We are all well aware that gamblers are losers on average. If they were not it would not pay the casinos, race courses, bookmakers and lotteries to supply them with gambling opportunities. Nor would governments be able to tax gambling winnings as heavily as they do were not gamblers as eager as they are to gamble on the unfavourable terms they do, made all the more unfavourable by heavy taxes on their winnings.
What is not as clear is why gamblers on average prove so willing to apparently throw away their income. The answer is they enjoy the process, the frisson of perhaps winning big and sometimes doing so. The vast majority of gamblers, perhaps more than 98% of them in SA, are well able to limit their losses to a small proportion of their incomes. On average about 1% of disposable incomes are spent on gambling activity of all kinds in SA.
Technically gamblers who trade off expected losses for the pleasures they receive are not risk averse as is conventionally assumed; they are risk loving, playing a game for which the outcomes are not normally distributed around zero. The outcomes are very much skewed to the right hand side of the distribution: many small losses with a small probability of a few big wins.
And so gamblers accept much less than the mathematical odds implied by a normal distribution of outcomes for the opportunity to win big. Or in other words they pay up for the chances they take. It has been established conclusively for US race tracks that the actual odds of a 100-1 outsider winning a race is about 160-1. Researchers with lots of data and computer power at their disposal have calculated the expected betting return from all US horse races run between 1992 and 2001.
These results were shown by Chris Holdsworth in a recent report written for Investec Securities (Long shot bias and the equity market, Investec Securities, 18 April 2011)) that extends the analysis to the equity market in SA.
The worst bets on the US race courses, in the sense of what you can expect to get back on the basis of historical outcomes, have been on the longest shots. US punters on the races should expect to lose over $60 for every 100 to 1 bet they make. The “fair odds” would have been about 160 to 1, that is 100-1 long shots win only once in 160 attempts not 100.
The explanation for this willingness of gamblers to pay above the theoretical odds for the chance of a big win is surely their taste for risk. They value the small chance of a big win much more than they fear a small (even near certain) loss.
Gamblers who play a lottery, that typically pays out only about 50% of what is taken in, do so for the same reason – for the chance of a really big life changing win. They are in fact risk loving rather than risk averse and pay up accordingly. Government controls over the supply of lottery type games and bookmakers as well as, more recently, online gambling in the US of course prevents the potential gambling competition from improving these especially poor lottery odds or indeed the odds on the race track or the spread at football games.
Holdsworth found that analogous to the results of the gambling research, investors “over pay” for the opportunity to invest in companies listed on the JSE with well above market average PE ratios. The attraction of the high flying companies for risk lovers is that when the companies with high PEs actually grow their earnings even faster than the market expected, as implicit in high trailing multiples, the return can be spectacularly good. And so the risk lovers looking at the far right distribution of potential outcomes drive up valuations and generally overvalue and pay above the theoretical normal distribution odds for the average high PE stock. The ordinary risk averse investor is deterred as much or more so by expected losses as much as they are encouraged by expected returns. This is not necessarily so for the risk lovers.
Holdsworth pursued the analogy of the taste for high PE stocks with the taste for long shots on the race course in the following way
To quote his explanation of the method he used
“……..At the beginning of each year from 1994 to 2010 we ranked the constituents of the ALSI by 12m trailing P/E. We then measured the return of each of these stocks over the subsequent 12 months including dividends. We grouped the stocks into deciles based on their P/E within each year. For each year we then had ten equally weighted portfolios based on starting P/E. If our classification is correct then the dispersion of one year returns for the stocks in the high P/E decile should have a much larger tail on the right hand side than that of the low P/E stocks. The top 1% of returns for the high P/E stocks (represented as 0.99 percentile in the chart below) were above 750%. Top 1% of returns for the low P/E portfolio were just under 400%. …….high P/E stocks have a higher propensity for very large returns over one year than low P/E stocks. If this characteristic attracts risk seeking investors, as we think it does, that would imply a lower expected return.
For each year in our sample we measured the return of each decile relative to the average of all the deciles. We then summed up the averages for each decile across the 17 years in the sample. The chart for average returns for each decile is remarkably similar to the horse race chart above .The cost of gambling in the market is high. On average the high P/E portfolio underperformed the average of the deciles by 10% p.a. over the 17 year window. This portfolio would have contained some spectacular winners but their outperformance would have been drowned out by the remaining large number of constituents with sub par performance. The low P/E portfolio, while containing fewer stellar performances, would have outperformed the average of the deciles by just under 4% per annum. Like long shot horses, investors have consistently paid over the odds for high P/E stocks……”
Holdsworth explained that as with gambling it would be sensible for risk lovers in the share market to strictly limit the number of long shots taken and the scale of the investment made in them. Accordingly there also would be no point in holding a number of such stocks. The more diversified the portfolio of very high PE stocks the larger the chance of realising the predicted well below normal returns, if the history of past performance on the JSE is relevant. It would be the equivalent of taking all the tickets in a lottery or raffle where the prize is worth less than the tickets sold.
We would suggest with Holdsworth that investing in high PE stocks it is similar to making a bet where expected losses can be sustained in the hope of a big win. Such action is clearly not for the faint hearted with limited wealth at their disposal.
These results, as with betting odds, should not be regarded as representing market inefficiency or market failure. Rather they represent competitively determined outcomes, given the important presence in the market place of risk loving behaviour.
The risk averse can benefit from these risk loving propensities by betting mostly on the shorter odds favourites that come in the form of the well established blue chips. They have proven track records and whose earnings are not likely to deviate greatly from expectations. These companies will not be expected to shoot the lights out as will be reflected in their average PE rating.
And in the share market, unlike the gambling markets, investors would be playing a positive (after much lower expenses including taxes on winnings) game where the sum of the gains can be realistically expected to exceed the sum of the losses over time. Past performance indicates very clearly as much, as Holdsworth has painstakingly confirmed.
Such advice would not come as a surprise to the typically cautious fund manager or advisor. They might be pleased to know that their experience and intuition is indeed very well supported by past performance on the JSE.
To view the graphs and tables referred to in the article, see Daily Ideas in todays Daily View: Daily View – The price of luck: Why betting on the long shots or the high PE companies is expensive