The low volatility effect in the technology sector

1

The low volatility effect in the technology sector

Bachelor thesis by Boas Plinck

Name: Boas Plinck

Student number: 10814612

Field of study: Economics and Business

Track: Finance and Organisation

Supervisor: Liang Zou

Year: 2017/2018

University of Amsterdam

Abstract: Previous studies found that low volatility stocks were associated with a better performance. This study attempts to find evidence for the low volatility effect with a sector based approach by investigating the technology sector. Two years of daily stock returns from technology sector are used to find a negative effect of volatility on stock performance. No significant evidence is found for a negative relation of stock return volatility, market risk and idiosyncratic volatility on stock performance.

2 Statement of Originality

This document is written by Student Boas Plinck who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of

3 Introduction

Over the years anomalies have been found that challenge the theory of efficient markets. One of those is the low-volatility anomaly or low-volatility effect. The theories surrounding the risk and return trade-off all assume market efficiency. In the past years anomalies are found challenging the market efficiency. One of those anomalies is the Low-volatility effect or low-volatility anomaly. Studies found low low-volatility stocks performing better.

Clark de Silva and Thorley (2006) performed research on the characteristics of minimum variance portfolio’s. They constructed a minimum variance portfolio constructed from the 1000 largest US stocks. The construction of the portfolio reduced volatility significantly but it did not decrease in returns. With the lower volatility the minimum variance portfolio had returns on the same level or higher than the market portfolio. The results of Clarke de Silva and Thorley were opposing the expectation from asset pricing theories.

Blitz and van Vliet (2007) Performed further research on the anomaly that low volatility stocks perform better. Blitz and van Vliet constructed portfolios of stocks based on the volatility of their stock returns. The portfolio’s constructed were 10 portfolio’s all with stocks equally weighted and selected based on their stock return volatility. The performance of the different portfolio’s was measured with Alpha and the Sharpe ratio, where both an higher alpha and a higher Sharpe ratio indicate better performance of a portfolio. Blitz and van Vliet found that both the alpha and Sharpe ratio of the portfolio’s increased with the decline of their volatility. Note that no optimal weights were assigned to the stocks within the portfolio’s, stocks were only selected based on volatility. The relation between risk and return Blitz and van Vliet found were the opposite of the theoretical relation provided by CAPM.

Baker and Haugen (2012) found that the low volatility effect is occurring worldwide. Baker and Haugen examined the relation between risk and return within markets of 21 developed countries and 12 emerging markets. They found that there exists a worldwide negative relation between risk and return. In addition Baker and Haugen found that low volatility portfolio’s globally were associated with higher Sharpe ratio’s.

. This paper investigates the technology sector for evidence on the low volatility anomaly. Individual stocks of technology companies are used to find a possible negative relation of volatility, market risk and idiosyncratic volatility on the performance of stocks selected from the Nasdaq technology industry list. This study tries to find that relation using linear

regression models. First earlier studies are reviewed to provide a theoretical framework about the pricing of risk, possible reasons for the low-volatility anomaly to occur and possible drivers of stock returns and performance. Secondly the research method for this paper is discussed. Regression models are used to test for evidence on the low-volatility anomaly.

4 The model test the relation of volatility on stock performance using the standard deviation of stock returns as an explaining variable. The stock performance is measured by the alpha and the Sharpe ratio of the stocks. In addition two more regression models are formulated tot test the effects of market risk and idiosyncratic risk. Model 2 uses the CAPM beta as an

explaining variable to measure market risk. Model 3 uses the idiosyncratic risk as an explaining variable. Explanation about the used variables is added to the methodology paragraph as well as explanation on the used data. The risk and performance measures are calculated from the daily stock returns of the technology firms along with a market proxy and a risk-free rate. The results of this study find no evidence for a low volatility anomaly within the technology sector. The results of the model will be presented in the following paragraph. The results of this study find no evidence for a low volatility anomaly within the technology sector as the model does not provide significant results. The model testing a negative relation of market risk on performance does not provide significant results as well. Based on the insignificant results for the second model no evidence is found for an influence of market risk on performance. Finally the third model did not provide evidence for a negative relation of idiosyncratic risk on performance as the results were not significant as well. The

implication of the results will be discussed in the conclusion and in addition a discussion paragraph is added where the poor performance of the used models is discussed. Literature Review

According to Berk and DeMarzo (2014) Investors are compensated for risk. They state that stock returns mostly outperform bond returns, while stock returns bear more risk. The total risk of an asset according to Berk and DeMarzo is its volatility which is measured by the standard deviation of the assets return. According to berk and DeMarzo the Risk of an asset is split into two parts: the market or systemic risk and the idiosyncratic or firm-specific risk. Berk and DeMarzo state that investors are only compensated for market risk an asset bears. The idiosyncratic risk can be diversified, if a portfolio is constructed of large number of stocks the firm the idiosyncratic risk will eventually cancel out. Berk and DeMarzo argue that

investors do not obtain a risk premium for idiosyncratic risk because investors will arbitrage away that risk premium. Market risk cannot be diversified away because market-wide events will affect every asset in the market. Because of this Berk and DeMarzo argue that investors are interested in an assets sensitivity towards market volatility. The sensitivity of an asset towards market risk is denoted by Beta from the Capital Asset Pricing model (CAPM). Berk and DeMarzo Describe Beta as the expected change in the return of an asset given a change in the return of the market. The capital asset pricing model (CAPM) uses beta to estimate Expected returns. An asset should provide an investor with more risk premium if Beta increases.

5 Dutt and Humphery-Jenner (2012) find that the returns of low volatility stocks are higher. They found evidence that that firms with low volatility stock tend to have an high operating performance. Dutt and Humphery-Jenner state that Low volatility companies have better excess to capital as they are seen as to be less risky to provide capital to. Because of the better excess to capital it is easier for low volatility companies to invest and generate a better operating performance. The higher operating performance will eventually lead to a higher stock return for this companies according to Dutt and Humphery-Jenner.

According to efficient market theory the low volatility anomaly should be arbitraged away by investors taking advantage of the opportunity. Taking advantage of the opportunity involves shorting the underperforming high volatility stocks. Baker, Bradley an Wurgler (2011) argue that high volatility stocks tend to be from small companies and therefore are less available to borrow for a short position. On the other hand investors should assign a larger weight to low volatility stocks in their long portfolio. Baker, Bradley and Wurgler (2011) argue that fund managers prefer high volatility stocks because they will increase the expected return of the fund since they have to beat a certain benchmark. This in combination with the low

availability for shorting high volatility stocks causes the low volatility anomaly to persist. Testing the CAPM by Black, Jensen and Scholes (1972) showed that low beta assets earn higher Alpha’s than high beta assets.

Research of Fama and French (1992) showed that the relation between beta and stock returns is weak. Fama and French performed research on NYSE stocks in the period of 1941-1990. They found that when portfolios were based on Beta there is no effect of beta on average stock returns. If portfolios were formed based on size beta has a positive effect on average returns and size has a negative effect on average returns. According to Fama and French the effect of Beta on stock returns is related to size because there is again no relation between beta and average returns when the relation between beta and size is filtered out. According to CAPM theory the expected return of a stock is based on its correlation with market risk. In the CAPM formula the E(r) only depends on Beta*market risk premium and the error term. The idea behind this is that firm specific risk is hedged away by diversification and therefore not priced in by investors. Some studies found that a (negative) premium for firm specific risk does exist. Khovansky and Zhylyevskyy (2013) performed research on the idiosyncratic risk premium. They used the stock returns of individual stocks to investigate the relation between their returns and idiosyncratic volatility. They found a significant positive premium for idiosyncratic volatility on daily returns a non-significant negative premium for weekly returns and significant negative premiums for monthly, quarterly and yearly returns. This results could be an explanation for the low volatility anomaly. If idiosyncratic volatility increases with volatility this negative premium would decrease returns. And if this negative premium is not included in asset pricing models this would cause the asset to perform worse

6 than expected and decrease abnormal returns.

Ang et al (2006) regressed stock returns on VIX. VIX is an index that is constructed from put and call options on the S&P100. It is constructed to represent the volatility of a synthetic at the money option for the S&P100. Ang et al found a risk premium of -1% for exposure to market volatility risk. In addition Ang et al found that stocks with high idiosyncratic risk earn abnormally low returns.

Besides the earlier describe effect of size found by Fama and French (1992) studies found other effects that might influence stock returns or stock performance. Fama and French (1992) additionally describe that stocks with a higher book-to-market ratio have higher returns.

Bhandari (1988) found a positive premium for debt to equity ratio on stock returns after controlling for beta. Bhandari argues that an increase in debt/equity ratio will increase the risk of stocks. Since Bhandari controlled for beta the premium for the debt/equity ratio is not captured by beta.

A study of Basu (1983) showed that stocks with higher earnings/price ratios are associated with higher stock returns.

Methodology

Models and Hypothesis

To test for the low volatility effect in the technology three regression models are formulated. The models are regressing Performance measures on Total volatility, beta and Idiosyncratic volatility. The measures for Performance are the CAPM alpha and the Sharpe Ratio. All models are controlling for size (market capitalization) and book-to-market ratio as suggested by Fma and French (1992), leverage ratio as found by Bhandari (1988) and PE ratio

following Basu (1983).

Earlier research for example described a negative relation between volatility and stock performance. Clark de Silva and Thorley (2006) found that the formation of a minimum variance portfolio reduced the volatility below the level of the market volatility while having the same or larger returns. Blitz and van Vliet (2007) described that both alpha and Sharpe ratios decrease when volatility increased along portfolio’s sorted on volatility. This implies a negative relation of volatility on stock performance. To test whether total volatility has a negative effect on stock of technology stocks a regression is formulated. For model 1 the stock performance is the dependent variable being explained by the total volatility and control variables. The CAPM alpha is a measure of performance and known as the abnormal return of an asset. In addition the Sharpe ratio is used as a performance measure in model 1. For the total volatility the standard deviation of stock returns will be used.

7

Alpha = β0 + β1*volatility + β2 *size + β3*Book/market + β4*leverage ratio + β5*PE ratio + ε

Sharpe ratio = β0 + β1*volatility + β2 *size + β3*Book/market + β4*leverage ratio + β5*PE ratio

+ ε

The model is constructed for this paper to find evidence for the negative relation of volatility on performance as implied by results from Clark de Silva and Thorley (2006) and Blitz and van Vliet (2007).

The null hypothesis of B1 being zero implies no relation between the performance of a stock and its volatility. The alternative hypothesis of B1 being lower than zero implies a negative relation between the performance of a stock and its volatility.

H0: β1=0, H1: β1<0

Testing the effect of the CAPM Beta on performance.

Model 2 tests the relation between the CAPM beta of technology stocks and their performance. Theoretically there should be a positive relation between beta and stock returns. The beta denotes an assets sensitivity towards market volatility and since this is not diversifiable investors earn a risk premium for bearing that risk. However some studies show that the relation is less strong than in theory according to Black, Jensen and Scholes (1972). This implies that investors earn less premium for the market risk they take on. This means that an increase in beta will cause a decrease in stock performance. To test whether a negative relation exists of sensitivity towards market risk on the performance of technology stocks model 2 is formulated. Model 2 is a regression with stock performance as dependent variable. For model 2 again the CAPM alpha and the Sharpe ratio are used. In model 2 the dependent variable is explained by the CAPM beta along with the same control variables used in model 1.

Alpha = β0 + β1*CAPM beta + β2 *size + β3*Book/market + β4*leverage ratio + β5*PE ratio + ε

Sharpe ratio = β0 + β1*CAPM beta + β2 *size + β3*Book/market + β4*leverage ratio + β5*PE

ratio + ε

The model is constructed for this paper to find evidence for the negative relation of volatility on performance as implied by results from Black, Jensen and Scholes (1972) and Fama and French (2007).

For model 2 the null hypothesis is B1 being equal to zero implying no effect of the CAPM beta on stock performance. The alternative hypothesis is B1 being less than zero implying a negative relation of the CAPM beta on stock performance.

8

H0: β1=0, H1: β1<0

Testing the effect of Idiosyncratic volatility on performance.

Model 3 tests the relation of idiosyncratic volatility on stock performance. In theory investors do not obtain a risk premium for bearing idiosyncratic risk. . Khovansky and Zhylyevskyy (2013) found a negative risk premium for idiosyncratic risk. Evidence for the negative risk premium was found by Ang et (2006) al as well. To test for a possible negative relation between idiosyncratic volatility and stock performance model 3 is formulated. Model 3 uses performance measure alpha as dependent variable being explained by idiosyncratic volatility. The model controls for the same variables as model 1. The Sharpe ratio is not used as a performance measure since idiosyncratic volatility will theoretically already decrease the Sharpe ratio. In addition model 3 will use stock returns as a dependent variable as well. If a significant negative relation is found between idiosyncratic volatility and returns this will provide evidence for a negative risk premium for bearing idiosyncratic volatility.

Alpha = β0 + β1*Idiosyncratic volatility + β2 *size + β3*Book/market + β4*leverage ratio +

β5*PE ratio + ε

Sharpe Ratio = β0 + β1*Idiosyncratic volatility + β2 *size + β3*Book/market + β4*leverage ratio

+ β5*PE ratio + ε

The model is constructed for this paper to find evidence for the negative relation of volatility on performance as implied by results from Ang et al (2006) and Khovansky and Zhylyevskyy (2013).

The null hypothesis for model 3 is B1 being equal to zero implying no effect of Idiosyncratic volatility on stock performance and stock return. The alternative hypothesis is B1 being less than zero implying a negative relation of idiosyncratic volatility on stock performance and stock returns.

H0: β1=0, H1: β1<0

Explanation of the used variables

The total volatility is the total risk witch an asset bears. The total volatility will be measured by the standard deviation of stock returns. The risk of an asset is split into Market risk and Idiosyncratic risk

Beta is the amount of Volatility a stock has in relation to the market. Beta denotes the assets sensibility to market volatility. In theory investors are compensated only for taking on market risk as Firm specific risk can be cancelled out by diversification. In the CAPM model the beta is used to determine the Expected Return of an asset given the market return. The formula for Beta used:

9

𝛽

=

𝑐𝑜𝑣(𝑟

, 𝑟

)

𝜎

Cov (ri,rm) denotes the covariance between the market return and the stock return. Var(rm)

denotes the Variation of the market return. In addition to the market risk the total volatility exists of idiosyncratic volatility.

Idiosyncratic volatility is the firm-specific risk that can be diversified and therefore

theoretically no risk premium is obtained for bearing idiosyncratic risk. For the idiosyncratic risk the idiosyncratic variance is used as an explaining variable. The idiosyncratic variance is obtained by:

𝜎

= 𝛽

∗ 𝜎

+ 𝜎

Gives:

𝜎

= 𝜎

− 𝛽

∗ 𝜎

𝜎

𝜎

𝑖

2 _{denotes the total variance of returns, β denotes}

the sensitivity of the asset towards market volatility and

𝜎

𝛽

∗ 𝜎

alpha is used. Alpha stands for the abnormal return of an asset. It is the return in excess of the return expected by the asset pricing model. The return on itself is no good measure for performance because the risk of assets varies. The alpha gives an indication of performance given the amount of risk. For this paper the CAPM alpha is used. Alpha will be all return above or below the return expected by the CAPM. The formula for alpha used:

𝐴𝑙𝑝ℎ𝑎 = 𝑟

− (𝑟

+ 𝛽 ∗ (𝑟

− 𝑟

))

Another measure for performance is the Sharpe ratio. The Sharpe ratio divides the excess return of a portfolio by the variance of the portfolio. This way the Sharpe ratio measures the amount of return for a unit of risk. Given the trade-off between risk and return this is a good measure for performance. The Sharpe ratio is mostly used to measure the performance of portfolios, but in this study it is used to measure the performance of single stocks. If a stock awards a larger return for a unit of risk the stock is more desirable for investors. The formula for the Sharpe ratio used:

𝑆ℎ𝑎𝑟𝑝𝑒 𝑅𝑎𝑡𝑖𝑜 =

𝐸(𝑟

) − 𝑟

𝜎

10

Exhibit 1: Descriptive statistics for all used variables.

N Minimum Maximum Mean Std. Deviation

Alpha 53 -.00109 .00315 .00008 .00071 Volatility 53 .01116 .03431 .01783 .00492 Beta 53 .23628 .71560 .49025 .09126 Idiosyncratic Volatility 53 .000000005 .00116 .00029 .00021 Book to Market 53 .01941 1.77418 .33384 .30363 Market Cap (millions) 53 5303395 587219541 69271046 125889101 Sharpe Ratio 53 -.03663 .13776 .02389 .03297 Return 53 -.00092 .16512 .00353 .02263 Leverage Ratio 53 0 150.53397 34.20378 24.42149 PE Ratio 53 0 238.50588 39.01089 45.47689

The Descriptive statistics of the used variables are shown in Exhibit 1. Volatility denotes the total volatility of stocks measured by the standard deviation of stock returns. The

idiosyncratic volatility is measured by the idiosyncratic variance of stock returns. The use of respectively standard deviation and variance variables explains the difference in magnitude between volatility and idiosyncratic volatility.

Data

To be able to perform the research data was collected with Data Stream. Via data stream daily stock prices were collected for 53 stocks from the technology sector for 2015 and 2016, 520 daily observations were used. The stocks of the 53 largest companies of Nasdaq’s technology industry list were selected. In addition the Nasdaq was used as a market proxy where the daily return on the Nasdaq served as rm variable. From the stock prices returns

were calculated to be able to compute the necessary variables. The risk-free rate was obtained from the website of Fama and French, used was the daily rf variable from the Fama

and French model. The daily Fama and French rf was equal to zero for 2015 and 2016 and

therefore not included in Exhibit 1.

For all stocks the market capitalization, book value of equity, leverage ratio and PE ratio were collected from DataStream to be used for the control variables.

11 Results

Results for Model 1 using Alpha.

Exhibit 2: Regression output for model 1 with dependent variable Alpha. Shown are the coefficients with their standard error, t-value and the p-value of the coefficient. Model statistics are shown below the regression output.

Variable Coefficient Std. Error t-value Sig.

Volatility -.006 .021 -.288 .775

Book to Market .000 .000 .614 .542

Market Cap 3.520E-13 .000 .429 .670

Leverage Ratio -5.525E-7 .000 -.132 .895

PE Ratio 3.858E-6 .000 1.711 .094 Constant -3.813E-5 .000 -.085 .993 F- value .637 Prob. F .672 R2 _.063 Adjusted R2 _-.036

Alpha was used and the independent variable was volatility along with control variables. The regression results are shown in Exhibit 2. The regression found a non-significant coefficient of -0.006 for volatility. With a p-value of .775 this coefficient is highly insignificant. Based on this result there is no evidence that the coefficient for volatility does significantly defer from zero. This means that vidence of a relation of stock return volatility on performance measure alpha. Of the control variables only the coefficient for the PE ratio was significant to the 10% level. With a F-value of .637and P-value of .672 the total model is highly unusable to predict the relation between volatility and alpha.

12

Results for Model 1 using Sharpe ratio.

Exhibit 3: Regression output for model 1 with dependent variable Sharpe ratio. Shown are the coefficients with their standard error, t-value and the P-value of the coefficient. Model statistics are shown below the regression output.

Variable Coefficient Std. Error t-value Sig.

Volatility -.734 .997 -.751 .456

Book to Market .007 .016 .447 .657

Market Cap 2.599E-11 .000 685 .496

Leverage Ratio -5.285E-5 .000 -.273 .786

PE Ratio .000 .000 1.678 .100 Constant .028 .021 1.336 .188 F- value .756 Prob. F .586 R2 _0.74 Adjusted R2 _-0.24

Model 1 tested the relation of total volatility on performance. As dependent variable besides alpha the Sharpe ratio was used and the independent variable was volatility along with control variables. The regression results are shown in Exhibit 3. The regression found a non-significant coefficient of -0.006 for volatility. With a p-value of .465 this coefficient is less but still highly insignificant. Based on this result there is no evidence that the coefficient for volatility does significantly defer from zero. This means that evidence of a relation of stock return volatility on the Sharpe ratio of stocks. Of the control variables only the coefficient for the PE ratio was significant to the 10% level. With a F-value of .756 and P-value of .586 the total model is highly unusable to predict the relation between volatility and the Sharpe Ratio.

13

Results for Model 2 using Alpha.

Exhibit 4: Regression output for model 2 with dependent variable Alpha. Shown are the coefficients with their standard error, t-value and the p-value of the coefficient. Model statistics are shown below the regression output.

Variable Coefficient Std. Error t-value Sig.

Beta .001 .001 .820 .416

Book to Market .000 .000 .612 .543

Market Cap 3.359E-13 .000 .416 .679

Leverage Ratio -6.836E-7 .000 -.164 .870

PE Ratio 3.456E-6 .000 1.532 .135 Constant -.001 .001 -.978 .333 F- value .763 Prob. F .581 R2 _.075 Adjusted R2 _-.023

Model 2 tested the effect of sensitivity of a stock towards market volatility on performance of the stock. As dependent variable the alpha was used and the independent variable was beta along with control variables. The regression results are shown in Exhibit 4. The variable measuring sensitivity towards market risk is describes as beta. The regression found a non-significant coefficient of .001 for Beta. With a P-value of .416 this coefficient is highly insignificant. Based on this result there is no evidence that the coefficient for volatility does significantly defer from zero. This means that evidence of a relation of sensitivity towards market volatility on the alpha of stocks is not found by performing this test. None of the control variables had a significant effect in the regression. With a F-value of .763 and P-value of .581 the total model is highly unusable to predict the relation between volatility and alpha

14

Results for Model 2 using Sharpe ratio.

Exhibit 5: Regression output for model 2 with dependent variable Sharpe ratio. Shown are the coefficients with their standard error, t-value and the P-value of the coefficient. Model statistics are shown below the regression output.

Variable Coefficient Std. Error t-value Sig.

Beta .063 .051 1.216 .230

Book to Market .006 .016 .377 .708

Market Cap 2.688E-11 .000 .722 .474

Leverage Ratio -5.849E-5 .000 -.305 .761

PE Ratio .000 .000 1.379 .174 Constant -0.14 .027 -.528 .600 F- value .951 Prob. F .457 R2 _0.92 Adjusted R2 _-0.05

As dependent variable besides alpha the Sharpe ratio was used and the independent

variable was beta along with control variables. The regression results are shown in Exhibit 5. The regression found a non-significant coefficient of .063 for Beta. With a P-value of .230 this improved in significance but is still insignificant. Based on this result there is no evidence that the coefficient for volatility does significantly defer from zero. This means that evidence of a relation of sensitivity towards market volatility on the Sharpe of stocks is not found by performing this test. None of the control variables had a significant effect in the regression. With a F-value of .951 and P-value of .457 the total model is highly unusable to predict the relation between volatility and the Sharpe ratio.

15

Results for Model 3 using Alpha.

Exhibit 6: Regression output for model 3 with dependent variable Alpha. Shown are the coefficients with their standard error, t-value and the p-value of the coefficient. Model statistics are shown below the regression output.

Variable Coefficient Std. Error t-value Sig.

Idiosyncratic Volatility

-.036 .502 .072 .943

Book to Market .000 .000 .576 .568

Market Cap 3.789E-13 .000 .459 .648

Leverage Ratio -5.491E-7 .000 -.129 .898

PE Ratio 3.780E-6 .000 1.674 .101 Constant .000 .000 -.407 .686 F- value .621 Prob. F .685 R2 _.062 Adjusted R2 _-.038

Model 3 tested the effect of idiosyncratic volatility on performance of the stock . As dependent variable the alpha was used and the independent variable was Idiosyncratic volatility along with control variables. The regression results are shown in Exhibit 6. The regression found a non-significant coefficient of -.036 for Beta. With a P-value of .943 this coefficient is highly insignificant. Based on this result there is no evidence that the coefficient for idiosyncratic volatility does significantly defer from zero. This means that evidence of a relation of idiosyncratic volatility on the alpha of stocks is not found by performing this test. None of the control variables had a significant effect in the regression. . With a F-value of .621 and P-value of .685 the total model is highly unusable to predict the relation between volatility and Alpha.

16

Results for Model 3 using stock returns.

Exhibit 7: Regression output for model 3 with dependent variable stock returns. Shown are the coefficients with their standard error, t-value and the p-value of the coefficient. Model statistics are shown below the regression output.

Variable Coefficient Std. Error t-value Sig.

Idiosyncratic Volatility

-9.336 16.206 -.576 .567

Book to Market .004 .011 -.395 .694

Market Cap -7.764E-12 .000 -.292 .772

Leverage Ratio -9.558E-5 .000 -.292 .490

PE Ratio 6.607E-5 .000 -.695 .369 Constant .009 .010 .907 .389 F- value .406 Prob. F .842 R2 _.041 Adjusted R2 _-.061

As dependent variable besides alpha the stock returns were used and the independent variable was idiosyncratic volatility along with control variables. The regression results are shown in Exhibit 7. The regression found a non-significant coefficient of -9.336 for

Idiosyncratic volatility. With a P-value of .567 this coefficient improved in significance but is still highly insignificant. Based on this result there is no evidence that the coefficient for idiosyncratic volatility does significantly defer from zero. This means that evidence of a relation of idiosyncratic volatility on results of stocks is not found by performing this test. None of the control variables had a significant effect in the regression. With a F-value of .504 and P-value of .771 the total model is highly unusable to predict the relation between

volatility and the Sharpe ratio. Conclusion.

To answer whether volatility has a negative effect on stocks from the technology sector a number of regressions were performed. Based on the results of the regressions no conclusions about the relation between risk and performance can be made. All models proved to be insignificant and deliver insignificant coefficients.

17 Model one used total volatility or total risk to explain the variation in alpha and the Sharpe ratio. The alpha and Sharpe ratio were used as measures for performance. Based on the adjusted F-value and the P-value for the model can be concluded that the model is unusable to explain the variance in Alpha and Sharpe ratio. In addition the regression coefficients for volatility as well as all control variables were insignificant. This means no negative relation as described in earlier studies like Blitz and van Vliet (2007) and Clark de Silva and Thorley (2006) was found.

Model two used the sensitivity towards market volatility to explain the variation in alpha and the Sharpe ratio. The CAPM beta was used to measure sensitivity towards market risk. Based on the R2 _{and P-value of the modal can be concluded that the model is not usable to}

explain the variance in Alpha and Sharpe ratio. The coefficients for the CAPM beta proved to be insignificant, for none of the control variables significant coefficients were found. The results from model 2 provide no evidence that a negative relation of sensitivity towards market risk on performance exists for stocks in the technology sector. This results are against expectations based on studies from Fama and French (1992) and . Black, Jensen and Scholes (1972).

Model three used idiosyncratic volatility to explain the variation in alpha and stock returns. Based on the R2 _{and P-value of model three can be concluded that the model is not usable}

to explain the variation in Alpha and Stock returns. The coefficients for the CAPM beta

proved to be insignificant, for none of the control variables significant coefficients were found. The results from model three provide no evidence that a negative relation of idiosyncratic volatility on performance exists for stocks in the technology sector. In addition no significant relation was found between idiosyncratic volatility and stock returns. This implies that no evidence was found for a negative risk-premium for idiosyncratic volatility described by Khovansky and Zhylyevskyy (2013).

Given the insignificant results provided by the models no evidence was found for the existence of a low-volatility anomaly in the technology sector.

Discussion

Performing this study did not provide any significant results. None of the models were significant and usable to explain the dependent variables. None of the coefficients for the explaining variables were significant either. The fact that no significant results were found could be explained by a number of reasons.

The study used a sample of 53 companies from the technology sector. 53 is a small sample to perform research on, it could be that the sample size was too small to provide significant results. An indication for the sample being too small for the number of explaining variables is the fact that the adjusted R2 _{is negative for all models. In addition the small sample implies}

18 that the results cannot provide results that could be generalized to the entire technology sector.

The selection of the sample could also have implications for the results, the 53 largest technology companies were selected. Not selecting smaller tech firms might have resulted in less variation within the sample. The average beta within the sample was 0.49 so the market risk for the sample was at the lower bound and the average volatility along the sample was 1.8%. In comparison the lowest volatility portfolio from the study of Blitz and van Vliet (2007) has a beta of 0.69 and a volatility 2.9%. Note that the portfolios of Blitz and van Vliet have diversification benefits and the sample of this study has not since the research is on single stocks. It could mean that the sample of this study consists of companies with very similar and low risk profiles and therefore no relation of risk on performance could be found. The fact that in this paper the measures for risk and performance were computed based on daily variables could have contributed to the insignificance of the results. Khovansky and Zhylyevskyy (2013) found a significant positive premium for idiosyncratic volatility on daily returns a non-significant negative premium for weekly returns and significant negative premiums for monthly, quarterly and yearly returns. This shows that different results can be found for different time intervals and it could be possible that this paper did not provide significant results because only daily intervals were used.

19 Blitz, D. van Vliet, P. (2007). The Volatility Effect. Journal of Portfolio Management, 34(1), pp. 102-113

Dutt, T. Humphery-Jenner, M.(2012) Stock return volatility, operating performance and stock returns: International evidence on drivers of the ‘low volatility’ anomaly. Journal of Banking and Finance, pp. 100-1017

Baker, M. Bradley, B. Wurgler, J. (2011) Benchmarks as limits to arbitrage: understanding the low-volatility anomaly. Financial Analyst Journal, 67 (1), pp. 40-54.

Khovansky, S. Zhylyevskyy, O. (2013) Impact of idiosyncratic volatility on stock returns: A cross-sectional study. Journal of Banking and Finance, 37(8), pp. 2064-3075

Fama, E. French, K. (1992) The cross-section of expected stock returns. Journal of Finance, 47(39), pp. 427-465

Clarke, R. De Silva, H. Thorley, S. (2006) Minimum-variance Portfolio Composition. Journal of Portfolio management, 37(2), pp. 31-46

Ang, A. Hodrick, R. Xing, Y. Zhang, X. (2006) The cross-section of volatility and expected returns. The Journal of Finance 61(1). pp. 259-299.

Basu, S. (1983) The relationship between earnings yield, market value, and return for NYSE common stocks: Further evidence. Journal of Financial Economics 12. pp. 129-156.

Bhandari, L. (1988) Debt/Equity ratio and expected common stock returns: Empirical evidence, Journal of Finance 43. pp. 507-528.

Jensen M., Black, F. Scholes, M. (1972), The capital asset pricing model: some empirical tests. Studies in the Theory of Capital Markets edited by M. C. Jensen

Baker, N. Haugen , R.(2012), Low Risk Stocks Outperform within All Observable Markets of the World. Available at SSRN: https://ssrn.com/abstract=2055431