19-08-2012, 11:08 AM
Hi all,
I was thinking if one was assessing potential investment candidates for buy and hold capital appreciation, how would one go about doing the valuation since the returns are to come mainly from the future?
Standard finance textbook preaches the discounted projected free cash flow model for valuing companies intrinsic value.
In Mohnish Pabrai "Dhandho Investor", he uses a variation where instead of the standard terminal value computation, he assumes a trading multiple on future projected cashflow for terminal value.
Problem is with DCF projection models, a large part of the intrinsic value comes from terminal value... but given that most business models grow obsolete overtime e.g. Kodak, Video Rentals even so called standard terminal value assumptions like "assume terminal growth rates at GDP/Rate of Inflation" can prove to be too optimistic. Other problems include high sensitivities to assumed growth rates and discount rate.
How do we overcome this problem? (Without some form of terminal value computation, the DCF model cannot work...)
Mohnish method was to buy at a 50% margin of safety on the calculated intrinsic value which appear to me to be quite self contradicting since it means his projected DCF model accuracy and reliability is very low... (might as just set a very low CF/PE mulitple as price target)
Some of my newbie thought on the above as below:
i) There is no need to come up with precise intrinsic value figures... think valuation instead in more probabilistic terms of whether it is possible to achieve desired IRR in light of current available facts .
ii) First come up with a desired rate of return say 15% per annum and an investment horizon say 5 years. In this case, to double one $ in 5 years time.
iii) Look at the current stock price today. Say company sells at $10 per share with 3-5 years average free cash flow per share of $1 (can use some other more relevant proxy metric like EPS, book value, important thing is metric need to be representative and adjusted for cyclicality and one off factors)
iv) Also look at the representative dividend payout ratio. Say company has been paying 20% of FCF for some years.
v) Since goal is to receive $2 for every $1 invested. It means selling at an approximate price of $19 in 5 years times assuming 20 cents in dividend per year.
vi) Since company need to sell at $19 in 5 years time... we can reverse engineer the business fundamental needed to justify the price and can cross check whether such fundamentals are achievable realistically in light of the industry environment and company track record.
vii) In this case, the company will need to generate FCF in the range of S$1.70 to S$2.00 depending on one guess of the future multiple (can look at historical trading range for reference) at which it would trade. More important would be thinking about the associated business fundamentals such as the revenue base, margins, amount of invested assets etc needed to produce that range of FCF.
viii) Think and assess business wise what management has to do in order to attain those revenue base, margins, level of invested assets in light of its current business environment.
ix) In this hypothetical case, the company has to either double the revenue or the margins or some function of both. It also probably has to double its invested operating assets unless margins i.e. ROA can be expanded/increased. Operations wise, this mean management taking actions to increase market share, raising prices, introducing new products, increased productivity on infrastructure/capital equipment etc.
x) Application wise, the method inclined towards a negative selection process. For example, the candidate company existing fundamental profile may be one of a niche provider with little growth space and already high margins/ROA, thus one can reject the company as an investment since it is not probable the fundamentals can improve to such an extent within 5 years (unless the share price of the company now falls like 40% to 50%). One can also use this method to reject things like companies with already very high market share unless new products will be introduced (say 45% of existing products market share) or are demonstrated price takers or stretched balance sheets (unable to get financing to invest the necessary amount of assets) etc. One can also reject if management plans for the future seem to not reconcile with what need to be done.
xi) Positive selection would seem to occur only when the current share price is very low in relation to existing fundamentals and it seem highly probable from one assessment that the improvements to fundamentals needed to reach/justify the desired exit price are not too aggressive and is within reality.
xii) In the end, my method is still banking on future fundamental performance which is uncertain and unknown. Different from DCF modelling only in that I do not attempt to come up with specific year to year forecasts nor compute a specific intrinsic value, nor seek some specific % MOS from computed intrinsic value. The go and no go is almost entirely determined by probabilistic assessment of business factors driving the chosen valuation metric.
Not sure if I being naive or illogical here or is over simplifying the valuation process... what am i missing here? Appreciate and be grateful for any comments fellow buddies may have. Thanks and sorry for the long post.
I was thinking if one was assessing potential investment candidates for buy and hold capital appreciation, how would one go about doing the valuation since the returns are to come mainly from the future?
Standard finance textbook preaches the discounted projected free cash flow model for valuing companies intrinsic value.
In Mohnish Pabrai "Dhandho Investor", he uses a variation where instead of the standard terminal value computation, he assumes a trading multiple on future projected cashflow for terminal value.
Problem is with DCF projection models, a large part of the intrinsic value comes from terminal value... but given that most business models grow obsolete overtime e.g. Kodak, Video Rentals even so called standard terminal value assumptions like "assume terminal growth rates at GDP/Rate of Inflation" can prove to be too optimistic. Other problems include high sensitivities to assumed growth rates and discount rate.
How do we overcome this problem? (Without some form of terminal value computation, the DCF model cannot work...)
Mohnish method was to buy at a 50% margin of safety on the calculated intrinsic value which appear to me to be quite self contradicting since it means his projected DCF model accuracy and reliability is very low... (might as just set a very low CF/PE mulitple as price target)
Some of my newbie thought on the above as below:
i) There is no need to come up with precise intrinsic value figures... think valuation instead in more probabilistic terms of whether it is possible to achieve desired IRR in light of current available facts .
ii) First come up with a desired rate of return say 15% per annum and an investment horizon say 5 years. In this case, to double one $ in 5 years time.
iii) Look at the current stock price today. Say company sells at $10 per share with 3-5 years average free cash flow per share of $1 (can use some other more relevant proxy metric like EPS, book value, important thing is metric need to be representative and adjusted for cyclicality and one off factors)
iv) Also look at the representative dividend payout ratio. Say company has been paying 20% of FCF for some years.
v) Since goal is to receive $2 for every $1 invested. It means selling at an approximate price of $19 in 5 years times assuming 20 cents in dividend per year.
vi) Since company need to sell at $19 in 5 years time... we can reverse engineer the business fundamental needed to justify the price and can cross check whether such fundamentals are achievable realistically in light of the industry environment and company track record.
vii) In this case, the company will need to generate FCF in the range of S$1.70 to S$2.00 depending on one guess of the future multiple (can look at historical trading range for reference) at which it would trade. More important would be thinking about the associated business fundamentals such as the revenue base, margins, amount of invested assets etc needed to produce that range of FCF.
viii) Think and assess business wise what management has to do in order to attain those revenue base, margins, level of invested assets in light of its current business environment.
ix) In this hypothetical case, the company has to either double the revenue or the margins or some function of both. It also probably has to double its invested operating assets unless margins i.e. ROA can be expanded/increased. Operations wise, this mean management taking actions to increase market share, raising prices, introducing new products, increased productivity on infrastructure/capital equipment etc.
x) Application wise, the method inclined towards a negative selection process. For example, the candidate company existing fundamental profile may be one of a niche provider with little growth space and already high margins/ROA, thus one can reject the company as an investment since it is not probable the fundamentals can improve to such an extent within 5 years (unless the share price of the company now falls like 40% to 50%). One can also use this method to reject things like companies with already very high market share unless new products will be introduced (say 45% of existing products market share) or are demonstrated price takers or stretched balance sheets (unable to get financing to invest the necessary amount of assets) etc. One can also reject if management plans for the future seem to not reconcile with what need to be done.
xi) Positive selection would seem to occur only when the current share price is very low in relation to existing fundamentals and it seem highly probable from one assessment that the improvements to fundamentals needed to reach/justify the desired exit price are not too aggressive and is within reality.
xii) In the end, my method is still banking on future fundamental performance which is uncertain and unknown. Different from DCF modelling only in that I do not attempt to come up with specific year to year forecasts nor compute a specific intrinsic value, nor seek some specific % MOS from computed intrinsic value. The go and no go is almost entirely determined by probabilistic assessment of business factors driving the chosen valuation metric.
Not sure if I being naive or illogical here or is over simplifying the valuation process... what am i missing here? Appreciate and be grateful for any comments fellow buddies may have. Thanks and sorry for the long post.