Surviving the Value Quicksand
Value investing refers to buying stocks at less than their intrinsic value. Warren Buffett is considered an extremely successful value investor. While Buffett continues to perform well, many of his value investing counterparts have struggled over the last five years as technology stocks, in particular, have attracted investment capital. Value Guru David Einhorn, for instance, has underperformed the S&P 500 by nearly 50%. The Russell growth index has outperformed the Value index by 38% over the last five years.
I first encountered value investing when I began working at JPMorgan Investment Management. I was intrigued by the method that they adopted, the so-called ‘Dividend Discount Model’ or present-value calculation. JPMIM employed a large number of equity analysts to forecast the dividend profile for a company which was then equated with the company’s current market price to get the implied discount rate. This ‘dividend discount rate’ (DDR) was central to the value philosophy since an increase in a company’s stock price would cause the DDR to fall (assuming the analyst dividend forecasts remained constant) thereby making the long-run rate of return lower. DDR’s should rotate from high (cheap) to low (expensive) as prices moved lower to higher. This ‘rotation mechanism’ is logically attractive and my first job at JPMIM was to understand whether the DDR actually ‘rotated’ the way people believed it should from a quantitative perspective. It didn’t…
One only had to look at the data to realise that the DDR did not mean-revert since many of the critters would just hang around the same level for years indicating the stock was forever-cheap or forever-expensive. Moreover, different analysts displayed different forecast biases meaning one set of DDRs were systematically higher than another set. A few weeks studying the dataset led me to the conclusion that value investing can ‘get stuck’ when the assumptions determining the level of the DDR are wrong. Getting stuck happens when you hear comments like ‘I’ve never found a cheap tech stock’ or ‘why hasn’t the market realised that this stock is a screaming buy’. In its most brutal form, getting stuck is when Guru’s like David Einhorn lose all their assets and their reputation is ruined. Getting stuck is like sinking in quicksand…
The solution to this DDR-level problem, to me, was and still is, simple. Rather than focus on levels, it is better to focus on changes. It is trite time-series practice that, if you suspect that a variable has an unstable or inestimable mean, you can eliminate the problem by simply differencing it away. DDRs, like share prices, can meander aimlessly for months and years until new information hits which then jump…so it is the change in the DDR that carries information as opposed to the level.
So how does this differencing idea work? A neat result from statistics is that if you think that a stock should trade at 10 times earnings, then the change in the stock price should also be 10 times the change in earnings. The ’10 times’ is a parameter that you can estimate using either levels or changes and it should be the same. If it turns out that this parameter estimate is wildly different when using levels v changes then you have a problem in your valuation process. Put another way, if for some reason your pessimism never makes a tech stock look cheap, no one is going to question the value-based wisdom of buying tech stocks that have just gone down by 25% – they may not be cheap but they just got a whole lot cheaper!