PDF version

Valuation of Intellectual Property in Pharmaceuticals


Introduction

Intellectual Property of one form or another is a key component of the value of almost all businesses. In the pharmaceutical sector, where an individual element of Intellectual Property such as a patent on an early stage compound or a newly conceived product delivery system may enable the company to earn future revenues in the hundreds of millions or even billions of dollars per year – or bring in nothing at all if it cannot be developed to market – the great majority of strategic decisions require knowledge of the potential value that that piece of Intellectual Property represents. Such strategic decisions may relate, among others, to fund raising, agreeing product licensing terms with partners, evaluating mergers and acquisitions, litigation proceedings, selling or buying parts of businesses, making portfolio prioritisation decisions within the organisation, determining patenting strategy, or preparing and evaluating capital investment proposals.

In the majority of cases, the Intellectual Property concerned is directly linked to a defined product opportunity, usually represented by an ongoing R&D development project. This paper therefore will concentrate on the discussion of possible methods for evaluating such a project.

The need for such evaluations is clear. How to do them is much less obvious. This is because so many things can happen to a pharmaceutical development project that will influence its worth. It may fail at any stage in development; the costs of its development will vary depending on what each stage reveals; the time that the development process will take can vary by years. The product may not gain a product licence; or take much longer than expected to do so; or do so in one country but not another; or get approval for a limited indication only. Once on the market its price can be influenced by a variety of factors which are currently unknown – from new price control legislation that is introduced to the price of other products launched in the same market before this one reaches it. The share of the market it can win will depend on its competitive profile when it finally completes development and what kind of other competing products have emerged by then or are introduced later. The size of the market will vary depending on trends in disease incidence, health care policy and population growth.

In the face of such uncertainties, which are made worse by the fact that the time-scales that need to be considered are enormously long – the product may not reach the market for, say, seven years, and not reach peak sales for fourteen years – the temptation is to say that the task is impossible, or that if it is attempted the results will be meaningless.

However, to dismiss the task as impossible will not do, in the face of the vital need for an evaluation of some kind. To dismiss results as meaningless is only justified if they have been generated using an inappropriate methodology. The purpose of this paper is to review some of the methodologies used, to assess their appropriateness and describe the application of some of them.


Peak sales

Probably the simplest method of giving an idea of the worth of a product is to estimate a peak sales figure that may be achieved. Such figures are often quoted in commentaries on pharmaceutical companies. This is normally a very straightforward assessment based on the sales of existing, competing products in the market where the product will be sold. If, for example, the current market leader is known to have a market share of 20% and sales of $260 million and the product to be assessed is thought to be more effective than the market leader, then it may estimated that the new product would have a share of 25% worth, at a price 33% higher, $430 million.

A slightly more sophisticated approach, and one which has to be used where there is not an existing market that the product will enter, would be to take an estimate of the number of patients who suffer from the condition to be treated, estimate how many of these may be prescribed the product and at what price it will be sold. For example, if there are 2 million sufferers from the condition, and 25% of them may receive the product which could be sold at $200 per year then the product would have sales of $100 million.

This approach has value where only the most rapid snapshot of a product’s potential is required – it would commonly be used for example in a financial analyst’s comments on the progress of a particular company. It is also valuable as a “sanity check” for the more sophisticated approaches that may be adopted, such as those described below, or as a means of broadly comparing one product with another. For any other purpose however it has little worth. It tells the observer nothing about the many other factors that go to build up the real total value of a product – its development, marketing and production cost, for example, or the chances that it will not even reach the market. Nor does it represent a meaningful value for the Intellectual Property at the present time – there is no reason to suppose that the product would be sold for a figure equivalent to its peak sales either now or at any time in the future.

In its defence it can be said that in the specific case of the pharmaceutical industry, where the gross profit from a product is generally in excess of 80% of sales value (well within the margin for error of the forecast involved), and the effort to sell it may be relatively marginal, it is a better predictor of earning capacity than in most industries.


Traditional NPV approach

The most useful and widely accepted way of gaining a meaningful estimate of the total value of a project/product, including the cost of its development, and the risks associated with its development and marketing is to calculate a Net Present Value of its cash flows. This is a well established methodology that has two main components:

• A forecast of the cash flows that a product will generate over its life-time (including the negative cash flows during its development period)

• A discount factor, which is applied to these cash flows, and which reflects, as a minimum, the cost of money over time (the fact that $10 million earned next year is worth more than $10 million earned in 2005)

Traditionally this has been approached in a relatively simplistic way. It involved estimating a single cash flow for the product and a single discount factor which dealt with not only the cost of money but also all the various risks associated with the development and marketing process.

So, for example, a product may be estimated to be likely to have a cash flow with the following appearance:




To this set of numbers is then applied a discount factor of, say, 40%. This can be done using a simple spread-sheet function, and it gives in this case a NPV of $2.25 million.

This may seem surprisingly low for a product that has peak sales of $100m and maximum development expenditure in any one year of $15m. The reason for the low figure is the high discount factor of 40%, which has been used to represent, as described above, both the cost of money and the risks inherent in the project. (This particular discount rate of 40% has been chosen because it is frequently used as a hurdle rate in assessing biotech investments.)


Simulation-based NPV

The simple NPV approach described above is widely used, but has several obvious disadvantages:

• It only allows for one cash flow estimate, although, as has already been described, the possibilities for different cash flows are virtually endless, given all the variables that could affect performance. (It is often said of such a single figure estimate of sales or cash flow that the only thing you can be sure about is that it is wrong.)

• It bundles all risk factors into one single element – the discount rate - which is generally an industry standard figure and is a very crude representation of the actual risks to which the product is exposed.

• It yields a single figure result which is likely to be very far removed from the actual outcome – if the product succeeds in getting to market its overall cash flow will be worth much more than $2.25 million. But if it fails in development it will be worth much less. One of the least likely actual outcomes is in fact a return of $2.25 million.

What is needed to improve on this is an approach that takes account of the many different cash flows that are possible for a given project.

The most common way of doing this is to use simulation. This method sounds complicated but is at least in principle quite simple. It involves the creation of a spreadsheet model of the product and its market which looks like any other but has one significant difference – many of the inputs do not take the form of single figures, but are ranges of numbers. So, for example, instead of putting into the cell containing the price of the product a single figure of $35, a formula is inserted which allows the price to lie somewhere between $30 and $50, with the most likely figure being $35.

In a similar way the launch date may be entered as any of, say, three different years, with a probability attached to each. For example:

• Launch in 2004 - Probability 20%

• Launch in 2005 - Probability 50%

• Launch in 2006 - Probability 30%

Any other element of the model – development cost, market share etc. – can be entered in the same way.

All that is needed once the model is set up like this is to run it many times – generally several thousand times – and the final result will encompass the overall effect of several thousand individual scenarios for the performance of that product. This method can even include the possibility of complete failure for the product. If for example it is thought that a project has only a 20% chance of moving successfully from one stage of development to the next, a cell can be created in the model which has a 20% chance of containing the number 1 and an 80% chance of containing the number 0. If when the model is run it finds a 0 in the cell, the rest of the model after that development stage will return zeros – the situation in which the project fails in development will have been modelled.

This final result will not of course be a single figure – it will be a range of possible figures, covering all the reasonably likely outcomes for the product. So the results for sales (assuming in this case that the product always reaches the market) may look like this:





While the NPV output (including here the possibility that the product might fail in development, and only discounting for the cost of money) might be:





This very interesting chart quite clearly reveals a great deal more about the product than the single figure produced by the simple NPV approach. In fact it illustrates very well the point about the single figure result already mentioned above - that the most likely outcomes lie very well away from the single figure (which in this case would be about $12 million). In particular it reveals the exposure to loss that is inherent in the project, with about 73% of outcomes falling below zero, 25% of them even below a loss of $30 million, and the single most likely outcome being a loss of $20 million.


This is a much more effective way of assessing the value of a product because it:

• Uses assumptions of risk which are specific to the project and its market, rather than some standard factor;

• Provides a result in which there can be a good deal of confidence because it is clear that all probable variables have been taken into account. This is very useful where, for example, a project team is providing the assumptions for a model and different members of the team have different opinions about some of the elements. There is no need to decide between the opinions - they can all be incorporated.

• Shows both the risk exposure that is inherent in the project and the potential upsides if all goes well. This is great use in assessing a portfolio of potential projects - two projects with an average NPV of $12 million can have very different possible upsides and downsides, and it is important for a company to be able to assess both of these against its strategic willingness to take a greater risk to get a greater return.

It is of course still possible to calculate a mean figure across all of these outcomes to give a snapshot of the product if comparisons across several projects are to be made. It is also possible of course to create sub-scenarios which examine the impact of making completely different assumptions for some inputs. For example, if a project may emerge with a wide indication or a narrow one, this could be dealt with by providing a very wide range of market shares in the main model, covering both possibilities. However this may provide a too vague an answer, and it may be better to run a simulation assuming just the narrow indication and compare it with one which assumed the wide indication.


Simulation and decision trees

This simulation method encompasses all the aspects of a project in one model. It is very powerful and informative, but to some degree can seem to be a "black box" within which many things happen which are not quite clear to observers. To provide greater clarity and a more intuitive structure, a decision tree can be used for some parts of the calculation. In the simplified example below, simulation has first been used to obtain a range of potential performances of the product in the market. This range has then been simplified down to three outcomes, shown at the top right of the diagram: one where the product profile is good and the market receptive (which is assumed to have a 25% chance of occurring), one where the profile and/or market is moderately favourable (40% chance) and one where one or both of these factors is unfavourable (35% chance).

The rest of the tree deals with the costs and probabilities associated with developing the product to launch, showing in each case the event which determines where the project goes next (e.g. "Phase III succeeds") and the probability of that event occurring (e.g. 0.60). Hidden within the model is the cost of taking the project through Phase III.




The result is a highly informative picture of the various events that can occur and the cost or benefit of each. It shows, for example, that there is a 5% chance that the project will earn around $450 million (top right), but a 10% chance that it will lose $50 million (if it fails at the end of a high cost Phase III programme). Its overall value now is $30.4 million. If it is successful in Phase II that will rise to $99.1 million.


Option pricing

Recently attention has focussed on an enhancement of the probabilistic approach described above which draws on a technique used in the field of commodities and share dealing - option pricing.

Options exist where an arrangement between two parties is made at a particular date which gives one party the right (but not the obligation) to buy an asset from the other party at a given future date at a given, pre-agreed price. The buying party pays to the selling party an amount on the date of the initial agreement in recognition of the receipt of this right.

The benefit to the selling party is the receipt of money up-front and knowledge of the price at which the ultimate sale will be made (if the option is taken up). This has cash-flow advantages for the seller and has other consequences depending on relationship between the value of the asset at the point of exercise of the option and the pre-agreed price:

• If the asset proves to be worth less than the price, the buyer will not take up the option and the asset will stay in the hands of the seller. In this case the seller has at least received some return on the asset in the form of the initial payment at the point of granting the option, and of course has the right to seek an alternative buyer for the asset in the way that he would have been doing if no option deal had been agreed. However at that point the asset may have lost so much value because of new data emerging since the deal was agreed that he cannot sell it at all, or at a reasonable return.

• If the asset proves to have a value equivalent to, or rather more than, the agreed price, and the option is taken up, the seller will have made a profit on the deal because he has received both the pre-agreed price and the up-front payment when the deal was agreed.

• If the asset is worth very much more than the price, the seller will not be able to profit from all of this gain in value because he is obliged to sell the asset at less than its full value.

The buyer on the other hand experiences the complementary benefits:

• If the asset is worth less than the price, he will not want to exercise the option and will lose money, because he has paid to obtain the option right in the first place.

• If the asset is worth about the agreed exercise price, he will lose the amount that he paid up-front as a means of guaranteeing his option right. Presumably this was a price worth paying to have exclusive access to the product when its true value became clearer.

• If the asset is worth the pre-agreed price plus the initial fee paid for the option right, then he has not lost any money and had greater control over the buying process than he would have had if he had bought on the open market.

• If the asset is worth more than these two prices put together, he has made a profit on the deal.

The whole arrangement underlying an option deal has a value that reflects all of these possible outcomes. In the financial field there is a widely accepted methodology for calculating this value which is based on the interaction of four key factors:

• Time - the reason for entering into an option is to permit a buy decision to be deferred until better information about the value of the asset concerned is available. The longer the decision can be deferred, the more the option is worth, since information will always improve over time.

• "Volatility" - an option over an asset which is likely to vary a lot in value is worth more than one for a highly predictable asset. This is because the option permits a buyer to walk away if the asset's value has fallen by the date of exercise - losing no more than the initial payment - and only to buy if it has gone up in value. Because of this, the buyer can concentrate solely on the upside variability of the asset. The greater this volatility is, the better.

• The value put on the asset which is to be bought at the point when the arrangement is initially made.

• The amount the buyer must pay when he exercises his option (the exercise price).

A formula, known as the Black-Scholes formula, is used to derive from these four input elements a value for the whole option arrangement at the point when the arrangement is entered into. It should be noted that the Black-Scholes formula requires cash-flows to be discounted at the risk-free rate of interest, since the element of risk is otherwise accounted for.

It can be seen from the factors listed above that such an approach could have value in the pharmaceutical field, where projects in R&D are subject to a huge array of influencing factors which can change their worth dramatically - they are highly volatile. At the same time their time-scales are very long, with a great deal more being learned about the potential of any project as it moves through the R&D process, and there are clear decision points from time to time as phases of development are competed, when management action can reduce the risk of losing large amounts of money.

This suggests that the methodology may be useful in the evaluation of a potential product within the R&D portfolio of its originator. Under this approach the company is considered to be buying options in its own product at each stage of development.

As an example, it could be assumed that the company is considering a project which is now at the end of Phase I, with one more decision to be made within the development process - at the end of Phase II. In this case the components of the option are as follows:

• The asset to be purchased is the product once on the market. This has a value today, at the end of Phase I, which is its Net Present Value (NPV) at this point, and a volatility.

• The exercise price is the amount of further development money that must be spent after the decision to proceed beyond the end of Phase II has been made.

• The time is the period between now and the end of Phase II

Using these elements, Black-Scholes can potentially give a value for the option. Of course the company will also have to pay to buy the option in the first place - by spending the amount of money needed to get through Phase II. This amount must be offset against the value of the option acquired. The net figure will be the value of the project today, when considered as an option.

As a further development of this approach it is possible to recognise the multi-phase character of R&D and to design an analysis that evaluates an individual project as a series of two or more nested options - as an option on an option.

These approaches could have advantages over the more traditional NPV approach, in that the NPV methodology, by discounting for time and for uncertainty, punishes more long term and more volatile projects, while option pricing does exactly the opposite - rewarding long time scales (and the chances this provides to make decisions as information becomes available) and volatility. Mechanically, there are two key differences between the two approaches:

• The options approach uses low discount rates to calculate present values. This maintains a much higher intrinsic value in the asset to be bought than using a discount rate which is weighted for risk.

• The options approach calculates the project value by applying a special factor to this asset value. The NPV approach calculates a final value by subtracting the NPV of the up-front costs from the asset value. In many circumstances the value of the asset is reduced less by the option factor than by the deduction of up-front costs.

In principle this approach is very attractive. However there are several reasons why it is not necessarily as useful in practice. The Black-Scholes formula was developed to describe a different environment, where purchasing decisions are driven by rather different considerations than those that apply in pharmaceutical development. It is considered by analysts to be a brilliant but over-simplified methodology, which ignores several important factors. It requires crucially the input of a volatility rate which should be taken from the changes in pricing of equivalent products over time on the open market (as with futures trading), but in pharmaceuticals there is virtually never an adequate market comparator on which to base the setting of that factor. Finally it is very much a "black box" into which certain data is fed and out of which an answer comes without it being very clear why. It is therefore often difficult to persuade others involved in or affected by the valuation process to accept that the answer is valid.


Conclusion

Valuation of Intellectual Property is not a precise art. It is frequently necessary to use several different methods (and the range of approaches described above is far from exhaustive) to arrive at a broad consensus position. In general however the author's experience is that the most meaningful results are obtained by using the simulation technique, perhaps allied as described to a decision tree, wherever time is available to do the work required to establish the key factors that will influence performance and to build the appropriate model. The Option Pricing approach has yet to show that it is robust and convincing enough to take over.

Top