Hi, Three Questions on the section about Shiller's test of the semi-strong form of the Efficient Markets Hypothesis (EMH): 1. This CT8 post: Ch1 & Ch8 | Actuarial Education (acted.co.uk) explains Shiller's observed "systematic forecast errors" to mean systematically over-estimating or under-estimating the price. Does this mean that Shiller's discounted dividend model always overestimated the price of each equity at each month over the 100 years? 2. Could one argue that the results of his test also contradicts the strong form of EMH - since insiders (eg Directors) may know future dividends (in the same way that Shiller knew the 100 years of dividend data) but it still wouldn't help them predict the future share price? 3. Please explain these criticisms of Shiller's methodology (page 12 of notes): "1) the choice of terminal value for the stock price 2) the use of a constant discount rate 3) bias in estimates of the variances due to autocorrelation 4) possible non-stationarity of the series, ie the series may have stochastic trends which invalidate the measurements obtained for the variance of the stock price." Wouldn't the terminal value be the observed share price at end of 100 years? And what is meant by 3) and 4)? Thanks in advance!
On reflection, a simple answer to Question 2 is that disproving semi-strong EMH automatically disproves strong EMH (since if markets don't auto consider public information they certainly don't consider confidential information) . But appreciate any answers to Questions 1 and 3 (above).
Yes, exactly on Question 2! For Question 1, not quite. What Shiller's discounted dividend model is doing is underestimating the variance of the stock price. In other words, what was being observed was that historic prices were being underestimated by the model in good times and overestimated by the model in bad times. For Question 3: - the terminal value is the discounted value of dividends if the share is held in perpetuity. However, Shiller used an arbitrary value for the terminal value in the model. - 3) means that the model did not take into account that there is a correlation between stock prices in different periods, which would have affected the estimated variances. - For 4), Shiller's model assumed stationarity, ie the variance of the dividends is constant (and independent of the time). However, this may not be true in practice.
