Appendix 11A: Derivation of Equation (11.6a) (11.1) By Cheng Few Lee Joseph Finnerty John Lee Alice C Lee Donald Wort.

Slides:



Advertisements
Similar presentations
Solving Absolute-Value Equations
Advertisements

What two numbers will give you a product of 64 and a quotient of 4?
BALANCING 2 AIM: To solve equations with variables on both sides.
Factoring Trinomials When a=1 ALWAYS check for GCF first! Factor trinomials in the standard form ax²+ bx + c Solve equations in the standard form ax²+
Chapter 11 Performance-Measure Approaches for selecting Optimum Portfolios By Cheng Few Lee Joseph Finnerty John Lee Alice C Lee Donald Wort.
Differential Equations
Appendix 27A: An Alternative Method to Derive The Black-Scholes Option Pricing Model (Related to 27.5) By Cheng Few Lee Joseph Finnerty John Lee Alice.
SYSTEMS OF EQUATIONS.
Absolute-Value Equations and Inequalities
Solving Equations A Solution
Solving Fraction Equations by Multiplying
Appendix 13A. Alternative Specifications Of APT By Cheng Few Lee Joseph Finnerty John Lee Alice C Lee Donald Wort.
Solving Linear Equations in One Variable
Chapter 27 Itô’s Calculus: Derivation of the Black–Scholes Option Pricing Model By Cheng Few Lee Joseph Finnerty John Lee Alice C Lee Donald Wort.
Section 6.1 Rational Expressions.
1 VI. Sample Techniques for Computing Optimal Portfolios.
Chapter 18 Decision Tree and Microsoft Excel Approach for Option Pricing Model By Cheng Few Lee Joseph Finnerty John Lee Alice C Lee Donald Wort.
Copyright 2013, 2010, 2007, Pearson, Education, Inc. Section 7.2 Solving Systems of Linear Equations by the Substitution and Addition Method.
Part 2.  Review…  Solve the following system by elimination:  x + 2y = 1 5x – 4y = -23  (2)x + (2)2y = 2(1)  2x + 4y = 2 5x – 4y = -23  7x = -21.
Roots & Zeros of Polynomials I How the roots, solutions, zeros, x-intercepts and factors of a polynomial function are related.
Patterns and Sequences
Previously, we learned that adding two numbers together which have the same absolute value but are opposite in sign results in a value of zero. This can.
Identity and Equality Properties. Properties refer to rules that indicate a standard procedure or method to be followed. A proof is a demonstration of.
Warm Up #4 1. Evaluate –3x – 5y for x = –3 and y = 4. –11 ANSWER
GCSE Revision 101 Maths Quadratics © Daniel Holloway.
9.2 Solving Systems of Linear Equations by Addition BobsMathClass.Com Copyright © 2010 All Rights Reserved. 1 Step 1.Write both equations in the form Ax.
Inverses and Systems Section Warm – up:
Implicit Differentiation 3.6. Implicit Differentiation So far, all the equations and functions we looked at were all stated explicitly in terms of one.
1 Appendix 10 A: A Linear-Programming Approach to Portfolio- Analysis Models (Related to ) By Cheng Few Lee Joseph Finnerty John Lee Alice C Lee.
By Cheng Few Lee Joseph Finnerty John Lee Alice C Lee Donald Wort Appendix 3A: Logarithms and their Properties.
Appendix 8A Graphical Analysis in Markowitz Portfolio- Selection Model: Three- Security Empirical Solution By Cheng Few Lee Joseph Finnerty John Lee Alice.
4.8 – Solving Equations with Fractions
Dear Power point User, This power point will be best viewed as a slideshow. At the top of the page click on slideshow, then click from the beginning.
Dear Power point User, This power point will be best viewed as a slideshow. At the top of the page click on slideshow, then click from the beginning.
Appendix 2A. Simple regression and multiple regression By Cheng Few Lee Joseph Finnerty John Lee Alice C Lee Donald Wort.
 In this lesson we will go over how to solve a basic matrix equation such as the following: These are matrices, not variables.
Solving Quadratic Equations Using the Quadratic Formula Part 2.
Appendix 9A Empirical Evidence for the Risk-Return Relationship (Question 9) By Cheng Few Lee Joseph Finnerty John Lee Alice C Lee Donald Wort.
Solving Equations Containing First, we will look at solving these problems algebraically. Here is an example that we will do together using two different.
Slide Copyright © 2009 Pearson Education, Inc. 7.2 Solving Systems of Equations by the Substitution and Addition Methods.
The Substitution Method Objectives: To solve a system of equations by substituting for a variable.
Elimination using Multiplication Honors Math – Grade 8.
Adding two numbers together which have the same absolute value but are opposite in sign results in a value of zero. This same principle can be applied.
© 2010 Pearson Education Inc.Goldstein/Schneider/Lay/Asmar, CALCULUS AND ITS APPLICATIONS, 12e– Slide 1 of 33 Chapter 3 Techniques of Differentiation.
Solving 2 step equations. Two step equations have addition or subtraction and multiply or divide 3x + 1 = 10 3x + 1 = 10 4y + 2 = 10 4y + 2 = 10 2b +
Solving Systems of Equations The Elimination Method.
Modified Bessel Equations 홍성민. General Bessel Equation of order n: (1) The general solution of Eq.(1) Let’s consider the solutions of the 4.
Section 7.6 Functions Math in Our World. Learning Objectives  Identify functions.  Write functions in function notation.  Evaluate functions.  Find.
4-6 Solving Equations with Fractions What You’ll Learn ► To solve one – step equations ► To solve two – equations.
Solving Equations involving Fractions
Differential Equations
Solving Linear Equations in One Variable
Solving Systems of Equations
Section 3.7 Implicit Functions
Chapter 3 Techniques of Differentiation
Solving Systems of Equations
Identity and Equality Properties
9.6 Solving Rational Equations
Solving One-Step Equations
Solving Linear Equations in One Variable
Appendix 6A: Monthly Returns For The Wilshire 5000 Equity Indexes
Example 1 b and c Are Positive
Solving Systems of Equations
Solving Systems of Equations
Appendix 5A: WORKSHEETS FOR YIELD CURVES (Section 5.4.2) By
Linear Word problems.
Solving Systems of Equations by the Substitution and Addition Methods
Solving Systems of Equations
Tutorial 4 Techniques of Differentiation
Presentation transcript:

Appendix 11A: Derivation of Equation (11.6a) (11.1) By Cheng Few Lee Joseph Finnerty John Lee Alice C Lee Donald Wort

The objective function L as defined in Equation (18.3) can be rewritten: Then, following the product and chain rule, we have: (11A.1) 2

Multiplying Equation (11A.1) by,, and rearranging yields: (11A.2) Defining 3

Yields Therefore: 4

Define Hi=kWi, where the Wi are the fractions to invest in each security and the Hi are proportional to this fraction. Substituting Hi for Wi: There is one equation like this for each value of i. This is Equation (11.6a) in the text. 5

Appendix 11B: Derivation of Equation (11.2) By Cheng Few Lee Joseph Finnerty John Lee Alice C Lee Donald Wort

Following the optimization procedure for deriving Equation (18A.2) in Appendix 18A: Let and, solving for any, (11B.1) 7

Multiplying both sides of the equation by : ( 11B.2 ) Adding together the n equation of this form yields: ( 11B.3 ) By substituting Equation (11B.3) into Equation (11B.1), Equation (11.10) is obtained. 8

Appendix 11C: Derivation of Equation (11.15) (11.3) By Cheng Few Lee Joseph Finnerty John Lee Alice C Lee Donald Wort

(11C.1) where and for all i. The justification of this equation can be found in Elton et al. (1976). Assuming all stocks that would be in an optimal portfolio (called d) can be found, and then arranging these stocks as i=1, 2, ….., d, for the subpopulation of stocks that make up the optimal portfolio: (11C.2) This appendix discusses the use of performance measure to examine the optimal portfolio with short sales not allowed. Therefore, Equation (11B.1) from Appendix 11B must be modified 10

Multiplying both sides by, summing over all stocks in d, and rearranging yields: (11C.3) Notice since the set d contains all stocks with positive Hi: 11

Using (11C.4), the following equation for is obtained after substitution and rearranging from Equation (11C.1): And Let: Since, the inclusion of ; can only increase the value of Hi. Therefore, if Hi is positive with can never make it zero and the security should be included, If Hi<0 when, positive values of can increase Hi. However, because the product of and Hi must equal zero, as indicated in Equation (11C.1), positive values of imply Hi=0. Therefore, any security Hi<0 when must be rejected. Therefore, Equation (11.16) in the text can be used to estimate the optimal weight of a portfolio. (11C.4) (11C.5) 12

Appendix 11D: Quadratic Programming and Kuhn-Tucker Conditions By Cheng Few Lee Joseph Finnerty John Lee Alice C Lee Donald Wort

The quadratic programming algorithm is based on a technique from advanced calculus called Kuhn–Tucker conditions. This technique can solve the optimal portfolio with short sales not allowed. The problem can be stated as Subject to where To find the maximum value of F(W), there are two case of the optimal weights W=(W 1, W 2,…W n ) in Figure 11.D.1 14

If F(W) is a function of W i in Figure 11.D.1(a), then the maximum value of F(W) occurs at P, where at positive optimal weight (W i > 0). However, in Figure 11.D.1(b), the maximum feasible value of F(W) occurs at P ʹ instead of P because of short sell not allowed, then at optimal weight (W i = 0). (a) (b) F(W) Wi P P P’ Figure 11.D.1 Value of the Function F(W) as W i Changes 15

Therefore, in general, we can obtain the optimal weights under conditions We can rewrite the conditions as five Kuhn–Tucker conditions: When maximum feasible value occurs on, A i is positive and W i is equal to zero. If maximum feasible value occurs on, then A i is equal to zero and W i is positive. 16

In the following part, we will solve the optimal weights under Kuhn–Tucker conditions by Excel. Given an example in Figure 11.D.2: initial weights W = (0.1, 0.2, 0.7), can be calculated by the Equation (11A.2) in Appendix 11A and the value of the excess return and covariance matrix as below. Figure 11.D.2 Solver Function in Excel 17

Then we use “Solver” function in Excel to find the optimal weights W. Set target cell B5 equal to the value of 1 (the fifth Kuhn–Tucker condition), select the range from B2 to C4 as change cells, and select the first and second Kuhn– Tucker conditions as the constraints. Figure 11.D.2 Solver Function in Excel (Continued) 18

For the third and fourth Kuhn–Tucker conditions, press “Options” into Solver Options, select “Assume Non-Negative” and press “OK.” Then go back to Solver Parameters and press “Solve.”. Figure 11.D.2 Solver Function in Excel (Continued) 19

If first solve cannot find a solution, use “Solver” function again until the Solver find a solution as Figure11.D.3. Figure 11.D.3 Optimal Weights by Solver Function 20

Appendix 11E: Portfolio Optimization with Short-Selling Constraints By Cheng Few Lee Joseph Finnerty John Lee Alice C Lee Donald Wort

The traditional mean-variance efficient portfolios generate appealing characteristics. However, it is frequently questioned by financial professionals due to the propensity of corner solutions in portfolio weights. Therefore, adding portfolio weighting limits help portfolio managers to fashion realistic asset allocation strategies. 22

Suppose risky asset investments can be characterized as a vector of multivariate returns of N securities, R T. The expected risk premiums and variance-covariance of asset returns can be expressed as a vector  and a positive definite matrix V, respectively. Let Ω be the set of all real vectors w that define the weights of assets such that w T 1=1, where 1 is an N-vector of ones. The expected return of the portfolio is  p =w T  and the variance of the portfolio is  p = w T Vw. Considering all constraints given the objective to minimize the portfolio’s risk, the efficient frontier can be then expressed as a Lagrangian function: As described in the previous section, the optimal portfolio weights are a function of means, variances, and covariances of asset returns. (11E.1) 23

Consider a generalized case with N assets, let A=1 T V -1 R, B=R T V -1 R, X=1 T V -1 1, and  =BX-A 2, the solution of the above quadratic function is (see Pennacchi (2008)) w p = [(X  p – A)(V -1 R) + (B – A  p )(V -1 1)] The short-selling constraints are further considered in Markowitz model. The inequalities that represent non-negative portfolio weights are w 1  0, w 2  0, ……, w N  0, where w=[w 1, w 2,……, w N ]. The solution of this constrained diversification is to incorporate Equation (11E.3) in Equation (11E.1). (11E.2) (11E.3) 24

Considering a three-asset case exemplified in the previous section, we have the following objective and constraint functions: Replacing the variables in Equation (11E.4), we can rewrite Equation (11E.1) as where i=1, 2, 3, are Lagrange multipliers (LMs). A Numerical Example (11E.4) (11E.5) 25

The corresponding complementary slackness conditions are The above complementary slackness conditions indicate that either inequality constraints should be active at a local optimum or the corresponding Lagarange variable should equal zero. For differentiable nonlinear program, solutions W i, i=1, 2, 3, satisfy the Kuhn–Tucker conditions if they fulfill complementary slackness conditions, primal constraints, and gradient equation Any combination of W i, i = 1, 2, 3, for which there exist a corresponding LMs satisfying these conditions is called a Kuhn–Tucker point. (11E.6) (11E.7) 26

Our portfolio model has the following objective function gradient and those of the five linear constraints are Therefore the gradient equation part of Kuhn–Tucker conditions is (11E.8) (11E.9) (11E.10) 27

plus the primal constraints as part of the conditions and complementary slackness conditions in Equation (11E.6). (11E.11) 28

Notice that the weights are functions of five LMs and bounded by inequality constraints. Since the functions of solutions of W i are multidimensional, we cannot show their relation on a graph. One may start from any feasible point and then search an improving feasible direction chased by implementations of feasible and small steps of LMs. The stop of an improving feasible search not necessarily represents the current Kuhn–Tucker point is the global optimum but suggests it is a local optimum. Since there is no close-form solution function for each weight, one may need to continue the search until no improvement can be found. In our case, if there is an improvement in objective function. (11E.12) 29

We use the same data set to construct the nonconstrained (NC) efficient frontier and short-selling constrained (SS) efficient frontier, which is shown in Figure 11E.1. In Figure 11E.1, the SS optimal diversification is a subset of the NC portfolio. Figure 11E.1 Efficient Frontiers 30

The global minimum variance (MV) for the two kinds of portfolio is identical in this case. The information is listed in Table 11E.1. The conclusion of the same MV under different constraints does not always happen when the coefficients of correlation among securities and the relative magnitudes of return among assets change. Note that the optimal diversification strategies are sensitive to the variation in the first two moments of asset returns in the portfolio. E(R)(%)  Portfolio Weight w 1 (%) w 2 (%) w 3 (%) Table 11E.1 Minimum-Variance Portfolio 31

As the portfolios on efficient frontiers shown in Table 11E.2, assuming the long- term annual risk-free interest rate is 4%, the portfolios of corner solutions are less mean-variance efficient than the ones without negative and extremely positive weights. E(R) (%)  Sharpe Ratio Portfolio Weight w 1 (%) w 2 (%) w 3 (%) −25.82− −100.53− − Table 11E.2 Portfolios on Efficient Frontier 32