Why VCs don’t value companies by discounting cash flows

Last week as part of our 50 Questions series I wrote about how VCs value businesses by working back from expected exit values, using various industry rules of thumb and responding to market dynamics (hotness of deal and market).  In the comments Senith asked why we don’t use discounted cash flow (DCF) analysis which as he says is “the most sound method [for valuing businesses] according to most textbooks”.  I’m going to address that today.

DCF analysis estimates the value of an investment, project, or company by adding up the projected cash flows over time and then discounting them to account for the time value of money (£1 in ten years is worth less than £1 today) and also for the risk that the projected cash flows might not be achieved.  Purists will correctly argue that the discounting should only reflect the time value of money and that the cash flows should be adjusted to reflect the risk, maybe using scenario analysis, but in practice most people conflate the risk and the time value of money into a single discount rate.  The wikipedia article on DCF is pretty good.

A DCF analysis of the value of a VC investment in a startup needs four inputs:

  • The amount of the investment (including any follow-ons)
  • The projected return when the investment is sold
  • The years in which each of these cash flows will occur
  • The discount rate

The problem is that whilst the methodology might be theoretically more rigorous than the simpler working back from projected exit values using a target multiple return the input variables are sufficiently uncertain that the extra complexity doesn’t bring you any extra accuracy.  Figuring out the right discount rate is particularly difficult and no better in practice than working with multiples.

One of the strengths of DCF analysis is that it takes account of time, which is very important with projects that are long and/or of variable length.  That is less important for startup investing when the projected exit is nearly always 3-5 years away.

Ultimately VCs are measured on IRRs (Internal Rates of Return) and we do link our multiple analysis to a projected IRR for each deal, but that is typically done towards the end as a final check using simple lookup tables.  I mention it now for completeness given the link between IRR analysis and DCF analysis, but I won’t go into detail because I think this post is long enough and technical enough already.

Enhanced by Zemanta
  • The trick to using DCF to model high variability outcomes is to not type in fixed forecasts in the boxes. Rather create a model which links the various boxes (the mere act of doing this will give you invaluable insight into the company’s cash generation dynamics).

    Then you simply run the model under various scenarios to see how the valuation changes.

  • Given the uncertainty of the scenarios the extra rigour wouldn’t give a better answer.

  • It depends on what you want to achieve.

    From my experience it can be very enlightening to see how revenues, fixed and variable costs, and ultimately CF play out under various scenarios.

    You start thinking about the company as a system – and half the trick to successful investing is understanding how the ‘system’ will behave in different circumstances.

  • That can make sense (and be valuable)with companies with a at least few million of revenue and a sophisticated finance dept, but we are straying into private equity rather than venture territory

  • Just as the laws of nature apply to objects big and small, the laws of finance do not discriminate based on company size.

    Questions such as “will costs rise in line with revenues” represent the very essence of cash flow analysis.

    Take Google and Salesforce. It’s no coincidence Google has minimal human sales or customer support staff. Machines do most of the work which means Google transforms revenues into cash very efficiently, even as it grows. Salesforce on the other hand, whilst being an excellent company, needs to add support and sales staff as its business expands.

    A thorough cash flow analysis of even an embryonic business is not a fruitless endeavour. The final answer(s) make not be reliable, but the exercise can yield insights which go far beyond a simple number preceded by a pound sign.

  • We look at the scalability and inherent profitability of companies and factor that in to our exit analysis, but for small comapnies detailed analysis in a spreadsheet rarely yields additional insight

  • There’s certainly value in building operational models of the business to learn how it works, how it scales, forecasting etc, but that’s a slightly different topic from whether DCF is a useful means of valuing a startup as an investor.

    Also, all the sensitivity analyses in the world aren’t going to help with the fact that most of the value in a DCF typically lies in the terminal (ongoing) value component, which normally assumes steady state growth (makes no sense in most VC investable startups), or uses some kind of multiple, which raises the obvious question of why you didn’t just stick with a multiple in the first place…

  • I have two seperate comments, so I’m going to put them in two threads.

    First, the idea that DCF is somehow a “proper valuation”.

    Back when I was an investment banker, we hired a number of MBA students. We spent a year deprogramming them from their rigorous and narrow-minded ways.

    The practical truth is that the value of something is the amount someone is prepared to pay for it. As an M&A banker, a DCF is an incredibly valuable negotiation tool. It forces you to:
    – create a set of financial projections which set out all of your assumptions for the business’s operational cash flows
    – to select a group of companies to act as the control group for the beta
    – to think about long term growth rates and when it reaches a steady state

    All of these give a number. Only academics and MBAs think this is a valuation. It’s a starting point. You’ve set out your reasons for valuations, the other side can see them, and then everyone can start negotiating.

    For me, DCFs are a way of formalising your sense of numbers, not an output. In the end, for startups, I would say that the operational forecasting bit is useful. The added process of discounting cash flows is not (which I think is what Nic is saying above, although I’m not 100% certain).

  • Comment 2:
    I was a little surprised to realise that you were talking about discounting the cash flows of the venture investment, rather than the company.

    If you do mean about the investment, then I agree wholeheartedly. Your inputs are so subjective (£ in, £ out, number of years til exit, discount rate), that it can’t add value to your process.

  • Comment 3:
    I work with companies that are a earlier stage than you, generally, and I usually say this:
    “A three year forecast will not tell you where you will be in three years time; you have to build one anyway”

    By going through the process, you start to understand how operational metrics like CPA, conversion rates and ARPU feed through to the bottom line. You learn which of your metrics are most important and which ones are levers that you can pull easily. You can start to benchmark against the opposition to look for quick wins.

    The “Profit” figure in the cell at the bottom right of your three year projections is irrelevant. The insight you gain into your business by building it is useful.

  • While I agree with much of what you say, Steve Blank has convincingly argued that different rules (or more accurately, purposes) apply to startups than corporates.
    I paraphrase his thoughts to be “a startup is a company that has not found its business model.” Moving too rapidly to corporate thinking just because you are, say, three years old is a mistake. You need to be sure you have got your business model working and ready to scale before you do that.

  • Exactamundo

  • The two are related, but it is the investment cashflows that matter to an investor. Estimating them from the comapnies projected cashflows would add another imprecise step, particularly as we generally assume one more round of financing than in the plan.

  • Hi Ced – you can do DCFs on negative cash flows if there is a big positive number coming from the exit at the end.

  • Yes – maybe I should have said this in the post. Because our startups are too far from stability to calculate a terminal value we would have to input an assumed exit value with all the inaccuracies that brings.

  • Hi Amir – good to hear from you!

    I suspect the growth and value of Google would have been underestimated leading to an underestimate of the value at each investment round. Thinking about it another way the VCs all made out like bandits so maybe that is exactly what happened 🙂

  • I would say that combing the VC method (if by that you mean the methods I mention) and then calculating the IRR is probably the most thorough and rigorous approach.