See discussions, stats, and author profiles for this publication at: [Link]

net/publication/50344001

Dynamic equity asset allocation with liquidity-adjusted market risk criterion:

Appraisal of efficient and coherent portfolios

Article in Journal of Asset Management · March 2011

DOI: 10.1057/jam.2010.28 · Source: OAI

CITATIONS READS

27 188

1 author:

Mazin A. M. Al Janabi
Tecnológico de Monterrey
131 PUBLICATIONS 1,633 CITATIONS

SEE PROFILE

All content following this page was uploaded by Mazin A. M. Al Janabi on 01 May 2017.

The user has requested enhancement of the downloaded file.

Original Article

Dynamic equity asset allocation

with liquidity-adjusted market risk
criterion: Appraisal of efficient and
coherent portfolios
Received (in revised form): 15th September 2009

Mazin A.M. Al Janabi

is associate professor of finance and has several years of real-world experience in financial markets. He has written extensively on
contemporary topics in trading, market and credit risk management. His research and consulting activities address practitioner
and regulatory issues in financial risk management and derivative securities.

Correspondence: Mazin A.M. Al Janabi, Faculty of Business and Economics, United Arab Emirates University, PO Box 17555,
Al-Ain, UAE
E-mail: [Link]@[Link]

ABSTRACT This article extends research literature related to the evaluation of modern
portfolio risk management techniques by providing a broad modeling of dynamic equity
asset allocation under the supposition of illiquid and adverse market settings. This study
analyzes, from a fund manager’s perspective, the performance of liquidity adjusted
risk modeling in obtaining efficient and coherent equity trading portfolios subject to
realistic operational constraints as specified by the fund manager. Specifically, the article
proposes a re-engineered and robust approach to equity optimal portfolio selection, in a
Liquidity-Adjusted Value at Risk (L-VaR) framework, and particularly from the perspective
of trading portfolios that have both long and short trading positions or for trading
portfolios that consists merely of long positions. Moreover, in this article, the authors
develop a dynamic portfolio selection model and an optimization algorithm that allocates
equity assets by minimizing L-VaR subject to the constraints that the expected return,
trading volume and liquidation horizon should meet the budget limits set by the fund
manager.
Journal of Asset Management (2011) 12, 378–394. doi:10.1057/jam.2010.28; published online 10 March 2011

Keywords: emerging markets; financial engineering; GCC; liquidity risk; portfolio

management; value at risk

INTRODUCTION AND market upheavals has prompted new

OVERVIEW research underlining the necessity for market
participants to develop reliable dynamic
The significance of assessing market risk of portfolio management and risk assessment
portfolios of financial securities has long methods and algorithms. In measuring
been acknowledged by academics and market risk of trading portfolios, one
practitioners. In recent years, the growth of technique advanced in the literature involves
trading activities and instances of financial the use of Value at Risk (VaR) models that

& 2011 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 12, 6, 378–394
[Link]/jam/
 Dynamic equity asset allocation

ascertain how much the value of a trading over time and the problem is formulated as
portfolio would plunge, in monetary terms, a constrained utility maximization problem
over a given period of time with a given over a period of time. To this end, a dynamic
probability as a result of changes in market programming technique is applied to derive
prices. Nowadays, VaR is by far the most the Hamilton-Jacobi-Bellman equation
popular and most accepted risk measure (HJB) and the method of Lagrange multiplier
among financial institutions, however, is used to tackle the constraint. Moreover,
whether or not there is a best way to a numerical method is proposed to solve
estimate VaR is still debatable. From a the HJB-equation and hence the optimal
portfolio market risk point of view, VaR constrained portfolio allocation. Under
faces some major difficulties. Three of the this formulation, the author argues that
most researched and discussed issues are the investments in risky assets are optimally
non-normal behavior of market returns, reduced by the imposed VaR constraint.
volatility clustering and the impact of Finally, in a relatively recent study,
illiquid securities. The effect of the latter Alexander and Baptista (2008) examine the
on portfolio risk management and dynamic impact of adding a VaR constraint to the
asset allocation is the main focus of this problem of an active manager who seeks to
article. outperform a benchmark while minimizing
Several authors have investigated the tracking error variance (TEV) by using the
use of VaR for the selection of optimum model of Roll (1992). As such, the authors
portfolios and for active portfolio management. obtain three main results. First, portfolios
For instance, Campbell et al (2001) develop on the constrained mean-TEV boundary
a portfolio selection model that allocates still exhibit three-fund separation, but the
financial assets by maximizing expected weights of the three funds when the
return subject to the constraint that the constraint binds differ from those in Roll’s
expected maximum loss should meet the model. Second, the constraint mitigates
VaR limits set by the risk manager. Similar to the problem that when a manager seeks
the mean-variance approach, a performance to outperform a benchmark using the
index like the Sharpe index is constructed. mean-TEV model, he or she selects an
Furthermore, when expected returns are inefficient portfolio. Finally, when short
assumed to be normally distributed the sales are disallowed, the extent to which the
model provides almost identical results to constraint reduces the optimal portfolio’s
the mean-variance approach. As such, the efficiency loss can still be notable but is
empirical analysis has been provided by using smaller than when short sales are allowed.
two risky assets: US stocks and bonds, and Indeed, their results on the usefulness of a
the empirical results highlight the influence VaR constraint justify the belief that the fund
of both non-normal characteristics of the management industry is increasingly using
expected return distribution and the length VaR to (1) allocate assets among managers,
of investment time horizon on the optimal (2) set risk limits, and (3) monitor asset
portfolio selection. allocations and managers.
In another study, Yiu (2004) looks at In effect, the conventional VaR approach
the optimal portfolio problem by imposing to computing market (or trading) risk of
VaR as a dynamic constraint. This provides a portfolio does not explicitly consider
a way to control risks in the optimal portfolio liquidity risk. Typical VaR models are based
and to fulfill the requirement of regulators on modern portfolio management theory
on market risks. Furthermore, the VaR and assess the worst change in mark-to-
constraint is derived for some risky assets plus market portfolio value over a given time
a risk-free asset and is imposed continuously horizon but do not account for the actual

& 2011 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 12, 6, 378–394 379
 Al Janabi

trading risk of liquidation. In general, of any particular stock market had been
customary fine-tunings are made on an ad hoc employed, as other authors have done
basis. At most, the holding period (or heretofore. The basic argument is that
liquidation horizon) over which the VaR specific country indices may have, compared
number is calculated is adjusted to ensure to individual stocks, a more predictable
the inclusion of liquidity risk. As a result, structure owing to aggregation. Third,
liquidity trading risk can be imprecisely unlike most empirical studies in this field,
factored into VaR assessments by assuring this study employs a thorough and credible
that the liquidation horizon is as a minimum trading risk management model that
larger than an orderly liquidation interval. considers risk analysis under normal, severe
Moreover, the same liquidation horizon is (crisis) and illiquid market conditions. The
employed to all trading asset classes, albeit principal advantage of employing such a
some assets may be more liquid than others. model is the ability to capture a full picture
To address the above deficiencies, in this of possible loss scenarios of actual equity
research we characterize trading risk for trading portfolios. Fourth, this article
emerging equity markets by using a proposes a new approach to optimal and
multivariate Liquidity-Adjusted Value at coherent portfolio selection and within an
Risk (L-VaR) approach that focuses on the L-VaR framework. To this end, an L-VaR
modeling of optimum L-VaR under the approach is introduced to allocate equity
notion of illiquid and adverse market assets by minimizing L-VaR subject to
conditions and by exercising different meaningful operational and financial
correlation factors and liquidity horizon constraints. The focus on L-VaR as the
periods. The overall objective of this article appropriate measure of portfolio risk allows
is to construct different equity portfolios, risk managers and fund managers to assign
which include several stock markets indices the desired liquidity horizon and to allocate
of the Gulf Cooperation Council (GCC) long and short trading assets according to
zone, and to evaluate the risk characteristics realistic market trading conditions. In this
of such a portfolio besides examining an sense, the current work is closer to the real
optimization algorithm process for assessing behavior of fund managers who employ, in
efficient and coherent market portfolios. some pertinent way, active and dynamic asset
The literature on testing volatility, allocation trading strategies. To the present
expected returns and risk measurement of authors’ knowledge, none of the studies in
the GCC equity markets has been relatively the literature have done this type of empirical
meager, inconclusive and providing mixed analysis. Another contribution of the article
results. As such, and in contrast to all existing is to provide a new approach to estimating
published literatures pertaining to the the fund manager’s risk parameters.
application of advanced risk-return analysis Accordingly, a robust optimization algorithm
and dynamic asset allocation, this article is introduced to calculate risk tolerance in
intents to make the following main the L-VaR asset allocation model under the
contributions to the academic literature in notion of different liquidity horizons and
this specific equity risk management field. correlation factors.
First, it represents one of the limited The rest of the article proceeds as follows.
numbers of research papers that empirically The next section lays out the salient features
examines equity trading risk management and derives the necessary quantitative
using actual data of the six GCC financial infrastructure of L-VaR, and its limitations.
markets. Second, a database of the six GCC The subsequent section analyzes the overall
states indices is utilized whose behavior is results of the different empirical tests and
presumably more diverse than if equity assets discusses the process and infrastructure that

380 & 2011 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 12, 6, 378–394
 Dynamic equity asset allocation

support large-scale quantitative-based Incorporating equity liquidity

investing and the role of an optimization trading risk into L-VaR models
engine in this process. This section also The simplest way to account for liquidity
reflects on efficient and coherent portfolios trading risk is to extend the holding period
construction. The last section remarks on of illiquid positions to reflect a suitable
conclusions and recommendations. liquidation period. An adjustment can be
made by adding a multiplier to the VaR
measure of each trading asset type, which at
THEORETICAL FOUNDATION
the end depends on the liquidity of each
OF L-VAR MODELS FOR individual security. Nonetheless, the
EQUITY PORTFOLIO RISK weakness of this method is that it allows for
MANAGEMENT subjective estimation of the liquidation
period. Furthermore, the typical assumption
Equity trading risk management of a 1-day horizon (or any inflexible time
with a general parametric L-VaR horizon), within VaR framework, neglects
approach any calculation of trading risk related to
A simplified calculation process of the liquidity effect (that is, when and whether
estimation of VaR risk factors for a single a trading position can be sold out and at
and multiple assets’ positions is illustrated what price). A broad VaR model should
(Al Janabi, 2005, 2007) as follows: incorporate a liquidity premium (or liquidity
risk factor). This can be worked out by
formulating a method by which one can
VaRi ¼ a  Mark-to-Market Value of
unwind a position, not at some ad hoc rate,
Equityi  si ð1Þ
but at the rate that market conditions is
optimal, so that one can effectively set a risk
where a is the confidence level (or in value for the liquidity effects. In general,
other words, the standard normal variant at this will raise significantly the VaR, or the
confidence level a) and si is the standard amount of economic capital to support the
deviation (volatility) of the equity that trading position.
constitutes the single position. The Mark-to- In fact, if returns are independent and
Market Value of Equity indicates the amount they can have any elliptical multivariate
of trading position in equity i. distribution, then it is possible to convert the
For multiple equity assets or a portfolio of VaR horizon parameter from daily to any
equity assets, VaR is a function of each t-day horizon. The variance of a t-day return
individual equity’s risk and the correlation should be t times the variance of a 1-day
factor between the returns on the individual return or s2 ¼ f (t). Thus, in terms of
equities, and as follows: standard deviation (or volatility), s ¼ f (Ot)
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi and the daily or overnight VaR number [VaR
VaRP ¼ jVaRjT jrj jVaRj ð2Þ (1day)] can be adjusted for any t-day
horizon as:
pffi
This formula is a general one for the VaR ðt  dayÞ ¼ VaR ð1  dayÞ t ð3Þ
calculation of VaR for any portfolio regardless
of the number of equities. It should be noted The above formula was proposed and
that this formula is presented in terms of used by J.P. Morgan in their earlier
matrix-algebra – a useful form to avoid RiskMetricsTM method (1994). This
mathematical complexity, as more and more methodology implicitly assumes that
securities are added. liquidation occurs in one block sale at the

& 2011 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 12, 6, 378–394 381
 Al Janabi

end of the holding period and that there is sadj liquidity risk factor or standard
one holding period for all assets, regardless of deviation of the illiquid equity trading
their inherent trading liquidity structure. position.
Unfortunately, the latter approach does not
consider real-life trading situations, where The proposed approach assumes that the
traders can liquidate (or re-balance) small equity trading position is closed out linearly
portions of their trading portfolios on a daily over t days and the variance owing to
basis. The assumption of a given holding liquidity risk over t days is the sum of the
period for orderly liquidation inevitably variance (s2i , for all i ¼ 1, 2, y, t) of the
implies that assets’ liquidation occurs during liquidity risk on the individual days.
the holding period. Accordingly, scaling Moreover, we can assume with reasonable
the holding period to account for orderly accuracy that asset returns are independent
liquidation can be justified if one allows the and identically distributed (iid ) and serially
assets to be liquidated throughout the uncorrelated along the liquidation horizon,
holding period. thus:
In this work we present a re-engineered
approach for calculating a closed-form 
s2adj ¼ s21 þ s22 þ s23 þ   
parametric L-VaR with explicit treatment of 
liquidity trading risk. The proposed model þs2t2 þ s2t1 þ s2t ð4Þ
and liquidity scaling factor is more realistic
and less conservative than the conventional For this special linear liquidation case and
root-t multiplier. In essence the suggested under the assumption that the variance of
multiplier is a function of a predetermined liquidity risk of the first trading day decreases
liquidity threshold defined as the maximum linearly each day we can obtain:
position that can be unwound without
disturbing market prices during 1 trading t 2    
t1 2 2 t2 2 2
day. The essence of the model relies on the s2adj ¼ s21 þ s1 þ s1
t t t
assumption of a stochastic stationary process  2  2  2 !
and some rules of thumb, which can be of 3 2 1
þ þ s21 þ s21 þ s21 ð5Þ
crucial value for more accurate overall t t t
trading risk assessment during market stress
periods when liquidity dries up. To this end, Evidently, the additional liquidity risk
a practical framework of a methodology factor depends only on the number of days
(within a simplified mathematical approach) needed to sell an illiquid equity position
is proposed below with the purpose of linearly. In the general case of t-days, the
incorporating and calculating of illiquid liquidity factor is given by the following
assets’ horizon L-VaR, detailed along these functional expression:
lines:
In order to take into account the full  2    
illiquidity of equity assets (that is, the t t1 2 t2 2
s2adj ¼f þ þ
required unwinding period to liquidate an t t t
asset) we define the following (Al Janabi,  2  2  2 !
3 2 1
2008): þ þ þ þ ð6Þ
t t t
t number of liquidation days (t-days to
liquidate the entire equity asset fully) To calculate the sum of the squares, it is
s2adj variance of the illiquid equity trading convenient to use a short-cut approach.
position From mathematical series the following

382 & 2011 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 12, 6, 378–394
 Dynamic equity asset allocation

relationship can be obtained: be estimated from the total trading position

2 2 2 2 size and the daily trading volume that
ðtÞ þ ðt  1Þ þ ðt  2Þ þ    þ ð3Þ
can be unwound into the market without
t ðt þ 1Þ ð2t þ 1Þ significantly disrupting equity market prices;
þ ð2Þ2 þ ð1Þ2 ¼ ð7Þ
6 and in actual practices it is generally
Hence, after substituting equation (7) into estimated as:
equation (6), the following can be achieved:
Total Trading Position Size of Equityi
1 t¼
s2adj ¼ f 2 ðtÞ2 þ ðt  1Þ2 þ ðt  2Þ2 þ    Daily Trading Volume of Equityi
t ð11Þ
þð3Þ2 þ ð2Þ2 þ ð1Þ2 or s2adj
In real practices the daily trading volume
  of any trading equity asset is estimated as the
ð2t þ 1Þ ðt þ 1Þ
¼f ð8Þ average volume over some period of time,
6t
generally a month of trading activities. In
Accordingly, from equation (8) the effect, the daily trading volume of assets can
liquidity risk factor is obtained as: be regarded as the average daily volume or

(rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi)
1 
sadj ¼f 2
ðtÞ2 þ ðt  1Þ2 þ ðt  2Þ2 þ    þ ð3Þ2 þ ð2Þ2 þ ð1Þ2 or
t
(rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi)
ð2t þ 1Þ ðt þ 1Þ
sadj ¼f ð9Þ
6t

The final result of equation (9) is of the volume that can be unwound under a
course a function of time and not the severe crisis period.
square root of time as employed by some In essence, the above liquidity scaling
financial market’s participants based on the factor is more realistic and less conservative
RiskMetricsTM methodologies. The above than the conventional root-t multiplier and
approach can also be used to calculate L-VaR can aid financial entities in allocating
for any time horizon. Likewise, in order to reasonable and liquidity market-driven
perform the calculation of L-VaR under regulatory and economic capital requirements.
illiquid market conditions, it is possible to In order to calculate the L-VaR for the
use the liquidity factor of equation (9) and full trading portfolio under illiquid market
define the following: conditions (L  VaRPadj ), the above
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi mathematical formulation can be extended,
ð2t þ 1Þ ðt þ 1Þ with the aid of equation (2), into a matrix-
L  VaRadj ¼ VaR ð10Þ algebra form to yield the following:
6t
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
where VaR ¼Value-at-Risk under liquid L  VaRPadj ¼ jL  VaRadj jT jrj jL  VaRadj j ð12Þ
market conditions and L-VaRadj ¼ Value-at-
Risk under illiquid market conditions. The The above mathematical structure
latter equation indicates that L-VaRadj4VaR, (in the form of two vectors, |LVaRadj|,
and for the special case when the number of |LVaRadj|T and a correlation matrix |r|)
days to liquidate the entire equity assets is 1 can facilitate the mathematical modeling and
trading day, then L-VaRadj ¼ VaR. Indeed, programming process so that the equity
the choice of the liquidation horizon can trading risk manger can specify different

& 2011 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 12, 6, 378–394 383
 Al Janabi

liquidation horizons for the whole portfolio to compute L-VaR. Historical database
and/or for each individual trading equity (of more than 5 years) of daily closing index
according to the necessary number of days levels, for the period 17/10/2004–22/05/
to liquidate the entire asset completely. 2009, are assembled for the purpose of
carrying out this research and further for the
construction of market and liquidity risk
ANALYSIS OF PRICE RISK management parameters. The analysis of data
EXPOSURE OF EQUITY and discussions of relevant findings and
results of this research are organized and
PORTFOLIOS – EMPIRICAL explained as follows:
EVIDENCE FROM THE GCC
STOCK MARKETS
In this work, database of daily return of six Equity trading risk management
GCC stock markets’ main indicators with L-VaR modeling algorithm
(indices) are gathered, filtered and adequately In order to illustrate the linkage between
adapted for the creation of relevant inputs the theoretical constructs of L-VaR and its
for the calculation of all risk factors. practical application and value as a tool
Historical database of daily indices levels is for equity trading risk management, the
drawn from Reuters 3000 Xtra Hosted following hypothetical trading portfolio
Terminal Platform. The total numbers of with full case study is presented. Using the
indices that are considered in this work are definition of L-VaR in the third section and
nine indices; seven local indices for the six under the assumption that a given equity
GCC stock markets (including two indices portfolio has both long and short-selling
for the UAE markets) and two benchmark trading positions, Table A1 illustrates a
indices, detailed as follows: DFM General practical risk report for the coverage of
Index (UAE, Dubai Financial Market equity trading risk management activities
General Index), ADSM Index (UAE, Abu of a hypothetical equity portfolio consisting
Dhabi Stock Market Index), BA All Share of several indices of the GCC stock markets.
Index (Bahrain, All Share Stock Market Asset allocation and L-VaR analysis are
Index), KSE General Index (Kuwait, Stock performed under the assumption that local
Exchange General Index), MSM30 Index indices represent exact replicas of diversified
(Oman, Muscat Stock Market Index), portfolios of local stocks for each GCC stock
DSM20 Index (Qatar, Doha Stock Market market, respectively. Furthermore, all risk
General Index), SE All Share Index (Saudi analyses are performed at the one-tailed 97.5
Arabia, All Share Stock Market Index), per cent level of confidence over different
Shuaa GCC Index (Shuaa Capital, GCC liquidation periods.
Stock Markets Benchmark Index), and Shuaa In this first full-case analysis study the total
Arab Index (Shuaa Capital, Arab Stock portfolio value is AED10 million (UAE
Markets Benchmark Index). Dirham) with different asset allocation
Moreover, in this work index returns are percentage. The analysis is carried out with
defined as R i,t ¼ ln(Pi,t)ln(Pi,t1) where R i,t 1-day liquidity horizon – that is, 1-day to
is the daily return of index i, ln is the natural unwind all equity trading positions fully.
logarithm; Pi,t is the current level of index i, Furthermore, Table A1 illustrates the effects
and Pi,t1 is the previous day index level. of stress-testing (that is, L-VaR under severe
Furthermore, for this particular study we market conditions) and different correlation
have chosen a confidence interval of 95 per factors on daily L-VaR calculations. The
cent (or 97.5 per cent with ‘one-tailed’ loss L-VaR engine’s report depicts also the
side) and several liquidation time horizons overnight (daily) unconditional volatilities,

384 & 2011 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 12, 6, 378–394
 Dynamic equity asset allocation

which are calculated as the standard deviation different correlation factors in any L-VaR
of the percentage change in the index and stress-testing exercises. This is because
level (daily returns) of the nine indices, in existing trends in correlation factors may
addition to their respective sensitivity factors break down (or change signs) under adverse
(or the beta factors) vis-à-vis the benchmark and severe market movements, caused by
index. Crisis market daily volatilities (or unforeseen financial or political crises. In
downside-risk) are calculated and illustrated theory, the case with correlation r þ 1 should
in the report. These daily severe downside- provide the maximum L-VaR numbers
risk volatilities represent the maximum (AED 284 816 and AED 1 900 595) as a
negative returns (losses), which are perceived result of the fact that under these
in the historical time series, for all stock circumstances total L-VaR of actual trading
market indices. In essence, this approach can portfolio is the weighted average of
aid in overcoming some of the limitations of individual L-VaRs of each equity trading
normality assumption and can provide a position. Furthermore, the degree of
better analysis of (L-VaR) especially under risk-diversification (namely, the effects of
severe and illiquid market settings. The diversified L-VaR) of this hypothetical equity
effects of short-selling (albeit short-selling is trading portfolio can also be deduced simply
currently not permitted in the GCC stock as the difference in the values of the two
markets) are depicted in Table A1. One of greatest L-VaRs – that is the L-VaR of unity
the interesting results of this study is the way correlation case versus the L-VaR of
in which L-VaR numbers have decreased. empirical correlation case (AED 64 052 or
This behavior might be explained by the way 18.36 per cent for the normal market
in which the overall portfolio is funded – in condition case). However, it is appealing to
other words, long positions have been note here that for this particular case, L-VaR
funded with short-selling of other stocks under correlation r þ 1 is less than L-VaR
(or indices) and consequently have led to under empirical correlation case owing to
reduction in the overall risk exposure. the impact of short selling of some equity
The L-VaR modeling results are calculated trading positions. In addition, the overall
under normal and severe market conditions sensitivity factor (beta factor) of this long/
by taking into account different correlation short equity portfolio is indicated in this
factors (empirical, zero and negative/positive report as 0.209, or in other words, the total
unity correlations between the various risk equity portfolio value, with actual asset
factors). Therefore, with 97.5 per cent allocation ratios, has little positive sensitivity
confidence, the actual equity trading with the benchmark index (Shuaa Arab
portfolio should expect to realize no greater Index). Moreover, expected returns and
than AED 348 868 decrease in the value risk-adjusted expected returns (under normal
over a 1-day time frame. In other words, the and severe market conditions) are also
loss of AED 348 868 is one that an equity included in the L-VaR risk analysis report.
portfolio should realize only 2.5 per cent
of the time. If the actual loss exceeds the
L-VaR estimate, then this would be
considered a violation of the estimate. Optimization of efficient and
Furthermore, the analysis of L-VaR under coherent portfolios for an equity
illiquid market conditions is performed trading risk management unit
with four different correlation factors: using L-VaR modeling algorithm
empirical, zero and positive/negative unity, One of the basic problems of applied finance
respectively, and for long and short trading is the optimal selection of assets, with the
positions. Indeed, it is essential to include aim of maximizing future returns and

& 2011 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 12, 6, 378–394 385
 Al Janabi

constraining risk by appropriate measure. markets and particularly under illiquid and
To this end, Markowitz (1959) illustrated adverse market conditions. As such, in
that, for a given levels of risk, one can this research we look at the optimization
identify certain groups of equity securities problem from a different realistic operational
that maximize expected return. He angle. In view of that, the enigma is
considered these optimum portfolios as formulated by finding the portfolio that
‘efficient’ and referred to a continuum of minimize L-VaR, with expected return,
such portfolios in dimensions of expected trading volume and liquidation horizons are
return and standard deviation as the efficient constrained according to the requirements of
frontier. Accordingly, for asset allocation the fund manager. As such, the focus in this
purposes, fund managers should choose work is on the forecast of risk measure,
portfolios located along the efficient frontier. rather than on expected returns for two
Optimized portfolios do not normally reasons: first, several studies have analyzed
perform as well in practice as one would the forecasts of expected returns in the
expect from theory. For example, they are context of mean-variance optimization (see
often outperformed by simple allocation for instance Best and Grauer, 1991). The
strategies such as the equally weighted common opinion is that expected returns are
portfolio (Jobson and Korkie, 1981) or not easy to foresee, and that the optimization
the global minimum variance portfolio process is very sensitive to these variations.
(Jorion, 1991). Simply put, the ‘optimized’ Second, there exists a general notion that
portfolio is not optimal at all. Portfolio L-VaR, in a wide sense, is simpler to assess
weights are often not stable over time but than expected returns from historical data.
change significantly each time the portfolio In this work we develop a model for
is re-optimized, leading to unnecessary optimizing portfolio risk-return with L-VaR
turnover and increased transaction costs. constraints using realistic operational and
Moreover, these portfolios typically present financial scenarios and conduct a case study
extreme holdings (‘corner solutions’) in a on optimizing equity portfolios of the six
few securities while other securities have GCC stock markets. The case study shows
close to zero weight. It is well documented that the optimization algorithm, which is
(Michaud, 1989) that mean-variance based on linear programming techniques,
optimizers, if left to their own devices, is very stable and efficient in handling
can sometimes lead to unintuitive portfolios different liquidity horizons and correlation
with extreme positions in asset classes. factors. Moreover, the approach can tackle
Consequently, these ‘optimized’ portfolios large number of equity securities and rational
are not necessarily well diversified and fund management scenarios. L-VaR risk
exposed to unnecessary ex post risk management constraints (reduced to linear
(Michaud, 1989). The reason for these constraints) can be used in various
phenomena is not a sign that mean-variance applications to bound percentiles of loss
optimization does not work but rather that distributions.
the modern portfolio theory framework is Essentially, our approach is a straightforward
very sensitive to small changes in the inputs. extension of the classic Markowitz mean-
Consequently, for more than four variance approach, where the original risk
decades a wide body of knowledge has measure, variance, is replaced by L-VaR.
been accumulated about the performance, The task is attained here by minimizing
strengths and weaknesses of this approach L  VaRPadj , while requiring a minimum
when applied to equity portfolios. However, expected return subject to several financially
much is less known about portfolio meaningful operational constraints. Thus,
optimization techniques in emerging equity by considering different expected returns,

386 & 2011 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 12, 6, 378–394
 Dynamic equity asset allocation

we can generate an efficient L-VaR frontier. indicates an (nx1) vector for all i ¼ 1, 2, y, n.
Alternatively, we can also maximize returns The rationality of imposing the above
while not allowing for large risks. For the constraints is to comply with current
purpose of this research, the optimization regulations that enforce capital requirements
problem is formulated as follows: on investment companies, proportional to
From equation (10) we can define VaR and/or L-VaR of a trading portfolio
liquidation horizon factor (LHFi) for each besides other operational limits (for instance,
trading asset as: volume trading limits).
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ð2ti þ 1Þ ðti þ 1Þ
LHFi ¼ ð13Þ Empirical optimization of efficient
6ti and coherent portfolios – The case
of long and short equity trading
To compute efficient portfolios we solve
for the following quadratic programming positions
formulation: In this first study, the optimization process is
based on the definition of L-VaR as the
Minimize: L  VaRPadj minimum possible loss over a specified time
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi horizon within a given confidence level. The
¼ jL  VaRadj jT jrj jL  VaRadj j iterative-optimization modeling algorithm
solves the problem by finding the market
ð14Þ positions that minimize the loss, subject
Subject to the following operational and to the fact that all constraints are satisfied
financial budget constraints as specified by within their boundary values. Further, in
the fund manager: all cases the liquidation horizons as indicated
in Tables A2–A5 are assumed constant
X
n throughout the optimization process. For
Ri xi ¼ RP ; li pxi pui ; the sake of simplifying the optimization
i¼1 ð15Þ
algorithm and thereafter its analysis, a
i ¼ 1; 2; . . . ; n volume trading position limit of 10 million
AED is assumed as a constraint – that is the
X
n
xi ¼ 1:0 ; li pxi pui ; equity trading entity must keep a maximum
i¼1 ð16Þ overall market value of different equities
of no more than 10 million AED (between
i ¼ 1; 2; ::::; n
long and short-selling positions). As such,
jLHFjX1:0 ; 8i ; i ¼ 1; 2; . . . ; n ð17Þ Figure A1 provides evidence of the empirical
L-VaR efficient frontiers (under 1-day and
X
n
Vi ¼ VP ; i ¼ 1; 2; . . . ; n ð18Þ 10-days liquidation horizons, respectively)
i¼1 defined using a 97.5 per cent confidence
level. As mentioned above, the optimal
Here RP and VP denote the target portfolio selection is performed by relaxing
portfolio mean return and total portfolio the short sale constraint, for the different
volume, respectively, and xi the weight or equity assets. On the other hand, efficient
percentage asset allocation for each asset. The portfolios cannot always be attained (for
values li and ui, i ¼ 1, 2, y, n denote the example short selling without realistic lower
lower and upper constraints for the portfolio boundaries on xi) in the day-to-day real-
weights xi. If we choose li ¼ 0, i ¼ 1, 2, y, n, world portfolio management operations and,
then we have the situation where no hence, the fund manager should establish
short selling is allowed. Moreover, |LHF| proactive coherent portfolios under more

& 2011 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 12, 6, 378–394 387
 Al Janabi

realistic and restricted budget constraints, the simplifying assumptions of the theory and
detailed as follows: the realities of the world.
In order to illustrate the composition of
 Total trading volume (between long and
coherent portfolios [1], [2], [3] and [4],
short equity trading positions) is 10
Tables A2–A5 point out the asset allocation
million AED.
weights for all equity assets in all the
 Asset allocation for long equity trading
liquidation periods under consideration in
position varies from 10 per cent to 100
addition to their expected return and
per cent.
sensitivity factor. Similarly, the four tables
 Asset allocation for short equity trading
depict L-VaR and the L-VaR/Volume ratio
position varies from 10 to 60 per cent.
under normal market settings and with the
 All liquidity horizons for all equities are
assumption of three different correlation
kept constant according to the specified
factors. In this way, fund managers should
values as indicated in Tables A2–A5.
employ risk measures that allow them to take
Now the weights are allowed to take negative decisions, which would produce a risk
or positive values, however, since arbitrarily budget lower than a specific target. Thus,
high or low percentages have no financial this analysis is substantially a generalization
sense, we determined to introduce lower of the Markowitz analysis that permits one
an upper bound for the weights and in to determine the asymmetric aspect of risk.
accordance with reasonable trading practices. In any case, the benefit of portfolio
Furthermore, for comparison purposes and optimization critically depends on how
as the endeavor in this work is to minimize accurately the implemented L-VaR risk
L-VaR subject to specific expected returns, measure is forecasted.
we decide to plot L-VaR versus expected
returns and not the reverse, as commonly
well accustomed to in the various portfolio Empirical optimization of efficient
management literatures. Accordingly, it is and coherent portfolios – the case of
worthy of note that the four benchmark long equity trading positions
portfolios (coherent portfolios [1], [2], [3] For the sake of comparison of the first
and [4]) are noticeably located way off empirical optimization procedure with
from the efficient frontiers as indicated in another real-word operational constraint,
Figure A1. This is because financially and a second optimization process is carried out;
operationally real-world investment however, this time by considering long
considerations make it unlikely that a trading trading positions only and by imposing
portfolio will behave exactly as theory a restriction on short sales of equities.
predicts. Imperfections such as restriction on Similarly, in this second study, the
long and short trading positions, total trading optimization algorithm is based on the
volume and liquidation horizons make it definition of L-VaR as the minimum possible
unlikely to create an efficient equity trading loss over a specified time horizon within
portfolio. Thus, the fund manager should a given confidence level. Further, in all
apply active strategies in order to earn excess cases the liquidation horizons as indicated
returns. These considerations are especially in Tables A6–A9 are assumed constant
relevant for individual fund managers who throughout the optimization process.
may spread their trading positions across a Likewise, for the sake of simplifying the
few securities. Nevertheless, the elegance optimization routine and thereafter its
and compelling logic of the theory prompt analysis, a volume trading-position-limit of
attempts to apply the theory even though 10 million AED is assumed as a constraint –
practitioners recognize the variance between that is the equity trading entity must keep a

388 & 2011 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 12, 6, 378–394
 Dynamic equity asset allocation

maximum overall market value of different portfolios. Finally, and in order to illustrate
equities of no more than 10 million AED of the composition of coherent portfolios [1],
long trading positions. As such, Figure A2 [2], [3] and [4], Tables A6–A9 point out
provides evidence of the empirical L-VaR the asset allocation weights for all equity
efficient frontiers (under 1-day and 10-days assets in all the liquidation periods under
liquidation horizons, respectively) defined consideration in addition to their expected
using a 97.5 per cent confidence level. As return and sensitivity factor. Similarly, the
indicated above, the optimal portfolio four tables depict L-VaR and the L-VaR/
selection is performed by imposing a short Volume ratio under normal market settings
sale prohibition constraint, for the different and with the assumption of three different
equity assets. As a result, efficient portfolios correlation factors.
cannot always be attained in the day-to-day
real-world portfolio management operations
and, hence, the fund manager should CONCLUSION REMARKS
establish proactive coherent portfolios under Given the fact that mean-variance optimizers
more realistic and restricted budget have serious financial deficiencies, which
constraints, detailed as follows: could often lead to financially meaningless
‘optimal’ portfolios, in this article we
 Total trading volume of long equity
examine how to determine the optimal
trading positions is 10 million AED.
portfolio choice for an equity fund manager
 Asset allocation for long equity trading
under the assumption of different liquidation
position varies from 0 to 60 per cent.
horizons and by implementing different
 All liquidity horizons for all equities are
long trading scenarios or a combination
kept constant according to the specified
of long/short equity trading strategies. In
values as indicated in Tables A6–A9.
this research, we develop an optimal and
In this particular case the weights are allowed coherent portfolio selection model that
to take only positive values, however, as implements a downside risk constraint rather
arbitrarily high or low positive weights could than standard deviation alone. In our
have no financial investment sense, we approach, downside risk is written in terms
determined to introduce lower an upper of portfolio L-VaR, so that additional risk
bound for the weights and in accordance resulting from any non-normality and
with reasonable trading practices. As a result, illiquid assets may be used to estimate the
it is worth mentioning that the four portfolio L-VaR. This enables a much more
benchmark portfolios (coherent portfolios generalized framework to be developed,
[1], [2], [3] and [4]) are located somehow with the distributional assumption most
more closely to the efficient frontiers than appropriate to the type of financial assets
in the previous optimization case as indicated to be employed. We then provide a robust
in Figure A2. As discussed above, it seems portfolio optimization technique using
that this optimization phenomenon could L-VaR as a risk measure subject to the
not be attained for long and short trading imposition of financially and operationally
position because financially and operationally meaningful constraints. In the final section of
real-world fund management considerations this article, we describe the selection process
make it unlikely that a trading portfolio will for equity coherent portfolios, of either
behave exactly as theory predicts; however, merely long positions or a combination of
for long trading positions it seems that it is long/short trading positions, and provide the
possible to get closer to the efficient frontier composition of each trading portfolio. In
and synchronize to a certain degree the addition to the standard L-VaR optimization
performance of efficient and coherent algorithm that are adjusted for liquidity risk,

& 2011 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 12, 6, 378–394 389
 Al Janabi

related topics such as multi-horizon portfolio asset and the set of portfolio weights. The
selection models and robust L-VaR empirical results are interesting in terms of
optimization approaches with various theory as well as practical applications.
correlation factors are discussed.
This article extends previous approaches
to optimization problems with L-VaR ACKNOWLEDGEMENTS
constraints. In particular, the suggested This work has benefited from a financial
approach can be used for minimizing support in the form of a summer-grant (for
L-VaR under several budget constraints. the summer semester of the academic year
Furthermore, multiple L-VaR constraints 2009–2010) from the Faculty of Business and
with various unwinding liquidation periods Economics (FBE), United Arab Emirates
and correlation factors can be used to shape University, Al-Ain, UAE. The usual
the profit/loss distribution. In some cases, disclaimer applies.
the mean-variance optimizations are highly
unstable, that is, small changes in the input
assumptions can lead to large changes in REFERENCES
the solutions. As such, in this work, the Alexander, G. and Baptista, A.M. (2008) Active portfolio
management with benchmarking: Adding a value-at-risk
optimization algorithm is formulated by constraint. Journal of Economic Dynamics & Control 32:
finding a set of portfolios that minimize 779–820.
L-VaR subject to a given expected returns, Al Janabi, M.A.M. (2005) Trading risk management: Practical
applications to emerging-markets. In: S. Motamen-
trading volume and liquidity horizons. Samadian (ed.) Risk Management in Emerging Markets. UK:
To this end, the L-VaR’s risk function is Palgrave/Macmillan, pp. 91–136.
constrained by a downside risk measure in Al Janabi, M.A.M. (2007) On the use of value at risk for
addition to several operational constraints managing foreign exchange exposure in large portfolios.
Journal of Risk Finance 8(3): 260–287.
such as total volume, long and short trading Al Janabi, M.A.M. (2008) Integrating liquidity risk factor
positions, equity asset allocation weight into a parametric value at risk method. Journal of Trading
factors, and liquidity unwinding periods. 3(3): 76–87.
Best, M.J. and Grauer, R.R. (1991) On the sensitivity of
Our model is a straightforward re-engineered mean-variance-efficient portfolios to changes in asset
extension of the classic Markowitz mean- means: Some analytical and computational results. Review
variance approach, where the original risk of Financial Studies 4: 315–342.
Campbell, R., Huisman, R. and Koedijk, K. (2001) Optimal
measure, variance, is replaced by L-VaR and portfolio selection in a value-at-risk framework. Journal
by guaranteeing minimum expected return of Banking and Finance 25: 1789–1804.
under different liquidation horizons. This Jobson, J.D. and Korkie, B.M. (1981) Putting Markowitz
approach can aid in solving some of the real- theory to work. Journal of Portfolio Management 7: 70–74.
Jorion, P. (1991) Bayesian and CAPM estimators of the means:
world trading dilemmas under adverse Implications for portfolio selection. Journal of Banking and
market conditions: when liquidity dries up; Finance 15: 717–727.
correlations factors switching signs; and the Markowitz, H. (1959) Portfolio Selection: Efficient Diversification
of Investments. New York: John Wiley.
incorporation of non-normal distribution of Michaud, R.O. (1989) The Markowitz optimization enigma:
assets’ returns in risk measurement. The Is ‘optimized’ optimal? Financial Analysts Journal 45(1):
empirical findings indicate that the new 31–42.
Morgan Guaranty Trust Company (1994) RiskMetrics-Technical
approach performs adequately over the entire Document. New York: Morgan Guaranty Trust Company,
sample period and across alternative Global Research.
investment horizons. The results indicate Roll, R. (1992) A mean/variance analysis of tracking error.
that the fund manager’s L-VaR depends on Journal of Portfolio Management 18: 13–22.
Yiu, K.F.C. (2004) Optimal portfolios under a value-at-risk
the minimum expected return, individual constraint. Journal of Economic Dynamics & Control 28:
L-VaR positions, liquidity horizons of each 1317–1334.

390 & 2011 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 12, 6, 378–394
 Dynamic equity asset allocation

APPENDIX

Analysis of different trading

portfolios and asset liquidity risk
management

Table A1: Equity trading risk management and control report (L-VaR analysis, full case study)

Asset Allocation and Liquidity-Adjusted Value at Risk (L-VaR) Report

Liquidity Daily Daily
Market Value Asset Allocation Holding Volatility Volatility Sensitivity
Stock Market Indices in AED Percentage Horizon (Normal) (severe) Factor
Daily Liquidity-Adjusted Value at Risk (L-VaR) in AED
DFM General Index $ 6,000,000 60.0% 1 1.93% 12.16% 0.58 [Normal Market Conditions]
ADSM Index $ 4,000,000 40.0% 1 1.42% 7.08% 0.40 Correlation = Empirical Correlation =1 Correlation =0 Correlation = -1
BA All Share Index $ 1,000,000 10.0% 1 0.59% 3.77% 0.06 348,868 284,816 323,171 357,433
KSE General Index $ (2,000,000) -20.0% 1 0.76% 3.74% 0.14 3.49% 2.85% 3.23% 3.57%
MSM30 Index $ 2,000,000 20.0% 1 0.84% 8.70% 0.10
DSM20 Index $ 3,000,000 30.0% 1 1.53% 8.07% 0.31 Diversification Benefits
SE All Share Index $ (4,000,000) -40.0% 1 2.08% 11.03% 0.98 $ (64,052) -18.36%
Shuaa GCC Index $ - 0.0% 1 1.45% 8.10% 1.05
Shuaa Arab Index $ - 0.0% 1 1.28% 7.57% 1.00 Daily Liquidity-Adjusted Value at Risk (L-VaR) in AED
Total Portfolio Value in AED $ 10,000,000 100% [Severe (Crisis) Market Conditions]
Correlation = Empirical Correlation =1 Correlation =0 Correlation = -1
2,064,245 1,900,595 1,900,404 1,900,212
Expected Return and Risk-Adjusted Return 20.64% 19.01% 19.00% 19.00%
Trading Portfolio Expected Return 0.115% Diversification Benefits
Risk-Adjusted Expected Return (Normal) 3.28% $ (163,650) -7.93%
Risk-Adjusted Expected Return (Severe) 0.55%
Overall Sensitivity Factor: Portfolio of Stock Indices
0.209

Table A2: Fund manager coherent market portfolio [1] (case analysis of long and short trading positions)
Liquidation Market value Asset Allocation
Market Index period ( in days) in AED ($) per market (%)

DFM General Index 3.0 (3 000 000) 30

ADSM Index 3.0 8 000 000 80
BA All Share Index 4.0 6 000 000 60
KSE General Index 3.0 (2 000 000) 20
MSM30 Index 5.0 6 000 000 60
DSM20 Index 6.0 (1 000 000) 10
SE All Share Index 3.0 (4 000 000) 40

Correlation factor L-VaR ( Normal) L-VaR/Volume (%) Expected return

Empirical 364 382 3.6 0.027%

1.0 79 582 0.8 Sensitivity factor
0.0 446 245 4.5 0.215

& 2011 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 12, 6, 378–394 391
 Al Janabi

Table A3: Fund manager coherent market portfolio [2] (case analysis of long and short trading positions)
Liquidation Market value Asset allocation
Market Index period ( in days) in AED ($) per market (%)

DFM General Index 3.0 (2 000 000) 20

ADSM Index 3.0 2 000 000 20
BA All Share Index 4.0 6 000 000 60
KSE General Index 3.0 3 000 000 30
MSM30 Index 5.0 (2 000 000) 20
DSM20 Index 6.0 (3 000 000) 30
SE All Share Index 3.0 6 000 000 60

Correlation factor L-VaR ( Normal) L-VaR/Volume (%) Expected return

Empirical 388 386 3.9 0.045%

1.0 251 735 2.5 Sensitivity factor
0.0 400 015 4.0 0.469

Table A4: Fund manager coherent market portfolio [3] (case analysis of long and short trading positions)
Liquidation Market value Asset allocation
Market Index period ( in days) in AED ($) per market (%)

DFM General Index 3.0 3 000 000 30

ADSM Index 3.0 2 000 000 20
BA All Share Index 4.0 3 000 000 30
KSE General Index 3.0 (2 000 000) 20
MSM30 Index 5.0 1 000 000 10
DSM20 Index 6.0 (1 000 000) 10
SE All Share Index 3.0 4 000 000 40

Correlation factor L-VaR ( Normal) L-VaR/Volume (%) Expected return

Empirical 327 292 3.3 0.091%

1.0 431 111 4.3 Sensitivity factor
0.0 290 531 2.9 0.513

Table A5: Fund manager coherent market portfolio [4] (case analysis of long and short trading positions)
Liquidation Market value Asset allocation
Market Index period ( in days) in AED ($) per market (%)

DFM General Index 3.0 10 000 000 100

ADSM Index 3.0 (6 000 000) 60
BA All Share Index 4.0 6 000 000 60
KSE General Index 3.0 (2 000 000) 20
MSM30 Index 5.0 (6 000 000) 60
DSM20 Index 6.0 3 000 000 30
SE All Share Index 3.0 5 000 000 50

Correlation factor L-VaR ( Normal) L-VaR/Volume (%) Expected return

Empirical 592 173 5.9 0.132%

1.0 641 378 6.4 Sensitivity factor
0.0 677 918 6.8 0.746

392 & 2011 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 12, 6, 378–394
 Dynamic equity asset allocation

Table A6: Fund manager coherent market portfolio [1] (case analysis of long trading positions)
Liquidation Market value Asset allocation
Market Index period ( in days) in AED ($) per market (%)

DFM General Index 3.0 — 0

ADSM Index 3.0 — 0
BA All Share Index 4.0 1 000 000 10
KSE General Index 3.0 — 0
MSM30 Index 5.0 3 000 000 30
DSM20 Index 6.0 6 000 000 60
SE All Share Index 3.0 — 0

Correlation factor L-VaR ( Normal) L-VaR/Volume (%) Expected return

Empirical 326 080 3.3 0.052%

1.0 393 852 3.9 Sensitivity factor
0.0 316 525 3.2 0.149

Table A7: Fund manager coherent market portfolio [2] (case analysis of long trading positions)
Liquidation Market value Asset allocation
Market Index period ( in days) in AED ($) per market (%)

DFM General Index 3.0 2 000 000 20

ADSM Index 3.0 2 000 000 20
BA All Share Index 4.0 1 000 000 10
KSE General Index 3.0 1 000 000 10
MSM30 Index 5.0 1 000 000 10
DSM20 Index 6.0 1 000 000 10
SE All Share Index 3.0 2 000 000 20

Correlation factor L-VaR ( Normal) L-VaR/Volume (%) Expected return

Empirical 231 332 2.3 0.085%

1.0 401 063 4.0 Sensitivity factor
0.0 180 256 1.8 0.369

Table A8: Fund manager coherent market portfolio [3] (case analysis of long trading positions)
Liquidation Market value Asset allocation
Market Index period ( in days) in AED ($) per market (%)

DFM General Index 3.0 5 000 000 50

ADSM Index 3.0 2 000 000 20
BA All Share Index 4.0 — 0
KSE General Index 3.0 — 0
MSM30 Index 5.0 — 0
DSM20 Index 6.0 — 0
SE All Share Index 3.0 3 000 000 30

Correlation factor L-VaR ( Normal) L-VaR/Volume (%) Expected return

Empirical 372 645 3.7 0.118%

1.0 502 583 5.0 Sensitivity factor
0.0 319 158 3.2 0.550

& 2011 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 12, 6, 378–394 393
 Al Janabi

Table A9: Fund manager coherent market portfolio [4] (case analysis of long trading positions)
Liquidation Market value Asset allocation
Market Index period ( in days) in AED ($) per market (%)

DFM General Index 3.0 6 000 000 60

ADSM Index 3.0 — 0
BA All Share Index 4.0 — 0
KSE General Index 3.0 3 000 000 30
MSM30 Index 5.0 1 000 000 10
DSM20 Index 6.0 — 0
SE All Share Index 3.0 — 0

Correlation factor L-VaR ( Normal) L-VaR/Volume (%) Expected return

Empirical 334 077 3.3 0.134%

1.0 400 751 4.0 Sensitivity factor
0.0 324 028 3.2 0.319

1000000

900000

800000

700000
Portfolio [4]
L-VaR in AED

600000

500000
Portfolio [2]
Portfolio [1]
400000
Portfolio [3]
300000

200000
Efficient Portfolios 1-Day Liquidation Period
100000 Efficient Portfolios 10-Day Liquidation Horizon
Fund Manager Coherent Market Portfolios
0
0.00% 0.05% 0.10% 0.15% 0.20% 0.25% 0.30%
Expected Return

Figure A1: Efficient and coherent portfolios with liquidity-adjusted VaR (case of long and short trading positions).

900000

800000

700000

600000
L-VaR in AED

500000
Portfolio 3
400000 Portfolio 1 Portfolio 4

300000 Portfolio 2

200000
Efficient Portfolios 1-Day Liquidation Period
100000 Efficient Portfolios 10-Day Liquidation Period
Fund Manager Coherent Market Portfolios
0
0.03% 0.05% 0.07% 0.09% 0.11% 0.13% 0.15% 0.17% 0.19%
Expected Return

Figure A2: Efficient and coherent portfolios with liquidity-adjusted VaR (case of long trading positions).

394 & 2011 Macmillan Publishers Ltd. 1470-8272 Journal of Asset Management Vol. 12, 6, 378–394
View publication stats