1. Chapter 2 Discounted Dividend Valuation

2. Challenges  Defining and forecasting CF’s  Estimating appropriate discount rate

3. Basic DCF model  An asset’s value is the present value of its (expected) future cash flows

4. Comments on basic DCF model  Flat term structure of discount rates versus differing discount rates for different time horizons  Value of an asset at any point in time is always the PV of subsequent cash flows discounted back to that point in time.

5. Three alternative definitions of cash flow  Dividend discount model  Free cash flow model  Residual income model

6. Dividend discount model  The DDM defines cash flows as dividends.  Why? An investor who buys and holds a share of stock receives cash flows only in the form of dividends  Problems: Companies that do not pay dividends. No clear relationship between dividends and profitability

7. DDM (continued)  The DDM is most suitable when: the company is dividend-paying the board of directors has a dividend policy that has an understandable relationship to profitability the investor has a non-control perspective.

8. Free cash flow  Free cash flow to the firm (FCFF) is cash flow from operations minus capital expenditures  Free cash flow to equity (FCFE) is cash flow from operations minus capital expenditures minus net payments to debtholders (interest and principal)

9. Free cash flow  FCFF is a pre-debt cash flow concept  FCFE is a post-debt cash flow concept  FCFE can be viewed as measuring what a company can afford to pay out in dividends  FCF valuation is appropriate for investors who want to take a control perspective

10. FCF valuation  PV of FCFF is the total value of the company. Value of equity is PV of FCFF minus the market value of outstanding debt.  PV of FCFE is the value of equity.  Discount rate for FCFF is the WACC. Discount rate for FCFE is the cost of equity (required rate of return for equity).

11. FCF (continued)  FCF valuation is most suitable when: the company is not dividend-paying. the company is dividend paying but dividends significantly differ from FCFE. The company’s FCF’s align with company’s profitability within a reasonable time horizon. the investor has a control perspective.  FCF valuation is very popular with analysts.

12. Residual income  RI for a given period is the earnings for that period in excess of the investors’ required return on beginning-of-period investment.  RI focuses on profitability in relation to opportunity costs.  A stock’s value is the book value per share plus the present value of expected future residual earnings

13. Residual income (continued)  RI valuation is most suitable when: the company is not dividend-paying, or as an alternative to the FCF model. the company’s FCF is negative within a comfortable time horizon. the investor has a control perspective.  RI valuation is also popular. The quality of accounting disclosure can make the use of RI valuation error-prone.

14. Which is best, DDM, FCF, or RI?  One model may be more suitable for a particular application.  Analyst may have more expertise with one model.  Availability of information.  In practice, skill in application, including the quality of forecasts, is decisive for the usefulness of an analyst’s work.

15. Discount rate determination  Jargon Discount rate: any rate used in finding the present value of a future cash flow Risk premium: compensation for risk, measured relative to the risk-free rate Required rate of return: minimum return required by investor to invest in an asset Cost of equity: required rate of return on common stock

16. Discount rate determination Weighted average cost of capital (WACC): the weighted average of the cost of equity, after-tax cost of debt, and cost of preferred stock

17. Two major approaches for cost of equity  Equilibrium models: Capital asset pricing model (CAPM) Arbitrage pricing theory (APT)  Bond yield plus risk premium method (BYPRP)

18. CAPM  Expected return is the risk-free rate plus a risk premium related to the asset’s beta:  E(R i ) = R F +  i [E(R M ) – R F ]  The beta is  i = Cov(R i,R M )/Var(R M )  [E(R M ) – R F ] is the market risk premium or the equity risk premium

19. CAPM  What do we use for the risk-free rate of return? Choice is often a short-term rate such as the 30-day T-bill rate or a long-term government bond rate. We usually match the duration of the bond rate with the investment period, so we use the long-term government bond rate. Risk-free rate must be coordinated with how the equity risk premium is calculated (i.e., both based on same bond maturity).

20. Equity risk premium  Historical estimates: Average difference between equity market returns and government debt returns. Choice between arithmetic mean return or geometric mean return (see Table 2-2 p. 50) Survivorship bias ERP varies over time ERP differs in different markets (see Table 2-3 p. 51)

21. Equity risk premium  Expectational method is forward looking instead of historical  One common estimate of this type: GGM equity risk premium estimate = dividend yield on index based on year-ahead dividends + consensus long-term earnings growth rate - current long-term government bond yield

22. Arbitrage Pricing Theory (APT)  CAPM adds a single risk premium to the risk-free rate. APT models add a set of risk premiums to the risk-free rate:  E(R i ) = R F + (Risk premium) 1 + (Risk premium) 2 + … + (Risk premium) K  (Risk premium) i = (Factor sensitivity) i × (Factor risk premium) i

23. Arbitrage Pricing Theory (APT)  Factor sensitivity is asset’s sensitivity to a particular factor (holding all other factors constant)  Factor risk premium is the factor’s expected return in excess of the risk-free rate.

24. APT models  One popular model is the Fama-French three factor model using company- specific attributes: RMRF – return on equity index minus 30 day T-bills SMB (small minus big) – return on small cap portfolio minus return on large cap portfolio HML (high minus low) – return on high book-to-market portfolio minus return on low book-to-market portfolio

25. APT models  The Burmeister, Roll, and Ross (BIRR) model uses five macroeconomic factors Confidence risk Time-horizon risk Inflation risk Business-cycle risk Market timing risk

26. Using BIRR model  Use BIRR model to calculate required return on the S&P 500 (data in example 2-4, p 53)  The required return is: r = 5.00% + (0.27×2.59%) – (0.56×0.66%) – (0.37×4.32%) + (1.71×1.40%) +(1.00×3.61%) r = 9.74%

27. Sources of error in using models  Three sources of error in using CAPM or APT models: Model uncertainty – Is the model correct? Input uncertainty – Are the equity risk premium or factor risk premiums and risk- free rate correct? Uncertainty about current values of stock beta or factor sensitivities

28. BYPRP method  The bond yield plus risk premium method finds the cost of equity as: BYPRP cost of equity = YTM on the company’s long-term debt + Risk premium  The typical risk premium added is 3-4 percent.

29. Build-up method  Cost of equity is the risk-free rate plus one or more risk premiums, one or more of which is usually subjective rather than theoretically sound.  For example, cost of equity is risk-free rate + equity risk premium +/- company-specific risk premium  BYPRP is an example of this.  Buildup method sometimes used for stocks that are not publicly traded.

30. Dividend discount models (DDMs)  Single-period DDM:  Rate of return for single-period DDM

31. More DDMs  Two-period DDM:  Multiple-period DDM:

32. Indefinite HP DDM  For an indefinite holding period, the PV of future dividends is:

33. Forecasting future dividends  Using stylized growth patterns Constant growth forever (the Gordon growth model) Two-distinct stages of growth (the two- stage growth model and the H model) Three distinct stages of growth (the three- stage growth model)

34. Forecasting future dividends  Forecast dividends for a visible time horizon, and then handle the value of the remaining future dividends either by Assigning a stylized growth pattern to dividends after the terminal point Estimate a stock price at the terminal point using some method such as a multiple of forecasted book value or earnings per share

35. Gordon Growth Model  Assumes a stylized pattern of growth, specifically constant growth: D t = D t-1 (1+g) Or D t = D 0 (1 + g) t

36. Gordon Growth Model  PV of dividend stream is:  Which can be simplified to:

37. Gordon growth model  Valuations are very sensitive to inputs. Assuming D 1 = 0.83, the value of a stock is:

38. Other Gordon Growth issues  Generally, it is illogical to have a perpetual dividend growth rate that exceeds the growth rate of GDP  Perpetuity value (g = 0):  Negative growth rates are also acceptable in the model.

39. Expected rate of return  The expected rate of return in the Gordon growth model is:  Implied growth rates can also be derived in the model.

40. PV of growth opportunities  If a firm has growing earnings and dividends, it can be worth more than a non-growing firm:  Value of growth = Value of growing firm – Value of assets in place (no growth)  OR

41. Gordon Model & P/E ratios  If E is next year’s earnings (leading P/E):  If E is this year’s earnings (trailing P/E):

42. Strengths of Gordon growth model  Good for valuing stable-growth, dividend-paying companies  Good for valuing indexes  Simplicity and clarity, also helps understanding of relationships between V, r, g, and D  Can be used as a component in more complex models

43. Weaknesses of Gordon growth model  Calculated values are very sensitive to assumed values of g and r  Is not applicable to non-dividend-paying stocks  Is not applicable to unstable-growth, dividend paying stocks

44. Two-stage DDM  The two-stage DDM is based on the multiple-period model:  Assume the first n dividends grow at g S and dividends then grow at g L. The first n dividends are:

45. Two-stage DDM (cont)  Using D n+1, the value of the stock at t=n is  The value at t = 0 is

46. Two-stage DDM example  Assume the following values D 0 is $1.00 g S is 30% Supernormal growth continues for 6 years g L is 6% The required rate of return is 12%

47. Two-stage DDM example

48. “Shortcut” two-stage DDM (not in the book)  If g S is constant during stage 1, this works:  For g S =30%, g L =6%, D 0 =1.00 and r=12%

49. Using a P/E for terminal value  The terminal value at the beginning of the second stage was found above with a Gordon growth model, assuming a long-term sustainable growth rate.  The terminal value can also be found using another method to estimate the terminal value at t = n. You can also use a P/E ratio, applied to estimated earnings at t = n.

50. Using a P/E for terminal value  For DuPont, assume D 0 = 1.40 g S = 9.3% for four years Payout ratio = 40% r = 11.5% Trailing P/E for t = 4 is 11.0  Forecasted EPS for year 4 is E 4 = 1.40(1.093) 4 / 0.40 = 1.9981 = 4.9952

51. Using a P/E for terminal value

52. Valuing a non-dividend paying stock  This can be viewed as a special case of the two- stage DDM where the dividend in stage one is zero:  Forecasting the length of stage one and the dividend pattern in stage two are the challenges.

53. The H model  The basic two-stage model assumes a constant, extraordinary rate for the super-normal growth period that is followed by a constant, normal growth rate thereafter.

54. The H model  Fuller and Hsia (1984) developed a variant of the two-stage model where the growth rate begins at a high rate and declines linearly throughout the super- normal growth period until it reaches the normal growth rate at the end. The normal growth rate continues thereafter.

55. The H model  The value of the dividend stream in the H model is:  V 0 = value per share at time zero  D 0 = current dividend  r = required rate of return on equity  H = half-life of the high growth period (i.e., high growth period = 2H years)  g S = initial short-term dividend growth rate  g L = normal long-term dividend growth rate after year 2H

56. H model example  For Siemans AG, the inputs are: Current dividend is €1.00. The dividend growth rate is 29.28%, declining linearly over a sixteen year period to a final and perpetual growth rate of 7.26%. The risk-free rate is 5.34%, the market risk premium is 5.32%, and the Siemens beta, estimated against the DAX index, is 1.37. The required rate of return for Siemens is: r = r f + b i (r m – r f ) = 5.34% + 1.37(5.32%) = 12.63%.

57. H model example  Using the H model, the value of the company is:  V 0 = 19.97 + 32.80 = €52.77.  If Siemens experienced normal growth starting now, its value would be €19.97. The extraordinary growth adds €32.80 to its value, which results in Siemens being worth a total of €52.77.

58. Three-stage DDM  There are two popular version of the three-stage DDM The first version is like the two-stage model, only the firm is assumed to have a constant dividend growth rate in each of the three stages. A second version of the three-stage DDM combines the two-stage DDM and the H model. In the first stage, dividends grow at a high, constant (supernormal) rate for the whole period. In the second stage, dividends decline linearly as they do in the H model. Finally, in stage three, dividends grow at a sustainable, constant rate.

59. Three-stage DDM with three distinct stages  Assume the following for IBM: Required rate of return is 12% Current dividend is $0.55 Growth rate and duration for phase one are 7.5% for two years Growth rate and duration for phase two are 13.5% for the next four years Growth rate in phase four is 11.25% forever

60. Three-stage DDM with three distinct stages

61. Spreadsheet modeling  Spreadsheets allow the analyst to build very complicated models that would be very cumbersome to describe using algebra.  Built-in functions such as those to find rates of return use algorithms to get a numerical answer when a mathematical solution would be impossible or extremely complicated.

62. Spreadsheet modeling  Because of their widespread use, several analysts can work together or exchange information through the sharing of their spreadsheet models.

63. Finding rates of return for any DDM  For a one-period DDM  For the Gordon model  For the H-model

64. Finding rates of return for any DDM  For multi-stage models and spreadsheet models it can be more difficult to find a single equation for the rate of return. Trial and error is used instead of an equation. Using a computer or trial and error, the analyst finds a discount rate such that the present value of future expected dividends equals the current stock price.

65. Finding r with trial & error  Johnson & Johnson’s current dividend of $.70 to grow by 14.5 percent for six years and then grow by 8 percent into perpetuity. J&J’s current price is $53.28. What is the expected return on an investment in J&J’s stock?

66. Finding r with trial & error  For a good initial guess, we can use the expected rate of return formula from the Gordon model as a first approximation: r = ($0.70  1.145)/$53.28 + 8% = 9.50%. Since we know that the growth rate in the first six years is more than 8 percent, the estimated rate of return must be above 9.5 percent.  Let’s use 9.5 percent and 10.0 percent to calculate the implied price.

67. Finding r with trial & error The present value of the terminal value = V 6 / (1+r) 6 = [D 7 /(r-g)]/(1+r) 6 The calculations for 9.5% and 10.0% are shown in the table. Actual r is 9.988%.

68. Strengths of multistage DDMs  Can accommodate a variety of patterns of future dividend streams.  Even though they may not replicate the future dividends exactly, they can be a useful approximation.  The expected rates of return can be imputed by finding the discount rate that equates the present value of the dividend stream to the current stock price.

69. Strengths of multistage DDMs  Because of the variety of DDMs available, the analyst is both enabled and compelled to evaluate carefully the assumptions about the stock under examination.  Spreadsheets are widely available, allowing the analyst to construct and solve an almost limitless number of models.

70. Strengths of multistage DDMs  Using a model forces the analyst to specify assumptions (rather than simply using subjective assessments). This allows analysts to use common assumptions, to understand the reasons for differing valuations when they occur, and to react to changing market conditions in a systematic manner.

71. Weaknesses of multistage DDMs  Garbage in, garbage out. If the inputs are not economically meaningful, the outputs from the model will be of questionable value.  Analysts sometimes employ models that they do not understand fully.  Valuations are very sensitive to the inputs to the models.

72. Weaknesses of multistage DDMs  Subjective assessments may be better than systematic, quantitative assessments in some cases.  Programming and data errors in spreadsheet models are very common. These models must be checked very thoroughly.

73. Weaknesses of multistage DDMs  The choice of model should be made very carefully. There is a tendency to grab a model, put in the data, get the results, and use them without carefully justifying the logic of the underlying model and the appropriateness and realism of the values inserted into the model.

74. Equity durations (not in the book)  Duration is a measure of the interest rate risk of fixed income securities. The concept of duration can also be adapted to equities.  The mathematical definition of duration is  The percentage change in price is

75. Gordon model duration (not in the book)  The stock price is  The derivative with respect to r is  The duration is

76. Forecasting growth rates  There are three basic methods for forecasting growth rates: Using analyst forecasts Using historical rates (use historical dividend growth rate or use a statistical forecasting model based on historical data) Using company and industry fundamentals

77. Finding g  The simplest model of the dividend growth rate is: g = b x ROE whereg = Dividend growth rate b = Earnings retention rate (1 – payout ratio) ROE = Return on equity.

78. Finding g  The ROE, found with the duPont model is:  The growth rate can also be expressed as:

79. DDMs and portfolio selection  Investment managers have used DCF models, including dividend discount models as part of a systematic approach to security selection and portfolio formation.  If a manager just chooses the most undervalued securities without any risk discipline, his selections might concentrate on a particular (or a few) risk factors. He might often fail to meet his risk objective. A risk control discipline must be used.

80. DDMs and portfolio selection  Sort stocks into groups according to the risk control methodology. For example, put stocks into groups of similar beta risk.  Rank stocks by expected return within each group using a DCF methodology. Rank stocks from highest to lowest expected return within each sector grouping.

81. DDMs and portfolio selection  Select portfolio from the highest expected return stocks consistent with the risk control methodology. All selected securities are equal weighted, but more important sectors have a larger number of securities; the result is approximate sector neutrality.

82. DDMs and portfolio selection  Six analysts follow a universe of 250 stocks.  Company uses a three-stage, H-model.  For each, an analyst estimates 1) the initial growth rate, 2) the length of the initial phase, and 3) the length of the transitional phase.  Initial growth rate estimated with duPont model.

83. DDMs and portfolio selection  Length of initial phase and transitional phase based on fundamental analysis.  Growth rate for maturity phase assumed to be the same for all stocks.  Stocks assigned to five beta quintiles.

84. DDMs and portfolio selection  Company invests in the top return quintile in each beta quintile (10 stocks in each beta quintile).  Method had superior returns for several years.