Implementing Basic Valuation

Part I: Discounted Cashflows

Jiro E. Kondo
McGill University - Desautels Faculty of Management and Northwestern University - Kellogg School of Management [email protected] FINE-443: Applied Corporate Finance McGill University - Desautels Winter 2012

Do Not Distribute or Copy in Any Part Without the Prior Consent of the Author

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part I

Outline for Topic 1 (Part I)

Review of NPV w/ WACC

Obtaining After-Tax Cashows From Earnings Forecasts Obtaining the Weighted-Average Cost of Capital (WACC)

Estimating the Cost of Capital in Practice

Practical Strategies for Estimating Discount Rates: Equity and Debt Other Considerations: (i) Multiple Divisions, (ii) Private Companies, (iii) Lack of Diversication Mini Case Study (Part 1): Discounting Public Pension Liabilities

Forecasting Cashows: Key Drivers

Strategies for Forecasting Cashows: (i) Top-Down vs. Bottom-Up, (ii) Decompositions Forecasting Terminal Values Mini Case Study (Part 2): Forecasting Public Pension Liabilities

DCF is Not A Number: Sensitivity Analysis

Determining a Plausible Range of Values

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ia

Topic 1 (Part Ia): NPV w/ WACC

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ia

NPV w/ WACC: Outline of Technique

Theorem
A common approach to valuing projects/companies is discounted cashow analysis (DCF). The most common version of this is NPV w/ WACC. This procedure involves the following steps: Step 1: (a) Forecast the components of expected future earnings and (b) use these to obtain after-tax cashows using: CFt = (1 T ax) EBIT Dt + T ax DEP Rt CAP Xt W Ct (1)

Step 2: (a) Get the after-tax discount rates (for valuing the cashows estimated above) using the weighted-average cost of capital (WACC) formula: E [r ] = (D/V ) (E [rD ] (1 T ax)) + (E/V ) E [rE ] (2)

where (b) the values of E [rD ] and E [rE ] are obtained from market prices (e.g., yields on debt of similar risk) or using an asset pricing model (e.g., the CAPM). Step 3: Take the output from Steps 1 and 2 and combine into the NPV formula. Review of NPV w/ WACC: Our rst goal in the course is to re-gain comfort with this procedure using examples.
Jiro E. Kondo (McGill) Implementing Basic Valuation Winter 2012

Lecture Note 1

Part Ia

NPV w/ WACC (Step 1): Obtaining After-Tax Cashflows

Components of the Cashow Formula From (1): CFt = (1 T ax) EBIT Dt + T ax DEP Rt CAP Xt W Ct {z } | {z } | {z } |
Accounting Capital Expenditures Other

Accounting: Only operations and ignores depreciation. Capital Expenditures: Capital purchases are straightforward, but make sure you factor in taxes in the case of capital sales... CAP Xt = Sales Pricet T ax (Sales Pricet Book Valuet ) | {z }
Capital Gain on Salet

(3)

Other: In this course, we focus on account receivable, accounts payable, and inventory. W Ct = ARt + IN Vt APt (4)

In your accounting courses, you will see that working capital is more generally dened as current assets minus current liabilities. The cashow formula at the top of this page is valid in this more general case as well.
Jiro E. Kondo (McGill) Implementing Basic Valuation Winter 2012

Lecture Note 1

Part Ia

NPV w/ WACC: Additional Notes

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ia

NPV w/ WACC: Obtaining After-Tax Cashflows

Exercise 1: The Jamie Blanchard Goalie School is considering the purchase of 5 brand new puck shooting machine systems to help train its students to become good goaltenders. These systems, known as The Boni Showdown 1 on 1, cost $9,000 each and are expected to last 10 years. For accounting purposes, they would be depreciated on a straight-line basis over 10 years down to a salvage value of zero. However, after 10 years, they would be sold for $1,000 each. Currently, the school does not have any puck shooting systems and instead hires local hockey players to shoot pucks at their goalies. The incremental costs due to this labor are $3,000 per year and purchasing the puck shooting system would eliminate this source of cost. All other labor and operations costs total to $25,000 per year regardless of whether the puck shooting system is purchased or not. By moving to high-tech shooting systems, the school expects to increase enrollment in its school from 95 to 100 students per year and the cost of enrollment (tuition) from $800 to $820 per year. Every year, 80% of students pay their tuition at the end of that year while the other 20% pay their tuition at the end of the following year. Assume that the goalie school pays 28% of its accounting prots as taxes every year. The after-tax discount rate to use for valuing the schools cashows is 8% (expressed as an EAR). Should the goalie school purchase these puck shooting machine systems? What is the payback period and IRR on this project? Why is NPV w/ WACC favorable to these metrics?

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ia

Practical Tip: Getting Organized w/ Cashflows

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ia

Depreciation in The Real World (USA): MACRS Schedule

Source: Internal Revenue Service (2007)

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ia

NPV w/ WACC: Obtaining After-Tax Cashflows

Exercise 2: Redo Exercise 1 with the 10-year MACRS depreciation schedule from the previous slide.

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ia

NPV w/ WACC (Step 2): Obtaining After-Tax Discount Rates

Components of the WACC Formula (2): Portfolio weighted-average of required rates of return on debt and equity... E [r ] = (D/V ) (E [rD ] (1 T ax)) +(E/V ) E [rE ] {z } |
The Tax-Benet of Debt

Portfolio Weights: Fraction of (total) rm value that is represented by debt (D/V ) and equity (E/V ). Capital Asset Pricing Model (CAPM): A model of expected returns... E [ri ] = rf + i (E [rm ] rf ) {z } |
Risk Premium Beta

(5)

where (E [rm ] rf ) is the market risk premium, rf is the riskless rate, and i is the beta of investment i. i attempts to capture nondiversiable risk by measuring how much investment is returns and the market portfolios returns: i = Cov[ri rf , rm rf ] V [ rm ] (6)

Of course, there are other approaches and details to getting E [rD ] and E [rE ] in practice. We will discuss these later on in this Topic.
Jiro E. Kondo (McGill) Implementing Basic Valuation Winter 2012

Lecture Note 1

Part Ia

NPV w/ WACC: Obtaining After-Tax Discount Rates

Exercise 3: Your rm operates in the fast food industry. This industry only has one type of project: the production of nasty food. Your rms debt has a market value of D = 240M and a current market capitalization of E = 960M . You have estimates for the beta of your rms debt and equity: E = 0.7 and D = 0. The risk free rate is 4% and the market risk premium is 5%. Your rm has a tax rate of 30%. Using the CAPM and WACC, what is the appropriate discount rate for nasty food production projects? Solution: To compute the discount rate for your rm, we need to compute E [rD ] and E [rE ] for the rm. We will do so using the CAPM: E [r D ] = = = E [r E ] = = = rf + D (E [rm ] rf ) 0.04 + 0 0.05 4% rf + E (E [rm ] rf ) 0.04 + 0.7 0.05 7.5%

Now, we can use the WACC formula to get the discount rate: E [r ] = (240/1200) (0.04 (1 0.3)) + (960/1200) (0.075) = 6.56%
Jiro E. Kondo (McGill) Implementing Basic Valuation Winter 2012

Lecture Note 1

Part Ia

NPV w/ WACC: Obtaining After-Tax Discount Rates

Key Assumption in Previous Exercise: The project in question was a typical rm project. In other words, the discount rate for the project would equal the discount rate for the rm as a whole. Alternative Scenarios: What if the discount rate for the rm as a whole cannot be estimated (e.g., because the rm is private) or the project in question is not a typical rm project (e.g., because the rm has many dierent divisions)? If you want a ne-tuned discount rate, you must use an alternative approach. One popular approach is to estimate discount rates using comparables. Direct Comparables: Find other companies whose risk resembles that of the project in question and whose discount rate for the rm as a whole can be estimated. These are generally focused companies (pure plays) in the projects industry. Synthetic Comparables: This approach will be highlighted with step-by-step instructions in Homework 1.

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ia

NPV w/ WACC: Obtaining After-Tax Discount Rates

Exercise 4: You are considering a proposal to start a new airline. To evaluate the proposal, you will need an estimate of the cost of capital in the airline industry. You have run CAPM regressions on the stock returns of 5 existing airlines that are comparable to the operations of the new business you are considering. You have also obtained estimates of the betas of the debt for these 5 rms. The risk free interest rate is 4% and the market risk premium is 5%. All rms have a tax rate of 30%. Company Southwest Airlines Co. SkyWest, Inc. Alaska Air Group, Inc. Mesa Air Group, Inc. Continental Airlines, Inc. Equity Beta 1.13 1.69 1.80 3.27 3.76 Debt Beta 0.00 0.15 0.15 0.30 0.40 Debt-Equity Ratio 0.15 1.05 1.06 3.52 5.59

Why are the equity betas (and debt betas) so dierent across rms? Does this imply that the various airline companys have very dierent levels of risk? Use this information to estimate the beta and the cost of capital for a typical project in the airline industry?

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ia

NPV w/ WACC: Two Approaches w/ Comparables

Approach #1: Average the comparables discount rates... E [r ] = 1 E [r 1 ] + ... + E [r N ] N (7)

where E [r n ] is the estimated discount rate for comparable n. Approach #2: Average the comparables unlevered betas and lever up using your rms capital structure... A = 1 1 N A + ... + A N and E [r ] = E [rA ] (D/V ) E [rD ] T ax (8)

n is the unlevered (asset) beta of comparable n. Ive assumed a constant D/V in where A the second equation.

Is There A Best Choice Between The Two Options? Not necessarily. The best option depends on your view of whats most comparable across rms: business risk (i.e., E [rA ]) or discount rates (i.e., E [r ]).
Jiro E. Kondo (McGill) Implementing Basic Valuation Winter 2012

Lecture Note 1

Part Ia

NPV w/ WACC: Fully-Fledged Exercise

Exercise 5: Bart and Mogul Inc. (BM) is planning to enter the online book-selling business. Doing so is expected to require $50M in upfront investments which are to be split equally between now (at t = 0) and next year (at t = 1). These investment costs entirely fall under capital expenditures and will be depreciated over 5 years (from t = 2 till t = 6). These capital assets will never be sold. Following this setup period, the company is expected to have yearly operational expenses that will start at $20M (at t = 2) and will rise by 10% per year for the next 8 years (until t = 10). Meanwhile, its operational revenues will start at $17.5M (at t = 2) and will rise by 30% per year for the next 8 years (until t = 10). After this period, everyone will be reading e-books and BMs book-selling business will cease to generate any cashows (i.e., lets assume that they wont enter the online e-book market). BM believes that the risk involved in this project lies exactly between those of Amazon.com (AMZN) and Barnes and Noble (BN). AMZN and BNs stocks both have (equity) betas of 1.5. However, 30% of AMZNs rm value comes from debt while this number equals 45% for BN. Both AMZNs and BNs debt have (debt) betas of zero. The riskless rate is 5% while the expected return on the market portfolio equals 15% (both rates are expressed as EARs). All three rms (BM, AMZN, and BN) have a corporate tax rate of 35%. BMs other lines of business are primarily center around Hoola-hoop production and sales. Historically, it has applied a discount rate of 10% on those projects. BM should not use the 10% discount rate for the current project under consideration. Why? More specically, determine the discount rate it should be using (given the statements above). Should BM undertake this project?

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ib

Topic 1 (Part Ib): Estimating the Cost of Capital in Practice

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ib

Cost of Capital: Two Practical Approaches to Estimation

Components of the WACC Formula (2): Portfolio weighted-average of required rates of return on debt and equity... E [r ] = (D/V ) | {z }
D -Weight (3)

(E [rD ] (1 T ax)) | {z }
After-Tax Cost of Debt (2)

(E/V ) | {z }
E -Weight (3)

E [ rE ] | {z }
Cost of Equity (1)

To get the cost of capital for a rm/project, we must estimate each component of this formula. Estimate w/ Market Data of Traded Assets: Uses market data (e.g., returns or promised yields) on your own company or comparable companies along with other aggregate market data to generate an estimate of your projects cost of capital. The key is that you use information about risk reected in market valuation dynamics to gauge risk. In other words, you are exploiting the collective brain of participants in asset markets. Estimate w/ Accounting Data: Uses accounting data (e.g., earnings or cashows) on your own company or comparable companies along with other aggregate accounting data to generate an estimate of your projects cost of capital. In most of these approaches, you are not exploiting the collective knowledge of markets and the validity of the approaches hinges on making additional assumptions. Remark: It is common to use dierent approaches for estimating the dierent components of a projects cost of capital (e.g., estimating E [rD ] and E [rE ]). In practice, some approaches are more geared for estimating the cost of equity while other are targeted towards assessing the cost debt. As a result, we will discuss estimating the cost of equity and the cost of debt separately.
Jiro E. Kondo (McGill) Implementing Basic Valuation Winter 2012

Lecture Note 1

Part Ib

Cost of Equity: The Capital Asset Pricing Model (CAPM)

Components of the CAPM Formula (5): The most commonly used model for obtaining the cost of equity is the CAPM... E [ ri ] = rf |{z}
Riskless Rate

is Equity Beta

i |{z}

(E [rm ] rf ) {z } |
Mkt Risk Premium

where stock is beta is given by formula (6) earlier in the slides. Implementation: In order to use the CAPM in practice, we must estimate each component of the CAPM formula. For each component, we will consider a few approaches to estimation.
Equity Beta: (i) Own Returns, (ii) Comparable Returns, (iii) Accounting Approach. Market Risk Premium: (i) Historical Data, (ii) Survey-Based, (iii) Model-Implied (Variety of Options Here). Riskless Rate: (i) Yields on Riskless Government Debt, (ii) Yields on Riskless Government Debt Plus An Illiquidity Premium.

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ib

Cost of Equity w/ Market Data: Estimating Equity Betas (i)

Estimating a Traded Firms Equity Beta: The simplest and most common approach is to run an ordinary least-squares regression with time-series data on stock is return: rit rf t = ai + bi (rmt rf t ) + {z } |
Systematic

|{z}
Idiosyncratic

it

(9)

With enough data, the coecient bi equals Cov [ri rf , rm rf ]/V [rm ] which is exactly the formula for stock is beta (i ) from (6). We call rit rf t and rmt rf t the (realized) excess returns of stock i and the market, respectively. As discussed earlier, the coecient ai is called the alpha of an investment i. The CAPM predicts that no investment has a positive or negative alpha (market eciency). Some useful facts: The systematic (non-diversiable) component of excess returns is bi (rmt rf t ) which has a
2 2 2 = i M variance sys,i

(Systematic Variance)
it

The idiosyncratic (i.e., diversiable) component of excess returns is

2 2 2 2 denoted by idio,i = i = i sys,i

which has a variance

(Idiosyncratic Variance)

An assets total variance always equals its systematic variance plus its idiosyncratic variance. The R-square of the regression can be interpreted as the fraction of total variance that is systematic:
2 2 R2 = sys,i /i

How can we interpret 1-R2 ?

Implementing Basic Valuation Winter 2012

Jiro E. Kondo (McGill)

Lecture Note 1

Part Ib

Visualization: Estimating Amazons Equity Beta

Heres a scatter plot displaying the excess returns of Amazon and the market from May 2000 to April 2005. Lets do a CAPM-based characterization and decomposition of Amazons risk.

rAt rF t = 0.0014 + 1.3053 (rM t rF t ) + At R2 = 0.326 M = 0.0451 and Amazons beta (A ) A = 0.1032

Amazon Excess Return

Mkt Excess Return

Systematic StDev: 1.30532 0.04512 = 0.0589

2 Idiosyncratic StDev: Solves 0.10322 = 0.05892 + idio,A idio,A = 0.0847

Notice: Systematic Variance/Total Variance = 0.05892 /0.10322 = 0.326 = R2 !

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ib

Excel Tip: Using the Analysis ToolPak to Run Regressions

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ib

Cost of Equity w/ Market Data: More on Equity Betas (i)

Question: How many years of data should you use? What frequency of returns?
Betas are not constant over time so you dont necessarily want to use the longest possible horizon of returns when estimating betas. Its common to use ve years of monthly data or a couple of years of weekly data.

Question: What is the market portfolio and the riskless rate in CAPM regressions?
For the market portfolio, its common to use the S&P500 index. Using other indexes like the Wilshire 5000, the NYSE Composite Index, or even the full index of all NYSE, NASDAQ and AMEX rms will give fairly similar beta estimates. Could also use (or construct) a world stock index or include other asset classes when computing returns. Unfortunately, data on these are tougher to come by. For the riskless rate, its common to use the promised yields on 30-day Treasury bills or another longer-term US govt bond.

Remark: You can also buy more sophisticated beta estimates from certain data vendors (e.g., from Value Line, Barra, etc).

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ib

Visualization: How AT&Ts Equity Beta Changed Over Time

Figure: AT&Ts Beta

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ib

Cost of Equity w/ Market Data: Estimating Equity Betas (ii)

An Alternative Method w/ Mkt Data: Perform the same analysis but using the returns of comparables. Important Caveat: Take into account leverage dierentials across comparables... E V E = 1 N E1 V1
1 E + ... +

EN VN

N E

(10)

1 = ... = N = 0 where xn denotes the value of x for comparable n. Ive assumed D = D D and a constant D/V in this formula.

The comparables approach can be especially useful if:

There is an identiable and appropriate set of comparables for the rm/project (obvious!). The company in question is private or the project is atypical for the rm. In this case, it is impossible to estimate an equity beta for the rm/project using own returns since there is no data on these returns (either at the rm or the division-level). The estimate of the equity beta for the company is very uncertain (large standard errors on slope coecient of the CAPM regression). This can happen if the rm only recently had an IPO or if it has a large amount of idiosyncratic risk.
Jiro E. Kondo (McGill) Implementing Basic Valuation Winter 2012

Lecture Note 1

Part Ib

Cost of Equity w/ Market Data: Comparables Example (Time?)

Exercise 6: It is January 2010 and the Desautels Hospitality Group is the target of a possible acquisition and its board is entertaining a few oers from private equity rms and large publicly traded companies. In order to estimate a fair value of the company, it must determine the cost of capital for the company. To do so, it will rst estimate its beta of equity using the following comparables: Name Intercontinental Hotels Starwood Hotels & Resorts Windham Hotels McCormick & Schmicks Ruths Hospitality Industry Lodging Lodging Lodging Restaurant Restaurant D/V 0.14 0.26 0.43 0.16 0.38 D 0.10 0.10 0.15 0.20 0.20 E 1.85 2.20 3.00 2.50 3.15

Throughout this exercise, assume that 70% of DHGs value is due to its lodging line of business and 30% is due to its restaurant line of business. Also assume that the debt of all the comparables has a beta of zero and that DHG has no existing debt. Get D/V for each comparable using its most recent balance sheet and market capitalization data. (Disregard the fact that it is not January 2010...) Calculate a bottom-up (equity) beta for DHG using equity return data for the comparables from January 2005 till December 2009. A bottom-up beta involves decomposing a company into its lines of business (e.g., Lodging and Restaurants in DHGs case) and estimating the betas for each lines of business using comparables. Then you combine the line of business betas into a rm-wide beta (using the formula for the beta of a portfolio).

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ib

Cost of Equity w/ Accounting Data: Estimating Equity Betas (iii)

Estimating a Nontraded Firms Beta: Aside from using comparables, another approach that is (somewhat) used is to run an ordinary least-squares regression with time-series data on one of rm is accounting numbers. For example, we could use percentage changes in net income (N I ): N Iit N Ii,t1 = ai + bi | N Imt + it |{z} N Im,t1 {z } Idiosyncratic

(11)

Systematic

where bi is often used as the estimate of asset is beta (i ) from (6). It is often called an accounting beta. Heres an example of accounting beta estimation for the private company InfoSoft (as of 1998:Q2): Period 1992:Q1 1992:Q2 1992:Q3 1992:Q4 1993:Q1 1993:Q2 1993:Q3 1993:Q4 1994:Q1 InfoSoft 7.5% 8.3% 8.8% 7.9% 14.3% 16.5% 17.1% 13.5% 11.5% Mkt -1.3% 2.2% 2.5% 3.0% 3.6% 5.1% 5.5% 6.2% 4.3% Period 1994:Q2 1994:Q3 1994:Q4 1995:Q1 1995:Q2 1995:Q3 1995:Q4 1996:Q1 1996:Q2 InfoSoft 12.3% 13.0% 11.1% 18.6% 24.1% 17.5% 16.0% 27.0% 21.3% Mkt 4.7% 4.5% 4.2% 7.1% 8.5% 6.0% 5.0% 8.1% 7.0% Period 1996:Q3 1996:Q4 1997:Q1 1997:Q2 1997:Q3 1997:Q4 1998:Q1 1998:Q2 InfoSoft 22.5% 20.0% 17.1% 22.2% 17.8% 14.5% 8.5% 3.5% Mkt 7.2% 6.0% 5.8% 8.0% 6.1% 4.5% 1.3% -0.5%

Running the above OLS regression yields a beta of 2.15.

Jiro E. Kondo (McGill) Implementing Basic Valuation Winter 2012

Lecture Note 1

Part Ib

Cost of Equity w/ Accounting Data: Caveat Regarding (iii)

When is this approach likely to be most valid? Should it be used in a broad set of circumstances?

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ib

Cost of Equity: Estimating the Market Risk Premium (i) and (ii)
Using Historical Data: If we assume that the past is reective of the future, we can use historical data on returns to get the market risk premium... 1926-2010 E [r i ] i Geo Ari Risk 12.1% 16.7% 32.6% 9.9% 11.9% 20.4% 10.8% 14.1% 27.2% 5.5% 5.9% 9.5% 5.4% 5.5% 5.7% 3.6% 3.7% 3.1% 1970-2010 E [r i ] Geo Ari 12.5% 15.1% 10.0% 11.6% 11.1% 13.8% 8.7% 9.3% 8.0% 8.2% 5.6% 5.6% i Risk 23.4% 17.9% 20.7% 11.7% 6.6% 3.1%

Asset Class i Small Stocks Large Stocks Mkt Portfolio (S&P500) LT Govt Bonds MT Govt Bonds Treasury Bills

To get the market risk premium, take the historical return on the market portfolio minus a horizonmatched riskless rate (i.e., T-Bill for short-term project, LT Govt Bond for long-term project). Using a Survey of Forecasters: The following are surveyed estimates from 2005 (published in the WSJ) for expectations forecasts through 2050... Forecaster William Dudley Jim Glassman David Rosenberg Jeremy Siegel Robert Shiller Organization Goldman Sachs JP Morgan Merrill Lynch Wharton Yale Stocks 5.0% 4.0% 4.0% 6.0% 4.6% Govt Bonds 2.0% 2.5% 3.0% 1.8% 2.2% Corp Bonds 2.5% 3.5% 4.0% 2.3% 2.7% Mkt Risk Premium 3.0% 1.5% 1.0% 4.2% 2.4%

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ib

Cost of Equity: Estimating the Market Risk Premium (iii)

(Forward) Dividend Strips: A dividend strip (a.k.a. an index dividend future) is a standardized contract where, at maturity, the buyer pays the futures price, which is determined today, and the seller pays the dollar amount of dividends during a calendar year. Relevant Market: Over the past decade, these contracts have been traded over-the-counter for various indices (e.g., S&P500, Nikkei 225, DJ Eurostoxx 50, etc.). Remark: Like other forward and futures contracts youve seen (e.g., forwards lending contracts), no cash changes hands initially and all cashows are exchanged at maturity. Using Dividend Strips to Get Model-Implied Market Risk Premia: Lets introduce some notation... Fm,T : The forward dividend strip price on a contract for index m that matures in T years. Dm,T : The expected total dividend payments on index m in year T . This has to be estimated using a forecasting model E.g., Forecast growth rates: Dm,T = Dm,0 (1 + gm,T )T rf,T : The T -year riskless rate of return (e.g., from the US Treasury yield curve). rm,T : The T -year expected return on index m for its dividends at year T . These variables are related to each other according to the following formula: Fm,T = Dm,T (1 + rf,T )T T (1 + rm,T ) | {z } {z } Pay Later FV |
PV of Future Dividend

and

m,T | {z }
Mkt Risk Prem

= rm,T rf,T

(12)

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ib

Cost of Equity: Additional Notes on Mkt Premium (iii)

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ib

Visualization: Market Risk-Premium w/ Dividend Strips (USA)

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ib

Visualization: Market Risk-Premium w/ Dividend Strips (JPN)

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ib

Cost of Equity: Estimating the Riskless Rate (i) and (ii)

Approach (i): Use US Govt Treasuries.
Implementation (i.a): Use the 30-day T-Bill. It is the closest thing to a riskless security out there. Implementation (i.b): Use another US Govt Treasury whose maturity roughly matches the horizon of your project. Implementation (i.c): Use time-appropriate riskless rate according to the US Treasury yield curve (e.g., the yield on a 2-yr T-Bond to get the riskless rate used for a 2-yr cashow). (Best complements the use of time-varying market premia obtained from Dividend Strips).

Approach (ii): Adjust (i) To Account For An Illiquidity Premium.

How can we obtain a reasonable illiquidity premium for this adjustment? Krishnamurthy and Vissing-Jorgensen (2011) suggest looking at the historical spread between safe assets that are illiquid (e.g., FDIC insured CDs) and safe assets that are liquid (e.g., T-Bills). From 1926-2008, this spread has averaged 0.46%. Are FDIC insured CDs the right benchmark? To perform the adjustment to (i), we simply add the illiquidity premium:

rf |{z}
Riskless for Illiquid

Treasury rf | {z } From (i)

Illiq |{z}
Illiquidity Adj.

(13)

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ib

Cost of Debt w/ Market Data: Credit Ratings and Beyond

Approach (i): Use Your Firms Cost of Debt or That of Ratings-Matched Firms. Remark: The latter requires your rm to have a credit rating. Table I in the next slide gives historical data on credit spreads (i.e., avg. yield minus the T-Bill rate) from the 1995-1999 period across dierent credit ratings and maturities. From this table, we can (roughly) compute the cost of debt, E [rD ], as follows: 0 B 1 + E [rD ] = [1 (1 Pbkrpt ) (1 Rbkrpt )] B @1 {z } |
Adjustment Due to Possibility of Default

1 + rf |{z}
Treasury

+ {z

Credit | {z }
Own Credit Class

C C A }

(14)

Promised Return

where Pbkrpt is the probability of bankruptcy/default and Rbkrpt is the average recovery rate to creditors conditional on default. Approach (ii): Use Some-Type of Econometric Analysis. We can also try to predict credit spreads using a regression model which predicts credit spreads as a function of rm-specic variables, macroeconomic variables. See Table II in a few slides. Similarly, you can get a predicted rating using rm-specic variables and more advanced econometric techniques (e.g., ordered multinomial model). Most useful for rms that dont have a credit rating (e.g., because all their debt is private).
Jiro E. Kondo (McGill) Implementing Basic Valuation Winter 2012

Lecture Note 1

Part Ib

Cost of Debt w/ Mkt Data: Credit Ratings and Yield Spreads

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ib

Visualization: Beware... Credit Spreads Change Over Time!

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ib

Cost of Debt w/ Market Data: Evidence From Campbell-Taksler

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ib

Cost of Debt w/ Market Data: Evidence (Contd)

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ib

Mini Case (Part 1): Discounting Public Pension Liabilities

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ib

Mini Case (Part 1): Key Issues

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ib

Mini Case (Part 1): Analysis

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ib

WACC Weights: How Do We Get the Market Value of Debt?

Market Value of Equity: Easy for publicly traded rms... E = Shares Outstanding Price Per Share (15)

Market Value of Debt: Far more complicated because a lot of a companys debt is not publicly traded (e.g., syndicated loans, lines of credit). How can we deal with this? Two (imperfect) approaches. Approach (i): Just use the book value of the companys debt (e.g., found in balance sheet). Approach (ii): A more sophisticated approach tries to incorporate average maturity and coupon payments using back-of-the-envelope PV calculations.
Let DBV be the book value of the companys debt (from (i)). Furthermore, let T be the average maturity of a companys debt and C be the companys most recent interest expenses (e.g., found in the income statement). A reasonable estimate of the market value of debt is: D= C DBV 1 + 1 T E [r D ] (1 + E [rD ]) (1 + E [rD ])T | {z } | {z }
PV of Coupons Annuity PV of Principal

(16)

What assumptions are made in this calculation?

Jiro E. Kondo (McGill) Implementing Basic Valuation Winter 2012

Lecture Note 1

Part Ib

Additional Notes: Overview of Implementing WACC...

Components of the WACC Formula (2): Portfolio weighted-average of required rates of return on debt and equity... E [r ] = (D/V ) | {z }
D -Weight (3)

(E [rD ] (1 T ax)) {z } |
After-Tax Cost of Debt (2)

(E/V ) | {z }
E -Weight (3)

E [ rE ] | {z }
Cost of Equity (1)

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ib

Caveat: Is WACC Appropriate for Private Companies?

Two Cases: Case #1: Company owners are well-diversied. Case #2: Company owners are not well-diversied.

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ic

Topic 1 (Part Ic): Forecasting Cashflows: Key Drivers

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ic

Fundamental Approaches to Forcasting Cashflows

Common Across Approaches: Forecasting the income statement for EBIT Dt and the balance sheet for CAP Xt and W Ct . To do this, you must:
Determine Key Drivers: Usually involves focusing on one or two important line items and projecting the others based on a forecast ratio relative to these. Question: Can you think of qualitative and quantitative ways to do this?

Bottom-Up: Focus on the operations of the company/project and, for example, forecast the key drivers of demand to determine future revenues and required capital expenditures. (Micro-Approach) Top-Down: Focus on the markets the company/project is involved in and, for example, forecast the key drivers of market size and share to determine future revenues and required capital expenditures. (Macro-Approach)

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ic

Forecasting Cashflows: Items to Forecast

Line Item Revenues COGS SG&A Depreciation EBITD Capital Expenditures Accounts Payable Inventory Accounts Receivable

Reported In Income Statement Income Statement Income Statement Income Statement Income Statement Balance Sheet Balance Sheet Balance Sheet Balance Sheet

Comments Often a focus of the forecaster. Often based on a forecast ratio of revenues. Sometimes also a focus of the forecaster. Often based on a forecast ratio of revenues. Often based on CAPX forecasts. Mechanically based on previous line items. Often a focus of the forecaster. Often based on a forecast ratio of revenues. Often based on a forecast ratio of revenues. Often based on a forecast ratio of revenues.

Remark: The science and art of forecasting is truly multidisciplinary...

Example #1: Revenue forecasting often involves marketing and strategy knowledge. Example #2: COGS often involves operations knowledge. So pay attention in your other classes... and this one. ;-)

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ic

Forecasting Cashflows: Bottom-Up Approach For NetFlix

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ic

Additional Notes: NetFlix Cashflows

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ic

Forecasting Cashflows: Top-Down Approach For Ford

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ic

Additional Notes: Ford Cashflows

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ic

Practical Tip: Decompose Forecasting By Major Lines-of-Business

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ic

Comparables Approaches to Forcasting Cashflows

Forecasting w/ Comparables: Jeremy Stein (2001) describes a strategy for obtaining forecasted cashow estimates using a panel regression technique. We slightly adapt his process here. The steps are: Step #1: Identify a set of N comparables for your rm/project. Step #2: Estimate a forecasting model of expected cashows using these comparables and historical data. The model Stein uses is:
0 @ CFnt An,t1 1 0 CFn,t1 An,t2 1 0 CFn,t2 An,t3 1 0 CFn,t3 An,t4 1 0 CFn,t4 An,t5 1 A + nt (17) A = aq + b @ 1 A+b @ 2 A+b @ 3 A+b @ 4

where n indexes comparables and t is measured in quarters. There are N T observations in this regression where T is the number of quarters of data included. (Stein uses 5 years of data so T = 20 in his case) Step #3: Use the estimated forecasting model to forecast future cashows for your rm/project (additional details on next slide). Question #1: How do we identify comparables? Stein splits his sample of rms into terciles of size, then terciles of protability, then terciles of industry risk, then terciles of equity volatility. This creates 34 = 81 buckets of comparables. Sometimes referred to as 34 -Comparables. Alternatives suggestions? There are many reasonable options.
Jiro E. Kondo (McGill) Implementing Basic Valuation Winter 2012

Lecture Note 1

Part Ic

Comparables Approaches to Forcasting Cashflows (Contd)

Question #2: How do we forecast future cashows using the panel regression results? Show approach via example. Suppose the estimates of the forecasting model (17) are:
0 @ CFnt An,t1 1 0 CFn,t1 An,t2 1 0 CFn,t2 An,t3 1 0 CFn,t3 An,t4 1 0 CFn,t4 An,t5 1 A + nt (18) A = aq +0.5 @ A +0.2 @ A +0.1 @ A +0.3 @

Why is the fact that b4 > b3 not surprising? Get CFn,t+1 : Suppose aq for that quarter is -0.001. Let the three previous normalized cashows be 0.05, 0.03, 0.02, 0.06. Ant = 250m. Then:
CFn,t+1 = 250m (0.001 + 0.5 0.05 + 0.2 0.03 + 0.1 0.02 + 0.2 0.06) = 12.5m {z } | Equals 0.05 (19)

Get CFn,t+2 : Suppose aq for that quarter is -0.004. Let the three previous normalized cashows be 0.05, 0.03, 0.02, 0.06. An,t+1 = 250m. Then:
CFn,t+2 = 250m (0.004 + 0.5 0.05 + 0.2 0.05 + 0.1 0.03 + 0.2 0.02) = 10m {z } | Equals 0.04 (20)

Note that we have assumed zero asset growth in this step. If you want to incorporate asset growth, you need to build a predictive model of assets along with cashow. This Techique is Useful Beyond Cashow Forecasting: See Stein paper...
Jiro E. Kondo (McGill) Implementing Basic Valuation Winter 2012

Lecture Note 1

Part Ic

Forecasting Cashflows: Steins 34 -Comparables for Dell (2001)

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ic

Forecasting Cashflows: 1-Yr CF Dists Using 34 -Comparables

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ic

Mini Case (Part 2): Forecasting Public Pension Liabilities

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ic

Mini Case (Part 2): Key Issues

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ic

Mini Case (Part 2): Analysis

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ic

Visualization: Value of An Existing Workers Liability

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ic

Visualization: Forecasted (Aggregate) Liabilities Over Time

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ic

Summary of Novy-Marx and Rauhs Main Findings (Table IV)

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ic

Terminal Value: A Way to Avoid Forecasting All Cashflows

Terminal Value and NPV: Rather than forecasting all cashows for a company/project, it can be useful to stop the detailed cashow forecasts after an early period and estimate the continuation or terminal value of the company/project after the last forecasted cashow. Given these, NPV is given by: N P V = CF0 + CF1 CFT + ... + (1 + E [r ])1 (1 + E [r ])T + T VT (1 + E [r ])T (21)

In essence, terminal value is given by: T VT = CFT +1 CFT +2 + + ... (1 + E [r ])1 (1 + E [r ])2 (22)

Estimating Terminal Value: A Gordon Growth Approach... We can use a growing perpetuity formula to estimate terminal values... CFT (1 + gstable ) T VT = (23) (E [r ] gstable ) where gstable is often estimated from stable (i.e., mature) comparables in the companys industry. This growth rate can be estimated from historical data or from comparables protability and investment policy. One way that the latter can be quantied is using the following formula adapted from the Gordon growth model: stable ROA stable gstable = P BR (24) is the fraction of cashow that is reinvested in the rm and ROA is calculated as the ratio where P BR of cashow to book value of assets.
Jiro E. Kondo (McGill) Implementing Basic Valuation Winter 2012

Lecture Note 1

Part Ic

Terminal Value: Additional Notes

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Ic

HQ: Should We Completely Rely on the Forecasts of Others? Dilbert: Probably Not...

Many practitioners and academics agree with Dilbert.

Poterba-Summers (1995) survery and nd that CEOs use excessively high discount rates partially as fudge factors to account for optimistic forecasts by division managers. Proper way to account for this? Two points of view... See syllabus for detailed references. Fischer Black (1998) argues that we should adjust division managers CF forecasts downwards. Richard Ruback (2011) argues that we should occasionally adjust both division managers CF forecasts downwards and discount rates upwards.
Jiro E. Kondo (McGill) Implementing Basic Valuation Winter 2012

Lecture Note 1

Part Id

Topic 1 (Part Id): DCF Is Not A Number: Sensitivity Analysis

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Id

Sensitivity Analysis: Two Main Applications

Sensitivity Analysis: How Does Value Vary w/ Alternative Baseline Assumptions? What Baseline Assumptions? Cashow Drivers: Prices (revenue and/or cost), quantities, capacity, tax-rates, etc. Discount Rate Drivers: Betas, risk-premia, tax-rates, etc. Main Applications: Three often used applications are... Identify Key Drivers of Value: What parameters most inuence company/project value? Should inuence which variables you focus most on in the cashow forecasting and discount rate estimation stage. (I wont go over an example of this though...) Get Trigger Points For Decision-Making: E.g., At what cost of carbon emissions credits should a power plant switch to emissions-minimizing energy production facilities? Should inuence your capital budgeting decisions and let you know which market variables to focus on in the short-term. Assess a Range of Plausible Values: E.g., Whats a plausible range of fair values for an acquisition? Would the loss implied by the lower range be acceptable? Plan of Attack: Well gain comfort w/ this topic via examples.

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Id

Trigger Points in Consumer Choice: Mazda3 vs. Toyota Prius

Exercise 7: You have decided to purchase either a Mazda3 or a Toyota Prius and will base your decision on minimizing the NPV of car-related payments. You consider two types of costs (car payments and fuel costs) and one type of revenue (car sale after 4 years). The details of these costs are included in the table below. Assume that gas costs $4.20 per gallon and that the discount rate is 5% expressed as a monthly-APR. Category Car Costs Fuel Use Subcategory Downpayment: Monthly Loan Pay.: # of Months: City Mileage: Monthly City: Hwy Mileage: Monthly Hwy: Revenue: At Month: Mazda3 $3,000 $333.33 36 21 400 29 200 $10,000 48 Prius $4,300 $396.10 48 48 400 45 200 $16,750 48

Car Sale

Which car should you buy? At what (trigger) gas price do you switch from purchasing a Mazda3 to buying a Toyota Prius? Side Question: Where can we get a forecast for the revenue of the future car sale?
Jiro E. Kondo (McGill) Implementing Basic Valuation Winter 2012

Lecture Note 1

Part Id

Breaking Down The Problem Into Simple Steps

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Id

Excel Tip: Using Solver To Do Algebra

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Id

This Can Be Relevant...

Chart: Chicago Area Gas Prices

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Id

Typical Sensitivity Analysis: InBev Buys Some BUD

Exercise 8: On July 13, 2008, Anheuser-Busch (BUD) agreed to be purchased by InBev in a transaction that would form the worlds leading global brewer. BUD shareholders would receive $70 per share in cash (for an aggregate equity value of $52bn). Citigroup Global Markets acted as a nancial advisor to BUD in this transaction and, among other things, performed a valuation analysis to assess the fairness of InBevs oer. The DEFM14A table below provides cashow estimates for the rm from the end of 2008 till the end of 2012.

Citi estimates a WACC of 7.5% for BUD and a stable growth rate of 2.0% starting at the end of 2012. If BUDs debt has a market value of $10.1bn (at the end of 2008), whats Citis estimate of BUDs equity value (at the end of 2008)? (Assume the transaction takes place at the end of 2008) Of course, Citi isnt certain of its WACC and stable growth rate estimates. Perform a sensitivity analysis with a WACC range of +/- 1.5% (increments of 0.5%) and a stable growth range of +/- 0.5% (increments of 0.25%).
Jiro E. Kondo (McGill) Implementing Basic Valuation Winter 2012

Lecture Note 1

Part Id

Excel Tip: Using Data Tables To Perform Sensitivity Analysis

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Id

Determine a Sensible Range for Vars in Sensitivity Analysis?

Comparables-Based Approaches: Some Examples...
WACC: E.g., Range for rms WACC implied by the range of WACCs estimated for comparable rms (both pure comparables and synthetic ones constructed from a bottom-up analysis). High Growth Period for a Start-Up: E.g., Range for terminal date implies by 10th and 90th percentiles of high growth period for a sample of past IPOs. (Must dene: (i) growth of what (e.g., sales)? (ii) when the phase ends (e.g., two consecutive years of sales growth below 5%)?)

Statistics-Based Approaches: Some Examples...

Beta: E.g., Range for rms beta implied by a condence interval around beta estimate from a , + ]. (This is a 68% condence interval. Generally avoid the regression. Usually use [ 95% interval since this often produces a very wide range) Mkt Premium: E.g., Range for mkt premium implied by a condence interval around the historical average of realized market returns. Again, often use +/- 1 standard deviation of this return.

Qualitative Approaches: An Example...

Survey-Based: E.g., Range for market premium implied by WSJs survey of forecasters in Topic 1 (Part 1B). The range was 1.0%-4.2%.
Jiro E. Kondo (McGill) Implementing Basic Valuation Winter 2012

Lecture Note 1

Part Id

Getting a Range for WACC: One Approach

Exercise 9: Lets estimate a range for the WACC of BUD ourselves. In performing the calculation, you will use the CAPM to estimate BUDs cost of equity (with the S&P500 as the market portfolio and 5-years worth of returns using price data from June 1, 2003 till July 1, 2008). Assume that BUD has a cost of debt of 5.7% and that the WACC-weights are equal to those implied by the transaction terms (i.e., D = 10.1bn and E = 52bn). Considering only a range implied by variation in beta, is the WACC range consistent with Citis estimate of beta? (BTW, another of BUDs advisors on the transaction, Goldman Sachs, estimated a WACC of 8%) How would your range change if you included variation in the market premium?

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Id

Getting a Range for WACC: Additional Notes

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Id

Overview of Topic 1 (Part I)

Review of NPV w/ WACC:

Estimating the Cost of Capital in Practice:

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Id

Overview of Topic 1 (Part I): Continued

Forecasting Cashows: Key Drivers

DCF is Not A Number: Sensitivity Analysis

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012

Lecture Note 1

Part Id

Overview of Topic 1 (Part I): Additional Notes

Jiro E. Kondo (McGill)

Implementing Basic Valuation

Winter 2012