What Empiricism Is and Isn't
I buy cheap stocks because of the extensive, statistically-significant evidence of outperformance of backtested portfolios of cheap stocks on any conceivable valuation metric. In other words, I buy cheap stocks because it has worked reliably in the past.
Seth Klarman once said:
Value investing is at its core the marriage of a contrarian streak and a calculator.
Well, along the same lines, I like to think of an empiricist as a historian with a calculator. And I think it's an admirable position. There are many ways to "know" something, but the rigorous tests and statistically significant findings of replicated studies following the scientific method are the most reliable. Farnam Street has a good recent article referencing a speech by Atul Gawande of Checklist Manifesto fame (I shamelessly use investment checklists, by the way, and surely will write about that more at some point). The article is about the importance of thinking scientifically and the skeptical, curious way of life that creates, as well as the surprising lack of scientific thinking in the world today given the remarkable technology and endless information we now have available like never before.
Empiricism is not simply looking at something that worked in the past. Observations about the past can be grouped into three categories:
- Empirical evidence
An anecdote is the weakest evidence and looks something like this:
I know that it's not a good idea to select the reduced-benefit survivorship option on our pension because my husband's parents did that with his father's pension and the guy is still kicking at 95 years old.
Patterns can also be extremely weak:
I know that I should buy this IPO because, well, look at Google, Apple, Intel, and Microsoft!
Source: YCharts, illustrative effects courtesy of author
Anecdotes, patterns, and empirical evidence are closely related. They all represent experience(s) or observation of experience(s). The difference is quantity. An anecdote is a single experience, which means very little. Patterns range from a few experiences to thousands. Anything less than statistically compelling evidence is still just a pattern, and a very dangerous one at that because of the poor behavior it can incite. Empirical evidence is the stuff of rigorous academic studies. Very large sample sizes. Replicated studies confirming the results. Peer reviewed. The works. Anecdotes are pretty harmless because most people know it's just one story and that many things could have happened. Patterns are dangerous because, without a calculator - and obviously we live much of our lives and take in much information without a calculator in hand, we have little sense of what's statistically significant evidence.
Further, some kinds of experiences can be more memorable than disconfirming experiences. For example, take IPOs. A successful IPO like Google (NASDAQ:GOOG), Microsoft (NASDAQ:MSFT), Intel (NASDAQ:INTC), or any other "next big thing" company is obviously very memorable confirming evidence of the success of investing in IPOs. The failures are less memorable because most of those companies don't exist anymore and aren't on our screens. And we naturally don't like to dwell on the negativity and losers like that.
There's also the problem of feasibility. Sure many IPOs perform well from the offering price they are sold to the initial stockholders at, but there is often only a tiny initial float so big institutional investors may not be able to take a sizable position in many IPOs in the offering and I bet all of that good performance come on the first trade on the first day from initial investors to average Joes without connections to the offering. I would not be surprised if there were findings that IPO performance measured from the close of the first day's trading is dramatically worse than that from the offering price and as I said, that's the performance that matters for most investors contemplating investment in a new issue.
Thus, without considering sample size and sampling method, it'd be pretty easy to think that IPOs are a fruitful source of compelling investment ideas, but in fact backtests of IPOs show they underperform by like 4% a year. The pattern is more memorable but far less reliable than the disconfirming empirical evidence of a bad base rate and thus overshadows it.
Common Investment Approaches That Empirical Evidence Doesn't Seem to Support
So back to why I buy cheap stocks. I guarantee you can find evidence of some kind that any investment approach or investment criterion is effective. Almost all of this information is noise though. Most of it is anecdote and pattern because that evidence is easier to observe and report on. There are far fewer scientific studies performed than there are human observations.
Let's briefly discuss some other methods that haven't necessarily worked in investing, but are commonly used. There are many other things investors look for that has not been backed by empirical evidence. Quality factors like earnings growth, sell-side recommendations, high returns on invested capital, etc. are not predictive of outperformance. In fact, in backtests of net-net stocks, the unprofitable net-nets actually outperform the profitable ones. In one of behavioral economist Richard Thaler's studies, he found that the lowest P/B deciles of the market (the dogs) subsequently showed the fastest earnings growth. Joel Greenblatt is a god of finance and his work helped me enormously in my development. I strongly recommend The Little Book that Beats the Market. The Magic Formula has worked well over long periods, but it's not perfect. According the Quantitative Value by Tobias Carlisle and Wesley Gray, the results of a portfolio of cheap stocks selected using the EV/EBIT metric alone actually delivers better results than Magic Formula, which also uses ROIC as a quality factor and gives it 50% weight in stock selection. So the quality factor may actually detract from the performance slightly. Why is this? It doesn't seem to make sense. I think this is because of the powerful reversion to the mean, cyclical dynamic in free market economies, which makes good investing extremely counterintuitive. In short, looking at all of these things makes sense, but the empirical evidence I've seen suggests it doesn't really work.
The Case for Buying Cheap Stocks
So what does the empirical evidence say works in investing? I won't lay out the numbers and studies here. That would be kind of redundant since I'm not the one who did the studies and there are already great books and white papers that synthesize the results. Here's a classic white paper from Tweedy Browne that I think is the best source I've seen. By far the most evidence has been found that buying the cheapest stocks in the market on almost any conceivable metric - P/B, EV/FCF, P/S, P/E, you name it - works very powerfully.
I just finished A Random Walk Down Wall Street, the manifesto of market efficiency. Even in this book, the antithesis of active investing, the author Burton Malkiel admits numerous times that there is compelling evidence that cheap stocks outperform. I followed his argument very carefully. He brings up the point numerous times, but simply writes it off as being explained by extra risks. He believes cheap stocks outperform because they are riskier. If we define risk as permanent impairment of capital, perhaps at the position level cheap stocks are riskier. I wouldn't be surprised to find, for instance, that a greater percentage of stocks trading below book at any given time subsequently file for bankruptcy in the next 3-5 years compared to the market population. However, risk at the position level and at the portfolio level can be very different things. At the (diversified) portfolio level, risk and return are much more correlated than they are at the position level because volatility and individual idiosyncratic risks like a bad luck bankruptcy are largely diversified away. There's also the element of time. Risk changes when your investment horizon is 10 years because then a drawdown in year 2 doesn't matter too much.
Having established all of that, I'm ready to say what I wanted to say: I believe a diversified, long-investment-horizon portfolio of very cheap stocks, with cheap defined as trading at the lowest multiples of a variety of commonly used valuation metrics will reliably deliver better returns with less risk than market portfolios and portfolios of expensive stocks. And I believe empirical evidence produced by rigorous studies trying to answer the question "What has worked in investing?" point to buying cheap stocks most often and with the most compelling evidence. That is why buying cheap stocks is so central to how I invest.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.