instance, the structural break of the financial services sector affects its stock returns negatively in the short run, while that of the natural financial services sector also yields adverse impact on its returns, but in the long-run only. Therefore, since the structural breaks respectively occur in 2010M06/2011M11 and 2008M12 (see Tables 5 and 7), they coincide with the oil price shock and global financial crisis of the respective years.
Specifically, these external world events singularly reduce the stock returns of their respective sectors by 35.78% (short-run) and 13.57% (long-run).This tendency is confirmed by the sudden falls in the trends of the stock returns of the financial services and natural resources sectors after their respective spike in April, 2010, andAugust, 2008.Meanwhile, interest rate expectedly drives the stock returns of each sector positively in both short-and long-run.

Robustness Check
This study further assesses the robustness of the regression models in order to determine whether the estimates are sensitive to different variable measurement. To do this, we carry out another round of estimation with and without breaks having replaced the Brent oil price with the West Texas Intermediate (WTI) oil price as a new proxy for the global oil price. For the sake of space, we report only the regression results, as given in Tables 9 and 10, but their associated Wald and Bai-Perron (2003) tests for the determination of the significance of asymmetries and structural breaks respectively are available upon request. As seen in Table 9 which reports the regression without breaks, there are relative changes in the responses of some sectors' stock returns to asymmetries compared to the original results reported in Table 4. However, it is discovered that the stock returns of natural resources, oil and gas, and general services sectors are unresponsive to the WTI oil price as is the case of the original model. In addition, the signs are largely positive for the significant estimates. This sums up to indicate that our regression estimates when structural breaks are not put into consideration are robust to different oil price measure.
For the other scenario, six sectors are sensitive to structural breaks when WTI oil price is made a proxy for world crude oil price. This is unlike the case of the original model, although the two sectors (financial services and natural resources) under the Brent oil price still show sensitivity to breaks when WTI oil price is used. The summary of the results is presented in Table 10. Their variable estimates are mostly significantly positive, while their breaks are largely negative and significant, thus implying that the regression estimates are sensitive to different measure of oil price given the role of structural breaks.

Symmetry Asymmetry Symmetry Asymmetry Asymmetry Asymmetry Asymmetry Symmetry Symmetry Asymmetry
Long-run Estimates     Beyond the aggregate measure of the stock market performance, it is discovered that there may be significant differences in the manner with which the global crude oil market narrows down to influence sectoral stock market indicators. Each sector has its own peculiarity, thus making their responses to external indicators, such as oil price

Conclusion and Implications for Policy
shocks not to necessarily follow the same pattern. This tendency is seen in the results earlier reported when oil price changes only matter in influencing the stock returns of some sectors only in the short-run or long-run, as the case may be, and others in both periods. The magnitude of effects and significance in the two periods also vary across sectors. Hence, accounting for sectoral behaviour in oil price-stock nexus proves crucial.
This study therefore re-assesses the nexus between oil price and the stock market on sectoral analyses. It basically tries to discover if there are significant differences across sectors. Also, since there is possibility of certain structural breaks along the time path of some variables, as is mostly notable for high frequency data such as is applied in this study, another round of analyses is made to give consideration to this tendency. Thus, the two models (with and without structural breaks) are estimated for both short-and long-run. Meanwhile, in order to show how best the sectoral stock returns respond to oil price shocks, we conduct both the symmetric and asymmetric models depending on which one is suitable to a particular sector as judged by the Bai-Perron (2013) test. This makes our study to employ both symmetric ARDL model developed by Pesaran et al. (2001) and its asymmetric version following the approach of Shin et al. (2014) in decomposing oil price into positive and negative changes.
Our analyses show that without considering structural breaks, the stock returns of majority of the sectors respond to shocks in oil price. In addition, these responses are positive across board, largely stronger in the long-run, and are irrespective of whether the model is symmetry or otherwise. On the other hand, only two sectors are significantly affected by structural breaks, each sharing the symmetric and asymmetric models. Their results are largely positive and stronger in the long-run too, except the negative impacts of the breaks in their respective periods. These significant results are consistent with studies including , Bouri et al. (2016), among others.
The lack of or weak response to asymmetry by some sectors also give support to studies including Cong et al. (2008), etc.
In rendering appropriate policies following the outcome of our findings, we first put a call through to relevant policy analysts, governments, investors and other policy makers that form their investment decisions on empirical studies like this to give more consideration to sectoral analysis. It is not untrue that knowing the pass-through of oil price shocks to the market for financial instruments is relevant for stock market assessments in terms of risk hedging and portfolio management .
However, this proves more valid only when the sectoral stock response to oil price shocks is known with certainty, rather than the aggregate measure of stock market performance that gives little regards to sub-component variations. Therefore, this is a plus to investors with sound knowledge of the sectoral stock-oil price relationship, as they can rationally determine the best sector to invest and commit their hard-earned funds, how and when to make this rational choice in the face of the risks associated with the upward and downward swings in oil price.
In addition, our findings reveal that certain sectors are unresponsive to asymmetric tendencies in oil price movement. Put differently, reactions to asymmetries in oil price changes vary across sectors. This has two implications-one for potential investors and the other for the stock market or general financial analysts and forecasters. For the potential investors, it particularly provides financial security to the risk averse. In other words, investors that take negative posture to risks due to their 'financial fear' are better guided to embrace investment in sectors with low response to asymmetries in oil price, although this may come with lower returns as reported in our regression tables. This perhaps follows the common axiom of 'the higher the risk, the higher the returns'.
Secondly, our results provide useful information and guide to stock market professionals for meaningful and effective stock forecasts. Our findings make them understand the need to incorporate the risk tendency in their forecasts of the future movements of stock returns since some sectors appear to be distinct in their behaviour to asymmetries.