Our equity strategist, Lars Kreckel, is not too worried about equity valuations. However, with the CAPE (cyclically-adjusted price-earnings) ratio the highest since the dot.com boom, many clients fear the stock market is set to decline. But while valuations might look expensive on this measure, there’s a difference between the ‘right’ long-term price and the price investors are willing to pay ‘right now’. Even with higher interest rates, some of our models suggest equity valuations could hold up if economic volatility remains low and growth continues to be solid.
The CAPE ratio in the chart is the S&P500 price divided by a 10-year moving average of earnings. The measure was invented by Graham and Dodd and made famous by Nobel prize winner Robert Shiller; it has historically had predictive power for long-term returns. The current level appears elevated, even if we adjust for methodological changes to the way companies report earnings (in particular write offs).
Yet asset prices are cyclical. They undergo booms and busts, just like economies. What determines the price investors are willing to pay for equities at different stages of the economic cycle? A couple of recent papers tried to model this issue:
Their models highlight structural and cyclical factors that influence valuations over time. I summarise them as GRIP:
Investors are typically willing to pay more for equities if: growth is strong, interest rates are low, inflation is close to target and economic uncertainty is low.
The Fed paper sees equities as slightly overvalued today. By contrast, Minack’s analysis suggests equities are still significantly cheap.
After deconstructing their models, I thought Minack’s was particularly interesting.
Minack's model has three advantages. First, it has a better historic ‘fit’. Second, several of the variables the Fed paper uses are not known in ‘real time’ (such as the neutral rate of interest and potential growth). Minack's measure of uncertainty is also simpler (he uses five-year GDP growth volatility while the Fed uses the change in the estimated neutral interest rate).
Finally, Minack models the earnings ‘yield’ rather than the price-to-earnings ‘ratio’. This is more appropriate when using bond yields to model equity valuations. This might sound trivial but it's not. A linear decline in the equity earnings yield from 5% to 4% to 3% to 2% implies a non-linear acceleration in the PE ratio from 20 to 25 to 33 to 50 times earnings. With bond yields low, shouldn't PE multiples be high?
To my surprise, it wasn’t lower bond yields that pushed up the estimate of fair value in recent years. Instead, it was a decline in economic volatility. Minack used the five-year standard deviation of US GDP growth. That collapsed in 2015 as the great recession dropped out the data set.
One criticism is that this is arbitrary. Why five years? Why not 10? A 10-year measure of GDP volatility is still high as it captures the last recession. It's hard to argue for a sudden re-rating of equities. But the big picture is that investors are risk averse when they fear a recession. They become more confident when the memory of recession fades, driving equity valuations higher. Equities should therefore do well until inflation rises above target, monetary policy is tightened and recession risks build – something that could occur in 2019.
I have tweaked the insights from the San Francisco Fed and Minack for my own CAPE model variant. The main adjustment I made is to take into account weaker structural growth. While Minack's model pointed to significant upside risk to equities I find valuations slightly cheap – even if interest rates rise further from here – so in line with Lars' findings.
As one of many inputs into our investment process, we look at several different valuation indicators. My model suggests we might not need to fear the CAPE for now. But everything hinges on being able to spot the pick up in economic volatility and onset of recession in advance.