The Performance of Options-Based Investment Strategies:
Evidence for Individual Stocks During 2003─2013
Michael L. Hemler
University of Notre Dame
Thomas W. Miller, Jr.
Mississippi State University
OIC Financial Advisors Forum
Hotel Fontainebleau
Miami Beach, FL
May 8, 2015
UNIVERSITY OF
NOTRE DAME
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
1
Prologue: What Motivated This Study?
• In Spring 2013 Alan Grigoletto of the Options Industry Council (OIC) contacted Mike
Hemler regarding a potential options research project.
• Alan noted that large pension funds used options to alter the risk-return profile of
their investment portfolios, e.g., the Ontario Teachers’ Pension Plan and the New
Jersey Public Employees’ Pension Plan
• Alan also noted that existing studies on the performance of option strategies
typically focused on covered calls or collars using index (mostly S&P 500) options.
• Alan suggested that large institutional investors would welcome a study that
compared the performance of the standard “buy and hold” strategy versus option
strategies that include long equity. They would be especially interested in a study
that focused on individual stocks, rather than equity indexes.
• See the CBOE website http://www.cboe.com/micro/buywrite/ for references on
options-based strategies and benchmark indexes. In addition, see the CBOE March
2015 Risk Management Conference: http://www.cboe.com/rmc/2015/Day-1Session-2-Speth.pdf and http://www.cboe.com/rmc/2015/Day-1-Session-2-BlackSzado-projector.pdf.
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
2
Introduction
• Options are a convenient and effective tool for portfolio
managers who want to tailor equity positions, augment
income, or limit risk in changing market conditions.
• A popular strategy is the covered call (or, buy-write) strategy:
it combines a long equity position with a short call position.
– Whaley (2002) examines the CBOE Buy-Write Index (BXM) during 1988-2001,
and concludes that “BXM outperformed the S&P 500 portfolio by
approximately 0.2% a month on a risk-adjusted basis.”
– Hill, Balasubramanian, Gregory, and Tierens (2006) analyze various S&P 500overwriting strategies, concluding that such strategies “have very favorable
performance characteristics at a range of risk levels.”
– Kapadia and Szado (2007) examine the buy-write strategy on Russell 2000,
concluding “the buy-write strategy can outperform the index.”
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
3
Introduction
• Another popular strategy is the collar: it combines a long equity
position, a short call position, and a long put position.
• A collar is simply a covered call that adds a long put position.
– The option parts are sometimes known as “split-strike” conversions.
– One can form a “zero-cost” collar, and combined it with long equity.
– The resulting payoff profile resembles a vertical bull spread.
• Collar studies include:
– Szado and Schneeweis (2010) analyze collars based on the PowerShares QQQ
exchange-traded fund. With their method, they conclude that collars gave
superior returns and reduced risk compared to the buy-and-hold strategy.
– Szado and Schneeweis (2012) analyze collars across a range of asset classes,
including equity, currency, commodity, real estate, and fixed income. They
conclude that “for most of the asset classes considered, an options-based
collar strategy … provides improved risk-adjusted performance and significant
risk reduction.”
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
4
Introduction
• In this study, we investigate the performance of optionsbased strategies.
• Our analysis differs in two major respects:
– We compare the relative performance of four (not one) optionsbased strategies versus the corresponding long equity strategy.
– We focus on individual stocks, not indexes (e.g., S&P 500).
• S&P 500 options are European; they cannot be exercised early.
• Options on individual stocks are American; they can be exercised early.
• This early exercise feature complicates the analysis if:
– There are dividends and the option strategy involves a short call position.
– The option strategy involves short put positions.
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
5
Payoff Diagrams for Strategies We Investigate
(Note: We shall not be holding strategies until expiration.)
Covered Call:
Collar:
Long Stock, Short Call
Long Stock, Long Put, Short Call
Protective Put:
Covered Combination:
Long Stock, Long Put
Long Stock, Short Put, Short Call
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
6
What Do We Do and What Do We Find?
• What Do We Do?
– We examine the relative performance of four option strategies and long equity
using ten stocks commonly used in 401(k)s.
– We examine performance during 2003-2013, both over the entire period and
the first and second halves.
– We use four popular performance measures (Sharpe ratio, Jensen’s alpha,
Treynor ratio, and Sortino ratio).
– For simplicity we ignore early exercise and transaction costs.
• What Do We Find?
– In general, the covered combination and covered call strategies outperform
the long stock strategy
– In general, the long stock strategy outperforms the collar and protective put
strategies.
– Our findings are preliminary because we do not have definitive answers to:
• What is the impact of early exercise?
• What is the effect of transaction costs?
• To what extent does the Leland (1999) critique alter the interpretation of CAPM-based
performance measures?
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
7
The Stocks in The Study
•
We wanted a small, representative sample that might especially interest large
institutional investors who typically “buy and hold” stocks.
•
We also wanted a “third party” to choose the stocks in order to avoid any
accusation of “cherry picking” firms for our sample.
•
We use a list of ten stocks given in the CNBC story of September 12, 2012,
entitled “Widely Held Stocks in 401(k)s.”
– See http://www.cnbc.com/id/48939384/page/1.
•
CNBC asked Morningstar to compile a list of the top ten firms in 401(k)s.
•
In decreasing order: ExxonMobil, Comcast, Berkshire Hathaway (Class A),
Oracle, Microsoft, Coca-Cola, Amazon, Wells Fargo, Google, and Apple.
– There are no exchange-traded options on Berkshire Hathaway (Class A), but there are such
options on Berkshire Hathaway (Class B).
– Rather than drop Berkshire Hathaway, we substitute Class B for Class A.
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
8
Sample Period and Data Sources
•
Our sample period runs from January, 2003 through August, 2013, which is the
last month for which we have options data.
– There are 50 observations for Berkshire Hathaway, 108 observations for Google, and
128 for all other stocks due to different dates for introduction of option trading.
•
For convenience, in our sample:
– We call 2002-2007 the first half.
– We call 2008-2013 the second half.
•
Our stock and T-bill data are from the Center for Research in Security Prices
(CRSP).
– Stock returns are value-weighted returns including distributions.
– T-bill returns are based on a thirty-day target maturity.
•
Options data are from OptionMetrics, with the exception of early exercise data
from the Options Clearing Corporation (OCC).
•
Options returns are based on bid-ask midpoints.
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
9
How Do We Implement the Option Strategies?
• We choose options that are at least 5% out of the money, but with strike
prices as close as possible to the opening stock price on the day strategy is
initiated.
• We focus on monthly returns.
– We roll over positions monthly on the last business day.
– When we choose options for a given month, we pick options expiring the next month.
– For instance, to calculate the return for a covered call during January, 2003, we choose
the call traded throughout January which expires in February, 2003.
January Trading Days
With Feb.
(1)
Expiration
February Trading Days
(2)
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
With Mar.
Expiration
10
Empirical Results
• We compute monthly returns for the four option strategies and the
long equity strategy for each of the ten stocks.
• In general, we find that compared to the long equity strategy:
– The covered call yields: lower standard deviations of return and (often)
higher mean returns.
– The protective put yields: lower standard deviations of return and
lower mean returns.
– The collar yields: lower standard deviations of return and
lower mean returns.
– The covered combination yields: higher standard deviations of return and
higher mean returns.
– Returns for all options strategies exhibit extreme non-normality, either
skewness or kurtosis or both. This fact raises major issues when interpreting
standard performance measures such as Jensen’s alpha.
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
11
Comparing Mean Returns and Standard Deviations of
Returns Between Each Option Strategy and Long Equity
• For each stock and option strategy, calculate the differential
between:
– the mean return of the option strategy and the mean return of the long equity
strategy.
– the standard deviation of returns from the option strategy and the standard
deviation of returns from the long equity strategy.
• For each option strategy, average the differentials across all ten
stocks.
• The results cluster in four groups, one for each of the strategies.
– The covered combination tends to have the highest mean and standard
deviation.
– The collar tends to have the lowest mean and standard deviation.
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
12
Exhibit 1
Average Differential Means and Standard Deviations of Returns Between Four Option Strategies
and the Corresponding Long Equity Strategy for Ten Stocks
(January, 2003 – August, 2013)
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
13
Performance Measures
• Exhibit 1 provides information regarding risk and return, but it does
not explicitly address the issue of whether the risk-return tradeoff
is favorable.
• To investigate performance, we utilize four popular measures:
–
–
–
–
Sharpe ratio
Jensen’s alpha
Treynor ratio
Sortino ratio
• We compute these four measures for each of the five strategies and
ten stocks.
– For each stock we rank the five strategies from 1 to 5 in terms of
decreasing performance.
– We then average the rankings across all ten stocks. Exhibit 2 presents the
results.
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
14
Exhibit 2
Average Rankings of the Relative Performance of Four Option Strategies
and the Corresponding Long Equity Strategy for Ten Stocks
(January, 2003 – August, 2013)
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
15
Empirical Results
• The results of Exhibit 2 are striking in their consistency.
• The same general theme holds for the entire period 2003-2013
and both sub-periods 2003-2007 and 2008-2013.
• From best to worst, the strategies rank as follows:
–
–
–
–
–
Covered combination
Covered call
Long equity
Collar
Protective put
• Although the consistency of these results is persuasive, there
are complicating issues (which we shall address) that prevent
these results from being as compelling as one might believe at
first glance.
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
16
How Do Cumulative Returns Compare
Across These Five Strategies?
• Does one get the same relative rankings using cumulative returns as
one gets with the aforementioned performance measures?
• Exhibit 3 displays the cumulative return results for all five strategies
and ten stocks over 2003-2013.
• The covered combination dominates by ranking first for all stocks
except Google.
• The covered call beats long equity for all stocks except Apple,
Amazon, and Google.
• Overall, the average rankings presented in Exhibit 3 agree with the
average rankings summarized in Exhibit 2.
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
17
Exhibit 3
Rankings of Performance of Cumulative Returns from Five Strategies for Ten Stocks
(January, 2003 – August, 2013)
TICKER
AAPL
AMZN
BRK
CMCSA
GOOG
KO
MSFT
ORCL
WFC
XOM
OBS
128
128
50
128
108
128
128
128
128
128
Average:
Strategy Ranking, by Ticker
Covered Protective
Call
Put
Collar
Combo
3
4
5
1
3
4
5
1
2
5
4
1
2
5
4
1
4
3
5
2
2
5
4
1
2
5
4
1
2
5
4
1
2
4
5
1
2
5
4
1
2.4
4.5
4.4
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
1.1
Long
Stock
2
2
3
3
1
3
3
3
3
3
2.6
18
Cumulative Return on a $1 Investment in AAPL
Note the Y-Axis
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
19
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
20
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
21
First Complicating Issue:
Transaction Costs
• Consider the covered combination strategy.
• It involves two options positions, and those option positions
are rolled over monthly.
• Including transaction costs (i.e., bid-ask spreads) will reduce,
or possibly even eliminate, the apparent advantage that the
covered combination has over the (already held) long equity
position.
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
22
Second Complicating Issue:
Non-Normality of Returns
•
Leland (1999) explores this issue in detail, using a simple example to examine both
the covered call and protective put strategies.
•
Leland demonstrates that using CAPM-based performance measures where there are
nonsymmetrical returns will:
–
–
Overrate “rebalancing” strategies (which reduce or “sell” skewness), i.e., covered call
Underrate “momentum” strategies (which increase or “buy” skewness), i.e., protective put.
• His demonstration foreshadows our findings exactly! That is, he predicts:
– The covered call strategy will outperform long equity.
– The protective put strategy will underperform long equity.
•
Note: Leland’s (1999) examples consider only strategies buying or selling options on
the market, but similar results are likely for options on individual securities.
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
23
Second Complicating Issue:
Non-Normality of Returns, Cont.
• Leland warns: “Any investment manager can ‘game’ the CAPM
performance measurement by selling options or rebalancing.”
• Leland proposes a performance measure that is robust to
non-normality of returns.
• Our preliminary investigations suggest that using Leland’s
proposed measure does not change the relative rankings of
the strategies.
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
24
Third Complicating Issue:
Early Exercise
• Had returns been adjusted for early exercise:
– Strategies involving (only) selling options, like the covered combination and
covered call, would obtain lower returns.
– Strategies involving (only) buying options, like the protective put, would obtain
higher returns.
• Consequently:
– The covered combination and covered call strategies might perform worse than
they currently appear to perform (based on Exhibit 2)
– The protective put strategy might perform better than it currently appears to
perform (based on Exhibit 2)
– The collar strategy might perform better or worse given that it involves both buying
and selling options.
• NOTE: Portfolio managers can, if they wish, avoid early exercise by:
– Using stocks that do not pay dividends (short call strategies)
– Using European-style FLEX options (short option strategies).
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
25
Early Exercise
• The OCC and Professor Robert Whaley gave us access to OCC
data on early exercise used in Barraclough and Whaley (2012)
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
26
“Costs of Competition Send Amazon Profit Down 58%”
http://www.nytimes.com/2006/07/26/technology/26amazon.html?_r=0
Date
7/3/2006
7/5/2006
7/6/2006
7/7/2006
AMZN
Implied
Closing
Put Price
Option
Option
Volatility Delta
Ask
Option
Lower
Option
Open Contracts
of the
of the Strike, Price,
Bid-Ask Bound:
Volume Interest Exercised Option Option
X
S_Ask Midpoint X - S_ask
146
1,562
494
1,007
6,700
6,855
8,049
8,201
0
0
0
0
0.4615
0.4573
0.4511
0.4572
-0.2366
-0.3148
-0.3318
-0.3781
-0.5115
-0.5629
0.6938
-0.9575
8/1/2006
8/2/2006
8/3/2006
8/4/2006
8/7/2006
8/8/2006
8/9/2006
8/10/2006
8/11/2006
8/14/2006
8/15/2006
8/16/2006
8/17/2006
8/18/2006
8/19/2006
39,855
6,004
3
12
0
6,968
5,808
0
2
10
401
0
60
25
6,622
3,653
3,493
3,496
3,487
3,483
2,880
2,884
2,884
2,883
2,876
2,565
2,576
2,581
40,860
6,164
0
7
4
7,571
5,804
0
1
7
299
1
0
0
2,581
0.8118
-0.9443
4.52
3.39
3.14
2.70
… …
0.6095
0.6323
-3.57
-2.09
-1.79
-1.10
… …
0
0
62,720
45,722
8,150
1,387
0.95
1.30
1.35
1.60
… …
… …
15,073
17,027
20,482
11,693
6,914
6,759
38.57
37.09
36.79
36.10
… …
… …
4,499
6,877
65,554
43,326
8,039
53
… …
… …
… …
… …
… …
7/24/2006
7/25/2006
7/26/2006
7/27/2006
7/28/2006
7/31/2006
35
35
35
35
BidAsk
Midpoint
Minus
Lower
Bound
35
35
35
35
35
35
34.31
33.58
26.18
26.53
27.17
26.86
2.50
2.93
8.75
8.45
7.80
8.15
0.69
1.42
8.82
8.47
7.83
8.14
1.81
1.51
-0.07
-0.02
-0.03
0.01
35
35
35
35
35
35
35
35
35
35
35
35
35
35
26.30
26.09
26.64
27.26
26.79
26.35
26.22
26.47
26.02
26.54
27.75
27.92
29.16
29.11
8.75
8.90
8.35
7.70
8.20
8.60
8.80
8.55
9.00
8.40
7.25
7.10
5.90
5.90
8.70
8.91
8.36
7.74
8.21
8.65
8.78
8.53
8.98
8.46
7.25
7.08
5.84
5.89
0.05
-0.01
-0.01
-0.04
-0.01
-0.05
0.02
0.02
0.02
-0.06
0.00
0.02
0.06
0.01
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
27
Summary of Our Results to this Point
• Long equity generally outperforms the collar and protective put.
• The covered call generally outperforms long equity.
• The covered combination generally outperforms long equity.
‒
‒
Premature to conclude covered combination outperforms long equity in practice.
The apparently superior performance could reflect the absence of:
‒
‒
‒
An early exercise adjustment
transaction costs
There is also the Leland (1999) critique involving nonsymmetrical returns.
• Further analysis, however, should enable us to provide stronger
evidence on the relative rankings of these options-based
strategies versus long equity.
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
28
Where Do We Go From Here?
• We could:
– Increase the number of stocks and lengthen the time period.
– Estimate and incorporate transaction costs.
– Estimate Leland’s (1999) performance measure
• OR alternatives such as Modigliani and Modigliani (1997), Stutzer (2000)
• OR the “manipulation-proof” measure of Ingersoll et al. (2007).
– Adjust returns for early exercise.
• The biggest concern for us all remains:
What is the best way to measure the
performance of options-based strategies?!
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
29
Questions and Comments?
Copyright © 2015 Michael L. Hemler and
Thomas W. Miller, Jr.
30