Using Cash Flow in a Quantitative Model

Publish date:

Cash flow is a tempting alternative to earnings in valuing companies. Focusing on cash flow rather than earnings allows investors to build a better estimate of private market value, and cash flow ratios more fully adjust for risk arising from balance sheet and operating leverage.

Cash flow is certainly popular among professional money managers. According to a recent

Merrill Lynch

quantitative poll quoted in


, price-to-cash flow is a more widely watched ratio for institutional investors than price-to-earnings. When I worked as a buy-side quant, cash flow was the most heavily weighted factor in my five-factor stock ranking system.

This preference for cash flow is linked in part to the professed "value" orientation of most institutional money managers. Growth stock managers, to whom individual investors are perhaps more closely attuned, are likely to emphasize earnings over cash flow.

Peter Lynch

, for example, has written that investors should focus on "earnings, earnings, earnings."

Mr. Market Uses Earnings Estimates

Consensus valuation is clearly based on earnings estimates. This is shown on a bottom-up company level by the high r-squared for prices versus estimated earnings, and from a top-down perspective when looking at the market's estimated earnings yield relative to fixed-income yields.

The r-squared of price versus 12-month forward estimated earnings for companies in the

Russell 1000

is 0.74, meaning 74% of the variability in price is explained by the level of estimated earnings. This is far higher than the r-squared for prices versus trailing earnings (0.52), estimated cash flow (0.52), trailing cash flow (0.43), book value (0.54) or sales (0.16). There is also a strong relationship between earnings estimates and overall market valuation, as shown in a prior Quant View


Defining the Cash Flow Ratio for a Quantitative Model

One of the biggest problems with cash flow is definition, and there are virtually as many formulations of cash flow as there are investors using the ratios.

The definitions used in quantitative ranking systems are likely to differ markedly from those used by individual investors or analysts. Instead of sitting down with a stack of 10Qs, a quant is working with large electronic databases provided by the likes of






, and the timeliness and quality of the data sets are as important as the algorithm.

Cash flow ratios including changes in working capital, for example, are very difficult to implement for a broad range of companies, because the data irregularities swamp the information content in the results. If half the companies resulting from a screen are there because of extraordinary items or data inaccuracies, the screen isn't useful in its raw form.

In my ranking system, I originally (1987-1990) used a simple price/cash flow ratio, with cash flow defined as nine-month forward estimated earnings + estimated depreciation. Earnings estimates were from I/B/E/S, depreciation estimates were from the Value Line Estimates and Projections database. We caught a lot of takeovers with this ratio.

With the


blowup in 1990, the leveraged buyout era ended and our price/cash flow ratio became pretty ineffective. Companies with high debt were poison, so I backtested an alternative cash flow ratio including long-term debt in the numerator, (price+ltd)/cash flow. My backtest showed comparable performance with less volatility, so I switched to the new formula. The early 1990s were a bad period for value money managers, and this formulation, along with the concept of keeping our sector bets neutral, helped us perform in line with the broader market.

In 1993, the bias against high debt had worked its way through the system -- I tracked the relative performance of debt-laden companies -- and I took out the bias against debt by moving to a definition of cash flow as EBITDA/Total Cap. The actual formula was:

(est pretax income + est depreciation & amortization + trailing 12 months interest expense) / (price + all debt + 80% of lease obligations)

This last formula was probably most similar to the enterprise value ratios popular today.

Cash Flow for Financials Pose a Problem

The debt and interest expense values available on most electronic databases for banking and financial companies are sketchy and misleading. As a result, I excluded interest expense and debt from the cash flow ratios for financial companies in my ranking system.

This adjustment turned the cash flow ratio for financial companies into a glorified price-to-earnings ratio, obviously not an optimal solution. Most importantly, by excluding interest expense and debt, the financial companies' cash flow ratio wasn't adjusting for financial leverage.

I was stumped. If you've worked out a good way to formulate a cash flow ratio that adjusts for financial leverage in banking and financial companies, drop me a

line and I'll mention it in the next Quant View.

Ted Murphy ( operates the MarketPlayer Web site. Prior to MarketPlayer, he was a partner at

Equinox Capital Management.