Biotech finance 101: for finance professionals

Richard Murphey, 10/16/2018

There are many good resources online to learn about finance, but analyzing biopharma companies -- especially pre-commercial companies -- requires some knowledge that you can't find in traditional finance courses.

In this next series of posts, I'll introduce some biopharma-specific financial concepts. This post will look at finance skills and techniques you'll need to work in investment banking or corp dev. Subsequent posts will deal with more advanced concepts that you'll need to work in investing or equity research.

This post assumes some intermediate knowledge of finance and valuation techniques and is geared towards current or aspiring investment bankers or corporate finance professionals. If you are a scientist who is not familiar with finance, you may need to do some googling to follow some of these concepts, but you can figure it out. Finance is much easier than biology, though it does take practice.

If you'd like to reference a sample model as we go through the post, you can download a model of a gene therapy called Zolgensma here. Zolgensma, priced at $2.1M per patient, is the most expensive drug in the world.

If you're interested in reading new posts I write on biotech, startups and VC, sign up for my mailing list:

Forecasting P&L for development stage biopharma

Your first “toe in the water” of biotech finance will likely be forecasting the P&L for a pre-revenue drug company. Like most other things in banking, the math here is pretty easy, and the challenge lies in making good estimates and assumptions. This is the first thing that got me interested in biotech – I felt like making these models was much more intellectually interesting than doing 10 refi models for acute care companies.

Basically, take the same P&L model template you use for any other industry, and insert a bunch of rows above your revenue line. You will do a detailed revenue build, starting with number of patients the drug is designed to treat, then layering on pricing, then estimating an adoption curve.

Revenue build

An easy way to get a feel for this is to look at equity research reports for pre-revenue biotech companies. Many will lay out the revenue build. This requires no science knowledge to understand, though you'll need to learn a few terms like incidence and prevalence, and get familiar with concepts like different subpopulations of disease, lines of therapy / treatment algorithms, how to read prescribing info, get introduced to drug pricing, basic concepts like royalties and gross to net discounts, etc. Most of this info can be found by googling.

A basic formula is:

  • Start with the population of your geography of interest (ie # of people in the US)
  • Then layer on the epidemiology data – incidence and prevalence – to estimate how many of these people have the disease
  • Then layer on any filters for subpopulations – are you targeting patients who have failed two previous lines of therapy? Patients with a particular genetic mutation? Patients exhibiting certain symptoms?
  • Then filter for accessible patients. How many of these patients are diagnosed? How many see a doctor regularly? How many have insurance that will cover the treatment?
  • Then filter for how many patients will be treated
  • Then filter for how many patients will be treated with your drug, vs another drug
  • Then multiply by price per patient (you should look at gross and net pricing, ie including discounts to insurance companies, wholesalers and retailers)

In most cases, you can get pretty reasonable estimates for “peak sales” if you do this.

Uptake / market penetration curve

The hardest thing is usually predicting uptake / sales trajectory. You have an estimate for peak revenue, and you know revenue starts at zero before the drug is approved, but what happens in between?

Investors commonly get this wrong – even investors who specialize in estimating pace of launch. I know some investors that, for rare diseases where only ~50 physicians treat the disease of interest (yes, you can develop billion+ dollar drugs treating these rare diseases), will call all 50 physicians and ask how many patients they treated in the quarter. These same investors have made terrible calls estimating launches for drugs in bigger markets (Regeneron’s Praluent is a classic example of the market being very wrong on a prominent drug launch).

It can help to look at the market from the bottom up. How many physicians treat this disease, how many patients does they typical physician see, and what are the alternative treatments. For a good launch, you need a drug that is meaningfully better for patients than the standard of care – otherwise physicians won’t prescribe it, and insurance companies won’t pay for it. Assuming the company has a useful drug, the next step is to see what kind of sales force the company can afford, and how many physicians that sales force can reach.

You won't be expected to have the right answer to this in IB. You'll likely benchmark some similar launch curves, or use management’s assumptions as a starting point. You’ll also probably look at datasources like IMS Health’s database.

This will give you your basic topline for the years after the drug is approved.

Cost of approval

Then you need to model what it will take to get the drug approved. This article is a pretty good starting point for understanding how long each stage of drug development lasts, what it costs to get through each stage, and the probability of success at each stage (it is a paywalled Nature article, although you can see the relevant figures). For more detail on the drug development process, FDA’s website has an overview of drug development for patients that is written in approachable terms. I also wrote a post on the basics of drug discovery and development.

You’ll want to refine your estimates based on the disease the company is treating, their trial design, and several other variables. If they are developing a cardiovascular drug, they will likely need to conduct large Phase 3 studies with thousands of patients to get approval. If they are developing a drug to treat a life-threatening rare disease, they may be able to get FDA approval with studies of only a few dozen patients.

Look at the company’s investor presentation or SEC filings to get a sense of what their specific path to FDA approval is. If they aren’t public, look at public companies developing products to treat the same disease. You can also look through to see the kinds of studies that other companies did. Google is also your friend.

Then you need to estimate probability of success. Probability of success varies widely based on stage of drug, disease, quality of preclinical disease models, and types of patients you enroll in your studies.

This is impossible to do accurately, which is why you see some biotech stocks going up 100%+ -- or going to zero – overnight when study results are released. Being good at this is probably the single most valuable skill you can have in biotech, and the reason funds like to hire PhDs. If you are just a banking analyst and aren’t putting millions of dollars behind your estimates, you can either use the company’s estimate, or use this article as a starting point.

The easiest way to build DCF models

Build robust biotech valuation models in the browser. Then download a fully built excel model, customized with your inputs.

How to learn it

The best way to learn this, other than talking to the biotech team at your firm, is to read equity research reports for pre-revenue biotech companies, read investor presentations, and SEC filings. Google anything you don’t understand. Try to identify which assumptions in the equity research models are BS – a lot of valuations in biotech right now are at levels that can’t be explained by fundamentals, at least with somewhat conservative assumptions, (ie there is a lot of strategic value / M&A speculation baked into a lot of prices), so many ER models that aren’t sell recommendations (ie most of them) will probably have some holes, and some of the holes may be large.

Once you get a sense of the terminology and basic techniques, try building your own models using just SEC filings and google (and maybe the investor presentations).

In the next post, we discuss how to value the P&L you just forecasted.