"p" or "β": How Clinical Success is Affected by the Macroeconomy

In the footnotes to my last blog post “Cost of Biotech Capital: Incorporating Development Risk into VC’s Target Return,” I contemplated which biotech start-up risks are uncorrelated with the general market and hence “diversifiable”. Many assume that a given clinical trial’s binary probability of technical or regulatory success (“p”) is one such risk.¹ As described in the last post, short-term oriented financial analysts (many at hedge & PE funds trying to gauge the outcome of binary events) assign a probability of success ("p") to a given clinical trial and do not expect that value to fluctuate with the general market (i.e. S&P 500). In the short run (event to occur in < 2 years), they are likely correct. In the long run (event to occur in >2 years), however, “p” is often influenced by general political and economic conditions. This is the timeframe considered by VCs and BioPharma companies. Below are three of the ways in which “p” can find its way into “β” in the long run and, ultimately, drive up the long term cost of capital for biotech.

(1)    FDA

FDA requirements affect clinical trial design and the choice of primary and secondary endpoints. All things equal, a stricter FDA will lead to a lower “p” or probability of what the FDA considers technical success. The FDA is in turn affected by the general political environment and hence the macroeconomy.

(2)    Availability of Capital

The free availability of capital can have dynamically opposed effects on “p”. On the one hand, more cash means better and bigger clinical trials which boosts “p”. On the other hand, too much capital leads to poor investment discipline and many companies getting funded that shouldn’t. This phenomenon is similar to what occurred in mortgages, where the free availability of low interest rate IO mortgages with no credit checks manifested in historical default rates 3-5 years later.   

(3)    Clinical Trial Recruitment

The status of the economy likely influences who and how many people agree to participate in clinical trials. One theory argues that in good times, more people can pay for experimental treatment or top-notch medical care themselves and decide not to participate in clinical trials. In bad times, the reverse may be true. Better recruitment may lead to better clinical trial results and a higher “p”. Likely (2) Availability of Capital and the existence of many competing trials is another factor in recruitment success.   

There are likely many other ways in which general economic conditions impact biopharma’s probability of technical success over time. In the long run, “p” shows up in “β”. Counter to my last blog post, this is an argument for adjusting biotech’s long-run cost of capital for binary technical risk. While 58.7% IRR still seems too high, perhaps the industry standard ~30% IRR is a reasonable happy medium.  

1.       “p” is not to be confused with the statistical term “p-value” which determines how likely a clinical trial result is due to chance.

Cost of Biotech Capital: Incorporating Development Risk into VC’s Target Return

Although the average biotech investor probably understands the term “cost of capital” and general CAPM theory, there still seems to be much confusion around how to incorporate the binary risk of drug development into discount rates and required IRRs. CAPM and portfolio theory tells us that investors are only compensated for correlation to general macroeconomic risks – i.e. risks we can’t diversify away. Thus, taking on so-called “idiosyncratic risks”, such as the binary risks of clinical trials, should not theoretically garner an investor any extra return (assuming the clinical trial risks are not correlated across a wide portfolio of investments, which may or may not be an appropriate assumption¹). Why then, do we see VCs demand gross² IRRs of 35%-60% per each investment (dramatically higher than CAPM would dictate) to compensate for drug development? The answer lies in the difference between a portfolio’s ultimate expected return (i.e. a VC investor’s cost of capital) and an individual investment’s target return.

                Andrew Metrick, my former professor at the Wharton School, clearly explains the difference in his book “Venture Capital and the Finance of Innovation” (http://www.amazon.com/Venture-Capital-Finance-Innovation-Metrick/dp/0470074280). The general financial theory he describes is also used widely in the BioPharma industry to calculate eNPV (expected or risk-adjusted NPVs). We used this theory at Genzyme, and I know it is used at many other major BioPharma companies to value their pipelines. The basic idea is to probability-adjust cash flows (or the numerator) for the binary risk of drug development (probability of technical success) and not incorporate that risk into discount rates (or the denominator). For example, to determine the value of an early-stage acquisition target, you would multiply the stream of potential cash flows by the probability of each of them actually occurring and then discount back using a discount rate determined by either (i) the general CAPM equation R_Discount = R_F + β(R_M - R_F) or (ii) a more elaborate model, such as the Pastor-Stambaugh model (PSM) that incorporates additional factors for value, size, and liquidity (for simplicity, assume no debt). The discount rate used by most BioPharma’s is ~10-15%. However, many VCs will use the following equation to move the idiosyncratic (or binary) risk (i.e. “p”) into the target discount rate (i.e. Target Return):

p/(1+ R_Discount)^Time = 1/(1+Target Return)^Time

For a project that has a p = 20% chance of exiting for some fixed amount of money and an 80% chance of exiting for $0 in Time = 5 years, with a standard R_Discount of 15%, the Target Return or Target Discount Rate = 58.7%.

.20/(1+.15)^5=1/(1+.587)^5

When VCs say their discount rate or required IRR is 58.7% to adjust for development risk, this really translates into the 15% discount rate we normally think of adjusted for probability of technical success. No investor actually expects to receive a 58.7% return for the entire portfolio. As is often said in the VC world, the winners have to make up for the losers so the returns can average out to a reasonable compensation for taking macroeconomic or non-diversifiable risk (and possibly adjustments for size, value, and liquidity).

Notes

 (1) These days, the big question is which biotech start-up risk is diversifiable and which is correlated to the general economy (i.e. macroeconomic risk) and affects β. Certainly capital availability, which plays a significant factor in the success of biotech start-ups, is greatly tied to the macroeconomy. However, this is perhaps already captured in the “size” factor of the PSM model. Other factors, such as FDA regulatory requirements or access to patients for clinical trials, are also tied to the general political environment and perhaps cannot be diversified away.

(2) Note that a VC’s gross IRR and net IRR differ due to fund management fees and carry.