Assessing US equities

One of the frustrations about being short stocks in the current environment is that one hears a constant refrain of how "cheap" they are. Indeed, Macro Man saw a US equity strategist from a C-list house a couple of weeks ago, and this guy was ashamed to admit that his analysis of the numbers would allow him to envisage "only" a 30-40% rally over the next year.

Now, eighteen months ago, Macro Man took a look at US stocks and liked what he saw. And so he remained constructive on equities through the end of last year. Early in 2008, however, he changed his mind on the fate of the US consumer, and thus his opinion on equities. Ever since, he's been biased towards neutral to short positioning in indices. Still, when in an introspective mood the other week, he realized that it was time to get cracking with trying to develop a more robust version of his old equity beta portfolio. In 2006, his work prompted him to go long despite his underlying suspicions about the sustainability of earnings. Would 2008 generate the same result?

Now, at first glance, US equities don't look particularly cheap at all. Punching up the SPX on his Bloomberg, Macro Man finds that it has a trailing P/E ratio of around 23...that hardly has "buy of the century" written on it, does it? The man from the C-list shop showed Macro Man a chart suggesting that the P/E was actually about 14 if you strip out all the writedowns, and that this level represented good value.

Uhhh...OK. Presumably if we strip out the writedowns, we should also strip out the accounting profits in 2005-2007 that were caused by the securities that were written off? And if we're stripping out the writedowns, surely we also need to strip out the "profits" generated by the fall in the financials' liabilities? Finally, Macro Man's friends at Merrill Lynch and Citi might also take issue with the notion that writedowns should be stripped out because they are a "one off."

Regardless, perhaps trailing P/Es don't tell the whole story. We also need to look forward, not back. And absent the kind of failsafe crystal ball that Macro Man has yet to locate in his 15 years in this business, that means relying on analysts. Now, simply looking at expected earnings growth isn't going to cut it, because of the Panglossian cloth from which equity analysts are cut: they almost always expect earnings growth. However, we can look at how their optimism changes over time; the chart below shows the six month momentum on analysts' rolling 12 month forward EPS estimates for the SPX. And what we observe here is that analysts' expectations have only slowed this much during recessions. And in each of the previous two recessions, downgrade momentum troughed before equities broke out to the topside. Chalk one up for the "not so cheap" camp!
Now granted, SPX earnings yields (the inverse of the PE) look pretty attractive relative to bond yields, even when using trailing data. This was the metric that Macro Man used in his old beta plus portfolio. But using that metric to evaluate equities carries the implicit assumption that bonds are fairly priced. And if there's one lesson of all that's happened since last July, it's that bonds of all description have not been fairly priced for quite some time. Comparing the S&P's earnings yield with another metric, the cost of living, yields a less satisfying result; the real earnings yield is less than 1%, down sharply from the levels in excess of 3% that prevailed this time last year. Again, not exactly compelling, is it?
Macro Man took these factors, and a couple of other fundamental inputs (no sentiment or momentum data allowed), and developed a couple of simple models to determine the attractiveness of US equities. Now, one thing that he found, which backed up his prior results, was very clear: just because stocks are not a buy does not make them a sell. In fact, he could find no consistent signals to market time the stock market from the short side without torturing the data to the degree that he violated the Geneva Convention.

So the output of his efforts is basically a signal that says "go long" or "stay flat." The first of these was a scorecard approach, wherein each of his factors provide a positive (+1) or negative (-1) signal, and the sum of the factors is totted up, much like in an investment manager's scorecard. Gratifyingly, Macro Man found a strong positive correlation between his scorecard reading and subsequent market returns. And it was pretty clear from sifting through the data that for any reading below +3, you might as well go to Vegas and put all your money on red, as equities are not a winning proposition.
Macro Man was quite pleased to see that his intuition was mirrored by his research result; after a long period in which US stocks looked very attractive, the scorecard recently fell through the key threshold and actually registered a negative reading in April.


He also took a different approach, taking the same basic factors but using regression analysis to generate a forecast market return over the next twelve months. The r-squared of this analysis was high enough to suggest statistical significance but low enough that Macro Man could be pretty sure he wasn't over-fitting what were, after all, a pretty simple set of factors.

He then compared this return forecast with a measure of trailing volatility (though perhaps VIX would be better) to get a "forecast Sharpe ratio." The benefit of this methodology over the scorecard one is that it accounts for the magnitude of the signal from each of the factors, rather than simply assigning a binary rating. It should provide an idea of just how good or bad the outlook for stocks is. And currently, the outlook is, ahem, less than rosy, with negative returns forecast for the next twelve months. Macro Man found that it's really only worth being long equities when the forecast Sharpe is in excess of 1; over the past 20 years, a strategy of being long the market when this ratio is above 1 and flat when below has delivered a risk-adjusted return of....err...1.0.

Now, this methodology has been kept deliberately simple to avoid the traps that many of the equity quants found themselves in last summer. But Macro Man approached it with a tabula rasa; he was more than prepared to accept a signal that the US market was a buy, to implement it, and then trade around it. But he can't make the numbers lie. On the basis of what's worked since 1988 (and indeed since 1970, on the basis of a stripped-down model that excludes analyst expectations), US stocks may be cheap, they may be expensive; but one thing they are clearly not is a buy.