Decomposing the Impact of

Portfolio Constraints | August 2009

Jennifer Bender
Jyh-Huei Lee
Dan Stefek

Introduction
Portfolio managers may use an array of constraints when employing mean-variance optimization
for portfolio construction. These include the “long-only” constraint, turnover constraints, sector
constraints, size constraints, and beta constraints, to name a few. However, using constraints
may potentially prevent a manager from getting the most out of his return forecasts in portfolio
construction.

This paper analyzes the impact of constraints on portfolio return and risk, extending the insights
of previous research in this area, including Clark et al (2002), Grinold and Easton (1998), and
Scherer and Xu (2007). We show that constraints move a manager’s portfolio away from the
optimal unconstrained portfolio in two ways. First, they may rein in or increase the risk of the
portfolio without impairing its information ratio. Second, they may force the portfolio to take
unwanted bets that incur risk but yield no return.

As a result, a constrained portfolio consists of positions that are aligned with the manager’s
alphas and positions that are orthogonal to the alphas but are adopted to satisfy the constraints.
We illustrate how to measure the risk and return arising from each of these sources and how to
drill down to examine the contributions of individual constraints.

The Basic Framework

The basic framework follows Grinold and Easton (1998). In the presence of constraints, the
standard active optimization problem is:

λ
Maximize h′α − h′Σh
2 (1)
st Ah ≤ b

where h represents the vector of active portfolio holdings, α is a vector of the manager’s
forecast returns, Σ is the asset covariance matrix and λ is the aversion to risk. The matrix A
1
and vector b represent the asset bounds as well as any other linear constraints .

If there were no constraints, the solution to the optimization problem would be:

1
hU= Σ −1α (2)
λ

This portfolio achieves the maximum ex-ante information ratio, IRU , for the given alphas. In the
presence of constraints, the optimal solution to the problem in (1) becomes:

1
Note that our methods may be easily extended to include more general, convex constraints.

MSCI Barra Research

© 2009 MSCI Barra. All rights reserved. 1 of 9
Please refer to the disclaimer at the end of this document.
 Decomposing the Impact of Portfolio Constraints
| August 2009

1
hC = Σ −1 (α − A′π ) (3)
λ

where π is the vector of dual or shadow prices of the constraints. Each shadow price, πk ,
represents the change in utility per unit increase in the constraint bound, bk .

Substituting in Equation (2), we get:

1
hC = hU − Σ −1 A′π
λ (4)

Equation (4) shows that the optimal constrained portfolio hC is the difference between two
portfolios: the unconstrained portfolio hU that reflects the manager’s information, and the
1
constraint portfolio h X = Σ −1 A′π . This is depicted in Figure 1.
λ

Figure 1: Decomposing the Constrained Portfolio

1
hX = Σ −1 A′π
hC λ

hU

The constraint portfolio h X can be rewritten as a weighted sum of the k individual constraint
portfolios: hX = ∑ π k hX ,k . The k th constraint portfolio is:

1
hX ,k= Σ −1 Ak' (5)
λ

In equation (5), hX , k is the portfolio with the smallest risk per unit exposure to the constraint k.
Only binding constraints contribute to the constraint portfolio. If a constraint is not binding, the
shadow price of the constraint is zero and the corresponding constraint portfolio makes no
contribution to h X .

MSCI Barra Research

© 2009 MSCI Barra. All rights reserved. 2 of 9
Please refer to the disclaimer at the end of this document. RV0609
 Decomposing the Impact of Portfolio Constraints
| August 2009

A Closer Look

So far we have decomposed the constrained portfolio hC into two parts: the unconstrained
optimal portfolio hU and the constraint portfolio h X . We are interested in how constraints force a
manager to take bets that are not aligned with his alphas (i.e. with the unconstrained optimal
portfolios). We can see this by resolving the constraint portfolio into a part that is aligned with the
2
optimal unconstrained portfolio, hX , A , and a part that is orthogonal to it, hX ,O. , as shown in
Figure 2. We now have:

hC =hU − hX , A + hX ,O
(6)

To get the constrained solution we first subtract hX , A from the unconstrained optimal portfolio.
Since hX , A is aligned with hU , the result is a portfolio with a different risk and return but with the
same information ratio (IR) as the unconstrained optimal portfolio. We then must add hX ,O .
Since hX ,O is orthogonal to hU , adding it increases the risk of the portfolio but does not
change its return. Thus, hX ,O represents the unwanted bets forced on the manager by the
constraints.

Figure 2: The Impact of Constraints

hX
hC
h X ,O

hI hX , A

hU
Like Clarke et al. (2002), who show that the optimal constrained portfolio can be written as the
sum of two uncorrelated portfolios, we see that the constrained portfolio, hC ,is the sum of the
two components: hC= hI + hX ,O . The portfolio hI represents the manager’s information. The
portfolio hX ,O reflects the distortions or inefficiencies created by the use of constraints. The two
portfolios are calculated as follows: (see Appendix for details)

hI = β C ,U hU (7A)

2
By orthogonal, we mean uncorrelated: hX′ ,O ∑ hU =
0

MSCI Barra Research

© 2009 MSCI Barra. All rights reserved. 3 of 9
Please refer to the disclaimer at the end of this document. RV0609
 Decomposing the Impact of Portfolio Constraints
| August 2009

hX ,=
O hC − βC ,U hU (7B)

where β C ,U is the beta of the constrained portfolio to the unconstrained portfolio hU .

To better understand how each constraint contributes to the uncompensated risk, we can write
hX ,O as a sum of k individual constraint portfolios, hX ,O ,k ’s, which contribute risk without any
return:

=hX ,O ∑ 
k
β h − h  π
 
k ,U U X ,k k (8)
hX ,O ,k

where

hX ,k ′ΣhU Ak Σ −1α
β k ,U =
= for each constraint k
σ U2 α ′Σ −1α

This detailed decomposition allows us to isolate the contribution of each individual constraint to
the distortionary effect of the set of constraints.

Attributing Risk and Return

We can see how much ex-ante risk and return of an optimal portfolio comes from the manager’s
information and how much comes from the constraints by using the basic Litterman (1996)
decomposition. Active risk is decomposed as follows:

hI′Σhc hX′ ,0 Σhc

σc
= + (9)
σc

σc

 
information constraints

The overall constraint risk is the sum of the risks arising from each individual constraint k:

hX′ ,0 Σhc hX′ ,0,k Σhc

=∑ (10)
σc k σc

Turning to return, we see that all of the expected alpha comes from the manager’s information:
hI′α . In this decomposition, there is no return associated with the constraints.

MSCI Barra Research

© 2009 MSCI Barra. All rights reserved. 4 of 9
Please refer to the disclaimer at the end of this document. RV0609
 Decomposing the Impact of Portfolio Constraints
| August 2009

An Illustration
We illustrate the decomposition framework with an example. We start with an active portfolio
optimized as of March 2008 using the Barra Short-Term US Equity Model (USE3S). The MSCI
US Prime Market 750 Index is used as the universe and benchmark. We use the following
constraints:
• All stock weights must sum to 1 ("Budget Constraint")
• Each stock's weight must lie between 0 and 1 ("Long-Only Constraint")
• The active portfolio should have no exposure to the Barra Earnings Variability factor.

The optimized constrained portfolio hC has an active return of 1.62% and annualized risk of
3.69%. In contrast, the optimal unconstrained portfolio hU , has an active return of 4.99% and
annualized risk of 7.45%.

The decomposition in Table 1 isolates the two components of the constrained portfolio: hI ,which
is in line with the manager’s information, and h X ,O , which is not. Recall that the constrained
portfolio hC can be written as a sum of the two components: hC = hI + h X ,O . We see that more
of the active risk is coming from constraints than from the manager’s information!
3
Table 1: The Information Content of Constraints: How Much Risk Without Return

Decomposition of the Annualized Contribution

Constrained Portfolio Active Return to Active
Risk
1.62% 1.59%
Information Portfolio hI
0.00% 2.10%
Constraint Portfolio h X ,O
- Long-only 0.00% 1.91%

- Earnings Variability 0.00% 0.22%

- Budget 0.00% -0.03%

The table further shows the contributions of individual constraints to risk and return. Not
surprisingly, the Long-Only constraint has the largest impact with respect to non-compensated
risk. The Earnings Variability constraint has the next largest impact followed by the Budget
constraint.

3
Note that we combine the individual binding holdings constraints into the Long-Only Constraint category.

MSCI Barra Research

© 2009 MSCI Barra. All rights reserved. 5 of 9
Please refer to the disclaimer at the end of this document. RV0609
 Decomposing the Impact of Portfolio Constraints
| August 2009

Let’s consider another example where the portfolio has the same level of active risk as our
previous example, but we allow the optimizer to take up to a 0.3% short position in each asset.
Table 2 gives the results.

Table 2: Relaxing the Long-Only Constraint Reduces Its Contribution to Active Risk

Decomposition of the Annualized Contribution

Constrained Portfolio Active Return to Active
Risk
2.30% 3.19%
Information Portfolio hI
0.00% 0.50%
Constraint Portfolio h X ,O
- Long-Only 0.00% 0.19%

- Earnings Variability 0.00% 0.28%

- Budget 0.00% 0.04%

Compared to the first example, this loosening of the Long-Only Constraint reduces its contribution
to active risk significantly from 1.91% to 0.19% - below that of Earnings Variability. Most of the
active risk now comes from the Information Portfolio, reflecting the better alignment of the overall
portfolio’s risk with the manager’s information.

Conclusion
Portfolio constraints may prevent an asset manager from getting the most out of return forecasts
in portfolio construction. We show that the use of constraints in optimization has two effects: one
is to rein in (or increase) the risk of the portfolio without diminishing its IR, and the other is to
force the portfolio to take additional bets that incur risk but garner no return. As a result, the active
risk of the constrained portfolio comes from both positions that are aligned with the manager’s
information and positions that are orthogonal to the information but are taken to satisfy the
constraints. We describe how to measure the risk and return coming from each of these sources.
Although our focus has been on ex-ante risk and return, the analysis extends to ex-post return as
well.

References
Clark, Roger, Harindra de Silva, and Steven Thorley (2002), “Portfolio Constraints and the
Fundamental Law of Active Management,” Financial Analysts Journal, 2002.

Grinold, Richard and Kelly Easton (1998), “Attribution of Performance and Holdings,” in
Worldwide Asset and Liability Modeling, eds. W.T. Ziemba, John M. Mulvey, Isaac Newton.

Litterman, Robert, Hot Spots and Hedges, The Journal of Portfolio Management, pages 52-75,
December. 1996.

Scherer, Bernd and Xiadong Xu (2007), “The Impact of Constraints on Value-Added,” The
Journal of Portfolio Management, 33(4):45–54, 2007.

MSCI Barra Research

© 2009 MSCI Barra. All rights reserved. 6 of 9
Please refer to the disclaimer at the end of this document. RV0609
 Decomposing the Impact of Portfolio Constraints
| August 2009

Appendix
This appendix provides further details underlying the discussion in the section “The Information
Content of Constraints”. There we show how the constraints alter the risk of an optimized portfolio
and force it to take bets that are not aligned with the manager’s alphas. The full decomposition
with equations is shown below in Figure A1.

Figure A1: The Decomposition of Constraint Effects

πk

hc′Σhu
where βC ,U =
hu′ Σhu

We can resolve hX ,O into individual constraints portfolios as follows:

1
hX ,O = hC − βC ,U hU = hU − Σ −1 A′π − βC ,U hU
λ
(A1)
1
=∑ Σ  β k ,U α − A  π k =∑  β k ,U hU − hX ,k  π k
−1 '

λ k
k
k

where

hX ,k ′ΣhU α ′Σ −1 Ak′
β k ,U
= = for each constraint k
σ U2 α ′Σ −1α

MSCI Barra Research

© 2009 MSCI Barra. All rights reserved. 7 of 9
Please refer to the disclaimer at the end of this document. RV0609
 Decomposing the Impact of Portfolio Constraints
| August 2009

Contact Information
clientservice@mscibarra.com

Americas

Americas 1.888.588.4567 (toll free)

Atlanta + 1.404.551.3212
Boston + 1.617.532.0920
Chicago + 1.312.675.0545
Montreal + 1.514.847.7506
New York + 1.212.804.3901
San Francisco + 1.415.576.2323
Sao Paulo + 55.11.3706.1360
Stamford +1.203.325.5630
Toronto + 1.416.628.1007

Europe, Middle East & Africa

Amsterdam + 31.20.462.1382
Cape Town + 27.21.673.0100
Frankfurt + 49.69.133.859.00
Geneva + 41.22.817.9777
London + 44.20.7618.2222
Madrid + 34.91.700.7275
Milan + 39.02.5849.0415
Paris 0800.91.59.17 (toll free)
Zurich + 41.44.220.9300

Asia Pacific

China Netcom 10800.852.1032 (toll free)

China Telecom 10800.152.1032 (toll free)
Hong Kong + 852.2844.9333
Singapore 800.852.3749 (toll free)
Sydney 61.2.9033.9333
Tokyo + 81.3.5226.8222

www.mscibarra.com

MSCI Barra Research

© 2009 MSCI Barra. All rights reserved. 8 of 9
Please refer to the disclaimer at the end of this document. RV0609
 Decomposing the Impact of Portfolio Constraints
| August 2009

Notice and Disclaimer

 This document and all of the information contained in it, including without limitation all text, data, graphs, charts
(collectively, the “Information”) is the property of MSCl Inc. (“MSCI”), Barra, Inc. (“Barra”), or their affiliates (including
without limitation Financial Engineering Associates, Inc.) (alone or with one or more of them, “MSCI Barra”), or their
direct or indirect suppliers or any third party involved in the making or compiling of the Information (collectively, the
“MSCI Barra Parties”), as applicable, and is provided for informational purposes only. The Information may not be
reproduced or redisseminated in whole or in part without prior written permission from MSCI or Barra, as applicable.

 The Information may not be used to verify or correct other data, to create indices, risk models or analytics, or in
connection with issuing, offering, sponsoring, managing or marketing any securities, portfolios, financial products or
other investment vehicles based on, linked to, tracking or otherwise derived from any MSCI or Barra product or data.

 Historical data and analysis should not be taken as an indication or guarantee of any future performance, analysis,
forecast or prediction.

 None of the Information constitutes an offer to sell (or a solicitation of an offer to buy), or a promotion or
recommendation of, any security, financial product or other investment vehicle or any trading strategy, and none of
the MSCI Barra Parties endorses, approves or otherwise expresses any opinion regarding any issuer, securities,
financial products or instruments or trading strategies. None of the Information, MSCI Barra indices, models or other
products or services is intended to constitute investment advice or a recommendation to make (or refrain from
making) any kind of investment decision and may not be relied on as such.

 The user of the Information assumes the entire risk of any use it may make or permit to be made of the Information.

 NONE OF THE MSCI BARRA PARTIES MAKES ANY EXPRESS OR IMPLIED WARRANTIES OR
REPRESENTATIONS WITH RESPECT TO THE INFORMATION (OR THE RESULTS TO BE OBTAINED BY THE
USE THEREOF), AND TO THE MAXIMUM EXTENT PERMITTED BY LAW, MSCI AND BARRA, EACH ON THEIR
BEHALF AND ON THE BEHALF OF EACH MSCI BARRA PARTY, HEREBY EXPRESSLY DISCLAIMS ALL
IMPLIED WARRANTIES (INCLUDING, WITHOUT LIMITATION, ANY IMPLIED WARRANTIES OF ORIGINALITY,
ACCURACY, TIMELINESS, NON-INFRINGEMENT, COMPLETENESS, MERCHANTABILITY AND FITNESS FOR
A PARTICULAR PURPOSE) WITH RESPECT TO ANY OF THE INFORMATION.

 Without limiting any of the foregoing and to the maximum extent permitted by law, in no event shall any of the MSCI
Barra Parties have any liability regarding any of the Information for any direct, indirect, special, punitive,
consequential (including lost profits) or any other damages even if notified of the possibility of such damages. The
foregoing shall not exclude or limit any liability that may not by applicable law be excluded or limited, including
without limitation (as applicable), any liability for death or personal injury to the extent that such injury results from the
negligence or wilful default of itself, its servants, agents or sub-contractors.

 Any use of or access to products, services or information of MSCI or Barra or their subsidiaries requires a license
from MSCI or Barra, or their subsidiaries, as applicable. MSCI, Barra, MSCI Barra, EAFE, Aegis, Cosmos,
BarraOne, and all other MSCI and Barra product names are the trademarks, registered trademarks, or service marks
of MSCI, Barra or their affiliates, in the United States and other jurisdictions. The Global Industry Classification
Standard (GICS) was developed by and is the exclusive property of MSCI and Standard & Poor’s. “Global Industry
Classification Standard (GICS)” is a service mark of MSCI and Standard & Poor’s.

© 2009 MSCI Barra. All rights reserved.

About MSCI Barra

MSCI Barra is a leading provider of investment decision support tools to investment institutions worldwide. MSCI Barra
products include indices and portfolio risk and performance analytics for use in managing equity, fixed income and
multi-asset class portfolios.
The company’s flagship products are the MSCI International Equity Indices, which include over 120,000 indices calculated
daily across more than 70 countries, and the Barra risk models and portfolio analytics, which cover 56 equity and 46 fixed
income markets. MSCI Barra is headquartered in New York, with research and commercial offices around the world.

MSCI Barra Research

© 2009 MSCI Barra. All rights reserved. 9 of 9
Please refer to the disclaimer at the end of this document. RV0609