,
19 tweets,
4 min read
Read on Twitter
0/ I’ve had a love-hate relationships with DCFs. I swore by them, then promised I’d never use them and now use them selectively. I’ll talk about my experiences and how I use them now, in this thread —>
1/ Many years ago when valuing a thermal power plant I decided to use a 2 stage DCF. Higher discount rate for the riskier construction period and lower discount rate for the more stable operating period. Imagine my surprise when the resultant value was obnoxiously high!
2/ The negative cash flows in the construction period got discounted at high rates making them less negative and the positive cash flows of the operating period got discounted at the lower rate making them higher. The terminal value of the final yr cash flow was insanely high 🤦🏻♂️
3/ That day was a lesson on how not to manage risk by discounting with higher rates. I went from DCF advocate to the other side of the spectrum.
4/ The 2nd issue w/DCF is it has a natural positive bias. I have never once seen anyone (me included) forecast declining revenues or margins or worsening wcap metrics resulting in declining cash flows in an investing world where cycles will catch up with you sooner or later
5/ The 3rd issue w/DCF is accuracy of forecasts. I have seen forecasts being met only once in last 12y as a professional investor. It was pure luck, not for original thesis, and only for a division of the company where at a consol level the company missed its forecasts. Badly.
6/ The 4th issue is random selection of the discounting period. Ive seen 3y, 5y, 7y, 10y and 30y. Where you are in the cycle is ignored. The investment banker favourite is 5y. You are almost always at the peak here 😎
7/ 5th: DCFs cannot apply to every business. In my hall of shame is a DCF applied to a shipping co. In 2008. At the peak of shipping cycle. Go figure! ... DCFs work best when earnings are relatively stable. Not cyclicals. Not startups. Not financials.
8/ 6th: Discount rate sensitivity can wildly swing your range. Ref the ex in tweet 1. Also cash flow is assumed at the end of the period which is not reality. Take mid-period. But its still an approximation. Better than quarterly which is an overkill
9/ 7th: If terminal value is 60-70% of your DCF value one or more of the following is true (a) something smells wrong (b) you are paying too high (c) you are working off the bankers model. It could be any one of the above points from tweet 1-8
10/ 8th: You get lost in the math. If you have to calculate so much you end up taking away time from thinking and get sucked into a spreadsheet whirlpool. Your returns are likely going to go drown too ... so how to use / modify / selectively use them?
11/ A: Rather than company, I mostly use them for units ie one store, one tower, one MT of capacity, etc. as if they were set up greenfield. It helps understand and value unit economics in some cases. There are exceptions, cannot use this everywhere
12/ B: When using A above always cross-check the implied multiple by dividing the resultant DCF metric (Mcap or EV) with the relevant P&L metric to get the implied PE, EV/EBITDA, EV/EBIT, EV/Sales.
14/ C: B above will likely be a number for a good asset (since we naturally have a bias for projecting increasing cash flows). Repeat B for flat, declining, average units. And see how the implied multiples begin to move.
15/ You should do C because companies will have a mix of good, average, ugly units. This gets lost when you apply DCF on company level cash flows. It also helps you see multiples in a whole new light
16/ D: I don’t use complex formulae to arrive at a discount rate. I test with a range of rates starting with co’s interest rate, always cross-checking what the implied multiples from B show.
17/ When the prevailing market multiple matches the implied multiple at disc rate=cost of cos debt, you are technically better off having a position in the cos debt rather than equity. But then markets don’t operate on formulae. It’s a signal nonetheless.
18/ E: Terminal value (TV). Do iterations with and without TV and see implied multiples from B. Ref tweet 9 above
19/ Lastly, valuations are about ranges and are only clear at extremes. Need to be triangulated across different methods. There is a huge factor of experience and art to it as well. Like beauty its in the eyes of the beholder and mkts don’t operate on formulae
