Oil Prices and the Cross-Section of Stock Returns Dayong Huang Bryan School of Business and Economics University of North Carolina at Greensboro Email: d_huang@uncg.edu Jianjun Miao Department of Economics Boston University Email: miaoj@bu.edu First version: May 2016 Abstract We investigate the pricing of oil price changes in the cross-section of stock returns using the US daily data from 1986 to 2015. We find that stocks with different exposures to oil price changes do not have a significant return difference in the full sample and in the subsample from 1986 to 2001. There is a negative return spread between a portfolio of stocks with the lowest oil betas and a portfolio of stocks with the highest oil betas in the subsample from 2002 to 2010 and this spread can be explained by the Fama-French three factors plus a momentum factor. For the subsample from 2011 to 2015, the return spread is positive and can be explained by the FamaFrench three factors plus a volatility factor. For both subsamples, the return spread cannot be explained by the Fama-French five-factor model. We also find that oil betas predict future stock returns negatively in the past two years. Keywords: Oil prices, factor model, return spread, cross section. JEL classification: G12 1 1 Introduction Oil prices have been volatile, especially in the past decade. Figure 1 presents the daily nominal front month crude oil futures prices over the sample period from January 1, 1986 (the earliest date when the daily oil price data is available) to December 31, 2015. Oil prices were around $20 per barrel until the end of 2001. Over the next six years, oil prices skyrocketed, reaching more than $140 per barrel in the middle of 2007, and then crashed, falling sharply to around $40 per barrel in July 2008. Oil prices gradually rose to around $100 per barrel in mid2014 and plummeted to around $30 per barrel in the end of 2015. Oil prices affect the macroeconomy (see Hamilton (2009) and Blanchard and Gali (2009) among others) and hence should be related to aggregate stock market movements. Bernanke (2016) finds that oil price changes and aggregate stock returns were positively correlated in the past five years and this positive correlation is puzzling because a decrease in oil prices should be good news for oil-importing economies such as the US. This issue has received a considerable amount of attention in both the academia and the general public and most discussions have been focused on the relationship between oil prices and aggregate stock returns. The impact of oil price changes on the cross-section of stock returns has been underexplored. The goal of our paper is to investigate this issue. We first estimate oil beta for each individual stock using daily data in each month and then form 10 portfolios sorted on oil beta. We define the LMHO portfolio as the one that buys decile 1 of stocks with the smallest oil betas and sells decile 10 of stocks with the highest oil betas. We track the value-weighted returns in the following month. Next we estimate the 2 average excess returns on all these portfolios using monthly data. We also estimate abnormal returns (alphas) based on the five-factor model of Fama and French (2015) using monthly data. This model performs better than the three-factor model of Fama and French (1993, 1996) and is used as our benchmark. We find that both excess returns and abnormal returns on the LHMO portfolio are insignificant in the full sample, implying that different exposures to oil price changes do not generate a significant difference in average stock returns in the full sample. One reason for this result is that the excess return on the LMHO portfolio switches signs over time. The average annual excess return alternates between negative and positive signs from 1986 to 2001, stays positive from 2002 to 2010, and becomes positive from 2011 to 2015. We then study these three subsamples. We find that the average excess return on the LMHO portfolio is 0.2% per month and insignificant in the subsample from January 1986 to December 2001. In the subsample from January 2002 to December 2010, the average excess return on the LMHO portfolio is -0.71% per month with a t-value of -1.33. Although this number is not statistically significant, it is economically significant as the implied annual return is a quite impressive number of -8.52%. After adjusting for risks using the Fama-French five factors, we find that the abnormal return is -0.99% per month and marginally significant with a t-value of -1.75. The GRS test (Gibbons, Ross, and Shanken (1989)) rejects the null hypothesis that the regression intercepts of the 10 portfolios are all equal to zero, with the p-value equal to 0.10. Thus the Fama-French five-factor model has a hard time to explain the cross-sectional return spread in this subsample. Using the Fama-French three factors (Fama and French (1993, 1996)) plus a momentum factor of Jagadeesh and Titman (1993) and Carhart (1997), we find that the abnormal return on 3 the LMHO portfolio is slightly lower and is equal to -0.87% per month with a t-value of -1.71. The GRS test gives a p-value of 0.32 and cannot reject the null hypothesis of joint zero intercepts. We also find that stocks with the lowest (highest) oil betas have a negative (positive) slope on the momentum factor. Since the average return on the momentum factor is positive, the negative loading on the momentum factor explains most of the negative average return of the LMHO portfolio. For the subsample from January 2011 to December 2015, the average excess return on the LMHO portfolio is 1.08% per month and highly significant with a t-value of 2.17. The abnormal return relative to the Fama-French five-factor model is 1.12% per month and significant at the 5% level. The GRS test rejects the hypothesis that the regression intercepts of the 10 portfolios are all equal to zero, with the p-value equal to 0.04. Thus the Fama-French five-factor model cannot explain the return spread in this subsample. Using the Fama-French three factors plus the momentum factor or the volatility factor (Ang et al (2006)) reduces the return spread and the significance of the GRS test with p-values of 0.09 and 0.13, respectively. Thus, the volatility factor plays an important role, but the abnormal return on portfolio 10 with the highest oil betas is still significant at the 5% level. We find that portfolio 10 has a significant positive loading on the value factor, but the larger positive loading on the volatility factor relative to portfolio 1 with the lowest oil betas helps explain the relatively lower return on portfolio 10 in the subsample from January 2011 to December 2015. What types of firms have high oil betas for this subsample? Using the Fama-MacBeth (1973) regressions, we find that high oil beta stocks tend to be large stocks, value stocks, stocks that grow investments aggressively, stocks that have low momentum, and stocks that are 4 volatile. But oil betas do not help predict future returns in this subsample. When restricting to the more recent subsample from January 2014 to December 2015, we find that high oil betas predict low future returns, even after many control variables are considered, including the size, the book-to-market ratio, asset growth, operating profitability, momentum, and stock’s total volatility. We also find that our results in this subsample are quite robust after controlling for these potential cross-sectional pricing effects in two-dimensional sorting. Examining industry-level evidence, we find that the oil industry has the largest positive average oil beta in the previous month and earns negative average excess returns and negative abnormal returns in the following month relative to the Fama-French five factor model. The retail industry has the smallest (negative) average oil beta in the previous month. It earns positive average excess returns and positive abnormal returns in the following month relative to the Fama-French five factor model. This result is consistent with the intuition that oil producers are hurt by the decline of oil prices, but consumers benefit from it. Bernanke (2016) argues that the recent close relationship between aggregate stock returns and oil prices may be due to the fact that they both react to a common aggregate demand factor. Thus it is possible that our oil beta is measured inaccurately if we ignore variables that capture fluctuations in aggregate demand. To deal with this issue, we follow Bernanke’s methodology, originally suggested by Hamilton (2014), to decompose oil price changes into a demand component and a supply component. We examine how oil price changes due to aggregate demand and aggregate supply affect the cross-section of stock returns in the recent period from 2011 to 2015, as this subsample gives a significant LMHO return. Furthermore, we control for each stock’s exposure to uncertainty measured by the VIX, 5 as stock prices and oil prices can both react to uncertainty in addition to fluctuations in aggregate demand. We then form 10 portfolios sorted separately by betas of the demand component (or demand component after controlling for uncertainty) and the supply component of oil price changes. We find that the abnormal return on the LMHO portfolio related to demand shocks is insignificant only after controlling for uncertainty, but it is still significant (1.05% per month) at the 5% level for the supply component. This result shows that the cross-sectional variation in stock returns due to demand-driven oil price changes after controlling for uncertainty can be explained by the Fama-French five-factor model. But the variation of stock returns due to the supply-driven oil price changes is still left unexplained. Our paper makes two important contributions to the literature. First, we are the first to investigate the impact of oil price changes on the cross-section of stock returns using some new common risk factors proposed recently in the literature. Second, we show that the FamaFrench five-factor model cannot explain the cross-section of stock returns related to oil price changes in the two subsamples from 2002 to 2010 and from 2011 to 2015. But the FamaFrench three factors plus a momentum factor or a volatility factor are useful. This finding should stimulate further studies. Most studies in the literature focus on the relation between oil price changes and aggregate stock returns (Jones and Kaul (1996), Huang, Masulis, and Stoll (1996), Sandorsky (1999), and Kilian and Park (2009)). These studies find some evidence for relations at various leads and lags, but weak results in contemporaneous correlations. Ready (2015) proposes a novel method for classifying oil price changes as supply or demand driven and finds that demand shocks are strongly positively correlated with market returns, while supply shocks have 6 a strong negative correlation. He also examines the impact of these shocks on industry stock returns. Driesprong, Jacobsen, and Maat (2008) find that oil price changes predict stock market returns worldwide. They show that investors’ underreaction to information in oil prices causes a rise in oil prices to lower future stock returns. Cross-sectional studies that incorporate oil price changes include Chen, Roll, and Ross (1986), Ferson and Harvey (1993), and Demirer, Jategaonkar, and Khalifa (2015). These studies find mixed evidence for the pricing effect of oil price changes. The reminder of the paper proceeds as follows. In Section 2 we discuss the data and methodology. Section 3 presents empirical results. Section 4 concludes. 2 Data and Methodology 2.1 Data We use daily data from January 1, 1986 to December 31, 2015. Our stock returns data correspond to CRSP common shares (SHRCD = 10 or 11) from the NYSE, AMEX and NASDAQ is the one-month Treasury bill rate. Firm (EXCH = 1 or 2 or 3). The risk-free rate fundamentals and accounting variables are taken from the Compustat database. Following Davis, Fama, and French (2000), Cooper, Gulen, and Schill (2008), Xing (2008), Fama and French (2008, 2015), Novy-Marx (2013), and Hou, Xue and Zhang (2015), we construct the following four fundamental variables in a standard way: firm size (MC), book-to-market ratio (BM), operating-profit-to-book-equity ratio or operating profitability (OPE), and asset growth (AG) or investment. 7 Other control variables we use include the stock price momentum (MOM), reversal (REV), and individual stock’s total volatility (TVOL). Following Jegadeesh and Titman (1993), at the end of each month t, we compute each stock’s cumulative return from month t-13 to t-2, and take it as the stock price momentum. Reversal is each stock’s return in month t-1. The total volatility of a stock is the variance of daily stock returns in each month. The five factors of Fama and French (2015) include the original three factors of Fama and French (1993, 1996) and two new factors. The original threes factors are the market excess return (MKT) defined as the return on the value-weighted market portfolio of all NYSE, AMEX, and NASDAQ stocks minus the risk-free rate, a size factor (SMB) defined as the return on a portfolio of small cap stocks minus the return on a portfolio of large cap stocks, and a value factor (HML) defined as the return on a portfolio of high-book-to-market stocks minus the return on a portfolio of low-book-to-market stocks. The two new factors are an asset growth factor or investment factor (CMA) which captures the difference between returns on portfolios of conservative and aggressive firms in terms of investments, and a profitability factor (RMW) which captures the difference between returns on portfolios of robust and weak firms in terms of operating profitability. Additional common factors used in our analysis are: the momentum factor (MOMF) which is the return on a portfolio that buys the past winner and sells the past loser, and the total volatility factor (VOL) which is the return on a portfolio that buys stocks with high total volatility and sells stocks with low total volatility. All these factor data are downloaded from Ken French’s website (http://mba.tuck.dartmouth.edu). We use the data of front month crude oil futures on West Texas Intermediate Oil, taken from the Genesis Financial Technologies and compute daily oil price changes (or the return on 8 oil) as , = ln( / ), where is the oil price at date t. As Huang, Masulis, and Stoll (1996) and Sadorsky (2001) point out, using oil futures prices has several advantages over using spot prices. For example, spot prices are more heavily affected by temporary random noise than futures prices. We find that using spot prices does not affect our results significantly. To estimate the demand component of oil price changes, we use the data of copper futures from Genesis Financial Technologies and the yields of the US 10 year T-bond from the Saint Louis Federal Reserve Bank. We use the percentage changes in the VIX from the Federal Reserve Bank at Saint Louis to measure uncertainty. Figure 1 presents the time series of oil prices and copper prices and shows that they follow a similar pattern. 2.2 Methodology We first estimate oil betas using daily data over the previous month. At the end of each month, we regress daily stock returns on daily market excess returns and daily returns of crude oil futures prices in excess of the risk free rate using the following equation where − = + + ( , − )+ , (1) is the return on stock between date t-1 and date t. The slope of the return on crude oil futures is the oil beta. The estimated oil beta may vary across months. Using these oil betas, we rank all firms from low oil betas to high oil betas in each month and follow their excess returns in the following month. We then form 10 portfolios based on the rank of oil betas. The return on a portfolio is the value-weighted average of the returns of 9 all firms in that portfolio. We also track the portfolio level oil betas at the time of portfolio formation and portfolio level betas in the following month after portfolio formation. The portfolio-level oil beta is the value-weighted average of oil betas of all stocks that belong to each portfolio. The LMHO portfolio is the one that longs portfolio 1 with the lowest oil betas and shorts portfolio 10 with the highest oil betas. The return on the LMHO portfolio is also equal to the return spread between portfolio 1 and portfolio 10. At the time of portfolio formation, we delete firms whose prices are less than five dollars, and firms whose market capitalization are less than the 10 percentile of market capitalization of NYSE firms. The market capitalization breakpoints of NYSE firms are available at Kenneth French’s website. We drop these firms to avoid micro-structure issues as they are hard to trade. Our results are stronger when we include these micro-cap firms. Our main testing methods follow the classical method of Fama and French (1996, 2015). We test if the excess returns on decile portfolios formed on oil betas can be explained by the recently proposed five factors of Fama and French (2015) using time-series regressions with monthly data. If the returns on oil portfolios can be explained by these five common factors, the joint test on the abnormal returns or alphas using the GRS test of Gibbons, Ross and Shanken (1986) shall be insignificant. In addition, we test alternative common factor models. One of them is the Fama and French three factors plus a momentum factor which buys past winner and sells past losers (Jagadeesh and Titman (1993) and Carhart (1997)); the other is a model that includes the Fama-French three factors and a volatility factor that buys stocks with high volatility and sells stocks with low volatility (Ang et al (2006)). 10 To control for other firm fundamentals or characteristics, we perform double sorting, in which we first sort on oil betas, and then independently sort on a fundamental or a characteristics, and we take the intersections of the two sorts to form portfolios. We track the return on the portfolios in the following month. We rely on the Fama and MacBeth (1973) regression to identify if oil beta has incremental power to explain future stock returns. As a robustness check, we decompose changes in oil prices into a component due to changes in aggregate demand and a component due to changes in aggregate supply, as in Bernanke (2016). In each month, we project daily oil prices changes on daily copper price changes , and daily changes in 10-year Treasury bond rate ∆"# using the following regression − , = $% + $ ( − , +& . ) + $ ∆"# (2) We take the fitted value as oil price changes related to demand, denoted by residuals as oil price changes related to supply, denoted by ( , ' , , and the . Similarly, to remove the effect of uncertainty on oil price changes, we run the following regression , − = $% + $ ( , − ) + $ ∆"# + $) ∆*"+ + , (3) where ∆*"+ is the percentage change in the daily VIX. The fitted values are the oil price changes related to the changes in demand and uncertainty, denoted by are oil price changes related to supply, denoted by with each one of ' , , ( , , ', , , and (, , (, , . We then replace , , and the residuals , in equation (1) to estimate the corresponding oil beta. We form portfolios using these oil betas and repeat the preceding analysis. 11 ', 3 Empirical Results 3.1. Results from One-Dimensional Sorting If oil price changes affect the cross-section of stock returns, we should find that firms with different exposures to oil price changes earn different returns. We first examine this issue in the full sample from 1986 to 2015. The first row of Table 1 presents the average monthly excess returns on the decile portfolios sorted by the past oil betas over the previous month. This row shows that portfolio 1 with the lowest oil betas earns an average return of 0.45% per month (tvalue = 1.39) and portfolio 10 with the highest oil beta stocks earn an average return of 0.38% per month (t-value = 1.05). The return spread between portfolio 1 and portfolio 10 is 0.07% per month (t-value = 0.29). The low t-value shows that the return spread is insignificant. After adjusting for the Fama-French five factors, we find that the return spread becomes negative and insignificant. The first row of the bottom panel in Table 1 presents the alphas of the decile portfolios after adjusting for the Fama-French five factors in the full sample. We find that the spread of alphas between portfolios 1 and 10 is -0.10% per month with the t-value equal to 0.38. To see why the average return spread is close to zero in the full sample intuitively, we compute the annual return spreads from 1986 to 2015, using the average of the monthly spreads. Figure 2 presents the result and shows that the return spread varies over time. It alternates between positive and negative signs from 1986 to 2001, stays negative from 2002 to 2010, and becomes positive from 2011 to 2015. The t-values for all years indicate that the return spreads are significant at the 5% level only in 1988 and 2014. This pattern shows that 12 there is almost no difference in average returns between low oil beta stocks and high oil beta stocks in the full sample and suggests that the weak result in the full sample is driven by the sign switching of the return spread in the full sample. We thus focus our analyses on three subsamples: 1986-2011, 2002-2010, and 2011-2015. Table 1 presents spreads in excess returns and in alphas for these three subsamples. We find the following result: From 1986 to 2001, the spread in excess returns is 0.20% per month (t-value = 0.61) and the spread in alphas relative to the Fama-French five-factor model is 0.17% per month (t-value = 0.54). Consistent with Figure 2, both spreads are insignificant. From 2002 to 2010, these spreads are -0.71% and -0.99% per month, respectively. The corresponding tvalues are 1.33 and -1.75, indicating that the abnormal returns are marginally significant. By contrast, from 2011 to 2015, these spreads are 1.08% and 1.12% per month and both are significant at the 5% level. Table 2 presents the portfolio-level oil betas at the time of portfolio formation and the immediate month after the portfolio formation. The purpose of this diagnosis is to ensure high oil beta portfolios at the time of portfolio formation continue to have high oil betas in the following month. We find that this pattern holds in all three subsamples. In particular, from 2011 to 2015, low (high) oil beta stocks on average have oil beta of -0.61 (0.68) at the time of portfolio formation. The beta spread of -1.29 between low and high oil betas is highly significant. In the month following portfolio formation, low (high) oil beta stocks on average have oil beta of -0.03 (0.14). The spread of oil betas is -0.17, much less than the oil beta spread at the time of portfolio formation, but this difference is also highly significant with a t-value of -7.37. The result in Table 2 is important because it is possible that, although they have high oil 13 betas at the time of portfolio formation, some stocks may become low oil beta stocks after portfolio formation, and this switch in betas makes it unclear whether it is high oil beta stocks or low oil stocks will have low returns in the future. 3.2 Can Return Spreads Be Explained by Common Risk Factors? The next three tables present results from testing if the excess returns on the portfolios sorted on oil betas can be explained by exposures to common risk factors. We consider the following three sets of parsimonious common factors: the Fama-French five factors, the FamaFrench three factors plus a momentum factor, and the Fama-French three factors plus a volatility factor. The purpose is to examine whether stocks with different exposures to oil price changes have anything to do with any of these common factors. If returns on oil portfolios are fully captured by their exposures to these common factors, we should expect insignificant abnormal returns and the GRS test will not be able to reject the null hypothesis of zero joint abnormal returns. Table 3 considers the Fama-French five-factor model. The notable findings are as follows. In the subsample from 1986 to 2001, the low oil beta stocks (portfolio 1) load negatively on the profitability factor with a coefficient of -0.37 (t-value = -2.44), and the high oil beta stocks (portfolio 10) load negatively on the asset growth factor with a coefficient of -0.59 (t-value = 6.07). The net result is that the LMHO oil portfolio loads positively on the asset growth factor and negatively on the profitability factor. This result suggests that high oil beta firms move in 14 lockstep with firms that are aggressive in asset growth and robust in profitability. The GRS test of 0.67 (p-value = 0.77) indicates that there is no abnormal return in this subsample. In the subsample from 2002 to 2010, both high oil beta stocks and low oil beta stocks have negative loadings on the asset growth factor and profitability factor. This result suggests that they both tend to be firms with aggressive investments and weak profitability. The GRS test statistic is 1.66 with a p-value of 0.10, suggesting abnormal returns are marginally significant. In the subsample from 2011 to 2015, the loadings of the returns on high oil beta stocks (portfolio 10) on the value factor, the size factor, the asset growth factor, and the profitability factor are 1.37 (t-value = 4.60), 0.37 (t-value=1.81), 0.41 (t-value = 1.34), and -1.48 (t-value = 3.35), respectively. This result implies that the returns on high oil beta stocks tend to co-move positively with the returns on small value firms or firms with either aggressive investment or weak profitability. The GRS test statistic is 2.13 with a p-value of 0.04. The null hypothesis of zero joint abnormal returns is rejected in this subsample. The result above shows that the Fama-French five-factor model fails to price the returns on the oil portfolios in the recent sample from 2011 to 2015. We now test if a momentum factor or a volatility factor, together with the usual Fama-French three factors, can resolve this challenge. Table 4 presents the results from using the Fama-French three factors and the momentum factor of Carhart (1997). In the subsample from 2002 to 2010, the returns on high oil beta stocks (portfolio 10) co-move positively with the returns of past winners. For the portfolio with the highest oil beta, the loading on the momentum factor is 0.10 and highly 15 significant with a t-value of 2.10. The loadings on the momentum factor increase monotonically from portfolio 1 with the lowest oil betas to portfolio 10 with the highest oil betas. This explains why stocks with high oil betas have high returns. The p-value of the GRS test is 0.32, suggesting that the Fama-French three factors plus the momentum factor can explain the variation in returns on portfolios with different oil betas in the subsample 2002-2010. By contrast, in the subsample from 2011 to 2015, the returns on high oil betas stocks comove negatively with the returns of past winners, or co-move positively with past losers. The loading on the momentum factor for portfolio 10 is -0.13 and its t-value is -1.59. The GRS test statistic is 1.76 with a p-value of 0.09, rejecting the hypothesis that alphas of the decile portfolios are all equal to zero. Thus the Fama-French three factors plus the momentum factor cannot explain the variation in returns on portfolios with different oil betas in the subsample 2011-2015. Table 5 presents the results from using the Fama-French three factors and a volatility factor. We find that in all three subsamples, the returns on both high and low oil beta stocks tend to co-move positively with the return on stocks with high total volatilities. In the subsample from 2011 to 2015, the high oil beta stocks have the largest exposure to the volatility factor. The coefficient is 0.34 and highly significant (t-value = 3.98). This can help explain why stocks with high oil betas have low returns in the subsample 2011-2015. The GRS test statistic is 1.61 with a p-value of 0.13, indicating that the null hypothesis of zero joint abnormal returns on the decile portfolios cannot be rejected. But the alpha of portfolio 10 with the highest oil betas is still significant with a t-value of-2.04. 16 Interestingly, the loadings on the volatility factor are not monotonic with oil betas. It is possible that among stocks that have large total volatility, some are sensitive to oil price changes, while some are not and simply vary by themselves. Using oil betas as a sorting criterion allocates some stocks with high total volatility into high oil beta stocks and others with high total volatility into low oil beta stocks. In summary, we find that the Fama-French five-factor model cannot explain the crosssectional variation in stock returns with different exposures to oil price changes in the subsamples 2002-2010 and 2011-2015. For the subsample from 2002 to 2010, either the momentum factor or the volatility factor, together with the Fama-French three factors, can explain this cross-sectional variation. By contrast, the recent sample from 2011-2015 is more challenging. We find that the Fama-French three factors plus the momentum factor do not help. The volatility factor together with the Fama-French three factors can marginally explain the cross-sectional variation with the GRS test p-value equal to 0.13. Moreover, the alpha of portfolio 10 with the highest oil betas after controlling for these factors is highly significant. 3.3 Results from Cross-Sectional Regressions What types of firms have high oil betas? We run the Fama and MacBeth (1973) regressions as follows. We regress oil betas at the time of portfolio formation, by sample periods, on the following fundamentals including market capitalization, book-to-market ratio, asset growth, operating profitability, price momentum in the past 12 month, and each stock’s total volatility at the time of portfolio formation. Table 6 presents the results. In the full sample 17 from 1986 to 2015, high oil beta stocks tend to be large stocks, value stocks, stocks that grow investments aggressively, stocks that have weak profitability, and stocks that are volatile. There are variations in the subsamples. From 1986 to 2001, only operating profitability explains oil betas. In particular, higher oil beta firms have lower operating profits. From 2002 to 2010, almost all variables are significant in explaining oil betas and the signs seem intuitive. The new addition is that, high momentum is associated with high oil betas. From 2011 to 2015, high oil beta stocks are large stocks, value stocks, stocks with aggressive investments, low momentum, and high volatility. In this sample, the adjusted R-square is 11.52%. Moreover, we find that a similar pattern holds for the most recent period from 2014 to 2015. The explanation power is even stronger as the adjusted R-square is 17.96%. But the total volatility is only marginally significant with the t-value equal to 1.41. How does the oil beta perform in predicting future returns? Following Fama and MacBeth (1973), we regress future one-month returns on current oil betas, fundamentals, and other control variables. Table 7 presents the results. In the full sample from 1986 to 2015, oil beta has a coefficient of -0.002 (t-value = -1.11). This coefficient has a t-value of -0.8 when we include control variables. Among the control variables, asset growth, profitability, and total volatility are significant. This result is consistent with our previous finding that oil price changes do not have a strong relationship with the cross-section of stock returns in the full sample. We now consider subsample results. From 1986 to 2001, oil beta by itself has a coefficient of -0.0034 (t-value = -1.95), but it becomes insignificant when control variables are included. Among the control variables, asset growth, momentum and total volatility are significant. From 2002 to 2010, oil beta by itself has a coefficient of 0.0035 (t-value = 0.95). 18 Asset growth, profitability, and total volatility are significant. From 2011 to 2015, oil beta by itself has a coefficient of -0.0054 (t-value = -1.16). In this period, the only control variable that is significant is the momentum. The preceding results indicate that oil beta does not help forecast future returns in both our full sample and the three subsamples. So what is it that makes oil price changes received so much attention lately? We then consider the most recent years from 2014 to 2015. We find that oil beta by itself has a coefficient of -0.0194 (t-value = -1.83). In this period, other control variables are insignificant. The size, book to market ratio, asset growth, and gross profitability, as emphasized in the recent asset pricing studies (Fama and French (2015) and Hou, Xue and Zhang (2014)), do not perform well in forecasting future returns; their t-values are 0.56, -0.91, 1.01, and 0.08, respectively. Momentum has a t-value of 0.6 and volatility has a t-value of -0.86. In summary, the results presented in Tables 6 and 7 seem consistent with those reported in Tables 3, 4, and 5. Returns on portfolios formed on oil betas seem to be captured by common factors such as the Fama-French three factors plus the momentum factor or the volatility factor. Evidence from 2011 to 2015 and especially from 2014 to 2015 is more challenging for these common factors. The following two subsections will investigate in more detail the cross-section of returns in the more recent period from 2014 to 2015 and will consider other common factors coming from aggregate demand or aggregate supply shocks from 2011 to 2015, as suggested by Bernanake (2016). 19 3.4 A Closer Look at the Evidence from 2014 to 2015 This subsection focuses on the two years from 2014 to 2015 since this period seems to pose the most serious challenge to the risk-based explanations of the effect of oil price changes on the cross-section of stock returns. We consider several robustness checks and also present some industry-level evidence. We first show that the cross-sectional variation in stock returns with different exposures to oil price changes remains large after we control for various fundamental variables. We sort stocks into five quintiles based on their oil betas in the previous month. We then independently sort stocks into five quintiles based on one of the following control variables: size, book-tomarket ratio, asset growth, gross profitability, momentum, short-term reversal, and total volatility. We then obtain 5 × 5 portfolios. Portfolio 15 is the one that buys quintile 1 and sells quintile 5 along one dimension of sorting in increasing orders. Table 8 presents the twodimensional sorting results. In the double sorting on oil beta and size, the excess returns and abnormal returns (relative to the Fama-French five-factor model) on portfolio 15 along the oil beta dimension are large in four of the size quintiles. They are above 1.2% per month and typically 2.5 standard deviations from zero. We find similar results from double sorting using the book-to-market, gross profitability, asset growth, and even momentum or short term reversal. Double sorting on the oil beta and momentum shows that, the abnormal return on portfolio 15 along the oil beta dimension is 2.19% per month (t-value = 2.06) among the low momentum stocks, and is 2.33% per month (t-value = 3.28) among the high momentum stocks. Finally, controlling for the total 20 volatility does not seem to undermine our results in 2014 to 2015. The abnormal return on portfolio 15 is significant in high volatility portfolio. It is around 2.40% per month (t-value = 2.07) among stocks with high volatility betas. As a comparison, the lower right block presents the volatility-oil beta sort in the reverse order. We find that, controlling for oil beta, the abnormal returns on portfolio 15 along the volatility dimension, in general, are smaller than the counterparts when we first control for volatility. The abnormal returns are insignificant in four of the five oil beta quintiles. In quintile 5 with the highest oil betas, the abnormal return is 1.54% per month (t-value = 1.90). Our finding suggests that both oil betas and total volatility help explain the cross-sectional variation in stock returns and that oil betas are more informative than total volatility. Are these high oil beta firms simply oil producers? If yes, the effect of oil price changes seems easier to understand. We identify industries, to which the high and low oil beta firms belong. We consider 48 industries identified by the SIC codes as in Fama and French (1997). We use the 80th percentile and 20th percentile as cutoffs to define high oil beta firms and low oil beta firms. Panel A of Table 9 shows that the firm-month observations of high oil beta firms in the oil industry account for the largest fraction (15%) of the total firm-month observations of high oil beta firms from 2014 to 2015. The other top nine industries that have high oil beta firms are Business Services (9.6%), Others (8.5%), Machinery (4.6%), Retail (4.6%), Electronic Equipment (4.3%), Drugs (4.0%), Banking (3.9%), Chemicals (3.6%), and Utilities (3.4%). These 10 industries account for 59% of the firm-month observations that have high oil betas. This result shows that high oil beta firms are not simply oil producers. 21 Panel A of Table 9 also presents the top 10 industries that have low oil beta firms. We find that, except for Petroleum and Natural Gas, Machinery, and Chemicals industries, other top 10 industries that have high oil beta firms are also in the list of top 10 industries that have low oil beta firms. The remaining three top 10 industries that have low oil beta firms are Transportation, Insurance, and Medical Equipment. Panel B of Table 9 presents the average oil betas by industries from 2014 to 2015. We find that, the 10 industries with the largest oil betas are the following: Petroleum and Natural Gas, Coal, Mine, Fabricated Products, Gold, Machinery, Entertainment, Electrical Equipment, Shipbuilding and Railroad Building, Equipment and Utility. These industries have positive oil betas and low returns in the following month, as most returns in the next month are large and negative, except for the Entertainment industry. Interestingly the oil industry (Petroleum and Natural Gas) is not the one with the largest negative abnormal returns (alphas) after adjusting for the Fama-French five factors. The two industries that have the largest negative alphas are the Fabricated Products and Coal, albeit they are relatively small in that the number of firms in those industries is not large. Considering the number of firms and market capitalization, we find that the two industries having negative alphas still point to Petroleum and Natural Gas and Machinery. The average alphas of all firms in the oil industry from 2014 to 2015 is -0.48% (tvalue = -2.64). This industry covers 1985 firm-month observations among 49140 total observations. The average market capitalization of the oil industry is 168 billion. The Machinery covers 1519 firm-month observations, and its average firm market capitalization is 116 billion. The 10 industries with the smallest oil betas, from the lowest to the highest, are Retail, Personal Services, Textiles, Transportation, Restaurants and Hotels, Other, Printing and 22 Publishing, Agriculture, Defense, and Pharmaceutical Products. These industries have negative oil betas and seem to have large positive returns in the following month. The two industries that have the largest alphas are Guns and Retail, and Retail seems to count heavily in the market in terms of market capitalization and the number of firms. Retail covers 2588 firmmonth observations and its average market capitalization is 83 billion. Its abnormal return is 2.08% per month (t-value = 18.69). 3.5. Demand and Supply Driven Oil Price Changes Bernanke (2016) argues that the recent strong correlation between oil price changes and aggregate stock returns may be explained by the hypothesis that both are reacting to a common factor, namely, a softening of global aggregate demand, which hurts both corporate profits and demand for oil. In this subsection we follow his methodology, which is originally suggested by Hamilton (2014), to decompose the changes in oil prices into a component due to demand shocks and a component due to supply shocks. We consider daily data in the sample period from January 1, 2011 to December 31, 2015. This sample period is almost the same as that in Bernanke (2016). Unlike his study of aggregate market returns, our analysis focuses on the impact of oil price changes on the cross-section of stock returns. We use copper futures prices and the US 10 year T-bond rate to capture demand changes. The rational of using copper prices suggested by Hamilton (2014) and Bernanke (2016) is that the drop of the price of copper has nothing to do with the success of people getting more oil out of the rocks in Texas and North Dakota. Softness in demand for commodities like 23 copper may be an indicator of new weakness in the economy. Similarly, yield on 10-year US Treasury bond indicates weakness in the world economy. In unreported results we have also considered the value of the US dollar to capture demand changes as in Hamilton (2014) and Bernanke (2016). We find that the results are similar. We regress the oil price changes on copper prices changes and the 10-year T-bond rate in each month using equation (2). The fitted value is the oil price changes as predicted by the demand changes and the residual is treated as oil price changes related to supply shocks. Using oil price changes related to demand and supply shocks, we calculate two corresponding oil betas. We form portfolios sorted on these two betas in the previous month. We then conduct a similar analysis as in previous subsections. Table 10 presents the results. The upper panel presents excess returns and the bottom panel presents abnormal returns using the Fama and French five-factor model. The first block and the second block present the returns of portfolios formed on oil betas related to demand and oil betas related to supply. The excess return on the LMHO portfolio formed on oil betas related to the demand component of oil price changes is 0.73% per month (t-value = 1.65). The excess return on the LMHO portfolio formed on oil betas related to the supply component of oil price changes is 1.00% per month (t-value = 1.65). The abnormal returns are 0.98% per month (t-value = 2.13) and 0.95% per month (t-value = 2.01). This result indicates that the abnormal returns based on the Fama-French five-factor model are still significant for both the demanddriven oil price changes and supply-driven oil price changes. 24 Bernanke (2016) argues that the recent stock market fluctuations have been accompanied by elevated volatility. When uncertainty or volatility elevates, it is possible that investors retreat from both oil and stocks, causing stock returns and oil price to move together. To test whether our results are affected by missing variables such as uncertainty, we augment our previous demand and supply decomposition with daily percentage changes in the volatility as measured by the VIX. Specifically, to estimate oil beta related to demand-and-uncertaintydriven oil price changes, we compute the fitted values from regressing daily oil price changes on copper price changes, 10-year T-bond rate changes and VIX percentage changes using equation (3), and treat the fitted values as demand-and-uncertainty-driven oil price changes. The residuals are treated as supply-driven oil price changes. We then regress individual stock returns on these oil price changes to estimate oil betas. The third blocks of Panels A and B of Table 10 present the excess and abnormal returns on portfolios formed on oil betas related to demand-and-uncertainty-driven oil price changes and on oil betas related to supply-driven oil price changes. The excess return on the LMHO portfolio formed on oil betas related to the demand-and-uncertainty-driven oil price changes is 0.33% per month (t-value = 0.83). The abnormal return based on the Fama-French five-factor model is 0.58% per month (t-value = 1.43). This result shows that, when controlling for uncertainty, the demand-driven oil price changes do not generate significant abnormal returns. The intuition is as follows. It is possible that, even though individual stocks have different exposures to the demand-and-uncertainty-driven oil price changes, they have similar exposures to the uncertainty component in the demand-and-uncertainty-driven oil price changes. Hence they have similar stock returns because volatility explains the abnormal returns. 25 The forth blocks of Panels A and B of Table 10 present the results for the supply-driven oil price changes. The excess return on the LMHO portfolio formed on oil betas related to the supply-driven oil price changes is 1.11% per month (t-value = 2.22). The abnormal return based on the Fama-French five-factor model is 1.05% per month and significant with t-value = 2.23. Thus the return spread due to different exposures to supply-driven oil price changes cannot be explained by the Fama-French five-factor model. 4 Concluding Remarks Our extensive empirical analyses have shown that the risk-based theory using the FamFrench five-factor model fails to explain the cross-sectional variation in stock returns with different exposures to oil price changes in subsamples 2002-2010 and 2011-2015. The FamaFrench three factors plus a momentum factor or a volatility factor can help explain the crosssectional variation. But there are still some unresolved questions for future research. For example, the abnormal return on the portfolio with the highest oil betas is significant. One may also consider other ways of decomposition of demand and supply shocks to oil price changes as in Kilian (2009) and Ready (2015). It is possible that there are other explanations for the return spreads in the past five years. First, it is possible that stocks with high oil betas earn lower future returns because they pay well when future investment opportunities are worsened by the intertemporal capital asset pricing model of Merton (1973). If high oil beta stocks provide useful hedges, their expected returns should be lower. As Campbell et al (2012) point out, there are two types of 26 deterioration in investment opportunities: declining expected returns and rising volatilities. In our sample from 2011 to 2015, we find weak support for this view because the correlation between oil price changes in the current month and the market return and volatility shocks in the next month are negligible. Second, the impact of oil price changes on stock returns may be due to behavioral biases. Blanchard (2016) argues that herding behavior can explain the recent positive relation between aggregate stock returns and oil price changes. The recent decline of oil prices was associated with elevated uncertainty. When some investors sold stocks in panic, others followed suit, causing the stock market to fall. Driesprong, Jacobsen, and Maat (2008) provide a behavioral story based on investors’ underreaction to oil price changes to explain why the oil price changes predict future aggregate returns negatively. We may use a related behavioral story to understand our cross-sectional results that stocks with high oil betas earn low future returns in the past five or two years. A decline of oil prices causes investors to be more likely to sell stocks with high oil betas. This may be due to herding behavior accompanied with elevated uncertainty. Investors’ underreaction to oil price changes causes the price of stocks with high oil betas to continue to fall. Thus these stocks should earn lower future returns. This story may be consistent with our finding of the importance of momentum and volatility factors and the recent narrative in the media. We leave a serious test of this story for future research. 27 References Ang, Andrew, Robert J. Hodrick, Yuhang Xing, and Xiaoyan Zhang, 2006, The cross-section of volatility and expected returns, Journal of Finance 51, 259-299. Bernanke, Ben, 2016, The relationship stocks and oil prices, http://www.brookings.edu/blogs/ben-bernanke/posts/2016/02/19-stocks-and-oil-prices#ftn2 Blanchard, Olivier, 2016, The price of oil, China, and http://voxeu.org/article/price-oil-china-and-stock-market-herding stock market herding. Blanchard, Olivier, and Jordi Gali, 2009, The Macroeconomic effects of oil price shocks: why are the 2000s so different from the 1970s? In J. Gali and M. Gertler (eds.), International Dimensions of Monetary Policy, University of Chicago Press (Chicago, IL), 373-428. Campbell John Y., Giglio S, Polk C, and Turley R., 2012, An Intertemporal CAPM with stochastic volatility. Working Paper, Harvard University. Carhart, Mark, 1997, On persistence in mutual fund performance, Journal of Finance 52, 57–82. Chen, Nai-fu, Richard Roll, and Stephen A. Ross, 1986, Economic forces and the stock market, Journal of Business 56, 383-404/ Cooper, Michael, Huseyin Gulen, and Michael Schill, 2008, Asset growth and the cross-section of stock returns, Journal of Finance 63, 1609–52. Davis, James, Eugene Fama, and Kenneth French, 2000, Characteristics, covariances, and average returns: 1929 to 1997, Journal of Finance 55, 389–406. Demirer, Riza, Shrikant P. Jategaonkar, and Ahmed A.A. Khalifa, 2015, Oil price risk exposure and the cross-section of stock returns: the case of net exporting countries, Energy Economics 49, 132-140. Driesprong, Gerben, Ben Jacobsen, and Benjamin Maat, 2008, Striking oil: Another puzzle, Journal of Financial Economics 89, 307-327. Fama, Eugene, and James MacBeth, 1973, Risk, return, and equilibrium–empirical tests, Journal of Political Economy 81, 607–636. Fama, Eugene, and Kenneth French, 1993, Common risk factors in the returns on stocks and bonds, Journal of Financial Economics 33, 3-56. Fama, Eugene, and Kenneth French, 1996, Multifactor explanations of asset pricing anomalies, Journal of Finance 51, 55–87. 28 Fama, Eugene, and Kenneth French, 2008, Dissecting anomalies, Journal of Finance 63, 1653– 1678. Fama, Eugene, and Kenneth French, 2015, A five-factor asset pricing model, Journal of Financial Economics 116, 1–22. Ferson, Wayne E., and Campbell R. Harvey, 1993, The risk and predictability of International equity returns, Review of Financial Studies 6, 527-566. Gibbons, Michael R., Steven A. Ross, and Jay Shanken, 1989, A test of the efficiency of a given portfolio, Econometrica 57, 1121-1152. Hamilton, James, 2009, Causes and consequences of the oil shock of 2007-08, Brookings Papers on Economic Activity 1, 215-259. Hamilton, 2014, Oil prices as an indicator of global economic conditions, http://econbrowser.com/archives/2014/12/oil-prices-as-an-indicator-of-global-economicconditions Hou, Kewei, Chen Xue, and Lu Zhang, 2015, Digesting anomalies: An investment approach, Review of Financial Studies, 28, 650-705. Huang, Roger D., R.W. Masulis, and H.R. Stoll, 1996, Energy shocks and financial markets, Journal of Futures Market 16, 1-27. Jones, Christopher M, and Kaul, 1996, Oil and stock markets, Journal of Finance 51, 463-491. Jegadeesh, Narasimhan, and Sheridan Titman, 1993, Returns to buying winners and selling losers: Implications for stock market efficiency, Journal of Finance 48, 65–91. Kilian, Lutz, 2009, Not All Oil Price Shocks Are Alike: Disentangling Demand and Supply Shocks in the Crude Oil Market, American Economic Review 99, 1053–69. Kilian Lutz, and Cheolbeom Park, 2009, The impact of oil price shocks on the US stock market, International Economic Review 5, 1267-1287. Newey, Whitney, and Kenneth West, 1987, A simple, positive definite, heteroskedasticity and autocorrelation consistent covariance matrix, Econometrica 55, 703–708. Novy-Marx, Robert, 2013, The other side of value: the gross profitability premium. Journal of Financial Economics 108, 1–28. Ready, Robert, 2015, Oil prices and the stock market, forthcoming in Review of Finance. 29 Sadorsky, Perry, 1999, Oil price shocks and stock market activity, Energy Economics 21, 449-469. Sadorsky, Perry, 2001, Risk factors in stock returns of Canadian oil and gas companies, Energy Economics 23, 17-28. Stambaugh, Robert, Jianfeng Yu, and Yu Yuan, 2012, The short of it: investor sentiment and anomalies, Journal of Financial Economics 104, 288–302. Xing, Yuhang, 2008, Interpreting the value effect through the Q-theory: an empirical investigation, Review of Financial Studies 21, 1767–1795. 30 Table 1. Returns and Abnormal Returns of Portfolios Formed on Oil Betas, 1986-2015 To estimate oil betas, we regress each stock’s daily excess returns on the market excess return and daily percentage changes in the price of crude oil futures (WTI) in excess of the risk-free rate. The slope on oil price changes is the oil beta. At the end of month t-1, we form decile portfolios based on the oil betas of all stocks, and track their returns in the following month. Portfolio excess return is value-weighted average of the returns on stocks in each portfolio. Decile 1 (L) includes stocks that have the lowest oil beta and decile 10 (H) has the highest oil betas. Column LMHO lists returns on a portfolio that buys decile 1 and sells decile 10. The upper panel lists excess returns (Ret) in monthly percentage and their t-values (t) and the bottom panel lists abnormal returns (Alpha) in monthly percentage and their t-values (t) after adjusting risk exposure using the fivefactor model of Fama and French (2015). At the time of portfolio formation, we delete firms that have negative or zero book-to-market ratio, that have prices less than five dollars, or whose market capitalizations are less than the 10 percentile market capitalization among the NYSE firms. Stocks returns are from the CRSP. Factor data, risk-free rate and NYSE size breakpoints are from Kenneth French’s website. The WTI crude oil future price is from the Genesis Financial Technologies. The sample is daily from January 1, 1986 to December 31, 2015. 1986-2015 1986-2001 2002-2010 2011-2015 1 (L) 2 3 4 5 6 7 8 9 10 (H) LMHO Ret 0.45 0.67 0.68 0.72 0.62 0.51 0.70 0.58 0.73 0.38 0.07 t 1.39 2.50 2.77 3.06 2.59 2.25 2.86 2.30 2.66 1.05 0.29 Ret 0.57 0.86 0.81 0.90 0.66 0.54 0.89 0.62 0.71 0.37 0.20 t 1.25 2.38 2.33 2.62 1.93 1.68 2.51 1.71 1.83 0.74 0.61 Ret -0.01 -0.01 0.26 0.21 0.15 0.18 0.21 0.40 0.82 0.69 -0.71 t -0.02 -0.03 0.56 0.48 0.33 0.40 0.48 0.86 1.64 1.01 -1.33 Ret 0.89 1.31 1.02 1.09 1.34 1.03 0.97 0.80 0.65 -0.19 1.08 t 1.58 2.56 2.16 2.47 3.13 2.53 2.02 1.51 1.01 -0.25 1.94 -0.19 -0.01 -0.02 0.02 -0.07 -0.14 0.00 -0.02 0.15 -0.09 -0.10 1986-2015 Alpha t -1.22 -0.11 -0.22 0.21 -0.80 -1.50 -0.04 -0.23 1.28 -0.46 -0.38 1986-2001 Alpha -0.03 0.16 0.01 0.08 -0.16 -0.18 0.09 -0.01 0.02 -0.20 0.17 t -0.15 1.08 0.07 0.67 -1.18 -1.24 0.75 -0.05 0.12 -0.89 0.54 Alpha -0.35 -0.39 0.01 -0.07 -0.10 -0.14 -0.15 0.04 0.64 0.64 -0.99 t -1.21 -1.85 0.04 -0.45 -0.61 -1.00 -1.03 0.28 2.64 1.60 -1.75 Alpha -0.17 0.22 -0.03 0.07 0.35 0.18 -0.06 -0.26 -0.41 -1.29 1.12 t -0.70 1.12 -0.18 0.47 2.87 1.28 -0.36 -1.29 -1.58 -3.32 2.17 2002-2010 2011-2015 31 Table 2. Pre-Formation and Post-Formation Oil Betas, 1986-2015 To estimate oil betas, we regress daily excess returns on the market excess return and daily percentage changes in the price of crude oil futures (WTI) in excess of the risk-free rate. The slope on oil price changes is the oil beta. At the end of month t-1, we form decile portfolios based on the oil betas of all stocks, and track their oil betas in the portfolio formation month (Beta Pre) and in the following month (Beta Post). Portfolio level oil betas are value-weighted average of the oil betas of stocks in each portfolio. The t-values are computed using time series information. Decile 1 (L) includes stocks that have the lowest oil beta and decile 10 (H) has the highest oil betas. LMHO represents a portfolio that buys decile 1 and sells decile 10. Port. 1 (L) 2 3 4 -0.68 -0.35 -0.21 -0.12 Beta Pre -29.31 -27.16 -24.76 -18.27 t 0.00 -0.01 -0.01 -0.01 Beta Post -0.51 -2.37 -3.90 -2.44 t 5 6 7 8 9 10 (H) LMHO 1986-2001 -0.04 0.04 0.11 0.21 0.35 0.68 -1.36 -8.46 9.50 24.20 30.73 32.72 32.94 -33.32 0.00 0.00 0.00 0.01 0.01 0.03 -0.03 -0.87 -0.31 -0.24 2.18 2.07 3.66 -3.20 -0.52 -0.27 -0.17 -0.10 Beta Pre -29.37 -25.64 -22.18 -16.92 t Beta Post -0.03 -0.03 -0.03 -0.03 -3.24 -5.76 -5.67 -6.26 t 2002-2010 -0.04 0.02 0.09 0.17 0.29 -7.62 5.50 16.91 23.93 28.09 -0.02 -0.03 -0.01 0.01 0.06 -5.43 -7.31 -3.11 1.47 7.97 0.57 30.62 0.18 9.96 -1.09 -33.86 -0.21 -9.34 -0.61 -0.30 -0.18 -0.10 Beta Pre -18.50 -18.20 -15.31 -10.69 t Beta Post -0.03 -0.04 -0.02 -0.02 -2.90 -5.29 -3.28 -4.50 t 2011-2015 -0.03 0.03 0.10 0.19 0.32 -3.86 4.36 10.95 15.56 19.17 -0.02 0.00 0.00 0.02 0.06 -2.64 -0.23 0.65 2.98 5.68 0.68 19.90 0.14 7.62 -1.29 -22.35 -0.17 -7.37 32 Table 3. Alphas and Betas of Oil Portfolios using the Fama and French Five Factors The table presents alphas and betas of decile portfolios formed on oil betas as testing assets returns. We use the Fama and French (2015) five-factor model that includes the market excess returns (MKT), a size factor (SMB), a value factor (HML), an asset growth factor (CMA), and a profitability factor (RMW). We present alphas, betas and their t-values. LMHO stands for the difference in returns between portfolio 1 and portfolio 10. The GRS and its pvalue are joint tests of alphas following Gibbons, Ross and Shanken (1989). Returns are in monthly percentage. Panel A: 1986-2001 Port alpha MKT SMB HML CMA RMW t(alpha) t(MKT) t(SMB) t(HML) t(CMA) t(RMW) 1 2 3 4 5 6 7 8 9 10 -0.03 0.16 0.01 0.04 -0.16 -0.18 0.07 -0.01 0.02 -0.20 1.05 0.97 1.01 0.99 1.01 0.92 1.03 1.00 1.09 1.21 0.18 0.06 -0.01 -0.05 -0.09 -0.13 -0.07 -0.01 0.05 0.22 -0.11 0.05 0.04 0.14 -0.02 0.14 0.13 0.06 0.14 0.00 0.16 0.23 0.26 0.17 0.14 0.13 0.07 -0.13 -0.16 -0.59 -0.37 -0.20 -0.04 -0.06 0.17 -0.03 -0.04 -0.06 -0.04 0.06 -0.15 1.08 0.07 0.40 -1.18 -1.24 0.63 -0.05 0.12 -0.89 19.13 25.19 34.91 35.25 29.04 24.96 34.69 31.31 28.48 21.23 2.66 1.22 -0.24 -1.40 -2.09 -2.76 -1.86 -0.37 1.07 3.05 -0.98 0.68 0.64 2.56 -0.25 1.95 2.12 0.96 1.87 -0.04 1.70 3.50 5.33 3.47 2.31 2.09 1.31 -2.37 -2.45 -6.07 -2.44 -1.87 -0.45 -0.83 1.80 -0.34 -0.55 -0.69 -0.38 0.38 LMHO 0.17 -0.16 -0.03 -0.10 0.75 -0.42 0.54 -1.97 -0.35 -0.65 5.51 -1.96 GRS 0.67 p-value 0.77 alpha MKT SMB HML CMA RMW t(alpha) t(MKT) t(SMB) t(HML) t(CMA) t(RMW) 1 2 3 4 5 6 7 8 9 10 -0.35 -0.39 0.01 -0.07 -0.10 -0.14 -0.15 0.04 0.64 0.64 1.06 1.03 0.94 0.93 1.00 0.98 1.01 0.99 1.01 1.10 0.38 0.21 -0.04 -0.07 -0.14 -0.08 -0.01 0.09 -0.03 0.20 0.20 0.25 0.16 0.00 -0.15 0.00 -0.07 -0.13 -0.07 -0.03 -0.36 -0.13 -0.10 0.05 0.05 0.12 0.23 0.08 -0.01 -0.44 -0.32 -0.18 -0.03 0.13 0.22 0.15 0.11 0.15 -0.22 -0.68 -1.21 -1.85 0.04 -0.45 -0.61 -1.00 -1.03 0.28 2.64 1.60 13.33 17.77 22.00 21.17 22.20 24.68 24.93 23.65 15.17 10.05 3.13 2.42 -0.57 -1.11 -2.01 -1.26 -0.18 1.47 -0.30 1.21 1.62 2.78 2.37 0.05 -2.08 -0.04 -1.02 -1.92 -0.65 -0.19 -2.13 -1.04 -1.14 0.52 0.57 1.44 2.67 0.91 -0.09 -1.93 -1.78 -1.40 -0.28 1.31 2.21 1.70 1.21 1.59 -1.45 -2.76 LMHO -0.99 -0.05 0.18 0.24 0.09 0.36 -1.75 -0.29 0.75 0.96 0.27 1.05 GRS 1.66 p-value 0.10 Panel B: 2002-2010 Panel C: 2011-2015 alpha MKT SMB HML CMA RMW t(alpha) t(MKT) t(SMB) t(HML) t(CMA) t(RMW) 1 2 3 4 5 6 7 8 9 10 -0.17 0.22 -0.03 0.07 0.35 0.18 -0.06 -0.26 -0.41 -1.29 1.02 1.03 0.99 0.95 0.95 0.85 1.01 1.09 1.21 1.32 0.36 0.12 0.07 -0.10 -0.21 -0.05 0.06 0.01 0.18 0.37 -0.42 -0.27 -0.20 -0.30 0.01 -0.08 -0.04 0.13 0.62 1.37 -0.07 0.02 0.09 0.01 0.00 -0.13 0.04 -0.09 -0.17 0.41 -0.11 -0.02 0.08 0.49 0.14 0.24 0.02 -0.10 -0.64 -1.48 -0.70 1.12 -0.18 0.47 2.87 1.28 -0.36 -1.29 -1.58 -3.32 14.14 17.58 21.24 22.65 26.52 20.68 22.20 18.58 15.93 11.59 2.82 1.14 0.82 -1.27 -3.31 -0.64 0.79 0.11 1.31 1.81 -2.24 -1.76 -1.62 -2.69 0.08 -0.77 -0.36 0.82 3.11 4.60 -0.36 0.15 0.71 0.04 -0.05 -1.19 0.33 -0.59 -0.85 1.34 -0.41 -0.10 0.46 3.01 1.03 1.52 0.09 -0.44 -2.18 -3.35 LMHO 1.12 -0.31 -0.01 -1.79 -0.48 1.37 2.17 -2.03 -0.03 -4.54 -1.18 2.33 GRS 2.13 p-value 0.04 33 Table 4. Alphas and Betas of Oil Portfolios using the Momentum Factor The table presents alphas and betas of decile portfolios formed on oil betas as testing assets returns. The factors are a momentum factor and the Fama and French (1993, 1996) three factors, which include the market excess returns (MKT), a size factor (SMB), and a value factor (HML). The momentum factor (MOMF) is the return on a portfolio that long the 10th decile and short the 1st decile of stocks formed on past stock price momentum. We present alphas, betas and their t-values. LMHO stands for the difference between portfolio 1 and portfolio 10. The GRS and its pvalue are joint tests of alphas following Gibbons, Ross and Shanken (1989). Returns are in monthly percentage. Panel A: 1986-2001 Port alpha MKT SMB HML MOMF t(alpha) t(MKT) t(SMB) t(HML) t(MOMF) 1 2 3 4 5 6 7 8 9 10 0.14 0.28 0.05 0.03 -0.20 -0.12 0.14 0.00 -0.03 -0.27 1.08 0.99 1.02 1.01 1.00 0.93 1.03 1.00 1.09 1.18 0.10 -0.03 -0.09 -0.10 -0.11 -0.17 -0.09 0.02 0.09 0.39 -0.28 0.01 0.11 0.18 0.14 0.16 0.11 -0.02 0.07 -0.19 -0.13 -0.06 0.01 0.02 0.07 -0.02 -0.03 -0.04 -0.01 -0.05 0.65 1.82 0.45 0.27 -1.52 -0.82 1.22 0.00 -0.20 -1.12 21.58 26.83 35.25 37.29 31.83 26.62 37.42 33.40 30.07 20.22 1.59 -0.61 -2.43 -3.04 -2.87 -3.97 -2.78 0.41 2.10 5.45 -3.84 0.22 2.62 4.50 2.98 3.25 2.85 -0.52 1.30 -2.23 -4.54 -2.75 0.79 1.45 4.16 -0.82 -1.96 -2.13 -0.49 -1.54 LMHO 0.0041 -0.09 -0.29 -0.09 -0.08 1.20 -1.16 -2.91 -0.77 -1.69 GRS 0.79 p-value 0.65 alpha MKT SMB HML MOMF t(alpha) t(MKT) t(SMB) t(HML) t(MOMF) 1 2 3 4 5 6 7 8 9 10 -0.55 -0.47 -0.04 -0.03 -0.04 -0.07 -0.04 0.10 0.59 0.32 1.02 1.01 0.91 0.88 0.98 0.96 0.99 1.00 1.12 1.34 0.50 0.25 0.02 -0.04 -0.12 -0.09 -0.07 0.07 -0.13 0.12 -0.01 0.16 0.10 0.02 -0.10 0.06 0.03 -0.06 -0.05 -0.20 -0.13 -0.05 -0.06 -0.03 0.00 0.01 0.04 0.04 0.10 0.10 -2.07 -2.38 -0.26 -0.17 -0.23 -0.49 -0.29 0.70 2.68 0.83 14.92 19.45 24.55 22.23 23.42 26.17 26.74 26.80 19.56 13.15 4.43 2.95 0.34 -0.64 -1.70 -1.47 -1.13 1.13 -1.38 0.71 -0.12 2.01 1.68 0.31 -1.50 1.06 0.59 -1.01 -0.59 -1.28 -4.21 -2.22 -3.42 -1.79 -0.08 0.83 2.52 2.31 3.76 2.10 LMHO -0.87 -0.32 0.38 0.19 -0.23 -1.71 -2.40 1.74 0.92 -3.79 GRS 1.17 p-value 0.32 alpha MKT SMB HML MOMF t(alpha) t(MKT) t(SMB) t(HML) t(MOMF) 1 2 3 4 5 6 7 8 9 10 -0.17 0.25 0.00 0.11 0.33 0.09 -0.05 -0.26 -0.49 -1.11 1.02 1.02 0.98 0.94 0.96 0.88 1.01 1.09 1.23 1.23 0.39 0.11 0.03 -0.13 -0.22 -0.02 0.05 0.05 0.27 0.35 -0.50 -0.31 -0.18 -0.08 0.10 0.12 -0.04 0.06 0.35 0.45 -0.02 -0.02 -0.01 -0.01 0.02 0.05 0.00 -0.02 0.01 -0.13 -0.67 1.27 0.03 0.71 2.67 0.66 -0.29 -1.26 -1.79 -2.61 14.11 17.44 20.84 20.81 26.63 21.67 22.07 18.53 15.43 9.98 3.52 1.26 0.47 -1.86 -3.97 -0.39 0.72 0.53 2.24 1.82 -3.36 -2.61 -1.85 -0.89 1.39 1.46 -0.45 0.49 2.12 1.79 -0.36 -0.56 -0.29 -0.46 0.78 2.05 -0.08 -0.46 0.28 -1.59 LMHO 0.94 -0.22 0.04 -0.95 0.11 1.73 -1.38 0.18 -2.93 1.08 GRS 1.76 p-value 0.09 Panel B: 2002-2010 Panel C: 2011-2015 34 Table 5. Alphas and Betas of Oil Portfolios using the Volatility Factor The table presents alphas and betas of decile portfolios formed on oil betas as testing assets returns. The factors are a volatility factor and the Fama and French (1993, 1996) three factors, which include the market excess returns (MKT), a size factor (SMB) and a value factor (HML). The volatility factor (VOL) is the return on a portfolio that longs the 10th decile and shorts the 1st decile of stocks formed on total volatility. We present alphas, betas and their tvalues. LMHO stands for the difference between portfolio 1 and portfolio 10. The GRS and its p-value are joint tests of alphas following Gibbons, Ross and Shanken (1989). Returns are in monthly percentage. Panel A: 1986-2001 Port alpha MKT SMB HML VOL t(alpha) t(MKT) t(SMB) t(HML) t(VOL) 1 2 3 4 5 6 7 8 9 10 -0.02 0.16 0.04 0.03 -0.11 -0.19 0.03 -0.07 0.00 -0.22 1.04 1.01 1.05 1.03 1.03 0.96 1.07 1.01 1.05 1.08 -0.03 -0.01 -0.03 -0.04 -0.03 -0.11 -0.03 0.01 0.02 0.18 -0.02 0.02 0.01 0.07 -0.03 0.08 0.03 0.00 0.19 0.16 0.15 -0.01 -0.07 -0.07 -0.09 -0.06 -0.07 0.00 0.08 0.23 -0.10 1.02 0.31 0.28 -0.88 -1.39 0.30 -0.56 0.03 -0.97 18.77 24.87 34.20 36.11 29.94 25.89 36.39 30.80 27.44 18.09 -0.39 -0.21 -0.65 -1.07 -0.65 -2.17 -0.78 0.30 0.39 2.21 -0.18 0.23 0.23 1.47 -0.42 1.22 0.63 -0.03 2.77 1.55 3.31 -0.39 -2.63 -2.85 -3.34 -2.13 -2.78 0.18 2.60 4.83 LMHO 0.19 -0.04 -0.21 -0.18 -0.08 0.59 -0.42 -1.74 -1.16 -1.18 GRS 0.90 p-value 0.54 Alpha MKT SMB HML VOL t(alpha) t(MKT) t(SMB) t(HML) t(VOL) 1 2 3 4 5 6 7 8 9 10 -0.48 -0.46 -0.03 -0.04 -0.05 -0.11 -0.08 0.08 0.60 0.42 0.97 1.02 0.93 0.93 1.03 1.04 1.06 1.02 1.01 1.01 0.29 0.18 -0.05 -0.05 -0.09 -0.02 0.03 0.13 -0.06 0.06 0.09 0.20 0.13 0.03 -0.10 0.04 0.00 -0.09 -0.11 -0.23 0.16 0.04 0.03 -0.01 -0.04 -0.08 -0.09 -0.04 0.01 0.18 -1.76 -2.25 -0.17 -0.23 -0.34 -0.79 -0.60 0.56 2.55 1.08 11.34 15.90 19.86 19.23 20.51 24.80 24.54 22.29 13.66 8.32 2.42 2.06 -0.69 -0.81 -1.31 -0.35 0.48 2.07 -0.60 0.36 0.78 2.43 2.24 0.56 -1.61 0.74 -0.08 -1.52 -1.13 -1.47 3.34 1.17 1.22 -0.56 -1.34 -3.43 -3.81 -1.80 0.16 2.69 LMHO -0.90 -0.05 0.23 0.31 -0.02 -1.65 -0.27 0.95 1.44 -0.25 GRS 1.47 p-value 0.15 alpha MKT SMB HML VOL t(alpha) t(MKT) t(SMB) t(HML) t(VOL) 1 2 3 4 5 6 7 8 9 10 0.00 0.28 -0.03 -0.02 0.27 0.06 -0.11 -0.29 -0.24 -0.80 0.92 1.00 0.99 1.01 1.00 0.91 1.04 1.11 1.11 1.04 0.19 0.06 0.05 -0.01 -0.13 0.09 0.11 0.06 0.04 -0.16 -0.40 -0.26 -0.17 -0.11 0.04 0.00 -0.06 0.08 0.41 0.82 0.14 0.04 -0.01 -0.08 -0.06 -0.08 -0.04 -0.01 0.16 0.34 0.02 1.36 -0.16 -0.15 2.24 0.40 -0.68 -1.40 -0.93 -2.04 12.16 15.41 18.94 20.95 25.66 20.50 20.67 16.79 13.26 8.34 1.44 0.53 0.59 -0.14 -2.00 1.19 1.30 0.52 0.29 -0.76 -3.15 -2.42 -1.97 -1.32 0.69 0.02 -0.73 0.74 2.94 3.96 2.60 0.81 -0.36 -2.43 -2.19 -2.49 -1.20 -0.17 2.74 3.98 LMHO 0.80 -0.11 0.35 -1.22 -0.21 1.47 -0.66 1.17 -4.21 -1.71 GRS 1.61 p-value 0.13 Panel B: 2002-2010 Panel C: 2011-2015 35 Table 6. Oil Beta and Firm Characteristics The table presents slopes and t-values (t) from the Fama-MacBeth (1973) regression of regressing oil betas at the time of portfolio formation on stock characteristics. Following Fama and French (2015), the following accounting variables are used: market capitalization (MC), book-to-market ratio (BM), asset growth (AG), and operating profitability (OPE). Additional control variables include the price momentum in the past 12 month (MOM) and stocks total volatility (TVOL). The Adj. R2 is the adjusted R-squared. The Obs. is the number of observations. Our sample is monthly from 1986 to 2015. Each Subsample starts from January to December. The t-values are Newey-West (1987) with four lags. Three (two, one) stars indicate significance at 1% (5%, 10%) level. Sample 1986-2001 2002-2010 2011-2015 2014-2015 1986-2015 Intercept 0.0157* -0.0445* -0.0375 -0.042 -0.0114 t 1.84 -1.75 -1.02 -1.41 -1.01 MC -0.0007 0.0050* 0.0065* 0.0091*** 0.0022* t -0.76 1.8 1.68 2.98 1.79 BM 0.0028 0.0045 0.0176*** 0.0281*** 0.0058*** t 1.49 1.3 2.91 3.1 3.1 AG -0.002 0.0149*** 0.0071*** 0.0079* 0.0046*** t -1.21 4.23 2.74 2 2.71 OPE -0.0019** -0.0030*** 0.0003 0.0011 -0.0019*** t -2.13 -3.98 0.33 0.96 -3.33 MOM -0.0015 0.0486** -0.0494* -0.0863* 0.0056 t -0.47 2.52 -1.81 -1.78 0.66 TVOL -1.6137 7.0221* 20.8625* 25.4074 4.7585* t -0.58 1.72 1.93 1.41 1.67 Adj R2 0.0248 0.0576 0.1152 0.1706 0.0499 Obs. 444432 227060 115403 46967 786895 36 Table 7. Predicting Future Returns Using the Oil Betas The table presents slopes and t-values (t), using the Fama-MacBeth (1973) regression, from regressing the next-month excess returns on the oil betas and controlling variables. Following Fama and French (2015), the following accounting variables are used: market capitalization (MC), book-to-market ratio (BM), asset growth (AG), and operating profitability (OPE). Additional control variables include the price momentum in the past 12 month (MOM) and stocks total volatility (TVOL). The Adj. R2 is the adjusted R-squared. The Obs. is the number of observations. Our sample is monthly from 1986 to 2015. Each Subsample starts from January to December. The t-values are Newey-West (1987) with four lags. Three (two, one) stars indicate significance at 1% (5%, 10%) level. Sample Intercept t Oil Beta t MC t BM t AG t OPE t MOM t TVOL t Adj R2 obs 1986-2001 0.0075** 0.0113* 2.12 1.91 -0.0034* -0.0017 -1.95 -1.02 -0.0003 -0.44 0.0013 0.99 -0.0021*** -3.28 0.0006 1.42 0.0054*** 3.49 -2.9287*** -4.74 0.0025 0.0477 483959 443197 2002-2010 0.0056 0.0105 0.86 1.28 0.0035 0.0022 0.95 0.65 -0.0007 -1.2 0.0007 0.47 -0.0016** -2.05 0.0006** 2.16 -0.0037 -0.77 -1.7677* -1.76 0.0084 0.0483 235721 226610 2011-2015 0.0120** 0.0139* 2.42 1.8 -0.0054 -0.0049 -1.16 -1.03 -0.0003 -0.58 -0.0005 -0.5 -0.0013 -1.6 0.0002 1.49 0.0041** 2.07 -0.5453 -0.93 0.0106 0.0414 143800 138707 37 2014-2015 0.0032 -0.0038 0.66 -0.36 -0.0194* -0.0198* -1.83 -1.81 0.0006 0.56 -0.0018 -0.91 0.001 1.01 0 0.08 0.0015 0.6 -0.7513 -0.86 0.0242 0.0655 49321 46881 1986-2015 0.0073** 0.0106** 2.56 2.52 -0.002 -0.0014 -1.11 -0.8 -0.0004 -0.88 0.0009 1.01 -0.0016*** -3.79 0.0005** 2.25 0.0025 1.39 -2.1767*** -4.66 0.0059 0.0469 839435 784995 Table 8. Abnormal Returns of Portfolios Formed on Oil Betas and Control Variables, 2014-2015 At the end of month t-1, we form qunitile portfolios based on the oil beta of each stock, and independently form quintile portfolios based on one of the following control variables: market capitalization (MC), book-to-market ratio (BM), asset growth (AG), operating profitability (OPE), momentum (MOM), short-term reversal (REV), and total volatility (TVOL). Portfolios are intersections from the two previous independent sorts. Portfolio excess returns (Ret) are value-weighted average of the returns on stocks in each portfolio. Alphas are computed using the Fama and French (2015) five factors. The t() represents t-values. Quintile 1 includes stocks that have the lowest values and quintile 5 includes stocks that have the highest values. The portfolio 15 is a portfolio that buys quintile 1 and sells quintile 5. Returns are in monthly percentage. MC 1 1 1 2 2 2 3 3 3 4 4 4 5 5 5 OIL 1 5 15 1 5 15 1 5 15 1 5 15 1 5 15 Ret 0.03 -0.38 0.41 0.72 -0.76 1.47 0.51 -0.86 1.37 0.58 -1.29 1.87 1.12 -0.81 1.93 t(Ret) 0.02 -0.33 0.61 0.70 -0.71 2.26 0.56 -0.74 1.83 0.71 -1.30 2.44 1.52 -0.96 2.84 Alpha 0.44 -0.01 0.45 0.80 -0.91 1.71 0.19 -1.00 1.18 0.03 -1.98 2.02 0.38 -1.34 1.71 t(Alpha) 1.04 -0.01 0.57 2.18 -1.87 2.32 0.58 -1.72 1.63 0.08 -3.72 2.58 1.52 -2.72 2.75 BM 1 1 1 2 2 2 3 3 3 4 4 4 5 5 5 OIL 1 5 15 1 5 15 1 5 15 1 5 15 1 5 15 Ret 1.25 -0.13 1.38 1.12 -0.84 1.96 0.97 -1.49 2.46 0.36 -1.07 1.44 -0.51 -0.82 0.31 t(Ret) 1.65 -0.15 2.53 1.26 -0.95 2.51 1.10 -1.44 2.92 0.48 -1.04 1.60 -0.71 -0.79 0.29 Alpha 0.26 -1.23 1.49 0.25 -1.83 2.07 0.73 -1.84 2.57 0.44 -1.21 1.65 -0.37 -0.96 0.59 t(Alpha) 0.66 -2.63 2.27 0.73 -3.25 2.96 1.38 -2.84 2.98 0.88 -2.03 1.75 -0.60 -1.58 0.56 AG 1 1 1 2 2 2 3 3 3 4 4 4 5 5 5 OIL 1 5 15 1 5 15 1 5 15 1 5 15 1 5 15 Ret 0.32 -1.27 1.59 0.50 -0.66 1.16 0.87 -0.62 1.49 1.49 -1.45 2.94 1.20 -0.68 1.89 t(Ret) 0.40 -1.43 1.99 0.70 -0.77 1.50 1.04 -0.78 2.17 1.93 -1.42 2.87 1.28 -0.56 2.09 Alpha -0.05 -1.42 1.38 0.30 -1.09 1.39 0.16 -0.98 1.15 0.72 -1.78 2.50 0.42 -2.09 2.51 t(Alpha) -0.15 -2.22 1.67 0.79 -2.10 1.91 0.41 -2.49 1.66 1.23 -2.86 2.40 0.83 -3.23 3.78 OPE 1 1 1 2 2 2 3 3 3 4 4 4 5 5 5 OIL 1 5 15 1 5 15 1 5 15 1 5 15 1 5 15 Ret 0.43 -0.96 1.39 1.00 -1.28 2.29 1.04 -0.74 1.78 0.79 -0.61 1.40 0.80 0.27 0.54 t(Ret) 0.48 -0.89 1.45 1.25 -1.26 2.48 1.23 -0.72 2.02 1.02 -0.69 2.08 1.05 0.33 0.99 Alpha 0.19 -1.22 1.41 0.57 -1.87 2.44 0.22 -1.16 1.38 0.25 -1.36 1.61 0.12 -0.29 0.41 t(Alpha) 0.29 -1.57 1.21 1.46 -2.89 2.84 0.52 -1.80 1.84 0.49 -2.95 2.03 0.27 -0.63 0.61 38 MOM 1 1 1 2 2 2 3 3 3 4 4 4 5 5 5 OIL 1 5 15 1 5 15 1 5 15 1 5 15 1 5 15 Ret 1.13 -1.11 2.24 0.36 -0.71 1.07 0.41 -0.53 0.94 0.83 -0.29 1.11 1.72 -0.29 2.01 t(Ret) 0.98 -0.88 2.00 0.47 -0.89 1.58 0.64 -0.63 1.41 1.03 -0.35 1.90 1.77 -0.27 2.89 Alpha 0.90 -1.29 2.19 -0.34 -0.95 0.61 -0.01 -1.23 1.22 0.35 -1.08 1.42 0.76 -1.57 2.33 t(Alpha) 0.95 -1.54 2.06 -1.06 -1.86 0.93 -0.03 -2.85 1.72 0.98 -2.44 2.40 1.74 -2.30 3.28 REV 1 1 1 2 2 2 3 3 3 4 4 4 5 5 5 OIL 1 5 15 1 5 15 1 5 15 1 5 15 1 5 15 Ret 0.91 -0.99 1.90 1.85 -0.45 2.30 0.28 -0.33 0.62 0.91 -0.99 1.90 0.55 -0.50 1.05 t(Ret) 1.07 -0.84 1.94 2.24 -0.48 3.28 0.39 -0.34 0.83 1.01 -1.06 2.59 0.72 -0.53 1.49 Alpha 0.34 -1.35 1.68 1.39 -1.23 2.62 -0.39 -0.91 0.53 0.48 -1.72 2.20 -0.25 -1.06 0.81 t(Alpha) 0.57 -1.69 2.14 3.40 -1.81 3.23 -0.79 -1.80 0.64 0.93 -3.19 2.66 -0.56 -1.91 1.01 TVOL 1 1 1 2 2 2 3 3 3 4 4 4 5 5 5 OIL 1 5 15 1 5 15 1 5 15 1 5 15 1 5 15 Ret 0.68 -0.45 1.12 0.26 -0.63 0.89 0.88 -0.50 1.38 0.97 -1.16 2.13 0.99 -1.49 2.48 t(Ret) 0.98 -0.56 1.72 0.47 -0.67 1.28 1.07 -0.56 1.59 1.12 -1.13 2.62 0.95 -1.30 2.20 Alpha 0.21 -0.40 0.62 0.01 -1.08 1.09 0.14 -1.06 1.19 0.32 -2.08 2.40 0.45 -1.94 2.39 t(Alpha) 0.55 -0.81 0.97 0.03 -1.69 1.43 0.33 -2.31 1.67 0.91 -3.30 3.20 0.59 -2.38 2.07 OIL 1 1 1 2 2 2 3 3 3 4 4 4 5 5 5 TVOL 1 5 15 1 5 15 1 5 15 1 5 15 1 5 15 Ret 0.68 0.99 -0.31 0.71 -0.04 0.76 0.66 0.90 -0.24 0.16 0.83 -0.67 -0.45 -1.49 1.04 t(Ret) 0.98 0.95 -0.38 1.21 -0.04 0.74 1.16 0.73 -0.21 0.25 0.64 -0.72 -0.56 -1.30 1.14 Alpha 0.21 0.45 -0.24 0.23 -0.70 0.93 0.13 0.57 -0.44 -0.11 0.12 -0.23 -0.40 -1.94 1.54 t(Alpha) 0.55 0.59 -0.28 0.56 -0.88 1.07 0.53 0.55 -0.40 -0.45 0.16 -0.29 -0.81 -2.38 1.90 39 Table 9 Oil Betas and Industries, 2014-2015 Panel A presents industries to which the high and low oil beta firms belong. High (low) oil betas use 80th percentile and 20th percentile cutoff values of oil betas. The column “Industry” indicates the numerical number of the industry in the Fama and French (1997) 48 industries. The column “# of Obs.” presents the number of firm-month observations. The column “%” presents the fraction of firm-month observations of a particular industry out of the total firm-month observations of high (or low) oil beta firms. Panel B presents value-weighted oil betas and the Fama-French five-factor alphas by industry. We rank the oil betas of 48 industries from high to low and report the average of next month return (Ret), the average of oil betas at current month (Oil Beta), the Fama-French five-factor abnormal returns, and their respective t-values of the 10 industries with the largest and smallest oil betas. We compute the five-factor alphas for each stock before we take averages by industry. The column “Ind” is the industry’s numerical code in the Fama and French 48 industries. The column “MC” is the value-weighted average of June market capitalization. The column “Obs” is the number of observations in each industry from 2014 to 2015. The row “All” averages (valueweighted) across all firms using all observations (49140) from 2014 to 2015. All returns are in monthly percentage. Panel A: 2014-2015 Industry # of Obs. 30 1479 34 952 48 841 21 457 42 432 36 427 13 398 44 387 14 351 31 339 Total High Oil Beta % Name 15.0% Oil Petroleum and Natural Gas 9.6% BusSv Business Services 8.5% Other 4.6% Mach Machinery 4.4% Rtail Retail 4.3% Chips Electronic Equipment 4.0% Drugs Pharmaceutical Products 3.9% Banks Banking 3.6% Chems Chemicals 3.4% Util Utilities Industry # of Obs. 34 1132 48 1132 44 723 13 714 42 702 36 452 40 333 45 330 12 304 31 295 58.0% Low Oil Beta % Name 11.5% BusSv Business Services 11.5% Other 7.3% Banks Banking 7.2% Drugs Pharmaceutical Products 7.1% Rtail Retail 4.6% Chips Electronic Equipment 3.4% Trans Transportation 3.3% Insur Insurance 3.1% MedEq Medical Equipment 3.0% Util Utilities 59.0% 40 Panel B: Average Beta and Abnormal Returns by Industry, 2014-2015 Ind Name Ret Oil Beta Alpha MC t(Ret) t(Oil Beta) t(Alpha) t(MC) Obs 30 Oil Petroleum and Natural Gas -1.19 0.2862 -0.48 167840 -6.74 42.06 -2.64 45.51 1985 29 Coal Coal -5.24 0.2785 -5.52 6884 -3.32 5.46 -8.53 17.23 67 28 Mines Non-Metallic and Industrial Metal Mining -1.70 0.1631 -1.12 20881 -2.27 6.19 -2.12 24.35 194 20 FabPr Fabricated Products -2.86 0.1433 -9.12 2201 -2.31 3.79 -15.51 21.27 78 27 Gold Precious Metals -1.24 0.1168 1.25 12588 -0.57 2.43 2.29 29.31 40 21 Mach Machinery -0.08 0.0653 -0.65 116531 -0.52 13.06 -5.58 40.64 1519 7 Fun Entertainment 1.08 0.0512 2.47 40221 2.31 2.77 12.76 32.93 445 22 ElcEq Electrical Equipment -0.61 0.0443 -1.74 21365 -2.28 5.32 -5.28 25.87 510 25 Ships Shipbuilding, Railroad Equipment 0.22 0.0388 0.07 5456 0.26 1.84 0.15 24.97 128 31 Util Utilities 0.24 0.0354 1.84 23859 1.75 7.63 21.89 62.05 1984 13 Drugs Pharmaceutical Products 1.18 -0.0435 1.07 129250 6.76 -3.24 5.63 63.08 1882 26 Guns Defense 1.75 -0.0489 3.25 45787 3.69 -3.17 20.30 30.79 110 1 Agric Agriculture 1.75 -0.0496 -3.59 2821 2.05 -2.11 -1.99 18.89 89 8 Books Printing and Publishing -0.29 -0.0514 0.46 28049 -0.73 -4.30 1.23 31.13 392 48 Other Other 1.10 -0.0582 1.60 49593 6.06 -9.02 2.85 45.58 3698 43 Meals Restaraunts, Hotels, Motels 0.50 -0.0616 -0.96 48816 2.27 -8.93 -4.20 40.85 859 40 Trans Transportation 0.22 -0.0637 0.24 39547 1.17 -8.66 2.01 46.53 1269 16 Txtls Textiles 0.78 -0.0667 -6.07 8860 1.16 -3.11 -16.62 20.44 103 33 PerSv Personal Services -0.87 -0.0681 1.89 4771 -2.25 -4.15 2.46 27.16 541 42 Rtail Retail 0.61 -0.0713 2.08 83445 4.62 -14.82 18.69 49.89 2588 1 to 48 All 0.48 0.0015 1.49 100607 15.45 1.14 23.73 177.25 49140 41 Table 10. Returns and Abnormal Returns of Portfolios Formed on Alternative Oil Betas, 2011-2015 This table presents returns on decile portfolios formed on four sets of alternative oil betas. The original oil betas are slopes from a regression of daily excess returns on the market excess return and daily percentage changes in the price of crude oil future (WTI) in excess of the risk-free rate. The first set of oil betas are slopes from a regression in which the daily percentage changes in the oil futures price in the previous regression is replaced by the oil futures price changes related to demand. The second set of betas uses oil price changes related to supply. The third and fourth sets include daily percentage changes in the VIX as an explanatory variable in estimating oil betas. To estimate the oil price changes related to demand and supply shocks, in each month, we regress daily oil price changes on daily percentage changes in the daily percentage change in the price of copper futures and the change in 10 year Treasury bond rate, and the daily percentage change in the VIX. The fitted value is treated as the oil price changes related to demand shocks and the residual is treated as the oil price changes related to supply shocks. Panel A presents excess returns and panel B presents abnormal returns. Decile 1 (L) includes stocks that have the lowest oil beta and decile 10 (H) has the highest oil betas. LMHO represents the return of a portfolio that buys decile 1 and sells decile 10. The copper futures price is from the Genesis Financial Technologies. The VIX is from the St. Louis Fed. The sample is daily from 2011 to 2015. All returns are in monthly percentage. Panel A: Excess Returns, 2011-2015 1 (L) 2 3 4 5 6 7 Portfolios formed on beta of the demand component of oil price changes 1.10 1.04 1.16 0.75 1.09 0.96 0.96 1.89 2.14 2.52 1.72 2.36 2.07 2.16 8 9 1.09 2.25 1.16 2.07 0.37 0.55 0.73 1.65 Portfolios formed on beta of the supply component of oil price changes 0.91 1.08 1.20 1.10 1.01 1.30 1.12 1.62 2.05 2.55 2.78 2.36 2.95 2.35 0.75 1.39 0.63 1.04 -0.09 -0.13 1.00 2.11 Portfolios formed on beta of the demand component of oil price changes (Controlling for VIX changes) 1.12 0.89 0.98 0.90 1.06 0.86 1.00 0.97 1.06 0.79 1.94 1.91 2.05 2.02 2.41 1.93 2.18 2.05 1.88 1.19 0.33 0.83 Portfolios formed on beta of the supply component of oil price changes (Controlling for VIX changes) 0.91 1.03 1.08 1.19 1.07 1.35 1.00 0.84 0.62 -0.20 1.63 1.91 2.44 2.94 2.58 2.95 2.06 1.64 0.98 -0.28 1.11 2.22 42 10 (H) LMHO Panel B: FF 5 factor alphas, 2011-2015 1 (L) 2 3 4 5 6 7 Portfolios formed on beta of the demand component of oil price changes 0.20 0.03 0.16 -0.23 -0.02 -0.06 0.07 0.72 0.16 1.23 -1.54 -0.15 -0.41 0.52 8 9 10 (H) LMH 0.13 0.63 0.10 0.45 -0.78 -2.43 0.98 2.13 Portfolios formed on beta of the supply component of oil price changes -0.09 -0.01 0.12 0.17 0.11 0.38 0.12 -0.39 -0.07 0.67 1.12 0.71 2.85 0.87 -0.27 -1.68 -0.52 -1.87 -1.04 -2.72 0.95 2.01 Portfolios formed on beta of the demand component of oil price changes (Controlling for VIX changes) 0.17 -0.03 -0.08 -0.02 0.11 -0.10 0.03 0.00 -0.07 -0.41 0.60 -0.13 -0.56 -0.13 0.76 -0.66 0.23 0.00 -0.29 -1.57 0.58 1.43 Portfolios formed on beta of the supply component of oil price changes (Controlling for VIX changes) -0.10 -0.15 0.05 0.30 0.15 0.38 -0.02 -0.14 -0.54 -1.15 -0.45 -0.79 0.32 1.70 1.14 2.74 -0.14 -0.91 -1.98 -3.02 1.05 2.23 43 198601 198607 198702 198708 198803 198809 198904 198910 199005 199011 199106 199112 199207 199301 199308 199403 199409 199504 199510 199605 199611 199706 199712 199807 199902 199908 200003 200009 200104 200110 200205 200212 200306 200401 200408 200502 200509 200603 200610 200704 200711 200805 200812 200907 201001 201008 201102 201109 201203 201210 201304 201311 201405 201412 201506 Figure 1. Daily WTI Crude Oil Futures Prices and Copper Futures Prices, 1986 to 2015 This figure plots the daily nominal front month crude oil future prices (WTI CL) and copper future prices (Copper HG) from 1986 to 2015. Data is from the Genesis Financial Technologies. WTI CL 1986-2015 160 140 120 100 80 60 40 20 0 44 198601 198607 198702 198708 198803 198809 198904 198910 199005 199011 199106 199112 199207 199301 199308 199403 199409 199504 199510 199605 199611 199706 199712 199807 199902 199908 200003 200009 200104 200110 200205 200212 200306 200401 200408 200502 200509 200603 200610 200704 200711 200805 200812 200907 201001 201008 201102 201109 201203 201210 201304 201311 201405 201412 201506 Copper HG 1986-2015 500 450 400 350 300 250 200 150 100 50 0 45 Figure 2. Returns of Portfolios Formed on Oil Betas, 1986-2015 To estimate oil betas, we regress daily excess returns on the market excess return and daily percentage changes in the price of crude oil futures (WTI) in excess of the risk-free rate. The slope on oil price changes is the oil beta. At the end of month t-1, we form decile portfolios based on the oil beta of each stock, and track their return in the following month. Portfolio excess return is value-weighted average of the returns of stocks in each portfolio. Decile 1 (L) includes stocks that have the lowest oil beta and decile 10 (H) has the highest oil betas. LMHO represents the return on a portfolio that buys decile 1 and sells decile 10. At the time of portfolio formation, we delete firms that have negative or zero book-to-market ratio, that have prices less than five dollars, or whose market capitalizations are less than the 10 percentile market capitalization among the NYSE firms. Stocks returns are from the CRSP. Factor data, risk-free rate and NYSE size breakpoints are from Kenneth French. The WTI crude oil future price is from the Genesis Financial Technologies. The sample is daily from 1986 to 2015. We average monthly returns in each year to obtain returns by calendar years. The left panel presents returns and the right panel presents t-values in each year. LMHO return LMHO t-value 0.03 2.5 0.025 2 0.02 1.5 0.015 1 0.01 0.5 0.005 0 0 -0.5 -0.005 -0.01 -1 -0.015 -1.5 -0.02 -2 46