Modern Portfolio Optimization:

A Practical Approach Using an

Excel Solver Single-Index Model
JEFF GROVER AND ANGELINE M. LAVIN
It is illegal to make unauthorized copies of this article, forward to an unauthorized user or to post electronically without Publisher permission.
The Journal of Wealth Management 2007.10.1:60-72. Downloaded from www.iijournals.com by NEW YORK UNIVERSITY on 07/21/15.

JEFF GROVER odern Portfolio Theory (MPT) tions to this rule-based system, in conjunction
is an associate professor at
Indiana Wesleyan Univer-
sity in Marion, IN.
[email protected]

ANGELINE M. LAVIN
M is based on the idea that com-
binations of assets have the
potential to provide better
returns with less risk than individual assets.
While MPT provides excellent insights into
with Sharpe’s methodology and the Sharpe
Ratio, can be simply implemented using the
Microsoft Excel Solver function. This process
provides an excellent example of how the Excel
Solver can be used to solve a mathematical
is an associate professor at which assets should be included in an investor’s optimization problem, which is especially ben-
Beacom School of Business, optimal portfolio, understanding the under- eficial to investors as they face the challenge of
University of South Dakota
in Vermillion, SD.
lying statistical techniques in portfolio opti- determining how to efficiently optimize their
[email protected] mization presents a rigorous challenge. This is mutual fund portfolios. This article uses TIAA-
especially true in determining the standard CREF mutual fund data as an example to illus-
deviation of a portfolio using the Markowitz trate the MPT concept as a practical approach
methodology. The investor’s knowledge of the to optimizing an investment portfolio to obtain
theoretical basis of MPT presents a concep- maximum return with minimal risk. However,
tual constraint on his/her ability to understand the model can be applied to any set of mutual
the underlying constructs of MPT, which is fund options to create an optimal fund alloca-
required to determine the asset investment tion. The process outlined in this article may
strategy that will optimize the portfolio. also prove useful for defined contribution plan
One solution to overcoming such investor providers as they work to implement the pro-
constraints is to provide a rule-based hierar- visions outlined in the Pension Protection Act
chal system as a primer to establishing an equiv- passed by the U.S. Congress in 2006. The Act
alent linear form of the Markowitz [1959] calls for greater access to professional advice
quadratic objective function which, when opti- about retirement for investors and gives workers
mized, affords an optimal solution for an invest- greater control over how to invest their retire-
ment portfolio. Due to the complexities of ment account funds. The model presented in
calculating the variance-covariance matrix and this article has the potential to provide useful
the computer memory resources required to information to professional advisors as well as
solve this quadratic objective function, Sharpe individual investors.
[1972] devised a single-index model that nicely In 1952, Markowitz first proposed a
replicates the Markowitz function. The single- method by which an investor could optimize
index model allows for the computation of betas, the individual components of a portfolio to
expected returns, and residual variances with a generate portfolio returns greater than the
market index return and variance. The solu- sum of the returns on the individual portfolio

60 MODERN PORTFOLIO OPTIMIZATION: A PRACTICAL APPROACH USING AN EXCEL SOLVER SINGLE-INDEX MODEL SUMMER 2007
 components. In 1959 he wrote a textbook on efficient that the co-movement of assets with each other is more
diversification of investments. In a review of this text- important than the individual security characteristics when
book, Tintner [1959] pointed out that Markowitz’s key forming portfolios of assets. According to Elton and
concept is the efficient portfolio. “If a portfolio is efficient, Gruber [1997], who provide an excellent overview of
it is impossible to obtain a greater average return without portfolio theory, the M-V framework has remained the
incurring greater standard deviation; it is impossible to cornerstone of MPT.
obtain smaller standard deviation without giving up return The development of M-V portfolio theory created
on average” Markowitz [1959, p. 22]. This optimization a need to estimate model inputs such as correlation coef-
process is a rare economic phenomenon that exists only ficients, and index models became the principal tool for
It is illegal to make unauthorized copies of this article, forward to an unauthorized user or to post electronically without Publisher permission.
The Journal of Wealth Management 2007.10.1:60-72. Downloaded from www.iijournals.com by NEW YORK UNIVERSITY on 07/21/15.

when a portfolio is established in return-risk space so that estimating covariance. Sharpe [1967] popularized the
the optimal mix of the securities results in optimal port- market model, a variant of the single-index model, for
folio returns. Use of the Markowitz methodology, which evaluating the risk and return characteristics of portfolios.
is quadratic in nature, requires an understanding of oper- Two of the most basic questions in portfolio analysis
ations research methodologies. Sharpe [1972] introduced are the number and specific types of securities to include
a single-index model that overcame the difficulties asso- in the portfolio. Mao [1970] was the first to use theoret-
ciated with implementing the quadratic algorithm. The ical analysis, in the form of a heuristic switching algo-
latter methodology suggests a linear framework that pro- rithm, to solve this optimization problem. However, Mao’s
vides a more simplistic method for the practical imple- procedure could not be used for negative or zero beta
mentation of portfolio optimization using the Microsoft securities, it did not determine the optimal allocation
Excel Solver. This method can be used to demonstrate the among the various securities, and it would potentially
optimization process for mutual fund investors and enable require enormous amounts of computer time. Sharpe
them to take control of their future wealth accumulation. [1972] proposed a slight modification to Mao’s method-
ology that would allow for negative betas and for the
Motivation selection of securities as well as the allocation of funds
among them. Sharpe and Stone [1973] extended Sharpe’s
A search of current software available to perform study to include non-market risk, which was a limitation
portfolio optimizations returns a limited number of cost of Sharpe’s single-index model, and captured the essence
effective alternatives. The method developed in this article of portfolio selection as a linear program.
yields a user-friendly mechanism with which investors can Elton, Gruber, and Padberg [1976] set out to oper-
perform the portfolio optimization process using Microsoft ationalize MPT, which although meant to be a practical
Solver in its Excel application. The intent is to have the tool, had primarily developed as a normative, theoretical
investor download mutual fund daily closing prices, the construct. They suggested that the implementation dif-
13-week U.S. Treasury Bill annual percentage returns, and ficulties lay in: 1) estimating the correlation matrices,
a market index return, and then have the Excel applica- 2) the time and costs in generating efficient quadratic
tion calculate the optimal portfolio combination of these portfolios, and 3) educating portfolio managers to relate
funds. The goal of this article is to provide a rigorous to risk-return tradeoffs expressed as covariances and stan-
approach to the theoretical application of this optimiza- dard deviations. They continued to expand the body of
tion process using techniques proposed by Sharpe [1972]. knowledge and proposed a constant correlation model.
Bawa, Elton, and Gruber [1979] extended their study and
Literature Review introduced simple procedures for multivariate portfolio
optimization. Chen [1983] challenged the Bawa et al.
The decision of how to allocate capital across dif- [1979] study’s limitation of standard error estimation tech-
ferent asset classes is a key issue that all investors face. niques, but Alexander [1985] later refuted these claims
Markowitz is widely considered to be the father of MPT. and supported the Elton et al. [1976] study. Kwan [1984]
His early work was the first to frame portfolio choice as further extended MPT by including the three existing
a mean-variance (M-V) optimization problem that yields portfolio optimization techniques—the single, multiple
an efficient frontier of potential investment choices for a and correlation index models—in a single common algo-
rational investor. His critical insight was the realization rithm. Green and Burton [1992] offered a solution to

SUMMER 2007 THE JOURNAL OF WEALTH MANAGEMENT 61

 extreme weightings caused by a dominant single factor portfolio of funds inside of a defined contribution retire-
in equity returns, thus reducing residual risks. ment plan. In the context of a tax-deferred or tax-exempt
Both the theoretical construct of portfolio opti- portfolio, there are no tax implications associated with
mization and its practical application have continued to portfolio rebalancing, and periodic asset allocation changes
evolve since Markowitz first proposed the M-V frame- can be made without creating taxable gains. The model
work in the 1950s. While ever more sophisticated methods developed in this article can be used to optimize any
are continually developed, it is important to recognize that mutual fund portfolio, regardless of whether the portfolio
simplicity may be the key to assisting individual investors is held in a tax-deferred, tax-exempt or taxable account.
who are faced with asset allocation decisions in their It is important to note that rebalancing a taxable fund
It is illegal to make unauthorized copies of this article, forward to an unauthorized user or to post electronically without Publisher permission.
The Journal of Wealth Management 2007.10.1:60-72. Downloaded from www.iijournals.com by NEW YORK UNIVERSITY on 07/21/15.

defined contribution retirement plans. In fact, a recent portfolio may create tax consequences, which can either
study by Iyengar, Huberman, and Jiang [2004] quoted on be positive or negative. Despite the fact that rebalancing
the TIAA-CREF website offers evidence that the com- a taxable fund portfolio has the potential to create taxable
plexity of the asset allocation decision often leads employees gains, periodic portfolio balancing is still recommended.
to delay savings plan enrollment. Rugh [2003] reports that In fact, studies have shown that periodic rebalancing can
the majority of TIAA-CREF participants direct at least actually improve portfolio returns by forcing investors to
half of their contributions to equity and make no changes sell assets that have appreciated in value and buy assets
in their allocations over the ten to twelve year periods of that have underperformed. Most financial advisors rec-
time studied. He also notes that the average participant ommend that clients review their portfolios at least annu-
allocation has become more diverse over the past ten years. ally and rebalance as needed upon review. This is especially
true when one considers the effects of inflation on the
Data portfolio. Reilly and Brown [2006] suggest that inflation
is the major cause of reductions in investment value.
The TIAA-CREF defined contribution variable
annuity retirement plan funds were selected for this analysis Simple Techniques for Determining an
because the company is a well-known provider of low- Optimal Portfolio
cost mutual funds as well as a major provider of defined
contribution retirement plans. In fact, Rugh [2003] reports The portfolio optimization challenge, as suggested
that TIAA-CREF is the largest pension provider in the by Mao [1970], is to maximize an objective function such
U.S., managing $300 billion in total assets for more than that an investor can create an optimal portfolio providing
two million individuals. The plans, which are provided the maximum return for the minimum risk. Mao sug-
through the employer of the respective investor, are gested such an algorithm, but it was limited by the inability
referred to as retirement (or group retirement) annuity to solve for a negative beta. Although negative betas appear
contracts. Investment amounts are dependent on con- less frequently than positive ones, they do exist, and a
tractual agreements between the employee and employer. mathematical resolution is required to efficiently opti-
Typically, contributions are made on a tax-deferred basis, mize a given sample of securities. The proper selection
which means that pre-tax dollars fund the accounts, but of the market index is critical to minimizing the number
the employees pay taxes on withdrawals. Information of securities with negative betas. Sharpe [1972] resolved
about the funds can be obtained from the TIAA-CREF this issue using his single-index methodology. The process
website, http://tiaa-cref.org/product_profiles/ras.html. developed in this article applies Sharpe’s methodology to
The purpose of this article is to demonstrate the concepts identify undervalued and overvalued securities using the
of MPT in portfolio optimization using this TIAA-CREF capital asset pricing model (CAPM) as the screening tool
fund data. However, the model developed in this article for the portfolio optimization process. This study will
can be applied to optimize any mutual fund portfolio. consider only undervalued securities as input variables for
portfolio optimization. After screening out overvalued
Tax Implications of Portfolio Rebalancing securities, the Sharpe Ratio is operationalized as an objec-
tive function and the portfolio returns and standard devi-
The specific example employed in this article assumes ations are used to establish an optimal or tangent portfolio
that the fund portfolio being optimized is a tax-deferred following the Sharpe [1972] CAPM methodology. We

62 MODERN PORTFOLIO OPTIMIZATION: A PRACTICAL APPROACH USING AN EXCEL SOLVER SINGLE-INDEX MODEL SUMMER 2007
 compute this objective function by maximizing the Sharpe Equation (4) computes a portfolio standard devia-
Ratio. This tangent or optimal portfolio, i.e., the port- tion using the Sharpe method. This function minimizes
folio with minimum risk and maximum return, will be the square of the sum of products of the security betas
a linear combination of the portfolios along the trace of and their respective weights summed with the security
the efficient frontier of portfolio combinations. We define residual variances.
the Sharpe Ratio in Equation (1) as:
1/2
 N

σ P =  βP2σ m2 + ∑ X m2 σ ei2  (4)
( RP − R f )  i =2 
(1)
σP
It is illegal to make unauthorized copies of this article, forward to an unauthorized user or to post electronically without Publisher permission.
The Journal of Wealth Management 2007.10.1:60-72. Downloaded from www.iijournals.com by NEW YORK UNIVERSITY on 07/21/15.

where σ 2ei is computed as σ i − βi σ M , is the residual vari-

2 2 2
The Sharpe Ratio, which measures the excess return
ance of the portfolio.
per unit of total risk, is the return of the portfolio minus
Substituting in the classical Sharpe Ratio, as pre-
the risk-free rate divided by the standard deviation of the –
sented in Equation (2), for R P and σP , we present our
portfolio. We calculate the tangent portfolio using the
modified Sharpe Ratio as:
Sharpe Ratio as the maximization objective function,
given in Equation (2), which takes the form:
α P + βP Rm – R f
θ= (5)
(β σ )
1/2
(RP − R f )
2
P
2
m
+ ΣiN= 2 X m2 σ im2
θ = (2)
Maximize σP
where Rf is the risk-free rate of return as determined by
The constraints for this function include the the 13-week U.S. Treasury-bill.
following: Since the goal is to maximize return while mini-
mizing risk, we use the Sharpe method for computing
n returns and standard deviation to determine the global
1. ∑ xi = 1, for i = 1, … n. minimum variance portfolio. We operationalize the Sharpe
1
Ratio presented in Equation (2) with the maximization
2. xi ≥ 0, for i = 1, … n. function presented in Equation (5).

The first constraint requires the sum of all weighted THE SINGLE-INDEX MODEL
coefficients to be equal to one and the second requires
these weights to be greater than or equal to zero because While many investors will be most interested in the
short-selling of securities is not allowed. outcome of the optimization process described in this
Equation (3) gives the portfolio return using the section, we suspect that some will be interested in the
Sharpe method (Elton, Gruber, Brown, and Goetzmann actual process and how it is achieved using the Microsoft
[2003]). The portfolio return is the sum of products of the Excel Solver. Therefore, this article both reports the out-
security alphas and their respective weights, plus the sum come and details the process.
of products of the security betas and their respective The method presented and demonstrated in this
weights with the product of the market index return. article gives the optimal procedure for selecting a port-
folio when the single-index model is accepted as the best
way to forecast the covariance structure of returns. The
R P = α P + βP R m (3) input variables for the objective function are the calcu-
lated betas, residual variances, and alphas for the indi-
where the security weights are designated by Xi, the vidual securities and the market index. The optimization
portfolio alpha is computed as αP = ΣNi=1Xi αi, the port- process, which utilizes five years of monthly return data
folio beta is computed as βP = ΣNi=1Xi βi, and the average from December 2000 to December 2005, is outlined in
–
return on the market index is designated as R m. steps below so that it can be replicated by interested

SUMMER 2007 THE JOURNAL OF WEALTH MANAGEMENT 63

 – –
investors using readily available mutual fund, market index, a. Compute alphas as αi = Ri – βi Rm, where Rm is the
and risk-free rate data. average return on the market index and βi and Ri
The optimization process uses an eight-step method, are the beta and average return on the ith fund.
which is conducted as follows: b. Compute residual variances as σ 2ei = σ 2i – βi2σ 2m,
Step 1: Compute natural log returns of a market which is the variance of the fund minus the product
index and the selected set of mutual funds (either TIAA- of the beta squared of the individual fund and the
CREF funds or another group of mutual funds) using variance of the market index.
end-of-month closing prices, and normalize the returns
of the 13-week U.S. Treasury-bill into monthly returns. Step 5: Set up the Microsoft Excel Solver by:
It is illegal to make unauthorized copies of this article, forward to an unauthorized user or to post electronically without Publisher permission.
The Journal of Wealth Management 2007.10.1:60-72. Downloaded from www.iijournals.com by NEW YORK UNIVERSITY on 07/21/15.

Step 2: Compute security averages, variances, covari-

ances, and betas as follows:
a. Establishing the objective function to maximize the
ΣiN=1Ri modified Sharpe Ratio as presented in Equation (5).
a. Compute averages as N b. Equally weighting the starting values for Xi and
σ im
b. Compute Betas as βi = σ m2 , computed as the covari- establishing the following initial constraints:
ance of security i returns with the returns of a market
index divided by the population variance of the i. The weights are greater than or equal to zero.
market index. ii. The sum of the starting values of Xi is equal to one.
Σ N ( R − R )2
c. Compute the market variance as σ M2 = m =1 Nm m
d. Compute the covariances of the individual mutual Step 6: Compute the optimal or tangent portfolio
ΣiN=1, m =1 ( Ri – Ri ) ( Rm – Rm )
funds as σ im = N by using Tangent Estimates, Forward Derivatives, and
Newton Search for the Solver Options, and the default
Step 3: Use the CAPM methodology to compute values for Max Time, Iterations, Tolerance, and Conver-
excess returns, identify individual fund valuations and gence in the Solver dialog box.
retain those that are undervalued. Funds that are over- Step 7: Establish the portfolio return, beta, and stan-
valued should not enter into the optimization process dard deviation as presented in Equations (3) and (4).
unless short-selling is allowed. Since most mutual funds Step 8: Optimize the mutual fund portfolio by max-
do not allow short-selling due to legal constraints, this imizing Equation (5) and determine the optimum or tan-
study does not include this option. We will evaluate this gent portfolio by using the computed weights of the
process by computing the CAPM required or expected undervalued mutual funds.
return, the excess returns, and the valuation of the indi-
vidual mutual funds as follows: THE OPTIMAL TIAA-CREF PORTFOLIO
– PRACTICUM IN EXCEL
a. Compute the CAPM as RCAPM = Rf + (Rm – Rf )
βi, where Rf is derived from 60 months of returns This section reports the specific findings developed
using the 13-week U.S. Treasury-bill. The market using the TIAA-CREF defined contribution variable
index average return is derived from the average of annuity retirement plan funds as a practicum in evalu-
60 monthly returns, and βi is derived from Step 2b. ating the eight-step optimization process outlined in the
b. Compute excess returns, (Ri – RCAPM), where Ri is previous section. End of month closing prices of the
computed as in Step 2a. TIAA-CREF funds and the Russell 3000 market index,
c. Determine individual fund valuations as either under, RUA, from December 2000 through December 2005,
over, or fairly valued using the CAPM methodology. are used in the optimization. The TIAA-CREF closing
Thus, if the excess return is greater than zero, the prices were downloaded from www.tiaa-cref.org, the
fund is selected for inclusion in the optimal port- RUA prices from http://finance.yahoo.com, and the U.S.
folio mix. Treasury-bill returns from www.treasurydirect.gov.
Step 1: The monthly natural log returns of RUA
Step 4: Compute the individual mutual fund alphas and TIAA-CREF funds were computed and reported in
and residual variances as follows: Exhibit 1.

64 MODERN PORTFOLIO OPTIMIZATION: A PRACTICAL APPROACH USING AN EXCEL SOLVER SINGLE-INDEX MODEL SUMMER 2007
 The Journal of Wealth Management 2007.10.1:60-72. Downloaded from www.iijournals.com by NEW YORK UNIVERSITY on 07/21/15.
It is illegal to make unauthorized copies of this article, forward to an unauthorized user or to post electronically without Publisher permission.

SUMMER 2007
EXHIBIT 1
Monthly Percentage Returns Computed from Monthly Fund Closing Prices

THE JOURNAL OF WEALTH MANAGEMENT

65
 EXHIBIT 1 (Continued)
It is illegal to make unauthorized copies of this article, forward to an unauthorized user or to post electronically without Publisher permission.
The Journal of Wealth Management 2007.10.1:60-72. Downloaded from www.iijournals.com by NEW YORK UNIVERSITY on 07/21/15.

Notes:
1) CREF IL-Bond is an inflation linked bond fund.
2) ^RUA is the Russell 3000 Index.
3) These returns do not include distributed dividends.
4) End of month closing price security data for the variable annuities were obtained from http://www.tiaacref.org/product_profiles/ras.html and the end of
month closing fund data for ^RUA from http://finance.yahoo.com.

Step 2: Fund averages, variances, covariances, and ii. The sum of the starting values of Xi is equal
betas were computed and are reported in Exhibit 2. to one.
Step 3: Fund valuations were identified using the
CAPM, and undervalued funds were retained and are These values are reported in Exhibit 3.
reported in Exhibit 2. Step 6: The Solver was then loaded with the fol-
Step 4: The TIAA-CREF fund alphas and residual lowing inputs:
variances were computed and are reported in Exhibit 3.
Step 5: The Excel Solver is set up as follows: a. “Set Target Cell:” Input the Tangent or Sharpe
Ratio in Cell F17.
a. The objective function is established to maximize the b. “Equal to: ‘Max’”
modified Sharpe Ratio presented in Equation 5. c. “By Changing Cells” Input Cell Range E6:E12.
b. The starting values for Xi are equally weighted, and d. “Subject to the Constraints:”
the following initial constraints were established:
i. Set Cell E15 = 1. This sums the individual
i. The weights are set to values greater than or equal weights to 1 or 100%, satisfying the first con-
to zero. straint to Equation (2).

66 MODERN PORTFOLIO OPTIMIZATION: A PRACTICAL APPROACH USING AN EXCEL SOLVER SINGLE-INDEX MODEL SUMMER 2007
 EXHIBIT 2
Monthly Data to Determine Undervaluation or Overvaluation of Funds
It is illegal to make unauthorized copies of this article, forward to an unauthorized user or to post electronically without Publisher permission.
The Journal of Wealth Management 2007.10.1:60-72. Downloaded from www.iijournals.com by NEW YORK UNIVERSITY on 07/21/15.

Notes:
1) RF = 0.188%, RM = –0.018%, σ 2M= 24.251. The monthly average risk-free rate was computed by dividing each monthly U.S. Treasury-bill annual
investment rate percentage by 12 to convert the APR to monthly returns.
2) The Russell 3000 Index (^RUA) was used as the market index fund. This is the index that TIAA-CREF uses to evaluate and benchmark their funds.
3) CREF IL-Bond is an inflation linked bond fund.

ii. Set Range E6:E12 >= 0, which satisfies the of the undervalued TIAA-CREF funds. These values are
second constraint to Equation (2). reported in Exhibit 4.
iii.Used Tangent Estimates, Forward Derivatives,
and Newton Search for the Solver Options EXCEL SOLVER PRACTICUM RESULTS
and the default values for Max Time, Itera-
tions, Tolerance, and Convergence in the The proportional investment percentages of the
Solver dialog box. selected undervalued funds in the tangent portfolio using
the hybrid Sharpe [1972] approach outlined in this article
Note: These weights would need to be adjusted suggests that the TIAA-CREF investor should purchase
if future undervalued securities change. These 81.484% of the TIAA Real Estate fund, 11.658% of the
values are reported in Exhibit 3. CREF Bond fund, and 6.859 % of the CREF Inflation-
Linked Bond fund. Even though the CREF Social Choice,
Step 7: The portfolio return, beta, and standard Stock, Global, and Equity funds were undervalued
deviation as presented in Equations (3) and (4) were com- according to the CAPM results, the funds did not opti-
puted and are reported in Exhibit 4. mize due to minimal residual variances. The suggested
Step 8: The TIAA-CREF portfolio was optimized allocation of funds optimizes the return and minimizes
by maximizing Equation (5) and determining the the risk of a retirement portfolio of TIAA-CREF funds.
optimum or tangent portfolio using the computed weights This allocation strategy is the outcome of the historical

SUMMER 2007 THE JOURNAL OF WEALTH MANAGEMENT 67

 EXHIBIT 3
Solver Optimization Results
It is illegal to make unauthorized copies of this article, forward to an unauthorized user or to post electronically without Publisher permission.
The Journal of Wealth Management 2007.10.1:60-72. Downloaded from www.iijournals.com by NEW YORK UNIVERSITY on 07/21/15.

Notes:
1) This exhibit reports the results of the Microsoft Solver optimization calculation which includes the nine TIAA-CREF funds together with the information
required to execute the Solver application, the valuation results (overvalued or undervalued) according to the Capital Asset Pricing Model, and the Sharpe Ratio,
which was maximized to obtain the tangent or efficient portfolio.
2) Only undervalued funds were used for the optimization process.
3) Beta∗ was computed as Fund i’s beta times its weight, Residual Variance∗ as the residual variance times its weight, and Alpha∗ as Alpha times its weight.
4) Undervalued funds are those with an actual 60-month average return that exceeded the expected CAPM return.

data period chosen for this analysis. The real estate market ally employed the optimal mix of the tangent portfolio
demonstrated outstanding performance during the equity in the TIAA Real Estate fund, the CREF Inflation-Linked
bear market of 2000–2002, and this performance drove Bond fund, and the CREF Bond fund as suggested, would
the allocation model, which was based on the historical have enjoyed excellent returns from December 2000 to
return data from December 2000 to December 2005. December 2005. However, we caution the reader that
Indeed, a TIAA-CREF investor who might have actu- past performance is not necessarily indicative of future

68 MODERN PORTFOLIO OPTIMIZATION: A PRACTICAL APPROACH USING AN EXCEL SOLVER SINGLE-INDEX MODEL SUMMER 2007
 The Journal of Wealth Management 2007.10.1:60-72. Downloaded from www.iijournals.com by NEW YORK UNIVERSITY on 07/21/15.
It is illegal to make unauthorized copies of this article, forward to an unauthorized user or to post electronically without Publisher permission.

SUMMER 2007
EXHIBIT 4
Efficient Frontier Results

THE JOURNAL OF WEALTH MANAGEMENT

69
 E X H I B I T 4 (Continued)
It is illegal to make unauthorized copies of this article, forward to an unauthorized user or to post electronically without Publisher permission.
The Journal of Wealth Management 2007.10.1:60-72. Downloaded from www.iijournals.com by NEW YORK UNIVERSITY on 07/21/15.

Note: This exhibit reports the results of the Microsoft Solver maximization optimization process including the minimum variance portfolio, the maximum return
portfolio, and the Tangent or Efficient Portfolio, designated as∗.

results. Indeed, an 82% real estate allocation going for- portfolios. The number 30 was arbitrarily selected to
ward may not prove to be optimal. It is important to keep capture the complete maximum-minimum portfolio
in mind that a critical assumption in this work is that the return range. The weights of successive tangent portfo-
variance-covariance structure of these securities over the lios at each increment along the frontier were recorded
past five years will continue for the next five years. In prac- along with the respective Sharpe Ratios, returns, and
tice, that assumption will be violated. Therefore, investors standard deviations. We recorded these iterative steps in
should re-optimize their portfolios quarterly, at a min- Exhibit 4 and then identified the portfolio with the max-
imum. Periodic rebalancing is feasible within the TIAA- imum Sharpe Ratio. This trace method identified the
CREF fund family because there is no charge to the portfolio with the maximum Sharpe Ratio as having the
investor for rebalancing within the plan. same weights as the portfolio identified by the Solver
As a check on the Solver model, we validated that process, thus validating that the computed weights cor-
the tangent portfolio computed using the Excel Solver rectly represent the optimal portfolio. This validation
is, in fact, the optimal portfolio from the efficient fron- provides confidence that subsequent Solver calculations
tier that includes all possible combinations of tangent will produce the optimal portfolio combination given a
portfolios. We computed the two endpoints of the effi- specified set of mutual funds.
cient frontier as the minimum variance and maximum The efficient frontier, with the optimal portfolio
return portfolios. To evaluate the efficient frontier across identified for visual reference, is shown in Exhibit 5. From
the return range, we increased the minimum portfolio an investing point of view, the beauty of this Excel Solver
variance in relationship to the difference between the program is that it allows investors to easily optimize their
maximum and minimum portfolio variances and divided portfolio on a monthly or quarterly basis, if desired, by
this difference by 30, the number of possible efficient following the simple steps described and illustrated above.

70 MODERN PORTFOLIO OPTIMIZATION: A PRACTICAL APPROACH USING AN EXCEL SOLVER SINGLE-INDEX MODEL SUMMER 2007
 EXHIBIT 5
Efficient Frontier Graph
It is illegal to make unauthorized copies of this article, forward to an unauthorized user or to post electronically without Publisher permission.
The Journal of Wealth Management 2007.10.1:60-72. Downloaded from www.iijournals.com by NEW YORK UNIVERSITY on 07/21/15.

Note: The efficient frontier is developed using the monthly average natural log returns and standard deviations. The arrow points to the optimal or tangent port-
folio, which is the portfolio with the maximum return and minimum risk.

CONCLUSION loading current monthly mutual fund, index fund, and

Treasury-bill raw data into Excel, and implementing the
This article details a user-friendly Excel optimiza- M-V efficient portfolio optimization process explained
tion tool that mutual funds investors can use to maxi- in Sections 2 and 3 of this article.
mize their retirement wealth. Use of this tool iteratively The spreadsheet model developed in this analysis
to periodically rebalance a mutual fund portfolio has the provides an optimal mutual fund mix for this portfolio
potential to enhance retirement wealth. The ability to based on the historical data for the mutual funds in the
optimize on a regular basis may be especially beneficial tax-deferred TIAA-CREF defined contribution variable
in a national economy that is questioning the viability of annuity retirement plan. The optimization process can be
future allocations of social security to retirees. This article applied to any set of mutual funds, whether the funds are
provides a straightforward computational method that uti- held in tax-advantage or taxable accounts. However, rebal-
lizes the Microsoft Excel Solver to solve a classic port- ancing of taxable fund portfolios must take into account
folio optimization problem. The optimization is outlined tax consequences created by the rebalancing process. The
in detail for ease of replication. Implementation of the process outlined in this article is a useful and easy tool for
model is relatively straight-forward and requires down- investors to use in periodically optimizing their retirement

SUMMER 2007 THE JOURNAL OF WEALTH MANAGEMENT 71

 portfolio. A copy of the Excel Solver Model used in this Kwan, C. “Portfolio Analysis Using Single-Index, Multi-Index,
article is available for use upon request. and Constant Correlation Models: A Unified Treatment.” Journal
of Finance, Vol. 39, No. 5 (1984), pp. 1469–1484.
REFERENCES
Mao, C. “Essentials of Portfolio Diversification Strategy.” Journal
Alexander, G., and B. Resnick. “More on Estimation Risk and of Finance, Vol. V, No. 5 (1970), pp. 1109–1121.
Simple Rules for Optimal Portfolio Selection.” Journal of Finance,
Vol. 40, No. 1 (1985), pp. 125–134. Markowitz, H. “Portfolio Selection.” Journal of Finance, (March
1952), pp. 77–91.
It is illegal to make unauthorized copies of this article, forward to an unauthorized user or to post electronically without Publisher permission.

Bawa, V., E. Elton, and M. Gruber. “Simple Rules for Optimal

The Journal of Wealth Management 2007.10.1:60-72. Downloaded from www.iijournals.com by NEW YORK UNIVERSITY on 07/21/15.

Portfolio Selection in a Stable Paretian Market.” Journal of Finance, ——. Portfolio Selection: Efficient Diversification of Investments.
Vol. 34, No. 4 (1979), pp. 1041–1048. New York: John Wiley & Sons, Inc., 1959.

Chen, S., and S. Brown. “Estimation Risk and Simple Rules Reilly, B., and K. Brown. Investment Analysis and Portfolio Man-
for Optimal Portfolio Selection.” Journal of Finance, Vol. 38, agement, 8th ed. U.S.A.: Thomson South-Western, 2006.
No. 4 (1983), pp. 1087–1094.
Rugh, J. “Participant Contributions and Asset Allocations
Elton, E., and M. Gruber. “Modern Portfolio Theory, 1950 to During the Recent Bull and Bear Market: Evidence from
Date.” Journal of Banking and Finance, Vol. 21, No. 11/12 (1997), TIAA-CREF.” Benefits Quarterly, 2003, pp. 55–67.
pp. 1743–1759.
Sharpe, W. “A Simplified Model of Portfolio Analysis.” Man-
Elton, E., M. Gruber, S. Brown, and W. Goetzmann. Modern agement Science, Vol. 13 (1967), pp. 277–293.
Portfolio Theory and Investment Analysis. New York: John Wiley
& Son, Inc., 2003. ——. “Simple Strategies for Portfolio Diversification: Com-
ment.” Journal of Finance, Vol. VII, No. 1 (1972), pp. 127–129.
Elton, E., M. Gruber, and M. Padberg. “Simple Criteria for
Optimal Portfolio Selection.” Journal of Finance, Vol. XI, No. ——. “Simple Strategies for Portfolio Diversification: Com-
5 (1976), pp. 1341–1357. ment, A Correction.” Journal of Finance, Vol. VII, No. 3 (1972),
pp. 773.
——. “Simple Rules for Optimal Portfolio Selection: The
Multi Group Case.” Journal of Financial and Quantitative Analysis, Sharpe, W. and B. Stone. “A Linear Programming Formulation
Vol. XII, No. 3 (1977), pp. 329–345. of the General Portfolio Selection Model.” Journal of Financial
and Quantitative Analysis, Vol. VIII, No. 4 (1973), pp. 621–636.
Green, R., and H. Burton. “When Will Mean-Variance Effi-
cient Portfolios Be Well Diversified?” Journal of Finance, Vol. 47, Tintner, G. “Review of the Book Portfolio Selection: Efficient
No. 5 (1992), pp. 1784. Diversification of Investments.” Journal of Finance Book Reviews,
1959, pp. 447.
http://finance.yahoo.com, retrieved on October 23, 2006.
Treasury Direct Data, http://www.treasurydirect.gov, retrieved
http://www.tiaa-cref.org/product_profiles/ras.html, retrieved on January 18, 2006.
on January 24, 2006.
U.S. Congress. The Pension Protection Act of 2006: Ensuring
http://www.treasurydirect.com, retrieved on October 23, 2006. Greater Retirement Security for American Workers.

Iyengar, S., G. Huberman, and W. Jiang. “How Much Choice

is Too Much: Contributions to 401(k) Retirement Plans.” In To order reprints of this article, please contact Dewey Palmieri at
O. Mitchel and S. Utkus, eds., Pension Design and Structure: New [email protected] or 212-224-3675
Lessons from Behavioral Finance. Oxford, UK: Oxford Univer-
sity Press, 2004, pp. 83–96.

72 MODERN PORTFOLIO OPTIMIZATION: A PRACTICAL APPROACH USING AN EXCEL SOLVER SINGLE-INDEX MODEL SUMMER 2007