19 views

Uploaded by api-3737025

- Prediction Error Methods for Time-domain Blind Identification of Multichannel FIR Filters
- MS221-201010
- Problems 2
- Teori Akuntansi Kel 2
- Method of Least squar with example -hsmallik.docx
- Final Exam With Results
- Method of Least squar with example -hsmallik.docx
- 152_2009WT2_2
- C2050_PDF_C11.pdf
- Crankshaft Analysis
- 431-hw6-key
- UT Dallas Syllabus for math2333.0u1.08u taught by (xxx)
- senior project paper
- Algorithm for Designing Free-Form Imaging Opti
- Differential Equations
- Cambridge A Level Further Mathematics Syllabus code 9231
- algebra_seminar
- The Variety of Integrable Killing Tensors on the 3-Sphere - Schobel
- AISC K Factors
- Analyisis of 3d Truss

You are on page 1of 7

Midterm Exam

Answer Key

Instructor: Norman Schürho¤

You have just …nished your master’s studies at HEC with success. Not surprisingly, you

have been hired as an investment advisor for OBS (Ordinary Bank of Switzerland). Your

…rst client is a wealthy Saudi-Arabian who wants you to do some asset allocation. The asset

classes he is considering are bonds, stocks, and real estate. You have done some preliminary

analysis and determined that the payo¤s of the three types of securities in three possible

scenarios are 2 3

2 1 1

A=4 2 3 2 5;

2 2 3

where each column is a di¤erent asset, and each row is a di¤erent scenario.

Your goal is to determine the portfolio allocation in each asset class, denoted 2 R3 , given

the client’s target payo¤ b 2 R3 in each scenario. In other words, you want to determine

such that A = b. To make sure that this is feasible, you determine the following:

a) Calculate jAj.

b) Is A invertible? Justify brie‡y.

c) Calculate A 1 .

d) What is the optimal portfolio given b = (b1 ; b2 ; b3 )0 ?

Answer to Question 1

a) det(A) = 10:

b) Yes, since det(A) = 10 6= 0.

c) 0 1

1 1 1

2 2 2

A 1

=@ 1

5

2

5

1

5

A

1 3 4

5 5 5

d) A = b ) = A 1b )

1 1 1

1 = b1 + b2 b3

2 2 2

1 2 1

2 = b1 b2 + b3

5 5 5

1 3 4

3 = b1 b2 + b3

5 5 5

1

Question 2 (10 points)

For each of the following questions, either provide a function and show brie‡y that it satis…es

the required conditions, or show why you cannot …nd such a function:

a) Can you provide an example of a function f : R 7! R where f is continuous, yet f

fails to attain either a maximum or a minimum?

b) Can you provide an example of a function f : X 7! R where X is neither closed nor

bounded, f is discontinuous, yet f attains both a maximum and a minimum?

c) Can you provide an example of a function f : X R 7! R where X is nonempty,

closed and bounded, f is continuous, yet f fails to attain either a maximum or a minimum?

Answer to Question 2

a) Yes, take for example f (x) = x3 . f (x) is continuous, is de…ned in R, but attains neither

its maximum (+1) nor its minimum ( 1).

b) Yes, take for example X = R++ and

1 if x is rational,

f (x) =

0 otherwise.

f (x) is discontinuous everywhere, yet it attains its maximum 1 and its minimum 0, and

it is de…ned over the open and unbounded interval (0; 1).

c) No. A continuous function f de…ned over a closed and bounded set X R attains, by

Weierstrass’Theorem, both its maximum and its minimum.

2

Question 3 (10 points)

Determine whether the following statements are true or false (by “false” I mean not neces-

sarily true). In each case, brie‡y explain your answer and provide some intuition:

a) If a Kuhn-Tucker multiplier is positive, then the corresponding constraint is binding.

b) If a Kuhn-Tucker multiplier is zero, then the corresponding constraint is non-binding.

c) If a constraint is non-binding, then the corresponding Kuhn-Tucker multiplier is zero.

d) If a constraint is binding, then the corresponding Kuhn-Tucker multiplier is positive.

Answer to Question 3

a) True. By the complementary slackness conditions: j (gj (x) bj ) = 0; 8j = 1; :::; m. If

we have j > 0, then gj (x) bj = 0 must be true.

b) False. The constraint could be “just”binding. In this case the marginal value of relaxing

the constraint is zero, i.e., j = 0. The complementary slackness conditions still hold in this

case, since both j = 0 and gj (x) bj = 0.

c) True. By the complementary slackness conditions: j (gj (x) bj ) = 0; 8j = 1; :::; m. If

gj (x) bj 6= 0; then j = 0 must be true.

d) False. The constraint could be “just”binding. In this case the marginal value of relaxing

the constraint is zero, i.e., j = 0. The complementary slackness conditions still hold in this

case, since both j = 0 and gj (x) bj = 0.

3

Question 4 (15 points)

This is your …rst day as controller of the new biotech company VALGEN located in Bex,

Valais. Your team has already done a lot of work on optimizing production, but they are

not very fresh on optimization techniques. Luckily, you are because you have just taken a

course on mathematics at HEC.

The team of controllers has determined that given the …rm’s production function Y =

f (K; L) = K 1=2 L1=2 with K =Capital, L =Labor, the …rm’s cost function is C(Y ; r; w) =

minK;L frK + wL subject to K 1=2 L1=2 = Y g. Note that r is the cost of capital that the chief

…nancial o¢ cer has provided you, and w is the company’s hourly wage rate including social

charges that the accounting department has provided. Hence, the minimum cost depend on

the output level Y and the factor prices r and w.

The chief …nancial o¢ cer (CFO) of VALGEN has asked you to provide him with the

following information:

a) Find the inputs K and L that minimize cost.

b) What are the minimum cost for Y = r = w = 1?

To make sure that the …rm has enough cash to pay all bills once demand rises, the CFO

now asks you to do some sensitivity analysis. In short, the CFO wants to know by how

much the costs change as output rises:

@C

c) Find @Y (1; 1; 1).

The CFO is skeptical that the accounting department has done a good job in coming

up with the factor prices. He comes back to you and asks for some more analysis. In

particular, he asks the following:

d) Show that (with Y = 1) K (r; w) and L (r; w), the input demand functions, exist for

(r; w) near (1; 1). Otherwise, there would be a problem in hiring su¢ cient workers.

e) How do the cost depend on the factor prices? That is, …nd @C@r

(1; 1; 1) and @C

@w

(1; 1; 1).

Answer to Question 4

The minimum cost function (i.e., the value function of the cost minimization problem) is

de…ned by

C(w; r; Y ) = min wL + rK : KL = Y 2 :

K;L

a) Solving for L(w; r; Y ) and K(w; r; Y ), using Lagrange’s method, one obtains

1=2

rY 2

L(w; r; Y ) = ;

w

1=2

wY 2

K(w; r; Y ) = :

r

b) Replace these into the cost function to get the minimum cost function,

C(w; r; Y ) = 2(wr)1=2 Y:

C(1; 1; 1) = 2(1)1=2 1 = 2:

4

@C @C

c) First, obtain @Y

= CY (w; r; Y ) = 2(wr)1=2 , and evaluate it at (1; 1; 1) to get@Y

(1; 1; 1) = 2:

1=2 1=2

d) The input demand functions are L (w; r; 1) = wr , K (w; r; 1) = w

r

. Let (w; r) =

(1; 1), then L h(1; 1; 1) = K (w;

i r; 1) = 1. We can rewrite these functions as f : R2++ ! R2++ ,

1=2 1=2

f : (w; r) ! wr ; wr . Note that the function f is well de…ned for positive factor

prices, r > 0; w > 0, and it is continuous and di¤erentiable in a neighborhood of (1; 1).

1=2 1=2

e) Obtain @C @r

= wr Y > 0 and @C@w

= wr Y > 0. Note that the minimum cost

function is increasing in both factor prices. Evaluate it at (1; 1; 1) to get @C

@r

(1; 1; 1) = 1,

@C

@w

(1; 1; 1) = 1.

5

Question 5 (15 points)

You have just become a fund manager at the new CIBERIUS hedge fund in Geneva. Your

managing director asks you to come up with a novel investment strategy that exploits return

predictability in the stock market of several developing countries. You remember from your

…nance classes at HEC that stock returns can have di¤erent volatility. That means, they

are heteroskedastic. So you consider the following generalized least squares problem. Let

y=X +

where

E( ) = 0;

E( 0 ) = 2 ;

covariance matrix 2 (i.e., is a positive de…nite symmetric matrix).

The problem is you cannot …nd your econometrics book which would tell you what the

generalized least squares estimator of is. To solve the problem of …nding an estimator for

yourself, you take three steps. First, you de…ne the standardized error term

P :

a) Apply a Matrix Decomposition Theorem to …nd a matrix P such that

0 2

E( )= I:

That means you try to …nd a matrix P such that is homoskedastic (= equal variance).

(Hint: Remember from Econometrics that the expectations operator is linear.)

b) Now use P to derive a least-squares estimator of . (Hint: You have derived a

least-squares estimator of in the exercise sessions of your mathematics course at HEC.)

Answer to Question 5

We have

0

E( ) = E(P (P )0 )

= E(P 0 P 0 )

= P E( 0 )P 0

= 2P P 0:

Let the spectral decomposition of be = C C 0 , with

0 1

1 0

B .. C ;

C= c1 c2 cn , = @ ... ...

. A

0 n

6

where fci gi=1;:::;n: are the eigenvectors and f i gi=1;:::;n: are the eigenvalues.

De…ne

1

0

P 2C :

1 1 1 1 1 1

Then P P 0 = C 0 ( 2 C 0 )0 =

2 2C

0

C 2 = 2C 0C C 0C 2 = I since C 0 C = I.

1 1

Also, P 0 P = ( 0

2C )

0 0

2C = C

1

C = ((C 0 )

0 1

C 1 1

) = (C C 0 ) 1

= 1

.

Now,

y = X +

, Py = PX + P

, y =X +

with y P y, X P X, P and E( 0 ) = 2 I.

This is now a standard OLS problem, for which we already know the solution

= (X 0 X ) 1 X 0 y

= ((P X)0 P X) 1 (P X)0 P y

= (X 0 P 0 P X) 1 X 0 P 0 P y

= (X 0 1 X) 1 X 0 1 y:

- Prediction Error Methods for Time-domain Blind Identification of Multichannel FIR FiltersUploaded byKiran Kumar
- MS221-201010Uploaded bySean Calvert
- Problems 2Uploaded byEric Parker
- Teori Akuntansi Kel 2Uploaded bysri
- Method of Least squar with example -hsmallik.docxUploaded bySagar Rawal
- Final Exam With ResultsUploaded byБатсүх Отгонцэцэг
- Method of Least squar with example -hsmallik.docxUploaded bySagar Rawal
- 152_2009WT2_2Uploaded bySteven Shin
- C2050_PDF_C11.pdfUploaded byBalaji GU
- Crankshaft AnalysisUploaded bypathakshashank
- 431-hw6-keyUploaded bysalim
- UT Dallas Syllabus for math2333.0u1.08u taught by (xxx)Uploaded byUT Dallas Provost's Technology Group
- senior project paperUploaded byapi-240620289
- Algorithm for Designing Free-Form Imaging OptiUploaded byvuliencn
- Differential EquationsUploaded bySelva Kumar
- Cambridge A Level Further Mathematics Syllabus code 9231Uploaded byTasnia Tushi
- algebra_seminarUploaded byLorin Atzberger
- The Variety of Integrable Killing Tensors on the 3-Sphere - SchobelUploaded byJama Hana
- AISC K FactorsUploaded bycklcon
- Analyisis of 3d TrussUploaded bysarsaro10
- 2245chap2.pdfUploaded byRoy Vesey
- Chapter 3 Kelompok 1Uploaded byFelliciaIkko
- AHPUploaded byNajihah Zakaria
- Jonathan Tennyson, Steven Miller and C. Ruth Le Sueur- TRIATOM: programs for the calculation of ro-vibrational spectra of triatomic moleculesUploaded byPassamm
- Chapter 2 - StrainUploaded byRick
- Eigenvalues and EigenvectorsUploaded bystevenspillkumar
- imf.pdfUploaded byGiovanni Gozzini
- Syllabus f.mathsUploaded byMuneeb Zafar
- JournalUploaded byCesar Latorre
- Second Order Differential Equations With CoeficientUploaded byDhany SSat

- Ch18 Modèles d'équations SimultanéesUploaded byapi-3737025
- Stata Introduction to StataUploaded byapi-3737025
- Ch12 Interprétation Des Tests Orientés régressionUploaded byapi-3737025
- Ch13 Les Tests d'HypothèsesUploaded byapi-3737025
- Learning Eco No Metrics With GaussUploaded byapi-3737025
- Ch11 Tests Basés Sur La régression de Gauss-NewtonUploaded byapi-3737025
- Ch20 Racines Unitaires Et CointégrationUploaded byapi-3737025
- Ch17 La méthode Des Moments généraliséeUploaded byapi-3737025
- Regression With StataUploaded byapi-3737025
- Practice2 SolutionsUploaded byapi-3737025
- Ch21 Les Expériences de Monte CarloUploaded byapi-3737025
- Ch16 itUploaded byapi-3737025
- Ch19 Modèles de régression Pour Données ChronologiquesUploaded byapi-3737025
- Stata TutorialUploaded byapi-3737025
- Ch15 Variables dépendantes Limitées Et ivesUploaded byapi-3737025
- Ch5 Méthodes Asymptotiques Et Moindres Carrés Non LinéairesUploaded byapi-3737025
- Ch7 Variables Instrument AlesUploaded byapi-3737025
- A Lea to Ire 4Uploaded byapi-3737025
- A Lea to Ire 5Uploaded byapi-3737025
- Ch2 Modèles de régressions Non LinéairesUploaded byapi-3737025
- Ch6 Régression Gauss-NewtonUploaded byapi-3737025
- annexe BUploaded byapi-3737025
- Ch10 AutocorrélationUploaded byapi-3737025
- Ch8 Maximum de VraisemblanceUploaded byapi-3737025
- Ch3 Inférence Dans Les Modèles de régression Non LinéairesUploaded byapi-3737025
- Ch4 Méthodes Et Théories AsymptotiquesUploaded byapi-3737025
- Ch9 Maximum de Vraissemblance Et Moindres Carrés généralisésUploaded byapi-3737025
- Ch1 géométries Moindres CarrésUploaded byapi-3737025
- annexe AUploaded byapi-3737025
- A Lea to Ire 67Uploaded byapi-3737025

- hw5soln_06.pdfUploaded bysubash1111@gmail.com
- Edexcel C2 MS Jan 2011Uploaded byIssam Saif
- Roads to Infinity - The Mathematics of Truth and ProofUploaded byslark1
- Dagger Theory.pdfUploaded byWilliam Campbell
- 0906.2185Higher order fractional derivativesUploaded bySARA
- tfyyUploaded byLincoln Chau
- k-bonacciUploaded byopenid_AePkLAJc
- C1.6 Sequences and SeriesUploaded byAnuradha Damale
- Math53 WorksheetUploaded byAndrew Berger
- Trial TERM 2 MATHEMATICS T 2013Uploaded byZuraini Arshad
- Group TheoryUploaded byMaria Mitrofan
- KLTUploaded byVineeth Kumar
- 3 Kontrak Lth Math Thn 3 4 5 6Uploaded byX - LevelOfficial
- Jan-2017_4HRUploaded byVidula Education
- Paul C. Bressloff- Metastable states and quasicycles in a stochastic Wilson-Cowan model of neuronal population dynamicsUploaded byNeerFam
- md_135_7_071008Uploaded byhermez19
- Logical ReasoningUploaded byVishali Raj
- Metro StateUploaded byangelsong27
- Pauls Online Notes _ Differ..Uploaded byFawwad Qureshi
- Errores de GamsUploaded byXimenaBarrera
- Ken D. Olum and Delia Schwartz-Perlov- Anthropic prediction in a large toy landscapeUploaded byPollmqc
- Multiplicative InverseUploaded bydearbhupi
- RudjitoPhDUploaded bymaucoi
- Midas Gen Analysis ReferenceUploaded byNayeemuddin Khaja
- Monte Carlo Simulation Formula in ExcelUploaded byAna Marks
- ConEd Sustainable Tomorrow Systems Thinking GuidebookUploaded byIrfan Fachri M
- pow 3Uploaded byapi-202876241
- The Research Competence Development of Students Trained In Mathematical Direction.pdfUploaded byYosua Ricky
- NCERT Solutions Class 6 Mathematics MensurationUploaded byNCERT CBSE Study Material
- Why I Think Andy Beal is a Bully by Denise SubramaniamUploaded byDenise Subramaniam