ssrn-id1481415
TRANSCRIPT
-
8/10/2019 SSRN-id1481415
1/23Electronic copy available at: http://ssrn.com/abstract=1481415
An Endogenous Volatility Approach
to Pricing and Hedging Call Options with
Transaction Costs
Leonard C. MacLean Yonggan Zhao William T. Ziemba
September 28, 2009
JEL Classication: B23 C15 C61 G13
Leonard C. MacLean: School of Business Administration, Dalhousie University, Halifax, NS,
Canada B3H 3J5. Email: [email protected] Yonggan Zhao: RBC Center for Risk Management and School of Business Administration, Dal-
housie University, Halifax, NS, Canada B3H 3J5. E-mail: [email protected]. William T. Ziemba: Sauder School of Business, University of British Columbia, Vancouver, BC,
Canada, V6T 1Z2 (Emeritus), and Visiting Professor, Mathematical Institute, Oxford University, 24-
29, St. Giles Street, Oxford, OX1 3LB, UK. E-mail: [email protected].
-
8/10/2019 SSRN-id1481415
2/23Electronic copy available at: http://ssrn.com/abstract=1481415
An Endogenous Volatility Approach
to Pricing and Hedging Call Options with TransactionCosts
Abstract
Standard delta hedging fails to exactly replicate a European call option in the
presence of transaction costs. We study a pricing and hedging model similar to the
delta hedging strategy with an endogenous volatility parameter for the calcula-
tion of delta over time. The endogenous volatility depends on both the transaction
costs and the option strike prices. The optimal hedging volatility is calculated
using the criterion of minimizing the weighted upside and downside replication
errors. The endogenous volatility model with equal weights on the up and down
replication errors yields an option premium close to the Leland (1985) heuristic
approach. The model with weights being the probabilities of the options money-
ness provides option prices closest to the actual prices. Option prices from the
model are identical to the Black-Scholes option prices when transaction costs are
zero. Data on S&P 500 index cash options from January to June 2008 illustrate
the model.
2
-
8/10/2019 SSRN-id1481415
3/23
1. Introduction
In the theory of option pricing, the goal is to create a replicating portfolio whose pay-
off equals the value of the option at some exercise date. In the Black-Scholes-Merton
(BSM) framework, delta hedging achieves that goal. The celebrated BSM formula
for pricing a European call option is based on two assumptions: (i) the underlying
stock price follows geometric Brownian motion; and (ii) trading has neither restric-
tions nor costs. Assumption (ii) does not hold and the theory fails, so the replicating
strategy has hedging error. One symptom of this failure is implied volatility skew-
ness. It is observed that equity option prices are consistently higher (lower) than the
Black-Scholes-Merton model prices for in- (out-of-) the-money call options, so implied
volatilities of in- (out-of-) the-money options are higher (lower) than at-the-money
ones.
Several researchers have considered the shortcomings of the BSM pricing formula.
An alternative to the geometric Brownian motion approach is the binomial pricing
model (Cox, Ross, and Rubinstein 1979.) Boyle and Vorst (1992) designed a perfect
hedging strategy in the binomial model with transaction costs. The perfect hedge is
possible due to the assumption of a binomial process for the underlying stock price.
They also developed a similar risk neutral valuation approach that is a two state
Markov process. Edirisinghe, et al. (1993) developed a general replicating strategy,
in the framework of optimization, by minimizing the initial cost subject to the hedging
portfolio payoff being at least as large as the options payoff. They indicated that it is
not necessarily optimal to revise the portfolio at each rebalance time.
Accepting the geometric Brownian motion assumption, but introducing transac-
tion costs on trading , Leland (1985) proposed a strategy in which the volatility for
the calculation of delta is increased by a term which is related to the transaction cost
rate and the length of the rebalance interval. As the option premium is positively
related to the volatility, the intention in augmenting the volatility was to increase
the option premium with trading frequency for a given proportional transaction cost
rate. However, as pointed out by Kabanov and Safarian (1997), this strategy does not
3
-
8/10/2019 SSRN-id1481415
4/23
guarantee a perfect replication of a call option in the limit (continuous trading). The
mathematical error in the Leland claim of perfect replication is discussed by Zhao
and Ziemba (2007a and 2007b) and Leland (2007).
A neglected aspect of the Leland approach is that the strike price of an option
does not appear in the expression for the augmented volatility, even though the im-plied volatilities are not constant across all levels of strike prices. This suggests that
relating the augmented volatility to the options strike price can adjust for skewness.
In this paper, we develop an option pricing and hedging model based on portfolio
replication techniques. The objective is to modify the Leland approach by linking the
hedging volatility to both transaction costs and the strike prices. Instead of using
an exogenous augmented volatility, we endogenize the volatility in an optimization
problem. The hedging volatility is selected to minimize the weighted mean absolutereplication error. The total replication error is divided into two quantities, the down-
side shortfall and the upside gain, which are weighted by the probabilities that the
option is in or out of the money, respectively. As the probability of the moneyness of
the option increases, the downside shortfall is considered to be more signicant. Simi-
larly, as the probability of the moneyness increases, the upside gain is considered less
signicant. With the endogenized volatility, it is possible to examine whether and
how transaction costs explain the skewness of equity options across a range of strikeprices. Since the probability that the call option is in the money decreases with the
level of the strike price, the weighing of the downside shortfall and the upside gain
generates skewness of the hedging volatilities across different strike prices.
The endogenous volatility approach is compared with the Leland and the Black-
Scholes approaches in numerical experiments. Using S&P 500 cash options, simula-
tion results show that option prices for the endogenous volatility model are closer to
the actual prices than prices from other models, especially for deep in-the-money and
deep out-of-the-money options. This result is consistent with the volatility skewness.
Option prices from the endogenous volatility model are identical to the Black-Scholes
option prices if transaction costs are ignored.
4
-
8/10/2019 SSRN-id1481415
5/23
2. A Pricing and Hedging Model
The Black-Scholes-Merton theory for options pricing has important restrictions in
terms of the dynamics of the underlying stock price and the frictionless trading mar-
ket. Lelands modication of the theory to incorporate transactions costs also has
limitations. In this section a new approach to option pricing with transactions costs
is developed. As in Lelands model, transactions costs affect the hedging volatility,
and the geometric Brownian motion and trading frequency assumptions are the same
as those in the BSM model. An optimization problem linking volatility to strike price
is dened and the corresponding option prices and hedging errors are related to the
BSM and Leland approaches .
2.1. Volatilities and the Hedging Strategies
Standard option pricing theory suggests that the implied volatilities for all option con-
tracts with different strikes should be the same in an orderly market. However, evi-
dence from the equity option market has shown that implied volatilities are heavily
skewed. Market prices of options are usually higher (lower) than the Black-Scholes-
Merton prices for in (out of) the money call options. This market imperfection has
been documented as volatility skewness in Rubinstein (1994).One reason for this imperfection is the presence of transaction costs. Leland (1985)
considered the idea that the increment of the option premium induced by an aug-
mented volatility can offset the necessary transaction costs. However, the hedging
portfolio does not replicate the option payoff. A possible explanation for the price
deviation with Lelands approach is that the augmented volatility does not depend
on the options strike prices. In this section we develop a method which relates the
hedging volatilities to both transaction costs and strike prices. We use the following notation:
5
-
8/10/2019 SSRN-id1481415
6/23
q round trip transaction cost.
K option strike price.
T option exercise date.
r risk free rate.
N () cumulative normal distribution function. instantaneous stock volatility.
hedging volatility for the delta hedging strategy.
To set up an alternative adjusted volatility model within the BSM setting, we
assume the dynamics of the underlying security follow geometric Brownian motion.
Let S t be the security price at time t. Then the price dynamics are
dS S
= r dt + dZ, (1)
where Z is a standard Brownian motion. This setting indicates that we are work-
ing under a risk neutral probability measure. We only consider hedging that takes
place at a nite number of time points. Two popular strategies are: (i) the delta
hedging strategy with xed time rebalancing, and (ii) the delta hedging strategy with
times determined by the price movement in the underlying security. We consider
xed times, so that comparisons with the BSM and the Leland models are based onthe alternative approaches to the hedging volatility.
Let i , i = 1 , ...n , be the time epochs when hedging trades take place for a given
horizon with 0 = 0 and n +1 = T . While the option writer holds N ( d i ) shares at
trading epoch i , where d i = ln
S iK +( r +
12
2 )( T i ) T i
, the amount in the trading account
is, for i = 1 , . . . ,n ,
B 0 = C 0
(1 + q/ 2) S 0
B i +1 = B i er i (1 + sgn (N ( d i +1 ) N ( d i )) q/ 2) S i +1 (N ( d i +1 ) N ( d i )) ,where C 0 is the initial price of the call option given by the BSM formula
C 0 = S 0 N ( d 0 ) Ke rT N ( d 0 T ) .
6
-
8/10/2019 SSRN-id1481415
7/23
Hence, the terminal payoff of the hedging portfolio at the options expiration date is
P T = B n e r ( T n ) + (1 q/ 2) S T N ( d n ) . (2)The focus is on the volatility estimate . In BSM the estimate is simply the stan-
dard deviation of stock returns and is independent of transaction costs and strike
prices. The hedging volatility proposed by Leland (1985) is
= 1 + 2 q twhich depends on the transaction cost q but not on the strike prices. In our endoge-
nous approach, the hedging volatility will depend on both the transactions cost and
strike price.
2.2. Hedging Error and Endogenous Volatility
The payoff on the call option with strike price K is C T = ( S T K )+ at the expira-tion date T . For a trajectory of stock prices, the delta hedging strategy has replication
error, with absolute value |P T C T |. The mean absolute error is E [|P T C T |]. Theupside and downside replication errors are E [(P T C T )+ ] and E [(P T C T ) ], re-spectively. The rationale for separating the error into scenarios with gains and lossesis the different perceptions or utility that an investor would assign to such scenarios.
Investors may view the upside replication gains differently from the downside repli-
cation losses. Empirical evidence shows that the implied volatilities for equity options
are heavily skewed, which suggests that risk perception is not symmetric. It is pro-
posed that risk aversion depends on the moneyness of the option. If the probability
that the option expires in the money is large, investors may be concerned about the
downside replication losses, considering the issued option as a liability. Similarly, if the probability that the option expires out of the money is small, the replication gains
will not be signicant to the writer of the option. An investor would like to weigh the
upside gains and downside losses according to the options moneyness.
7
-
8/10/2019 SSRN-id1481415
8/23
Let K be the probability that the option with strike price K ends valueless at the
expiry date, i.e.,
K = Pr[ S T K ],which implies that
K = N (ln( S 0 /K )+( r 12
2
) T T ).
With the xed parameters = {S 0 ,K,r,T,K,,q, 1 , ... n }, consider the weightedmean errors criterion for determining volatility
( , ) = K ( ) E (P T C T )+ | , + (1 K ) E (P T C T ) | , . (3)
Equation ( 3) can be viewed as the expected performance, in terms of hedging error,
using the probability measure induced by the moneyness of the option. Since the
probability that the option is out of the money at expiration increases with the level
of the strike prices, the interest in hedging the downside error should decrease as the
option strike price increases. Correspondingly, the investor prefers to charge a higher
premium for deep in the money options and lower premium for deep out-of-the-money
options than the BSM options prices.
The performance measure depends on several parameters. The main attention
is on the hedging volatility parameter . So, holding other parameters constant, the
objective is to determine the adjusted volatility with the best performance. The writer
of the call option will solve the nonlinear optimization model
min> 0
( , ) . (4)
The solution to ( 4) depends on the parameter vector and particularly the transaction
costs and the strike prices. There are some important properties of the endogenous
volatility problem as an alternative to the BSM or Leland hedging approaches.
The BSM option price is a special instance of the solution to the optimization
model ( 4) if transaction costs are zero. Moreover, as the trading interval ap-
proaches zero, which implies a continuous trading in the limit, both the upside
replication gains and the downside replication losses must tend to zero, since
8
-
8/10/2019 SSRN-id1481415
9/23
an exact hedge can be achieved in the framework of continuous trading without
transaction costs. Thus, the optimization model ( 4) is consistent with the BSM
model.
If the weights reecting risk aversion are selected as K ( ) = 0 .5 (indifference
between gains and losses), so that the average absolute replication error is min-
imized, the optimal volatility is close to Lelands formula, which approximately
minimizes mean squared errors.
The weights from the moneyness probability will introduce a correction for the
bias of hedging prices. The weights have an effect of tilting the prices, aligning
them closer to true prices.
Since there is only one variable in the nonlinear optimization model ( 4), an efcientalgorithm can be applied to the solution for the optimal endogenous volatility. A
Matlab code was written to solve this problem.
3. Empirical Tests
The objective in proposing a new method for delta hedging is to address some of
the problems with the BSM and Leland approaches. The outcome measures are: (i)
volatility skewness; (ii) pricing bias; and (iii) hedging error. The endogenous volatility
approach is expected to improve performance on those dimensions over the standard
methods. Evidence of improved performance is provided in this section.
The comparator option pricing and hedging models have the same assumptions
for the stock dynamics - geometric Brownian motion, and option type - European call
option. Data on S&P 500 index options are used for the model input to compare the
simulation results with the actual option prices. The instantaneous stock volatilityis estimated as the sample standard deviation, based on the maximum likelihood
method. The risk free rate is set to be the average of the US six month term deposit
rates for that period. The index level was S 0 = 892 .5 on December 18, 2008. Esti-
mated as the sample standard deviation, the annualized volatility of the index return
9
-
8/10/2019 SSRN-id1481415
10/23
is 34.61%. The range of the strike prices used in this study is from 750 to 1100 in
25 point increments. The round-trip proportional transaction costs used to study how
the option prices are affected by the level of transaction cost rate are 0 (no cost), 0.3% ,
0.6% , 0.9% , 1.2% , and 1.5% . The options expired on June 18, 2009, which indicates
a total of 130 days for the hedging to take place. For the signicance of this study,the number of scenarios is set to 10000. We simulate 10000 paths of the underlying
asset prices using the estimated parameters from the data. We then apply a process
control strategy to implementing portfolio hedging over time, taking transaction costs
into consideration. The key innovation in the approach is the calculation of the ad-
justed volatility. Instead of exogenously specifying an adjusted volatility for the BSM
formula, we endogenize the hedging volatility for the calculation of delta hedging.
The optimal hedging volatility is determined by minimizing the weighted upside anddownside hedging errors at the expiry date of the call option.
For the rest of this section, we study model performance based on the different
criteria. For comparison with the BSM and the Leland models, we consider the en-
dogenized volatility model with two weighting schemes for the upside and downside
hedging errors, (i) EV1: equal weights, (ii) EV2: moneyness probability weights.
3.1. Volatility SkewnessIt is observed that equity option prices are consistently higher (lower) than the BSM
model prices for in (out-of) -the-money options, so implied volatilities of in (out-of) -
the-money call options are higher (lower) than at-the-money ones. A motivating prin-
ciple for endogenizing volatility is to link the hedging volatility not only to transaction
costs but also to strike prices. The hedging volatilities for the various models are given
in Table I and plotted in Figure 1. The BSM volatility, which was estimated as 34.61%
per annum, does not vary by transaction cost or strike price, the Leland volatilities
vary by transaction cost, and the endogenous volatilities vary by both transaction cost
and strike price.
From Table I, the hedging volatility increases with transaction costs for all models.
10
-
8/10/2019 SSRN-id1481415
11/23
Table I: Hedging Volatilities for Various Models
Model StrikeCost
0% 0.3% 0.6% 0.9% 1.2% 1.5%
Leland 0.3461 0.3649 0.3828 0.3999 0.4163 0.4321
EV1
725 0.3451 0.3761 0.4046 0.4307 0.4559 0.4799
750 0.3455 0.3746 0.4018 0.4271 0.4515 0.4748
775 0.3454 0.3732 0.3993 0.4241 0.4475 0.4701
800 0.3452 0.3724 0.3977 0.4214 0.4442 0.4660
825 0.3452 0.3715 0.3962 0.4193 0.4415 0.4629
850 0.3453 0.3709 0.3947 0.4175 0.4395 0.4604
875 0.3453 0.3701 0.3935 0.4159 0.4372 0.4578
900 0.3453 0.3696 0.3925 0.4142 0.4352 0.4555
925 0.3454 0.3689 0.3914 0.4131 0.4337 0.4537
950 0.3450 0.3682 0.3907 0.4120 0.4325 0.4522975 0.3447 0.3679 0.3901 0.4110 0.4313 0.4506
1000 0.3450 0.3676 0.3892 0.4101 0.4303 0.4497
1025 0.3448 0.3672 0.3888 0.4094 0.4293 0.4484
1050 0.3445 0.3667 0.3881 0.4084 0.4281 0.4472
1075 0.3440 0.3663 0.3876 0.4078 0.4273 0.4461
1100 0.3441 0.3660 0.3871 0.4070 0.4263 0.4449
EV2
725 0.3629 0.3949 0.4257 0.4552 0.4834 0.5107
750 0.3595 0.3894 0.4180 0.4452 0.4712 0.4966
775 0.3559 0.3844 0.4112 0.4369 0.4614 0.4852
800 0.3526 0.3795 0.4053 0.4298 0.4533 0.4760
825 0.3495 0.3756 0.4002 0.4237 0.4463 0.4683
850 0.3465 0.3721 0.3960 0.4188 0.4406 0.4616
875 0.3436 0.3685 0.3917 0.4139 0.4349 0.4554
900 0.3408 0.3648 0.3876 0.4097 0.4303 0.4502
925 0.3379 0.3617 0.3841 0.4057 0.4258 0.4454
950 0.3350 0.3583 0.3805 0.4015 0.4213 0.4404
975 0.3319 0.3549 0.3767 0.3972 0.4168 0.43591000 0.3293 0.3520 0.3734 0.3938 0.4133 0.4320
1025 0.3265 0.3491 0.3701 0.3900 0.4092 0.4276
1050 0.3236 0.3460 0.3668 0.3867 0.4056 0.4238
1075 0.3214 0.3435 0.3643 0.3838 0.4027 0.4206
1100 0.3182 0.3404 0.3610 0.3806 0.3992 0.4170
11
-
8/10/2019 SSRN-id1481415
12/23
00.005
0.010.015
500
1000
1500
1
0
1
2
H e
d g
i n g
V o
l a t i l i t y
BSM Model
Transaction CostsStrike Prices 00.005
0.010.015
500
1000
1500
0.35
0.4
0.45
0.5
Transaction Costs
Leland Model
Strike Prices
H e
d g
i n g
V o
l a t i l i t y
00.005
0.010.015
500
1000
1500
0.35
0.4
0.45
0.5
Transaction Costs
EV1 Model
Strike Prices
H e
d g
i n g
V o
l a t i l i t y
00.005
0.010.015
500
1000
15000.2
0.4
0.6
Transaction Costs
EV2 Model
Strike Prices
H e
d g
i n g
V o
l a t i l i t y
Figure 1. Hedging Volatilities for Alternative Models
The Leland hedging volatility does not depend on the strike price. As the strike price
increases, the volatility decreases for the endogenous models. The changing volatility
with changing strikes is greater for the EV2 model than the EV1 model. Also, the
volatility skewness is more pronounced for all levels of transaction cost with the EV2
model than with the EV1. This suggests that the option prices with the endogenous
method EV2 should be closer to actual prices.
3.2. Predicted Option Prices
In Table II , the actual option prices at various strike levels are compared to the model
prices at strike levels and transaction cost levels. For each model the prices which areclosest to the actual price are in bold.
There are several observations we can make based on the presented prices in Table
II and Figure 2. Within each model, the best price for in-the-money options is at the
high transaction cost level. In fact, as the strike price increases, the transaction cost
12
-
8/10/2019 SSRN-id1481415
13/23
Table II: Predicted Option Prices by Model
Model Strike Cost Actual 0% 0.3% 0.6% 0.9% 1.2% 1.5%
Leland
725 212.10 183.28 186.34 189.34 192.28 195.15 197.96750 193.95 164.92 168.36 171.69 174.93 178.08 181.15775 176.55 147.72 151.50 155.14 158.65 162.05 165.34
800 159.90 131.72 135.79 139.69 143.44 147.05 150.54825 144.00 116.94 121.26 125.37 129.31 133.10 136.75850 128.75 103.38 107.88 112.16 116.25 120.17 123.94875 114.65 91.01 95.64 100.03 104.22 108.24 112.10900 100.90 79.81 84.50 88.95 93.19 97.26 101.18925 88.55 69.72 74.41 78.87 83.13 87.22 91.15950 76.85 60.69 65.33 69.75 73.98 78.05 81.97975 66.20 52.64 57.18 61.52 65.69 69.71 73.58
1000 56.50 45.51 49.91 54.14 58.21 62.14 65.951025 47.75 39.22 43.45 47.53 51.48 55.31 59.021050 40.00 33.70 37.73 41.64 45.44 49.14 52.741075 33.20 28.88 32.68 36.40 40.04 43.60 47.071100 27.20 24.68 28.25 31.77 35.23 38.62 41.96
EV1
725 212.10 183.12 188.22 193.10 197.72 202.29 206.72750 193.95 164.82 170.16 175.29 180.19 184.96 189.60775 176.55 147.584 153.188 158.529 163.669 168.58 173.35800 159.90 131.53 137.42 142.97 148.18 153.23 158.08825 144.00 116.74 122.77 128.46 133.80 138.94 143.89850 128.75 103.21 109.31 115.01 120.45 125.72 130.71875 114.65 90.82 96.92 102.65 108.14 113.35 118.38900 100.90 79.63 85.66 91.36 96.74 101.94 106.99925 88.55 69.56 75.42 81.02 86.42 91.56 96.54950 76.85 60.42 66.15 71.70 76.99 82.06 86.96975 66.20 52.31 57.90 63.30 68.40 73.38 78.16
1000 56.50 45.25 50.56 55.67 60.65 65.52 70.241025 47.75 38.93 43.98 48.91 53.69 58.37 62.891050 40.00 33.36 38.12 42.81 47.36 51.83 56.231075 33.20 28.47 32.98 37.42 41.75 46.01 50.201100 27.20 24.31 28.47 32.63 36.70 40.73 44.72
EV2
725 212.10 186.00 191.41 196.82 202.16 207.37 212.54750 193.95 167.35 172.93 178.40 183.73 188.87 193.96775 176.55 149.69 155.46 160.99 166.36 171.51 176.57800 159.90 133.12 138.97 144.63 150.04 155.26 160.30825 144.00 117.73 123.71 129.37 134.80 140.04 145.14850 128.75 103.49 109.59 115.31 120.77 125.99 131.00875 114.65 90.39 96.52 102.20 107.64 112.79 117.81900 100.90 78.49 84.47 90.13 95.61 100.73 105.68925 88.55 67.70 73.61 79.19 84.57 89.59 94.47950 76.85 57.96 63.71 69.17 74.36 79.28 84.04975 66.20 49.24 54.76 60.04 65.03 69.84 74.52
1000 56.50 41.64 46.89 51.90 56.74 61.42 65.931025 47.75 34.91 39.90 44.63 49.18 53.65 57.971050 40.00 29.01 33.68 38.14 42.49 46.72 50.841075 33.20 24.08 28.36 32.55 36.61 40.64 44.541100 27.20 19.66 23.62 27.49 31.32 35.09 38.77
13
-
8/10/2019 SSRN-id1481415
14/23
00.005
0.010.015
500
1000
1500
0
100
200
Transaction Costs
BSM Model
Strike Prices
P r e
d i c t e d P r i c e s
00.005
0.010.015
500
1000
1500
0
100
200
Transaction Costs
Leland Model
Strike Prices
P r e
d i c t e d P r i c e s
00.005
0.010.015
500
1000
15000
100
200
300
Transaction Costs
EV1 Model
Strike Prices
P r e
d i c t e d P r i c e s
00.005
0.010.015
500
1000
15000
100
200
300
Transaction Costs
EV2 Model
Strike Prices
P r e
d i c t e d P r i c e s
Figure 2. Predicted Prices for Various Models
for the best matched price decreases. For almost all strike prices, the closest model
price is produced by the EV2 model for some transaction cost level.The options prices
from the EV2 model are very accurate for in-the-money options. For each of the deeply
in- or out-of-the-money options there exists a transaction cost level such that the EV2
model price is closest to the actual price, as bolded prices in the table indicate.
3.3. Hedging Error
Realized Error . The estimation error with the option price is a factor in the perfor-
mance of the hedging portfolio. The portfolio performance for the period of the data
based on option prices from the various models is provided in Table III , in which theerror is the difference between the portfolio value and the option payoff.
As expected the superiority of the EV2 model occurs for deeply in- or out-of-the-
money options for some transaction cost level, where the predicted option prices were
closer to actual option prices. In the presence of transaction costs, we propose that
14
-
8/10/2019 SSRN-id1481415
15/23
Table III: Realized Errors for the Various ModelsModel Strike Cost0% 0.3% 0.6% 0.9% 1.2% 1.5%
BSM
725 22.5076 16.6535 10.7993 4.9451 -0.9091 -6.76322750 23.4656 17.1585 10.8515 4.5444 -1.7626 -8.06976775 24.5768 17.8350 11.0932 4.3514 -2.3904 -9.13213800 25.8243 18.6799 11.5356 4.3912 -2.7532 -9.89758825 27.2804 19.7815 12.2826 4.7837 -2.7152 -10.21415850 29.2287 21.4312 13.6337 5.8362 -1.9614 -9.75894875 33.1912 25.1650 17.1387 9.1125 1.0862 -6.94007900 39.0564 31.2463 23.4363 15.6263 7.8163 0.00638925 42.5350 35.3584 28.1818 21.0053 13.8287 6.65216950 33.2905 26.9553 20.6202 14.2850 7.9498 1.61473975 26.9386 21.5411 16.1435 10.7460 5.3485 -0.04916
1000 20.9479 16.4414 11.9349 7.4284 2.9219 -1.584601025 15.3613 11.6266 7.8920 4.1574 0.4228 -3.311941050 10.7064 7.6393 4.5721 1.5050 -1.5621 -4.629271075 7.1018 4.5916 2.0813 -0.4289 -2.9392 -5.449481100 4.3057 2.2485 0.1913 -1.8659 -3.9231 -5.98030
Leland
725 22.5076 17.3185 12.1842 7.1064 2.0849 -2.8814750 23.4656 18.0064 12.6324 7.3402 2.1257 -3.0154775 24.5768 18.8555 13.2548 7.7647 2.3763 -2.9184800 25.8243 19.8748 14.0846 8.4359 2.9140 -2.4936825 27.2804 21.1794 15.2741 9.5382 3.9507 -1.5058850 29.2287 23.1092 17.2065 11.4878 5.9282 0.5081875 33.1912 27.1816 21.4014 15.8197 10.4125 5.1600900 39.0564 33.4957 28.1752 23.0584 18.1166 13.3272925 42.5350 37.5091 32.6753 28.0054 23.4773 19.0731950 33.2905 28.4193 23.7313 19.2008 14.8073 10.5340975 26.9386 22.2825 17.7660 13.3758 9.0997 4.9268
1000 20.9479 16.6096 12.3621 8.2014 4.1229 0.12141025 15.3613 11.2853 7.2711 3.3183 -0.5744 -4.40921050 10.7064 6.8227 2.9879 -0.7976 -4.5346 -8.22421075 7.1018 3.3705 -0.3231 -3.9764 -7.5888 -11.16071100 4.3057 0.7310 -2.8245 -6.3539 -9.8531 -13.3205
EV1
725 22.4737 17.6824 12.8881 8.1075 3.3890 -1.2812750 23.4427 18.4089 13.4228 8.4886 3.6329 -1.1493775 24.5422 19.2740 14.1002 9.0297 4.0470 -0.8364800 25.7722 20.3186 15.0034 9.8002 4.7411 -0.2089825 27.2218 21.6439 16.2623 11.0324 5.9642 1.0276850 29.1688 23.6207 18.2814 13.1426 8.1903 3.3498875 33.1130 27.7227 22.5619 17.6275 12.8521 8.2443900 38.9719 34.0349 29.3454 24.8440 20.5395 16.4045925 42.4621 37.9574 33.6717 29.5824 25.6122 21.7803950 33.2117 28.6778 24.3989 20.2977 16.3449 12.5262975 26.8911 22.4082 18.1300 14.0016 10.0332 6.1798
1000 20.9411 16.6487 12.5023 8.4926 4.6081 0.83231025 15.3874 11.2596 7.2508 3.3534 -0.4413 -4.14681050 10.7787 6.7591 2.8425 -0.9707 -4.6931 -8.33191075 7.2348 3.2901 -0.5613 -4.3119 -7.9801 -11.57061100 4.4629 0.6481 -3.1148 -6.7915 -10.4032 -13.9520
EV2
725 23.0339 18.2398 13.4942 8.7993 4.1593 -0.4118750 23.9992 18.9788 14.0360 9.1699 4.3763 -0.3236775 25.0496 19.8070 14.6680 9.6470 4.7232 -0.0901800 26.1889 20.7286 15.4459 10.3054 5.2987 0.4110825 27.5078 21.9278 16.5486 11.3554 6.3243 1.4482850 29.2652 23.7231 18.3945 13.2655 8.2966 3.4688875 32.9434 27.5547 22.3688 17.4011 12.5957 7.9738900 38.4696 33.4842 28.7553 24.2852 19.9256 15.7290925 41.6664 37.1496 32.8233 28.7018 24.6590 20.7609950 32.5135 27.9060 23.5343 19.3407 15.2839 11.3684975 26.4853 21.8731 17.4686 13.2248 9.1324 5.1825
1000 20.9214 16.4740 12.1872 8.0430 4.0253 0.11861025 15.8749 11.5550 7.3830 3.3314 -0.6139 -4.46191050 11.8874 7.6444 3.5492 -0.4359 -4.3187 -8.11231075 8.9112 4.7819 0.7577 -3.1533 -6.9899 -10.73381100 6.7489 2.7567 -1.1475 -4.9883 -8.7516 -12.4411
15
-
8/10/2019 SSRN-id1481415
16/23
investors view gains and losses from a hedging portfolio differently depending on the
moneyness of the option.
The performance of the hedging portfolio depends on the hedging volatility and
corresponding option price. The hedging volatility depends on the criteria for perfor-
mance of the hedging portfolio. The results by the alternative performance criteriaare considered separately.
Hedging Error by Mean Absolute Deviation . The values for the mean ab-
solute deviation criterion are illustrated in Figure 3 and Table IV. Those results by
strike price and transaction cost show that the EV1 is the best model for this criterion,
and also how far the other models are from the optimum.
00.005
0.010.015
500
1000
15000
10
20
Transaction Costs
BSM Model
Strike Prices
H e
d g
i n g
E r r o r
00.005
0.010.015
500
1000
15000
2
4
6
Transaction Costs
Leland Model
Strike Prices
H e
d g
i n g
E r r o r
00.005
0.010.015
500
1000
15001
2
3
4
Transaction Costs
EV1 Model
Strike Prices
H e
d g
i n g
E r r o r
00.005
0.010.015
500
1000
15002
3
4
5
Transaction Costs
EV2 Model
Strike Prices
H e
d g
i n g
E r r o r
Figure 3. Hedging Error by Mean-Absolute Deviation for Alternative Models
For the mean absolute deviation criterion the values for the Leland, EV1 and EV2
models are close, and very close in the mid-range of strikes. The errors increase in
transaction cost for each model. However, the pattern with strikes differs for the
models. Leland and EV1 have a hill pattern, whereas the EV2 has a valley shape. So
16
-
8/10/2019 SSRN-id1481415
17/23
Table IV: Hedging Errors by Mean-Absolute DeviationModel Strike Cost0% 0.3% 0.6% 0.9% 1.2% 1.5%
BSM
725 1.6407 2.8783 5.2932 7.9109 10.5519 13.1938750 1.8288 3.0946 5.6181 8.3779 11.1687 13.9605775 1.9962 3.2653 5.8775 8.7624 11.6789 14.5989800 2.1342 3.3941 6.0694 9.0506 12.0672 15.0854825 2.2536 3.5033 6.2089 9.2459 12.3277 15.4115850 2.3513 3.5721 6.2853 9.3471 12.4600 15.5777875 2.4198 3.6061 6.2789 9.3332 12.4482 15.5667900 2.4438 3.5846 6.2010 9.2172 12.2993 15.3873925 2.4463 3.5480 6.0840 9.0369 12.0576 15.0861950 2.4216 3.4604 5.9030 8.7616 11.6950 14.6363975 2.4000 3.3612 5.6764 8.4209 11.2447 14.0739
1000 2.3416 3.2474 5.4382 8.0507 10.7424 13.44021025 2.2814 3.1142 5.1566 7.6231 10.1681 12.72121050 2.1930 2.9503 4.8513 7.1602 9.5483 11.94541075 2.0688 2.7713 4.5351 6.6742 8.8971 11.13071100 1.9534 2.5938 4.2062 6.1805 8.2363 10.3026
Leland
725 1.6407 2.0093 2.7585 3.6462 4.5909 5.5668750 1.8288 2.1467 2.8142 3.6286 4.5098 5.4269775 1.9962 2.2600 2.8429 3.5828 4.3947 5.2487800 2.1342 2.3564 2.8672 3.5286 4.2714 5.0598825 2.2536 2.4435 2.8844 3.4728 4.1492 4.8748850 2.3513 2.5172 2.9008 3.4241 4.0320 4.6940875 2.4198 2.5676 2.8984 3.3609 3.9090 4.5103900 2.4438 2.5787 2.8702 3.2813 3.7692 4.3114925 2.4463 2.5693 2.8300 3.1995 3.6392 4.1294950 2.4216 2.5323 2.7636 3.0891 3.4809 3.9214975 2.4000 2.5028 2.7077 2.9935 3.3412 3.7377
1000 2.3416 2.4411 2.6266 2.8849 3.1949 3.55161025 2.2814 2.3788 2.5471 2.7790 3.0594 3.37821050 2.1930 2.2869 2.4430 2.6515 2.9062 3.19641075 2.0688 2.1616 2.3080 2.5000 2.7327 3.00021100 1.9534 2.0449 2.1834 2.3591 2.5706 2.8133
EV1
725 1.6394 1.8520 2.2112 2.6188 3.0467 3.4804750 1.8283 2.0033 2.3118 2.6736 3.0598 3.4582775 1.9953 2.1415 2.4051 2.7240 3.0723 3.4353800 2.1329 2.2585 2.4843 2.7629 3.0747 3.4063825 2.2522 2.3612 2.5574 2.8069 3.0880 3.3923850 2.3502 2.4485 2.6228 2.8454 3.0993 3.3742875 2.4186 2.5110 2.6727 2.8756 3.1097 3.3622900 2.4423 2.5351 2.6838 2.8716 3.0892 3.3254925 2.4453 2.5335 2.6694 2.8429 3.0417 3.2605950 2.4189 2.5084 2.6384 2.7993 2.9826 3.1847975 2.3971 2.4862 2.6100 2.7595 2.9308 3.1190
1000 2.3387 2.4269 2.5460 2.6872 2.8486 3.02541025 2.2783 2.3684 2.4842 2.6195 2.7711 2.93651050 2.1883 2.2817 2.3960 2.5268 2.6715 2.82711075 2.0627 2.1589 2.2736 2.4029 2.5442 2.69431100 1.9480 2.0432 2.1558 2.2828 2.4193 2.5634
EV2
725 2.0216 2.2873 2.7415 3.2578 3.7860 4.3291750 2.1075 2.3128 2.6716 3.0884 3.5213 3.9802775 2.1739 2.3407 2.6232 2.9690 3.3399 3.7325800 2.2249 2.3515 2.5901 2.8817 3.2057 3.5494825 2.2858 2.3929 2.5902 2.8419 3.1283 3.4393850 2.3529 2.4511 2.6257 2.8487 3.1021 3.3772875 2.4248 2.5168 2.6789 2.8839 3.1182 3.3711900 2.4854 2.5805 2.7323 2.9156 3.1372 3.3762925 2.5573 2.6456 2.7881 2.9573 3.1701 3.3953950 2.6184 2.7117 2.8471 3.0209 3.2269 3.4451975 2.7077 2.8076 2.9422 3.1186 3.3132 3.5133
1000 2.7635 2.8694 3.0132 3.1771 3.3591 3.55611025 2.8203 2.9219 3.0752 3.2584 3.4432 3.64471050 2.8468 2.9566 3.1151 3.2900 3.4834 3.68761075 2.7970 2.9290 3.0835 3.2730 3.4580 3.66681100 2.7927 2.9210 3.0869 3.2646 3.4609 3.6731
17
-
8/10/2019 SSRN-id1481415
18/23
the errors for Leland and EV1 decrease for the extreme strikes, whereas the errors
increase for EV2 in the extremes.
Hedging Error by Option Moneyness . With the mean absolute deviation,
gains and losses are perceived the same - equally weighted. The more realistic propo-
sition is that hedging error is perceived differently depending on the moneyness of theoption. Table V presents the weights by strike (moneyness) for the gains and losses.
When the option is deep in the money, the greater weight is on losses, while for deep
out of the money the greater weight is on gains.
Table V: Weighting probabilitiesStrike Pr : Gains 1 Pr : Losses
725 0.2426 0.7574750 0.2880 0.7120
775 0.3355 0.6645800 0.3840 0.6160825 0.4329 0.5671850 0.4814 0.5186875 0.5287 0.4713900 0.5743 0.4257925 0.6177 0.3823950 0.6586 0.3414975 0.6967 0.3033
1000 0.7319 0.26811025 0.7642 0.23581050 0.7935 0.20651075 0.8198 0.1802
1100 0.8435 0.1565
The values for the weighted mean absolute deviation criterion are illustrated in
Figure 4 and Table VI. Those results by strike price and transaction cost show that
the EV2 is the best model for the new criterion.
In Figure 4 the pattern with strikes is somewhat changed for the Leland and EV1
models, with increasing error for deep out of the money options. The pattern for
the EV2 model is ipped, so that the errors for deep in/out of the money options in
decreased. This is exactly the effect that underlies the hedging strategy.
18
-
8/10/2019 SSRN-id1481415
19/23
Table VI: Hedging Errors By Option MoneynessModel Strike Cost0% 0.3% 0.6% 0.9% 1.2% 1.5%
BSM
725 1.6326 4.2301 8.0050 11.9826 15.9835 19.9853750 1.8294 4.2786 7.9856 11.9288 15.9031 19.8784775 1.9959 4.2257 7.7988 11.6445 15.5219 19.4027800 2.1329 4.0928 7.4680 11.1491 14.8656 18.5837825 2.2526 3.9159 7.0351 10.4856 13.9810 17.4783850 2.3508 3.6877 6.5170 9.6948 12.9237 16.1575875 2.4214 3.4287 5.9224 8.7976 11.7335 14.6731900 2.4519 3.1337 5.2912 7.8484 10.4715 13.1006925 2.4602 2.8486 4.6715 6.9112 9.2186 11.5340950 2.4443 2.5499 4.0593 5.9848 7.9849 9.9930975 2.4300 2.2776 3.4793 5.1102 6.8205 8.5361
1000 2.3671 2.0209 2.9596 4.3201 5.7597 7.20541025 2.3072 1.7906 2.4836 3.6007 4.7963 6.00001050 2.2194 1.5692 2.0629 2.9643 3.9450 4.93461075 2.0970 1.3698 1.7039 2.4132 3.2065 4.01041100 1.9818 1.2009 1.3921 1.9453 2.5798 3.2250
Leland
725 1.6326 2.5594 3.8621 5.2996 6.7912 8.3115750 1.8294 2.5855 3.6881 4.9351 6.2468 7.5930775 1.9959 2.5823 3.4859 4.5449 5.6749 6.8462800 2.1329 2.5692 3.2933 4.1673 5.1222 6.1225825 2.2526 2.5584 3.1148 3.8188 4.6107 5.4520850 2.3508 2.5467 2.9603 3.5136 4.1517 4.8439875 2.4214 2.5265 2.8142 3.2335 3.7380 4.2956900 2.4519 2.4848 2.6733 2.9804 3.3635 3.8001925 2.4602 2.4351 2.5454 2.7626 3.0483 3.3829950 2.4443 2.3732 2.4190 2.5557 2.7558 3.0021975 2.4300 2.3290 2.3245 2.3962 2.5255 2.6999
1000 2.3671 2.2507 2.2129 2.2412 2.3157 2.43171025 2.3072 2.1840 2.1223 2.1159 2.1510 2.21811050 2.2194 2.0942 2.0199 1.9883 1.9943 2.02831075 2.0970 1.9774 1.8985 1.8538 1.8401 1.85221100 1.9818 1.8709 1.7928 1.7391 1.7102 1.7025
EV1
725 1.6736 1.9122 2.3184 2.8114 3.3067 3.8198750 1.8500 2.0516 2.3935 2.8069 3.2405 3.6882775 2.0178 2.1790 2.4655 2.8096 3.1947 3.5939800 2.1544 2.2781 2.5127 2.8166 3.1496 3.5085825 2.2647 2.3718 2.5711 2.8316 3.1237 3.4414850 2.3531 2.4508 2.6267 2.8513 3.1065 3.3849875 2.4146 2.5075 2.6672 2.8683 3.0984 3.3466900 2.4361 2.5295 2.6741 2.8518 3.0607 3.2895925 2.4390 2.5201 2.6483 2.8177 3.0030 3.2099950 2.3980 2.4813 2.6150 2.7709 2.9418 3.1290975 2.3617 2.4551 2.5872 2.7220 2.8873 3.0553
1000 2.3029 2.3875 2.4970 2.6344 2.7968 2.97071025 2.2264 2.3129 2.4317 2.5592 2.7118 2.86031050 2.1136 2.2030 2.3195 2.4415 2.5770 2.73071075 1.9612 2.0670 2.1901 2.3099 2.4407 2.57951100 1.8511 1.9435 2.0596 2.1683 2.2869 2.4172
EV2
725 1.3100 1.5115 1.8609 2.2589 2.6677 3.0744750 1.5897 1.7603 2.0685 2.4319 2.8189 3.2134775 1.8482 1.9936 2.2607 2.5856 2.9429 3.3143800 2.0611 2.1857 2.4168 2.7047 3.0261 3.3695825 2.2320 2.3385 2.5390 2.7942 3.0842 3.3958850 2.3505 2.4479 2.6236 2.8482 3.1039 3.3818875 2.4085 2.5015 2.6602 2.8613 3.0896 3.3375900 2.3950 2.4848 2.6275 2.8067 3.0113 3.2334925 2.3325 2.4153 2.5423 2.7006 2.8820 3.0784950 2.2085 2.2902 2.4078 2.5516 2.7127 2.8850975 2.0688 2.1484 2.2560 2.3835 2.5250 2.6760
1000 1.8934 1.9685 2.0651 2.1767 2.3006 2.43171025 1.7101 1.7788 1.8656 1.9658 2.0747 2.19001050 1.5034 1.5692 1.6490 1.7382 1.8334 1.93441075 1.3007 1.3625 1.4336 1.5125 1.5969 1.68461100 1.1130 1.1687 1.2326 1.3025 1.3761 1.4523
19
-
8/10/2019 SSRN-id1481415
20/23
00.005
0.010.015
500
1000
1500
0
10
20
Transaction Costs
BSM Model
Strike Prices
H e
d g
i n g
E r r o r
00.005
0.010.015
500
1000
1500
0
5
10
Transaction Costs
Leland Model
Strike Prices
H e
d g
i n g
E r r o r
00.005
0.010.015
500
1000
15001
2
3
4
Transaction Costs
EV1 Model
Strike Prices
H e
d g
i n g
E r r o r
00.005
0.010.015
500
1000
15001
2
3
4
Transaction Costs
EV2 Model
Strike Prices
H e
d g
i n g
E r r o r
Figure 4. Hedging Error by Option Moneyness for Alternative Models
4. Conclusion
In the last thirty years, nancial derivatives have grown from a marginal activity
to occupy center-stage position in nancial economic theory and nancial practice.
At the same time, mathematical nance has grown to be one of the main branches
of applied mathematics. The single largest credit for these remarkable developments
are due to Fisher Black, Myron Scholes, and Robert Merton, whos classic 1973 papers
gave a theory of how to price options. Without this prescription, option pricing would
have remained more of an art than a science, and trading in options would have been
less liquid and less important, as traders would have had a less rm idea on how
to fairly value and hedge the options. However, this great achievement rests on theassumptions of no arbitrage, lognormality for spot price dynamics, and frictionless
trading. In reality, even though the condition of arbitrage free and the assumption of
lognormality are arguably to be satisfactory most of the time, transaction costs always
exist. As well the evidence suggests that the option price depends on the moneyness
20
-
8/10/2019 SSRN-id1481415
21/23
of the option, that is the strike price.
In this paper a pricing method is introduced which links the volatility to both
the transaction cost and the strike price. The Black-Scholes-Merton formula is used,
with the endogenous volatility, to price options. The results are clear. The endoge-
nous volatility approach produces more accurate prices. The optimization problemfor calculating the endogenous volatilities is simple to implement so that using the
approach is practical.
21
-
8/10/2019 SSRN-id1481415
22/23
References
A HN , H., M. D AYAL , E. G RANNAN , and G. S WINDLE (1998): Option Replication with
Transaction Costs: General Diffusion Limits. Annals of Applied Probability 8(3),
676707.
B ENSAID , B., J. P. L ESNE , H. P AG ES , and J. S CHEINKMAN (1992): Derivative Asset
Pricing with Transaction Costs. Mathematical Finance 2, 6386.
B LACK , F., and M. S CHOLES (1973): The Pricing of Options and Corporate Liabilities.
Journal of Political Economy 81, 637654.
B OYLE , P., and T. V ORST (1992): Option Replication in Discrete Time with Transac-
tions Costs. Journal of Finance 47(1), 271293.
C ONSTANTINIDES , G., and T. Z ARIP HOPOUL OU (1999): Bounds on Prices of Contin-
gent Claims in an Intertemporal Economy with Proportional Transaction Costs
and General Preference. Finance and Stochastics 3, 345369.
C ONSTANTINIDES , G., and T. Z ARIP HOPOUL OU (2001): Bounds on Prices of Contin-
gent Claims in an Intertemporal Economy with Multiple Securities. Mathemati-
cal Finance 11, 331346.
C OX , J. , S. R OSS , and M. R UBINSTEIN (1979): Option Pricing: A Simplied Ap-
proach. Journal of Financial Economics 7, 229263.
D AVIS , M., and J. C LARK (1994): A Note on Super Replicating Strategies. Phil. Trans.
Roy. Soc. London Ser. A347, 485494.
D AVIS , M. H. A., V. G. P ANAS , and T. Z AR IPHOPO ULOU (1993): European Option
Pricing with Transaction Costs. SIAM Journal on Control and Optimization 31,
470493.
E DIRISINGHE , C., V. N AIK , and R. U PPAL (1993): Optimal Replications of Options
with Transactions Costs. Journal of Financial and Quantitative Analysis 28(1),
117138.
22
-
8/10/2019 SSRN-id1481415
23/23
G RANNAN , E. R., and G. S WINDLE (1996): Minimizing Transaction Costs of Option
Hedging Strategies. Mathematical Finance 6, 341364.
H OGGARD , T., A. E. W HALLEY , and P. W ILMOTT (1994): Hedging Option Portfolios
in the Presence of Transaction Costs. Advances in Futures and Options Research
7, 2135.
K ABANOV , Y. M., and M. M. S AFA RIAN (1997): On Lelands Strategy of Option Pric-
ing with Transaction Costs. Finance and Stochastics 1, 239250.
L ELAND , H. E. (1985): Option Pricing and Replication with Transaction Costs. Jour-
nal of Finance 40, 12831301.
L ELAND , H. E. (2007): Comments on Hedging errors with Lelands option model in
the presence of transactions costs. Financial Research Letters 4, 200202.
M ERTON , R. C. (1973): Theory of Rational Option Pricing. The Bell Journal of Eco-
nomics and Management Science 4(1), 141183.
R UBINSTEIN , M. (1994): Implied Binomial Trees. Journal of Finance 49, 771818.
S ONER , H. M., S. S HREVE , and J. C VITANI C (1995): There Is No Nontrivial Hedging
Portfolio for Option Pricing with Transaction Costs. Annals of Applied Probabil-ity 5, 327355.
T OF T , K. B. (1996): On the Mean-Variance Tradeoff in Option Replication with Trans-
action costs. Journal of Financial and Quantitative Analysis 31(2), 233263.
ZHAO , Y., and W. T. Z IEMBA (2007a): Comments on and Corrigendum to Hedging Er-
rors with Lelands Option Model in the Presence of Transaction Costs. Finance
Research Letters 4, 49 58.
ZHAO , Y., and W. T. Z IEMBA (2007b): Hedging Errors with Lelands Option Model in
the Presence of Transaction Costs. Finance Research Letters 4, 49 58.
23