The methodology for valuing biotech products using the risk-adjusted NPV (‘rNPV’) approach has been well established. The concept behind the rNPV method is taking into account the risk of success / failure in clinical trials for clinical development products. In this way, companies can prioritise their portfolio based on the expected value of each product.
However, clinical success rates are well documented (e.g. see Nature and BIO papers) and therefore the risk lies in the market performance of the product. This is further discussed below.
There are three key revenue and cost parameters that need to be modeled to arrive at the rNPV of a biotech product. These are the following:
Revenue parameters
- Effective addressable number of patients (i.e. after diagnosis rate, treatment rate and patient compliance has been accounted for)
- Peak market share
- Annual price per patient on an ex-factory basis
Cost parameters
- Cost of goods sold (COGS): typically 10% for biologics
- Selling, General and administrative expenses: ranging from 25% to 35% for biotech companies
- R&D costs: these may be post-marketing / Phase IV studies – usually these costs are minimal upon product commercialisation.
As we may see may revenue parameters 1. and 3. can easily be defined. A company that is developing a lung cancer product, can use incidence, survival and treatment rates to determine the effective addressable number of patients. In addition, the annual price to be charged can be based on similar products in the market, for example using the price of another product that is prescribed under the same line of treatment (as prices usually differ for drugs indicated for different treatment lines under the same disease).
Cost parameters can also be determined using industry benchmarks. For example, for a CNS or an oncology product one could check Biogen’s and Celgene’s COGS and SG&A margin, respectively.
Consequently, the most uncertain parameter is peak market share as it is linked with the product’s safety and efficacy results in clinical trials. However, for a drug that is in Phase I safety and efficacy outcomes are highly unknown. As a result the valuation model should provide sufficient sensitivity and stress testing on peak market share.
The purpose of this article is to provide a user-friendly Monte Carlo simulation model that can help a company make an informed decision on whether a biotech project is worth undertaking.
In the model that I have developed, the following assumptions are made:
Indication: Non-small Cell Lung (‘NSCL’) Cancer
Company location: US
Addressable market: 500,000 reflecting 10-year prevalence of 554,000 of which 90% is treatable (i.e. not at terminal stage or at very early stage)
Peak market share: 20%
Market share uptake: starting at 10% and increasing by 10% every year. Once uptake reaches 100% it stays at 100% for another year. Due to patent expiration and generic erosion it falls down to 50% the following year, then down to 20% the year after, then down to 10% the year after and finally down to 5% in the last year of the forecasting period.
Price at launch: $150,000, which is a typical annual price for an NSCL cancer product in the US.
COGS margin: 10%
SG&A margin: 35%
Probability of success in Oncology (source BIO): Phase I – 63%, Phase II – 25%, Phase III – 40%, Approval – 82%
R&D costs: Phase I – $30m, Phase II – $40m, Phase III – $100m, Approval – $2m
R&D period: Phase I – 2 years, Phase II – 3 years, Phase III – 4 years, Approval – 1 year
Tax rate: 26% (21% + state tax at 5%)
Working Capital: 15% of revenue
Discount rate: 25%
The assumptions above produce a risk adjusted value of $36m.
The important question is whether a peak market share of 20% is feasible or not. As mentioned previously the commercial outcome is a function of efficacy and safety results in clinical trials, which at the start of Phase I is highly uncertain.
This is where the Monte Carlo approach comes into play.
Assume in all possible scenaria the mean peak market share is 20% and the standard deviation is 10% (50% of mean). Now lets randomize the probability by using the RAND() excel function and dragging this formula down in 3,000 rows (3,000 randomised probabilities). By applying the NORM.INV function, where the mean is 20% the standard deviation is 10% and the probability is the Randomised probability calculated by the RAND function, the randomised peak market share can be obtained.
The final step is to create a data table where the row input is the randomised peak market share and the data table input is the risk adjusted value of $36m. Using the data table, a histogram can be plotted which shows in ranges of rNPV values and the count of these value ranges, including those that produce negative values. The distribution of value ranges is presented below:
The histogram shows that there are 417 possible zero or negative outcomes i.e. there is a probability of 14% that the project becomes worthless (zero or negative rNPVs).
By applying the Monte Carlo method, biotech companies are able to prioritise drug candidates that combine low probability of making a loss and high weighted average rNPV across all scenaria. Such prioritization is effective for generating high returns to investors, especially for public listed biotechs.
You may find the model with the assumptions and results discussed above in this link.
