Market efficiency Specific meaning of the term market

Market efficiency Specific meaning of the term “market efficiency” in financial economics: “security prices fully reflect all available information” So… this is “Informational Efficiency” Realistically: an efficient market is one in which information is quickly incorporated into the price. Tests of market efficiency refer (since Fama, 1970) to three progressively more demanding concepts: • Weak form • Semi-strong form • Strong form

Market efficiency tests The difference between the three lies in what information is taken into account: • Weak form is all info from past prices incorporated? • Semi-strong is all publicly available info incorporated? • Strong form is all info, public or private, incorporated?

Market efficiency • Weak form no additional info from past prices tests of return predictability • Semi-strong all publicly available info incorporated event studies • Strong form all info, public or private, incorporated event studies NB: if strong form is true, then the value of security analysis becomes suspect—though someone must be doing it

Return on equity of takeover targets around date of takeover attempt

Market efficiency Two aspects, both of which are relevant: • Stock picking • Timing (especially for the market as a whole)

Tests of return predictability 1. Time patterns • • Monday January – 33% (1962 -78) 5% (small firms) 3% (large firms) Why? Tax losses? (but worked before 1917)

Tests of return predictability 2. Prediction from past returns (a) Correlation / regression (best to use residual from CAPM or APT model to eliminate role of time-varying risk-premia) Quite a few show significant, albeit small, correlations – especially small shares)

Tests of return predictability 2. Prediction from past returns (b) Runs tests Ignore size of changes and just look at positive versus negative (avoids overemphasis on unusual extreme events)

Tests of return predictability 2. Prediction from past returns (c) Filter rules Such as: “buy when the stock has increased by (say) 0. 5% from its previous low; sell when stock has fallen by (say) 0. 5% from its previous high. ” (Based on the idea that small fluctuations don’t mean news – big fluctuations do)

Tests of return predictability 2. Prediction from past returns (d) Momentum For the next month, hold only the stocks that have appreciated over the past year (or could do this for shorter-term fluctuations) Used to generate good returns, though high transactions costs. Has faded Similar: “relative strength”: stocks that are high relative to recent average

Tests of return predictability 2. Prediction from past returns Most of these tests show small inefficiencies, (especially at very high frequencies – seconds) Most could not be exploited profitably by investors because of transactions costs (Transactions costs include the bid-ask spread employed by specialists to defend themselves against better-informed investors)

Tests of return predictability 3. Returns and firm characteristics The Size Effect (small cap have high returns) Market-to-book (value firms have high returns) Earnings-to-price (high earnings firms have high returns) (These could all be “factor-mimicking” characteristics – capturing additional dimensions of risk )

Tests of return predictability 3. Returns and firm characteristics The Size Effect 20% extra return on very small vs. large, 1936 -77 (risk-adjusted by CAPM) Only affects lowest quintile of firms Most of it happens in January!

Size effect Ibbotson, cited in BKM

Small firm portfolios Cochrane, EP 99

Tests of return predictability 3. Returns and firm characteristics Why the size effect? Maybe beta poorly measured Small stocks rarely traded – leads to bias in beta estimate Shrinking firm may have out-of-date beta Is CAPM a good enough risk adjuster? Using other risk factors (though not size) in an APT reduced the size factor a lot (11% to <2%) Key factor here: risk-premium on corporate bonds relative to govt bonds

Tests of return predictability 3. Returns and firm characteristics Why the size effect? Transactions costs especially high Compensation for liquidity Impossible to profit from effect?

Tests of return predictability 3. Returns and firm characteristics The Book-to-market effect Gap of 8% per annum between highest and lowest deciles of firms ranked by the ratio of their book value to market value (high B/M predicts high return) Maybe value firms are close to bankruptcy but recovered – so these firms are not a hedge for recession, hence high yield “Value” (high B/M) and “Growth” (low B/M) stocks (These terms not quite synonymous with B/M) The Earning-to-price effect But this seems to have no independent effect after size and market -to-book taken account of. The neglected firm effect

Book to market effect French, cited in BKM

Tests of return predictability 3. Returns and firm characteristics Questions to ask about any of these anomalies: • • • Is it real? (Data-mining) Are we just missing another factor (especially one correlated with investors’ consumption/welfare) (i. e. firm characteristics correlated with a risk-factor that the market prices but we have omitted) Mis-estimate of beta (or other factor loadings)? Transactions costs? Risk premia changing over time? (CAPM/APT use stable λ’s) Truly inefficient?