Question on DCF

Hi Hyperion and Tree, thanks for sharing your thoughts.

I buy quite "normal" stocks, so the beta cannot be -1 or 0. The beta for my stocks should be somewhere between 0 and 2.
That was why I decided to take an arbitrary value of 1. Is the CAPM still valid if I assume beta=1? Is it better if I try a range of values like for beta = 0.5 to 1.5?

The full formula I use for the discount rate is:
(Debt/Asset)*Rd + (Equity/Asset)*Re, where:
- Rd is the cost of debt (Can be calculated by taking interest expense/debt)
- Re = Rf + Beta (Rm-Rf)

If I use the "average 10 year E/P for the market index" as the discount rate, won't that mean that the cost of debt be ignored? (The Debt/Asset*Rd part)
How about if I use the average 10 year spread between the E/P for the market index and the risk-free rate as Rm-Rf? Would that be more accurate?

Sorry for the newbie questions.
Until now, I have only been using ratios like P/E and P/B in my analysis.
I am now trying to include DCF in my analysis, because I heard that it is a more objective valuation method (assuming that you get the inputs right!)
This is my first attempt, so I am have some doubts about the values I should use. Please let me know if I am not making sense.

(11-03-2015, 05:24 PM)HyperionTree Wrote: Dear gzbkel,

Hyperion says:
There are 3 choices for Beta, namely 1, 0 and -1, each with different assumptions. Beta = 1 means you expect your stock price to be move in 100% same direction as the market index. Beta = 0 means your stock's price movement is independent of the market index. Beta = -1 means your stock's price movement is in the opposite direction as the market index 100% of the time. So when index up your stock down.

Choosing Beta = 1 means, if the market index PE is high, your discount rate is low. Thus when the market is overvalued, you tend to also overvalue your stocks. This is risky.

Tree suggests:
You might want to use peer group long term PE instead if you want to ignore the CAPM because volatility, as you say, is not risk.

For example, if you have 5 stocks in the same industry, you can construct an index based on these stocks. Then you may calculate the average 10 year E/P and use this to be your discount rate. Without a peer group, you can use the average 10 year E/P for the market index to avoid overvaluing your stock when the market is up. To be safe, test a range of discount rates to calculate your fair value.

There are times when you will be unable to ascertain the discount rate for a stock due to various reasons like market uncertainty or risk free rate policy uncertainty or exchange rate effects. In this case, it is better to look for another stock idea.

Cheers,
Hyperion and Tree

Hi mrEngineer, thanks for the reply.

May I know how you arrive at the ballpark figure of 10%?
Do you vary it depending on factors like the stability of the company or the current risk-free rate?

Thanks!

(11-03-2015, 09:02 PM)mrEngineer Wrote: I have not done many calculations of WACC but I generally use 10% as a shortcut to quickly see what may be the magnitude of DCF. Possibly I may vary my WACC with a sensitivity analysis of +/- 0.5% for 6 increments. The most important thing in my view is to determine what is the margin of safety. Generally I have not seen for companies with P/E above 10 able to justify it's DCF per share value.

(12-03-2015, 08:01 AM)gzbkel Wrote: Hi Hyperion and Tree, thanks for sharing your thoughts.

I buy quite "normal" stocks, so the beta cannot be -1 or 0. The beta for my stocks should be somewhere between 0 and 2.
That was why I decided to take an arbitrary value of 1. Is the CAPM still valid if I assume beta=1? Is it better if I try a range of values like for beta = 0.5 to 1.5?
....

Hi gzbkel,

Hyperion says:
Beta = 1, will give Re = Rf + 1(Rm-Rf) = Rm. Which means you are willing to take the same risks as the market index can offer.

CAPM would be changed with your additional assumption on beta=1 because now beta is not really based on the original derivation in the CAPM model.

It would be easier to change the discount rate directly like 8% to 10% than to change beta for sensitivity analysis.

Tree says:

If I use the "average 10 year E/P for the market index" as the discount rate, won't that mean that the cost of debt be ignored? (The Debt/Asset*Rd part)

You may ignore cost of debt if:
1. You model debt interest and repayment into your cash flow.
2. You are calculating the value of the company to the equity holder only.
See source:
http://aswathdamodaran.blogspot.sg/2015/...-rate.html

The purpose of averaging 10 year E/P is to get the long term E/P over at least one investing cycle of boom and burst. So this is more conservative and would not overvalue a stock even if the current year E/P is too low.

How about if I use the average 10 year spread between the E/P for the market index and the risk-free rate as Rm-Rf? Would that be more accurate?

This question is not clear on what you want to use Rm-Rf for.

If Rm-Rf is used as a discount rate, it is not accurate. If Rm-Rf is used for calculating Beta, you have to follow the CAPM methodology which is to regress a time series of Re-Rf against Rm-Rf where the Rm is the yearly return of market index and not an average return of market index.

if Rm-Rf is used as a spread to add back to current Rf to get the discount rate, it means the assumption is that there is a fix spread between Rm and Rf. Tree is not sure whether this assumption will hold. Further, isn't it easier to just average out E/P than to add a spread to your assumptions?

Surprisingly, PE and PB are actually good indicators and the need of DCF is really questionable. For example, the famous 3 factor model by Fama and Franch shows that low PB together with small cap as a screening criteria, can generate a portfolio with excess return over the market significantly over time.

Are you really sure you need DCF?

Cheers
Hyperion and Tree

(12-03-2015, 10:14 AM)HyperionTree Wrote:

(12-03-2015, 08:01 AM)gzbkel Wrote: Hi Hyperion and Tree, thanks for sharing your thoughts.

I buy quite "normal" stocks, so the beta cannot be -1 or 0. The beta for my stocks should be somewhere between 0 and 2.
That was why I decided to take an arbitrary value of 1. Is the CAPM still valid if I assume beta=1? Is it better if I try a range of values like for beta = 0.5 to 1.5?
....

Hi gzbkel,

Hyperion says:
Beta = 1, will give Re = Rf + 1(Rm-Rf) = Rm. Which means you are willing to take the same risks as the market index can offer.

CAPM would be changed with your additional assumption on beta=1 because now beta is not really based on the original derivation in the CAPM model.

It would be easier to change the discount rate directly like 8% to 10% than to change beta for sensitivity analysis.

Tree says:

If I use the "average 10 year E/P for the market index" as the discount rate, won't that mean that the cost of debt be ignored? (The Debt/Asset*Rd part)

You may ignore cost of debt if:
1. You model debt interest and repayment into your cash flow.
2. You are calculating the value of the company to the equity holder only.
See source:
http://aswathdamodaran.blogspot.sg/2015/...-rate.html

The purpose of averaging 10 year E/P is to get the long term E/P over at least one investing cycle of boom and burst. So this is more conservative and would not overvalue a stock even if the current year E/P is too low.

How about if I use the average 10 year spread between the E/P for the market index and the risk-free rate as Rm-Rf? Would that be more accurate?

This question is not clear on what you want to use Rm-Rf for.

If Rm-Rf is used as a discount rate, it is not accurate. If Rm-Rf is used for calculating Beta, you have to follow the CAPM methodology which is to regress a time series of Re-Rf against Rm-Rf where the Rm is the yearly return of market index and not an average return of market index.

if Rm-Rf is used as a spread to add back to current Rf to get the discount rate, it means the assumption is that there is a fix spread between Rm and Rf. Tree is not sure whether this assumption will hold. Further, isn't it easier to just average out E/P than to add a spread to your assumptions?

Surprisingly, PE and PB are actually good indicators and the need of DCF is really questionable. For example, the famous 3 factor model by Fama and Franch shows that low PB together with small cap as a screening criteria, can generate a portfolio with excess return over the market significantly over time.

Are you really sure you need DCF?

Cheers
Hyperion and Tree

no need to be so complicated. agar agar can already.

Cost of capital = 10yr risk-free rate of govt debt + equity risk premium of 4-5%.

alternatively, you can use your desired minimum rate of return as DCF. 10% or 20% up to you.

so buy only when the market price hits your "intrinsic" value .

(12-03-2015, 08:05 AM)gzbkel Wrote: Hi mrEngineer, thanks for the reply.

May I know how you arrive at the ballpark figure of 10%?
Do you vary it depending on factors like the stability of the company or the current risk-free rate?

Thanks!

Honestly I can't remember. But I vaguely recall one of the expert forumers (either D.O.G or dennis or ???) concluded in the past that 10% is good enough for Singapore stocks. I forgot my disclaimer on the 10%.. it varies quite differently for hong Kong, European and us stocks. If you have access to Bloomberg terminals, you can easily obtain WACC and it's components like beta, equity risk premium and cost of debt (in my view cost of debt is fairly complicated as well for big companies as you do not know their term structure of their debt). Last time I checked in bloomberg for my European company that I worked in, the WACC ranges from 13 to 20%..

Hi Hyperion and Tree

Thanks for taking the time to reply.
I must admit your post got abit cheem for me halfway through, especially the part about "regress a time series of Re-Rf against Rm-Rf". I will go read up more about CAPM. Maybe your post will ring a bell after that.

I want to try DCF, because it is an absolute measure of value, as compared to PE and PB which needs to be compared to similar companies. PE is sensitive to one-time gains and losses, and it is based on accounting earnings rather than cash flow. P/B may not be also relevant to asset-light companies like SGX and Dairy Farm, and the book value is not realized until the company sells off the assets.
I still use PE and PB, but I want to experiment with using one more indicator.
Of course calculating the discount rate is the challenge. Based on the replies so far, I gathered that I should also try a range of values for the discount rate. So I need to find out:
- Appropriate way to calculate discount rate
- The range of values to use.

>This question is not clear on what you want to use Rm-Rf for.

I want to plug Rm-Rf into the Re = Rf + 1(Rm-Rf) formula to calculate the cost of equity.
After getting Re, I can calculate WACC using the (Debt/Asset)*Rd + (Equity/Asset)*Re formula.
So, my original idea for calculating Rm-Rf is to find the average spread between earnings yield and risk-free rate (E/P - Rf) for different period of time.
As you said, there is no guarantee that this spread stays constant, so I plan to take the average over a number of years, like 10 years.
After that, I will add the current risk-free rate (or the forecasted risk-free rate) to get Re.

Is this to be too convoluted? Better to "just average out E/P" as you suggested?

Just thinking out loud. Think I need to study more

(12-03-2015, 10:14 AM)HyperionTree Wrote:

(12-03-2015, 08:01 AM)gzbkel Wrote: Hi Hyperion and Tree, thanks for sharing your thoughts.

I buy quite "normal" stocks, so the beta cannot be -1 or 0. The beta for my stocks should be somewhere between 0 and 2.
That was why I decided to take an arbitrary value of 1. Is the CAPM still valid if I assume beta=1? Is it better if I try a range of values like for beta = 0.5 to 1.5?
....

Hi gzbkel,

Hyperion says:
Beta = 1, will give Re = Rf + 1(Rm-Rf) = Rm. Which means you are willing to take the same risks as the market index can offer.

CAPM would be changed with your additional assumption on beta=1 because now beta is not really based on the original derivation in the CAPM model.

It would be easier to change the discount rate directly like 8% to 10% than to change beta for sensitivity analysis.

Tree says:

If I use the "average 10 year E/P for the market index" as the discount rate, won't that mean that the cost of debt be ignored? (The Debt/Asset*Rd part)

You may ignore cost of debt if:
1. You model debt interest and repayment into your cash flow.
2. You are calculating the value of the company to the equity holder only.
See source:
http://aswathdamodaran.blogspot.sg/2015/...-rate.html

The purpose of averaging 10 year E/P is to get the long term E/P over at least one investing cycle of boom and burst. So this is more conservative and would not overvalue a stock even if the current year E/P is too low.

How about if I use the average 10 year spread between the E/P for the market index and the risk-free rate as Rm-Rf? Would that be more accurate?

This question is not clear on what you want to use Rm-Rf for.

If Rm-Rf is used as a discount rate, it is not accurate. If Rm-Rf is used for calculating Beta, you have to follow the CAPM methodology which is to regress a time series of Re-Rf against Rm-Rf where the Rm is the yearly return of market index and not an average return of market index.

if Rm-Rf is used as a spread to add back to current Rf to get the discount rate, it means the assumption is that there is a fix spread between Rm and Rf. Tree is not sure whether this assumption will hold. Further, isn't it easier to just average out E/P than to add a spread to your assumptions?

Surprisingly, PE and PB are actually good indicators and the need of DCF is really questionable. For example, the famous 3 factor model by Fama and Franch shows that low PB together with small cap as a screening criteria, can generate a portfolio with excess return over the market significantly over time.

Are you really sure you need DCF?

Cheers
Hyperion and Tree

mrEngineer, invest4aliving thanks for sharing.
I should try out the Bloomberg terminal the next time I go to the central library. Maybe I can save some time, instead of calculating myself (and possibly getting it wrong)