Sin-stock performance before, during and after the financial crisis of 2008

(1)

S

IN

-

STOCK

P

ERFORMANCE BEFORE

,

DURING

AND AFTER THE FINANCIAL CRISIS OF

2008

Jeremy Renoult 10250727 Organization & Economics BSc ECB University of Amsterdam

(2)

D

ECLARATION

This dissertation is the result of my own work and includes nothing, which is the outcome of work done in collaboration except where specifically indicated in the text. It has not been previously submitted, in part or whole, to any university of institution for any degree, diploma, or other qualification. Signed:______________________________________________________________ Date:___________________05/07/2016______________________________________________ Jeremy Renoult

(3)

INHOUDSOPGAVE

Page

1. Introduction

4 2. Related Literature

6 3. Hypothesis

8 4. Methodology

9 5. Data and Performing the Analysis

12 5. Results

13 6. Discussion and Concluding Remarks

19 7. References

21

(4)

1. Introduction

Pension funds in the US are under pressure to do socially responsible investments because they are under public scrutiny says Sethi (2005). Also in the Netherlands the largest trade union and several influential charities like Amnesty International called for an investigation to disclose where the largest Dutch banks invest their money. This led to the fair money-index where consumers can find information about all the large banks and insurance companies and if they invest socially responsible. This is a trend that is recognized by Geczy and Sambaugh (2003), they refer to the Social Investment Forum that estimates that in 2001 about $2.34 trillion dollars undergoes some kind of social screening which accounts for about 12%. After looking into the Social Investment Forums report of 2014 they show that in 2012 the total assets undergoing this screening grew to $13.3 trillion making up 21,5% of total assets. And in 2014 it even reached $21.5 trillion and a 30,2%. It also shows that US is leading this change in social responsible investing with 58,8% of investments being socially screened This would suggest that a large percentage of investments have a socially responsible preference. Geczy and Sambaugh (2003) conclude that the opportunity-cost of socially responsible investing can be significant when they tested mutual funds with the Carhart four-factor model. That is the case if even a small portion of the fund is SRI, if the complete portfolio is SRI the opportunity-costs are even higher. Common industries that are regarded as sinful are alcohol, tobacco, weapons, gaming/gambling and nuclear energy. The view is that companies making money by taking advantage of people or the environment through for instance addiction are unethical. For alcohol, tobacco and gaming/gambling it is evident that these products and activities can have a negative effect on the health of the consumer. Could the addictive properties of tobacco, alcohol and gambling/gambling result in a stable demand for these products and result in companies that are not as sensitive to economic fluctuations? If this is true these sin-stocks could make a good investment to diversify a portfolio. Salaber(2009) discusses the effect of sin-stock returns during different economic cycles. One of his conclusions is that sin-stocks earn an abnormally high return during recessions, but not during times of growth. He does however state that sin-companies do not perform better when compared to industry-comparable stocks. For instance the

(5)

food-industry can also be used to diversify risk during times of recession. So summarizing he concludes that sin-stocks do perform better than the market during economic downturn, but they are not the only type of stocks that have this characteristic. So this poses the question, does a sin-portfolio outperform a neutral-portfolio, and is it consistent under different economic conditions? The main focus is to analyse if a sin-portfolio comprised out of all the alcohol, tobacco and gaming stocks active in the S&P500, Nasdaq, AMEX and NYSE, active from 01-01-2004 until 01-01-2015 outperforms a neutral-portfolio consisting all remaining stocks from those four indexes, active in the same period. First measuring Jensen- α’s with an OLS-regression of the Carhart four-factor model and than comparing the two portfolios with an independent sample t-test. Then the sin-portfolio will be split into sub-periods to test if the portfolio makes an abnormal return in any or all of the periods, measured through the Jensen’s-α. The first period (T1) builds up to the crisis, the second period (T2) covers the financial crisis and the third period (T3) represents the recovery of the economy. T1 (pre-crisis): January 2004 – January 2007 = 37 months T2 (crisis): February 2007 – April 2011 = 50 months T3 (After-crisis): May 2011- January 2015 = 45 months The third analysis tests if the industries represented in the sin-portfolio have a significant Jensen’s-α, to see if a possible abnormal return of the sin-portfolio can be attributed to a specific industry. Finally if the third analysis shows that a certain industry makes a significant and abnormal return it will also be split up in the sub-periods to analyze it in different economic cycles (T1, T2 and T3). This is also analyzed by looking at the Jensen’s-α. The current literature investigates the performance of sin-stocks in various ways, and for different periods. Almost all research is done up until the financial crisis of 2007. An interesting question is to see if all the statements made by previous articles are still true if it is tested in and around the financial crisis of 2007. Furthermore the sin-portfolio consisting out of alcohol, tobacco and gaming stocks are barely tested for individual industries to see if maybe just one of the industries is the sole cause of the possible undervaluation of the entire sin-portfolio. Because of the changing preferences of investors concerning social responsibility this investigation might find differences that

(6)

the older literature did not. For instance the neglect of sin stocks might have increased because investors are becoming more averse to sin stocks. Or the changing preferences might have an influence on the anti-cyclical properties of the sin-stocks. Starting with a summary of existing literature comparing different views, results and methods of research. Taking everything into account the hypotheses are setup and explained. Then the exact model, portfolios and data are discussed extensively to give a clear view of the research.

2. Related Literature

Hong and Kascperczyk (2009), Fabozzi (2008), Merton (1987), Carvell and Strebel (1987), Ghoul and Guedhami (2011) and Duran and Koh (2013) all conclude that stocks related to ‘sin’ industries are neglected by investors because social norms and public pressures force them to. Socially responsible investing (SRI) is becoming the norm for institutional investors and the effects are on performance are still being studied. Kübler (2001) describes a phenomenon that he calls bandwagon norms. These norms are characterized by the fact that once a critical proportion of het population follows the norm, it becomes very costly in terms of reputation not to follow it. Bandwagon norms may affect investor’s reputations, which in turn may result in different investment decisions. This could be a probable cause for investors to neglect sin stocks as it may have a negative effect on their reputation. This is called the neglect-effect. Hong and Kacperczyk (2009) find that institutional investors shun sin-stocks and by performing a regression they prove that these sin-stocks are 15-20% under-priced. In their regression Hong and Kacperczyk (2009) measure institutional investor holdings of sin-stocks. They hypothesize that institutional investors like pension funds, banks and universities are less willing to hold sin-stocks than comparable non-sin-stocks due to their public nature. Fabozzi (2008) draws the same conclusion as his sin stock portfolio produced an annual return of 19%, outperforming market benchmarks. His article also concludes that sin stock is undervalued because of sin aversion of the average investor. Durand and Koh (2013) investigate the difference between sin and socially responsible firms. They discover a difference in financing. Sin firms seem to have difficulties finding equity and have to turn to debt financing, as the debt market is less prone to public opinion than the equity market. The sin firms are forced to have higher levels of cash,

(7)

ceteris paribus. And this may force them to have sub-optimal financial policies. So Durand and Koh (2013) conclude that public opinion affects sin firms negatively through their financing. So sin-companies have higher financing costs than neutral or socially responsible-companies, but Durand and Koh as well as Hong and Kacperczyk control for leverage in their research and still measure an abnormal positive return for sin-stocks. Ghoul and Guedhami (2011) conclude that high-SRI firms have lower cost of equity and lower perceived risk than low-SRI firms. They prove this by using a sample of 12,915 US firm-specific variations and controlling for other firm-specific determinants as well as industry and year fixed effects. Where Fabozzi (2008) argues that sin industries are more disciplined by public opinion, intensifying analysis and monitoring maybe making them stronger. On the other hand he mentions that the industries are barely regulated in terms of pricing, maybe making more profit with higher margins. Kim and Venkatachalam (2011) build on the research by Hong and Kacperczyk (2009). They investigate if the lower quality or absence of information on sin stocks is a factor causing the neglect of the stocks. They conclude that the neglect of sin stocks by market participants is not attributable to financial reporting factors. They even mention that sin stocks provide information of higher quality. Salaber(2009) discusses the effect of sin-stock returns during different economic cycles. One of his conclusions is that sin-stocks earn an abnormally high return during recessions, but not during times of growth. He does however state that sin-companies do not perform better when compared to industry-comparable stocks. For instance the food-industry can also be used to diversify risk during times of recession. So summarizing he concludes that sin-stocks do perform better than the market during economic downturn, but they are not the only type of stocks that have this characteristic.

(8)

3. Hypotheses

The following four hypotheses will be tested: 3.1 Hypothesis 1 Sin-portfolio against neutral-portfolio H0:

α

sin

≤

α

neutral H1:

α

sin

>

α

neutral The first hypothesis is that the sin-portfolio outperforms the neutral-portfolio, measuring the difference in the α’s between the two portfolios. So first the two α’s are calculated with the Carhart four-factor model performing an OLS-estimation. This provides the sin-α and neutral-α’s . Then an independent sample t-test is carried out to see if the difference between the different α’s is significantly different from zero. 3.2 Hypothesis 2 Sub periods of sin-portfolio H0:

α

sinT1

=

α

sinT2

=

α

sinT3 =

0

H1:

α

sinT1

and/or

α

sinT2

and/or

α

sinT3 >

0

The Second hypothesis states that the sin-portfolio- α is significantly different from zero for the three sub-periods. T1 is the period before the crisis, T2 the period of the crisis and T3 the period after the crisis. Tested by performing an OLS-estimation of the Carhart four-factor model. 3.3 Hypothesis 3 Industry specific portfolios

H0:

α

alcohol

=

α

tobacco

=

α

gaming =

0

H1:

α

alcohol

and/or

α

tobacco

and/or

α

gaming >

0

The third hypothesis is that the three industries that are represented in the sin-portfolio are individually significantly larger than zero. Test performed with an OLS-regression of the Carhart four-factor model.

(9)

3.4 Hypothesis 4

Sub-periods for tobacco-portfolio

H0:

α

industryT1

=

α

industrieT2

=

α

industrieT3 =

0

H1:

α

industryT1

and/or

α

industrieT2

and/or

α

industrieT3 > 0

Combining the tests of the sub-periods and the industries. If there is a specific industry that makes an abnormal return it will also be tested for the sub-periods to investigate if it varies per period.

4. Methodology

4.1 The Carhart Four-Factor Model. The preferred model to measure abnormal returns of portfolios is the Carhart (1997) four-factor model. The model is tested extensively and also used by the related literature. Fabozzi (2008), Hong and Kacperczyk (2009), Kampf and Osthoff (2007) and Salaber (2009) are a few examples that also use the model. The model is constructed using the Fama and French (1993) three-factor model and adds the factor one-year momentum anomaly by Jegadeesh and Titman (1993). So the four-factor model is consistent with a model of market equilibrium taking four risk factors into account. It can be interpreted as a performance attribution model. In this research Jenson’s-α

is the indicator for a possible abnormal return of the portfolio. The four independent factors (RMRF, SMB, HML and MOM) are control variables. The model is represented as follows: (

R

t −

R

ft) =

α

t +

β

RMRFRMRFt +

β

SMBSMBt +

β

HMLHMLt +

β

MOMMOMt +

ε

t 4.1.2 Variable description - The dependent variable (

R

t −

R

ft) is the excess return of a given portfolio, which is calculated by taking the value weighted portfolio return and subtracting the risk-free rate.

(10)

- The intercept (α) is the value that represents a possible abnormal return of the portfolio. If the intercept is significantly different from zero there is an indication that the portfolio makes an abnormally higher (positive intercept) or abnormally low (negative intercept) return. - The RMRF-coefficient (β) controls the reaction of the portfolio to the market (systematic-risk). A lower (β) means less volatility, so during the financial crisis the portfolio with the lower (β) will react less strongly to the market. - SMB controls for the proportion of small vs. big firms measured in market-capitalization. If the SMB-coefficient is positive the portfolio predominantly consist out of small-cap stocks. This will also only be discussed if significant and important. - HML controls for the proportion of high-value vs. low-value stocks. High-value meaning high book-to-market ratios and visa-versa. If the HML coefficient is positive the portfolio consist predominantly out of high-value stocks. SMB represents the proportion of small vs. big firms measured in market-capitalization. If the SMB-coefficient is positive the portfolio predominantly consist out of small-cap stocks. - MOM controls for the momentum factor. The main concept is that stocks (especially bad-performing stocks) exhibit a short-term momentum. So if the MOM-factor is positive winners will keep winning, and losers will keep losing. 4.2 Building Portfolios This research will focus on stocks in US indexes just like most of the relevant previous research like Hong and Kacperczyk (2009). To build the sin and neutral-portfolios all the stocks in the S&P500, Nasdaq, AMEX and NYSE that were active at least from 01-01-2004 until 01-01-2015 are gathered. 4.2.1 Building the Sin-portfolio Hong and Kacperczyk (2009) selected the alcohol, tobacco and gaming industries and omit adult services, gambling and weapons industries. They argue that the number of companies that are active in de adult services industry is so small it has no effect on the sample. They also omit the weapons industry because most Americans do not consider weapons to be sinful.

(11)

The sin companies are identified in the same manner that Hong and Kacperczyk (2009) and Kampf and Osthoff (2007) make their selection. Stocks with SIC codes 2100-2199 belong to the alcohol group, and with codes 2080-2085 belong to the tobacco group. The second method for identifying stocks is with NIACS classifications, which identifies gaming stocks by the following codes: 7132, 71312, 713210, 71329, 713290, 72112 and 721120. But because the S&P500 is the only index that provides NIACS classifications another method was needed to identify gaming stocks in the remaining three indexes. Therefore the final method to identify these missing gaming stocks is by using information from the official Nasdaq-website. The website provides lists of stocks on all three indexes of stocks in the gaming-industry. Combining these identification methods provides a list of 33 companies that are considered sin-stocks and are active in the entire period. 4.2.2 Building the neutral-portfolio As a reference three random neutral portfolios are constructed. Just like the sin-portfolio, the neutral-portfolio is constructed using 33 random companies from the S&P500, Nasdaq, AMEX and NYSE. The random selection is done by alphabetically ordering the data and choosing three random letters and takes the first 33 companies that are active from 01-01-2004 till 01-01-2015. The three random selected letters are C, P and Z. 4.3 Testing for sub-periods The period from 01-01-2004 and 01-01-2015 was not a period without economic turmoil. The financial crisis is somewhere in the middle. The Federal Reserves Bank of St. Louis recorded a timeline. Their timeline starts in February of 2007 with the announcement of Freddie Mac that they wouldn't buy the most risky subprime mortgages and mortgage-related securities. The timeline ends in April of 2011 with the U.S. Senate committee on Investigations releasing their final report on its inquiry into key causes of the financial crisis. The financial crisis period has a span of 50 months in total. The period leading up to the crisis considering the data used spans over 37 months from January 2004 till February 2007. The period after the crisis starts in May of 2011 till January 2015. T1 (pre-crisis): January 2004 – January 2007 = 37 months T2 (crisis): February 2007 – April 2011 = 50 months

(12)

T3 (After-crisis): May 2011- January 2015 = 45 months The previous regression using the same factors and data is done for all three sub periods. 4.4 Industry specific testing To test where the difference in the intercept (α) between the sin and neutral portfolio comes from the same regression is done but the sin-portfolio is analysed per industry. There are three sin-industries represented in the portfolio. The first is Alcohol with 12 companies, Tobacco representing 7 companies, and gaming representing 14. Adding up to the sin-portfolio total of 33 companies.

4.3 Data and performing the analysis

The monthly risk-free rate, SMB, HML, RMRF and MOM are available on the Ken French website. Compustat provides the S&P500 data, and CRSP provides the data of the other three indexes. Only companies that were active in the entire period from 01-01-2004 until 01-01-2015 are selected. As described all the stocks in all the portfolios are identified, providing lists of company names per portfolio. The names are entered into the portfolio-tool of CRSP, which produces the monthly value weighted portfolio return for the following portfolios: Sin-portfolio – complete period (January 2004 – January 2015) Sin-portfolio – T1 (January 2004 – January 2017) Sin-portfolio – T2 (February 2007 - April 2011) Sin-portfolio – T3 (May 2011 – January 2015) Portfolios per industry (Alcohol, Tobacco and Gaming)- complete period Portfolios per industry (Alcohol, Tobacco and Gaming)- T1 Portfolios per industry (Alcohol, Tobacco and Gaming)- T2 Portfolios per industry (Alcohol, Tobacco and Gaming)- T3 Random neutral-portfolios C, P and Z – complete period Random neutral-portfolios C, P and Z – T1 (January 2004 – January 2017) Random neutral-portfolios C, P and Z – T2 (February 2007 - April 2011) Random neutral-portfolios C, P and Z – T3 (May 2011 – January 2015)

(13)

Now all the variables in the Carhart four-factor model are gathered and all the regressions are performed carried out using SPSS. The dependent variable is the value weighted portfolio return minus the risk free rate, and the independent variables are RMRF, SMB, HML and MOM.

5. Results

5.1 Descriptive Statistics Table 1. Descriptive statistics of the excess portfolio returns on all the tested portfolios.

Portfolio N Minimum Maximum Mean Std.Deviation

Sin 133 -.1587 .08320 .00991 .0374 SinT1 38 -.0698 .0755 .0112 .0263 SinT2 50 -.1587 .0832 .0055 .0494 SinT3 45 -.0456 .0661 .0140 .0291 NeutralC 133 -.1995 .1648 .0019 .0534 NeutralCT1 38 -.0517 .0518 .0049 .0259 NeutralCT2 50 -.1995 .1146 -.0026 .0707 NeutralCT3 45 -.1044 .1648 .4464 .0488 NeutralP 133 -.1457 .1008 .0047 .0388 NeutralZ 133 -.1814 .1285 .0063 .0466 Alcohol 133 -.2010 .0887 .0078 .0379 Gaming 133 -.2974 .8095 .0099 .1220 Tobacco 133 -.1422 .1532 .0163 .0462 TobaccoT1 38 -.1276 .1523 .0164 .0462 TobaccoT2 50 -.1422 .0996 .0114 .0521 TobaccoT3 45 -.0629 .0889 .0192 .0404 RMRF 133 -.1723 .1135 .0062 .0421 RMRFT1 38 -.0406 .0454 .0060 .0219 RMRFT2 50 -.1723 .1019 .0024 .0570 RMRFT3 45 -.0759 .1135 .0106 .0354 SMB 133 -.0425 .0579 .0014 .0227 SMBT1 38 -.0400 .0531 .0016 .0226 SMBT2 50 -.0423 .0579 .0039 .0246 SMBT3 45 -.0425 .0428 .0016 .0205 HML 133 -.0967 .0765 .0008 .0231 HMLT1 38 -.0185 .0456 .0078 .0157 HMLT2 50 -.0967 .0765 .0026 .0309 HMLT3 45 -.0337 .0460 .0011 .0167

(14)

MOM 133 -.3458 .1245 .0006 .0462 MOMT1 38 -.0536 .0526 .0018 .0229 MOMT2 50 -.3458 .1245 .0024 .0693 MOMT3 45 -.0795 .0664 .0030 .0241 RF 133 .0000 .0044 .0011 .0015 RFT1 38 .0006 .0044 .0025 .0013 RFT2 50 .0000 .0044 .0011 .00148 RFT3 45 .0000 .0001 .00002 .00004 T1 = January 2004 – January 2017, T2 = February 2007 - April 2011 and T3 = May 2011 – January 2015. When looking and comparing the descriptive statistics the following differences are interesting. The mean-excess return for the sin-portfolio is always higher than that of the neutral-portfolio, also for the sub-periods (T1, T2 and T3). And the tobacco-portfolio has an even higher return than the sin-portfolio. The standard deviation of the sin-portfolio almost doubles during the crisis (T2), but drops to the same level of the first period (T1) after the crisis (T3). The neutral-portfolio’s standard deviation increases more than three-fold during the crisis (T2) and slightly drops after the crisis but is still a lot higher than during the pre-crisis period (T1). The tobacco portfolio’s standard deviations are higher compared to the sin-portfolio over the sub-periods, but are closer together indicating a volatility that is more stable. The seemingly stable-volatility with the highest mean-return makes the tobacco-portfolio interesting to investigate. The Intercept (α) is the indicator for abnormal return of the tested portfolio. A positive significant α suggests a positive abnormal return and vice-versa. The control factors (RMRF, SMB, HML and MOM) will be discussed if they are notable.

(15)

5.2 Results hypothesis 1 Table 2.

Carhart four-factor model output for the sin-portfolio, neutralC-Portfolio, neutralP-Portfolio and neutralZ-Portfolio.

Sin Neutral C Neutral P Neutral Z

Intercept(α) .006 (2,549)** -.005(-2,869)*** .00001581 (0,008) .00003052 (0,019) RMRF(β) .743 (12,095)*** 1.148 (21,994)*** .834 (15,559) *** 1,026(23,025)*** SMB -.264 (-2,483)** -.304 -3,361)*** -.374 (-4,036)*** .41 (0,0525) HML -.008 (-0,077) .513 (5,992)*** .074 (0,842) -.103 (-1,404) MOM .039 (0,758) .088 (2,023)** .017 (0,376) -.17 (-0,461) Obs. 33 33 33 33 Adj. R2 .679 .846 .693 .853 Absolute values of t-statistic in parentheses. * Significant at 10% level, ** significant at 5% level, *** significant at 1% level. Table 1. shows that the α of the sin-portfolio makes an abnormal monthly return of 0.6% and is significant at 1.2%. The neutral-portfolio makes a negative abnormal return and is significant at 1%. To see if the difference between the two portfolios is significant an independent sample t-test is performed in the next table. Table 3. Group statistics of the sin portfolio and neutralC-Portfolio.

Portfolio Mean N Std. Deviation Std. Error Mean

Sin .01 133 .037377224 .00324101 Neutral C .0019 133 .053390633 .00462956 Absolute values of t-statistic in parentheses. * Significant at 10% level, ** significant at 5% level, *** significant at 1% level.

(16)

Table 4.

Independent samples t-test comparing the sin portfolio and neutral-portfolio.

Equal Variances F Sig. t Mean difference

Assumed 8.711 .003 1.424 .0080496 Not Assumed - - 1.424 .0080496 Absolute values of t-statistic in parentheses. * Significant at 10% level, ** significant at 5% level, *** significant at 1% level. Comparing the t-statistic of 1.424 with the t-table shows that the difference between the sin-portfolio and the neutral-portfolio is significant between 10% and 5%. This result is not statistically strong enough to make claims, but it does show an indication of a difference between the two portfolios. This could mean that the sin-portfolio makes an abnormal return over the neutral-portfolio. 5.3 Results hypothesis 2 A new regression is performed for three sub periods with the sin-portfolio and with the neutral-portfolio because that is the most significant of the three neutral-portfolios T1 (pre-crisis): January 2004 – January 2007 T2 (crisis): February 2007 – April 2011 T3 (After-crisis): May 2011- January 2015 Table 5. Carhart four-factor model output for the sin-portfolio divided in sub-periods. Sin T1 T2 T3 Intercept(α) .009* (1.898) .004 (.952) .005 (1.386) RMRF(β) .832** (3.303) .761*** (8.941) .701*** (6.282) SMB -.363 (-1.330) -.053 (-.293) -.465 (-2.545) HML -.114 (-.381) -.029 (-.184) -.246 (-1.279) MOM -.016 (-.071) .028 (.421) .360 (2.545) Adj. R2 0,216 0,684 0,432 Std.Est.Er 0,02324 0,02782 0,02207 Absolute values of t-statistic in parentheses. * Significant at 10% level, ** significant at 5% level, *** significant at 1% level

(17)

Table 6. Carhart four-factor model output for the neutral-portfolio divided in sub-periods. Neutral T1 T2 T3 Intercept(α) -.003 (-.699) -0.001 (-.339) -.007* (-2.837) RMRF(β) 1.103*** (5.428) 1.112*** (15.035) 1.195*** (15.796) SMB -.524 (-2.384) X.512** (-3.226) .048 (.376) HML .173 (.717) .664*** (4.816) .556*** (4.100) MOM .239 (1.320) .134 (2.313) X.237 (-2.475) Adj. R2 0,477 0,883 0,911 Std.Est.Er 0,01876 0,02418 0,01451 Absolute values of t-statistic in parentheses. * Significant at 10% level, ** significant at 5% level, *** significant at 1% level. Comparing the results of the sub-periods of the sin and neutral-portfolio the following information is relevant. Starting with the sin-portfolio it shows that the only significant α at only 10% and with a positive monthly return of 0,009. Which is a high return, but not statistically strong as the significance level is low and so is the R-squared. Even though not significant, the α of T2 and T3 are not in line with the hypothesis because the expectation was that sin-stocks would perform relatively better during crisis. Could be investigated further. In T1 the sin-portfolio (0.009*) performs better than the neutral portfolio (-0.003), which doesn't say much. During the crisis in T2 the sin-portfolio (0.004) drops by more than half whilst the neutral-portfolio (-.001) performs better than the previous period. Then after the crisis in T3 the sin-portfolio (0.005) starts to recover slightly, whilst the neutral-portfolio (-0.007*) decreases. This suggests that the sin-portfolio recovers faster than the neutral-portfolio maybe indicating a time lag for the neutral-portfolio. The momentum-factors for the sin-portfolio are smaller and closer to zero than the factors of the neutral-portfolio. This could explain the lag as the neutral portfolio has more momentum whereby the neutral-portfolio takes longer to recover from the negative effects of the crisis. Unfortunately the α’s are not significant enough to be certain, but this may be something to investigate further.

(18)

5.4 Results hypothesis 3 The next regression separates the sin-portfolio into the three industries in the portfolio for the complete period. The first being alcohol the second tobacco and the third gaming/gambling. Table 7. Carhart four-factor model output for the industry specific portfolios.

Sector Alcohol Tobacco Gaming

Intercept(α) .001 (.266) .012*** (3.350) .000 (-.062) RMRF(β) .753*** (6.918) .546*** (5.190) 1.468*** (8.193) SMB 0.180 (1.582) -.278 (-1.525) 1.122*** (3.617) HML 0.571*** (8.780) .013 (.074) .252 (.855) MOM -.211*** (-3.773) .010 (.113) -.898 (-6.032) Adj. R2 0,557 0,177 0,653 Std.Est.Er 0,02802 0,04221 0,07192 Absolute values of t-statistic in parentheses. * Significant at 10% level, ** significant at 5% level, *** significant at 1% level. It stands out that the tobacco-industry (.012***) has a very large positive abnormal return, whilst the alcohol and gaming industry are not significantly different from zero. This suggests that the positive return of the sin-portfolio is attributable to the tobacco-stocks for a large portion. This is an interesting result that will be looked into further below. The β for tobacco is also the lowest for tobacco making it the least volatile of the three industries, and they are all significant at 1%. The MOM-coefficient of the tobacco industry is also closest to zero making it less prone to past results. The table suggests that tobacco is the superior investment option of the three. 5.5 Results hypothesis 4 Given the previous results the tobacco-portfolio is tested for the sub-periods to see how it performs under different economic cycles.

(19)

Table 8. Carhart four-factor model output for the tobacco-portfolios. Tobacco T1 T2 T3 Intercept(α) .002 (.250) .009 (1.360) .009 (1.477) RMRF(β) 1.200** (2.588) .516***(3.813) .701*** (3.856) SMB - .243 (-.491) .181 (.622) -1.088*** (-3.576) HML 1154** (2.131) -.209 (-.829) -.342 (-1.050) MOM -.376 (-.943) -.052 (-.491) .381 (1.657) Adj. R2 .128 .279 .255 Std.Est.Er .04279 .04428 .03488 Absolute values of t-statistic in parentheses. * Significant at 10% level, ** significant at 5% level, *** significant at 1% level. The results are for the intercepts are not significant. The numbers indicate that the tobacco industry makes a good and positive abnormal return during the crisis (T2) and also afterwards, it is strange however that it is lower before the crisis (T1). This could have a connection to the low momentum factor during the crisis; the results are not significant unfortunately. This could be tested in a later research. The β’s are the only coefficients that are significant and show a drop during the crisis (T2) meaning that the tobacco portfolio becomes less volatile during the financial crisis. This does not seem likely. The sample is too small to make assumptions that are likely.

6. Discussion and Concluding Remarks

6.1 Summary Overall it shows that the α-values of the sin-portfolio and sin-industry-portfolios are higher than the α-values of the neutral-portfolios, however they are not always significant. When the sample of companies in the portfolios drops the significance drops fast, so does the R-squared. The results are in line with the related literature like Hong and Kascperczyk (2009) and Fabozzi (2008). The β-coefficient is almost always significant at the 1% level, even when the sample becomes very small. These β’s show

(20)

that the sin- portfolio, alcohol and tobacco-portfolio are less volatile than the neutral-portfolio over the entire period and before, during and after the crisis. Also Salaber’s (2009) findings that sin-stocks make an abnormal return and can be used to diversify risk during times of recession seems to be confirmed. In the second set of regressions for the sub-periods a time lag seems to occur. During the crisis (T2) it shows that the sin-portfolio’s α isn’t affected as badly as the neutral-portfolio’s α. And in the period after the crisis (T3) the sin-portfolio’s α starts to increase, where the neutral-portfolio’s α decreases even further. This time lag is also reflected in the momentum-coefficients, which are not always significant. The time-lag interpretation is not yet statistically sound, but should be investigated further. 6.2 Implications The results seem to indicate that a sin-portfolio does make an abnormal return and the tobacco-portfolio might play a big role in this abnormal return. This research is not extensive enough to claim that socially responsible investing comes at a high cost but it does point in that direction. If the difference in return between sin and neutral-portfolios or even individual stocks can be proven a divide between ‘good’ and ‘bad’ investors might develop. Investors that are ‘good’ take the cost of socially responsible investing and ‘bad’ investors just go for the highest returns not taking ethics into account. 6.3 Limitations The data only used companies that are active during the entire period. The total list of identified sin-companies was a lot bigger. The data of most companies begins of stops somewhere in the period. So one could argue that there is a selection bias because only the stronger companies are in the portfolios. On the other hand, the same is done with the neutral portfolio. The data is only done for four US-indexes. Views on smoking, drinking and gambling could be very different in other parts of the world. 6.4 Suggestions A larger sample of sin-companies could strengthen the statistic power of the research, for instance data from more indexes and stock-markets could be used. Data becomes

(21)

more difficult to gather and preferences and social-norms per country influence the results. More control variables could be added, for the literature indicates that leveraging could play a role. And also include the companies whose data stops during the period. It might also be interesting to investigate if the tobacco-industry really has a β and momentum-coefficient that are close to zero taking a larger sample. Salaber (2009) also states that a comparable-industry (non-sin) could diversify risk and make an abnormal return exists. This is something that could also be investigated further.

(22)

7. References

Berman, D. (2002). Why sin is good; Tobacco, alcohol and gaming stocks can add sizzle to your portfolio? Money Sense, November. Carvell, A. & Strebel, P. (1987). Is There a Neglected Firm Effect? Jounal of Business & Accounting, 14(2), 229-304. Carhart, M. (1997). On persistence in Mutual Fund Performance. Journal of Finance, 52(1), 57-82. Damadaran (1985). Scarce information may cause not only higher variance, but also less positive skewness and greater kurtosis in the return distribution. Durand, R. & Koh, S. 2013. Saints versus Sinners. Does morality matter? Journal of International Financial Markets, Institutions & Money, 24, 166-183. Fabozzi, F. & Oliphant, B. (2008). Sin Stock Returns. Journal of Portfolio Management, Fall, 82-94. Fama, E. & French, K. (1992). Common risk factors in the returns on stock and bonds. Journal of Financial Economics, 33, 3-56. Geczy, C., Stambaugh, B. & Levin, D. (2003). Investing in socially responsive mutual funds. Unpublished working paper, University of Pennsylvania. Whartons. Frederick, W. (1995). Values, Nature and Culture in the American Corporation. Oxford University Press, New York. Ghoul, S., Guedhami, O., Kwok, C. & Mishra, D. (2011). Does Corporate responsibility affect the cost of capital? Journal of Banking & Finance, 35, 2388-2406.

(23)

Global Sustainable Investment Review 2014 - Social Investment Forum Hughes, P. & Thakor, A. (1992). Litigation Risk, Intermediation, and the Underpricing of Initial Public Offerings. The Review of Financial Studies, 5(4), 709-742. Hong, H. & Kacperczyk, M. (2009). The price of sin: The effect of social norms on markets. Journal of Financial Economics, 93, 15-36. Jegadeesh, N. & Titman, S. (1993). Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency. Journal of Finance, 48(1), 65-91. Kempf, A. & Osthoff, P., (2007). The Effect of Socially Responsible Investing on Portfolio Performance. European Financial Management, 13(5), 908-922. Kim, I. & Venkatachalam, M. 2011. Are sin stocks paying the price for accounting sins? Journal of Accounting, Auditing & Finance, 26(2), 415-442 Kübler, D. (2001). On the regulation of social norms. Journal of Law Economics & Organization, 17(2), 449-476. Prakash Sethi, S. (2005). Investing in Socially Responsible Companies is a Must for Public Pension Funds. Journal of Business Ethics, 56, 99-129. Robinson, M. Kleffner, A. & Bertels, S. (2008). The Value of a Reputation for Corporate Social Responsibility: Empirical Evidence. Working Paper, University of Michigan. Salaber, J. M. (2009). Sin stock returns over the business cycle. Available at SSRN 1443188.