You are on page 1of 6

WILMOTT

Aug 2001
DAVID BAKSTEIN’S

LET IT
Real-world markets do not always
subscribe to the Black-Scholes
FLOW
THE PRICING OF DERIVATIVES
assumption of complete liquidity. IN ILLIQUID MARKETS
Here David Bakstein finds a real-
will be, in general, a nonlinear monotonically increasing
istic method for parameterizing the function of trade size.
effects of finite liquidity and develops Secondly, liquidity is directly responsible for the degree
of market slippage. This means that, since every participant
models for pricing derivatives in a can observe the same market depth, every trade of any

T
less-than- perfect market one agent is felt throughout. If a large trader removes a
certain price level in its entirety then market makers may
wo of the underlying assumptions of, adjust their prices subsequently. This results in compara-
amongst others, the basic Black-Scholes or tively large individual transactions or influential market
CAPM economies are first, frictionless participants pushing the asset price in a certain direction, in
markets and second, that every agent is a some cases deliberately (see Taleb, 1997). Hence there is
price-taker. However, real-world markets a market manipulation effect associated with liquidity that
substantially deviate from these assumptions because, for goes beyond pure transaction costs. The two effects may
virtually all traded assets, there exist both bid-ask spreads have opposite signs, although for the market to be free of
and a limited market depth. The effects of the two on asset arbitrage, transaction costs have to be higher than the
dynamics are loosely referred to as liquidity, meaning the gains from market manipulation.
more of an asset that is tradable at tight spreads, the more Generally, however, there is no consensus approach to
liquid, and thus attractive, a market. A vast amount of the parameterization and measurement of the liquidity of
research is being currently conducted on how to measure, a market. The papers of Longstaff (1995), and Chordia,
parameterize, price and manage liquidity in most fields of Roll and Subrahmanyam (2000), use combinations of
finance. This includes the extension of basic arbitrage or bid-ask spreads, volume and open interest as a proxy to
equilibrium models to cover the case of finite liquidity. empirically investigate the effects on returns and distribu-
The latter has two main effects. First, it represents a tions of the underlying and options on it. The papers of
random transaction cost which is correlated with the Jarrow (1992), Schönbucher (1993), Frey (1996),
market’s dynamics. In general, a market consists of Almgren & Chriss (1999) and Huberman & Stanzl (2000),
competing buyers and sellers, who quote an asset’s propose liquidity models that feature a reaction function
transaction directions, prices and quantities. The most that models the immediate impact of a trade and the aver-
common exchange structures are, respectively, a age price paid per asset. It is also a function of both a
monopolistic market maker, an oligopoly of market liquidity scaling parameter and the trade size. A possible
makers or an order-driven market. In all cases, there will proxy for the former is explicitly given by Krakovsky
be layers of bid and ask quotes with the respective quan- (1999), as the ratio of notional traded to the relative
tities. The width and depth of the spreads represent change in the price of the underlying asset. This choice of
primarily a transaction cost for market makers (since they estimator has the advantage that, at the time of the trade,
will buy low and sell high) and, also, an insurance against the liquidity parameter is observable and predictable. The
asymmetric information. Generally, if many competing papers by Almgren & Chriss (1999), Price-Impact Func-
market participants that want to trade exist , bid-ask tions, Huberman & Stanzl (2000) and Optimal Liquidity
spreads tend to be narrow and market depth substantial Trading, Huberman & Stanzl (2000), further consider the
because low transaction cost will attract large volume. permanent slippage effect on the asset, by making its new
However, whereas it may be possible for agents to trade equilibrium price a function of both the previous and the
small quantities of an asset at the best possible price, the average transaction price. However, they only apply their
larger the trade size the more levels of market depth will models to optimal portfolio trading strategies.
have to be tapped. Hence the average transaction price

www.w i l m o t t. com for the Quant i t at i ve Fi n a n ce Co m m u n i t y, Te c h n o l o gy and Analys i s


1
David Bakstein’s Let it Flow Wilmott July 2001

VA LUATION AND HEDGING OF


OPTIONS (1)
One area of finance where liquidity is a significant factor
is the valuation and hedging of options. Even if no single where . The bond on the other hand will always yield the
trader has the intention to push the market in a certain riskless return , namely
direction, there exist agents who have to trade certain
quantities of the underlying in order to hedge their expo- (2)
sure to a portfolio of derivatives. If, as in the
Black-Scholes theory, they try and Delta-hedge, then for Moreover, we can set the initial values of stock and
options with non-smooth or even-discontinuous payoffs, bond equal to and , without loss of gener-
the Delta and Gamma, i.e. the amount of the underlying ality. Two key properties of the model are, firstly, the
they have to hold and add/remove, respectively, may absence of arbitrage provided that
become very large close to expiry or close to payoff and, secondly, that for appropriate choice of , and the
discontinuities. Since, in reality, markets only have risk-neutral probability the model’s first two moments,
limited liquidity, they will thereby automatically move the approximately, can be fitted to the corresponding
market in a certain direction. To avoid any mis-hedging, moments of continuous geometric Brownian motion
the respective ratios have to be adjusted for this feed-
back effect. This in turn will affect the price of the (3)
portfolio, since the risk-free amount that can be earned
on a replicating portfolio changes as well. Moreover, the where is the assets volatility and increments of
price of a portfolio of options is not the sum of the indi- standard Brownian motion. The same model is employed
vidual options. in the seminal paper by Black & Scholes (1973), as the
To incorporate the effects of finite liquidity into option model for the underlying asset.
prices and hedging strategies, we employ a discrete-time On top of this random process for the underlying we
model, based on binomial trees. For the transaction cost construct a controlled process that represents the effect
effect we make the observable asset price an exponen- of a large or influential trader on the market. We denote
tial function of the trade size, scaled by a liquidity this trader’s holding process in the stock by
parameter. For the permanent slippage effect, we take a and in the bond by . Both processes are
geometric average of the last observed and the average adapted to the filtration and one step
transaction price. This makes the model nonlinear. We ahead predictable with respect to it. The latter point
show that under certain realistic assumptions the trees entails, that the trader’s portfolio can be rebalanced in
become recombining and can be implemented. By between the random jumps of the underlying asset. If we
changing the sign of the option payoff, we derive natural now assume that represents the mid-market price,
bid-ask spreads of the option that arise from the degree then the best buying and selling prices will be above and
of illiquidity of the market for the underlying. Finally, we below, respectively. Also, if the quantity traded is larger
mention some further extensions to and applications of than the quantity offered at the best price, then more than
the basic model. one quote has to be filled in order to complete the trade.
This means that the average transaction price is a
THE BASIC MODEL monotonically increasing function of the trade size. We
The main building block for the pricing framework of define its process as a function of the current
derivatives and portfolio trades is a suitable model for the spot , liquidity and trade size . Intu-
underlying asset. We thus commence our analysis in a itively, the trade-reaction or price-impact function, in
discrete-time finite horizon economy where trading in addition to being monotonous and positively sloped with
assets takes place at times . The respect to the trade size, should have the properties that
state of the economy is given by the finite set
and the revelation of the true state by
the increasing sequence of algebras . The Figure 1: Average Transaction Prices
initial set of states is , the eventual true state of
the economy is revealed as , . There are ,
two assets, namely a risky stock and a riskless
bond , whose respective processes are adapted to the
filtration and valued in . ,
Resorting to the widely used binomial model of Cox,
Ross & Rubinstein (1979), we will model randomness,
which represents the arrival of information and agents .
trading in the stock, by making the risky asset go up by a
fraction with probability or down by a fraction One possible function as already noted in Jarrow
with probability over one time step. Therefore (1992) and Frey (1996) is

www.w i l m o t t. com for the Quant i t at i ve Fi n a n ce Co m m u n i t y, Te c h n o l o gy and Analys i s


2
David Bakstein’s Let it Flow Wilmott July 2001

(4) , (7)

where is a liquidity scaling parameter and we as the model for the dynamics of the underlying. Even
suppressed the explicit dependence on the trajectory . though the latter is a computationally convenient model for
As an example, Figure 1 shows the exact average high-dimensional portfolio trading applications, it may
transaction price as a function of trade size for an order cause concerns when applied to the pricing of derivatives,
book with homogeneous equidistant market depth and mainly due to the fact that the spot of the underlying may
compares it with an estimate obtained from Equation (4). become negative with positive probability. In the standard
The total cash flow and implicit transaction cost are geometric Brownian motion this is only possible with zero
given by and , probability.
respectively. Unlike the transaction cost functions of
Boyle and Vorst(1992) and Edirisinghe et al (1993), THE HEDGING AND PRICING OF
Equation (4) is asymmetric, but does not require the VANILLA OPTIONS UNDER T H E
modulus sign, which, as we will see, makes it possible to BASIC MODEL
remove the path dependence and make the resulting Contingent claims are valued in reference to the initial
tree recombining. value of a portfolio strategy in the underlying risky and
In addition to the transaction cost effect, there is a riskless assets. This self-financing hedging strategy
1
market manipulation effect that is felt by all participants, , with , will exactly repli-
since the best quotes have been removed from the order cate or super-replicate any payoffs of the claim
book. Unlike Jarrow (1992 and 1994), Schönbucher . For a discrete time economy the valuation
(1993), Frey (1996) and Krakovsky (1999), the papers by of European vanilla-type contingent claims under our finite
Almgren & Chriss (1999) and Hubermann & Stanzl (2000) liquidity model can then be formulated as a non-linear
treat the reaction function as an instantaneous price programme with objective function
impact and they distinguish a permanent price update
effect, which is a function of both the previous equilibrium (8)
and the average transaction price. An intuitive explanation
is that large trades may not contain fundamental new infor-
mation, which would push the market to an untenable subject to initial holding, the self-financing and payoff
price level. A mathematically convenient model for this super-replication constraints
effect is to make the new equilibrium log-price a linear
combination of the two previous equilibrium and average , (9)
transaction log-prices or, equivalently, a geometric aver-
age of the two prices: (10)

(5)

If and constant, then the new observable ,


price is a convex combination. However, , realisti- (11)
cally, can be negative, since in general the average
transaction price is, depending on the trade direction, ,
below or above the last price traded, unless only one
level of market depth was filled. Combining the instanta- respectively, where the processes of are
neous trade-reaction (4), the permanent slippage (5) and given by (2) and (6). Because, in general, , and
reverting to the binomial representation (1) we obtain thus are functions of the present and past stock-
the price dynamics holdings, the problem is path-dependent and the n u m b e r
of variables as well as constraints is exponentially
growing as the number of time steps increases. As an
example, we consider the three period economy with the
set of states and the filtra-
2
. (6) tion. , ,
and .
Then the asset’s dynamics are
This model setup, albeit structured similarly, is differ-
ent from those of Almgren & Chriss (1999), and
Hubermann & Stanzl (2000) who, in their respective
papers, resort to arithmetic Brownian motion

www.w i l m o t t. com for the Quant i t at i ve Fi n a n ce Co m m u n i t y, Te c h n o l o gy and Analys i s


3
David Bakstein’s Let it Flow Wilmott July 2001

large trader a valid reason to exactly hedge his position.


Figure 2: Asset Tree for α≠0
The asset tree becomes recombining and thus feasible
to implement. The asset’s dynamics are visualized in
Figure 2. Under the special case where the
process reduces to Figure 3. The case entails that
any price impact due to large trades is a permanent
effect in its entirety.
Nevertheless, it represents a possibly large scale
nonlinear optimization problem. To solve for the holding
process we have to resort to an opti-
misation algorithm that will converge quickly. One
standard possibility is the Newton method:

, where

Figure 3: Asset Tree for α=0

is the Jacobian matrix. For the terminal condition (11)we


have to solve the system of implicit nonlinear functions

and

, where and are the number of down and


time steps, respect i v e l y, and
for notational conven-
ience. Furthermore, the intermediate self-financing
conditions (11) span the system

It becomes apparent that a one-period model has two


3
variables/constraints, a two-period model has six vari- and
ables/constraints and an -period model
variables/constraints. The controlled process makes the
asset tree bushy and thus hard to implement.
However, when we turn the inequality (11) into an equal-
ity for Markovian contingent claims, two distinct trajectories
with an identical number of up and down moves at a time
will result in identical holdings in stock and bond e.g. ,
. We refer to this condition as the
‘justified manipulation’ effect. Because market manipulation , and suppressing the explicit
or front-running are normally illegal, the condition gives the dependence on .

www.w i l m o t t. com for the Quant i t at i ve Fi n a n ce Co m m u n i t y, Te c h n o l o gy and Analys i s


4
David Bakstein’s Let it Flow Wilmott July 2001

Figure 4: Average Transaction price liquidity for the transaction will be good; the converse holds if
estimate with two parameters one follows the market. The reaction function (4) offers only
one scaling parameter and has a linear approximation for
small changes. Thus it may not offer enough flexibility to
account for distinct bid and ask liquidity. One simple modifica-
tion would be to replace (4) by

where , are the bid and ask liquidity, respectively,


and is the indicator function. Figure 4 shows an
actual order book snapshot and a two parameter esti-
mate.
This modification, however, makes the model path-
Figure 5: dependent and we cannot solve it through the tree
structure any longer. Instead, a large-scale dynamical
programming algorithm, possibly with approximations,
would have to be employed (see Edirisinghe, Naik &
Uppal,1992).

PARAMETERIZATION AND
CALIBRATION OF THE MODEL
Krakovsky (1999), explicitly defines liquidity as the reciprocal
of i.e. the sensitivity of the stock price to the quantity
traded. However, in this form the parameter is not dimension-
less and depends on the absolute size of both the quantity and
nominal stock price. Abetter measure would be to treat the
Calculating the liquidity-modified values for put product as a dimensionless variable. In this
options with time to expiry of one year, 50 time steps, case . The liquidity parameter therefore becomes
strike of 50, annualized riskless rate of 5 per cent and observable at the time of the trade, since the market depth is
volatility of 20 per cent gives the results, which can be visible. Figure 5 shows for all the subsequent trades in one
seen in the Results Tables published on the Wilmott particular trading day.
website. To make liquidity more comparable across different
The case implies that there is no permanent slippage stocks and markets we would need to make the denom-
effect. The other market participants did not consider the trade inator dimensionless as well. This could be done by
to be based on fundamental information. In this case the dividing it by the total quantity traded across the time
model resembles the pure transaction cost models of Boyle & interval in question. That means, that one’s own trades
Vorst (1992), Bensaid, Lesne, Pagès & Scheinkman (1992), are treated as a fraction of the total market. However the
and Edirisinghe, Naik & Uppal (1993). In fact, we can deduce total trade size in general is not predictable.
the value of the manipulation effect of illiquid markets by
subtracting the result of a particular choice of from the result CONCLUSION
for . We believe that our model offers a flexible, simple but real-
The calculated prices represent the seller’s price, i.e. istic approach to parameterizing liquidity. It relies on inputs
how much a writer would require or a buyer would need that are either directly observable or possible to estimate.
to pay for. By multiplying the payoffs by we obtain the Moreover, the speed of calculation entirely depends on the
buyer’s price, i.e. how much the customer would obtain for choice of optimization algorithm employed. Also, this model
entering into this position. These two prices, which due to may offer the framework for a number of related applications
the nonlinearity of the model will not be the same, repre- that primarily depend on liquidity.
sent natural bid-ask spreads that are founded on the
degree of illiquidity of the market for the underlying. The PORTFOLIO TRADING
tables above show the bid-ask spreads for different Portfolio trading is the liquidation or rebalancing of a large
scenarios. portfolio of one or more stocks. In general, the portfolio is
assumed to be large enough to move a market substan-
DISTINCT BID AND ASK LIQUIDITY tially, so that it has to be broken up into smaller chunks.
Typically, the market depth on the bid and ask side (and thus Sometimes an agent guarantees a client the liquidation
liquidity) is not equal. If there exist large imbalances this usually price in advance, usually in terms of a spread around the
leads to increased volatility and to price movements. In that volume weighted average price over a period of time.
case i.e. buying when everybody is selling and vice versa, the Depending if it is necessary to return any outperformance

www.w i l m o t t. com for the Quant i t at i ve Fi n a n ce Co m m u n i t y, Te c h n o l o gy and Analys i s


5
David Bakstein’s Let it Flow Wilmott July 2001

of the vwap to the client or not, the initial agreement repr FOOTNOTES
sents an option. The papers of Almgren & Chriss (1999) 1. This condition is for simplicity only. If there is an initial hold-
ing in the underlying, then the modifications if the model are
and Huberman & Stanzl (2000) deal with this problem by straightforward, but it has an effect on the valuation.
resorting to arithmetic Brownian motion, optimizing on 2. Strictly speaking, the filtration is given by the -alge-
objective function that trades off return against variance, bra of the given partition at every , i.e. all the unions and
complements of the elements of .
scaled by a risk-aversion parameter. The implementation 3. We do not count the holdings or constraints.
with our model would be straightforward.
REFERENCES
LIQUIDITY OPTIONS • Almgren & Chriss, Value under liquidation, Risk, Dec. 1999
• Bensaid, Lesne, Pagès & Scheinkman, Derivative Asset
Scholes (1999), defines liquidity options as the right or obli- Pricing with Transaction Costs, Mathematical Finance 2, 1992
gation to buy or sell a certain amount of an asset at the quoted • Black & Scholes, The Pricing of Options and Corporate
spot price, exercisable within a prespecified time window. Liabilities, Journal of Political Economy 2, 1973
• Boyle & Vorst, Option Replication in Discrete Time with
Under perfect liquidity, this amounts to a call or put option with Transaction Costs, The Journal of Finance 1, 1992
a strike price of zero. Hence it would theoretically amount to • Chordia, Roll & Subrahmanyam, Order Imbalance, Liquidity
the forward price of the asset. However, when liquidity is not and Market Returns, UCLA, Working Paper, 2000
• Cox, Ross & Rubinstein, Option Pricing: A Simplified
perfect this valuation does not hold any longer, since it may Approach, Journal of Financial Economics, Sept. 1979
not be the cheapest alternative to take a static hedge up-front. • Edirisinghe, Naik & Uppal, Optimal Replication of Options with
Transactions Costs and Trading Restrictions, March 1993
• Frey, The Pricing and Hedging of Options in Finitely Elastic
Markets, University of Bonn, Discussion Paper, 1996
E XOTIC OPTIONS IN ILLIQU I D • Frey & Stremme, Market Volatility and Feedback Effects
MARKETS from Dynamic Hedging, University of Bonn, Discussion
Paper, 1995
Taleb (1997), mentions (possibly illegal) practices of a • Huberman & Stanzl, Arbitrage-free Price-update and Price-
large trader front-running the client who holds positions Impact Functions, Yale, Working Paper, 2000
in the market. The author also mentions that clients • Huberman & Stanzl, Optimal Liquidity Trading, Yale, Working
Paper 2000,
require a liquidity rebate when entering into positions in • Jarrow, Market Manipulation, Bubbles, Corners, and Short
illiquid markets, especially when exposed to knock-out Squeezes, Journal of Financial and Quantitative Analysis,
barriers. Our model may be extended to exotic, possibly Sept. 1992,
• Jarrow, Derivative Security Markets, Market Manipulation
non-Markovian payoffs, so that the manipulation effect and Option Pricing Theory, Journal of Financial and Quantita -
can be extracted. tive Analysis, June 1994
• Krakovsky, Pricing liquidity into derivatives, Risk, Dec. 1999
STRIKE DETECTION • Lo & Wang, Implementing Option Pricing Models When Asset
Returns Are Predictable, The Journal of Finance 1, 1995
Finally, our model may prove useful for the inverse prob- • Longstaff, Option Pricing and the Martingale, Review of Financial
lem: given that certain large trades are observed, is it Studies, 4, 1995
• Scholes, Liquidity options,Risk, Nov. 1999
possible to deduce where the trader wants the asset • Schönbucher, The feedback effects of hedging in illiquid
price to be or what position (strike, barrier) is defended. markets, University of Oxford, MSc Thesis, 1993
Lo & Wang, 1995, show how options should be priced • Sircar & Papanicolaou, General Black-Scholes models
accounting for increased market volatility from hedging strate-
when asset returns are correlated. Another possibility gies, Applied Mathematical Finance 1, 1998
would be to perform a maximum likelihood analysis, after • Taleb, Dynamic Hedging, Wiley, 1997
having observed a sequence of large trades. • Contact: bakstein@maths.ox.ac.uk;
work supported by the EPSRC, Charterhouse, Socrates and
All these applications form part of current research and ESF;
development. Hopefully some interesting results can be The author wishes to thank: Sam Howison, Hyungsok Ahn,
expected in the future. Jeff Dewynne, Henrik Rasmussen and Paul Wilmott for help-
ful comments.

www.w i l m o t t. com for the Quant i t at i ve Fi n a n ce Co m m u n i t y, Te c h n o l o gy and Analys i s


6