Hacker News new | past | comments | ask | show | jobs | submit login
Valuing high-tech companies (mckinsey.com)
129 points by pmcpinto on Feb 29, 2016 | hide | past | web | favorite | 35 comments

I worked at a McKinsey once upon a time, and this is exactly how we did valuations. But this article elides the fact that with this kind of model, you can quite literally get any output you want from choosing your assumptions. In my view, it was better used as a "what assumptions would you have to believe to believe in a given valuation" rather than a valuation approach per se. It can be helpful to know that you'd have to have untenable assumptions to justify a given valuation. In any case, I bet the DCF model generates lower values than the private markets are paying, so it's only interesting to such companies in the public markets where you can act on your beliefs by going short.

Thanks for the comment. Though I am wondering, as a consulting firm not a finance firm, why does McKinsey model and value companies? Are they brought into double check?

I worked in a different consulting company.

We're brought in to double check, but also do conduct due diligence and valuation. There's some overlap between financial services and consulting companies. In some deals, you might have a bank on one side and a bank + consultants on the other.

Also, banks are good at doing standard, off-the-shelf valuations. They usually know little about the specific industry and use one-size-fits-all models.

If you want a detailed valuation, taking in account scenarios, industry changes, etc. you need people who know the industry pretty well, which usually means hiring consultants (who either are industry experts, or hire external industry experts).

In some cases, private equity firms also fill in the role of the bank in valuations. I've worked on a multi-party investment deal where one company had consultants (us), another had a bank, and another had a PE firm helping them. Looking at the different approaches and models from each company was a very interesting experience.

Exactly. I was a partner at McK and, at the time, valuation engagements were an important line-of-business, both for private equity firms or in tandem with bankers (M&A, IPOs...).

It's a strategy consulting firm and often strategy involves valuation; like should we buy a competitor or divest a division. Generally if the strategy went all the way through to an action, an investment bank would get involved to finalize things.

Would you trust a real estate agent who is paid only if the commission occurs? Both the buyers and sellers bankers only get paid on a transaction. Paying for the time can give an independent valuation. Also, many times valuations don't impact transactions.

That is kind of how it works when selling a house with a real estate agent.. they only get paid on a sale, and they went to sell as cheap as possible so they get their money.. see https://www.youtube.com/watch?v=17jO_w6f8Ck

That's my point - you don't trust them. And shouldn't trust investment banks who get paid on transactions for coming up with a correct price. (Hence you call McKinsey for the 2nd opinion since they get paid on time rather than transaction.)

> That's my point - you don't trust them.

Actually most people DO trust them. I would eat my hat if 90% of people don't agree to sell their house at what the listing agent suggests.

Fair enough. Maybe I'm just the cynic. The real estate agent trying to sell my Mom's house came up with a dozen reasons on why it should sell for mid-300s. I did the analysis on the same data and came up with mid-400s. We'll see where it lands.

Ya. Usually what it comes down to is you need to wait. I bet your agents price would sell within a month.

I asked for a higher price on my house, and it took about 90 days to turn it over... was worth the wait though, say 15k more in my pocket.

> ... it's only interesting to such companies in the public markets where you can act on your beliefs by going short.

I disagree. As Keynes stated, the market can stay irrational longer than you can stay solvent.

I'm curious - how did you determine defensible discount rates at McKinsey?

Former Bain consultant. We used comps from public companies to sense-check things.

The interest rate for an energy conglomerate would not be something you would also use for a non-diversified tech company like Dropbox. As long as we could sell the number (both internally and to the client).

But within like OP said, the range of defensible numbers is still quite wide.

Did you ever use WACC, or is that not realistic?

We did use WACC, which invariably led to arguments about how to correctly calculate WACC as well. For a division of a large company, it can be non-trivial to decide what the appropriate WACC is.

The title uses the word "high-tech", while the article uses "Hot startup", I doubt if Yelp can still be classified as a "Hot startup". Answering whether Yelp is truly "high-tech" does not matters since the article makes no assertion that relies on assuming Yelp to be "High-Tech".

The article is selling standard business analysis, packaged as if it was some kind of brilliant visionary insight. This "Retro analysis"completely ignores new streams of businesses that Yelp is getting into such as Food delivery. These typically have much higher revenue albeit at lower margins, than just advertising. Not factoring it is a gross miscalculation. The article also fails to compare Yelp with Grubhub. Which would have produced a much more realistic expectation.

The article reminds me of this video by Steve Jobs discussing how Marketing & Sales end up destroying companies. https://www.youtube.com/watch?v=-AxZofbMGpM

I don't think their analysis "ignores new streams of business". They specifically highlight how the different potential markets should be approximated.

Not only that but Yelp "getting into" a new line of business or revenue stream does not mean you get to say "it's a $50 billion market we're getting into so clearly we can add $5 billion to Yelp's value!"

Until it's generating something (e.g. cash on the balance sheet or some other tangible benefit) I don't see how it would affect valuation at all.

Totally agree in principle, but it does seem like just "getting into" the food delivery business is a humorously quick way to get a $X00M valuation in the private markets.

Well YELP is public. Public markets can also be quite irrational.

> Whether this price is appropriate depends on your confidence in the forecasts and their respective probabilities.

Did I miss something or did the authors not mention a discount rate, or Weighted Average Cost of Capital (WACC)? Unless I'm wrong, a discount factor is needed to calculate the present value, regardless of using a probability-weighted DCF or a simple DCF model.

If true, then without more information on the WACC, the price they state does not only depend on our confidence in the forecasts and their respective probabilities.

The target audience for this article is executives at corporate America, who understand very little of tech. It is helpful to understand how the rest of the world looks at tech.

For those of you who want to learn more and don't want to waste money on a finance text book, check out Aswath Damodoran's website. Pretty much everything you need to know about valuation finance can be found there, along with a lot of data.

Importantly he has a lot of sample models and calculations which will help make the principles "real" for those of you who prefer details to simplistic high-level concepts.

Do analysts ever share their data models publicly? It'd be interesting to see the data and be able to play with numbers along with these kinds of reports.

If you are looking for how these models are created, you can look at the work of well-regarded experts in the valuation field like Damodoran, whose techniques many at i-banks or consulting firms would learn at some point in their careers. This link is to Damodoran's post for LinkedIn and GoPro (with downloadable, completed Excel models, which can be used as templates for other valuations):


Yes, sell-side (i.e. brokers like Goldman, Morgan Stanley, etc.) analysts do share their models with clients. They do this both in 'hardcopy,' pasted into published research reports, as well as 'softcopy,' sending or publishing the Excel file. Not exactly mainstream distribution, but 'public' in the eyes of the SEC and something you can find if you look for it.

You'll never see any of their data models or data

Really? I find it very surprising that investors would follow such guidance without being able to see the numbers. How are businesses who produce this kind of report evaluated? Do people count the number of times they predicted correctly or incorrectly?

I have a good friend who worked in investor relations for a tech company that IPO'd. He worked with these analysts and they would send him models. I'm told that most of these analysts were juggling 5+ companies and couldn't devote much time to any particular company without neglecting the others. Due to this, their models were often very poor or were missing large, publicly known items.

Something to think about.

Some stories are incredible, for example the Dragon horror story that was featured here a few weeks ago:



> I find it very surprising that investors would follow such guidance without being able to see the numbers.


> How are businesses who produce this kind of report evaluated?

These reports are made by sell-side analysts, and the banks are evaluated by how much profit their buy-side analysts can make. See a disconnect?

> Do people count the number of times they predicted correctly or incorrectly?

Google "analyst ranks", so I guess once a year, but investing is mostly a relationship-based environment so the actual ranks might not make a difference that often

I will say Dave Wessels is real good at this

In my opinion, you can only value tech companies relative to each other, because as a whole their price/earnings ratios are all too high. For example, if you sum the value of the entire solar power industry, you get roughly the value of Twitter. That does not make sense to me.

The money is made by being able to predict the future, be that lottery numbers, disasters or the future earnings of companies.

Modelling startups is hard, because there is so much estimation and variance. But the fact that it's hard is why its so important.

Thanks for sharing, it's quite handy to see the thought process written down, quite educational.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact