Valuing US biotech

Lately I have been working on an intriguing project. A colleague, who is also our in-house biotech analyst, asked me if I could take a look at his sector in order to try and find some "leading indicators". Many investors, and indeed normal people, would probably find this sort of work rather dull - you can easily end up getting lost in a sea of numbers as your spreadsheets become larger and larger (indeed, in this particular project, I actually ran out of columns on Excel!). In my experience, people usually prefer to stock pick, rather than grind out the numbers for an entire industry or sector. Fortunately, being somewhat skeptical of the value-added potential of bottom-up stock-picking (and being a bit of geek), I actually quite enjoy these sorts of investigations. While the methodology can be long and laborious, it can be somewhat rewarding when you finally reach a conclusion.

So I set out by extracting all the necessary fundamental data from Bloomberg as far back as 2001. I had initially intended to go back further, but a brief experiment along these lines quickly showed that a)there was not enough data for each stock in the Bloomberg database, b)most of it would be distorted by the 1999 bubble anyway and c)any index returns would be distorted by a severe survivorship bias (still a problem going back to 2001, but less so).

I must admit that I did not really expect to find any "leading indicators". As someone who believes that markets are pretty efficient most of the time (although not perfectly efficient) I am generally skeptical of any variable that claims to have predictive power, especially if that claim is over short term returns (as I wrote in a previous post there are variables that appear to have long-term predictive power). Nevertheless, after some very extensive investigations over the past two weeks, it appears that I may have actually found such a variable: the EV/cash ratio.

The EV/cash ratio is the Enterprise Value of a firm divided by the cash and equivalents that the firm has on its balance sheet. Why should this variable be important? For most businesses it is not important at all. A supermarket, for example, does not typically hold a lot of cash on its balance sheet. Assuming it is well run, it will have a healthy positive cash-flow, so the amount of cash on the balance sheet should not have much relevance as far as valuation is concerned. However, for a biotechnology firm, especially one that is small, cash-flow is negative and the firm is not profitable. The business model is one of research and development, financed via the equity and debt markets, with the eventual aim of developing a blockbuster drug. This means that the cash on the balance sheet is actually very important as it gives some indication of how much resources the company has for research and development before it must go back to the market to raise more cash.

The interesting conclusion from the investigation is that if you had used the EV/cash ratio as a systematic trading strategy for the Biotech sector since 2001, you could have made substantial profits. On an aggregate level, there is a strong correlation between the EV/cash ratio and the price performance for the large, mid and small-cap US Biotech sectors. This is important because it suggests that the fluctuation, or mean-reversion, of the EV/cash ratio occurs more often than not through the price adjusting rather than the cash adjusting - a necessary condition for any ratio to be a useful indicator of future returns. Following on from this, comparing the aggregate EV/cash ratio with the subsequent 1-year return shows that buying when the EV/cash ratio is low and selling when it is high would have been a strongly positive return strategy since 2001. Finally, even within the mid and small-cap sectors, the trading strategy seems to work. For example, the one year rolling return of the small cap stocks trading on less than 2 times cash has outperformed, in almost every period, the return for those on greater than 8 or 10 times cash.

What I find most surprising about the results is that they work over a 1-year time period. Indeed, if I had accurate and reliable data going back to the beginning of the biotech industry, I would not be overly surprised if the correlations held over a 3, 5 or 10 year view. But 1 year is very intriguing, not to mention highly useful, at least from a fund management perspective.

But of course the results do not "prove" anything because the sample period is not long enough. So we shall have to wait to see if the relationship continues to hold. As we all know, history does not repeat itself - although it does tend to rhyme.