Security Selection Intrinsic Valuation安全选择的内在价值.pptVIP

* Value of relative valuation good: not as time-intensive as dividend discount models growing evidence that simple measures pay abnormal risk-adjusted returns (after controlling for other factors) bad difficult to hold “all else constant” across many securities different accounting methods across firms, and within a single firm across time different time-periods appear to have different results: sometimes growth stocks outperform * P/E Analysis PE’s of some loosely related retailers except for XOM and MO Data as of 1/10/2008 Who is cheap, who is expensive? * A Generalized Decomposition of the PE ratio Let dt = ktet = payout ratio * earning per share then, let e1 = e0(1+g1), e2 = e1(1+g2), etc… Note that g1 need not equal g2 , and that g can be positive or negative then, then, divide by e0 * P/E Decomposition Further simplification available if we let g be constant: Therefore, In order for P/E ratios to provide insight on relative value of one stock versus others, we must hold these factors constant * P/E Analysis - PEG ratio A simple control for g is to divide P/E ratio by earnings growth rate. I went backward here – found the PEG online and backed out the implied growth rate. Note the different PEG ranking relative to PE ratio The Motley Fools say: PEG 0-0.5 = BUY; PEG 0.5 - 0.65 = WEAK BUY PEG 0.65 - 1.00 = HOLD; PEG 1.00 = SELL * Deeper P/E Analysis More factors (k and required rate of return) are also likely to matter We could run a regression of the form, PE = ? + ?1(g) + ?2(k) + ?3(req return) + ? * P/E Analysis – Regression Screens Basic Steps in Regression Screens 1. Use all firms in industry/sector/universe - your choice, but be careful... 2. Regress P/E on its underlying parameters 3. Examine errors +? implies overvalued - ? implies undervalued Problems Regressions only use those stocks with a P/E ratio sample does not include negative earners Noisy estimates (garbage-in, garbage-out) Missing var