Technical Appendix (Not for publication) Generic and Brand Advertising Strategies in a Dynamic Duopoly

Σχετικά έγγραφα
α & β spatial orbitals in

Πανεπιστήµιο Κρήτης - Τµήµα Επιστήµης Υπολογιστών. ΗΥ-570: Στατιστική Επεξεργασία Σήµατος. ιδάσκων : Α. Μουχτάρης. εύτερη Σειρά Ασκήσεων.

One and two particle density matrices for single determinant HF wavefunctions. (1) = φ 2. )β(1) ( ) ) + β(1)β * β. (1)ρ RHF

(a,b) Let s review the general definitions of trig functions first. (See back cover of your book) sin θ = b/r cos θ = a/r tan θ = b/a, a 0

The following are appendices A, B1 and B2 of our paper, Integrated Process Modeling

Analytical Expression for Hessian

Variance of Trait in an Inbred Population. Variance of Trait in an Inbred Population

Example 1: THE ELECTRIC DIPOLE

Econ 2110: Fall 2008 Suggested Solutions to Problem Set 8 questions or comments to Dan Fetter 1

A Class of Orthohomological Triangles

Phys460.nb Solution for the t-dependent Schrodinger s equation How did we find the solution? (not required)

Multi-dimensional Central Limit Theorem

CHAPTER 25 SOLVING EQUATIONS BY ITERATIVE METHODS

Finite Field Problems: Solutions

Laplace s Equation in Spherical Polar Coördinates

HOMEWORK 4 = G. In order to plot the stress versus the stretch we define a normalized stretch:

2 Composition. Invertible Mappings

Some Theorems on Multiple. A-Function Transform

Matrix Hartree-Fock Equations for a Closed Shell System

Multi-dimensional Central Limit Theorem

Exercise, May 23, 2016: Inflation stabilization with noisy data 1

Every set of first-order formulas is equivalent to an independent set

The Simply Typed Lambda Calculus

EE512: Error Control Coding

Μηχανική Μάθηση Hypothesis Testing

Section 8.3 Trigonometric Equations

Solution Set #2

Section 7.6 Double and Half Angle Formulas

MATH 38061/MATH48061/MATH68061: MULTIVARIATE STATISTICS Solutions to Problems on Matrix Algebra

1 Complete Set of Grassmann States

Other Test Constructions: Likelihood Ratio & Bayes Tests

8.324 Relativistic Quantum Field Theory II

3.4 SUM AND DIFFERENCE FORMULAS. NOTE: cos(α+β) cos α + cos β cos(α-β) cos α -cos β

Homework 8 Model Solution Section

Matrices and Determinants

8.1 The Nature of Heteroskedasticity 8.2 Using the Least Squares Estimator 8.3 The Generalized Least Squares Estimator 8.

Ordinal Arithmetic: Addition, Multiplication, Exponentiation and Limit

ANSWERSHEET (TOPIC = DIFFERENTIAL CALCULUS) COLLECTION #2. h 0 h h 0 h h 0 ( ) g k = g 0 + g 1 + g g 2009 =?

Exercises 10. Find a fundamental matrix of the given system of equations. Also find the fundamental matrix Φ(t) satisfying Φ(0) = I. 1.

1 3D Helmholtz Equation

SOLUTIONS TO MATH38181 EXTREME VALUES AND FINANCIAL RISK EXAM

Fractional Colorings and Zykov Products of graphs

Areas and Lengths in Polar Coordinates

e t e r Cylindrical and Spherical Coordinate Representation of grad, div, curl and 2

Lecture 2: Dirac notation and a review of linear algebra Read Sakurai chapter 1, Baym chatper 3

ω ω ω ω ω ω+2 ω ω+2 + ω ω ω ω+2 + ω ω+1 ω ω+2 2 ω ω ω ω ω ω ω ω+1 ω ω2 ω ω2 + ω ω ω2 + ω ω ω ω2 + ω ω+1 ω ω2 + ω ω+1 + ω ω ω ω2 + ω

derivation of the Laplacian from rectangular to spherical coordinates

DESIGN OF MACHINERY SOLUTION MANUAL h in h 4 0.

Practice Exam 2. Conceptual Questions. 1. State a Basic identity and then verify it. (a) Identity: Solution: One identity is csc(θ) = 1

ST5224: Advanced Statistical Theory II

SOLUTIONS TO MATH38181 EXTREME VALUES AND FINANCIAL RISK EXAM

2. THEORY OF EQUATIONS. PREVIOUS EAMCET Bits.

Inverse trigonometric functions & General Solution of Trigonometric Equations

Approximation of distance between locations on earth given by latitude and longitude

Statistical Inference I Locally most powerful tests

Areas and Lengths in Polar Coordinates

Example Sheet 3 Solutions

Section 9.2 Polar Equations and Graphs

Constant Elasticity of Substitution in Applied General Equilibrium

4.2 Differential Equations in Polar Coordinates

George S. A. Shaker ECE477 Understanding Reflections in Media. Reflection in Media

ECE 308 SIGNALS AND SYSTEMS FALL 2017 Answers to selected problems on prior years examinations

TMA4115 Matematikk 3

Symplecticity of the Störmer-Verlet algorithm for coupling between the shallow water equations and horizontal vehicle motion

Math221: HW# 1 solutions

Solutions to Exercise Sheet 5

Concrete Mathematics Exercises from 30 September 2016

A Note on Intuitionistic Fuzzy. Equivalence Relation

8. ΕΠΕΞΕΡΓΑΣΊΑ ΣΗΜΆΤΩΝ. ICA: συναρτήσεις κόστους & εφαρμογές

Generalized Fibonacci-Like Polynomial and its. Determinantal Identities

Saigo-Maeda Fractional Differential Operators of the Multivariable H-Function

Second Order Partial Differential Equations

On a four-dimensional hyperbolic manifold with finite volume

Jesse Maassen and Mark Lundstrom Purdue University November 25, 2013

Tutorial Note - Week 09 - Solution

Oscillating dipole system Suppose we have two small spheres separated by a distance s. The charge on one sphere changes with time and is described by

Uniform Convergence of Fourier Series Michael Taylor

Instruction Execution Times

The Laplacian in Spherical Polar Coordinates

Durbin-Levinson recursive method

Galatia SIL Keyboard Information

The challenges of non-stable predicates

4.6 Autoregressive Moving Average Model ARMA(1,1)

b. Use the parametrization from (a) to compute the area of S a as S a ds. Be sure to substitute for ds!

w o = R 1 p. (1) R = p =. = 1

Nowhere-zero flows Let be a digraph, Abelian group. A Γ-circulation in is a mapping : such that, where, and : tail in X, head in

1 String with massive end-points

( y) Partial Differential Equations

Neutralino contributions to Dark Matter, LHC and future Linear Collider searches

ΗΥ537: Έλεγχος Πόρων και Επίδοση σε Ευρυζωνικά Δίκτυα,

ORDINAL ARITHMETIC JULIAN J. SCHLÖDER

An Inventory of Continuous Distributions

Fundamental Equations of Fluid Mechanics

Homework 3 Solutions

Vidyamandir Classes. Solutions to Revision Test Series - 2/ ACEG / IITJEE (Mathematics) = 2 centre = r. a

Forced Pendulum Numerical approach

CHAPTER 48 APPLICATIONS OF MATRICES AND DETERMINANTS

PARTIAL NOTES for 6.1 Trigonometric Identities

ΚΥΠΡΙΑΚΗ ΕΤΑΙΡΕΙΑ ΠΛΗΡΟΦΟΡΙΚΗΣ CYPRUS COMPUTER SOCIETY ΠΑΓΚΥΠΡΙΟΣ ΜΑΘΗΤΙΚΟΣ ΔΙΑΓΩΝΙΣΜΟΣ ΠΛΗΡΟΦΟΡΙΚΗΣ 6/5/2006

Problem Set 9 Solutions. θ + 1. θ 2 + cotθ ( ) sinθ e iφ is an eigenfunction of the ˆ L 2 operator. / θ 2. φ 2. sin 2 θ φ 2. ( ) = e iφ. = e iφ cosθ.

Transcript:

Tehnal Appendx (Not fo publaton Gene and Band Advetsng Stateges n a Dynam Duopoly Ths Tehnal Appendx povdes supplementay nfomaton to the pape Gene and Band Advetsng Stateges n a Dynam Duopoly. It s dvded nto the followng setons: A. Poofs n geate detal than n the Appendx. B. Poofs of ompaatve stats esults fo asymmet fms. C. Poofs of ompaatve stats esults fo symmet fms. D. Devatons fo the Extenson seton. A. Poofs n Geate Detal than n the Appendx. Poof of Poposton The Hamlton-Jaob-Bellman (HJB equaton fo fm { } s gven by V ps( bp dp3 ( a u ( u S3 3u3 S θ( a a S V max (A u a p V ( 3u3 S u S3 θ3( a a. S3 Fom ths the fst-ode ondtons fo u and a { } yeld espetvely V V V V u ( S a ( θ θ. 3 3 S S3 S S3 (A The fst-ode ondtons fo p and p yeld bp dp 0 d p bp 0. (A3 Solvng the two smultaneous equatons n (A3 we obtan the optmal pe of fm to be p d b 3 4bb dd We an theefoe smplfy the pe tems n (A so that. (A4

d b p ( bp dp b( m. Fom ths pont fowad we wll use 3 3 4bb dd m b denote the equlbum magns of the two fms. Substtutng (A and (A5 nto (A yelds d b ( 4bb dd V V V V V m S S and m d b b( 4bb dd ( θ θ3 ( 3 S S3 S S3 3 V V V V 3 V V V V S θ θ θ θ 3 S S S S 3 S S S S ( ( ( (. The lnea value funton V α βs γs3 { } satsfes (A6. The optmal band and gene advetsng desons n (A an now be ewtten as 3 3 (A5 to (A6 u ( β γ S a ( θ β θ γ. (A7 Substtutng V α βs γs3 nto (A6 and smplfyng we have α βs γs3 ms ( θβ θ3γ ( β γ S3 (A8 ( ( ( (. 3 3 β γ β γ S θβ θγ θβ θγ 3 3 Equatng the oeffents of S S3 and the onstant n equaton (A8 esults n the followng smultaneous equatons to solve fo α β and γ : ( ( ( (A9 α 3 3 θ β θ γ θ β θ γ θ β θγ 3 3 β m ( β γ ( β γ (A0 3 γ β γ (. (A Wtng out these equatons expltly fo { } we get ( ( ( (A α θ β θ γ θ β θ γ θ β θ γ

β β γ β γ m ( ( (A3 ( (A4 γ β γ ( ( ( (A5 α θ β θ γ θ β θ γ θ β θ γ ( ( (A6 β m β γ β γ (. (A7 γ β γ To solve subtat (A4 fom (A3 and (A7 fom (A6. Let y β γ and z β γ. Upon smplfyng we have the followng smultaneous quadat equatons n y and z : y yz y m 0 (A8 z yz z m 0. (A9 Substtutng fo z n tems of y n equaton (A9 afte solvng fo z n (A8 yelds the followng quat equaton n y : η η η η η (A0 4 3 y y 3y 4y 5 0 4 whee η 3 η 4 ( η 4 ( ( m m 4 η 8 m( and η 4 m. 5 3 The oots of equaton (A0 an be omputed usng Mathemata v4.0. Of the fou solutons fo y two ae magnay one s negatve and only one s always eal and postve. Ths s the unque Nash equlbum of the dffeental game. That soluton s gven by η ω y 4η 4 ω ω ω ( 7 ω ω5 ω6 ω ω5 ω6 5 6 (A 3

η 3 whee η ω 4η 3 ω η3 3ηη 4 ηη 5 ω3 η3 9ηηη 3 4 7ηη 4 7ηη 5 7 ηηη 3 5 3η 3 3 4 3 3 ω ( ω ω 4 ω ω 3 ω ω 4 5 ω6 3η ω 3 4 3 Knowng y we an ompute the followng: 3 4ηη 3 η 8η 4 and ω7. 3 η η η η γ y β y y z ( m y y γ z β z z. (A y One an see fom (A that β > 0 γ > 0 and β > γ esultng n postve values fo the ontols. Poof of Coollay To solve the smultaneous equatons: 3 α ( β γ (A3 8 β m ( β γ (A4 γ ( β γ (A5 m multply (A5 by and add to (A4. Ths yelds β γ. Substtutng ths nto (A5 esults n the followng quadat equaton n γ : m γ γ γ m γ m (A6 ( 3 0 9 (3 0. The two solutons of ths quadat equaton ae 3 m ± ( 6 m γ. (A7 9 To fnd out whh of the two oots to hoose we use the test that γ 0 when m 0. Ths s beause the value funton should be dentally equal to zeo when the goss magn s zeo sne the fm maes zeo poft n ths ase. Cheng wth (A7 t an be seen that 4

3 m ( 6 m γ (A8 9 s the only oot that satsfes ths ondton. m Knowng γ we an ompute β and α usng β γ and (A3 espetvely to be 3m ( 6 m β (A9 9 4 4 ( 3m 8 m (6 m ( 6 m α 3 4. (A30 08 An examnaton of equatons (A8-30 eveals that α > 0 β > 0 γ > 0 and β > γ so the ontols and the value funtons ae postve. Poof of Poposton To deve the optmal sales paths we substtute the esults fom Poposton nto the two state equatons to obtan the followng system of dffeental equatons: S S S S S (A3 S ( S ( S ( ( S (0 S. ( βγ ( β γ θ( ( θβ θγ ( θγ θβ (0 0 β γ β γ θ( θβ θγ θγ θβ 0 Fo expostonal ease denote ψ ( β γ ψ ( β γ ψ ( θβ θγ ( θ γ θβ. (A3 3 The dffeental equatons an now be ewtten as S ψs ψs θψ 3 S(0 S0 S ψ S ψ S θ ψ S (0 S. 3 0 Note fom (A33 that thee s no long-un equlbum n sales.e. S and S need not go to (A33 zeo. Solvng the smultaneous dffeental equatons n (A33 usng Mathemata v4.0 yelds 5

e S ( t ( ( S ( e ( S ( e (A34 ( ψ ψ t ( ψ ψ t ( ψ ψ t ψ ψ ψ 0 ψ ψ 0 ψ ψ ( ψ ψ ψ ( θψ θψ ( ψ ψ θ ψ ψ θ t t e ( ψ ψ t 3 ( ( e S ( t ( ( S ( e ( S ( e (A35 ( ψ ψ t ( ψ ψ t ( ψ ψ t ψ ψ ψ 0 ψ ψ 0 ψ ψ ( ψ ψ ψ ( θψ θ ψ ( ψ ψ θ ψ ψ θ t t e ( ψ ψ t 3 ( (. The long-un equlbum maet shaes ae gven by S( t S( t x lm x lm. t S( t S( t t S( t S( t Smplfyng and tang the lmts we have (A36 x. (A37 ( ( ( ( ( βγ ( β γ x β γ β γ β γ β γ Poof of Poposton 3 If the fm owns both bands ts deson poblem s max u( t a( t u( t a( t p( t p( t ( t e p( t S(( t bp( t dp( t p( t S(( t dp( t bp( t V (A38 0 ( a( t u( t ( a( t u( t dt s.t. S ( t u( t S( t u( t S( t θ( a( t a( t S(0 S0 (A39 S ( t u ( t S ( t u ( t S ( t θ ( a ( t a ( t S (0 S 0 whee V s the value funton of the fm a and u ae the gene and band advetsng desons espetvely of band S s the sales of band s the advetsng ost paamete and m the magn of band and s the fm s dsount ate. The HJB equaton s 6

p S( bp dp ps( dpbp ( u a ( u a V V max ( u S u S θ( a a u a p u a p S V ( u S u S θ( a a. S The fst-ode ondtons fo the optmal advetsng desons yeld V V V V u ( S a ( θ θ S S S S (A40 (A4 V V V V u ( S a ( θ θ. S S S S As befoe substtutng the solutons fom (A4-4 nto (A40 suggests a lnea value funton V α β S γ S wll solve the esultng patal dffeental equaton. The optmal m m m advetsng desons an now be ewtten as 0 Note fom (A43-44 that ethe m m m m (A4 u max{0 ( β γ S } a ( θ β θ γ (A43 u max{0 ( γ β S } a ( θ β θ γ. (A44 m m m m u o u s always postve. If β > γ m m > 0 u and u sne band s moe poftable. If β < γ the opposte s tue. Theefoe n a m monopoly total band advetsng need not be zeo. The monopolst an hoose the optmal advetsng desons to ensue the value funton n the monopoly ase s neve less than that n the ase of ompetton.e. V V V whee V and V ae the pofts n the ompettve ase. We theefoe have m α β γ α β γ α γ β (A45. m ms ms S S S S Equaton (A45 an be ewtten to yeld α ( α α ( β ( β γ S ( γ ( β γ S 0. (A46 m m m Sne equaton (A46 holds S 0 S 0 t must be the ase that α ( α α β β γ γ β γ (A47 m m m whee eah of the above oeffents s non-negatve. 7

Fom equaton (A7 the total gene advetsng n the ompettve ase s ( θ β θ γ ( θ γ θ β (A48 whle that n the monopoly ase s fom equatons (A43-44 ( θ β θ γ ( θ β θ γ. (A49 m m m m Subtatng equaton (A48 fom (A49 the dffeene between the total gene advetsng n the monopoly ase and that n the ompettve ase s ( θ ( β β θ ( γ γ ( θ ( β γ θ ( γ β (A50 m m m m whh fom equaton (A47 s geate than zeo. Theefoe the monopolst s total gene advetsng s geate than that unde ompetton. B. Poofs of Compaatve Stats Results fo Asymmet Fms. We epodue the ompaatve stats table fo onvenene. Table : Compaatve Stats Results fo the Asymmet Case Vaables 3 m m3 3 3 3 θ α????????? β?? γ u a? V????????? x Legend: nease; deease; unhanged;? ambguous. 8

To obtan the ompaatve stats fo γ and β we fst ty to obtan the esults fo y and z. Defne F( y z m y yz y 0 F ( y z m z yz z 0. Fo any paamete φ we an use the mplt funton theoem y F F F F F F φ y z φ z z φ z F F F F F F φ y z φ y y φ to detemne the ompaatve stats esult. Afte some smplfaton ths an be ewtten as y F y z y φ φ z F z y z φ φ (B (B (B3 whee the detemnant y z y > 0. z y z We now detemne the ompaatve stats wth espet to eah of the model paametes. y y z y y z m z 0 (B4 z y z z m y y z y y m 0 z (B5 z y z y z m 9

y y z y y( y z y z 0 (B6 z y z yz y y z y yz 0 z z (B7 z y z z( y z y y z y y z z y z yz y ( y z? yz( y z y y z y yz z z y z z? yz( y z z ( y z y y z y y z z y z yz y ( y z? yz( y z (B8 (B9 (B0 0

Theefoe y y z y yz z z y z z yz( y z.? z ( y z y z y z y z > 0 < 0 < 0 > 0 < 0 > 0 m m m m y z z y z y > 0 < 0 < 0 < 0 > 0 > 0. (B (B The ompaatve stats esults fo y on and and fo z on and ae ambguous. Note that y and z ae ndependent of and θ { }. To ompute the ompaatve stats fo γ β γ and β we use γ y β y y γ z β z z. (B3 Theefoe γ ( y y y y y 0 β ( y y ( > > 0 m m m m m m γ ( z z z z z 0 β ( z z ( < < 0 m m m m m m (B4 (B5 γ ( y y y y y 0 β ( y y ( < < 0 (B6 m m m m m m γ ( z z z z z 0 β ( z z ( > > 0 (B7 m m m m m m γ ( y y y 0 β ( y ( y y < < 0 (B8 y y y γ ( z z z z z 0 β ( z z ( > > 0 (B9

γ ( y y y y y 0 β ( y y ( > > 0 (B0 γ ( z z z 0 β ( z ( z z < < 0 (B z z z γ ( y y y ( y ( y y y β y y (B γ ( z z z z z 0 β ( z z ( < < 0 (B3 γ ( y y y y y 0 β ( y y ( < < 0 (B4 γ ( z z z ( z ( z z z β z z (B5 γ ( y y y β ( y ( y y (B6 y y y γ ( z z z z z 0 β ( z z ( > > 0 (B7 γ ( y y y y y 0 β ( y y ( > > 0 (B8 γ ( z z z β ( z ( z z. z z z (B9 Although the ompaatve stats fo γ on and n (B and fo γ on and n (B5 ae ambguous one an sgn these devatves by tansfomng the mplt equatons n (B n tems of γ and γ. Rewtng the equatons n (B usng z γ yelds the followng: y γ and J ( γ γ m γ γ γ γ 0 J ( γ γ m γ γ γ γ 0. (B30

Fo any paamete φ we have γ J J J J J J φ γ γ φ γ γ φ γ J J J J J J φ γ γ φ γ γ φ Afte some smplfaton ths an be ewtten as. (B3 γ γ γ J φ γ γ γ φ (B3 γ γ γ J φ γ γ γ φ whee the detemnant > 0. as follows: We an now ompute the ompaatve stats fo γ on and and fo γ on and γ γ γ γ γ γ γ γ γ γ γ γ (B33 γγ γ γγ γ γ γ γ γ γ γ γ γ γ γ γ (B34 γγ γγ γ 3

γ γ γ γ γ γ γ γ γ γ γ γ (B35 γγ γ γγ γ γ γ γ γ γ γ γ γ γ γ γ (B36 γγ. γγ γ Howeve the ompaatve stats of β on and and of β on and ae stll ambguous. Moeove note that γ β γ and β ae ndependent of and θ { }. We an now wte γ β γ β γ β γ β γ > 0 > 0 < 0 < 0 < 0 < 0 > 0 > 0 < 0 m m m m m m m m β γ β γ β γ β γ γ < 0 > 0 > 0 > 0 > 0 < 0 < 0 > 0 < 0 (B37 β γ γ β γ γ β γ γ β < 0 < 0 > 0 < 0 < 0 > 0 > 0 > 0 < 0 > 0. Fo the ompaatve stats esults fo α { } on the model paametes we note that ( ( ( (B38 α θ β θ γ θ β θ γ θ β θ γ ( ( ( (B39 α θ β θ γ θ β θ γ θ β θ γ α ( y θ y ( y θ y( z θ z (B40 4

α ( z θ z ( y θ y( z θ z. (B4 The ompaatve stats esults fo α { } on and ae easy to obtan. We have α α α α > 0 > 0 > 0 > 0. (B4 The ompaatve stats fo α { } on the othe model paametes ae ambguous gven the esults fo β and γ { }. The ompaatve stats fo y and z have the opposte sgns fo most of the model paametes so the devatves of the seond tem n equatons (B40 and (B4 ( y y( z z and θ θ ( y y( z z θ θ espetvely ae not easy to sgn. Fo the ompaatve stats fo the optmal band and gene advetsng desons we an see that sne u a and ( β γ S u ( β γ S a ( θ β θ γ and ( θ γ θ β most of the esults ae staghtfowad gven the ompaatve stats fo γ β { }. Fo the ompaatve stats of we ewte the optmal band advetsng desons as u w..t. and and of u w..t. and u ( β γ S y S u ( β γ S z S.(B43 Substtutng fo y n tems of mplt equatons: u and fo z n tems of S SS S S SS S u we have the followng G ( u u m u uu u 0 G ( u u m u uu u 0. (B44 The mplt funton theoem yelds 5

whee the detemnant > 0. Theefoe u G G G G G G u u u u u G G G G G G u u u u (B45 u u u u S SS S SS u u u u S SS SS S (B46 uu u SS S uu SS u u u u S SS S SS u u u u S SS SS S (B47 uu SS uu u SS S u u u u S SS S SS u u u u S SS SS S (B48 u uu u S SS S 0 6

The ompaatve stats of u u u u S SS S SS u u u u S SS SS S (B49 a and fashon. The esults ae summazed below. 0. u uu u S SS S a w..t. and espetvely an be obtaned n a smla u u u u u u u u u > 0 < 0 < 0 > 0 < 0 > 0 > 0 < 0 > 0 m m m m u u u u u u u a a < 0 > 0 < 0 > 0 > 0 < 0 < 0 > 0 < 0 m m a a a a a a a a a < 0 > 0 < 0 > 0 > 0 < 0 < 0 < 0 < 0 m m a a a a a a a a a > 0 > 0 < 0 > 0 > 0 > 0 < 0 < 0 > 0. θ θ θ θ (B50 Gven the ambguous ompaatve stats esults fo α most of the esults fo V ae also ambguous. Howeve t s possble to sgn the ompaatve stats of V fo the paametes n equaton (B4. The esults ae as follows: V V V V > 0 > 0 > 0 > 0. We now ty to obtan the ompaatve stats esults fo the steady-state maet shaes. The long-un equlbum maet shaes ae gven by x ( βγ ( β γ x β γ β γ β γ β γ (B5. (B5 ( ( ( ( Fo all the paametes exept and the ompaatve stats of x and x ae staghtfowad to obtan gven the ompaatve stats fo y and z. Fo and we ewte equaton (B5 as 7

y z x x. y z y z (B53 Defne l y l z. Theefoe l l x x. l l l l l Fo the ompaatve stats of x w..t. any paamete φ note that x l l theefoe l x φ l l l (. φ φ φ l l ( l l (B54 (B55 Rewtng the mplt equatons n tems of l and l yelds H ( l l m l ll l 0 H ( l l m l ll l 0. (B56 As befoe one an obtan the ompaatve stats of l and l w..t. and usng whee φ s any paamete. l H H H φ l l φ l H H H φ l l φ (B57 Pefomng the ompaatve stats of l and l fo any paamete φ yelds whee the detemnant > 0. l H l l l φ φ l H l l l φ φ (B58 8

Pefomng the ompaatve stats of l and l w..t. we have l ( l l l 3 l ( l l. l (B59 l ( l l 0 l l Theefoe > 0 and < 0. Moeove l l ( l 3 ( l l ( l l ll 3 ( l l (B60 ( l ( l ( l l > 0. x Fom (B55 and (B60 one an see that > 0 3. The ompaatve stats of x on and an also be obtaned usng the mplt equatons n (B56. These esults ae summazed below. x x x x > 0 < 0 < 0 > 0. (B6 Fo all the othe paametes the esults fo y also hold fo x. Sne x x the ompaatve stats esults fo x an be obtaned by flppng the esults fo x. C. Poofs of Compaatve Stats Results fo Symmet Fms. Fo onvenene we epodue the ompaatve stats esults n Table. We now 4 4 ( 3m 8 m (6 m ( 6 m α (C 3 4 08 3m ( 6 m β (C 9 9

3 m ( 6 m γ. (C3 9 Table : Compaatve Stats Results fo the Symmet Case Vaables m α? β γ u a V?? x Legend: nease; deease; unhanged;? ambguous. Tang the fst devatves of β and γ n equatons (C-3 wth espet to eah of the paametes yelds the followng ompaatve stats: β γ β γ β γ β γ > 0 > 0 < 0 < 0 < 0 > 0 > 0 < 0. (C4 m m The esults n (C4 follow n a staghtfowad manne. The ompaatve stats fo the optmal advetsng expendtues u and a { } ae easy to obtan wheneve the esults fo β and γ ae n the same deton. The ompaatve stats β and γ wth espet to and ae not of the same sgn. Howeve even n these ases the sgns of the devatves an be obtaned as n the asymmet ase usng the mplt equatons. The ompaatve stats esults of V and x { } on the model paametes an also be obtaned as fo asymmet fms. We obtan the followng ompaatve stats fo the optmal advetsng desons: 0

u u u u u u u u a > 0 < 0 < 0 < 0 > 0 > 0 < 0 < 0 > 0 m m m a a a a a a a a a > 0 < 0 < 0 < 0 < 0 < 0 < 0 > 0 > 0. m (C5 D. Devatons fo the Extenson Seton. a. Maet Potental The Hamlton-Jaob-Bellman (HJB equaton fo fm { } s gven by V ps( bp dp3 ( a u ( u S3 3u3 S θ( a a QSS S V max (D u a p V ( 3u3 S u S3 θ3( a a Q S S. S3 Fom ths the fst-ode ondtons fo u and a yeld espetvely V V V V u ( S a ( θ θ QS S. 3 3 S S3 S S3 As befoe one an solve fo the optmal pes and vefy that the solutons ae the same as those n equaton (A4. Substtutng equatons (A4 and (D nto (D yelds V V V V V m S Q S S S ( θ θ j ( ( 3 S S3 S S3 3 V V V V 3 V V V V S θ θ θ θ QSS 3 S S S S 3 S S S S ( ( ( ( (. The lnea value funton V α β S γ S3 satsfes (D3. The optmal band and gene advetsng desons n (D an now be ewtten as u ( β γ S a ( θ β θ γ QS S. (D4 3 3 Substtutng V α β S γ S3 nto (D3 and smplfyng we have S S ms Q S S S (D5 ( ( ( ( (. α β γ 3 ( θβ θ3 γ ( ( β γ 3 3 3 βγ β γ S θβ θγ θβ θγ QSS 3 3 (D (D3

Equatng the oeffents of S S3 and the onstant n equaton (D5 esults n the followng smultaneous equatons to solve fo α β and γ { } : ( Q ( ( Q (D6 3 α θβ θ3γ θβ θγ θβ θγ 3 ( ( ( ( ( (D7 3 3 β m θβ θ3γ β γ β γ θβ θγ θβ θγ 3 3 ( ( ( (.(D8 3 γ θβ θ3γ β γ θβ θγ θβ θγ 3 Wtng out these equatons expltly fo { } we get ( Q ( ( Q (D9 α θβ θγ θβ θγ θβ θγ ( ( ( ( ( (D0 β m θβ θγ β γ β γ θβ θγ θβ θγ ( ( ( ( (D γ θβ θγ β γ θβ θγ θβ θγ ( Q ( ( Q (D α θβ θγ θβ θγ θβ θγ ( ( ( ( ( (D3 β m θβ θγ β γ β γ θβ θγ θβ θγ ( ( ( (. (D4 γ θβ θγ β γ θβ θγ θβ θγ To solve subtat (D fom (D0 and (D4 fom (D3. Let y β γ and z β γ Smplfyng esults n the followng smultaneous quadat equatons n y and z :. y yz y m 0 (D5 z yz z m 0. (D6 Substtutng fo z n tems of y n equaton (D6 afte solvng fo z n (D5 yelds the followng quat equaton n y : η η η η η (D7 4 3 y y 3y 4y 5 0

4 whee η 3 η 4 ( η 4 ( ( m m 4 η 8 m( and η 4 m. 5 3 The oots of equaton (D7 an be omputed usng Mathemata v4.0. Of the fou solutons fo y two ae magnay one s negatve and only one s always eal and postve. Ths s the unque Nash equlbum of the dffeental game. That soluton s gven by η ω y 4η 4 ω ω ω ( 7 ω ω5 ω6 ω ω5 ω6 5 6 (D8 η 3 whee η ω 4η 3 ω η3 3ηη 4 ηη 5 ω3 η3 9ηηη 3 4 7ηη 4 7ηη 5 7 ηηη 3 5 3η 3 3 4 3 3 ω ( ω ω 4 ω ω 3 ω ω 4 5 ω6 3η ω 3 4 3 3 4ηη 3 η 8η 4 and ω7. 3 η η η η Knowng y we an ompute z ( m y y. Knowng y and z γ and γ y an be obtaned by solvng the followng smultaneous quadat equatons: ( y y ( y (( z (D9 γ θ γ θ γ θ γ (( z z ( y(( z. (D0 γ θ γ γ θ θ γ Afte γ and γ ae obtaned one an obtan β and β usng β y γ and β z γ espetvely. These solutons fo γ γ β and β an be used n equatons (D9 and (D to solve fo α and α. b. Othe Extensons The uent-value Hamltonan fo fm { } s H ps ( bp dp3 ( a u S λ( ω ( ω ω ( ω ( (D u( S3 S 3u3( 3 S 3 S3 S S a a S3 ( 3u3( 3 S ( 3 S3 u( S3 ( S a a S S µ ω ω ω ω (. 3

The optmal pes ae the same as those obtaned n equaton (A4. Fo the optmal advetsng desons the fst-ode ondtons fo fm ae H u ( ( u λ ( ω S ( ω S µ ( ω S ( ω S 0 (D 3 3 H λ S µ S 0. a S S S S 3 a (D3 Solvng the fst-ode ondtons fo the two fms yelds the optmal advetsng desons to be λ S µ S u ( S ( S ( a ( ω ω λ µ S S S S (D4 λ S µ S u ( S ( S ( a (. ω ω µ λ S S S S The adjont equatons fo the shadow pes λ ( t and µ ( t fo fm ae H H u H a λ λ S u S a S 3 3 3 3 ( ω u ω u ( a a S λ p ( b p d p ( λ µ 3 ( 3 3 3 3 S S ( S S ω (( ω S ( ω S ( λ µ ( 3 3 3 3 3 3 3 S λ S µ S S ( ( ( µ 3 λ3 S S S S S S S 3 3 3 3 ( H H u H a µ µ S u S a S 3 3 3 3 3 3 3 ( µ λ 3 3 ωu ( ω u ( a a S µ ( λ µ ( 3 3 3 S S 3 3 ( S S ( ω (( ω S ( ω S ( λ µ ( ( µ 3 λ3 3 3 3 3 3 3 3 S3 λ S µ S S ( ( ( µ 3 λ3. S S S S S S S 3 3 3 3 ( (D5 (D6 One an now substtute the optmal pes and advetsng desons nto the state equatons and the adjont equatons. Sne these dffeental equatons do not pemt losed-fom solutons we use numeal methods to solve fo λ ( t and µ ( t { }. These solutons an then be used n (D4 to obtan the optmal band and gene advetsng desons fo the two fms. 4