Finance Experimental IEOR Spring 2012 Mike Lipkin, Pankaj Mody
Lecture 5f
Dynamics
Consider the following scenarios: Stock XYZ; price, S0= 50.00; 3 weeks to go to expiration. Earnings date: 4 weeks away. For concreteness, we take the front month options to be the Junes. Which option generally has the higher implied vol, the Jun 50 C or Jul 50 C? Suppose that XYZ announces a change in the earnings announcement, moving the date ahead 1 week. What will happen to the implied vols? Suppose XYZ preannounces earnings today;
what will happen to the vols? Will it matter whether the announcement is better than expected, or worse?
Usually, only bad earnings gets preannounced.
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 2
Lecture 5f Some basics:
Dynamics
How many times a year are earnings announced? What would happen if a stock fails to announce earnings?
Imagine that earnings are coming out in 2 days (Jun expiry), and XYZ drops $3 to $47.00.
What will happen to the Jun 50 vol?
Suppose earnings are announced and XYZ drops $3 to $47.00.
What will happen to the Jun 50 vol?
What is the difference between these two scenarios?
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 3
Lecture 5f
Dynamics
There are two kinds of new information that get disseminated in the marketplace. They are scheduled events and unscheduled ones. It is often pretty easy to distinguish between the two. Let s try some examples:
Earnings Drug trial results Upgrades/downgrades by analysts Terrorist bombing in USA or Western Europe Articles in the news media Fed open market meeting/short rate change Mergers/take-overs/acquisitions State/federal actions for improprieties Corporate personnel changes (CEO, CFO, etc.)
Mike Lipkin, Alexander Stanton Page 4
Experimental Finance
Lecture 5f
Dynamics
One of the things which we should like to understand is how the volatility surfaces adjust themselves before and after both kinds of events. In a thorough research project, one would examine stocks in different industry groups, of different market caps, etc., and look for regularity. Is there an existing theory which addresses these concerns? No. Note: Theory is different than empirical results. Good (predictive) results will never get published!
Why???
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 5
Lecture 5f
Dynamics
Earnings announcements come (usually) at very specific, well-defined times. What frequency? For some stocks, earnings are a small effect;
which ones might these be?
For others, earnings announcements move the stock more than any typical daily move. As a result, the implied volatilities increase strongly heading into earnings. In this way, IVs are anticipative. The following is a graph of the IVs for CAT over a six-month interval. (Brown curve; ignore the blue.)
Can you identify the earnings dates? About how long before earnings does volatility appear to begin climbing?
Problem Set V explores this issue.
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 6
Lecture 5f
Dynamics
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 7
Lecture 5f
Dynamics
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 8
Lecture 5f
Dynamics
Day before LNKD earnings last week
Experimental Finance Mike Lipkin, Alexander Stanton Page 9
Lecture 5f
Dynamics
Day after LNKD earnings
Experimental Finance Mike Lipkin, Alexander Stanton Page 10
Lecture 5f
Dynamics
Drug announcements come in two varieties.
There are scheduled dates for stage trial announcements, but also sudden news releases.
I m not sure which one applies to the following, but you can see the potential for trading opportunities and blunders!
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 11
Lecture 5f
Dynamics
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 12
Lecture 5f
Dynamics
Monday, Mar 14, 2005 Interim Analysis of Phase III Trial Shows Avastin Plus Chemotherapy Extends Survival of Patients with First-Line Non-Squamous, Non-Small Cell Lung Cancer -- First Positive Phase III Results with an Anti-Angiogenesis Therapy in Lung Cancer --
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 13
Lecture 5f
Dynamics
When a corporate event happens suddenly and unexpectedly, a typical response in the market is to have a large size trading day. We have just seen this with DNA. However, size trading can accompany big increases or decreases in volatility and sometimes no change at all. The DNA event, a large upward price jump, was accompanied by a big spike in volume. Below are two spikes in volume coinciding with down moves.
What do you imagine may have happened with the following news event? Why?
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 14
Lecture 5f
Dynamics
McDonald s chairman, CEO dies unexpectedly Cantalupo suffers heart attack; fast-food giant taps replacement
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 15
Lecture 5f
Dynamics
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 16
Lecture 5f
Dynamics
When a news event is anticipated, such as earnings, there is a lag time for dealing with the event. The volatility must go up for earnings, drug announcements, etc.
Why? Can you think of a future, scheduled event which will reduce volatility? (We will discuss such an event in a later week.)
What would cause the volatility to go up slowly? In other words, why wouldn t the vol stay high from earnings to earnings? Let s take a look again at a blow up of the CAT preearnings chart:
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 17
Lecture 5f
Dynamics
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 18
Lecture 5f
Dynamics
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 19
Lecture 5f
Dynamics
This is why vol doesn t stay high from start to finish. Rising vol just means prices decline at a slower pace.
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 20
Lecture 5f
Dynamics
is It important to understand the change in volatility heading into earnings announcements. For typical curves of this sort there are two elements of interest:
The size of the change, and The characteristic time scale over which this change occurs.
Why would it be insufficient to only know one of these properties? Characteristic time scales can be eye-balled off the graph, however if the growth curve is exponential, it is conventional to identify the half-life of the curve, the time required to double in value (from a baseline). Is there a well-formulated theory of this effect in the literature?
The only one I know is: Johannes, Michael S. and Dubinsky, Andrew L., "EarningsAnnouncements and Option Prices" (June 2005). SSRN: http://ssrn.com/abstract=600593
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 21
Lecture 5f
Dynamics
In Problem Set V you explore earnings announcements and volatility growth. Enough about volatility before these events.
What can we say about volatility after these events?
The behavior of vol about scheduled and unscheduled events will generally be very different.
Why?
How do you expect CAT vol after earnings to compare with CAT vol well before earnings? (What does well before mean?) What are some of the consequences of this understanding? What about vol after the CEO of McDonald s dies suddenly?
(There may be a characteristic time post this event).
The following two slides show Hewlett-Packard (HPQ) through its earnings event: AMC 2/18/09, near months then mid-months.
Experimental Finance Mike Lipkin, Alexander Stanton Page 22
Lecture 5f
Dynamics
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 23
Lecture 5f
Dynamics
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 24
Lecture 5f
Dynamics
Now let s consider the vol surfaces. For simplicity let us restrict the discussion to one stock, one series. (For concreteness, we could imagine the XYZ Jun options with May being the front month.)
What is the usual shape of the volatility surface for this series? What will happen if the stock experiences a gradual price change which shifts the at-the-$? What will happen if the stock experiences a sudden price change which shifts the at-the-$?
Is there a theory which covers this behavior? No.
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 25
Lecture 5f
Dynamics
Let s be blunt about standard option pricing theory! It applies when every option is well-priced. ONLY! In other words, if conditions materially change, standard option theory will not be able to distinguish between the need to alter the parameters of the model used and the presence of arbitrage!
I am plenty redundant about this point!!!!!!!!!
When a stock drops dramatically, the vol often changes. But it can go down and up! A theory would be a dynamic theory, but there is no such theory currently. An attempt to patch statics to dynamics is sticky strike/sticky delta. Problem Set V explores this. The following two slides show recent flashcrashes: AAPL; MNKD
Mike Lipkin, Alexander Stanton Page 26
Experimental Finance
Lecture 5f
Dynamics
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 27
Lecture 5f
Dynamics
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 28
Lecture 5f What is sticky strike? What is sticky delta?
Dynamics
Sticky strike postulates that as the stock moves the vol skew stays put. This gibes with our intuition that as the stock moves lower the volatility might go up. But is this true? What if XYZ drops suddenly on uncertain news? What if XYZ drops suddenly because of definitive news (such as earnings or a drug trial results)? Will up moves be different than down moves?
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 29
Lecture 5f
Dynamics
Sticky delta postulates that as the stock moves the vol skew stays with the corresponding option, delta by delta. This gibes with our intuition that the at-the-$ options should have a depressed vol. Why? Should a time scale matter here? In other words, if the stock drifts gently up or down is this different than if the stock shoots quickly to another value? How would you define such a time scale? The same kinds of spikes can happen in the entire market s volatility. Here is a 3-year graph of the VIX. The data set I used ended with the onset of a vol spike in May 2006.
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 30
Lecture 5f
Dynamics
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 31
Lecture 5f
Dynamics So, here is a mini-quiz! The following slide is a picture of a stock I traded for a number of months in 2006. Can you look at it and deduce what happened to the volatility surface from before to after the event in question? One thing that did not change much was the realized vol on either side of the event! Why would the implied volatility not be a reflection of the realized volatility? The key story is that implied volatilities assimilate the expected movement over an extended time horizon. They are a poor man s representation of a jump process.
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 32
Lecture 5f
Dynamics
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 33
Lecture 5f
Dynamics
Here is a similar stock, in this case prior to an announcement:
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 34
Lecture 5f
Dynamics
Subsequent to an event, the vol may be ca. 60.
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 35
Lecture 5f
Dynamics
Here is VMW before the Jan 2008 earnings announcement:
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 36
Lecture 5f
Dynamics
What do you think happened to the vols after this event? Can you tell from the candlesticks what happened to the realized vol? We will come back to this product next time when we look at hardto-borrows because another exciting non-standard thing happened.
Experimental Finance
Mike Lipkin, Alexander Stanton
Page 37