Charlatanry in Valuation
Corporate valuation is not an exact science. While we
can observe the price of a stock, we cannot observe the value of a company. It
is a subjective measure, depending on the perception of risk and potential in
the eyes of the investors. This opens the door to many different valuation
approaches. To make it even more complex, we can account for probabilities of
drugs to reach a certain development stage, i.e. the success rates.
Understanding probabilities is a difficult task for the human mind and we will
see in the following article that some minds already stumble over much easier
concepts. We show some examples – sometimes funny, sometimes even frightening –
how valuation should not be performed. We have all these “concepts” in one or another
of our valuation projects. Some of these pitfalls have been discussed in
earlier newsletter articles; others will doubtlessly be the topic of future articles.
We will discuss the different methods with an example
company that has only one phase II project with the following parameters:
Phase II

2 years, US$10Mio, 40%

Phase III

3 years, US$50Mio, 60%

Review

1 year, US$4Mio, 90%

Sales

300 peak sales, 65% op
margin

Company

25% tax rate, 50
employees

The standard rNPV valuation at a discount rate of 17%
yields a pretax value of US$ 39 Mio, and a posttax value of US$ 25 Mio.
We
use the riskfree rate for discounting in real options. (AR 2004, Astra Zeneca)
The concept of riskfree discounting has been adopted in many publications about real options. The origin of this idea is always the similarity between real and financial options. Financial options are financial instruments, with an underlying that is traded and liquid. Consequently it is possible to design a portfolio with the financial option, the underlying asset, and a bond that allows to hedge away all risk (in theory). Therefore we can apply riskneutral discounting. In drug development this is clearly not the case. The underlying of the option is the potential future sales of the drug. There is no instrument in the market that allows efficient hedging of these sales, which are uncertain in time, size, and event. Companies using real options and riskfree discounting thus run the risk of significantly overvaluing their projects. Our example company would have a pretax value of about US$ 187 Mio if we use 5% risk free discounting.
If
you use success rates, then you have considered already the attrition risk and
you should discount with a pharma discount rate. (CFO of a listed biotech company)
This would mean that the discount rate should be the
same for pharma and biotech. The argument is that both pharma and biotech
companies depend on the same underlying market, drug sales. However, there are
two main differences between pharma and biotech. The value of a pharma company
is mainly made up by products that are already on the market. Biotech companies
have their projects still in R&D and their sales potential is not yet
observable. The sales projections for biotech companies with their projects in
development are highly uncertain, regardless of the success rates, which have
to be accounted for on top of that. The second difference is that a pharma
company’s value is driven by the next few years’ benefit. These years are
certainly easier to estimate than the value relevant years for a biotech
company, which lie 15 to 20 years in the future. And if the biotech company
wants to get there it has to finance the same period of lossmaking R&D.
This increased estimation uncertainty and the implicit leverage of mandatory
R&D expenditures before reaching profitability lead to a further increase
of the discount rate for biotech companies. Finally, we can also observe a
general riskaversion of investors. Investing in a biotech company is like
betting on the company’s lead compound, or on its technology. Investing in a
pharma company corresponds rather to placing bets on various products. Imagine
playing with a biased coin (15% chance for head, 90% chance for tail). If it’s
head you win $1,000, otherwise you don’t win anything. How much would you pay
to participate? The fair price would be $150, but typically you would say that
in most of the cases you will loose anyway, so better don’t put too much at
stake. And now compare this game with the following: you play 100 times head or
tails with the biased coin, and fro each head you get $10. Again, if all coin
flips show head you win $1,000. But as shown in the graph the chance of loosing
everything is much lower. The forecastability of the payoff has increased tremendously.
While for the first game we were sure to be off from the actual payoff ($0 or
$1,000) by a large amount, in the second game we are pretty sure that the final
payoff will be somewhere between $100 and $200. In a simulation of 1,000 scenarios
(cf. figure 1) the payoff never fell out of the range of $60 to $290. This is
quite a difference to the range of the first game, which is $0 to $1,000. As a
consequence, in the second game we are ready to pay a price that is closer to
the fair price of $150. The first game corresponds to investing in biotech,
where you place a bet on one project. The second game corresponds to investing
in pharma, where you are not so much dependent of the fate of one single
project.
Figure 1: Simulation of 1,000 biased coin flips.
The lack of diversification in a biotech company is usually punished by a higher discount rate. The success rate does not account for the risk aversion. The value of our example project becomes US$ 112 Mio if discounted at 9%.
Terminal Value is 125% of value (Copeland)
Tom Copeland is not the only author who claims that in
R&D intensive companies most of the company value lies in the terminal
value, i.e. that the main chunk of value stems from the time period beyond the
estimated cash flows. Usually it is assumed that from then onwards the company
is able to maintain stable annual cash flows. These stable cash flows are then
summarised in the socalled terminal value. If this terminal value is more than
100% this means that the company would be an unprofitable venture if we don’t
include the cash flows of the time period we do not analyse in detail. Probably
nobody who invests his own money would like to rely on such assumptions. In life
sciences terminal value has also another shortcoming: terminal value assumes a
fair degree of stability of the company. But drug development projects are
subject to a large amount of attrition and it is very difficult to estimate
when a company will have reached a stable state. A stable state corresponds to
a healthy pipeline that is able to compensate for attrition. Will the company
really be already that far by then? Even pipelines of big pharma companies do
not have this property. Is, e.g., Pfizer’s pipeline strong enough to compensate
for the loss of patent protection of Lipitor? We propose an alternative
methodology to capture the value of the company’s ongoing business: Add every
year new projects to the pipeline. These new projects, more or less standard
projects, come from business development (inlicensed projects) or from the
labs. We can estimate this feed rate by looking at the past performance of
business development and research of the company. This way the rather delicate
estimation of the moment when the company reaches a stable state is not an
input but rather an output. You find a more detailed description of terminal
value and why it is dangerous to rely on it in our September 2008
newsletter. If we set the time horizon at ten years then the terminal value
is US$ 50 Mio and the overall project value US$ 39 Mio. A surprising match with
our initial rNPV valuation, and also surprisingly close to Copeland’s figure
(here the terminal value is 129% of the value). However, if we push the time
horizon to 12 years then the company value already becomes US$ 49 Mio. But the
value of the company should not depend on our time horizon.
Value=DCF
without success rates (because compounds are not comparable) but discounted at
50%.
We often see this valuation practice being used for
early stage companies. A Harvard business school article by William Sahlman (“A
method for valuing highrisk, longterm investments”, 2003) gave this method
the apparent academic justification. The discount rate is decomposed into
various components that account for risk, liquidity, added value by the
investor, and cash flow adjustment because of uncertain business outcome. While
we completely agree with the first two components, i.e. risk premium and
liquidity adjustment, we object to the other premiums. The argument for the
value added premium goes as follows: The investor adds his experience to
improve the performance of the company. This premium can be avoided by estimating
the cash flows for the premoney valuation (which is also preinvestor, i.e.
without his valuable input) without the value adding effect of the investor and
his network. This is less arbitrary than a premium to the discount rate. The
cash flow adjustment should account for the attrition. We think it makes much
more sense to use success rates instead of a higher discount rate. Success
rates are available for any disease area, for any compound type, and for each
phase. Success rates allow differentiating between safer and riskier therapeutic
areas. A premium to the discount rate lacks a clear quantification. How much
should this premium be for a preclinical cancer compound compared to a phase II
Alzheimer drug? Finally, cash flows after launch of eth drug have the same
probability. It is unclear why nevertheless with every additional year the cash
flows should be worth about 50% less. For the sake of transparency and credibility
of the valuations we strongly argue to quantify (finally valuation is a
quantification exercise) the impact of each effect. But these effects should
rather have an impact on the cash flows than on the discount rate whose effect
on the overall valuation is difficult to judge. Our example company now has a
value of US$ 2.6 Mio.
In the continuation we exhibit a few even more
adventurous valuation methods:
Comparables
A socalled valuation specialist tried to find the
value of a company by comparing it to companies whose price is known (e.g. the
market capitalisation) based on the number of PhDs: Companies are comparable if
the number of PhDs is about the same. Unfortunately not each technology
requires the same amount of research, and not each researcher adds the same
value. Although completely meaningless, the number of PhDs seems to be a well
observable parameter that our colleagues happily used, because more
valuerelevant parameters were hard to get. We could also cite other
meaningless parameters like P&L, amount of burnt capital, etc. Comparables
are a wellaccepted method for price finding, but it should use criteria that
are indeed valuerelevant. Such criteria can be stage of lead compound, same
therapeutic area, similar technology, or same business model.
Value=#employees*US$1Mio
(business development of a private biotech company)
This valuation method falls into the same category
like the one just above. If it were that easy we would recommend that you hire
a few 1,000 employees. This way you boost your value by US$1bio. Our example
company would be worth US$ 50 Mio.
Posttax value=(1tax)*Pretax value (European investment bank)
This tax consideration is correct for companies that are profitable (and remain so). But for companies, which are not yet profitable the effect is actually worse, because they can only realise their tax assets once they reach profitability, without interests. And for companies that become profitable only in some cases (like biotech) the value destroying effect of taxes is even more devastating. We have discussed the impact of taxes in our May 2008 newsletter. In the example the posttax value is US$ 25 Mio, only 64% of the pretax value. The described method assumes a posttax value of 75% (=100%25%) of the pretax value, i.e. US$ 29 Mio.
Value=invested
capital compounded at target IRR. (“valuation specialist”)
This method puts a lot of trust into the management of the company. It assumes that the management could increase the company value at the target IRR since inception regardless of what actually happened. It is clear that the valuation depends on the current situation of the company and its future outlook; the past is irrelevant for the company value.
Value=DCF
without success rate * success rate (Analyst)
This analyst valued each project without success rates
but then felt that he had to adjust for the attrition rate of the projects.
Consequently he simply reduced the value by the success rate of the project.
This riskadjustment is wrong. In biotech the value reducing investments are
unfortunately more likely than the value adding positive revenues from sales.
This means that the value suffers more than just by the success rate if you
correctly adjust the cash flows for their risk. The example company would have
a value without success rates of US$ 240 Mio. Multiplying this with the success
rates (*40%*60%*90%) this would lead to a value of US$ 52 Mio.
Another popular method amongst analysts is the dynamic
P/E ratio and also receives a honourable mention in this list. It has been
discussed in detail in our July
2008 newsletter.
Method

Value
(US$ Mio)

rNPV pretax

39

rNPV posttax

25

Real options at risk
free

187

with pharma discount

112

Terminal value

3949

50% discount rate

2.6

1FTE=US$1Mio

50

pretax*(1tax)

29

risk adjustment

52
