In my Friday letter, Thoughts from the Frontline (you can view it here), we looked at how valuation and prices change over market cycles. As I mentioned and have written about extensively in my book Bull's Eye Investing, market cycles should be viewed in terms of valuation and not prices. But how does one capitalize on such a way of thinking? Today's "Outside the Box" will show how several valuation styles and categories have performed over different time intervals, how each styles compares to the other and how each style meshes with the other.
The piece titled "Just a Little Patience" is written by my good friend and fellow investment colleague, James Montier. James is the Director of Global Strategy at Dresdner Kleinwort Watterstein, a London and Frankfurt based investment bank. He is also a prolific writer and author of the book "Behavioral Finance - Insights into Irrational Minds and Markets."
Aided by data from the Quant department of his firm, James dissects a large amount of information in order to present a well-researched report on how value and growth strategies work over time. His conclusions show how patience (defined as a longer time horizon) favors the value investor and hurts the growth investor. One particular note of interest is where James shows the results of a value component in a growth strategy and vice versa.
A key insight to gain is that the prudent and disciplined investor is rewarded for not wavering in his investment methodology, while those that do achieve lower returns. This is one of the more in-depth editions of the year and I trust that you will find it to be "outside the box."
John Mauldin, Editor
Outside the Box
Just A Little Patience
This week we want to look at the benefits (and costs) of long-term investing from a portfolio point of view. Can value and growth managers exploit the noise to generate superior long-run returns? In order to answer this question I enlisted the help of Sebastian Lancetti of our Global Quant team (I would be lost without them!). I asked Sebastian to form quintiles based on simple trailing PE. The universe we used was MSCI Europe since 1991. We called the cheapest 20% of the universe 'value', and the most expensive 20% of the universe 'growth'. Of course you might argue that isn't a great definition of growth, and indeed it isn't, but the results don't look dramatically different if we use price to book, or any other valuation measure.
Each month, portfolios were formed and then let to run over various time horizons. Of course, this creates problems with overlapping sample periods, but the results don't alter materially if we try to avoid this.
As is my want I am going to start from the value perspective. Is time arbitrage a profitable pursuit for value investors? Certainly a priori one would expect so. As Ben Graham said, "Undervaluations caused by neglect or prejudice may persist for an inconveniently long time". As I have written many times before, when a value position is established, one can never be sure which potential return pathway will be taken. Effectively, any value position falls into one of three categories:
- those that enjoy a re-rating as the market more generally recognizes a mis-pricing has occurred (type I)
- those that generate a higher return via dividend yield, but are not immediately re-rated (type II)
- those that simply don't recover, the value traps. (type III)
So patience really should be a virtue for value managers as long as we are dealing with a type I or type II value stock. The chart below confirms just how strongly this is true. It shows the cumulative returns to an incredibly simple value strategy (buying the lowest 20% of the MSCI Europe ranked by trailing PE). The chart provides very graphic evidence for the rewards to patience.
The strategy tends to generate around 3% outperformance relative to the market in the first 12 months. But if you carry on holding for another year, this rises to 5.7% (a year two return of just over 2%). However, at three-year time horizons and beyond, the excess return pick-up is much sharper, running at the rate of 8-10% p.a. for years 3, 4, and 5!
Also noteworthy, is that the value strategy appears to start working from day one. This surprised me as I expected to see a period of underperformance or non-performance, rather than an immediate return to the value approach.
This finding sits well with the fact that successful value investors seem to display patience. The average holding period for our group of long-term outperforming value managers was 5 years, against the average holding period of just 11 months for stocks listed on the NYSE, and just over 1 year for the average US mutual fund.
A recent paper by Fama and French shines some light on the relative probability of each type of value situation occurring. They examine both the returns and the probabilities of transition amongst US value and growth stocks over the period 1926-2004.
They formed portfolios based around the interaction of price to book and size. In the table below, V = value, G = growth, N = neutral, B = big and S = small. The columns give four possible outcomes; the column labeled 'Same' refers to the group of stocks that one year later are in the same style and size basket as they were in the original sort. The 'Plus' column covers stocks that have moved into a higher price to book category, i.e. value stocks that have moved back towards market average pricing (or been acquired).The 'Minus' column depicts the opposite: here we find stocks that have been growth stocks and are now returning to, say, more normal valuations. The final column heading is 'change in size' and reflects stocks that switch between size groups from t to t+1 - that is small stocks that become big, and big stocks that become small.
So let's take BV (big value) as an example. The upper part of the table shows the average p.a. excess return over the market that such stocks have generated since 1926. So stocks that started in BV and then moved back towards the market average pricing, generate 17.5% outperformance p.a. (type I value stocks from our earlier discussion). However, stocks that start in BV and are still in the BV universe one year later have, on average, generated 3.4% p.a. outperformance (type II stocks). Value traps return an average -34.5% p.a.
Of course, these statistics mean nothing on their own. We need to know the frequency with which they occur, in order to gauge how important they are. A 17.5% p.a. excess return sounds very impressive, but it doesn't mean a lot if there is essentially no chance of it actually occurring. The middle part of the table details the average transition probabilities. For instance, in the BV portfolio, some 75% of stocks remain in the BV portfolio after a year (type II value stocks). Around 23% move to a higher price to book ratio, which moves them out of the BV portfolio (type I value stocks). A mere 2.3% of stocks end up being very serious value traps. (type III stocks).
In fact, Joseph Piotroski has gone even further. He found that a minority of value stocks created the US value premium. Piotroski uncovered the fact that only 42% of value stocks outperform the market on a one-year view. We now know from the Fama and French work that 23% of the value stocks re-rate towards a more normal multiple, and that this generates sizeable returns for the investor. This implies that only around 25% (or one in four) of the type II value stocks generate positive excess returns. This helps to explain why the type II value stocks only generate 3.4% p.a. excess returns: the category of type II value stocks is a broad spectrum that hides a multitude of sins.
In our own work we have found that 50% of the stocks in our simple value screen outperform over a one-year time horizon. This implies that investors who follow such strategies not only need to be patient, but also need to avoid narrow framing. If they look at the portfolio in totality then such investors are likely to be ok, even in the short-term. However, if they frame narrowly and start to look at the individual performance of stocks within the portfolio then problems could arise. With a relatively high percentage of stocks underperforming, the pain of running the strategy would be psychologically hard to bear, and result in the abandonment of the process.
One potential solution is to combine a value selection with a stop-loss system. We leave the investigation of this for another note. But it certainly looks like value investing could be said to be about avoiding losers.
The chart below shows the improvement in the percentage of correct calls as we extend the time horizon. Interestingly, the percentage of stocks outperforming the market rises over time, but not massively. So a 50% hit ratio rises to a 57% hit ratio when a five-year time horizon is used. This improvement presumably stems from a greater proportion of value stocks managing to turn their businesses around as time goes by.
So does time arbitrage pay for growth investors? The answer is a resounding no. Patience just results in more and more stocks being torpedoed. The chart below makes the point graphically. Over our sample period, growth stocks have lost, on average, 6% p.a. in the first year of portfolio construction. Year two sees a further 3% loss. In year three a further 2% is lost, and then things heat up with years 4 and 5 witnessing around 6% p.a. declines. As Peter Tasker might say, time is a killer not a healer for growth stocks.
In terms of the percentage of stocks that outperform, the numbers don't make for pretty reading. Only around 38% of the stocks in the growth portfolio have an above-market return. So successful growth investing is all about picking winners, i.e. identifying the stocks that can generate positive returns.
When we extend the time horizon we see the hit ratio decline. That is to say, the longer we wait, fewer and fewer stocks manage to outperform the index. Effectively, the risk of being hit by an earnings torpedo increases as time goes on.
Growth and momentum
One obvious way in which we might be able to improve on growth portfolios is to introduce a momentum filter. To test this we take the top 20% of stocks by PE and then split the portfolio into three, based around the past 6 months' price momentum lagged by one month.
The impact that this has is dramatic. The chart below shows the high and low momentum portfolios for the growth universe. The benefits of using momentum are massive. The median p.a. performance improvement is over 400bps.
Of course, the absolute returns are still negative, but the relative gains are massive. Now personally I've never been a great fan of the relative performance game (a future weekly will explore this in more depth), but the potential value added by inclusion of momentum into a growth strategy can't be ignored. Moreover, the best benefits of the momentum input are achieved when the holding period is extended!
In terms of improving the hit rate of the screen, the inclusion of momentum improves the percentage of stocks outperforming from 38% to 41% at the one-year time horizon.
Value for growth investors
Of course, there is an alternative approach which growth investors might consider - use value. The chart below shows the excess 12-month returns to our various quintiles, with each quintile further broken down into three value baskets. It shows that the cheapest of the value stocks achieve relatively little outperformance, with the majority of the value premium being generated by the other two groups.
However, look at the growth stocks. The cheapest of the growth stock significantly outperforms both the growth category generally (by nearly 3% p.a.) and the most expensive of the growth stocks (by nearly 7% p.a.) So growth investors would be wise to remember that value matters - even for growth stocks.
So bizarrely, growth works within the value universe, and value works within the growth universe. Ah, the ironies of investment management.
Value and momentum
So if momentum is so useful in a growth context, what is its impact in a value universe? The results are less impressive than their growth cousins. The median value added is 130bps p.a. Not inconsequential, but not in the same league at the growth results. The chart below shows the cumulative returns for the high and low momentum portfolios within the value universe.
It transpires the improvement in the hit ratio is roughly similar to that seen in the growth arena. The 12-month hit rate rises from 50% to just over 55%.
This finding that momentum works better for growth stocks than for value stocks was first documented by Cliff Asness for US stocks. The chart below, taken from Sebastian's magnum opus on momentum, provides a simple view of the conclusion for European stocks. Here each PE quintile has been split into three, based on price momentum. The difference between high and low momentum stocks in the value universe is around 4% p.a. However, the difference between the high and low momentum stocks in the growth universe is much higher, at around 12% p.a!
The fact that momentum adds relatively little to performance of a value portfolio, but adds major benefits to a growth portfolio, has implications for the portable alpha addicts out there (you know who you are).
If value managers find it hard to beat their benchmark (more on this below) but growth managers can do it far more easily, then a structure which is long value beta and then transports the alpha from a momentum using growth manager, should be fairly attractive.
Indeed, in a new paper Chan, Dimmock and Lakonishok show that across a wide variety of benchmark measures, growth managers seem to be better able to generate abnormal returns relative to their universe than value managers do. Chan et al found that the average alpha (from 1989-2001) for US growth managers is 2.6% p.a.; the corresponding number for value managers is 1.2% p.a. -entirely consistent with the finding outlined above.
A similar picture is painted by the chart below. Here we have taken data from Houge and Loughran who formed two groups of US fund managers based around their factor loadings on the value premium. Those in the highest loading quartile are value investors, those in the lowest quartile are growth.
The chart compares the returns on the simple stock decile rankings from 1965-2002, against the returns from the two manager groups (net of fees) over the same sample period.
The value universe showed a 16% p.a. return over the period. Value managers (net of fees) showed only 11% p.a. return. The growth universe showed a 10% p.a. return, whilst growth managers showed a 9.9% (net of fees) return. So value managers seem to have trouble actually capturing the value premium.
However, above and beyond all else, we have also shown that patience is certainly a prerequisite for value investors, whilst growth investors would be better off ensuring momentum is one of the factors in their stock selection process and remembering that they shouldn't ignore value.
Your learning patience is a virtue analyst,