Stock price volatility and dividend yield: Evidence from Sweden

(1)

Stock price volatility

and dividend yield:

Evidence from

Sweden

BACHELOR THESIS WITHIN: Economics NUMBER OF CREDITS: 15 ECTS

PROGRAMME OF STUDY: International Economics AUTHOR: Olena Deboi, William Sörensen

(2)

i

Bachelor Thesis in Economics

Title: Stock price volatility and dividend yield: Evidence from Sweden. Authors: O. Deboi and W. Sörensen

Tutor: Andrea Schneider Date: 2020-12-07

Key terms: Stock Price Volatility, Dividend Yield, Stock market

Abstract

This research aims to examine if a negative relationship exists between the dividend yield and stock price volatility of firms listed on the Swedish Stock exchange market, which is of utter interest and intrinsic for investors and financial analyst in the process of valuing a security’s and a stock portfolio's risk and return. The data that was utilized for this study consists of 52 companies for the period of 2010 to 2019 which makes up for 520 observations. A pooled regression model and a multiple ordinary least squares model was applied to test the relationship. The results show a negative relationship between the dividend yield and stock price volatility. On the other hand, the results indicate that there is a significant positive relationship between earnings volatility and stock price volatility. However, there is a negative relationship for leverage, market value and asset growth with stock price volatility.

(3)

ii

1. Introduction

_____________________________________________________________________________________ “Volatility matters, because it defines the uncertainty of the price at which an asset will

be liquidated.”

Krass 1999, p. 154

______________________________________________________________________

It is cited from the book written by Berk and DeMarzo (2017) that “dividend yield is the percentage return an investor expects to earn from the dividend paid by the stock”. While stock price volatility explains how the stock market fluctuates (Green, 2020). Therefore, dividend yield and stock price volatility are two imperative aspects within the field of corporate finance because they are of interest for shareholders and investors. The dividend yield is one of the core elements that have an influence on total return to shareholders and volatility. Hence, investors are interested in less volatile stocks with a stable or increasing dividend yield as it decreases the risk (Baskin, 1989). Against this background, it does not come as a surprise that the relationship between dividend yield and stock price volatility has been extensively discussed in the scientific literature (see, e.g., Baskin, 1989; Hussainey et al., 2011; and Gordon, 1960). Additionally, the topic is of interest and important for managers of organizations since they have to be mindful of keeping the liquidity stable for daily corporate operations and maintaining financial health for future capital investments, in order to augment operational capacity, capture a large market share as well as generate additional revenue to the firm (Roy, 2015, Hakeem & Bambale, 2016).

In conjunction with the above mentioned, Baskin (1989) shed light upon in his article regarding the importance of dividend yield by mentioning the fact that investors have to establish regularity between stock price volatility and dividends in order to be able to prognosticate and minimize the risk, which is an indispensable factor for investors to reach an investment decision as they are risk averse by nature (Hussainey et al., 2011). By continuing writing on the relationship and endeavouring to discern as well as illuminate important theories on the topic, Baskin (1989) referred to the Bird-in-Hand

(6)

2

theory established and developed by Gordon (1963) and Lintner (1962) who stated that an increase in dividends cause a decrease in risk for shareholder or a decrease in the stock price volatility, which consequently affects the cost of capital and the price of the stock. On the contrary, Miller and Modigliani (1961) introduced their view on the topic by establishing a theory called Miller & Modigliani Theory that claimed that in a perfect capital market a change in dividend yield has no influence on stock price, thus these two factors are irrelevant.

The aim of the following study is to explore the relationship between dividend yield and stock price volatility in the Swedish Stock market for the period between 2010-2019. We will include control variables, i.e. market value, leverage, earnings volatility and asset growth that determine stock price volatility as well as theories that explain them. Albeit there have been many discussions about the relationship between dividend yield and stock price volatility, researchers still cannot reach the consensus on the topic (Baskin, 1989). Therefore, we believe that this research will contribute to the existing literature and will shed light upon the theories and views, which will be beneficial and useful for Swedish investors and managers of companies to have a better understanding of how they can improve their current dividend policy. We will do our research about Sweden because similar research questions are not explored on the Swedish Market. Thus, the Research Question - that is going to be explored throughout this thesis is:

Does a correlation exist between Dividend Yield and Stock Price Volatility?

The remainder of the thesis is organized as follows: section two illustrates literature review on the relation between dividend yield and stock price volatility. Section three presents the theoretical framework. In section four, data and control variables are presented. Section five discusses methods that were utilized to answer the research question. An empirical analysis of the results is also included in section 5. The last section concludes the whole research that was done on the topic.

(7)

3

2. Literature Review

_____________________________________________________________________________________

The following section presents previous research regarding the relationship between stock price volatility and dividend yield. We structured our literature as follows: the first paragraphs present the earliest studies that were conducted in developed countries, after that, we analysed the results that were obtained in developing countries. This part presents countries that discerned a positive relationship and afterwards we have countries that found a negative relationship. It was crucial to include cases in developing countries too, to see how the results differ and what are the reasons for having different findings.

______________________________________________________________________

One of the earliest researches was conducted by Baskin (1989) on the USA market and evidence of the negative relationship between dividend yield and stock price volatility was obtained by investigating the effect of dividend policy on price volatility from the period 1967-1986 of 2344 US stocks. Due to the inverse relationship that Baskin found in his research, it was concluded that dividend policy has an impact on stock price volatility, which proved that MM Theory was not applicable during that time period, a theory which will be discussed more in detail in section 3. Moreover, Baskin added more control variables i.e. debt, market value and earning volatility which made his results reliable with more explaining power. His findings demonstrated that there is a significantly negative relationship between market value and stock price volatility, while earnings volatility and debt have positive relationships with stock price volatility. In other words, when the market value of a firm is larger, the stock price volatility will be lower, but when the debt and earning volatility increases the stock price volatility will decrease. In light of the above, Baskin (1989) came to the conclusion that it is of the highest importance to include additional control variables as they explain the relationship between dividend yield and stock price volatility.

On the other hand, Allen and Rachim (1996) study contradicts Baskin’s study. In their research, 173 firms on the Australian stock market in the period 1972-1985 were scrutinized. The result of their study demonstrated that there is no relationship between

(8)

4

stock price volatility and dividend yield by utilizing cross-sectional multiple regression. Also, they concluded that there is a positive relationship between earning volatility and stock price volatility, however, there is a negative relationship between pay-out ratio and stock price volatility. Moreover, the authors emphasized that such variables i.e. size of the company, earnings volatility, payout ratio and the level of debt are the most crucial determinants of stock price volatility. This part of findings goes in line with the results of Baskin (1989).

Hussainey et al. (2011) research was based on Baskin’s (1989) study and was conducted on non-financial firms in the UK in the period of 1998-2007. The relationship between changes in the share price and dividend policy was investigated by utilizing multiple regression analysis. The result of their research showed that dividend yield, as well as dividend payout, has a negative relationship with stock price volatility which implies that an increase in dividend yield and dividend payout will lead to a decrease in stock price volatility. In addition, Hussainey et al. (2011) has included for the control variables: growth rate, market value, earnings volatility and debt level as they discerned that those variables provide an explanation of the relationship with stock price volatility which goes in line with existing literature i.e. Baskin (1989) and Allen and Rachim (1996). Also, Profilet and Bacon (2013) conducted a research on the US market and presented in their respective studies that such control variables as market value, growth, leverage and dividend yield have a significant negative relationship with stock price volatility. In this study ordinary least square (OLS) multiple regression on panel data was applied. In light of foregoing, it can be seen that existing literature confirms the existence of a negative relationship between dividend yield and stock price volatility in developed countries.

By scrutinizing developing countries, it was noticed that researchers presented different results in comparison with what was introduced before. In the case of Pakistan, Nazir et al. (2010) conducted a research on Karachi Stock Exchange by using data from 2003-2008 of 73 firms examining the relationship between dividend yield and stock price volatility by applying the fixed effect and random effect models. Their research showed a mixed result: a positive relationship between dividend yield and stock price volatility was found; while dividend payout has a negative relationship with stock price volatility. Taking into consideration the fact that Pakistan is a developing country, the results of the

(9)

5

study completely contradict with the findings of developed countries i.e. Hussainey et al. (2011) and Baskin (1989). Moreover, their findings indicated that there is a significant positive relationship between control variables, such as: earning volatility, earnings per share, asset growth and stock price volatility. However, two remaining variables – market value and long-term debt have a negative relationship with the stock price volatility.

Similar results were obtained by Nguyen et al. (2020) in Vietnam on a sample of 260 listed firms on the Hochiminh Stock Exchange (HOSE) in the period 2009-2018. The author examines the relationship between share price volatility and dividend yield as well as the dividend payout ratio. The result of their study showed that there is a positive relationship between stock price volatility and dividend yield, but a there is negative relationship between stock price volatility and dividend pay-out ratio. At the same time, they presented that control variables i.e. growth, earning volatility and leverage has a significant positive relationship with stock price volatility, while market value has a negative relationship with stock price volatility, hence Nguyen results matched with Baskin’s (1989).

On the contrary, Hashemijoo et al. (2012) investigated the relationship between stock price volatility and dividend payout, in a sample of 84 firms listed in Bursa Malaysia, in the period of 2005-2010. The study indicated that there was a significant negative relationship between dividend yield and dividend payout with stock price volatility. They extended his research with additional control variables i.e. leverage, earning volatility, market value and growth as recommended by Baskin (1989). Therefore, it was found a negative relationship between market value and stock price volatility by utilizing multiple regression models and concluded in their research that dividend yield and market value have the biggest effect on stock price volatility among control variables. Similar results were presented by Safian and Ali (2012) on the same issue. The results of this research demonstrated a significant negative relationship of dividend yield with stock price volatility. A significant positive relationship between leverage and stock price volatility was also found by Safian and Ali (2012). In conclusion, it is intrinsic to emphasize that both studies presented the same results which matched with existing body of literature i.e. Baskin (1989) and Hussainey et al. (2011).

(10)

6

In light of the above, we can draw a conclusion that there has been a lot of discussion and research done in a different time frame and different countries, however, authors and researchers have not reached a consensus. One of the reasons could be that studies were conducted in different countries. Another reason could be different methods that were applied in the studies. Last but not least, different states of economy and financial systems also could have an impact on dividend yield and stock price volatility. In addition, it is crucial to emphasize that most of the studies found a negative relationship between dividend yield and stock price volatility, hence most of the results are in line with existing literature.

Also, given the predominantly negative relationship between dividend yield and stock price volatility, we can set up null and alternative hypotheses. Moreover, previous literature provides us with a solid foundation for our thesis and therefore the most efficient control variables were chosen according to previous research.

3. Theoretical Framework

_____________________________________________________________________________________

There are many theories that exist on the topic which explain the relationship between dividend yield and stock price volatility, i.e. agency cost, bird-in-hand fallacy, pecking order theory, clientele effect and signalling theory. In this section, it is intended to focus only on two theories, which are Miller and Modigliani theory (MM) and Bird-in-hand theory that vividly convey different perspectives on the same issue and therefore both theories do not come to the same conclusion. That was the main reason why we decided to focus only on these two theories. This will test which theory is applicable to real cases and data. First, we summarized Miller and Modigliani and then we discussed Bird-in-hand theory.

______________________________________________________________________

By painstakingly analysing the academic research paper conducted by Miller and Modigliani (1961) it was discerned that the relationship between dividend yield and stock

(11)

7

price volatility does not exist. One of the imperative and significant aspects to emphasize is that the theory was based and drew the conclusion on the assumption that perfect capital markets exist, the investors are rational and make a valuation of a company based on the discounted cash flow model, which is not the case in the real world as the markets involve taxes and transactional costs as well as investors that may not act rationally, which makes Miller & Modigliani Theory more theoretical rather than practical. Nevertheless, it is imperative to mention that imperfections can have an effect on a manager's decision when choosing particular policies under certain circumstances. Another important factor that has to be taken into consideration is that Miller and Modigliani made an argument that stated the following: earnings determine the value of the company and firm's investment policy defines the future cash flow, earnings and growth opportunities. According to the MM theory all decisions based on the investment policy, but not on the dividend policy.

Moreover, it was stated by Miller and Modigliani (1961) that a change in dividend rate beget a change in share price, due to investors assumption that if the dividends increase, it can be an indication that a company will generate higher earnings or will have higher opportunity growth and vice versa. However, the perception of dividend change can be wrongly interpreted, since the management of the organization might be just changing its payout target or in the worst case endeavouring to manipulate the price of the stock.

Other researchers, such as Brennan (1971),Black and Scholes (1974) and Adelfila, et al. (2004); supported findings that were presented by Miller and Modigliani. Even though MM theory had its followers i.e. Black and Scholes (1974) and Miller and Scholes (1982), other researchers did not agree with the conclusion that they drew on this issue.

In contrast, a theory established and expanded by Gordon (1960) and Lintner (1962), which is known as Bird-in-Hand Theory vividly conveys research findings that contradict the findings presented by Miller and Modigliani. According to this theory, the authors claimed that investors do care more regarding dividends that they receive today rather than capital gain in the future, due to the fact that shareholders prefer to be paid higher dividend yield and as a consequence take less risks in lieu of staying uncertain for a longer period of time, as in the world of information asymmetry and uncertainty investors prefer to take a safe card when investing (Gordon, 1960), (Al-Malkawi, 2007). Therefore, it is

(12)

8

an unsafe option for shareholders to choose capital gains over dividends, as investors understand that A bird (dividend) in the hand is worth two in the bush (capital gain).

Additionally, authors discerned that a reduction of dividend by one percent will require a capital gain and an increase in it by more than one percent owing to the fact that shareholders will require a compensation for a one-percent loss or decrease in dividends, thus managers will have more responsibilities and pressure from shareholders in order to deliver higher growth in the future, which cannot be guaranteed.

Gordon (1960) and Lintner (1962) drew a conclusion that a total return to shareholders comprises two factors, which are future growth rate of the dividend and dividend return, thus the value of the stock price can be measured by the Gordon Growth Model, which Baskin (1989) presented in his study. He analysed the duration effect and stated that “high dividend yields imply more near-term cash flow” which means that price volatility will decrease when the dividend yield is higher. It can be demonstrated by applying the Gordon Growth model. It is necessary to specify that the level of g (growth) in the dividends and an equity discount rate Ke is constant. Therefore, a stock price can be calculated by utilizing the next formula:

Pt – the value of stock,

Dt - dividend yield,

Ke - equity discount rate g - growth in the dividends

Step 1

𝑃_𝑡 = 𝐷𝑡+1 𝐾_𝑒−𝑔

(1)

Step 2: we are taking the first derivative with respect to Ke:

𝑑𝑃_𝑡 𝑑𝐾_𝑒 = − ( 𝐷_𝑡+1 𝐾_𝑒−𝑔) 2 (2)

(13)

9

Step 3: The third equation scrutinize the impact of discount rate on dividend yield:

−𝑑𝑃𝑡 𝑑𝐾𝑒 /𝑃𝑡 𝐾𝑒 = 𝐾𝑒/ 𝐷_𝑡+1 𝑃𝑡 (3)

Equation (3) demonstrates that high dividend yield stock will be less sensitive to fluctuation in discount rates and as a consequence will display lower price volatility, all other things being equal. Moreover, it is of the highest importance to mention two points. First point is that dividend growth was treated as a constant variable, while it could be that the effect is much more general. Therefore, there is a need for stability in the dividend yield as was discerned by Lintner (1956). Another important aspect that has to be taken into consideration is that interest rate sensitivity represents mostly undiversifiable risk (Baskin, 1989).

The main focus of this research is to test the hypothesis and to discern whether the relationship between dividend yield and stock price volatility exists and if the relationship is significantly negative or positive.

In light of the above, a deduction can be made since two theories that were mentioned previously are not in line with each other, we cannot predict what the final relation will be between the dividend yield and stock price volatility. However, based on the fact that the majority of previous studies revealed a negative relationship between dividend yield and stock price volatility, we are expecting to find the same results in the Swedish Stock Market.

(14)

10

4. Data

_____________________________________________________________________________________

In this section data will be presented and a description of all variables used in this study.

_____________________________________________________________________________________

Our analysis uses yearly panel data for the largest firms listed on the Stockholm stock exchange (OMX Large Cap) between the time period of 2010-2019. We chose this time period because we want to evaluate the relationship between stock price volatility and dividend yield after the financial crisis on a mature Swedish market. The study is focused on annualized data which means that each year is treated as a new observation for each company.

Companies included in the OMX large cap represent the largest companies in Sweden and therefore have a large number of investors, institutional investors and pension funds involved which is why this is of interest for national and international investors alike, as well as for managers that are mainly in control over the dividend payout policies.

The main variables of interest for our analysis are Stock price volatility and Dividend yield. In addition, we control for changes in stock price volatility by including control variables, namely Market Value, Earnings Volatility, Leverage and Asset Growth. We got access to the data for this study using Thomson Reuters DataStream. This is a platform that is commonly used for academic papers in the economic field. The use of secondary data saves us time because we do not have to read every company fillings report for each year and observation.

Only companies that have continuously been listed on the Stockholm exchange are included in the study. This is done to make our dataset balanced and only including mature companies. This will cause our dataset to suffer from survivorship bias since companies that did not fit the requirements are omitted from the sample. (Elton et al., 1996) This study excludes companies that are in the financial sector due to their heavily

(15)

11

regulatory environment which Hussainey et al. (2011) suggests. Subsectors in the financial sectors are banking, insurance, real estate, investment funds and mortgage companies. Last but not least, we encountered the issue of having 4 companies having missing data for several observation periods. We could gather the data for the missing observation from the company fillings but by doing this we would mix our data source which could reduce the reliability of our results. Instead we opted to remove these 4 companies with a missing observation so we could work with a balanced dataset. The 4 companies are not in the same industry and their missing data points seem to be random. Lastly, a vast number of companies have both A-shares and B-shares or even more shares listed on the exchange. We had to decide which of the shares would be used in our dataset otherwise we would face the risk of having biased results. Otherwise, a company with several classifications of their share-types would represents a larger percentage of the data sample thus giving biased results. The way we approached this is that we looked at the trading volume for each of the different shares and compared them. The share class with the highest trading volume would then be used in our study as this would provide a more accurate representation of the volatility. The final sample consists of 52 companies. Appendix 1 shows a list of all the included companies.

In the study by Baskin (1989), he emphasized the importance of including such control variables, as the logarithmic market value, leverage, asset and earnings volatility, because it is intrinsic when testing the significance of the relationship between dividend yield and stock price volatility to control for the impact of these external factors. Husseiney et al. (2011) included another control variable, asset growth which would account for companies that are in their growth stage usually have better investment opportunities and have a lower dividend yield.

Stock Price Volatility (SPV).

This is our dependent variable and it is based on the daily change of return on a stock price for a range of 1 year for each year. Then the daily standard deviation is calculated from the range of returns for 1 year, which is then annualized by taking the daily standard deviation times the square root of the number of trading days (DeMarzo and Berk, 2017). This value was directly accessed through Thomas Reuters DataStream. This is the most common way of calculating stock price volatility in the field of finance. Some previous

(16)

12

studies in the field are using a different formula to calculate volatility that is presented by Parkinson (1980) that utilizes the range from the highest price to the lowest price and divides it with their average and raised to the second power.

Dividend Yield (DY).

This is our focused variable and it is based on the dividend payout per share divided by the stock price of the share. This value is averaged over the year. This value was directly accessed through Thomas Reuters DataStream. According to our theoretical framework we expect it to either have a negative sign or a value of 0.

Logarithmic Market Value (MV).

This variable is based by the current market equity, which is averaged by the current shares outstanding times the stock price over the year. The market value is transformed using a logarithm with a base of 10 in order to account for magnitude (Baskin 1989). The market value was gathered directly through DataStream and the log transformation with a base of 10 was then applied. Larger companies are considered more stable and involve less risk thus we expect there to be a negative relationship between Market Value and the stock price volatility.

Leverage (LVG).

This variable is based on the long-term debt with a maturity over 1 year of a company divided by the total asset for each year. Long-Term debt and Total asset were directly gathered through DataStream and then a calculation was applied to get Leverage. According to Allen and Rachim (1996) there could be a relation between leverage and the stock price volatility as an increase in leverage could be seen as an increase in corporate risk. Therefore, we expect a positive relationship between leverage and stock price volatility.

Asset Growth (AG).

This variable is based on the change in total assets from the previous year divided by the previous year total asset (Hussainey et al. 2011). We expect asset growth to have a positive relationship with stock price volatility.

(17)

13

Earnings volatility (EV1).

As there is no clear defined formula for earnings volatility, we have decided to follow the steps presented by Baskin (1989).

√∑(𝑋 − 𝑋̅)2

𝑁 (4)

Where X is Earnings before Interest and taxes (EBIT) divided by Total Asset, 𝑋̅ is the average of EBIT divided by Total asset, and N is the total number of observations. This formula will provide a constant earnings volatility for each of the firms. Total asset is defined as the total value of all assets a firm owns and is presented in a company’s balance sheet.

We have also included a second formula in this study presented by Dichev and Tang (2009) to see if we would get similar results with different formulas for the control variable, however all the results from regressions with this formula will be presented in the appendix. The second formula presented by Dichev and Tang (2009) is calculated by taking the standard deviation of Earnings before interest and taxes (EBIT) for the most recent 5 years. Dichev and Tang (2009) argues that Earnings volatility has a persistence and predictive power up to 5 years into the future. The main problem utilizing this formula is the loss of observations, as the earliest regression that can be conducted will then be 2014. The abbreviation for this variable will be presented as EV2.

Earnings before interest and taxes (EBIT) and total asset data was directly accessed through DataStream and figures were calculated by following the steps presented by Baskin (1989) and Dichev and Tang (2009).

(18)

14

5. Method

_____________________________________________________________________________________

In this section descriptive tables of our dataset and a correlation matrix will be presented followed by the methodology of our main regression model. Afterwards the results will be presented and a discussion of the results.

5.1. Descriptive statistics

Table 1 provides a descriptive statistic of our dependent, independent and control variables of this study, with rows containing the mean, median, standard deviation, maximum value and minimum value for all the variables provided by this research.

While looking at table 1, that contains the descriptive statistics of our data for the Swedish stock exchange between 2010 to 2019 we have a mean stock price volatility of 24.95 percent for the given period, with a range from 10.95 percent to 53.88 percent. The median in the descriptive statistics provides us information in case our dataset suffers from large outliers which is seen in asset growth where the mean is 11.41 percent but the median is 6.30 percent. The range of this variable is between -65.49 percent and 575.25 percent and a standard deviation of 34.15 percent. However, our data for this variable is of similar nature as Hussainey et al (2011) study on the UK market that had a range of 426 percent and a standard deviation of 37.76 percent.

Table 1. Descriptive statistics

SPV DY EV1 LVG MV AG Mean 0.249505 0.028104 0.134416 0.166970 4.456796 0.114145 Median 0.238650 0.027250 0.101742 0.152008 4.435636 0.062950 Std. Dev. 0.069295 0.020656 0.101447 0.137637 0.571372 0.341493 Max 0.538800 0.202200 0.564831 0.914988 5.931179 5.752500 Min 0.109500 0.000000 0.018955 0.000000 2.832790 -0.654900

(19)

15

Observations 520 520 520 520 520 520

Companies 52 52 52 52 52 52

SPV: Stock Price volatility, DY: Dividend Yield, EV1: Earnings Volatility, LVG: Leverage, MV: Market Value, AG: Asset Growth

Table 2 provides a correlation analysis of all variables presented in this paper. The correlation analysis explains the relationship between independent variables and the dependent variable. In the case of stock price volatility and dividend yield the correlation value is -0.36. This result is in line with Baskin (1989) study of the US stock market (-0.643) and Hussainey et al. (2011) on the UK stock market (-0.2583). Allen and Rachim (1996) got a positive value (0.006) in their study on the Australian market which is in contrast to our findings. The correlation analysis also provides information whether we should raise concern if there could be potential multicollinearity between variables. The highest correlation of our dataset provided in table 2 is between stock price volatility and the control variable market value of -0.42. The variable is in line with theory in terms of the expected sign however, this could cause problems with the analysis of the regression.

Table 2 correlation analysis

Correlation SPV DY EV1 LVG MV AG SPV 1 DY -.367700 1 EV1 .268918 -.132293 1 LVG -.205393 .016470 .095446 1 MV -.423969 .148516 .021849 .249900 1 AG .005220 -.016097 .030733 .029281 -.129959 1

In appendix 3 and appendix 4, a descriptive statistic and a correlation analysis is provided for the reduced dataset that follows that calculates earnings volatility (EV2) following Dichev and Tang’s (2009) method. Comparing table 1, that includes the period of 2010

(20)

16

to 2019, with appendix 3, that includes the period of 2014-2019, one can see that there is not massive difference between the variables, with the exception of earnings volatility due to their differences in calculations. In appendix 3 the stock price volatility is lower (22.78 percent) than of the stock price volatility from table 1 (24.95 percent). This could be explained that table 1 includes a period of more volatility due to being closer to the financial crisis, while also including the flash crash, and the Greek debt crisis. Appendix 4 gives an overlook on the correlation analysis which shows a negative correlation between stock price volatility and earnings volatility which is different from the positive value shown in table 2. One can also see in appendix 4 a rather large positive correlation between earnings volatility and market value (0.42) which raises some concerns over multicollinearity of these control variables.

5.2. Regression model

This study is following Baskin (1989) using a multiple least square regression between the stock price volatility and dividend yield. The dependent variable is Stock price Volatility. A number of control variables are included to account for factors that might affect stock price volatility. As we are interpreting the results by using two different calculations of earnings volatility two different models will be tested.

Our final econometric models for interpretation are:

𝑆𝑃𝑉 = 𝛽₀+ 𝛽₁𝐷𝑌_𝑖𝑡+ 𝛽₂𝐿𝑜𝑔(𝑀𝑉)_𝑖𝑡+ 𝛽₃𝐿𝑉𝐺_𝑖𝑡+ 𝛽₄𝐴𝐺_𝑖𝑡+ 𝛽₅𝐸𝑉1_𝑖𝑡+ 𝜀 (5) 𝑆𝑃𝑉 = 𝛽₀+ 𝛽₁𝐷𝑌_𝑖𝑡+ 𝛽₂𝐿𝑜𝑔(𝑀𝑉)_𝑖𝑡+ 𝛽₃𝐿𝑉𝐺_𝑖𝑡+ 𝛽₄𝐴𝐺_𝑖𝑡+ 𝛽₅𝐸𝑉2_𝑖𝑡+ 𝜀 (6)

Where “i” captures the firm and “t” the year. SPV is Stock Price Volatility, DY is dividend yield, MV is Market Value, LVG is Leverage, AG is Asset Growth, EV1 is earnings volatility with the use of the formula presented by Baskin (1989), EV2 is earnings volatility using Dichev and Tang (2009) methodology and ε is the error term.

There are several methods of conducting a multi regression model. Because we are dealing with panel data, we have the option to choose between a Random effect model (REM), Fixed effect Model (FEM) or a pooled ordinary least square model.

(21)

17

In order to decide on which methodology, we should choose to conduct a Hausman test to assess whether or not a fixed effect model or a random effect model is appropriate. The Hausman test has the following hypothesis

𝐻₀: 𝑅𝑎𝑛𝑑𝑜𝑚 𝑒𝑓𝑓𝑒𝑐𝑡 𝑀𝑜𝑑𝑒𝑙 𝑖𝑠 𝑎𝑝𝑝𝑟𝑜𝑝𝑟𝑖𝑎𝑡𝑒 𝐻1: 𝐹𝑖𝑥𝑒𝑑 𝐸𝑓𝑓𝑒𝑐𝑡 𝑀𝑜𝑑𝑒𝑙 𝑖𝑠 𝑎𝑝𝑝𝑟𝑜𝑝𝑟𝑖𝑎𝑡𝑒

The results from conducting these tests shows that we should use a FEM at a 5% significance level when we are dealing with our second model, Appendix 2 shows the Hausman test. A pooled regression model is only appropriate for our first model since earnings volatility is constant over time for each firm which makes it impossible to conduct a Fixed effect model.

5.3. Result

First, we ran a regression of our main two variables, stock price volatility and dividend yield, which is shown in table 3. The results in table 3 indicates that there is a significant negative relationship between the stock price volatility and the dividend yield on the Swedish stock market, which is in line with previous studies by Baskin (1989) and Husseiney et al. (2011). The Durbin-Watson statistics is 0.47 which raises concerns for positive autocorrelation. Multiple control variables will be added to our initial regression to find further results.

Table 3 – Pooled OLS regression without control variables

Coefficient Std. Error t-Statistic Prob

C 0.2284173*** 0.003376 84.16868 0.0000

Dividend Yield -1.233549*** 0.096832 -12.73900 0.0000

F-stat. = 162.2822 DW-stat. = 0.47 𝑅2 = 0.14 Adj. 𝑅2 = 0.13

Periods included = 10. Included observations = 520. Total pool (balanced) observations = 1040 *=Significant at 10% level. **= significant at 5% level. ***=significant at 1% level

(22)

18

We followed our regression by utilizing a pooled OLS with our dataset. Because our data sample has fixed values for earnings volatility it will make regression using fixed effect model and random effect model obsolete, thus a pooled OLS is the proper method to use.

For the model, that utilizes EV1, a pooled regression was made and by looking at table 4, there is a significance influence on stock price volatility at a 5 percent level of significance for all independent variables. This means that 1 unit increase on one of the control variables will increase stock price volatility (SPV) by their respective coefficient. We can see on table 4 that dividend yield (DY) drives the volatility by -.936085. Earnings volatility (EV1) seems to also be a factor that drives stock price volatility (SPV) by 0.17608 which is in line with the theory and previous study by Baskin (1989) (1.265). However, leverage (LVG) has a negative coefficient meaning as leverage increases, the volatility decreases which is not in line with our theory and previous studies from Baskin (1989) (.162), Husseiney et al. (2009) (.2463) and Allen and Rachim (1996) (0.356). Market value (MV) is negative which is in line with theory and the result from Allen and Rachim (1996) (-0.022) and Husseiney et al. (2009) (-0.3405) but it is not in line with the results from Baskin (1989) (0.075). Asset Growth (AG) is negative which is not in line with the theory nor the results from Husseiney et al. (2011), however his results for asset growth were insignificant. The Durbin-Watson statistics shows a value of 0.49 which raises concerns for autocorrelation, however this cannot be seen in our correlation matrix presented in table 2.

Table 4 – Pooled OLS regression with control variables

C 0.460122*** 0.008123 56.64752 0.0000 Dividend Yield -0.936085*** 0.049436 -18.93538 0.0000 Earnings Volatility 0.173608*** 0.010001 17.35887 0.0000 Leverage -0.067113*** 0.007552 -8.886454 0.0000 Market Value -0.043816*** 0.001847 -23.71993 0.0000 Asset Growth -0.010173*** 0.002961 -3.435146 0.0006

(23)

19

Periods included = 10. Included observations = 520. Total pool (balanced) observations = 3120 *=Significant at 10% level. **= significant at 5% level. ***=significant at 1% level

For the second model, that utilizes EV2, a panel FEM was conducted, and by looking at appendix 5 dividend yield (DY) and market value (MV) are significant at 1 percent level and asset growth (AG) is significant at 5 percent level. The rest of the control variables are not significant even at the 10 percent level. Earnings volatility (EV2) has a negative coefficient which is not in line with theory and the findings of Husseiney et al. (2011) (.2809). The Durbin-Watson statistic is quite low (0.8246) which raises some concerns over some variables that might be autocorrelated. A robustness check took place by omitting leverage (LVG) to see if the results will be improved. Leverage was omitted because of it having the highest correlation with earnings volatility (EV2) which is shown in appendix 6. The following formula was then tested:

𝑆𝑃𝑉 = 𝛽₀+ 𝛽₁𝐷𝑌_𝑖𝑡+ 𝛽₂𝐿𝑜𝑔(𝑀𝑉)_𝑖𝑡+ 𝛽₃𝐴𝐺_𝑖𝑡+ 𝛽₄𝐸𝑉2_𝑖𝑡+ 𝜀 (6)

Where “i” captures the firm and “t” the year. SPV is Stock Price Volatility, DY is dividend yield, MV is Market Value, AG is Asset Growth, EV2 is earnings volatility using the methodology presented by Dichev and Tang (2009) and 𝜀 is the error term.

A new Hausman test was conducted which concluded that a REM was appropriate at the 5 percent level which can be seen in appendix 2. In this test dividend yield (DY) and market value (MV) remain significant at the 0.5 percent level. Asset growth (AG) is significant at the 5 percent level whereas earnings volatility (EV2) remains insignificant. The Durbin-Watson statistics remained low in this test as well which means there are concerns for autocorrelation, the results are displayed in appendix 6.

When comparing the results of the two models, we notice that the sign of EV1 and EV2 are different from each other where EV1 remains positive and in line with the theory whereas EV2 is negative, insignificant and not in line with theory. The rest of the variables have a similar coefficient sign. With the reduced number of observations and

(24)

20

the change in methodology, we conclude that this model is of less of value to our study. We will then only focus discussing the results with the regressions and formulas presented by Baskin (1989).

5.4. Discussion of Results

From the result of the paper's initial model, we can reject the theory that there is no relationship between dividend yield and stock price volatility, which means that dividend yield has a correlation on the volatility of the stock prices. The result shows that the Miller & Modigliani Theory does not hold for the Swedish Stock market between 2010 and 2019.

From the result of the paper's initial model there is a negative relationship between dividend yield and stock price volatility. Thus, this means that there is a negative correlation between stock price volatility and the dividend yield. A high dividend yield will decrease the stock price volatility, reducing the risk of holding such equities. This is in terms with the Bird in Hand Theory, which concludes that the higher the dividend yield, the less risk is perceived for investors and a negative sign of the coefficient is expected.

If we compare it with previous studies, most studies found a negative relationship between stock price volatility and the dividend yield. From the study conducted by Baskin (1989), he observed the US stock exchange over the time period from 1967 to 1986. His result signals a strong negative relationship between stock price volatility and the dividend yield. This relationship is in line with our conclusion. A difference between the study of Baskin and this study is that Baskin included a longer time period in their model and with the inclusion of smaller companies.

Hussainey et al. (2011) also found a negative relationship between stock price volatility and the dividend yield on the UK market from 1998 to 2007. This relationship matched with our findings. While their study and our study are conducted with the same number of observations, their study is conducted over the time period leading up towards the financial crisis whereas this paper is conducted over the time period after the financial crisis.

(25)

21

However, Allen and Rachim (1996), found instead a small positive relationship between stock price volatility and dividend yield on the Australian market from 1972 to 1985. This relationship is not in line with the results from this paper.

Because of these differential results there is a need for further studies in the field, if the location of the market plays a role, location of investors and the time period plays a significant role in how dividend yield has influence on the stock price volatility.

6. Conclusion

The results from this paper suggest that there is a significant negative relationship between the stock price volatility and the dividend yield on the Swedish stock exchange (OMX Large Cap) in the period of 2010 to 2019. The Miller & Modigliani theory does not hold for the time period and companies conducted in the study. However due to the limitations of our study, we cannot conduct if our results hold over a longer time period or if taxes play a substantial role on the results.

We also find that our control variables i.e. earnings volatility, leverage, asset growth and logarithmic market value, all play a significant role in stock price volatility over this time period. A positive statistically significant relationship can be seen for earnings volatility which goes in hand with the result presented by Baskin (1989), Allen and Rachim (1996) and Husseiney et al (2011). A negative statistically significant relationship can be seen for leverage, logarithmic market value and asset Growth. The result from leverage is not in line with our theoretical framework and not in line with results presented by Baskin (1989), Allen and Rachim (1996) and Husseiney et al. (2011). The result from Market Value matches with the theoretical framework and the results presented by Allen and Rachim (1996) and Husseiney et al (2011) but not in line with the result presented by Baskin (1989). Lastly asset growth is not in line with the theoretical framework nor the results presented by Husseiney et al. (2011).

The results of this paper are limited to equity companies listed on the Swedish stock exchange (OMX Large Cap) at the end of 2019 and have been continuously listed since

(26)

22

2010. Companies that have been delisted or had an Initial Public Offering (IPO) between this time period were dismayed. This study excludes the financial sectors as the current theoretical framework does not fit into their regulatory nature. The results of this paper are generalized over all sectors (excluding the financial sector) that are included in the Swedish stock exchange. Additionally, the findings of this research are based on a balanced dataset from one data source. Data that was not managed to be retrieved from the data source was discluded from the study. Companies that have more than 1 type of equity listed on the exchange got reduced to only include the type-class that had the highest trading volume. Last but not least, the model used for the result of this study is limited to control variables presented by Baskin (1989) and Husseiney et al. (2011). Also, a control variable for taxes that might affect stock price volatility is not included in this model.

This research can be extended to include companies that are included in the financial sectors such as banking, insurance, real estate, investment funds and mortgage companies listed on the Stockholm exchange. This could also be further researched by comparing all the different sectors, like utility, energy, financials, industry, communications, consumer goods, healthcare, technology, chemicals, materials, etc, and see if there is a comparative difference depending on the sectors involved. Thus, it could give valuable insight for investors whether a certain sector is more influenced by the dividend yield.

The data in this study was solely focused on the Swedish stock exchange listed on OMX Large Cap. To evaluate if our results hold true, the whole Swedish market could be looked at to include the smaller companies listed on exchanges such as Mid cap, Small cap, First North and even smaller stock exchanges such as Nordic SME and Spotlight stock market. This can also further be studied by comparing the Swedish exchange with the other Nordic exchanges to see if there is a comparative difference between the Nordic countries that have a similar economic standpoint and culture. Another interesting topic to do research onto is to see the relationship or impact on changes/volatility of dividends yields or dividends paid.

(27)

23

7. References

Adelfila, J. J., Oladipo J. A., & Adeoti, J. O. (2004). The Effect of Dividend Policy on the Market Price of Shares. International Journal of Accounting, 2(1), 49-62.

Allen, D.E., & Rachim, V.S. (1996). Dividend policy and stock price volatility: Australian evidence. Journal of Applied Economics, 6, 175-88.

Al-Malkawi, H. N. (2007). Determinants of corporate dividend policy in Jordan: an application of the Tobit model. Journal of Applied Accounting Research, 23, 44-70.

Baskin, J. (1989). Dividend policy and the volatility of common stock. Journal of

Portfolio Management, 15, 19-25.

Berk, J., & DeMarzo, P. (2017). Capital markets and the pricing of risk. In J. Berk & P. DeMarzo (Eds.), Corporate Finance (4th ed., pp. 350-388). Pearson.

Brennan, M. (1971). A note on dividend irrelevance and the Gordon valuation model. The

Journal of Finance, 26, 1115-1121.

Black, F., & Scholes, M. (1974). The effects of dividend yield and dividend policy on common stock prices and returns. Journal of Financial Economics, 1, 1-22.

Dichev, I. D. & Tang, V. W. (2009). Earnings volatility and earnings predictability.

Journal of accounting and economics, 47, 160-181.

Elton, E. J., Gruber, M. J., & Blake, C. R. (1996). Survivorship Bias and Mutual Fund Performance. The Review of financial studies, 9(4), 1097-1120.

Green, T. (2020). Stock Market Volatility Defined. The Motley Fool.

(28)

24

Gordon, M. J. (1960). Security and a Financial Theory of Investment. The Quarterly

Journal of Economics, 74(3), 472-492.

Gordon, M. J. (1963). Optimal Investment and Financing Policy. Journal of finance, 18, 264-272.

Hakeem, S. A., & Bambale, A. J, (2016). Mediating effect of liquidity on firm performance and dividend payout of listed manufacturing companies in Nigeria. Journal

of Economic Development, Management, I T, Finance, and Marketing, 8(1), 15-35.

Hashemijoo, M., Ardekani, A. M., & Younesi, N. (2012). The impact of dividend policy on share price volatility in the Malaysian Stock Market. Journal of Business Studies

Quarterly, 4(1), 111-129.

Hussainey, K., Mgbame, C. O., & Chijoke-Mgbame, A. M. (2011). Dividend policy and share price volatility: UK evidence. The Journal of Risk Finance, 12(1), 57-68.

Krass, P. (1999). The Book of Investing Wisdom. John Wiley & Sons.

Lintner, J. (1956). The distribution of incomes of corporations among dividends, retained earnings and taxes. American Economic Review, 46, 97-113.

Lintner, J. (1962). Dividends, earnings, leverage, stock prices and supply of capital to corporations. The Review of Economics and Statistics, 64, 243-269.

Miller, M. H., & Modigliani, F. (1961). Dividend policy, growth and the valuation of shares. The Journal of Business, 34, 411-33.

Miller, M. H., & Scholes, M. S. (1982). Dividends and Taxes: Some Empirical Evidence.

(29)

25

Nazir, M. S., Musarat, M., Waseem, N., & Ahmed, A. F. (2010). Determinants of Stock Price Volatility in Karachi Stock Exchange: The Mediating Role of Corporate Dividend Policy. International Research Journal of Finance and Economics, 55, 100-107.

Nguyen, T. H., Nguyen H. A., Tran, Q. C., & Le, Q. L. (2020). Dividend policy and share price volatility: empirical evidence from Vietnam. Growing Science Accounting (North

Vancouver), 6 (2), 67-78.

Profilet, K. A., & Bacon, F. W. (2013). Dividend policy and stock price volatility in the US equity capital market. ASBBS Proceedings, 20(1), 219.

Roy, A. (2015). Dividend Policy, Ownership Structure and Corporate Governance: An Empirical Analysis of Indian Firms. Indian Journal of Corporate Governance 8(1), 1-33.

Safian M. F. A., & Ali N. (2012). Dividend policy and share price volatility: Evidence from Malaysia. IEEE Colloquium on Humanities, Science and Engineering (CHUSER), 221-226.

(30)

26

Appendix 1 – List of included OMX Large Cap

Companies

AAK ABB Addtech AF

Alfa Laval Assa Abloy Atlas Copco Axfood

Astra Zeneca Beijer Ref Betsson Billerud Korsnas

Boliden Elekta Electrolux - B Ericsson - B

Getinge Hexagon Hennes & Mauritz Holmen - B

Hexpol Husqvarna - B Ica Gruppen Indutrade

Kindred Loomis Lundin Mining Lundin Energy

MTG - B Mycronic NCC - B NIBE

Nobia Nolato Peab Saab

Sandvik SCA - B Securitas Skanska

SKF - B Swedish Orphan

Biovitrum

SSAB - A Sweco - B

Swedish Match Tele2 - B Telia TietoEVRY

Millicom Trelleborg Vitrolife Volvo - B

Appendix 2 – Hausman Test

Hausman test variables. P-value

DY, LVG, MV, AG, EV2 0.0454

(31)

27

Appendix 3 – Descriptive Statistics (Second model)

SPV DY EV2 LVG MV AG Mean 0.227845 0.027570 1577544 0.172850 4.563235 0.119681 Median 0.218000 0.025900 577311.1 0.152085 4.492464 0.081800 Std. Dev. 0.054097 0.019171 2920973 0.143848 0.499704 0.240736 Max 0.394600 0.175200 24826366 0.914988 5.931179 2.256600 Min 0.109500 0.000000 15960.51 0.000000 3.084279 -0.654900 Observations 312 312 312 312 312 312 Companies 52 52 52 52 52 52

SPV: Stock Price volatility, DY: Dividend Yield, EV1: Earnings Volatility, LVG: Leverage, MV: Market Value, AG: Asset Growth

Appendix 4 – Correlation Analysis (Second model)

Correlation SPV DY EV1 LVG MV AG SPV 1 DY -0.384337 1 EV2 -0.068952 0.081600 1 LVG -0.215805 0.018923 0.149973 1 MV -0.475757 0.184914 0.421144 0.218526 1 AG 0.061895 -0.062527 -0.105923 0.041465 -0.186105 1

(32)

28

Appendix 5 – Panel fixed effect model.

C 0.543742*** 0.044003 12.35701 0.0000 Dividend Yield -0.440703*** 0.103637 -4.252374 0.0000 Earnings Volatility (EV2) -2.98E-10 8.29E-10 -0.359541 0.7195 Leverage -0.002695 0.022188 -0.121481 0.9034 Market Value -0.066012*** 0.009534 -6.923635 0.0000 Asset Growth -0.013236** 0.005891 -2.246648 0.0255

Periods included = 6. Included observations = 312. Total panel (balanced) observations = 312

*=Significant at 10% level. **= significant at 5% level. ***=significant at 1% level

Appendix 6 Robustness check – Panel random effect

model

C 0.511956*** 0.034040 15.03976 0.0000 Dividend Yield -0.477321*** 0.096808 -4.930584 0.0000 Earnings Volatility (EV2) 2.99E-10 7.76E-10 0.385239 0.7003 Market Value -0.059137*** 0.007318 -8.080619 0.0000 Asset Growth -0.013080** 0.005653 -2.313939 0.0213

Periods included = 6. Included observations = 312. Total panel (balanced) observations = 312

Stock price volatility and dividend yield: Evidence from Sweden