1
Combining factor and value investing
Mirroring Warren Buffett; a five-factor model, applied
to the FTSE 100
Abstract
Using a previous study, this paper aims to apply a model that mirrors the performance of investor Warren Buffett to the FTSE100. An outperformance of the FTSE100 is expected. The model is applied and one of the factors, describing the quality of an investment, is not significant.
By going long the predicted 30 best performers and shorting the predicted 30 worst performing stocks, an outperformance of the FTSE 100 with 25.87% is measured.
Bachelor Thesis Economics and Business
Specialization Finance and Organization July 1st, 2014, Amsterdam
Name: Bob Verhagen, 6058310 Thesis Supervisor: V. Malinova
2
1. Introduction
When people, companies, pension funds, governments and many other entities have an excess amount of cash there are, abstractly spoken, two ways to invest the money. Either, an investor assumes a market's return and invests in an index (tracker), or an investor selects a specific investment (fund) and assumes the return of that specific investment (fund). Usually, a selection is based on the higher expectations of the risk-adjusted returns of a (basket of) security('s). Investors, when selecting which securities to invest in, dedicate time and money to research individual securities (financial) background.
One of these investors, Warren Buffett, has an interesting track record of above average risk-adjusted return profiles for the past 30 years. From several of his statements in annual reports and at press releases, his appetite for qualitative companies for a good price has become known. Mirroring his stock picking ability, though, has proved nearly impossible. To mirror his performance, Frazzini, Kabiller and Pedersen (2013) launched a factor model which should copy Buffett's way of picking securities; high quality, low risk and high risk-adjusted return securities. They manage to mirror his risk-return profile, but express they can only do it with the benefits of hindsight; insight in Buffett's theories and principles. Investors, always looking for above average returns, can be expected to want to know if Frazzini, Kabiller and Pedersen (2013) have opened up Warren Buffett's returns to the public. Frazzini, Kabiller and Pedersen (2013) use a dataset with securities from the United States.
This paper will research if investing securities by using the model from Frazzini, Kabiller and Pedersen (2013) in the FTSE 100, during the time period from 01-01-2002 to 31-13-2008 yields an above market return.
In the literature review I will first explain the doctrine of efficient capital markets and the impossibility to beat the market. Investors can choose a risk-return profile along the capital market line, supported by the capital asset pricing model and with the existence of a riskless asset, which can also be shorted.
Questioning the capital asset pricing model and its assumptions, I introduce abnormalities in the pricing of securities, which open up opportunities for investors to outperform the risk-adjusted market return and earn above average returns. This is followed by the rejection of the efficient capital market hypothesis and the joint hypothesis problem. With the support of several studies, the Frazzini, Kabiller and Pedersen (2013) model is constructed to mirror the performance of Warren Buffett.
3
This model is applied to the FTSE 100 market index, to investigate outperformance within the London Stock Exchange. The investments are mirrored from 01-01-2002 until 31-12-2008, to mirror the holding period of a value investor.
2. Literature review efficient capital markets and outperformance
Investors are always looking to outperform the market. Whether or not this is possible, has been up for debate by supporters and opponents of the theory on efficient capital markets. For opponents, Warren Buffett's performance is part of the ‘proof’ of imperfect efficient capital markets. As the founder and manager of Berkshire Hathaway, an investment fund, he has been able to outperform the market over a long period of time (Lowenstein, 2008).
This section reviews the existing literature. First we move from the random walk of the capital markets to the efficient capital markets theory. In the efficient capital markets theory, investors can only choose a risk-return profile along the capital market line. Second, we look at flaws in the efficient market theory and potential ways to outperform the market. Third, we look at how these flaws are used to construct a factor model.
2.1 Efficient capital market theory
As one of the first to provide a solid view on competition and efficiency in (financial) markets, Samuelson (1965) and Mandelbrot (1966) find that past performance cannot be extrapolated to the price of a security in the future. This is supported by Samuelson's (1965) s random walk model (equation (1)).
(1) Xt = µ + Xt-1 + εt Where
- Χ is the price of a stock a time t, - μ depicts de arbitrary drift parameter, - ε depicts random disturbance
In the random walk model, the unspecific pattern of availability of newly available information is assumed. Therefore, the price of a security shows a random walking pattern, as it reacts on new publicly available information, which cannot be predicted, is available to all and is free to find out to investors.
Fama (1970) reviews the efficient capital market theory and, in particular, its three forms: weak, semi-strong and strong. In weak form efficient capital markets, prices reflect
4
historical capital market prices. All the historical financial information from a security is available to all investors. Based on the weak-form efficient capital market theory, it is impossible to beat the market by predicting the future price of stock, based on all stock price information. In the semi strong form of the efficient capital markets theory, prices adjust efficiently to new information which is publicly made available. As soon as new information, relevant to the company, is out in the open, this is reflected in the stock price. In the strong form of the efficient capital market theory all information, both publicly and privately available, is reflected in the price of a security. For the three forms, Fama (1970) shows extensive, yet contradictory, evidence for weak and semi-strong efficient capital markets.
To research investor’s ability to beat the efficient capital markets, academics encounter the problem of joint hypotheses. In a test of the efficient capital market theory, investors are simultaneously testing the efficiency of the market and the asset pricing model, driven by the risk preference of the investor (Sewell, 2011). The right performance of a security is found in an efficient capital market, where the security is rightly priced. If the security is an outperformer and the efficiency of capital markets is assumed, there are two potential cases:
(1) Capital markets are efficient, but the pricing model is wrong. This causes the wrong price for a security to be expected
(2) The capital pricing model is correct, but the capital markets are inefficient. This causes outperformance to be linked to the inefficiency of capital markets.
The problem with these two hypotheses is that to price a security in an efficient market, these two hypotheses would have to hold simultaneously. If outperformance is found, this cannot be linked to either inefficient capital markets or an incorrect pricing model. Therefore, we cannot reject the efficient capital market theory.
The strong-form of efficient capital markets contradicts with Buffett’s ability to outperform the capital markets and find a significant alpha (α), a measure for outperforming the market, for the long period of time that Warren Buffett has been investing (Lowenstein (2008) and Frazzini, Kabiller and Pedersen (2013)).
An investor, unable to use the inefficient capital markets to outperform under the strong-form of the efficient capital market theory, can only assume the level of risk that he is willing to take to match with the corresponding level of return. This is shown by the capital market line (Markowitz, 1959).
Markowitz's (1959) risk return profile corresponds with the uselessness of investing either time or money in stockpicking; an investor researching a stock has no extra return over
5
his chosen risk profile. Tobin (1958) shows that, as a result of the capital market line, an investor only needs to decide on the optimal risky asset portfolio and on the division of his investments between the risky asset portfolio as produced by the capital market line and the riskless asset. Sharp (1964) depicts this as picking the optimal point on the investment opportunity curve, optimizing risk-adjusted return to maximize an investors wealth. Lintner (1965) continues Sharp (1964) under the assumptions that all investors follow the mean-variance portfolio and agree on the true joint distribution, comprised of the value weighted market portfolio. In Lintner's (1965) system, only bearing the risk of the market is rewarded, as risks of individual securities are diversified and the market risk remains. This is reflected in the Sharp ratio, a measure for the risk-adjusted return of a security.
Black, Jensen and Scholes (1972) develop the Capital Asset Pricing Model (CAPM, described in equation (2)
(2) E (R) = R(f) + βi * (E(Rm) – R(f)) Where:
- E (R) is the expected return of a security, - Rf is the riskfree rate,
- βi is the beta of security I, the amount of risk, relative to the market risk, - E(Rm) is the market return
which states that securities have different riskprofiles relative to the marketrisk. Investors choose a certain risk level, relative to the market risk and receive a corresponding return, relative to the market return. If an investor buys a security with a higher risk, relative to the market risk, he can expect a higher return, relative to the market return.
Via the capital asset pricing model, investors choose a security with a specific beta, which corresponds to the risk of a security. In return, investors receive a higher return, whenever they take on more risk from the market. This method suits their risk preference when levered or de-levered to the desired level with the use of the riskless asset as also described by Tobin (1958). An excess in return (out- or underperformance) is then described by α and the resulting CAPM with α is described in equation (3).
(3) E(R) + α = rf + βi * (E(Rm) − Rf) Where:
- E (R) is the expected return of a security,
6
- Rf is the riskfree rate,
- βi is the beta of security I, the amount of risk, relative to the market risk, - E(Rm) is the market return
2.2 The outperformance of capital markets
Questioning the assumption that every investor has all information, and the resulting perfect and strong form of the efficient capital market, Grossman and Stiglitz (1980) argue that gathering and analyzing information has its price and takes time. As the price and time of gathering and analyzing the information has to be reflected in the price of a security, the latter cannot perfectly reflect all the publicly available information. An efficient market would not price in the time and money spend on gathering and analyzing information. Over all, the cost of gathering and analyzing information has to be less than the opportunity cost of a security to gain a positive return.
Black (1972) investigates the assumptions from the CAPM; that any investor can take either a long or a short position in a security of any desired size, that a riskless asset exists and that an investor is allowed to borrow any amount at the riskfree rate. He finds that a model with restricted borrowing is in line with the dataresults as presented by Black, Jensen and Scholes (1972), who find that a capital asset pricing model with restricted borrowing capacities best describes the data. The CAPM with restricted borrowing working best, is partially caused by restrictions on borrowing for pension and mutual funds and results in part of the investment society only being able to take long positions or not being able to fully adjust their adjusted return by the use of leverage up to the desired level of the risk-return profile. In the search for high risk-returns, the premium for long positions favours high risk high beta stocks over low risk low beta stocks (Frazzini and Pedersen, 2013).
2.3 Active investment strategies
Buffett depicted his investment style in 1989 as buying qualitatively well developed companies. In security's investments there are more strategies, all in a way questioning the specification of the CAPM and promoting active investments. In this paper, we will discuss two specific active investment strategies.
Intrinsic value investing is based on the principle that a securities market value will return to its intrinsic value. This intrinsic value is calculated by taking into account a variety of quantitative and qualitative measures, with a potential variety of underlying assumptions.
7
Factor investing is based on the principle that quantitative (merely financial) measures of a company and its market can predict the future market value of a company’s security. This way, specific securities can be selected from the available securities can be selected, based on their loading on specific factors.
2.3 Value investing
In 1934, Graham and Dodd (1934) introduced the concept of value investing. In their book, Graham and Dodd call investors to shift their attention from earnings and earning trends of companies to the compant at large. With value investing, an investor should take a look at the intrinsic value of a company. This way, an investor should buy a stock when it is below its intrinsic value and sell a stock when it is above its intrinsic value. To find the intrinsic value of a company, (Gordon, 1959) proposed a dividend discount model (DDM, see equation (4)).
(4)
𝑃 =
Where:- P is the price of a security
- D1 is the value of next year's dividend - r is the cost of capital
- g is the perpetual growth rate of the dividend
As the DDM assumes constant perpetual growth of the dividend and a constant cost of capital, a more comprehensive model is often used. One more comprehensive way is the Discounted Free Cash Flow model (DCF). Berk and DeMarzo (2007) introduce the DFCF as the present value of all future free cashflows. The DFCF model allows investors to predict the cash flows in his own model, which can be as detailed as desired and takes into account the specific costs of debt and equity of a company. The DFCF bottom-up approach may take more time, effectively requiring a greater difference between current price of a security and its intrinsic value (Grossman & Stiglitz, 1980).
The DFCF has the flaw that, in order to gain a real return, a security’s market price still has to adjust to its intrinsic value. Or, as Warren Buffett calls it: ‘Market price and
intrinsic value often follow very different paths – sometimes for extended periods – but eventually they meet’ (Berkshire Hathaway, 2013).
8
2.4 Buffett’s outperformance
Warren Buffett and his company Berkshire Hathaway have an interesting trackrecord. From 1976 to 2006 Berkshire Hathaway exceeded the S&P 500 in 27 of the 31 years, on average with 11.14% (Martin and Puthenpurackal, 2008). Berkshire Hathaway’s manager Warren Buffett has been one of the most successfull investors in the 20th century (Frazzini, Kabiller and Pedersen (2013) and Lowenstein (2008)). Frazzini, Kabiller and Pedersen (2013) find that Berkshire Hathaway has the highest Sharpe ratio of all U.S. mutual funds. According to Frazzini, Kabiller and Pedersen (2013), Buffett’s strong performance has two main reasons.
First, Warren Buffet picks high quality companies for a good price, instead of regular companies for a bargain. Berkshire Hathaway invests in both public and private companies. Frazzini, Kabiller and Pedersen (2013) find that Buffett’s great stockpicking abilities are more important for Berkshire Hathaway’s performance as an investment fund than his management skills in private companies. He has an average excess return of 11.8% of the public investments against a 9.6% average excess return from private holdings. Warren Buffett has made several statements in the past, describing his investment strategy, such as: ‘It’s far better to buy a wonderful company at a fair price than a fair company at a wonderful
price.’ (Berkshire Hathaway Inc., Annual Report 1989).
Second, Warren Buffet uses both embedded and cheap leverage. Embedded leverage, on average 1.6-to-1, works as lever to take on more risk in relatively riskless assets from qualitative companies (Frazzini and Pedersen (2012) and Frazzini, Kabiller and Pedersen (2013)). Berkshire Hathaway’s insurance companies provide the free float, upfront paid insurance premiums, as cheap leverage to the investment company (Frazzini, Kabiller and Pedersen, 2013).
2.5 Factor investing
Factor investing was originally introduced by Ross (1976) in his arbitrage pricing theory as a multi-factor model. As the factors in Ross’s multi-factor model could be determined again and again (as opposed to the factors in the CAPM). Each factor in the model holds a combination between risk and return, together making up for the risk-return profile of a single security of a company.
Banz (1981) finds that the securities of small firms tend to outperfom the securities of larger firms when controlling for risk-adjusted returns within a sample from the New York Stock Exchange. This effect, though, is not very stable over time. This size factor, as
9
described by SMB in equation (5), is one of the two factors used by Fama and French (1993) in their three factor model.
(5) Ri – Rf = α + β1*MKTt + β2*SMBt + β3*HMLt + εt Where:
- Ri is the return of security i
- Rf is the return of the risk-free asset - MKT is the market return
- SMB is the factor Small-minus-Big - HML is the factor High-minus-Low - ε is a random disturbance
The book-to-market equity factor, as introduced by Fama French (1993), is depicted by HML in equation (5). The book-to-market factor depicts the book value of a firm’s equity, relative to its market value. The effect of the book-to-market equity factor has been described by Stattman (1980), Rosenberg, Reid and Lanstein (1985) and Chan, Hamao and Lakonishok (1991). Rosenberg, Reid and Lanstein (1985) find that firms with a high book-to-market value have a higher risk-adjusted return than firms with a low book-to-market value. Chan, Hamao and Lakonishok (1991) support this hypothesis for the Japanese market.
Carhart (1997) extends the model as presented by Fama and French (1993) by finding that CAPM has its flaws in different momentums of the stock market. These flaws were presented by Jegadeesh and Titman (1993), when they researched a strategy based on the momentum of the market for securities, taking into account whether the market has risen in the past twelve months or not. Jegadeesh and Titman (1993) proof that well performing stocks from the past will outperform stocks with a bad track record in the upcoming three to twelve month period. That is why Carhart (1997) adds a fourth factor to the three factor model from Fama and French (1993), which is described by UMD in equation (6).
(6) Ri – Rf = α + β1*MKTt + β2*SMBt + β3*HMLt + β4*UMDt + εt Where:
- Ri is the return of security i
- Rf is the return of the risk-free asset - MKT is the market return
10
- HML is the factor High-minus-Low - UMD is the factor Up-minus-Down - ε is a random disturbance
This factor is higher than one if a security follows an upward trend after having had an upward momentum in the past and is lower than one if a security follows a downward trend after having had a period of upward momentum.
2.6 Combining factor and value investing
The previous four strategies were factors which are found relatively easy in available financial data. When doing a comprehensive security data analysis, the search for quality by value investors is harder. Ever since Graham (1949) introduced value investing, investors have tried to take quality into account, when valueing a security. Combining the the qualities of both, the easyness of factor investing and the comprehensiveness of value investing, is harder.
2.6.1 Defining quality in a factor model
To extend the four factor model presented at (6), Assness, Frazzini and Pedersen (2013) capture quality in a factor, existing of a variety of measures. They measure profit, growth, safety and the payout of a security through a variety of accounting statistics.
Novy-Marx (2012) finds that gross profitability significantly explains the return of securities. Novy-Marx (2012) also subscribes to Buffett’s 1989 annual report message by saying: ‘Buying high quality assets without paying premium prices is just as much value
investing as buying average quality assets at discount prices’. Basu (1977) finds that firms
with a low price/earnings (P/E) ratio outperform firms with a high P/E ratio. Baker and Wurgler (2002) find that firms that repruchase stock, outperform companies that do not repurchase stock. Mohanram (2005) finds that a firm that shows high growth, outperforms a firm that shows low growth. George and Hwang (2010) show that firms with a low debt to equity ratio or low leverage tend to outperform firms with higher leverage. Ohlson (1980) shows that firms with a higher leverage show a greater risk for bankruptcy and therefore, firms with lower leverage tend to show higher risk-adjusted returns. Sloan (1996) shows that firms with high accruals do not see their accruals fully reflected in their security prices. Therefore they underperform relative to firms with lower accruals. Richardson, Sloan, Soliman and Tuna (2005) find that unreliable accruals over time lead to unreliable earnings
11
performance and therefore, firms with unreliable accruals tend to underperform when compared with their peers.
Assness, Frazzini and Pedersen (2013) combine these accounting measures into a factor, that adds QMJ to equation (6), resulting in equation (7). In this model, the factor QMJ is higher than one if a company has a higher quality, relative to its price, and is lower than one if the company has a lower quality, relative to its price.
(7) Ri – Rf = α + β1*MKTt + β2*SMBt + β3*HMLt + β4*UMDt + β5*QMJt + εt Where:
- Ri is the return of security i
- Rf is the return of the risk-free asset - MKT is the market return
- SMB is the factor Small-minus-Big - HML is the factor High-minus-Low - UMD is the factor Up-minus-Down - QMJ is the factor Quality-minus-Junk - ε is a random disturbance
2.6.2 Defining cheap leverage in a factor model
To extend the five factor model presented at (7), Frazzini, Kabiller and Pedersen (2013) add a factor for the cheap leverage used on low risk and low beta stocks.
As presented by Black, Jensen and Scholes (1972), low beta stocks tend to outperform high beta stocks in risk-adjusted returns. Frazzini and Pedersen (2013) use this knowledge to construct a factor which is long low-beta stocks and short high-beta stocks, which they call betting against beta (BAB). This factor has an above average risk-adjusted return and is described by the BAB factor in equation (8). The betting against beta factor invests is long low-beta securities and shorts high-beta securities.
(8) Ri – Rf = α + β1*MKTt + β2*SMBt + β3*HMLt + β4*UMDt + β5*QMJt + β6*BABt + εt
12
2.7 FTSE 100
Frazzini, Kabiller and Pedersen (2013) use data from the US stock market, from 1976 to 2011 to measure Warren Buffett’s performance. For this time period, they find an average excess return over the T-Bill rate of 6.1%.
The FTSE 100 is comprised of the 100 stocks with the highest market capitalization with a primary listing on the London Stock Exchange. The FTSE was introduced in January 1984 and was created to give a better insight into market sentiment, when compared to the existing FT Index of 30 blue-chip stocks. From January 1st 2002 until the 31st of December 2008, the FTSE 100 (London Stock Exchange, 2014) fell from 5323.8 to 4561.8, a 14.3% drop, while the S&P 100, comprised of the 100 stocks most often traded in the United States, fell from 588.98 to 431.54, a drop of 26.7%.
2.8 Conclusion
The research highlighted in this literature review has shown that the capital markets are not perfectly efficient and several strategies have been highlighted to outperform the market, as Warren Buffett has done over a long period of time. The strategies can be used as factors in a factor model. The factor model that Frazzini, Kabiller and Pedersen (2013) use to mirror Warren Buffett, has outperformed the US stock market over the 1976 to 2011 time period. In the period from the 1st of January 2002 until the 31st of December 2008, the FTSE 100 dropped less in value than the S&P 100.
3. Outperforming the FTSE 100with a factor model
The aim of this paper is to apply the model as presented by Frazzini, Kabiller and Pedersen (2013) to the FTSE 100.
For this study, we focus on the FTSE 100 index of stocks from the London Stock Exchange. The FTSE 100 is comprised of the 100 largest stock (by market capitalization) as listed on the London Stock Exchange. We use the FTSE 100 to mirror the performance of an investor on the UK stock market. The investment period, from 01-01-2002 until 31-12-2008 is chosen to ensure:
- A period when there is sufficient data for the FTSE 100 available. The data is both available because the FTSE 100 exists (since 1984) and the data programs have data on it.
- A holding period long enough to enable the price of a stock to move to the intrinsic value. - A period which is not directly in a (financial) crisis such as the credit crunch, which started in 2008, or the dot.com bubble, which happened in the late 1990's.
13 Based on the model, 30 stocks will be predicted to perform best in the upcoming seven year time period and 30 stocks will be predicted to perform the worst during the upcoming seven year time period.
The predicted 30 best performing stocks are then bought as part of a fictive portfolio, weighted equally in market value at the time of acquisition, and held for the holding period of seven years. The predicted 30 worst performing stocks are bought with borrowed money (shorted), weighted equally in market value at the time of acquisition, and will be resold at the end of the holding period, while the borrowed money has to be repaid. The total return of the portfolio is determined by the combination of both the long and the short portfolio. This Buffett- portfolio will then be compared with the return of the FTSE 100 overall for the same time period.
Based on the literature review, outperformance of the FTSE 100 is expected, in line with the outperformance by the model from Frazzini, Kabiller and Pedersen (2013).
This paper can contribute to the existing literature by identifying a new market, using a new set of data, that holds opportunity for investors applying the Warren Buffett model to the UK stock market. Also, identifying quality of a stock in the UK stock market as a significant to the stock price contributes to the existing literature.
Hypotheses
To find out if the basket of stocks picked from the FTSE 100 by the Frazzini, Kabiller and Pedersen (2013) model, as presented in equation (8) would have been able to outperform the overall index, hypotheses (1) is tested.
Hypotheses (1)
H0 : Using the Frazzini, Kabiller and Pedersen (2013) model (equation (7)), an investor cannot outperform the FTSE 100 index by selecting a basket of individual stocks
Accepting H0 means the stocks picked by the Frazzini, Kabiller and Pedersen (2013) model from the FTSE 100 at the 1st of January 2002 did not significantly outperform the FTSE 100 Index after the holding period of 7 years, at the 31st of December 2008.
Also, the significance of the quality factor for a stock in the FTSE 100 is tested.
Hypotheses (2)
H0 : The factor Quality minus Junk (QMJ) is not significant for stocks in the UK stock market.
14
Accepting H0 means that the factor QMJ cannot significantly load on the returns of UK stocks.
Model
The Frazzini, Kabiller and Pedersen (2013) model is used as it introduces quality and betting against beta into a factor model to mirror Warren Buffett. The idea of the model is that it is relatively easy to use
The model (equation (8)) consists of six factors, constructed with data from the capital markets.
(8) Ri – Rf = α + β1*MKTt + β2*SMBt + β3*HMLt + β4*UMDt + β5*QMJt + β6*BABt + ε In (8), MKT is equal to the return of the market.
The factors SMB (small minus big) and HML (high minus low) are constructed simultaneously by dividing all stocks into three BE/ME groups and two market capital groups (see Table (1) in the appendix). The construction of the SMB and HML factors is as follows: - SMB= 1/3 (Small Value + Small Neutral + Small Growth) – 1/3 (Big Value + Big Neutral + Big Growth).
- HML= 1/2 (Small Value + Big Value) – 1/2 (Small Growth + Big Growth).
UMD, up minus down, is equal to: equal weight average of firms with highest 30% past year return - equal weighted average of firms with the lowest 30% past year return QMJ, quality minus junk, is equal to the return of the 30% highest quality stocks – return of the 30% lowest quality stocks. Where quality is determined by:
- Quality = z(Profitability + Growth + Safety + Payout) - Profitability = z (zGPOA + zROE + zROA + zNI) Where:
- GPOA = gross profit over assets - ROE = return on equity
- ROA = return on assets - NI = Net income
- Growth = z (z∆GPOA + z∆ROE + z∆ROA + z∆NI) Where:
15
- ∆ROE = the difference in return on equity between t and t-1 - ∆ROA = the difference in return on assets between t and t-1 - ∆NI = the difference in net income between t and t-1
- Safety = z (zBAB + zIVOL + zLEV + zBankruptcy + zEVOL) Where:
- BAB = the beta of a stock
- IVOL = the idiosyncratic volatility of a stock - LEV = leverage ratio of a stock
- Bankruptcy = the bankruptcy risk of a stock - EVOL = the return on equity volatility - Payout = z (zEISS + zDISS + zNPOP) Where:
- EISS = equity issuance in the previous year - DISS = debt issuance in past year
- NPOP = net payout over profits
BAB = (𝑟 − 𝑟 ) − (𝑟 − 𝑟 )
Where:
- βL is the average beta of the stocks with the 50% lowest beta - βH is the average beta of the stocks with the 50% highest beta - RL is the return on stocks with a low beta
- RH is the return on stocks with a high beta
4. Descriptive statistics
First, the constituents of the FTSE 100 at the 1st of January 2002 are retrieved from Thomson Reuters Datastream. Datastream provides a list of all the 100 constituents names at 01-012002 in an Excel file. See appendix 1.2 for the list of constituents. For these companies, the data points are retrieved as in Buffett's Alpha.
The price year 1 reflects the price of a stock at 01-01-2002, which is on average £12,54 and has a standard deviation of £22,35.
16
The price year -1 reflects the price of a stock at 01-01-2001, which is on average £13.97 and has a standard deviation of £22,80. The price of stocks has, on average, risen from 01-01-2001 to 01-01-2002.
The book equity, the value of equity at the balance sheet of the company at 01-01-2002, is on average £7495,49 million, with a standard deviation of £18048,05 million.
The market equity, equal to the number of shares outstanding times the shareprice, is at 01-01-2002 £20138,99 million, with a standard deviation of £43678,85. The book equity divided by the market equity is therefore equal to 0,47.
The gross profit over assets, the gross profit of a company in the financial year 2001, divided by the assets at 01-01-2002, is on average 6,94%, with a standard deviation of 9,04%.
The return on equity, equal to the net income divided by the total equity, is on average 60%, with a standard deviation of 431,98%. This high number is partially caused by one outlier. When controlling for this one outlier, the return on equity is 16,24% with a standard deviation of 29,36%.
The return on assets, the net income divided by the total assets, is equal to 3,91%, with a standard deviation of 4,83%.
Cash flow over assets, the cash flow from operations (EBITDA, earnings before interest, taxes, depreciations and amortizations) divided by the assets is on average 7,23%, with a standard deviation of 8,69%.
The EBITDA margin, equal to the earnings before interest, taxes depreciations and amortizations divided by the total revenue, is on average 21,48%, with a standard deviation of 15,38%.
The gross profit over assets growth, equal to the growth in gross profit over assets from 01-01-1997 to 01-01-2002, is on average -77,82%, with a standard deviation of 48,07%. Overall, the FTSE 100 declined by 23% in the period from 01-01-1997 until 01-01-2002.
The return on equity growth, equal to the growth in return on equity from 1997 to 01-01-2002, is on average 46%, with a standard deviation of 607%.
The return on assets growth, equal to the growth in the return on assets from 01-01-1997 to 01-01-2002, is on average -35,70%, with a standard deviation of 105,00%.
The cash flow over assets growth, equal to the growth in cash flow over assets from 01-01-1997 to 01-01-2002, is on average -25,15%, with a standard deviation of 161,10%.
The EBITDA margin growth, the growth in EBITDA margin from 01-01-1997 to 01-01-2002, is on average -89,89%, with a standard deviation of 32,78%.
17
The idiosyncratic volatility, the unsystematic or firm specific risk of a stock, at 01-01-2002 is on average 36,92, with a standard deviation of 23,47.
The leverage ratio, the percentage of assets financed by debt, at 01-01-2002, is on average 41,46%, with a standard deviation of 32,87%.
The bankruptcy risk, the potential of a company to go bankrupt, at 01-01-2002, is on average 0,58, with a standard deviation of 0,43.
The return on equity volatility, the volatility movements of the return on equity of a stock, over the financial year 2001, is on average 9,72 with a standard deviation of 13,24. The equity issuance, the amount of equity issued by a company in the financial year 2001, is on average £60,96 million, with a standard deviation of £405,65 million.
The debt issuance, the amount if debt issued by a company in the financial year 2001, is on average £496,35 million, with a standard deviation of £2421,61 million.
The payout to shareholders, the amount of money paid out to the shareholders from 01-01-2001 to 01-01-2002, is on average £232,26 million, with a standard deviation of £2039,10 million.
5. Results
First, we will take a look at the results of the estimation. Based on the estimation results, 30 stocks are selected. The selected stocks receive a virtual investment at 01-01-2002, in an equal weight, and are to be kept for the holding period of 7 year, until 31-12-2008.
For the Frazzini, Kabiller and Pedersen (2013) model (as presented at (8)), the statistics in table (1) are found.
Table (1)
This table reports the factors from the Frazzini, Kabiller and Pedersen (2013) model, as applied to the FTSE 100 data. The factors are loaded for the returns from the year 01-01-2001 to 01-01-2002. The dependent variable in the model is the return on a stock Ri.
SMB -0.026 **
HML -0.004***
UMD -0.176
QMJ 0.039**
18
Obs. 100
Adj.-R2 0.31
Note: *, ** and *** depict statistical significance at the 1, 5 and 10% level, respectively. The average return from 01-01-2001 to 01-01-2002 is a negative return of 8.68%. Small companies underperform large companies, as reflected by the factor SMB. On average, they had a 2,66% lower return over 2001. Companies with a high book to market value on average underperformed companies with a low book to market value. They had an average of 0,41% lower return over the year 2001. Companies that were up in the previous year, on average underperformed companies that were down in the previous year by 17,64%. The factor UMD is not significant at 10%. Companies that were qualitatively well developed, which ranked high on the QMJ factor, outperformed their low quality counterparts on average by 4,00%. Companies with a low leveraged-beta underperformed companies with a high leveraged-beta, on average by 1%.
Based on the results from table (1), the factors of the model are used to predict the best performing stocks at the end of the holding period. The result, the predicted 30 best and 30 worst performing stocks for the period of 01-01-2002 to 31-12-2008, are virtually bought in an equally market weighted basket.
For the predicted 30 best and 30 worst performers, the actual average returns are calculated and shown in table 2.
Table (2)
This table shows the return of the market in the period from 01-01-2002 until 31-12-2008, as well as the return of the predicted 30 best performing stocks and the predicted 30 worst performing stocks.
Market return 1.23%
Predicted 30 best performing stocks 42.69%
19
5.1 Buffett's return
Mirroring the performance of Buffett's Alpha (Frazzini, Kabiller, & Pedersen, 2013), we are both long the predicted 30 best performing stocks with borrowed money, as well as selling short the predicted worst performing stocks. With a leverage ratio of 1.4 (Frazzini, Kabiller, & Pedersen, 2013), the return of the predicted 30 best performing stocks would be 42.69%*1.4 = 59.76%. The predicted 30 worst performing stocks, that are shorted, would yield a return of (-) 3.98*1.4=(-)5.57% With a portfolio that is 50% long the predicted 30 best performing stocks and that is 50% short the predicted 30 worst performing stocks, the average of the portfolio is 0.5*59.76+0.5*(-)5.57=27,10%. This portfolio has outperformed the market in the period from 01-01-2002 to 31-13-2008 by 27.10%-1.23%=25.87%.
The outperformance of the model is in line with the Frazzini, Kabiller and Pedersen (2013). Frazzini, Kabiller and Pedersen (2013) have split their returns in periods of five years. Therefore, an exact comparison between the returns is impossible.
The hypotheses (1), which states that the model as presented in equation (8) is not able to outperform the overall FTSE 100 market, can be rejected at a 1% significance level.
Hypotheses (2), which states that the factor Quality minus Junk (QMJ) is not significant for stocks in the UK stock market, can be rejected at a 5% significance level. Although this paper has found outperformance by the model, this paper also has its limitations. The data used in this paper has only a limited time plan. Capturing quality, for example, by measuring growth could be done by taking either a longer time period, or taking more data from within the time period. These limitations also correspond with the ease of use of the model. Filling in the data points with factor investing should require less time.
Also, an investor would be able to rebalance his portfolio at each desired moment that trade of stocks is opened. This would have opened up the opportunity to exit stocks which have a less desired loading on the factor model and invest in stocks that have a more desired loading on the factor model.
6. Conclusion
In this paper, the investment strategy from Buffett's Alpha (Frazzini, Kabiller, & Pedersen, 2013), which supposedly mirrored the investment strategy of investor Warren Buffett, was applied to the FTSE100. The model from Frazzini, Kabiller and Pedersen (2013), as applied to the new dataset, has outperformed the market of the FTSE100 by 25.87%. The search for
20
quality paid off, with a significant factor QMJ. This is in line with the findings from with Asness, Frazzini and Pedersen (2014).
Further research can be done by in multiple ways.
One extension of the literature is the application of the Frazzini, Kabiller and Pedersen (2013) model to a new data set, comprised of more stocks. Second, describing quality more accurate by finding more significant data points can improve this research. Third, academics can extend this research over time, by choosing different start dates of the portfolio balancing. At last, academics can extend this research by rebalancing the portfolio of chosen stocks in the basket on a monthly, quarterly or yearly basis, instead of holding on the stocks for the holding period.
Bibliography
Baker, M., & Wurgler, J. (2002). Market Timing and Capital Structure. The Journal of
Finance , 57 (1), 1-32.
Banz, R. (1981). The relationship between return and market value of common stocks.
Journal of Financial Economics , 9 (1), 3-18.
Basu, S. (1977). Investment Performance of Common Stocks in Relation to Their Price-Earnings Ratios: A Test of the Efficient Market Hypothesis. Journal of Finance , 663-682. Berk, J., & DeMarzo, P. (2007). Corporate Finance. Boston: Pearson Education, Inc. Black, F. (1993). Beta and Return. The Journal of Portfolio Management , 20 (1), 8-18. Black, F. (1972). Capital Market Equilibrium with Restricted Borrowing. The Journal of
Business , 444-455.
Carhart, M. (1997). On Persistence in Mutual Fund Performance. The Journal of Finance , 52 (1), 57-82.
Chan, L., Hamao, Y., & Lakonishok, J. (1991). Fundamentals and stock returns in Japan. The
Journal of Finance , 43, 309-325.
Fama, E. F. (1970). Efficient Capital Markets: A Review of Theory and Empirical Work. The
21
Fama, E., & French, K. (1993). Common risk factors in the returns of stocks and bonds.
Journal of Financial Economics , 33 (1), 3-56.
Frazzini, A., & Pedersen, L. (2013). Betting against Beta. NBER Working Paper Series (WP
16601) .
Frazzini, A., & Pedersen, L. (2012). Embedded Leverage. NBER Working Paper No. 18558 , 60.
Frazzini, A., Kabiller, D., & Pedersen, L. (2013). Buffett's Alpha. Working papers series , 33. George, T., & Hwang, C. (2010). A resolution of the distress risk and leverage puzzles in the cross section of stock returns. Journal of Financial Economics , 96 (1), 56-79.
Graham, B. (1949). The Intelligent Investor. New York: Harper & Brothers. Graham, B., & Dodd, D. (1934). Security Analysis. York: McGraw-Hill.
Grossman, S., & Stiglitz, J. (1980). On the Impossibility of Informationally Efficient Markets.
The American Economic Review , 70 (3), 393-408.
Jegadeesh, N., & Titman, S. (1993). Returns to Buying Winners and Selling Losers. The
Journal of Finance , 48 (1), 65-91.
Jensen, M., Black, F., & Scholes, M. (1972). The Capital Asset Pricing Model: Some Emperical Tests. STUDIES IN THE THEORY OF CAPITAL MARKETS .
La Porta, R., Lakonishok, J., Shleifer, A., & Vishny, R. (1997). Good News for Value Stocks: Further Evidence on Market Efficiency. The Journal of Finance , 859-874.
Lintner, J. (1965). The Valuation of Risk Assets and the Selection of Risky Investments in Stock Portfolios and. The Review of Economics and Statistics , 13-37.
Lowenstein, R. (2008). Buffett: The Making of an American Capitalist. New York: 1st Main Street Book Ed.
Mandelbrot, B. (1966). Forecasts of future prices, unbiased markets and “martingale” models.
Journal of Business , 242-255.
22
Martin, G., & Puthenpurackal, J. (2008). Imitation is the sincerest form of flattery: Warren Buffett and Berkshire Hathaway. Working papes series , 51.
Mohanram, P. (2005). Separating Winners from Losers among Low Book-to-Market Stocks using Financial Statement Analysis. Review of Accounting Studies , 10 (1), 133-170.
Novy-Marx, R. (2012). The Other Side of Value: The Gross Profitability Premium. Working
paper .
Ohlson, J. (1980). Financial Ratios and the Probabilistic Prediction of Bankruptcy. Journal of
Accounting Research , 18 (1), 109-131.
Richardson, S., Sloan, R., Soliman, M., & Tuna, I. (2005). Accrual Reliability, Earnings Persistence and Stock Prices. Accepted Paper Series , 65.
Rosenberg, B., Reid, K., & Lanstein, R. (1985). Persuasive evidence of market inefficiency.
Journal of Portfolio Management , 9-17.
Samuelson, P. A. (1965). Proof That Properly Anticipated Prices Fluctuate Randomly.
Industrial Management Review , 41.
Sewell, M. (2011). History of the Efficient Market Hypothesis. Research Note , 1-14. Sharpe, W. (1964). Capital Asset Prices: A Theory of Market Equilibrium under Conditions of Risk. The Journal of Finance , 21-41.
Sharpe, W. F. (1964). Capital Asset Prices: A Theory of Market Equilibrium under Conditions of Risk. The Journal of Finance , 21-41.
Sloan, R. (1996). Do Stock Prices Fully Reflect Information in Accruals and Cash Flows about Future Earnings? The Accounting Review , 71 (3), 289-315.
Stattman, D. (1980). Book Values and Stock Returns. The Chicago MBA: , 4, 25-45. Tobin, J. (1958). Liquidity Preference as Behavior Towards Risk. The Review of Economic
Studies , 65-86.
23
1.1
Table (1)
Each group in the SMB and HML factors correspond to a certain group in market capitalization and book equity / market equity.
Market Cap BE/ME Group name
Smallest 50% Lowest 30% SmallValue
Smallest 50% Middle 30% to 70% SmallNeutral
Smallest 50% Highest 30% SmallGrowth
Largest 50% Lowest 30% BigValue
Largest 50% Middle 30% to 70% BigNeutral
Largest 50% Highest 30% BigGrowth
1.2
List of constituents of the FTSE 100 at 01-01-2002 (in alphabetical order):
1 3i Group PLC 51 Kingfisher PLC
2 Alliance & Leicester PLC 52 Ladbrokes PLC
3 Alliance Boots Holdings Ltd 53 Land Securities Group PLC
4 Allied Domecq Ltd 54 Lattice Group PLC
5 Anglo American PLC 55 Legal & General Group PLC
6 ARM Holdings PLC 56 LHR Airports Ltd
7 Associated British Foods PLC 57 Lloyds Banking Group PLC
8 AstraZeneca PLC 58 Logica Ltd
9 Aviva PLC 59 Man Group PLC
10 BAE Systems PLC 60 Marks and Spencer Group PLC
11 Barclays PLC 61 National Grid PLC
12 BG Group PLC 62 Next PLC
13 Bhp Billiton PLC 63 Oclaro Technology Ltd
14 Bip Industries Ltd 64 Old Mutual PLC
15 BOC Group Ltd 65 Pearson PLC
16 BP PLC 66 Prudential PLC
17 British American Tobacco PLC 67 Reckitt Benckiser Group PLC 18 British Land Company PLC 68 Reed Elsevier PLC
24
20 BT Group PLC 70 Rio Tinto PLC
21 Cable & Wireless Communications PLC 71 Rolls-Royce Holdings PLC
22 Cadbury Ltd 72 Royal Bank of Scotland Group PLC
23 Canary Wharf Group PLC 73 Royal Dutch Shell
24 Capita PLC 74 Royal Dutch Shell PLC
25 Carnival PLC 75 RSA Insurance Group PLC
26 Celltech Group Ltd 76 RWE Npower Holdings PLC
27 Centrica PLC 77 SABMiller PLC
28 Compass Group PLC 78 Santander UK PLC
29 Daily Mail and General Trust PLC 79 Schroders PLC
30 Diageo PLC 80 Schroders PLC
31 Dixons Retail PLC 81 Scottish & Newcastle Ltd
32 E.ON UK PLC 82 Scottish Power Ltd
33 EDF Energy Nuclear Generation Group Ltd 83 Severn Trent PLC 34 Electrocomponents PLC 84 Shire PLC
35 EMI Group Ltd 85 Six Continents Ltd
36 Enterprise Oil Ltd 86 Smith & Nephew PLC
37 Experian PLC 87 Smiths Group PLC
38 Friends Life FPG Ltd 88 SSE PLC
39 G4S 89 Standard Chartered PLC
40 GlaxoSmithKline PLC 90 Telefonica Europe PLC
41 Hanson Ltd 91 Telewest Communications Networks Ltd
42 Hays PLC 92 Tesco PLC
43 HBOS PLC 93 The Sage Group PLC
44 Imperial Chemical Industries Ltd 94 Thomson Reuters UK Ltd 45 Imperial Tobacco Group PLC 95 Unilever PLC
46 International Power Ltd 96 United Utilities Group PLC
47 Invensys PLC 97 Vodafone Group PLC
48 Invesco Holding Company Ltd 98 WM Morrison Supermarkets PLC
49 ITV PLC 99 Wolseley PLC
