5.1 Sin Portfolio Performance
This section examines the results obtained from time-series regressions of portfolios consisting of sin. I begin by performing series of regressions for the Sin Portfolio (the excess monthly return of a value-weighted portfolio consisting of alcohol, tobacco and gambling stocks). By performing regressions using the Capital Asset Pricing Model, the Fama-French Three-Factor Model and the Carhart Four-Factor Model, I test my hypothesis and examine whether the Sin Portfolio has provided abnormal positive returns over the period from 2000 to 2021.
Results of the first estimation based on Capital Asset Pricing Model are shown in Table 2. The first estimation shows the excess monthly return of value-weighted Sin Portfolio over the excess monthly return of a value-weighted market portfolio. Results show that the alpha coefficient is 0.43 percentage points monthly implying the abnormal returns over the market portfolio. Alpha is significant at 5% confidence level. Market beta measures how sin stock portfolio moves when the overall stock market increases or decreases. Beta is 0.69, which is significant at 1% confidence level, implying that the Sin Portfolio is less sensitive compared to the overall stock market.
The second estimation is based on Fama-French Three-Factor Model and results are shown in Table 2. The second estimation presents the monthly return of Sin Portfolio over the market portfolio and two other portfolios – SMB and HML. Results show that when two factors are added, alpha coefficient is slightly lower in the second regression at 0.42 percentage points, which is still significant at 5% confidence level. SMB coefficient is -0.17, significant at 10% confidence level, signaling that the Sin Portfolio is slightly weighted towards stocks with high market capitalization.
HML coefficient is not significant, implying that Sin Portfolio is not considerably loaded with either value or growth stocks.
The third estimation is based on Carhart Four-Factor Model presented in Table 2. In the third estimation, I add MOM to the regression. Four-Factor alpha is 0.40 percentage points, which stays significant at 5% confidence level. MOM is 0.08 which is significant at 10% confidence level, implying that Sin Portfolio is slightly weighted toward stocks that have performed better than the overall market.
Table 2. Sin Portfolio Regressions: 2000-2021
Notes: The table reports coefficients obtained from the time-series regressions of the Sin Portfolio (the excess monthly return of a value-weighted portfolio of sin stocks – alcohol, tobacco and gambling).t-statistics are in parentheses.
Each regression is estimated using monthly data for the period from January 2000 to September 2021. I estimate the coefficients with Capital Asset Pricing Model (i), Fama-French Three-Factor Model (ii) and Carhart Four-Factor Model (iii).ALPHA is the abnormal return over the market portfolio.MKTis the excess monthly return of the value-weighted market portfolio consisting of Western-European stocks.SMB is the return of a portfolio long small market capitalization stocks and short large market capitalization stocks.HML is the return of a portfolio long high book-to-market stocks and short low book-to-book-to-market stocks.MOM is the return of a portfolio long past 12-month highest return and short past 12-month lowest return.*** significant at 1%; ** significant at 5%; * significant at 10%.
(i) (ii) (iii)
The main finding in Table 2 is that in the estimations, alpha coefficient ranges between 0.40 and 0.43 percentages, which are all significant at the 5% confidence level. Significant results of alpha are in harmony with previous findings that suggest sin stocks outperform the market. Market beta ranges between 0.69 and 0.72 which is significant in all models at 1% level, showing that the Sin Portfolio is less volatile than the market portfolio, which has beta equal to one. Beta less than one
is typical in sin industries which are stated to be defensive. However, it is notable that the percentage of alcohol stocks in the Sin Portfolio is higher than other industries during the period.
R-squared remains at an appropriate level at around 51% in all models.
5.2 Industry portfolios and excluding each industry
In this section, I examine the characteristics of sin industries individually. I estimate regressions using the Carhart Four-Factor Model. I perform the regression for each value-weighted industry portfolio (Alcohol Portfolio, Tobacco Portfolio, Gambling Portfolio) separately and excluding each industry from the Sin Portfolio one by one. I estimate industries separately to further understand the characteristics and performance of each field of business. Excluding each industry one by one is essential to notify if there are industries that drive the results of the Sin Portfolio.
Results of regressions are shown in Table 3 below.
I begin by examining abnormal returns of each industry portfolio. Each industry portfolio has positive alpha coefficient, which implies that each industry has outperformed the market portfolio, stocks of Western Europe. Whereas there are no significant abnormal returns in Alcohol or Tobacco portfolios, Gambling returns 1.2 percentage points, which is significant at 1% level.
Results show that gambling has provided significant returns over the market portfolio and compared to other industries during the past two decades. Each industry returns market beta significant at 1% ranging from 0.64 to 0.78. Results suggest that each industry is less volatile than the overall market. Tobacco and Gambling portfolios have SMB coefficients of -0.45 and 0.63, respectively. Both results are significant at 1% level. Results imply that the Tobacco portfolio is considerably weighted towards stocks with high market capitalization. This is concordant in the real world since there are only a few major players in the tobacco industry who dominate the market and have acquired smaller companies. Gambling also shows HML coefficient of -0.25 which is significant at 1% level. This would imply that the portfolio is weighted towards stocks with low book-to-market ratio i.e., growth stocks.
I am also interested in how the results occur as I exclude each sin industry from Sin Portfolio one by one. Interestingly, whereas alpha is positive in each portfolio, it is not significant anymore after excluding gambling stocks from the Sin Portfolio. Results suggest that gambling stocks’ returns
are driving the Sin Portfolio’s performance at least partly, which is somewhat contradictory to previous studies. However, gambling stocks’ share in Sin Portfolio is rather small, as the alcohol stocks have the most substantial weight in the portfolio. Market betas range from 0.68 to 0.74 which are all significant at 1%, suggesting that each industry is considerably less volatile compared to the market portfolio.
Table 3. Industry specific regressions and Excluded industries: 2000-2021
Notes: The table reports coefficients obtained from the time-series regression of the excess monthly return of each industry portfolio (value-weighted portfolio of alcohol, tobacco or gambling) and return of the Sin Portfolio when industries are excluded one by one from the portfolio. For instance, ExcAlcohol is the Sin Portfolio excluding alcohol stocks.t-statistics are in parentheses. Each regression is estimated using monthly data for the period from January 2000 to September 2021. I estimate the coefficients with Carhart Four-Factor Model.ALPHA is the abnormal return over the market portfolio.MKTis the excess monthly return of the value-weighted market portfolio consisting of Western-European stocks.SMB is the return of a portfolio long small market capitalization stocks and short large market capitalization stocks.HML is the return of a portfolio long high market stocks and short low book-to-market stocks.MOM is the return of a portfolio long past 12-month highest return and short past 12-month lowest return.*** significant at 1%; ** significant at 5%; * significant at 10%.
Alcohol Tobacco Gambling ExcAlcohol ExcTobacco ExcGambling ALPHA 0.0035 0.0039 0.012*** 0.0056** 0.0044** 0.0034
(1.54) (1.14) (3.83) (1.98) (2.07) (1.56)
MKT 0.74*** 0.64*** 0.78*** 0.68*** 0.74*** 0.71***
(14.72) (8.54) (11.61) (10.92) (16.15) (14.96)
SMB -0.16 -0.45*** 0.63*** -0.23* -0.10 -0.24**
(-1.58) (-2.92) (4.61) (-1.82) (-1.09) (-2.46)
HML 0.06 0.18 -0.25** 0.051 0.026 0.087
(0.77) (1.49) (-2.37) (0.52) (0.35) (1.15)
MOM 0.042 0.18** -0.18** 0.14** 0.031 0.09*
(0.79) (2.25) (-2.43) (2.13) (0.63) (1.74)
R-square 0.49 0.24 0.45 0.33 0.54 0.49
Observations 260 260 260 260 260 260
5.3 Sin stocks during Covid-19
I use the Carhart Four-Factor Model to perform time-series regressions of portfolios consisting of stocks in sinful industries during the Covid-19 pandemic. I add a dummy variable to regression model, that equals one during the Covid-19 pandemic period from February 2020 to September 2021 and zero otherwise. I examine the results of the Sin Portfolio and each sin industry separately.
Results of the regressions are shown in Table 4 below.
To test my second hypothesis that states sin stocks have provided investors abnormal returns over the period of the Covid-19 pandemic, my main interest is in the alpha coefficient. All portfolios returned positive alpha, although neither the Sin, Alcohol nor Tobacco Portfolio show significant results. Gambling Portfolio has alpha of 0.94 percentage points, which is significant at 1%
confidence level. Portfolios have beta ranging from 0.63 to 0.75, which are all significant at 1%, implying that each industry has been less volatile than the overall market during the pandemic.
Tobacco and Gambling Portfolios’ SMB coefficients are -0.46 and 0.61, respectively. Both results are significant at 1% level.
Results suggest that sin stocks have provided abnormal returns during the pandemic in some industries. While the other portfolios have no significant returns over the market portfolio, the Gambling portfolio has highly significant abnormal returns. Results, along with studies suggest that the gambling industry has benefitted from the pandemic in their operations. Hodgins and Stevens (2021) discuss in their study that while for instance sport betting and variety of land-based gambling operations have suffered due to restrictions such as lockdowns to avoid the spread of the coronavirus, at the same time online gambling has increased by around 15%. Overall, the companies have low beta, which indicates that each industry is less volatile than the market portfolio, consisting of Western-European stocks. Results are in line with previous literature (see eg. Salaber, 2009; Troberg, 2016) and reinforce the defensive characteristics of sin industries. The defensive nature of sin stocks might be due to addictive traits of products and services that sin industries provide.
Table 4. Sin Regressions Covid-19
Notes: The table reports coefficients obtained from the time-series regressions of a Sin Portfolio (the excess monthly return of a value-weighted portfolio of sin stocks – alcohol, tobacco and gambling) and each sin industry separately.t-statistics are in parentheses. Regression is estimated using monthly data for the period from January 2000 to September 2021. I estimate the coefficients Carhart Four-Factor Model.ALPHA is the abnormal return over the market portfolio.MKT is the excess monthly return of the value-weighted market portfolio consisting of Western-European stocks.SMB is the return of a portfolio long small market capitalization stocks and short large market capitalization stocks.HML is the return of a portfolio long high book-to-market stocks and short low book-to-market stocks.MOM is the return of a portfolio long past 12-month highest return and short past 12-12-month lowest return.CDUM is a dummy variable that equals one during the Covid-19 pandemic period from February 2020 to September 2021 and zero otherwise. ***
significant at 1%; ** significant at 5%; * significant at 10%.
Sin Alcohol Tobacco Gambling
ALPHA 0.0033 0.0032 0.0032 0.0094***
(1.54) (1.30) (0.87) (2.90)
MKT 0.71*** 0.73*** 0.63*** 0.75***
(14.72) (14.36) (8.28) (11.10)
SMB -0.20 -0.17 -0.46*** 0.61***
(-1.58) (-1.61) (-2.95) (4.43)
HML 0.064 0.064 0.18 -0.23**
(0.77) (0.80) (1.53) (-2.19)
MOM 0.08 0.042 0.18** -0.18**
(0.79) (0.78) (2.25) (-2.44)
CDUM 0.007 0.004 0.007 0.023**
(0.99) (0.45) (0.59) (2.18)
R-square 0.53 0.49 0.24 0.46
Observations 260 260 260 260