Skip to main content

Savita Subramanian is the chief quantitative strategist at Merrill Lynch and her recent Quantitative Primer 2016 is among the best of the thousands of research reports I've ever read. It's a 240-page how-to guide on the most successful investing techniques – valuations, stock price momentum, institutional holdings analysis and technical analysis measures – for U.S. stocks using all performance data for the past 30 years.

With this column, I'll present Ms. Subramanian's conclusions on the right tool for the job – the valuation measures that have generated the highest risk-adjusted returns in U.S. market sectors – and then apply the findings to the Canadian market using the bank sector.

The table below shows the valuation metrics Ms. Subramanian identified as the best guides to generate the highest risk-adjusted returns in each sector. Performance is measured by average annual returns, and risk is measured by both standard deviation of performance and percentage possibility of negative returns. Special mention should be made for historical P/E and historical relative P/E, which are likely not applicable for most investors. Ms. Subramanian uses a series of complicated adjustments to account for the profit cycle and inflation in order to calculate historical price-earnings ratios.

U.S. SectorMost Successful Valuation Metric 1986 to Present
MediaForward P/E, EV/EBITDA
RetailingFCF/EV, Price/FCF
Consumer StaplesFCF/EV
EnergyHistorical P/E
BanksHistorical Relative P/E, Forward P/E
InsuranceHistorical Relative P/E, Forward P/E
Health Care EqtP/FCF, FCF/EV
Pharma and BiotechP/Sales
Capital GoodsFCF/EV
TechnologyFCF/EV, P/FCF
TelecomP/Sales, Forward P/E
UtilitiesP/Sales, EV/EBITDA

One of the biggest surprises in the report was that forward price-earnings ratios work better for investors in bank stocks than price-to-book ratios. Conventional wisdom is that price-to-book is the better tool for bank stocks but Ms. Subramanian found that using this measure creates too much volatility, increasing the odds of investor losses.

There are a number of issues in translated Merrill Lynch's findings on bank stocks to the Canadian market. The data are far less extensive – the forward estimated earnings data necessary to calculate forward price-earnings ratios are only available from 1993 for domestic banks – far less than the 30 years used by Ms. Subramanian. In addition, the Canadian banking sector is dominated by six banks whereas there are currently more than 5,000 commercial banks operating in the United States.

Those caveats aside, the charts below show that forward P/E has been an effective valuation tool for domestic bank investors, particularly in the post-financial crisis period. The charts show the price-to-book value and price-to-forward-earnings estimates ratio for the S&P/TSX bank index compared with the forward two-year returns for the index.

Correlation analysis (not shown) indicates that price-to-book ratio has been the better indicator of future bank index performance for the 13-year period shown as a whole. However, since September, 2009, forward-price earnings has been more effective at predicting domestic bank stock returns.

Follow Scott Barlow on Twitter @SBarlow_ROB.