Lecture 11 With Lecture Notes

Ecient Markets

(Welch, Chapter 11)

Oleg Shibanov

New Economic School

March, 2014

Did you bring your calculator? Did you read the chapter ahead of time?

Oleg Shibanov (NES) Ecient Markets March, 2014 1 / 51

Perfect vs. Ecient Markets

I An ecient market is one that sets the price correctly.

I Historically, there has been much confusion here: most people mean perfectmarkets when they say ecient markets.

Perfect Market) Ecient Marketbut

Perfect Market(== Ecient Market

I Our coverage is highly abbreviated. (An investments course covers marketeciency (ME) in much more detail.)


Covered Topics

I ME means roughly that \prices are set right".I The problem is: what does right mean? We need to have a good model for

the right target.

I Classication Schemes for Market Eciency (ME).I Classical Finance vs. Behavioral Finance.

I Advantages of ME: you can learn from market values; and you know whereyou can add value and where not.

I Event Study Methods.

I Arbitrage and Great Bets.


What does it mean to believe the market is ecient?

That the market uses all information in its setting of the current price, to give youwhatever the rate of return should be [according to model].


IllustrationA Specic Example: PepsiCo

The market estimates PepsiCo's expected value next year to be $55 per share. Italso estimates all other interesting characteristics, such as cash ows,market-betas, covariances, liquidity, etc.

Say the CAPM is the correct pricing model. Then the nancial market looks atPepsiCo's market beta, the risk-free rate, and the expected rate of return on themarket, and sets PepsiCo's expected rate of return. Say this CAPM expected rateof return is 10%.

The price today is $55=1:1 = $50 per share.

(Q: What would you conclude to have gone wrong if you saw a price of, say,$45.83 today?)

The General Case

The nancial markets estimate the statistical distribution of future cash ows,including their expected cash ow values, covariances, liquidity, and anything elsepossibly of interest.

The nancial market determines the appropriate expected rate of return, given allvalue-relevant characteristics.

The market sets today's price, so that the expected rate of return is as the modelstates.


What is the link between perfect markets and ecientmarkets?

If agents are not stupid (and there are innitely many smart ones), then anyperfect market will become ecient.

However, even in an imperfect market (e.g., with transaction costs), the marketcould still be ecient, setting prices at whatever level is appropriate.


You see that the price of IBM is such that you expect it toearn 20% over the next year. Can you conclude that themarket is inecient?


What sort of claims would reject ME?

An expected 10% in one day is contradicting ME, because it is not plausible thatany reasonable equilibrium model could tell you that you should earn 10% in oneday.


Is market eciency a stronger concept over short intervals(a day) or over long intervals (a decade)?

Over short intervals, we expect about a zero rate of return.

Over long intervals, our models of how stocks should be priced are very unreliable.

Thus, it is a stronger concept for short horizons.


In itself, is ME a very strong claim? As believer, how canyou dispute someone doubting it?

Nope. You can always claim that the model was wrong. Some people believe MEis close to religion, which for long periods it may be.


What types of markets are more likely to be (in-)ecient?

Think about extent of market perfection. The closer a market is to being perfect,the more likely it is to being ecient, too.

Of course, if the market is not perfect, there may not even be an ecient value.


Traditional Classications (EM)

Focuses on information availability:

Strong Form: Price reects all public and private information. You cannot outperform(\make money") even with insider information. (Noone believes this one.)

Semi-Strong Form: Price reects public, but not all private information. You cannot make moneywith public information.

Weak Form: Price reects enough public and private information that you cannot makemoney by plotting historical price patterns|but you could still make moneyanalyzing other aspects, such as company fundamentals.


More Modern Classication (EM)

Focuses on the relation between price reecting underlying value, and closelylinked to behavioral nance:

True believer: Price is always PV of the rm's cash ow.

Firm believer: Price deviates from PV, but this is not exploitable.

Mild believer: Price deviates from PV, and exploiting it is possible, giving you asan investor a mild edge.

Non believer: Price deviates strongly from PV, so investors can easily get rich.


Causality

True market eciency implies unpredictable stock prices, i.e., strong orsemi-strong form eciency.

Strong or semi-strong form unpredictability does not imply true (underlyingfundamentals-based) market eciency.

Take \unpredictability" loosely here. It could be that expected returns themselvesare time-varying, e.g., because the risk-prole is time-varying. In this case, it maybe predictable that you get higher average returns when risk is higher.Unpredictable here means \relative to expectations".


Why is the debate over ME so dicult to settle?

The signal-to-noise ratio is very low. See the next slides.


According to the CAPM or similar models, what do youexpect the expected rate of return on a trading day be?

Maybe 10%/255 0.04% per day (4bp per day). More realistically, 0.02% perday.

I This is based on random walk: E (P1) = P0 +m + , where m is small.

I Transaction costs will further reduce this 0.04% in an imperfect world. Aone-day roundtrip transaction will have you lose money.


What is the typical daily move up or down of a stock? Ofan Index?

I Around 1%-3% for stocks.

I Around 0.5% to 1% for portfolios, such as S&P500 or Dow-Jones.

Signal = Mean. Noise = Standard Deviation.


How does risk (standard deviation) grow with the holdingperiod duration (time) in a random walk?


What is a T-statistic that gives you statistical condence?


What kind of an investment edge does it mean to be aninvestments superstar?

If you can expect to outperform by an extra 2% per year, you are a star! This isless than 1 basis points a day.

Presume you are a true superstar.


Over 1-day, what is your expected performance T-statistic?

1 basis point/200 basis points, or 0.005.


Over 100 days, what is your expected T-statistic?

100 basis point/(200 10) basis points 0.2.


Over 10,000 days, what is your expected T-statistic?

10,000 basis point/(200100) basis points 0.5. This is 40 years.With about 160 years, you can get statistical condence!


What is the problem in testing for ME, and testingwhether you can outperform the market?

The problem is the low signal-to-noise ratio in rates of returns.

(End of topic)


What exactly is arbitrage?

I No possible negative cash ows|some positive cash ows with non-zeroprobability.

I A positive investment amount is a negative cash ow up-front. Therefore,investing in a Treasury is not arbitrage.


What is a great bet?

For example, a bet with equal probability payos of {$1 or +$100 million is notan arbitrage, but it is a great bet.


Who would prefer an arbitrage to a great bet?

Everyone would like to take any arbitrage opportunity there is.

Depending on risk aversion, not everyone would like to take a great bet.

But, given the choice of either-or, a non-innitely-risk-averse investor may prefer agreat bet to a tiny non-scaleable arbitrage.


Do you believe it is easy to nd either?

Nope.


Finance Models

All reasonable nancial models impose the belief

that there is an absence of both non-trivial great

bets and non-trivial arbitrage opportunities.


What is Technical Analysis?

People who look at patterns of historical stock prices and try to predict themarket with it.


What is the historical empirical evidence?

First-order: the U.S. nancial markets are reasonably ecient with respect topublic information. It is very dicult to get rich easily. Few funds manage tooutperform. It is close to random.

Second-order: There may be some \anomalies" that seem to oer a tiny bit morethan what seems reasonable. The two main equities-related anomalies are

I Momentum (at least a specic form thereof)|although much ofmomentum's average rate of return of 1% per month is probably simplycompensation for risk. We learned this in the nancial crisis, where thezero-investment momentum portfolio ($1 long, $1 short) lost more than $1 inone year!

I Value vs. growth|value rms prefer much better than glamorous growthstocks, but they did not do so in all situations.

There are non-equities and other more specialized anomalies, too.


If you can beat the market, who would you tell?

No one! Keep this information to yourself and trade to get rich.

The only problem occurs if you know how to beat the market but you are poor. Inthis case, you have a worse problem. If you tell someone too much detail, theydon't need you any more.


How much data do you need to prove to investors that youare good?

See before: 100 years for few-stock strategies, 30 years for large-portfoliostrategies.


How do (hedge/mutual) funds get started?

In-house for a couple of years. The bad ones close, the good ones are advertised.Survivorship bias.


How many funds should outperform the market 10 years ina row if none have skills?

About half every year (not exact, of course) | so about 1/1000 in 10 years


How many funds should outperform the market 10 years ina row if some have skills?

A few more than 1/1000 (not exact, of course).


Among existing, large funds, how many funds should haveoutperformed the market with/without skills for 10 years ina row?

A lot more than 1/1000 (not exact, of course) due to survivorship bias.

Tests of persistence of outperformance perform similarly poorly. That is, the fundsthat do the best in one periods usually do not do the best the following period.


Is Berkshire-Hathaway a good investment?

Not good, not bad. For sure, its historical performance and behavior is not asecret. Even if Buett is good at making money, this is certainly already known toinvestors, who have bid up the BH price to take this into acount.

(This was also explained at the end of Chapter 2.)


Who would get the rents from Buett's abilities?

The fund manager (or existing investors). There is more than enough outsidemoney to fund great managers. Money by itself does not bring anything great tothe table.


If you were an investment manager having made 5% peryear above your benchmark ve years in a row, what wouldyou think of your capabilities?

Very highly, of course. You would sco at academics like me.


What do you think of contingent compensation|you payme only if I give you a protable stock pick? Will this notremedy the problem of ignorant managers not wanting toget into the business?


What is the empirical evidence for ME?

The empirical evidence is pretty consistent with ME.

Momentum and maybe value/growth strategies have outperformed by just a littlebit|though there appears to be \fat-tail" risk.


Are there any corporate consequences of ME?

I You can learn from your own market value.

I You can learn from your competitors' values.

I You can learn from other values.

I You cannot add value by doing things that investors can do (or undo). [splits,dividends, etc.]

I You cannot make money by trying to time interest rates or gambling oncommodities.


As the CEO, what should you do if your shares areundervalued, relative to your (possibly private)information?

Don't use the CAPM.

Don't take projects.

Usually, repurchase your own shares with any extra cash you have|they are goodinvestments.


As the CEO, what should you do if your shares areovervalued, relative to your (possibly private) information?

Don't use the CAPM.

Raise capital and take all sorts of projects. If you can, invest in more projects likethe one you are currently investing in, but don't invest in negative NPV projects.You can always invest in Treasuries.


How long should it take for the market to incorporatenewly public information into the asset price?

Almost instant|note limits by transaction costs.


Should you be able to earn excess returns if you tradebased on information released two hours ago?

Nope. However, this is within transaction cost bounds.


Should you be able to earn excess returns if you tradebased on facts that were widely expected to occur?

Nope.


What should happen when a rm (or the government)makes an announcement?

The price adjusts really fast.

Note that the price response is just the resolution of uncertainty. It is not the fulleect.


Can we use ME to determine what sort of corporateactions or laws/regulations are good or bad?

Yes, but magnitudes are tough, because we rarely know expectations.


What kind of other events may be usefully explored?

Dividend changes, earnings announcements, etc..


Documents

Lecture 11 With Lecture Notes