ooo

Overreaction and bias in the stock marketOverreaction and bias in the stock market

Back to forecastingBack to forecasting

Current stock value = PV future dividends

P = DP = D11/(r -g)/(r -g)

D1 = next expected dividend

r = required return

g = expected dividend growth rate

Where does "g" come from?Where does "g" come from?

EarningsEarnings11 = Earnings = Earnings00 + (Ret)Earnings + (Ret)Earnings00(ROE)(ROE)

If the retention ratio (RetRet) remains constant over time,

EarningsEarnings11/Earnings/Earnings00 = Dividend = Dividend11/Dividend/Dividend00 = 1+g = 1+g

g = (Ret)(ROE)g = (Ret)(ROE)

The growth in dividend depends on:

• the proportion of earnings reinvested back into the company• ROE

Forecasting in a Forecasting in a

We try to guestimate:

• earnings growth• the length of the growth period

Mean reversionMean reversion

The tendency of earnings to revert to an average trend over the long run.

EXPERIMENT 1 (Dechow and Sloan) EXPERIMENT 1 (Dechow and Sloan)

Future vs. past earnings growthFuture vs. past earnings growth First year:

Rank NYSE stocks based on their P/E ratios.

Form ten portfolios from the cheapest to the most expensive stocks.

For each portfolio calculate the average growth in earnings for the last five years.

Then calculate the growth in earnings for the following five years.

Second year:

Re-rank the stocks according to P/E ratios and redo the above calculations.

Keep doing this for 23 years.

The relationship between P/E ratios and past and future earnings growth.The relationship between P/E ratios and past and future earnings growth.

The X-axis represents the ten portfolios, from the most expensive (1) to the cheapest (10).

The Y-axisrepresents past andfuture earningsgrowth.

DiscussionDiscussion

Firms that had earnings growing fast in the past (expensive stocks) will experience a relative slowdown in the future.

Firms that had earnings growing slowly in the past (cheap stocks) will experience a relative acceleration in the future.

Earnings appear to revert to the mean.

From bias to surpriseFrom bias to surprise

Overestimating the duration of the mean reversion and the true value of the average growth rate can cause the market to overreact.

That is, some prices will be pushed too high, while others will drop

too low.

True True earningsearnings

Case 1Case 1

today True horizon

Forecasted Forecasted earningsearnings

Case 1: DiscussionCase 1: Discussion

If:

True growth horizon = Forecasted growth horizon, and

True growth = Forecasted growth

No overreaction

True True earningsearnings

Case 2aCase 2a

today True horizon

Forecasted Forecasted earningsearnings

Case 2bCase 2b

today True horizon

True earningsTrue earnings

Forecasted earningsForecasted earnings

Case 2: DiscussionCase 2: Discussion

If:

True growth horizon < Forecasted growth horizon, and

True growth = Forecasted growth

The market would be surprised by:

- the relative poor performance of growth stocks :(

- the relative good performance of value stocks :)

Both surprises would be of equal magnitude.

Case 3aCase 3a

today True horizon

Forecasted earningsForecasted earnings

True earningsTrue earnings

Case 3bCase 3b

today True horizon

True earningsTrue earnings

Forecasted earningsForecasted earnings

Case 3: DiscussionCase 3: Discussion

If True growth horizon = Forecasted growth horizon, and True growth < Forecasted growth

The market would be surprised by the relative poor performance of all stocks.

pleasant surprises caused by growth stocks would be larger in

magnitude than the unpleasant surprises caused by value stocks.

Case 4aCase 4a

today True horizon

Forecasted earningsForecasted earnings

True earningsTrue earnings

Case 4bCase 4b

today True horizon

True earningsTrue earnings

Forecasted earningsForecasted earnings

If True growth horizon < Forecasted growth horizon, and True growth < Forecasted growth

The market would be unpleasantly surprised by the relative poor performance of growth stocks and the relative good performance of value stocks.

Unpleasant surprises caused by growth stocks would be larger in magnitude

than pleasant surprises caused by value stocks.

Case 4: DiscussionCase 4: Discussion

EXPERIMENT 2 (La Porta):EXPERIMENT 2 (La Porta):

Analysts' forecast revisions Analysts' forecast revisions Year 1:

- In April, rank NYSE stocks based on analysts' consensus about future growth

- Build ten portfolios (from low growth to high growth)

- The following April, compare the revised estimation with the original ones.

Year 2:

- Re-rank the stocks and redo the procedure.

Keep doing this for several years.

Revision of earnings forecasts

The X-axis shows the ten portfolios, from the lowest expected growth (portfolio 1) to thehighest expected growth (portfolio 10)

The Y-axisshows earningsforecasts in twoconsecutiveyears

Green: Current earnings as forecasted in the previous year

Red: Earnings revisions in the current year

DiscussionDiscussion

Investors appear to overestimate the growth rate and the growth horizon.

It appears earnings grow at a lower rate and revert to the mean faster than forecasted.