Revaluing value

PORTFOLIO POINT: We may see a new, lower range of valuations for risk assets generally, and equities specifically.
Getting the earnings cycle (E) right is only half the battle with equities: The other issue is how investors will value those earnings (P/E). We may, in my view, enter a new, lower range of valuations for risk assets generally, and equities specifically.

Valuation shifts are not that common, but neither are they unprecedented. Arguably, we have already seen one in this cycle. The popular “Fed model”, which compares the prospective earnings yield on equities to the 10-year Treasury yield, has shifted its centre of gravity. Exhibit 1 shows the ratio between the two yields. When the ratio is low, equities are relatively attractive. In fact, equities have never been cheaper (relatively speaking, and over the period shown).

However, it's also clear from Exhibit 1 that the ratio has entered a new lower range since 2002. The level that was a relative “buy” signal for stocks prior to 2002 has since been a “sell” signal. Put another way, having mean-reverted around a ratio of 1, the ratio now persistently reverts to about 1 3/4.

Rising risk-aversion may explain this shift. Yet that hardly seems plausible in a cycle where other risk spreads plumbed new lows. The Fed model shift may reflect concerns about the sustainability or quality of earnings. Certainly, both concerns seem warranted.

First, it's worth noting that the prospective earnings yield is based on consensus earnings forecasts. It's well known that those forecasts are almost always wrong (but if consistently wrong, that's not an issue). A bigger issue is what earnings are being forecast. Consensus forecasts aim for the earnings series that a company highlights. Famously, this has become “earnings before bad stuff”. Exhibit 2 shows the wide – and widening – gap between trailing “core” earnings and trailing GAAP earnings. In short, earnings today aren't what earnings were yesterday. By itself, this would warrant a shift in the Fed model “fair value”.

The second issue is the sustainability of earnings. I keep harping about an earnings bubble. Investors may be less willing to pay “normal” valuations for equities when earnings on almost every measure have been above long-term averages through this cycle.

Because of the bubble, I prefer to use valuation measures such as the Graham-Dodd P/E, which is based on a trailing 10-year earnings series. But even that metric may be affected by the bubble. As Exhibit 3 shows, the current 10-year average for earnings (relative to GDP) is well above the long-term average. The Graham-Dodd P/E is now around its long-term average of 16. But if profits decline to the long-term average, then, all else equal, the Graham-Dodd P/E is closer to 26 1/2.

The value of equities relative to GDP captures (approximately) the combination of elevated earnings and elevated valuations. On this basis, equities remain well above long-term averages (Exhibit 4). This series peaked in 2000 but remains above the long-term average. We may now be in a two-stage process of normalising this metric. The first phase (2000–02) corrected the valuation bubble. The second stage (starting this year) will correct the earnings bubble. This double-phase bear market may return valuations to historical norms.

Shift in absolute valuation could be associated with a new range for relative valuation measures. Note that relative valuations were at exceptional levels through the past 25 years. Exhibit 1 is based on prospective earnings yield, but earning forecasts are available only since 1985. Exhibit 5 shows the gap between trailing earnings and the real 10-year Treasury yield. The range seen through the past 25 years was clearly exceptional, and the average gap was well below the long-term average.

To be fair, long-term averages may not be relevant. What many long-term valuation series suggest is that valuations go through long cycles. Exhibit 5 alternates between a low yield gap (or risk premium) in the 1920s, 1960s, and 1990s and high risk premiums in the 1930s/1940s and 1970s/1980s. Not coincidently, the periods of low risk premiums have coincided with credit booms.

The issue going forward may be whether we enter one of the periods of low valuation/high risk premiums. I don't have a dogmatic view on that. The point is that investors should be aware that that is a risk, and the ranges that applied through the past 25 years have been, on a long view, exceptional.

Gerard Minack is chief market strategist of Morgan Stanley Australia.