0
Is managerial compensation a reliable predictor for future stock
returns?
Thesis MSc. Finance
Abstract
This study examines if managerial compensation in public firms are a reliable predictor of the stock performance. According to literature, proper incentives can lead to an increase in firm performance. This paper tries to determine if constructing stock portfolios based on managerial compensation can achieve abnormal returns. The portfolios will be based on the earnings and owner earnings, as well as bonus, salary and stock holding structures for upper management. Firms in the S&P 500 were studied for a period of 27 years. The strongest performing portfolio was constructed using PE values and stock holding by management. This portfolio realized an annual return of 8.8% over a holding period of 10 years while the benchmark portfolio realized 8.0%. This result was consistent throughout the holding period. Applying the Fama and French three-factor model indicates that his additional return can be attributed to risk-factors. Therefore we conclude that investment strategies based on the managerial compensation structures of stocks do not offer better risk-adjusted returns. This result is robust to a momentum effect following the Carhart four-factor model.
Keywords: Efficient Market Hypothesis, Three-factor model, Value investing, Managerial compensation
JEL classifications: C23, C33, G11, G14, G17, G35, G40
Name: Ewoud Nijhuis Student number: 2608839 Supervisor: dr. J.J. Bosma Date: 07-08-2020
1
Introduction
In 1949, Benjamin Graham’s The Intelligent Investor was published. With this book, the investment style of value investing was born and it still remains relevant to investment decision making around the globe. The book outlined how an investor could achieve satisfactory returns on his stock investments using fundamental analysis. Briefly it states that stocks do not always trade at their intrinsic value, but some stocks can be bought at a discount. The market price is sometimes significantly lower, or higher than its objective worth to the investor. This theory has been highly debated in academic circles. Generally, academics seem to agree on the idea that markets are efficient, and thus no mispricings can exist. (Fama & Malkiel, 1970) Proponents of this efficient market hypothesis state that value investing returns can be contributed to additional risk that these value stocks bring. Opponents of the EMH such as Shiller believe in return predictability. They believe that returns are predictable to some degree and that mispricings may exist in the stock market. In this paper, I will try to add to this debate by determining if value investing was worthwhile over the last few decades. Furthermore, I will expand on the value investing framework by introducing managerial compensation proxies to the selection criteria. Using the Fama and French three-factor model, I will determine if selecting stocks based on managerial salary, bonuses and stock holdings rewards the investor with higher risk-adjusted returns.
2
compensation proxies to account for the qualitative side of the fundamental analysis. It is found that the value investing strategy can still lead to abnormal returns, and that accounting for the stock holdings of management prior to your investment decision is a reliable indicator of future stock performance. In the literature therefore I present the literature known, as well as practices from well-known value investors as Warren Buffet and Philip Fisher. Finally, this paper will test the performance of the quantitative and qualitative value indicators by constructing portfolios and analyzing these over various holding periods up to 10 years. I will test if a value portfolio that incorporates compensation proxies outperforms a traditional value portfolio. In this paper, I will try to add to the debate between Fama and Shiller by determining if incorporating managerial proxies in a value investing framework leads to higher returns when controlled for certain risk factors. This will be based on the Fama and French three-factor model, constructed using the Fama-MacBeth two-step regression. Literature Review
Market efficiency
Investment strategies have been widely discussed in asset pricing literature. There is no consensus on whether or not such strategies could exist, and whether it is due to extra risk or not. Proponents and opponents of the EMH have debated this for decades, and they have yet to agree. Both parties agree that value stocks have historically outperformed growth stocks. However, Fama argues that the basis for this excess return is additional risk. To Shiller, investors are rewarded for cleverly exploiting an irrational market’s mispricing of securities. (Shiller, 1981) In this paper, I will try to add to the debate by determining if managerial proxies can predict returns while controlling for risk factors as defined by the Fama and French three-factor model (Fama & French, 1996).
3
market volatility. This beta factor is based on historical data. However, CAPM was found to be flawed by Fama and French (1996). They found that historically, small cap stocks and high book-to-market stocks outperformed large cap stocks and low book-to-market stocks respectively. Thus, they introduced a size and value factor to account for this finding. This resulted in their well-known three-factor model. (Fama & French, 1992; Fama & French 1993) It is important to note here that Fama and French find that the additional factors do not mean that there is market inefficiency. Rather they conclude that these are additional risk factors that should be accounted for. In this paper, I will estimate the performance of the stocks using the three-factor model. As a robustness check I will also use the four-factor model as proposed by Carhart (1997). This model includes a momentum factor. In addition to the three-factor model, Fama and French proposed a five-three-factor model in 2015. (Fama and French, 2015). This model also included factors for profitability and investment. This is however a rather new model, and thus analysis is better comparable when done using the three-factor model.
Opponents of the EMH mostly adhere to the behavioral finance. This model assumes that investors on the stock market behave irrationally, causing the market to be inefficient. Mispricings may occur, and there could even be capitalized on this by investors. In Shiller (1981) it is stated that the stock market is too volatile to be explained by a difference in information. He therefore concludes that there must be other factors driving the market. In 1988, Campbell and Shiller found that accounting data could accurately predict the present value of future dividends. They found however, that the stocks were not priced according to the expected dividend flow. This is unexpected, since this can be considered a reliable indicator of future stock performance. Shiller and Campbell conclude not always all information is reflected in the stock price, and thus leave room for return predictability. In a way, my thesis sets out to add an additional predictor for stock returns in the way of managerial compensation proxies.
4
Jegadeesh and Titman (1993) found that a portfolio that bought stocks that performed well in the past and sell stocks that performed poorly in the past generated significant positive returns. They found that the profitability was not due to their systematic risk or delayed stock price reactions to common factors, but rather due to initial mispricing by the market. This effect was caused by the combination of investors’ underreaction and overreaction to information. All these behavioral mispricings in the market allowed the value investing approach to achieve systematic abnormal returns. It is important to understand why the value premium is present. The following section will look at potential explanations.
Different studies give various explanations for the existence of the value premium. Asness et al. (2015) found that the value investing approach is more suited to small-cap stocks. However, they did find positive results for large-cap stocks, albeit smaller. Therefore, they argued that a part of the value investing returns was due to size effect. This is a common critique of the value investing approach. According to Reinganum (1980), Banz (1981), and Stattman (1980), the performance of stocks correlates more strongly to the firm size than their PE or PB ratios. These studies found evidence for a value premium and all noted that this premium is significantly smaller for large-cap stocks. (Fama & French, 1998) Bauman et al. (1998) studied the performance of value stocks versus growth stocks. By using four valuation ratios to define value stocks, they found that value stocks outperform growth stocks on an absolute and risk-adjusted basis. They also found a strong size effect. However, value stocks outperformed growth stocks in nearly all size categories.
Value investing
Value investing is an interesting subject of research within the academic world. Currently, most academics agree that the market is efficient and thus leave no room for investment strategies that reliably beat the market. In order to accurately describe the literature that is available, I will set out the literature on value investing, going into both sides of the fundamental analysis. Following this, I will explain the role of managerial compensation for shareholder returns.
Benjamin Graham outlined the foundation for the value investing approach in his 1949 book,
The Intelligent Investor. He stated that a portfolio constructed using this approach would be
5
stock compared to the profitability of the company. The qualitative analysis focuses on non-quantifiable qualities that would make a stock a good investment, such as its products, management, barriers to entry the market, and its competitive advantage. With the help of these two approaches, the value investor can buy stock in qualitatively good companies for a fair price. According to value investors, this approach should lead to abnormal returns in the long run. These abnormal returns achieved by value stocks are called the value premium. Quantitative value analysis
The main financial ratios that Graham took into account were the price-to-earnings (PE) ratio and the price-to-book (PB) ratio (PB). The PE ratio measures the market price of the company related to the earnings, or the net profit of the most recent book year. He uses a maximum PE value of 9.0, the lower the PE value, the better. The PB ratio compared the market price of the company to its book-value based on its balance sheet. As PE ratios can sometimes be misleading, PB can be used as an additional measure. Graham used to ratio of less than 1.2 as a rule of thumb. Furthermore, Graham required the companies to have positive earnings growth and to be paying out dividends.
Athanassakos (2012) defines the quantitative value investing process as a three-step process. First, one must identify possibly undervalued stocks by choosing stocks with low PE, PB, or other valuation-related metrics. Second, the intrinsic value of the stocks selected must be calculated. Finally, a buy decision must only be made when the value is below the current price by a predetermined margin of safety. The margin of safety is an important concept within the value investing approach. As the value of a stock is based on many variables, one should only invest in stocks which prices differ significantly from their intrinsic value. Yee (2008) states that “Margins of safety are typically 20% to 35% of the share price for value stocks, they could be over 50% for growth companies.” Seth Klarman (1991) added that “For a value investor a pitch must not only be in the strike zone, it must be in his ‘sweet spot.’” The margin of safety implies that investors should avoid excessive trading. Regarding the holding period that value-investors suggest, there is a clear consensus among the experts. According to Warren Buffett, the best time frame for holding a stock is forever. (Lowe, 1997). On holding periods, Rousseau and van Rensburg (2003) state that value investing becomes more reliable as the investment horizon gets longer. They find that value investors seem to be rewarded for their patience and time. This confirms Buffett’s statement. In this study, I will examine both short and long holding periods to determine if longer holding periods indeed deliver a more reliable and higher performance.
6
This study found a negative correlation between the PE ratio and future stock performance. This is thus a confirmation of the value investing approach. Basu stated that the relationship exists because of exaggerated investor reaction. According to the study, portfolios consisting of low PE firms earned higher absolute and risk-adjusted rates of return than their high PE counterparts. This research focused on the US market, but similar results were found in an international context by Bauman et al. (1998), Fama and French (1998), and Chan et al. (1991). While this first wave of academic research regarding value investing primarily focused on the P/E ratio, other studies were done with different quantitative measures. Capaul et al. (1993) and Chan et al. (1998) found that low P/B portfolios performed significantly better than their high P/B counterparts on both an absolute and a risk-adjusted basis. Since the P/E and P/B ratios measures were similar, the results were in line with the expectations. Piotroski (2000) showed that the mean return earned by an investor can be increased by 7.5% annually by applying a simple accounting-based fundamental analysis. To achieve this, he selected firms with high book-to-market ratios. Additionally, an investment strategy that bought expected winners and shorted expected losers generated a 23% annual return between 1976 and 1996. He found evidence that the market initially underreacted to known historical information and that this was eventually corrected after new earning reports. Overall, the evidence suggested that the market does not fully incorporate historical financial information into prices instantly, which can lead to attractively priced stocks for investors.
Qualitative value analysis
The qualitative analysis a value investor does for its potential investments depends on many factors, such as the market in which the firm is active, the products or services it sells, and the management team responsible for setting out a corporate strategy. In the book Common
Stocks and Uncommon Profits, Philip Fisher describes 15 points to look for in a company.
7
relations, depth to the management, and good communication. The largest part of Fisher’s investment criteria is thus related to the management team (Fisher, 2003).
Focusing on the qualitative approach of value investing is associated with Warren Buffett. While the initial philosophy of Graham was to buy companies that were objectively a bargain, Buffett stated that he preferred to buy wonderful companies at a fair price. This approach requires the company one invests in to be better than average, with a price that is not higher than one would expect. During the 1994 annual Berkshire Hathaway shareholder meeting, Buffett stated that he judges management in two ways. First, he looks at their past performance, considering how well they ran the business and how well they did as compared to competitors. Second, he looked at how the management treated shareholders in the past. Poor managers also turned out to be the ones that did not give much consideration to the shareholders. This implies that in his qualitative analysis, Buffett tries to seek out companies that minimize the principal-agent problem, which is a problem that arises when the interests of management and shareholders are not aligned.
Managerial compensation
Principal-agent theory states that an agent should have an incentive to work in the best interest of the shareholders. The principal-agent theory implies that the manager is not optimally intrinsically motivated to maximize the shareholder value. To maximize shareholder value, the interests of the managers and the shareholders should be aligned (Sappington, 1991). Therefore, instead of paying a simple salary, most firms opt to compensate their managers in other ways in order to align managerial interests with those of the shareholders. To do this, they might include options, restricted stock, or performance-based components in manager salaries. Furthermore, studies show that a possible takeover threat serves as a natural incentive for managers to perform. If the share price is low, the chance of a hostile takeover happening is larger. When this occurs, management is usually replaced. However, it has been found that this effect alone does not optimize management performance (Fama, 1980). Research has shown no clear winner between long-term compensation options, as different characteristics between firms require different compensation requirements. However, a clear relationship was found between long-term managerial compensation plans and an increase in shareholder wealth (Brickley et al., 1985).
8
performance. However, bonuses are generally linked to the performance of the firm in the previous year, which could incentivize the manager to act in such a way that short-term success prevails over long-term success goals. This may hurt the shareholder, especially those that have a long investment horizon. Guidry et al. (1999) have shown that managers often made decisions that maximized their short-term bonuses. This might mean that awarding term bonuses may even hurt long-term goals, as it seems that managers prefer short-term rewards over long-short-term rewards.
The most positive findings for increasing shareholder wealth involved stock-based compensation. This also held true for management that already owned a significant amount of stock. The findings for bonuses as an indicator of strong performance were ambiguous. Jensen (1990) concluded that bonuses made up 50 percent of the total salary, but, in most cases, the bonuses were not sensitive to the market value of equity. Therefore, I expect the most positive compensation indicator to be stock holdings. One might conclude that short-term bonuses incentivize the manager to not act in the interest of long-short-term shareholders. In the sample used in this study, multiple investment periods will be analyzed. I expect to see that bonuses are indeed a reliable indicator for short-term performance, but have diminishing, or even negative, returns in the long-run.
Abowd (1990) investigated the correlation between managerial compensation and firm performance. Abowd also investigated the relationship between managerial compensation and shareholder wealth. He found that there was a strong positive correlation in both cases. For a 10% increase in compensation, the shareholders were found to have a 4 to 12% increase in wealth the following year. Barkema et al. (1998) studied the relationship between managerial compensation and firm performance and came to similar conclusions.
Brickley et al. (1985) studied the relationship between managerial compensation and shareholder wealth. Based on an event study surrounding compensation announcements, they determined that an increase in CEO compensation was met with positive market reactions. This result was limited to long-term managerial compensation packages. However, between types of long-term compensation packages, they found no clear winner. This is because firms with different characteristics require different compensation requirements. They defined four different long-term compensation methods: overall plans, plans containing options, plans containing performance plan components, and plans containing restricted stock components.
9
presence of long-term managerial compensation packages will be used as a proxy to quantify the qualitative element of value investing. Because long-term managerial compensation plans align the interests of the agent and principal, I hypothesize that these plans can be a good proxy for the qualitative component of value investing. This information, combined with various quantitative financial ratio measures, allows for the construction of an investment strategy.
Jensen et al. (1990) used various compensation measures. Since shareholders cannot observe every move of a firm’s management, they do not know what actions can be taken to increase shareholder value. Agency theory thus predicts that compensation policy should be designed to give the management team incentives to maximize shareholder value. This will align the interests of the shareholders and the management, possibly solving the principal-agent problem. Jensen stated that in large firms, management tended to own less stock and have less compensation-based incentives than management in smaller firms. Furthermore, the relationship between CEO compensation and shareholder value has decreased from 1940 to 1990. The largest incentive came from the ownership of the firm’s stock. Bonuses did make up a significant part of the total compensation package of the CEO. However, studies have found that these bonuses are often not dependent on performance measures such as shareholder value, earnings, or sales. Finally, Jensen found a strong relationship between the previous year’s stock performance and salary revision in the following year. Therefore, it can be concluded that large changes in salary imply a strong incentive-based salary. This means that salaries that increased significantly based on the previous year’s performance did so because of a strong incentive during that year.
In a value investing framework, a good management team was found to be a strong indicator of a good investment. In Fisher (2003), most of the criteria used for selecting a stock was based on the qualities and integrity of the management. As the literature has shown, managerial compensation packages can be utilized to properly motivate managers. A proper compensation package can be an indicator of a good management team. Building on this idea, this paper will develop three proxies based on three different compensation instruments. These compensation instruments will be integrated into the value investing framework to determine if these components add value to the investment strategy.
10
It is a fundamental analysis of the business model, its management, and the future prospects of the company. Because long-term managerial compensation plans align the interests of the agent and principal, I hypothesize that these plans can be a good proxy for the qualitative component of value investing. Based on Jensen (1990), I will use the relative bonus payout, annual salary increase and stock holdings by upper management as the compensation proxies. These compensation instruments will be integrated into the value investing framework to determine if these components add value to the investment strategy.
Methodology Research Questions
This chapter outlines the research questions, hypotheses, and research methodology. The main research questions this paper will try to answer are as follows:
- Are compensation proxies reliable indicators for future stock performance within a value investing framework? If so, which compensation measures result in the highest performance gain?
- Do the managerial compensation proxies predict returns that the Fama French three-factor model does not?
To answer this question, stocks in the S&P 500 will be analyzed. These stocks share similar characteristics in size, geographical risk, and systematic risk, and thus are suited for such comparative analysis. To answer the research question, some other questions have to be answered first.
- What is the performance of the control group, the S&P 500?
- How does a portfolio based only on quantitative measures perform? - What quantitative measure delivers the best performance?
- How does a portfolio based only on qualitative measures perform? - Can the additional return realized be attributed to additional risk? - Do the effects of compensation proxies differ between holding periods?
11
revalue these stocks. According to Graham, mispricing tends to fix itself between 1 and 2.5 years. However, since his research is very dated, it can be assumed that the market has gotten more efficient in recent years. If mispricing ceases to exist after a certain number of years, we may observe that the abnormal returns are diminishing as the holding periods get longer. I expect that the managerial proxies significantly predict positive alphas.
Based on the literature, a portfolio constructed using a quantitative approach is likely to outperform the market. A portfolio based only on qualitative performance is not likely to outperform the market. The effect of holding periods is ambiguous. On the one hand, the returns should only be abnormal until the market fixes the portfolio mispricing. This should take a fixed amount of time, after which the returns will likely be in line with the market. However, Rousseau and van Rensburg (2003) state that rewards and reliability grow as the holding period grows. This predicts that a longer holding period is better. Furthermore, I expect that these excess returns are not predicted by the Fama and French model. Therefore, I expect that the managerial proxies reliably predict positive alphas in all holding periods. Quantitative Measures
This study aims to test whether stock selection on a value investing basis can lead to systematic abnormal results, or a value premium. The quantitative measures used will be based on the balance sheet, cash flow, and income statements of each company. In particular, two measures of each firm’s earnings will be taken to determine the portfolio composition: the reported net earnings and the owner earnings. In the 1986 Berkshire annual shareholder letter, Warren Buffett first described owner earnings. He stated that the true value a stock held for shareholders was the actual dollar amount that could be extracted annually from the firm while not interfering with operations. Buffett described an alternative way to state earnings, which varied from the reported earnings that follow form accounting standards. Buffett describes owner earnings as: (1) report earnings + (2) depreciation, depletion and amortization, and certain other non-cash chargers – (3) the average annual amount of capital expenditures + (4) change in net working capital. This can be interpreted as the reported earnings that are adjusted for charges that do not impact the money an owner could extract from the business without hurting the business itself. In this study’s sample, the owner earnings will be determined using formula 1.
𝑂𝐸 = 𝑁𝐼 + 𝐷𝐷𝐴 − 𝐶𝑎𝑝𝐸𝑥 + ∆𝑁𝑊𝐶 (1) OE = Owner earnings
NI = Net income
12
CapEx = Reported capital expenditure ΔNWC = Change in net working capital
The reported earnings can be found in a firm’s income statement. The depreciation, depletion, and amortization are cash flow items. The average capital expenditure can be found in the cash flow statement as well. Since it is hard to estimate the (forward) average capital expenditure, the total capital expenditure is taken. This is a conservative estimate, and thus incorporates a natural margin of safety into the owner earnings measure. Finally, the changes in working capital are defined as the net change in current assets and current liabilities. These can be found in the balance sheet, after which the change in net working capital can be calculated.
When making an investment decision, the current price of the security is an important factor. Therefore, the calculated earnings will be related to the current market price. This will be done using formula 2.
𝑃𝑂𝐸 = 𝑃
𝑂𝐸 𝑆⁄ (2) POE = Price / Owner earnings ratio
P = Security price at the moment of investment OE = Owner earnings
S = Number of shares outstanding
This ratio tells how much a share costs relative to the share of the profit it is entitled to. Thus, the lower this POE ratio is, the more value you get for your money. This is the core idea of value investing. In addition to the calculated owner earnings, this study will also explore the role of the reported net earnings. This measure can be found in the income statement and does not have to be altered for analysis. Using a similar approach to calculating the owner earnings per share, but substituting owner earnings for net income, formula 3 is formed.
𝑃𝐸 = 𝑃
𝑁𝐼 𝑆⁄ (3) PE = Price/ Earnings ratio
P = Security price at the moment of investment NI = Net income
S = Number of shares outstanding
13
good “bang for your buck.” In this study’s sample, it is important to isolate the firms that are below the median value for the respective ratio. Using this method also accounts for high- and low-valuation periods in the stock market, and is, therefore, less susceptible to be influenced by the timing of the market. It should be noted that the median PE ratio is around 16 for most of the sample years. The median POE lies lower at 12.5. The median is chosen for investment decisions, as we do not want outliers to influence our investment decisions, but rather take the most attractive half of the stocks available. In the sensitivity analysis, we will alter the upper bound for the ratios so that the robustness can be determined.
When applying the PE or POE metric, the value investing framework states that a lower PE or POE ratio implies a more attractive investment. However, there are some issues with very low ratios. First, negative ratios imply that negative earnings have been realized over the last period. Value investing requires a company to turn a profit, so stocks with negative PE or POE ratio must not be selected when applying the quantitative metric. Furthermore, simply having a positive PE or POE ratio does not instantly make the stock attractive. A very low ratio implies that the company is trading at only a few times its earnings. This can be for various reasons, but it can be assumed that this is a bad sign. Either the earnings are misstated, the stock is a penny stock, or there are very few common shares outstanding. Regardless of the reason, it is necessary to use a floor in the construction of the portfolios. For the PE and POE ratios, a minimum value of 3 will be required. This implies that the stock price is 3 times the annual earnings. Assuming a 100% payout rate, this implies that an investor would earn back his investment in only three years, excluding any capital gains. Any less than this is considered too good to be true and is thus excluded from the quantitative portfolios. Using a lower-bound of 3 isolates the outliers from the sample while not significantly influencing the result of the selection process. This lower-bound will also be analyzed in the sensitivity analysis to determine the role of this arbitrary value. Therefore, in our value portfolios, we will include stocks that satisfy the following conditions:
3 < 𝑃𝐸 ≤ 𝑃𝐸̃ 𝑃𝐸̃ = The median PE at the moment of the investment decision PE = The PE value for the security
3 < 𝑃𝑂𝐸 ≤ 𝑃𝑂𝐸̃ 𝑃𝑂𝐸̃ = Median POE at the moment of the investment decision
14
In practice, value investors usually complement their quantitative analysis with a qualitative study. Studies have shown that long-term compensation has a positive correlation with shareholder wealth. In this academic research, however, this component of value investing was not often quantified. This paper will test if a qualitative measure based on managerial compensation can add to the value investing framework. The rationale behind this is that the quality of a management team will likely be reflected in its compensation. Since good management is found to increase shareholder value, a portfolio that ensures well-managed companies might offer satisfying returns. Three proxies will be constructed based on the following: (1) the bonus of the executive relative to his base salary, (2) the annual salary increase the executive received in the last year, and (3) the percent of stock the management team holds in the company. All three factors are expected to positively forecast high returns. However, it is expected that the bonus will be especially relevant in the short-term, while the stock ownership will be a more reliable indicator for long-term gains. To determine these factors, this paper will look at the top 5 board members per firm based on total compensation. The average relative bonus of the firm will be calculated as stated in formula 4.
𝐴𝑅𝐵 = ∑ 𝐵𝑚 𝐶𝑚 𝑚 𝑛=1 𝑚 (4) ARB = Average relative bonus
B = Bonus C = Base salary
M = Number of executives for which data is available. Max 5
The annual base salary increase for each executive was obtained from the Compustat database. From this, the average annual base salary increase for the firm can easily be calculated. Finally, the third proxy based on the stock held by management will be calculated as a percentage of the total shares outstanding.
𝐴𝑆𝐻 = ∑ 𝐻𝑚 𝑆 𝑚 𝑛=1 𝑚 (5)
ASH = Average stock held by management expressed as a percentage of total shares outstanding Hm = Stock holdings of executive m
S = Amount of shares outstanding
15
For each year, twelve portfolios will be constructed on the first (trading) day of the year. These portfolios will be built according to the following schedule:
Table 1 - Portfolio construction measures and lower and upper bounds
Portfolio construction measures Portfolio Quantitative measure Lower bound Upper bound Compensation Measure Lower bound Upper bound
1 All stocks will be included
2 POE 3 Median
3 PE 3 Median
4 Bonus Median -
5 Salary Increase Median -
6 Stock Held Median 1
7 POE 3 Median Bonus Median
8 POE 3 Median Salary Increase Median
9 POE 3 Median Stock Held Median 1
10 PE 3 Median Bonus Median
11 PE 3 Median Salary Increase Median
12 PE 3 Median Stock Held Median 1
For each portfolio, the returns will be analyzed for up to 10 years. Following this, the total return for the holding period will be calculated each year, as well as the annual and average yearly return. The annualized return for an individual stock will be calculated using the following formula:
𝑅𝑛 = √𝑃𝑧
𝑃0
𝑧
(6)
Where Rn is the annual return for firm n, z is the holding period in years, Pz is the price at the end of the holding period and P0 is the price at the construction moment. The prices are adjusted for dividends and/or stock splits. The return calculated is thus the total return an investor would have gained during a holding period.
The annualized return for each portfolio will be calculated using the following formula: 𝑅𝑝 =𝑅1 + 𝑅2 + 𝑅3 + ⋯ + 𝑅𝑛
𝑛 (7)
Where Rp is the annualized portfolio return and n is the number of stocks. This formula is
16
is to say, a portfolio can be thought of as a portfolio holding exactly one share of each selected firm.
Measuring performance
Only measuring the returns of the portfolios does not indicate if their performance is due to higher risk or if the performance is objectively better. To determine if the excess return is explained by an existing risk-factor, the Fama and French three-factor model alpha will be calculated. Following this, I will regress these alphas on the managerial proxies. Furthermore, the Sharpe ratio will be calculated to assess the portfolio risk-adjusted performance. The Sharpe ratio measures the excess return of an asset compared to the standard deviation. The excess return here is defined as the return of the asset minus the risk-free return. Equation 8 gives the formula I use to calculate the Sharpe ratio. Here, R-Rf is the excess return of the
portfolio during the holding period. The risk free rates used in calculating the Sharpe ratio and Fama and French model are obtained from the Kenneth French data library.
𝑆 = 𝑅 − 𝑅𝑓 √𝑉𝑎𝑟[𝑅 − 𝑅𝑓]
(8)
17
our sample using a Durbin-Watson test, the regressions will be done using OLS and estimating standard errors.
First, a time-series regression will be done using the monthly returns as the dependent variable. The regressors in the time series will be the monthly risk factors as obtained from the Kenneth French data library. The regression equation is specified by equation 9. Here, 𝑅𝑖𝑡− 𝑅𝑓𝑡 is the excess return for month t, 𝛽1[𝑅𝑚𝑡 − 𝑅𝑓𝑡]is the market risk factor, 𝛽2(𝑆𝑀𝐵𝑡)
is the size factor and 𝛽3(𝐻𝑀𝐿𝑡) is the value factor. The time-series alphas are reported in the
descriptive statistics.
𝑅𝑖𝑡− 𝑅𝑓𝑡 = 𝛼𝑖 + 𝛽1[𝑅𝑚𝑡 − 𝑅𝑓𝑡] + 𝛽2(𝑆𝑀𝐵𝑡) + 𝛽3(𝐻𝑀𝐿𝑡) (9)
The alphas that are found in the time series will be used to determine the effect of the managerial proxies on the performance. This will indicate if return that is not explained by the model can be explained by the managerial proxies. Equation 10 shows this regression, where the alpha is the dependent variable, and the relative bonus, average salary increase and stock holdings by management are the independent variables. It should be noted that the constant in this regression is denoted by a, to avoid confusion. The three independent
variables are the proxies as calculated for the portfolio analysis
𝛼𝑖 = 𝑎𝑖 + 𝛽0𝐵𝑜𝑛𝑢𝑠 + 𝛽1𝑆𝑎𝑙𝑎𝑟𝑦 + 𝛽3𝑆𝑡𝑜𝑐𝑘 (10)
The betas that are found in regression 9 will then be used in a cross-sectional regression as specified by equation 11. Here, a cross-sectional regression is done for each month. This allows us to determine the reward for exposure to the various risk factors per asset. This regression will be done for each month, using monthly excess returns. Taking the average constant of all the regressions allows us to determine the cross-sectional alpha for each asset which is denoted by λ𝑖. Both the steps in the Fama-Macbeth procedure will be conducted for the different holding periods.
𝑅̅ − 𝑅𝑖 ̅̅̅ =𝑓 λ𝑖+λ1∗𝛽̂ +1 λ2∗𝛽̂ +2 λ3∗𝛽̂ (11) 3
These cross-sectional alphas can then be used as a dependent variable in a regression on the managerial proxies. For each firm, the average alpha will be regressed on the constructed managerial proxies. This will allow us to determine the effect of the managerial proxies on the excess return that is not based on the reward for the existing risk factors.
λ𝑖 = 𝑎𝑖 + 𝛽0𝐵𝑜𝑛𝑢𝑠 + 𝛽1𝑆𝑎𝑙𝑎𝑟𝑦 + 𝛽3𝑆𝑡𝑜𝑐𝑘 (12)
Finally the procedure will be repeated using the Carhart four-factor model as proposed by Carhart (1997). Equations 13 and 14 show the relevant formulas where 𝑏4(𝑀𝑂𝑀𝑡) is the
18
𝑅𝑖𝑡− 𝑅𝑓𝑡 = 𝛼𝑖+ 𝑏1[𝑅𝑚𝑡− 𝑅𝑓𝑡] + 𝑏2(𝑆𝑀𝐵𝑡) + 𝑏3(𝐻𝑀𝐿𝑡) + 𝑏4(𝑀𝑂𝑀𝑡) (13) 𝑅̅ − 𝑅𝑖 ̅̅̅ =𝑓 λ𝑖+λ1∗𝛽̂ +1 λ2∗𝛽̂ +2 λ3∗𝛽̂ +3 λ4∗𝛽̂ (14) 4
Hypotheses
I expect portfolios that combine a quantitative measure such as PE or POE and a qualitative measure such as relative bonus, salary increase or stock holdings to have better returns than the control group. Furthermore, I expect the stock holding effect to be the strongest predictor of firm performance. To determine if the additional return can be attributed to the Fama-French factors equations 10 and 12 will be used. The coefficients 𝛽0,1,2 represent the effect the respective managerial proxies have on the outperformance of the assets. Since the literature does not offer a clear consensus on the effect of the managerial proxies on firm performance the coefficients will be tested using a two-tailed t-tests as specified in equation 15, 16 and 17. 𝐻0: 𝛽0 = 0 𝐻1: 𝛽0 ≠ 0 (15) 𝐻0: 𝛽1 = 0 𝐻1: 𝛽1 ≠ 0 (16) 𝐻0: 𝛽2 = 0 𝐻1: 𝛽2 ≠ 0 (17) Sensitivity analysis
19
In summary, the analysis will be done in three parts. First, this study will analyze the performance of the various portfolios constructed using the value investing framework and the compensation measures. Next, the Fama and French three-factor model will be used to determine the alphas for each asset. Using the Fama-MacBeth procedure, both a time-series alpha and a cross-sectional alpha will be determined per asset. A cross section multiple linear regression on the short-, medium-, and long-term alphas of the stocks is done to determine the effect of the managerial proxies. Furthermore, the regressions specified rely on certain assumptions that will be tested for robustness.
Data
To conduct this study, three main datasets needed to be distinguished. First, there was the fundamental data for each firm on which the quantitative measures are based. Second, there was the compensation data for the respective firms, on which the qualitative proxies are based. Finally, there was the stock data for the sample during the sample period. The stock data will be used to measure portfolio performance. The data will be extracted using Wharton Research Data Services. This service allows access to the Compustat Capital IQ and Compustat ExecuComp databases. All the required data can be found in these databases. Only stocks that are in the S&P 500 on the first of January in a given year are considered. To begin the stock selection process, the S&P Compustat Capital IQ database was used to determine the S&P 500 constituents for each given year. The sample period chosen was from 1992 to 2019. This was for two reasons. First, the compensation data available in the ExecuComp database is only available from 1992 onward. Second, the sample period allows 27 years of data to be collected. This ensures that about three business cycles have taken place during the sample period, which greatly reduces any time period bias.
The fundamental data was obtained using the Compustat Capital IQ database. This database offers publicly available information about the company, mainly from annual reports. Information collected included, but was not limited to, the net earnings, depreciation, operational income, total market cap, total assets, total liabilities, net working capital, and capital expenditures. This data will be used the calculation of owner earnings as described in the methodology.
20
the CEO and CFO. After dropping the excess board members, an average of 2.98 executives per firm remained. Based on the data for these board members, the relative bonus, annual salary increase, and stock holdings were calculated as averages for each firm/year combination.
Finally, the stock data was collected from the Compustat CapitalIQ database. The Monthly Security Database offers monthly security prices for all the firms in our sample. Furthermore, the adjusted closing prices for each month are available, in which dividend and stock split effects have been accounted for. Using these prices, the returns for up to 10 years for each firm have been calculated, beginning at the various chosen starting points. Furthermore, the variance for the stock returns for each firm has been collected based on their monthly stock prices. If shares were delisted during the holding period, it was assumed that the return was 0 in the following years. This caused diminishing returns for longer holding periods, which will be addressed in the discussion. This paper assumes that the portfolios were created on the first trading day of the year. Thus, for each firm, an observation is made for each year. The moment of this observation is the first trading day of the year, coupled with the fundamental data of the year that had just ended. The compensation data, which is also based on the last year, is linked as well. It must, however, be noted that, in practice, these numbers are only available to the public several months later. The portfolios in this study, therefore, have a slight information asymmetry compared to an actual investor making his decision in this particular moment. As a final step, the stock price data was added. This allowed us to link returns for up to 10 years for each unique firm/year combination. Furthermore, the current stock price at the moment of the portfolio composition was determined. This allowed for an accurate calculation of a PE and POE that were relevant at that time.
Finally, the factors for the Fama and French three-factor model were downloaded from the Kenneth French website. These factors were used to determine the alpha for the assets and portfolios. The momentum factor was also collected from this website, which will be used in the robustness tests. The factors were collected on a monthly basis.
Descriptive statistics
21
differences between this sector and the rest of the sample, which caused outliers in the data. 292 observations were classified as utility based on the NAICS code. In table 1, the various portfolios are outlined, along with their total number of constituents and various return metrics including the FF3 alpha and Sharpe ratio.
22
Table 2 - Descriptive statistics per portfolio
Descriptive statistics per portfolio
Average annual return FF3 alpha Sharpe ratio
Portfolio
Number
23
Results
This section will outline the empirical results of my research. First the findings of the portfolios in terms of holding returns will be presented. Following this the regression analysis is described. Finally, a robustness test will be done to ensure the result is robust to the momentum effect. Furthermore, the final regression will also be conducted using only the CEO and CFO of the firm, as they are expected to bear the most predictive power in their compensation packages. I find the portfolio results to be mostly consistent throughout the methods, with POE, PE and stock holdings all being positive indicators. For different holding periods, other metrics turn out to be reliable indicators. Furthermore, contrary to my hypotheses I found the annual salary increase to often have a negative effect on the portfolio return. In the regressions, the only proxy that was found to significantly explain an alpha was the relative bonus. This was found to have a negative effect. Furthermore, when using only the CEO and CFO in the analysis, stock holdings were found to have a positive effect on the achieved alpha. The results imply that the excess return that we observe in the portfolios are largely explained by the risk factors. Therefore, no violation of the EMH is found, and no superior investment strategy can be formed upon managerial proxies. This implies that the additional performance that is observed can be attributed to additional risk. In describing the results in more detail, I will mainly focus on three time windows: the short-, mid- and long-term. Unless mentioned otherwise, I will base these findings on the 1-, 5- and 10-year returns or alphas.
Portfolio analysis
This section gives the results of the portfolio analyses. As can be seen in table 3 the best total return over the entire holding period was achieved by portfolio 12 which was based on the PE and stock holdings. This is also true for the short- and mid-term. For all three holding periods an investors would have been best of investing based on a combination of PE and stock holdings. Other portfolios see differing results throughout the various holding periods. This implies that some measures are better for short-term returns, while others are better suited for long-term horizon investments.
24
high 1-year results, but are no reliable indicator in the long-term. For the salary proxy, no clear results can be observed. In all holding periods incorporating annual salary increase in the stock selection process does not add value to the portfolio performance. The stock holding proxy outperformed the control portfolio on its own in all holding periods. Furthermore, adding the stock proxy to the quantities portfolios adds value in each holding period.
In the short-term, the PE + Stock holding portfolio realized the highest return. Portfolio 12 had a short-term holding return of 20.0% while the benchmark portfolio had a holding return of 15.9%. As this return was significantly higher than the benchmark we can conclude that the combination of PE and stock holding criteria led to abnormal results. In the mid- and long-term, this portfolio is still the best performing stock portfolio. The 10-year return of portfolio 12 was found to be 133.6%, while the benchmark realized 117.6% in returns. Furthermore, it is found that PE and stock holdings on their own perform better than their counterparts in isolation. Stock holdings are also found to improve the performance when combined with the POE ratio, albeit slightly less. Finally, it is found that selecting stocks based on annual salary increase and relative bonus does not lead to higher portfolio returns. This is contrary to what I hypothesized based on literature. We conclude the variables that seem to indicate high future stock returns to be POE and stock holdings.
Table 3 - Total Holding Returns
Total Holding Returns with reinvested dividends Holding period in years
Portfolio 1 2 3 4 5 6 7 8 9 10 (1) All stocks 1.159 1.279 1.403 1.528 1.654 1.753 1.852 1.955 2.063 2.176 (2) POE 1.153 1.276 1.390 1.487 1.608 1.718 1.819 1.939 2.048 2.152 (3) PE 1.177 1.302 1.433 1.565 1.701 1.807 1.902 2.013 2.129 2.223 (4) Bonus 1.176 1.295 1.413 1.536 1.659 1.753 1.847 1.931 2.027 2.130 (5) Salary 1.159 1.276 1.394 1.512 1.620 1.722 1.812 1.911 2.020 2.143 (6) Stock 1.174 1.304 1.456 1.580 1.696 1.780 1.874 1.960 2.063 2.216 (7) POE + Bonus 1.165 1.285 1.393 1.488 1.592 1.702 1.822 1.928 2.036 2.132 (8) POE + Salary 1.150 1.273 1.384 1.473 1.585 1.691 1.793 1.901 2.016 2.139 (9) POE + Stock 1.158 1.302 1.460 1.545 1.634 1.761 1.862 2.001 2.125 2.293 (10) PE + Bonus 1.195 1.319 1.443 1.574 1.700 1.797 1.892 1.976 2.077 2.159 (11) PE + Salary 1.176 1.300 1.426 1.549 1.666 1.767 1.871 1.983 2.100 2.209 (12) PE + Stock 1.205 1.350 1.519 1.658 1.791 1.881 1.955 2.054 2.183 2.336
25
bonuses as a selection criteria only improved the return in the short-term. High annual salary increases never increased the portfolio performance. This implies that high salary increases are a negative indicator of future stock performance. According to the portfolios, an investors with a long investment horizon would have been best of investing in low PE high stock portfolios.
Regression analysis
To determine if the additional returns in the portfolios can be attributed to additional risk or a superior investment strategy, the Fama and French three-factor model is used. The full regression results can be found in appendix III. In table 4, the time-series alphas obtained using the model are regressed on the managerial proxies. This alpha can be interpreted as the excess return that is not explained by the existing risk factors in the model. As can be seen in table 4, a negative significant coefficient is obtained for the bonus proxy for the 5- and 10-year alpha. This is consistent with our findings regarding the portfolios, where we observed diminishing returns for bonus portfolios in the mid- and long-term. The low R-squared indicates that only a very small part of the excess return is explained by our model. This implies that the managerial proxies do not add value to the Fama and French three-factor model. This is contrary to our finding that stock holdings predicted high returns. Apparently the excess return the stock holdings provide can be explained by one of the risk factors in the three-factor model.
Table 4 - Regression results for the Fama and French alpha on the managerial proxies
1-year 5-year 10-year
Fama French
alpha French alpha Fama French alpha Fama
Bonus -0.000 -0.001*** -0.000*** (0.000) (0.000) (0.000) Salary 0.000 0.000 -0.000 (0.000) (0.000) (0.000) Stock 2.894 2.015 -1.027 (4.039) (1.306) (1.089) _cons 0.003*** 0.001*** 0.002*** (0.001) (0.000) (0.000) Obs. 4314 3178 2171 R-squared 0.000 0.008 0.014
Standard errors are in parenthesis
*** p<0.01, ** p<0.05, * p<0.1
26
Table 5 - Regression results for the Carhart alpha on the managerial proxies
1-year 5-year 10-year
Carhart
alpha Carhart alpha Carhart alpha
Bonus -0.000 -0.001*** -0.000*** (0.000) (0.000) (0.000) Salary -0.000 0.000 -0.000 (0.000) (0.000) (0.000) Stock 5.983 1.722 -0.675 (4.789) (1.260) (1.103) _cons 0.003*** 0.002*** 0.002*** (0.001) (0.000) (0.000) Obs. 4314 3178 2171 R-squared 0.000 0.008 0.012
Standard errors are in parenthesis
*** p<0.01, ** p<0.05, * p<0.1
The cross sectional alphas as obtained by the Fama-MacBeth two-step regression are the dependent variable for the results in table 6. The two-step regression provided us with alphas for each month in the sample. The average of these alphas was taken as a dependent variable using the managerial proxies as regressors. The alpha here can be interpreted as the reward an investors could expect controlling for the risk rewards of the exist risk factors. Here we observe similar results to the time-series regression. The only significant variable is the bonus proxy in the short-term. No clear relation between the managerial proxies and outperformance of the Fama-French model can be found. Therefore, we conclude that the abnormal returns as observed in the portfolio results can be attributed to the existing risk-factors. This implies that the additional performance is not due to a superior investment strategy, but rather a riskier one.
Table 6 - Regression results for the cross-sectional alphas on the managerial proxies
Alpha 1-year holding period Alpha 5-year holding period Alpha 10-year holding period Bonus 0.000** -0.000 -0.000 (0.000) (0.000) (0.000) Salary -0.000 -0.000 -0.000 (0.000) (0.000) (0.000) Stock 1.361 0.673 0.204 (0.415) (0.703) (0.393) _cons 0.014*** 0.006*** 0.003*** (0.000) (0.000) (0.000) Obs. 4387 4387 4387 R-squared 0.005 0.001 0.001
Standard errors are in parenthesis
27
Sensitivity
The process of selecting managers for the proxy was highly arbitrary. To control for this process, the analysis was done using only the CEO and CFO. Here we assume that the CEO and CFO are the two highest paid members of the management team. In our data sample it was denoted for some firms, in which the assumption held true. It is interesting to see this effect as these two managers have the most influence on capital allocation and overall strategy. In table 7 and 8 the results for the time-series alphas can be found using both the Fama-French three-factor model as the Carhart four-factor model. We observe similar results for the bonus and salary proxy as in the original results. However, a positive coefficient for the stock holdings can be observed in the short-term. This implies that the stock holdings of the CEO and CFO are a reliable indicator of stock performance in the short-term. Moreover, it implies that part of the excess return of the Fama and French model is explained by the stock holdings. This is a significantly different result than the original model we estimated. This could be explained by the fact that stock holdings were diluted when using all available managers for analysis. In our sample, the most stock holdings were concentrated in the CEO and CFO. It will be interesting to see if the stock holdings of the CEO and CFO can add to the Fama and French model as an additional factor in future research.
Table 7 - Regression results for the Fama-French alphas on the managerial proxies using only the CEO and CFO
1-year 5-year 10-year
Fama French alpha Fama French alpha Fama French alpha Bonus (CEO + CFO) -0.001 -0.001*** -0.001***
(0.001) (0.000) (0.000)
Salary (CEO + CFO) 0.000 -0.000 -0.000
(0.000) (0.000) (0.000)
Stock (CEO + CFO) 14.183* 4.117 -2.054
(8.131) (2.784) (2.316)
_cons 0.003*** 0.001*** 0.002***
(0.001) (0.000) (0.000)
Obs. 7966 6002 4161
R-squared 0.000 0.005 0.010
Standard errors are in parenthesis
28
Table 8 - Regression results for the Carhart alphas on the managerial proxies using only the CEO and CFO
1-year 5-year 10-year Carhart
alpha Carhart alpha Carhart alpha Bonus (CEO + CFO) -0.003** -0.001*** -0.001***
(0.001) (0.000) (0.000)
Salary (CEO + CFO) -0.000 -0.000 -0.000
(0.000) (0.000) (0.000)
Stock (CEO + CFO) 82.969*** 3.242 -0.339
(13.061) (2.666) (2.317)
_cons 0.002* 0.002*** 0.002***
(0.001) (0.000) (0.000)
Obs. 7966 6002 4161
R-squared 0.005 0.006 0.009
Standard errors are in parenthesis
*** p<0.01, ** p<0.05, * p<0.1
Conclusion
It is found that stock holdings are a reliable indicator for future stock performance within a value investing framework. Positive results were found for portfolios based on stock holdings by upper management. Furthermore, combining value metrics such as PE and POE with the stock proxy found to increase the stock performance in the short-, medium- and long-term. In terms of performance, the PE + stock portfolio had the highest returns. In terms of the Sharpe ratio, which measures return as a function of risk, no clear winning indicator could be distinguished. This indicates that the stock holdings do offer more returns, but also offer an additional amount of risk.
This is confirmed by the regression analysis. Here we found no positive relation between any of the managerial proxies and the alpha as predicted by the Fama-French model. Moreover, the result was robust to the momentum effect when using the Carhart four-factor model to estimate the alphas. Comparable results were achieved when using the cross-sectional alphas as obtained using the Fama-MacBeth procedure. In the sensitivity analysis however, it was found that the stock holdings do provide a positive coefficient in the short-term when only considering the stock holdings of the top two managers in terms of compensation. This indicates that the concentration of stock holdings between management is important for a reliable indicator.
29
however due to additional risk. That is not to say that the strategy can not be used in practice. Risk-seeking value-investors might want to adopt the investment strategy. As is shown in the portfolio results, historically the high stock holdings portfolios did outperform the control group. However, it is unclear if this excess return can be attributed to the superiority of the investment strategy. It may be that the excess return may simply be a compensation for additional risk. The Fama-French alpha that was calculated using the Fama-MacBeth procedure indicated that this was indeed the case. No relation could be found between the unexplained excess return, the FF alpha, and the managerial proxies. Therefore, we can conclude that investing based on managerial proxies and simple valuation metric such as PE do not lead to abnormal returns in risk-adjusted terms.
Limitations
An extension to this research could be adding a factor to the three-factor model. Appending the model with a managerial ownership proxy may add value to the literature. In the sensitivity analysis, a positive coefficient can be found for the stock holdings. As the effect is also clearly positive in the portfolio results, I suggest future research on this proxy. In hindsight, an alternative method to my research could have been to extend the Fama and French three-factor model. Adding the stock holdings as a fourth factor might be able to explain the effect of the selection measure. This could be constructed as the other factors where high- and low stock portfolios are constructed and compared. It would be interesting to see if adding to the Fama and French model yields significant results using the stock holdings by management. The alphas produced by this model could be compared to the alphas of the original model. I recommend that this research is done using the stock holdings by the CEO and CFO, as this tends to be the most complete data and appears to be the most reliable indicator.
The larger a stock portfolio gets, the more its return tends to converge to the mean return of the market. Since a value investor believes he has an edge over the market, he wants to realize larger gains than the market does. Therefore, value investors advocate for focused portfolios rather than diversification. In practice, value investors may only hold a handful of stocks. There the results from the portfolio analysis may deviate from a strategy that might be employed in practice.
30
assumes that the annual numbers for last year are instantly known on the first trading day of the next year. In reality, these figures only become public several months later. Therefore, the portfolios in this paper were constructed with more information than would be available at the time. This information asymmetry might explain part of the higher returns.
31
References
Abowd, J. M. (1990). Does Performance-Based Managerial Compensation Affect Corporate Performance? Industrial and Labor Relations Review, 43(3), 52S-73S.
https://doi.org/10.2307/2523571
Asness, C., Frazzini, A., Israel, R., & Moskowitz, T. (2015). Fact, Fiction, and Value Investing.
The Journal of Portfolio Management, 42(1), 34–52.
https://doi.org/10.3905/jpm.2015.42.1.034
Athanassakos, G. (2012). Value Investing vs. Modern Portfolio Theory. Journal of Business &
Financial Affairs, 1(2), 1–3. https://doi.org/10.4172/bsfa.1000e110
Barkema, H. G., & Gomez-Mejia, L. R. (1998). Managerial Gompensation and Firm Performance: A General Research Framework. Academy of Management Journal, 41(2), 135–145. https://doi.org/10.5465/257098
Basu, S. (1977). Investment Performance of Common Stocks in Relation to Their Price to earnings Ratios: A Test of the Efficient Market Hypothesis. The Journal of Finance, 32(3), 663–682. https://doi.org/10.1111/j.1540-6261.1977.tb01979.x
Bauman, W. S., Conover, C. M., & Miller, R. E. (1998). Growth versus Value and Large-Cap versus Small-Cap Stocks in International Markets. Financial Analysts Journal, 54(2), 75–89. https://doi.org/10.2469/faj.v54.n2.2168
Brickley, J. A., Bhagat, S., & Lease, R. C. (1985). The impact of long-range managerial compensation plans on shareholder wealth. Journal of Accounting and Economics, 7(1–3), 115–129. https://doi.org/10.1016/0165-4101(85)90031-x
Calandro, J., Jr. (2009). Lessons for strategists in Graham & Dodd’s Security Analysis, 6th edition. Strategy & Leadership, 37(2), 45–49. https://doi.org/10.1108/10878570910941226 Capaul, C., Rowley, I., & Sharpe, W. F. (1993). International Value and Growth Stock Returns.
Financial Analysts Journal, 49(1), 27–36. https://doi.org/10.2469/faj.v49.n1.27
Carhart, M. M. (1997). On Persistence in Mutual Fund Performance. The Journal of Finance,
52(1), 57–82. https://doi.org/10.1111/j.1540-6261.1997.tb03808.x
Chan, C., Hamao, Y. A. S. U. S. H. I., & Lakonishok, J. O. S. E. F. (1991). Fundamentals and Stock Returns in Japan. The Journal of Finance, 46(5), 1739–1764.
https://doi.org/10.1111/j.1540-6261.1991.tb04642.x
De Bondt, W., & Thaler, R. (1985). Does the Stock Market Overreact? The Journal of Finance,
40(3), 793–805. https://doi.org/10.1111/j.1540-6261.1985.tb05004.x
Fama, E. F. (1980). Agency Problems and the Theory of the Firm. Journal of Political
Economy, 88(2), 288–307. https://doi.org/10.1086/260866
Fama, E. F., & French, K. R. (1992). The Cross-Section of Expected Stock Returns. The Journal
32
Fama, E. F., & French, K. R. (1993). Common risk factors in the returns on stocks and bonds.
Journal of Financial Economics, 33(1), 3–56. https://doi.org/10.1016/0304-405x(93)90023-5
Fama, E. F., & French, K. R. (1998). Value versus Growth: The International Evidence. The
Journal of Finance, 53(6), 1975–1999. https://doi.org/10.1111/0022-1082.00080
Fama, E. F., & French, K. R. (2015). A five-factor asset pricing model. Journal of Financial
Economics, 116(1), 1–22. https://doi.org/10.1016/j.jfineco.2014.10.010
Fama, E. F., & MacBeth, J. D. (1973). Risk, Return, and Equilibrium: Empirical Tests. Journal
of Political Economy, 81(3), 607–636. https://doi.org/10.1086/260061
Fama, E. F., & French, K.R. (1996). Multifactor Explanations of Asset Pricing Anomalies. The
Journal of Finance, 51(1), 55–84. https://doi.org/10.1111/j.1540-6261.1996.tb05202.x
Fisher, P. A., & Fisher, K. L. (2003). Common Stocks and Uncommon Profits and Other
Writings (2nd ed.). Hoboken, United States of America: Wiley.
Graham, B. (2005). Intelligent Investor. New York, United States: HarperCollins.
Greenwald, B. C. N., Kahn, J., Sonkin, P. D., van Biema, M., & van Biema, M. (2004). Value
Investing. Hoboken, NJ, United States: Wiley.
Guidry, F., J. Leone, A., & Rock, S. (1999). Earnings-based bonus plans and earnings management by business-unit managers. Journal of Accounting and Economics, 26(1–3), 113–142. https://doi.org/10.1016/s0165-4101(98)00037-8
Jegadeesh, N., & Titman, S. (1993). Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency. The Journal of Finance, 48(1), 65–91. https://doi.org/10.1111/j.1540-6261.1993.tb04702.x
Jensen, M. C., & Murphy, K. J. (1990). Performance Pay and Top-Management Incentives.
Journal of Political Economy, 98(2), 225–264. https://doi.org/10.1086/261677
Klarman, S. A. (1991). Margin of Safety. New York, United States: HarperBusiness. Lewellen, W. G. (1969). Management and Ownership in the Large Firm. The Journal of
Finance, 24(2), 299. https://doi.org/10.2307/2325296
Malkiel, B. G. (1962). Expectations, Bond Prices, and the Term Structure of Interest Rates.
The Quarterly Journal of Economics, 76(2), 197. https://doi.org/10.2307/1880816
Malkiel, B. G., & Fama, E. F. (1970). Efficient Capital Markets: A Review of Theory and Empirical Work. The Journal of Finance, 25(2), 383–417. https://doi.org/10.1111/j.1540-6261.1970.tb00518.x
Markowitz, H. M. (1991). Foundations of Portfolio Theory. The Journal of Finance, 46(2), 469–477. https://doi.org/10.1111/j.1540-6261.1991.tb02669.x
Piotroski, J. D. (2000). Value Investing: The Use of Historical Financial Statement Information to Separate Winners from Losers. Journal of Accounting Research, 38, 1–41.
33
Rousseau, R., & van Rensburg, P. (2003). Time and the payoff to value investing. Journal of
Asset Management, 4(5), 318–325. https://doi.org/10.1057/palgrave.jam.2240112
Rozeff, M. S., & Kinney, W. R., Jr. (1976). Capital market seasonality: The case of stock returns. Journal of Financial Economics, 3(4), 379–402. https://doi.org/10.1016/0304-405x(76)90028-3
Sappington, D. E. M. (1991). Incentives in Principal-Agent Relationships. Journal of Economic
Perspectives, 5(2), 45–66. https://doi.org/10.1257/jep.5.2.45
Shiller, R. J. (1981). Do Stock Prices Move Too Much to be Justified by Subsequent Changes in Dividends? The American Economic Review, 71(3), 421–436.
https://doi.org/10.3386/w0456
Yee, K. K. (2008). Deep-Value Investing, Fundamental Risks, and the Margin of Safety. The
34 Appendix I: Risk-free rate used in calculations
Table 9 - Risk free rates used to determine the risk-adjusted returns
10-Year US Treasury Rate
Year Average Yield Year Average Yield
1991 7.86% 2006 4.80% 1992 7.01% 2007 4.63% 1993 5.87% 2008 3.66% 1994 7.09% 2009 3.26% 1995 6.57% 2010 3.22% 1996 6.44% 2011 2.78% 1997 6.35% 2012 1.80% 1998 5.26% 2013 2.35% 1999 5.65% 2014 2.54% 2000 6.03% 2015 2.14% 2001 5.02% 2016 1.84% 2002 4.61% 2017 2.33% 2003 4.01% 2018 2.91% 2004 4.27% 2019 2.14% 2005 4.29%
Appendix II: Portfolio results
Portfolio 1: All firms in S&P 500
Variable Obs Mean Std. Dev. Min Max
