March 29, 2006

Common errors in DCF models

Found this great article from Michael Mauboussin, Chief Investment Strategist of Legg Mason Capital Management (LMCM). It is a 12 page article on the common errors investors commit in using the DCF (Discounted cash flow) model.

Personally my approach to valuation (which is not original and mainly developed from reading) is to create a DCF model for three scenarios. I extend the current business condition and create an as-is scenario. So the assumption is that the current growth rates, margins, competitive situation etc will continue as is. The second scenario is an optimistic scenario where in I try to calculate the intrinsic value using the most optimisitic assumptions for growth rates, margins, competitive intensity etc. The third scenario is the pessimistic scenario with poor growth rates, high competitive intensity etc.

I try to associate probability against each scenario and try to calculate the expected value.

So expected value is = intrinsic value (as is) * probability for ‘as is’ + instrinsic value (optimistic scenario)* probability for optimisitic scenario + intrinsic value (pesimistic scenario) * probability for pessimistic scenario.

I also cross check the above expected value with ratio based valuations.

The above approach forces me to think harder on all my assumptions. Also when the annual results are declared for any company I have invested in, I go back to my excel spreadsheet and relook at the numbers, assumptions etc and calculate the new intrinsic value again. This gives me an idea on whether I should sell, buy more or hold.

I am not able to post my valuation / analysis spreadsheet on the blog. If any one is interested, please e-mail me on


Anonymous said...

I've sent you an email to request for the spreadsheet. Please check


Prem Sagar said...

I am looking at the spreadsheet. I am trying to understand it fully. Fantastic work on the blog. Keep posting..

abhi said...

The hardest part in DFC model I find is compute a opportunity cost or cost to you capital? and make assumtion. It sometimes makes me feel that it is taking me away from at reality.
Model could easily becomes weaker becasue your assumtions are could be wrong even if you do logical analysis of assumtion.
Does any one share this opnion as well?

Rohit said...

I have taken the approach suggested by buffett/munger for the cost of capital issue. I have chosen a hurdle rate of 13% for my cost of capital. so if i buy at the intrinsic value, then i would theorotically make 13 % per annum. In addition i add a margin of safety to take care of inevitable mistakes

You are right that the model is as good as its assumptions. i have seen a similar question posed to buffett/munger and other investing greats and most of them respond that there are 2 ways to take care of it
- understand the industry/business as well as possible so that the cash flows can be estimated well.
- be conservative in making your assumptions

In addition, i think buffet's approach / analogy to 1 foot v/s 7 foot jumps is great. Dont try to evaluate or invest in companies which one cannot predict or understand. It would end in weak assumptions and wrong results from the model

Market Participant said...

While DCF models are nice, I tend to think that they are way too sensitive to assumptions about discount rates and cost of capital. It's possible to make a DCF model give you value you want based on what assumptions you use.

Determining the cost of equity capital is very tricky, and always full of handwaving.

What I prefer to do instead is to do an ROC decomposition.

(NOPAT/Sales)*(Sales/Operating Assets)*(Operating Assets/Equity)

And see how the company compares to other similar companies on each of those attributes, from doing that you can get a sense of relative value of the company. If the company has experienced shocks to one of these factors you can easily incorporate that into your analysis.

Rohit said...

i would say DCF should be one of the tools only and not the only tool. Also the no.s thrown by the DCF model should be compared with what one is getting from ratio based valuation models, compare the valuation with similar companies in the industry.
I have typically found the sensitivity based DCF models not too useful.
The model can throw up any number you want based on growth, competitive advantage period , ROE , Cost of capital and other assumptions.
So for me typically the DCF model comes towards the end, where i have some basis for the above assumptions (after having done the necessary qualitative analysis).