Which long-term smoothed (10-year average) stock valuation metric works best to screen stocks? In the end-of-September 2013 version of their paper entitled “On the Performance of Cyclically Adjusted Valuation Measures”, Wesley Gray and Jack Vogel compare the abilities of five valuation ratios expressed as yields to predict U.S. stock returns, as follows:

- 10-year average real earnings to market capitalization (CA-EM)
- 10-year average real book values to market capitalization (CA-BM)
- 10-year average real earnings before interest, taxes, depreciation and amortization to total enterprise value (CA-EBITDA/TEV)
- 10-year average real free cash flow to total enterprise value (CA-FCF/TEV)
- 10-year average real free gross profits to total enterprise value (CA-GP/TEV)

They adjust for inflation using the monthly U.S. Consumer Price Index (CPI). Each year on June 30 (with accounting data lagged to ensure availability), they sort stocks into tenths (deciles) based on each ratio. They then calculate monthly decile returns over the next 12 months based on equal initial weights, with either annual (next June 30) or monthly decile reformation. For monthly reformation, they recalculate market capitalizations (but not valuation ratios) each month. They also test a stock return momentum overlay, each month splitting each valuation ratio top decile into high-momentum and low-momentum halves, with momentum defined as cumulative return from 12 months ago to one month ago. They do not account for trading frictions or taxes. Using stock prices and accounting data for U.S. firms with at least 10 years of the required accounting data and market capitalizations above the 40th percentile for NYSE-listed stocks, and contemporaneous CPI data, during 1962 through 2012, *they find that:* Keep Reading