Momentum for short-term tactical play, CAPE for the longer run

We found the CAPE ratio gives a good signal for predicting longer-term equity returns in a model where we assume it will mean revert to its long-term average.

In one of our previous notes, we created a tactical portfolio and showed how it performed compared to its strategic benchmark based on three rule-based models applied to equities, bonds and commodities respectively.
In this note, we are revisiting our equity model specifically as it has been the least efficient in signalling timely investment opportunities in the equity complex. We first look at the CAPE ratio and examine if it can be improved. We then provide a practical example of its enhanced returns in a portfolio.

The CAPE ratio
One commonly used ratio to assess whether a stock or an equity index is over or undervalued is the price-to-earnings (PE) ratio. In order to make it comparable over time, the price is adjusted for inflation (real price), as is the earnings per share (real earnings), resulting in the real PE ratio.

The CAPE (cyclically adjusted PE) ratio is the result of real price divided by the 10-year average of real earnings. If the CAPE ratio is above the 10-year average of the real PE, the equity index is seen as overvalued and vice versa.
In this note, we have replaced the average real PE by the median real PE to remove unnecessary noise in the data. We then take the difference between the CAPE and the current median real PE as an indicator of over/under valuation. We call this our ‘valuation indicator’.
The previous chart shows a relatively strong correlation between the valuation indicator and the index annual return (here the MSCI US Index). Since January 2014, both lines started to diverge, suggesting that valuation multiples may not be as efficient as they used to be in predicting equity returns.

Shortening the average period of earnings used
In this section, we tested whether using a shorter average earnings period would improve the signal provided by the CAPE. Our analysis shows that using the median earnings over a 5-year period (new valuation indicator) rather than the traditional 10-year period (old valuation indicator) improves the correlation between the valuation indicator and the index return since January 2014.

We believe the market has been willing to pay a premium for earnings visibility and therefore willing to re-rate already elevated multiples. We have observed a similar re-rating in the past, between 2005 and 2008, prior to the great financial crisis. In the following sections, we will look at how it translates in terms of performance in a portfolio of 27 equity indices.

The CAPE model for the longer run
In this section, we have created two portfolios based on the previous valuation indicators and compared them to our equity benchmark, a subcomponent of the MSCI AC World with the same index methodology but 27 constituents out of 48. Both portfolios assume that the CAPE ratio will mean revert to its long-term average. In both cases, the model overweights the five cheapest indices and underweights the five most expensive.
Our analysis shows that both portfolios outperform the benchmark if it holds its positions for at least four years as illustrated in the chart below. We observe that the portfolio using the new valuation indicator has a slightly better information ratio than the portfolio using the old valuation indicator (model 1).

The below table shows the performance of portfolios with a holding period of four years. Model 1 outperforms the benchmark by 0.3% with a lower level of volatility, enhancing the Sharpe ratio to 0.14 from 0.12.

In addition to higher returns and lower volatility, model 1 recovers faster to previous peak and provides better protection against the downside risk.

A model for short-term tactical play
In this section, we look for a model that can outperform the benchmark over a holding period shorter than four years.
When model 1 is applied to a portfolio that rebalances every month, the Sharpe ratio declines to 0.10 as the annual return deteriorates.
For monthly returns, we find that a momentum strategy using the 5-year CAPE ratio works best. This new model (model 2) overweights the five most expensive indices while underweighting the cheapest five.

The above chart shows that model 2 outperforms model 1 and the benchmark by 0.7% and 0.4% per year, respectively, since January 1970. The Sharpe ratio improves from 0.12 for the benchmark, 0.10 for model 1 to 0.15 for model 2. The below table shows the performance of model 1 and model 2 when rebalancing every month.

Shortening the average period of earnings used from ten to five years in the valuation indicators shows a better correlation with the index annual return. In a portfolio of equity indices, a strategy based on a mean reversion of the CAPE outperforms the benchmark when the positions are hold over a period of four years at least. Using five-year average earnings rather than ten increases the outperformance slightly. Both portfolios based on the old and new valuation indicators enhance the Sharpe ratio from 0.12 for the benchmark to 0.143 on average.
However, over a shorter holding period, the CAPE model underperforms and we found that a momentum strategy that actually overweights (underweights) expensive (cheap) markets works best. A portfolio that rebalances monthly using this momentum strategy enhances the Sharpe ratio to 0.15.

Edith Southammakosane – Director –Multi-Asset Strategist – ETF Securities