MCPDerivations.nb 1 This notebook presents derivation of results for the paper "The Markov Consumption Problem" by Michael Sattinger. The first derivation is the differential equations appearing in 2.5 and 2.6.Utility in 2.2 is given by U@CD below with γ replaced by g. U@C_D := HC ^ gL ê g For optimal C1 and C2, the current value formulations given in 2.3 and 2.4 are : U@C1D + p1 ∗ HV2@AD − V1@ADL + Hr ∗ A + y1 − C1L ∗ D@V1@AD, AD − b ∗ V1@AD = 0 U@C2D + p2 ∗ HV1@AD − V2@ADL + Hr ∗ A + y2 − C2L ∗ D@V2@AD, AD − b ∗ V2@AD = 0 The first order conditions for optimal consumption appearing in 5.3 are Simplify@D@U@C1D + p1 ∗ HV2@AD − V1@ADL + Hr ∗ A + y1 − C1L ∗ D@V1@AD, AD − b ∗ V1@AD, C1DD Simplify@D@U@C2D + p2 ∗ HV1@AD − V2@ADL + Hr ∗ A + y2 − C2L ∗ D@V2@AD, AD − b ∗ V2@AD, C2DD C1−1+g − V1′ @AD C2−1+g − V2′ @AD The following solves the above first order conditions for C1 and C2, given by 5.4 in the appendix. Simplify@Solve@C1−1+g − V1′ @AD 0, C1DD Simplify@Solve@C2−1+g − V2′ @AD 0, C2DD Solve::ifun : Inverse functions are being used by Solve, so some solutions may not be found; use Reduce for complete solution information. 99C1 → V1′ @AD 1 −1+g More… == Solve::ifun : Inverse functions are being used by Solve, so some solutions may not be found; use Reduce for complete solution information. More… 1 −1+g == 99C2 → V2′ @AD The derivatives of Ci with respect to A Happearing as 5.5L are then : SimplifyADAV1′ @AD −1+g , AEE 1 SimplifyADAV2′ @AD −1+g , AEE 1 1 −1+g V1′′ @AD V1′ @AD−1+ −1 + g 1 −1+g V2′′ @AD V2′ @AD−1+ −1 + g In the above, ViA = Vi '@AD and ViAA = V1′′ @AD To find V1AA and V2AA, take the derivatives of 2.3 and 2.4 with respect to A, applying the envelope theorem to disregard changes in C1 and C2. MCPDerivations.nb 2 Simplify@D@U@C1D + p1 ∗ HV2@AD − V1@ADL + Hr ∗ A + y1 − C1L ∗ D@V1@AD, AD − b ∗ V1@AD, ADD Simplify@D@U@C2D + p2 ∗ HV1@AD − V2@ADL + Hr ∗ A + y2 − C2L ∗ D@V2@AD, AD − b ∗ V2@AD, ADD −Hb + p1 − rL V1′ @AD + p1 V2′ @AD + H−C1 + A r + y1L V1′′ @AD p2 V1′ @AD − Hb + p2 − rL V2′ @AD + H−C2 + A r + y2L V2′′ @AD Simplify@Solve@−Hb + p1 − rL V1′ @AD + p1 V2′ @AD + H−C1 + A r + y1L V1′′ @AD 0, V1′′ @ADDD Simplify@Solve@p2 V1′ @AD − Hb + p2 − rL V2′ @AD + H−C2 + A r + y2L V2′′ @AD 0, V2′′ @ADDD Hb + p1 − rL V1′ @AD − p1 V2′ @AD 99V1′′ @AD → == −C1 + A r + y1 −p2 V1′ @AD + Hb + p2 − rL V2′ @AD 99V2′′ @AD → == −C2 + A r + y2 The above are the solutions for ViAA appearing in 5.8 and 5.9. Substituting into the derivatives of Ci with respect to A above yields : Hb+p1−rL V1 @AD−p1 V2 @AD V1′ @AD −1+g −C1+A r+y1 SimplifyA E −1 + g 1 ′ −1+ ′ −p2 V1 @AD+Hb+p2−rL V2 @AD V2′ @AD −1+g −C2+A r+y2 SimplifyA E −1 + g 1 −1+ ′ ′ 1 −1+g HHb + p1 − rL V1′ @AD − p1 V2′ @ADL V1′ @AD−1+ H−1 + gL H−C1 + A r + y1L 1 −1+g H−p2 V1′ @AD + Hb + p2 − rL V2′ @ADL V2′ @AD−1+ H−1 + gL H−C2 + A r + y2L C1−1+g To express these differential equations in terms of consumption levels, use the first order conditions for consumption given above by C1−1+g − V1′ @AD = 0 and C2−1+g − V2′ @AD = 0. Substituting Ci−1+g for Vi '@AD into the differential equations for C1 and C2 yields expressions in terms of C1, C2 and A. 1 −1+g C1−1+g HHb + p1 − rL C1−1+g − p1 C2−1+g L SimplifyA E H−1 + gL H−C1 + A r + y1L −1+ 1 −1+ −1+g C2−1+g H−p2 C1−1+g + Hb + p2 − rL C2−1+g L SimplifyA E H−1 + gL H−C2 + A r + y2L −1+ 1 HC1−1+g L −1+g H−C2−1+g p1 + C1−1+g Hb + p1 − rLL H−1 + gL H−C1 + A r + y1L −1+ 1 HC2−1+g L −1+g H−C1−1+g p2 + C2−1+g Hb + p2 − rLL H−1 + gL H−C2 + A r + y2L These simplify to the differential equations in Result 1, equations 2.5 and 2.6. Appendix A .2. Equation 2.8 in Result 2. Assume singularity at As with C1 > 0, b > r.Then the numerator must be zero : MCPDerivations.nb 3 C2 y i i i j z j SimplifyASolveAj j z jr − b − p1 ∗ j j1 − j C1 { k k k −1+g yz y z z zz z 0, C2EE {{ Solve::ifun : Inverse functions are being used by Solve, so some solutions may not be found; use Reduce for complete solution information. More… 1 −1+g b + p1 − r 99C2 → C1 J N == p1 Appendix A .3. Derivation of Equation 2.10 in Result 3 Hslope of C2 at AminL. Applying l ' Hospital' s Rule, the derivative of the numerator of dC2 ê dA is Simplify@D@HC2@AD ê H1 − gLL ∗ Hr − b − p2 ∗ H1 − HHC2@AD ê C1@ADL ^ H1 − gLLLL, ADD 1 C2@AD −g JJ N J−H−1 + gL p2 C2@AD2 C1′ @AD + 2 C1@AD H−1 + gL C1@AD C2@AD g H−2 + gL p2 C1@AD C2@AD C2′ @AD + Hb + p2 − rL C1@AD2 J N C2′ @ADNN C1@AD At C2@AminD = 0, this reduces to : HHb + p2 − rL C2′ @ADL H−1 + gL The derivative of the denominator is Simplify@D@r ∗ A + y2 − C2@AD, ADD r − C2′ @AD Then 1 HHb + p2 − rL C2′ @ADL C2′ @AD = r − C2′ @AD H−1 + gL Dividing both sides by C2′ @AD yields an equation that can be solved for C2′ @AD : Hb + p2 − rL r − C2′ @AD = H−1 + gL Hb + p2 − rL SimplifyASolveAr − C2′ @AD == , C2′ @ADEE H−1 + gL b + p2 − g r 99C2′ @AD → == 1−g This is the expression for 2.10 in Result 3. Appendix A .4., derivation of dC1 ê dA at singularity As in Section 3 on numerical method. Applying l ' Hospital' s rule at the singularity, the derivative of the numerator of dC1 ê dA is : MCPDerivations.nb 4 i i i C1@AD y j j SimplifyADAHC1@AD ê H1 − gLL ∗ j j z z jr − b − p1 ∗ j j1 − j k C2@AD { k k 1−g yz y z z zz z, AEE {{ 1 H−1 + gL C2@AD2 C1@AD −g C1@AD g JJ N JH−2 + gL p1 C1@AD C2@AD C1′ @AD + Hb + p1 − rL J N C2@AD2 C1′ @AD − C2@AD C2@AD H−1 + gL p1 C1@AD2 C2′ @ADNN −1+g i i yz y i C2@AD y j z j z Since j j z z jr − b − p1 ∗ j j1 − j zz z= C1@AD k { k k {{ 0 from the proof of Result 2. i. in Appendix Section A .2, the derivative above reduces to 1−g i i yz y i C1@AD y j j z SimplifyAHC1@AD ê H1 − gLL ∗ DAj j z z z jr − b − p1 ∗ j j1 − j zz z, AEE C2@AD k { k k {{ −g C1@AD p1 C1@AD I M H−C2@AD C1′ @AD + C1@AD C2′ @ADL C2@AD − C2@AD2 The derivative of the denominator of dC1 ê dA is : Simplify@D@r ∗ A + y1 − C1@AD, ADD r − C1′ @AD Then applying l ' Hospital' s rule at As yields C1@AD i p1 C1@AD I M H−C2@AD C1′ @AD + C1@AD C2′ @ADL y j z C2@AD j z ′ j z C1 @AD = j z j z j− z ì Hr − C1 @ADL 2 j z C2@AD k { −g ′ This simplifies to C1@AD 1−g C1@AD i p1 I M I− C1′ @AD + C2′ @ADM y j z j z C2@AD C2@AD j z j − z C1 @AD = j z j z ′ j z Hr − C1 @ADL j z k { ′ HThe author would like to thank Gonçalo Pina for pointing out an error in the equation above, which has now been corrected. The expression in the paper is correct.L 1 C1@AD i b + p1 − r z y 1−g Let x = C1 @AD, rat = = j z from equation 2.8 in Result 2, j C2@AD p1 k { S2 = r ∗ A + y2 − C2@AD and C2′ @AD = dC2 ê dA = HC2@AD ∗ Hr − b − p2 ∗ H1 − HratL ^ Hg − 1LLL ê H1 − gLL ê S2 from 2.6. Making these substitutions yields ′ 1 i x=j j− Hp1 HratL1−g k Hr − xL y H− x + rat ∗ C1@AD ∗ HC2@AD ∗ Hr − b − p2 ∗ H1 − HratL ^ Hg − 1LLL ê H1 − gLL ê S2 LLz z { Solving yields a quadratic in x = dC1 ê dA : MCPDerivations.nb 1 i SimplifyASolveAx == j j− Hp1 HratL1−g k Hr − xL 5 y H− x + rat ∗ C1@AD ∗ HC2@AD ∗ Hr − b − p2 ∗ H1 − HratL ^ Hg − 1LLL ê H1 − gLL ê S2 LLz z, xEE { 1 99x → Irat−g Ip1 rat S2 − g p1 rat S2 − 2 H−1 + gL S2 , r ratg S2 + g r ratg S2 − HH−1 + gL S2 HH−1 + gL Hp1 rat − r ratg L2 S2 − 4 p1 rat1+g H−b rat + r rat + p2 H−rat + ratg LL C1@AD C2@ADLLMM=, 1 9x → Irat−g Ip1 rat S2 − g p1 rat S2 − r ratg S2 + g r ratg S2 + 2 H−1 + gL S2 , HH−1 + gL S2 HH−1 + gL Hp1 rat − r ratg L2 S2 − 4 p1 rat1+g H−b rat + r rat + p2 H−rat + ratg LL C1@AD C2@ADLLMM== 1 x1@C1_, C2_, A_D := Irat−g Ip1 rat S2 − g p1 rat S2 − 2 H−1 + gL S2 , r ratg S2 + g r ratg S2 + HH−1 + gL S2 HH−1 + gL Hp1 rat − r ratg L2 S2 − 4 p1 rat1+g H−b rat + r rat + p2 H−rat + ratg LL C1@AD C2@ADLLMM 1 x2@C1_, C2_, A_D := Irat−g Ip1 rat S2 − g p1 rat S2 − r ratg S2 + 2 H−1 + gL S2 , g r ratg S2 + HH−1 + gL S2 HH−1 + gL Hp1 rat − r ratg L2 S2 − 4 p1 rat1+g H−b rat + r rat + p2 H−rat + ratg LL C1@AD C2@ADLLMM By hand, this simplifies to : HH2 ∗ r − b − p1L + HHHH2 ∗ r − b − p1L ^ 2L + 4 ∗ p1 ∗ Hrat ^ H1 − gLL ∗ HC1As^ 2L ∗ Hr − b + p2 ∗ HHrat^ Hg − 1LL − 1LL ê HS2 ∗ H1 − gLLL ^ H1 ê 2LLL ê 2 Numerical check on simplification MCPDerivations.nb b = 0.03 r = .0270352 p2 = .0736241 ∗ 52 p1 = 0.24474850760576117` y1 = 1 y2 = .5 g = .5 rat = HHb + p1 − rL ê p1L ^ H1 ê H1 − gLL As = −15.438 C1As = r ∗ As + y1 C2As = C1As ê rat S2 = r ∗ As + y2 − C1As ê rat 0.03 0.0270352 3.82845 0.244749 1 0.5 0.5 1.02437 −15.438 0.582631 0.568767 −0.486137 1 Irat−g Ip1 rat S2 − g p1 rat S2 − 2 H−1 + gL S2 , r ratg S2 + g r ratg S2 + HH−1 + gL S2 HH−1 + gL Hp1 rat − r ratg L2 S2 − 4 p1 rat1+g H−b rat + r rat + p2 H−rat + ratg LL C1As C2AsLLMM HH2 ∗ r − b − p1L + HHHH2 ∗ r − b − p1L ^ 2L + 4 ∗ p1 ∗ Hrat ^ H1 − gLL ∗ HC1As ^ 2L ∗ Hr − b + p2 ∗ HHrat ^ Hg − 1LL − 1LL ê HS2 ∗ H1 − gLLL ^ H1 ê 2LLL ê 2 0.0601077 0.0601077 6