Monetary Policy in Pakistan: A Dynamic Stochastic General Equilibrium Analysis

Transcription

1 Moneary Policy in Pakisan: A Dynamic Sochasic General Equilibrium Analysis Ehsan U. Choudhri a and Hamza Malik b a Carleon Universiy (conac: ehsan_choudhri@carleon.ca), b Sae Bank of Pakisan (conac: hamza.malik@sbp.org.pk) Working Paper 12/389 June 212 Inernaional Growh Cenre London School of Economics and Poliical Science 4h Floor, Tower Two Houghon Sree London WC2A 2AE Unied Kingdom For media or communicaions enquiries, please conac Mazida Khaun mazida.khaun@heigc.org Direced and Organised by

2 Moneary Policy in Pakisan: A Dynamic Sochasic General Equilibrium Analysis Ehsan U. Choudhri and Hamza Malik* Carleon Universiy and Sae Bank of Pakisan March 212 We are graeful o Inernaional Growh Cenre for he suppor of his projec. We would also like o hank Aban Haq for helpful assisance.

3 Moneary Policy in Pakisan: A Dynamic Sochasic General Equilibrium Analysis Absrac A small-scale DSGE model for Pakisan is developed o analyze moneary policy in Pakisan. The model includes a financial secor and disinguishes beween high-income households who paricipae in he financial secor and low-income households who face borrowing consrains. In evaluaing differen moneary policy opions, he model akes ino accoun he consrain ha he Sae Bank has o saisfy he long-erm borrowing needs of he governmen, and hus canno independenly deermine he long-run rae of inflaion. The baseline model assumes ha fiscal policy adjuss primary budge surplus o sabilize governmen deb o GDP raio around a feasible arge level. This model is used o examine macroeconomic adjusmen o various shocks and o compare he macroeconomic performance of alernaive moneary policy rules. In his regime, moneary policy can play an imporan role in sabilizing inflaion and oupu. The paper also considers an alernaive policy regime in which fiscal policy does no aemp o sabilize governmen deb. In his case, moneary policy is consrained furher by he need o use ineres raes o conrol he growh of governmen deb. Feasible moneary rules under his consrain are found o produce much larger variabiliy in inflaion. 2

4 1. Inroducion Pakisan has recenly experienced high inflaion persising in double-digis, fiscal imbalances, low privae secor credi growh and sagnan economic growh. This paper explores how moneary policy should be formulaed in his difficul macroeconomic environmen. Dynamic Sochasic General Equilibrium (DSGE) models have emerged recenly as he sandard ool for evaluaing moneary policy and are increasingly used for policy advice o cenral banks in developed and many emerging counries. We develop a small-scale DSGE model for Pakisan and use i o evaluae differen moneary policy opions. Our model follows he sandard framework of he new open economy macroeconomic models, bu inroduces cerain variaions ha are appropriae for moneary policy analysis in Pakisan. ` Firs, we include a banking secor o incorporae financial fricions in he model. There are a number of models ha inroduce financial fricions of one ype or anoher in DSGE models (e.g., Bernanke, Gerler and Gilchris, 1999; Goodfriend and McCallum, 27; Canzoneri e al., 28), bu he focus of his lieraure is on modelling financial markes in developed counries. In his paper, we use a varian of he Canzoneri e al.'s (28) model o examine how financial fricions inerac wih moneary policy in a developing economy like Pakisan. Second, we depar from he represenaive-agen seup and disinguish beween households wih high and low incomes. High-income households paricipae in he financial marke (e.g., hold bank deposis and purchase governmen bonds). Low-income households, on he oher hand, face liquidiy consrains (do no borrow or lend) and hus do no inerac wih financial markes. A number of recen DSGE models (e.g., Gali e al., 27) have included liquidiy-consrained households (also referred o as "non-ricardian" households or "rule-of-humb" consumers) o allow deparures from he Ricardian equivalence proposiion, which implies ha a deb-financed ax 3

5 cu would no affec consumpion. We include such households also o explore wheher moneary policy acions impac households a low and high income levels differenly. Finally, as financial markes in Pakisan are no well inegraed wih foreign financial markes, we do no assume an ineres pariy relaion linking he domesic and foreign ineres raes. One major consrain for moneary policy in Pakisan arises from he need of he governmen o coninuously borrow from he Sae Bank. If fiscal policy relies on a permanen flow of revenue from money creaion (seignorage), he inflaion rae in he long run is deermined by he rae of growh of he moneary base needed o yield he long-run level of seignorage. In his case, moneary policy canno independenly deermine a long-run inflaion arge. A crucial quesion is wheher fiscal policy is prepared o adjus primary defici o sabilize governmen deb a some arge level. If fiscal policy sabilizes governmen deb, hen moneary policy can sabilize inflaion and oupu by following a convenional (Taylor-ype) rule whereby he real ineres rae is increased in response o an increase in inflaion above he long-run rae (deermined by seignorage requiremens) and o an increase in he oupu gap. However, he convenional ineres rae response o inflaion may no be desirable or even feasible if he governmen is no willing or able o adjus primary defici in response o deb growh. Such an inflexible fiscal policy - - ha subordinaes moneary policy o fiscal needs - - is referred o as "fiscal dominance" in he lieraure. Under fiscal dominance, a ighening of moneary policy in response o higher inflaion has been shown o lead o perverse resuls (Sargen and Wallace, 1981; Woodford, 21). Kumhof, Nunes and Yakadina (28) explore feasible ineres rae rules under fiscal dominance for a closed economy model. They show ha i is beneficial o include governmen deb as an argumen in he ineres rae rule and an opimal rule would require lowering (insead of raising) he real ineres rae in response o higher inflaion. Fiscal 4

6 dominance, moreover, leads o greaer inflaion variabiliy and loss of welfare han a fiscal policy ha sabilizes governmen deb. In evaluaing moneary policy rules for Pakisan, we consider boh fiscal regimes. In our baseline case, we assume ha fiscal policy deermines he long-run value of seignorage, bu adjuss axes in response o he deviaion of governmen deb from is arge level. The economy is assumed o be subjec o several inernal and exernal shocks including shocks o governmen expendiures. We explore wha moneary rule is appropriae under hese condiions. In he convenional DSGE model based on he New-Keynesian framework wihou financial fricions, moneary aggregaes do no play a role in he formulaion of moneary policy (Clarida, Gali, and Gerler, 1999; Woodford, 23). Our model, however, incorporaes financial fricions, and we examine how such fricions influence he ransmission mechanism for moneary policy. The model is also used o invesigae some issues ha have been widely debaed. For example, i has been argued ha supply and foreign price shocks have been an imporan source of inflaion in Pakisan. We use impulse response funcions derived from he model o idenify he conribuion of such shocks o inflaion. There is an ongoing debae in Pakisan ha huge borrowing requiremens of he governmen significanly crowd ou bank lending o he privae secor and impede invesmen. We also use impulse response analysis o examine he impac of shocks o governmen expendiure on credi and invesmen. We also explore moneary policy opions under fiscal dominance. We consider moneary rules similar o Kumhof e al. (28). As our model differs from here's, especially in including financial fricions and liquidiy-consrained households, we explore wha rules are feasible and appropriae in our model. Moneary rules in his regime require a decrease in he real ineres rae 5

7 in response o an increase in real governmen deb and his policy is found o produce excessive inflaion variabiliy. The model is presened in Secion 2 and calibraed o Pakisan's economy in Secion 3. Secion 4 analyzes moneary policy in Pakisan and Secion 5 concludes he paper. 2. The Model Our model is based on he sandard new-keynesian framework, bu inroduces a number of variaions o address cerain moneary policy issues in Pakisan. There is one composie good (consising of differeniaed home and foreign varieies), which is used by households, invesors and governmen. There are wo ypes of households denoed by H and L. Households of ype H have higher wage income, own firms and paricipae in financial markes: buy governmen bonds, hold bank deposis and ake bank loans o finance fixed expendiures on nondurables. Households of ype L ge lower wage income, are liquidiy consrained, and do no ransac in he financial markes (hold no asses excep currency). Capial goods producers underake invesmen decisions subjec o adjusmen coss and supply (insalled) capial o capial leasing firms who finance he addiions o capial by loans from banks. Banks require cash reserves and governmen bonds o provide converibiliy services for deposis and use labor o monior loans. Governmen uses lump-sum axes o raise revenue. 1 Domesic financial markes are no inegraed wih global financial markes, and households and banks are assumed no o hold foreign bonds. Finally, nominal rigidiies are inroduced by assuming ha here are adjusmen coss for boh prices and wages as in Roemberg (1982). 1 As our focus is on moneary policy issues and no on he efficiency coss of disorionary axes, such axes are no included in he model o simplify he analysis.. 6

8 Real variables are denoed by lower case leers and nominal variables by upper case leers. An aserisk is used o denoe foreign variables. 2.1 Households Assume ha here be a coninuum of households of ype and l (,1), respecively. The Uiliy funcion for household h of ype H is H and L, indexed by h (,1) s ch, s ( h) HC ( CU H, s( h) / Ps ) HD ( DH, s( h) / Ps ) HNnH, s( h) UH, ( h) E s,(1) where, in period, ch, ( h), nh, ( h ) are he household s consumpion and labor supply while CU ( h), and D ( h ) are he (end of period) holdings of currency and bank deposis. The H, H, budge consrain for he household is CU H, ( h) DH, ( h) BH, ( h) WH, nh, ( h)(1 ACWH, ( h)) PR ( h) ch, ( h) H, ( h) P P, (2) CU H, 1 ( h) (1 RD, 1 ) DH, 1( h) (1 R 1 ) BH, 1 ( h) (1 RL, 1) L( h) P where BH, ( h ) is he sock of governmen bonds (a he end of he period); Lh ( ) is he fixed amoun of bank loans; P is he price of one uni of he composie good; PR ( h ) represens he household's share of profis; H, ( h) sands for (lump sum) real axes. R D,, R and R L, are he ineres raes on bank deposis, governmen bonds and bank loans; W, ( h ) H is he wage rae; and AC W ( h) / W ( h) H H, H, 1 WH, ( h) 1 2 P 1/ P2 2 is he adjusmen cos for wages. This adjusmen cos funcion is based on he exension of he basic Roemberg model by Laxon and Peseni (23) 7

9 and accouns for he presence of inflaion. We use a similar funcion below for price adjusmen coss. The household chooses ch, ( h), CU H, ( h), DH, ( h), and WH, ( h ) o maximize uiliy subjec o he budge consrain and he demand for is labor service (discussed below) wih wage elasiciy equal o. Opimizaion by he household implies he following condiions: EP 1 ch, ( h) EcH, 1( h) (1 R ) P, (3) H, ( ) H, ( ) CU h c h R, (4) P 1 R HC, H, ( ) H, ( ) D, D h c h R R, (5) P 1 R HD, (1 AC ( h))( 1) W ( h) n ( h) Pc ( h) ( W ( h)) WH, H, HN H, H, Pc ( h) n ( h) AC ( h) ( ) ( ). P c ( h) n ( h) W ( h) H, 1 WH, 1 WH, h WH, 1 h 1 1 H, H, 2 AC W WH, H, ( h) ( h) (6) Equaion (3) is he sandard Euler equaion for ineremporal consumpion choice. Equaions (4) and (5) represen, respecively, he demand for real currency and real bank deposis as a funcion of consumpion and he opporuniy cos. Equaion (6) deermines he dynamics of wage adjusmen in he presence of adjusmen coss. Noe ha in seady sae, ACWH ( h), AC W WH, H, ( h) ( h) and AC W WH, 1 H, ( h), and (6) simplifies o ( h) WH, ( h) HNnH, ( h) ch, ( h). P ( 1) The corresponding uiliy funcion and he budge consrain for household l are 8

10 1 1 1 s cl, s ( l) LC ( CU L, s( l) / Ps ) LNnL, s ( l) U L, () l E s 1 1 1, (7) c l l CU ( l) W ( l) n ( l)(1 AC ( l)) CU ( l) L, L, L, WL, L, 1 L, ( ) L, ( ), (8) P P where AC W ( l) / W ( l) L L, L, 1 WL, ( l) 1 2 P 1/ P2 2. Noe ha household l does no hold bank deposis or governmen bonds, and does no receive any profis. Uiliy maximizaion by he household subjec o he budge consrain and labor demand (also discussed below) implies ha he choice of L, L, CU ( l) and W ( l ) saisfies L, ( ), ( ) L L, 1( ) 1 P LC cl, () l E P 1 CU l c l E c l P, (9) (1 AC ( l))( 1) W ( l) n ( l) Pc ( l) ( W ( l)) WL, L, LN L, L, Pc () l n ( l) AC ( l) ( ) ( ). P c ( l) n ( l) W ( l) L, 1 WL, 1 WL, l WL, 1 l 1 1 L, L, 2 AC W WL, L, () l () l (1) Noe ha household l 's demand for real currency (9) is of a differen form han ha of household h because he wo households face differen opporuniy coss. In seady sae, he wage-seing equaion (1) also simplifies o WL, ( l) LNnL, ( l) cl, ( l). Since he household canno borrow P ( 1) or lend, consumpion, cl, () l, is deermined by he consrain (8). 2.2 Banks 9

11 The specificaion of he banking secor is based on Canzoneri e al. (28). In heir paper, bank loans finance a fixed amoun of loans o households. We assume ha bank loans are also used o finance invesmen. Deposi creaion and producion of loans by a bank is deermined by he following liquidiy and monioring funcion: DH, CR B B, BD P P P 1, (11) L B, P n, (12) BL HB, where CR and B, B represens cash reserves and governmen bonds held by banks, L B, are bank loans, and n HB, is a bundle of labor services of H ype households defined as 1 ( 1)/ nhb, n HB, ( h) dh /( 1). We assume, for simpliciy, ha banks employ only H ype households. The balance shee of he banking secor is given by CR BB, LB, DH,. (13) P P Express he discouned value of profis for a bank as E CR (1 R ) B (1 R ) L (1 R ) D W s s B, s L, s B, s D, s H, s H, s, s n s HB, s E Ps 1 Ps, where s, denoe 1/(1 ) 1 1 he discoun facor, and WH, W H, ( h) dh represens he wage rae for he labor bundle. Banks choose he raios, BB, / DH, and CU H, / D H,, o maximize his value subjec o 1

12 he balance-shee consrain (13), and he liquidiy and monioring relaions (12) and (13). The opimal choice by banks implies ha B B, ( P / E P 1 )( RL, RD, ) WH, / ( BLP ) (1 ), (14) DH, ( P / E P 1 )( RL, R ) WH, / ( BLP ) CR ( P / E P 1 )(,, ), / (, ) RL RD WH BL P, (15) DH, ( P / E P 1 )(1 RL, 1) WH, / ( BL, P ) where WH, / ( BL, P ) represens he marginal cos of making a loan (in real value). As (14) and (15) show, he securiies o deposis and he cash reserves o deposis raios are influenced by ineres rae spreads and he expeced inflaion rae. 2.3 Capial goods producers and capial leasing firms We assume a sandard model of invesmen where capial producers make addiions o insalled capial in he presence of adjusmen coss. However, o relae invesmen o bank loans, we inroduce capial leasing firm who require bank loans o finance purchases of addiional insalled capial from capial producers. Le k represen he insalled capial sock a he beginning of period, and i invesmen in he period. In each period, capial goods producers buy previously insalled capial (afer depreciaion), k, from capial leasing firms, produce and sell new insalled capial, k 1 k i. Invesmen is subjec o he following adjusmen cos: AC I, I i 2 k 2 k (16) 11

13 where is he depreciaion rae. Alhough he price of a uni of i is P (he same as he price of a uni of c ), he price of a uni of insalled capial is differen because of capial adjusmen coss and is denoed by Q.Capial goods producers maximize he discouned value of profis equal o E, s Qs ( ks 1 ks) Ps ( is ACI, s) s. Subsiuing i for k 1 k, and using (16), he firs order condiion for his problem is Q P i 1I k. (17) Capial leasing firms ren insalled capial o firms producing he final good. In each period, hey disribue income from previously insalled capial o H households, and finance purchase of addiional insalled capial [ Q ( k 1 k )] by a loan from banks. Their profis from he acquisiion of addiional insalled capial in period 1 are [ E RE 1 (1 ) EQ 1 ] i (1 RL. ) Qi, where RE denoed he renal rae for a uni of capial. The opimal choice for invesmen saisfies E RE (1 ) E Q. (18) RL. Q Capial accumulaes as k 1 i (1 ) k, (19) and invesmen is linked o bank loans as Q i L L, (2) B, where L is he fixed amoun of loans o all households. 2.4 Composie good producers 12

14 Assume ha he composie good (used by household, governmen and invesors) is a CES bundle of home and foreign varieies produced by a coninuum of home firms indexed by f (,1), and foreign firms indexed by * f (,1). Leing z represen he amoun of he composie good, we have z c c i g, (21) H, L, /( 1) 1/ ( 1)/ 1/ ( 1)/ z (1 ) ( zd, ) ( zm, ), (22) 1 /( 1) ( 1)/ ( ) D z z f df, (23) D,, 1 /( 1) * ( 1)/ * ( ) M z z f df. (24) M,, I follows ha he price of he composie good and he demand for he domesic and impored varieies is (1 ) (1 ) 1/(1 ) (1 )( D, ) ( M, ) P P P (25) z (1 ) z ( P / P), z z ( P / P) (26) D, D, M, M, z ( f ) z ( P ( f ) / P ), z ( f ) z ( P ( f ) / P ) (27) * * D, D, D, D, M, M, M, M, Similarly, he foreign demand for expored variey is given by z z ( P / P ), z ( f ) z ( P ( f ) / P ) (28) * * * * * * X, X, X, X, X, X, 13

15 Le S denoe he (rupee/dollar) exchange rae. The prices of impored and expored varieies in he home and foreign markes are linked as P ( f ) S P ( f ), (29) * * * M. M, P ( f ) S P ( f ). (3) * X. X, The home variey of he composie good is produced according o he following producion funcion: y ( f ) n ( f ) n ( f ) k ( f ) (31) H L 1 H L y, HY, L, where n HY, and n L, are bundles of labor services defined as /( 1) 1 ( 1)/ nhy, n HY, ( h) dh, and /( 1) 1 ( 1)/ nl, n L, () l dl. The opimal choice of inpus implies he following demand funcions: n ( f ) y ( f ) MC / W, (32) HY, H H, n ( f ) y ( f ) MC / W, (33) L, L L, k ( f ) (1 ) y ( f ) MC / RE, (34) H L 14

16 1/(1 ) 1 1 where WL, W L, () l dl is he wage rae for he L ype labor bundle, and MC is he marginal cos of he composie oupu. Define nh, nhy, nhb,. The demand funcions for labor services of households h and l [used in seing WH, ( h) and WL, ( l ) in (6) and (1)] are n ( h) n ( W ( h) / W ), n ( l) n ( W ( l) / W ). (35) H, H, H, H, L, L, L, L, Oupu of he variey of a composie good equals y ( f ) z ( f ) z ( f ). (36) D, X, We assume he following adjusmen coss for domesic and expor prices: AC P ( f ) / P ( f ) P D, D, 1 PD, ( f ) 1 2 P 1/ P2 2 and AC P ( f ) / P ( f ) P X, X, 1 PX, ( f ) 1 2 P 1/ P2 2. Firms choose PD, ( f ) and PX, ( f ) o maximize he discouned value of he profis, s, s ( PD, s( f ) MCs ) zd, s( f )(1 ACPD, s( f )) ( PX, s( f ) MCs ) zx, s( f )(1 ACPX, s( f )), subjec o demand funcions in (27) and (28). Noing ha, 1,, 1 1/ (1 R ), he opimal prices are (1 ACPD, ( f )) ( 1) PD, ( f ) MC PD, ( f )( PD, ( f ) MC )( ACPD, ( f ) / PD, ( f )) z ( f) (37) P ( f )( P ( f ) MC )( AC ( f ) / P ( f )), D, 1 D, D, 1 1 PD, 1 D, (1 R ) zd, ( f ) (1 ACPX, ( f )) ( 1) PX, ( f ) MC PX, ( f )( PX, ( f ) MC )( ACPX, ( f ) / PX, ( f )) z ( f) (38) P ( f )( P ( f ) MC )( AC ( f ) / P ( f )). X, 1 X, X, 1 1 PX, 1 X, (1 R ) zx, ( f ) 15

17 In seady sae, boh prices are he same and equal marginal cos muliplied by a markup facor: PD, ( f ) PX, ( f ) MC. 1 In symmeric equilibrium, z ( f ) z, z ( f ) z, z ( f ) z, y ( f ) y ; * M, M, D, D, X, X, * PM, ( f ) PM,, PD, ( f ) PD,, PX, ( f ) PX, ; nh, ( h) nh, nhy, nhb,, nl, ( l) nl, ; and W ( h) W, W ( l) W. Finally, he curren accoun balance condiion is H, H, L, L, P z P z CF, (39) M, M, X, X, where CF is an exogenous ne capial inflow (including remiances). Assuming ha he home economy is small, foreign variables z, P and P are exogenous. * * * M, 2.5 Moneary and fiscal policy Define BP, BH, BB, as governmen bonds held in he privae secor, CU CU H, CU L, as currency held by public, and MB CU CR as he moneary base. The governmen s flow budge consrain is BP P ( g H, L, ) ( MB MB 1) (1 R 1 ) BP, 1, (4) where i is assumed ha he governmen does no pay ineres o he Sae Bank (i.e., ineres income from he Bank's holding of governmen securiies is ransferred o he governmen). Real seignorage equals ( MB MB 1) / P. We assume ha he long-run real seignorage is deermined by he fiscal auhoriy, bu we disinguish wo policy regimes. In he firs regime, fiscal policy adjuss axes o sabilize deb a 16

18 some arge level, and moneary policy can use a Taylor-ype rule wih an inflaion arge given by he long-run inflaion rae deermined by real seignorage. We call his regime seignorageconsrained moneary policy. In he second regime, fiscal policy does no adjus axes in response o deb growh and moneary policy aemps o sabilize deb. This regime is referred o as fiscal dominance. The firs regime is described by he following ax and ineres-rae rules: B B,, (41) P, 1 P H, H P 1 P ln(1 R ) ln(1 R) ln( / ) ln( y / y) ln,,, (42) r ry r, r ry where 1 R (1 r ) E P 1 / P is he gross nominal ineres rae, 1 R (1 r), P / P 1 (so ha he inflaion rae equals 1), r, is a moneary policy shock, and an overbar over a variable denoes he value in seady sae. In he ax rule, only he axes for ype H households are assumed o be adjused. In he second regime, here is no ax rule followed by he fiscal auhoriy and he ineresrae rule is modified o include reacion o deb growh as follows: B P, 1 P ln(1 R ) ln(1 R) r ln( / ) ry ln( y / y) rb lnr, P 1 P B. (43) Noe ha he signs of he coefficiens in (43) are no resriced o be posiive as his resricion may no longer be feasible under fiscal dominance. 2.6 Shocks 17

19 We consider four shocks in he model: hree inernal and one exernal shocks. The inernal shocks include shocks o real governmen expendiures ( g ), oal facor produciviy ( y, ) and o moneary policy rule ( r, ). The exernal shock is a shock o real foreign price of impors ( P * * M, / P ). Each shock is assumed o follow an AR (1) process, and he equaions for he variables subjec o shocks are given by ln g (1 g )ln g g ln g 1 xg,, (44) ln (1 )ln ln x, (45), y, y y y y, 1 y, ln r, (1 r )lnr r lnr, 1 xr,, (46) * * * * * * ln( PM, / P ) (1 PM )ln( PM / P ) PM ln( PM, 1 / P 1 ) xpm,, (47) where xg,, xy,, xr,, and x pm, are whie-noise shocks. As he model variables in nominal values are non-saionary, hey were convered o real values o obain a seady sae soluion for he model. The real version of he model is summarized in Appendix A. 3. Calibraion o Pakisan's Economy The values of a number of model parameers were chosen by calibraing he model o he daa for Pakisan's economy. Table 1 shows he average annual values for key financial and macro variables for Pakisan. The seady-sae values of model variables were mached wih his daa by appropriae choice of cerain parameers. The uni of ime in he model equals a quarer, and (where needed) he daa in he able were convered o quarerly values. The seady-sae 18

20 quarerly inflaion rae is assumed o be equal o 3% (12% annual rae). Survey daa suggess ha abou 7% of he non-agriculural workers are in he informal secor. Considering he disribuion of workers beween he informal and formal secors as a rough indicaor of he relaive size of L and H ype households, we le n 2n in seady sae. Household of ype L L H earn a lower wage, and we assume ha w w /3. We do no have informaion on wha L H proporion of oal currency is held by he wo ypes of households (or in informal and formal secors). We iniially assume ha he share of oal currency held by L ype households roughly corresponds o heir relaive wage income. The values of model parameers ha were no deermined by calibraion or he assumpions discussed above were seleced from oher sudies. The quarerly value of he real ineres rae (which deermines he discoun facor, ) is ypically assumed o equal.1 in DSGE models and we use his value in our model. 2 The value of he risk aversion parameer ( ) is generally assumed o be close o one in a number of recen DSGE models for emerging economies and we le i equal The inverse of he elasiciy of labor supply ( ) is se equal o 2., which is wihin he range of values assumed in hese models. The elasiciy of real currency and real bank deposis wih respec o real consumpion ( /, / ) are assumed o equal one, 4 The subsiuion elasiciies beween domesic and foreign goods is assumed o be 2., which is consisen wih he range of values ypically used in open economy macro models. The elasiciies of subsiuion beween varieies of he (home or foreign) differeniaed good and 2 Ahmed, Haider and Iqbal's (212) esimaes sugges ha he long-erm real ineres is lower in Pakisan. A lower value of his variable would no make a major difference o our resuls. 3 For example, see Ahmad e al.'s (212) model for Pakisan, Gabriel e al.'s (21) model for India, Shaari's (28) model for Malaysia, and Casro e al.'s (211) model for Brazil. 4 This is similar o he assumpion of a uniary income elasiciy of money demand. Also noe ha parameers and can be inerpreed as he inverse of he elasiciy of real currency and real bank deposis wih respec o heir opporuniy cos. 19

21 differeniaed labor services (of he wo households ypes), and, are assumed o equal 6., which implies a markup of 2%. Sudies of he frequency of changes in prices and wages in Pakisan sugges ha prices are less sicky han wages, and he degree of sickiness for boh variables may be less han in developed economies. In he Roemberg model of wage-price adjusmen used in his paper, he degree of sickiness depends on P for prices, and on H and for wages. A value of 4 for L hese parameers, which is roughly equivalen o a four-quarer average conrac lengh in a Calvo-ype model, is ypically assumed in DSGE models. We use a lower value of 2 for and, and a furher-reduced value of 1 for. H L P In equaions (44)-(47), we se he auoregressive coefficiens equal o.9. We se he sandard deviaions of he whie noise shocks o governmen expendiures and impor prices ( x g, and x pm. ) equal o.5. 5 We se he sandard deviaion of he whie noise shock o produciviy ( x y, ) equal o.25 o bring he variabiliy of oupu closer o he variabiliy of real GDP in Pakisan. In our sochasic simulaions comparing he performances of differen moneary policy rules, we consider rules wihou shocks and se he moneary policy shock ( x r, ) equal o zero. 4. Moneary Policy Analysis In analyzing moneary policy in his secion, we focus on a policy regime where moneary policy is consrained by fiscal needs o raise some revenue from money creaion. The fiscal auhoriy, however, adjuss axes o sabilize deb in he long run. This regime is 5 These values are suggesed by he derended annual daa on real governmen expendiures and foreign impor price since The auoregressive coefficien in he annual daa is lower, bu we use a higher value for our quarerly model. 2

22 represened by ax and ineres rae rules (41) and (42) wih he long-run inflaion arge in he ineres rae rule deermined by fiscal policy. We firs examine he ransmission mechanism of he model by deriving he dynamic response of key macro variables o differen shocks under baseline policy rules. Nex, we explore desirable rule by comparing he performance of he alernaive rules. Finally, we discuss moneary rules in he case of fiscal dominance where he fiscal auhoriy does no reac o changes in governmen deb and he cenral bank adjuss he ineres rae o sabilize governmen deb. This moneary policy regime is represened by he ineres rae rule (43). 4.1 Transmission Mechanism in he Baseline Case For he baseline case, we consider a moneary policy rule in which he ineres rae reacs only o inflaion and exhibis considerable ineria. We se he auoregressive coefficien ( rr ) equal o.9, he inflaion coefficien ( r ) equal o.5, and he oupu gap coefficien ( ry ) equal o zero. This rule is compared laer wih variaions ha allow smaller ineria, less or more aggressive response o inflaion and reacion o oupu gap. In he baseline ax rule, we assume a weak response of axes o deb growh and le he deb coefficien in he rule ( ) equal o.25. b To illusrae he ransmission mechanism for moneary policy, we race he dynamic effecs of a one-quarer shock o he ineres rae rule ha produces a one percenage poin reducion in he nominal ineres rae (expressed as a percenage rae on an annual basis) in he firs quarer. The ineres rae adjuss in he following quarers according o he ineres rae rule. The behavior of he ineres rae and he dynamic responses of seleced macro variables over 15 quarers are displayed in Figure 1. In he presence of nominal rigidiies (based on adjusmen 21

23 coss for wages and prices), he reducion in he nominal ineres rae also lowers he real ineres rae and emporarily increases boh he inflaion rae (expressed as a annual percenage rae) and he oupu gap (defined as a log deviaion from he seady sae level). The inroducion of he banking secor in he model allows moneary policy o influence ineres rae spreads. To explore his channel, he figure also shows he response of he real spread beween he bank loan rae and he ineres rae as well as he raio of bank loans o deposis. The figure shows ha he ineres rae reducion emporarily lowers he real loan rae spread and increases he loans o deposi raio. Alhough he loan-deposi raio increases by abou one poin iniially, he decrease in he real loan rae spread is quaniaively very small. To see wheher moneary policy affecs low- and high-income households differenly, he response of he consumpion and employmen of low-income relaive o high-income households is also exhibied in he figure. In he shor run, he relaive consumpion of low-income households increases marginally because of liquidiy consrains which make he consumpion of lowincome households respond more srongly o changes in income. The effec on he relaive employmen, on he oher hand, is negligible We nex examine he effecs of a one-quarer shock o real governmen expendiures equal o 5%. 6 The dynamic responses o his shock are shown in Figure 2. The fiscal shock leads o a small increase in boh he inflaion rae and he oupu gap in he shor run. One concern abou he increase in governmen expendiures is ha hey could crowd ou privae invesmen. The figure shows ha invesmen indeed decreases in he shor run, and his decrease offses 6 The magniude of his shock implies an iniial increase in he real value of governmen expendiure roughly equal o 1% of GDP. 22

24 much of he increase in governmen expendiures. The loan-deposi raio decreases as well. The fiscal shock increases he relaive consumpion of low-income household only marginally. Figure 3 illusraes he effecs of a one-quarer shock o produciviy equal o -1%. Adverse supply shocks are ofen hough o be a major source of inflaion. In he presen model, however, he negaive produciviy shock has lile effec on he inflaion rae because of nominal rigidiies. The shock does have a srong negaive impac on oupu. I also leads o a significan decrease in invesmen and loan-deposi raio. Ineresingly, he effecs of he negaive produciviy shock on hese variables are similar o hose of a posiive shock o governmen expendiures. Finally, he ransmission of a one-quarer shock o foreign impor price equal o 5% is exhibied in Figure 4. This shock emporarily increases he inflaion rae by abou ½ percenage poin. The effec of he impor price increase on inflaion is dampened somewha by he moneary policy response ha raises he ineres rae. The impor price increase does lead o a more significan negaive effec on oupu and invesmen. The model suggess ha adverse shocks o produciviy and impor prices are no an imporan source of inflaion. However, hey have a significan negaive impac on economic aciviy. 4.2 Performance of Differen Moneary Policy Rules To explore wha kind of moneary rules would be desirable, his secion compares he performance of he baseline rule wih alernaive rules incorporaing differen responses o inflaion, differen degree of ineria and a reacion o oupu gap. We compue wo ypes of performance indicaors from sochasic simulaions, in which he economy is subjeced o shocks o governmen expendiures, produciviy and impor prices. The indicaors of he firs ype are 23

25 based on he radiional approach, in which losses arise from he variabiliy of inflaion and oupu around heir arge values. The indicaors of he second ype use a welfare crierion based on household uiliy. Welfare effecs are ypically discussed for a represenaive household. In he presen model, however, we can examine wheher he welfare effecs are differen for he low-income (L) and high-income (H) households. The resuls are presened in Table 2. Firs, we consider a rule wih an inflaion coefficien equal o.15, which is less aggressive in fighing inflaion han he baseline rule. 7 The lessaggressive rule increases he variabiliy of inflaion (as measured by he sandard deviaion of he gap beween he acual and he fixed arge value)), bu decreases he variabiliy of oupu (measured in he same way). This rule also worsens he welfare of boh L and H households. To explore wheher a more aggressive ani-inflaion rule would perform beer han he baseline rule, we also show he resuls for a rule wih inflaion coefficien equal o.85. This rule lowers he variabiliy of inflaion and improves he welfare of boh households, bu he quaniaive improvemen over he baseline rule is relaively small. The able also shows ha a rule wih less ineres rae ineria (an auo-regressive coefficien of.6 insead of he baseline value f.9) would, like he less aggressive ani-inflaion rule, increase he inflaion variabiliy, reduce he oupu variabiliy and worsen welfare of each household. We nex examine he performance of rules ha also reac o he oupu gap. A rule wih an oupu coefficien equal o.25 (and he same inflaion and auo-regressive coefficiens as he baseline case) significanly reduces oupu variabiliy, bu a he cos of a subsanial increase in variabiliy of inflaion. This rule also affecs he wo households differenly: relaive o he 7 This value is close o he inflaion coefficien in an ineres rae rule esimaed for Pakisan using daa for a long period. 24

26 baseline case, he L household looses while he H household gains. These effecs are srenghened if he oupu coefficien is increased o Fiscal Dominance We now consider a fiscal regime ha does no adjus is primary surplus in response o growh of governmen deb and relies on borrowing from he cenral bank o mee is fiscal needs. In order o avoid acceleraion of money growh and inflaion under hese condiions, moneary policy aemps o sabilize governmen deb via ineres rae changes. For his case, we represen moneary policy by he ineres rae rule (43), which includes governmen deb as an argumen. In his rule, he real ineres rae mus decrease in response o an increase in real governmen deb in order o reverse deb growh by lowering ineres paymens. Figure 5 illusraes some feasible ineres rae responses o an increase in deb brough abou by a one-quarer shock o governmen expendiures equal o 5%. Par (a) of he figure shows he dynamic effecs of his shock when he ineres rae rule responds negaively o real governmen deb and does no reac o inflaion. We le, and.1. This rule brings abou he necessary reducion in he real ineres rae, bu his adjusmen also involves a sharp increase in inflaion and oupu. In fac, he iniial increase in inflaion is sufficienly large (is over 5%) o cause governmen deb o decrease in real erms (and he nominal ineres rae o increase) iniially. We nex examine an ineres rae rule ha sill responds negaively o real governmen deb, bu also responds posiively o inflaion. For his case, we le.5, and.1. 8 Par (b) of he figure shows ha his rule decreases he impac on r rb inflaion and oupu, bu i now akes longer for he real governmen deb o converge o is arge r rb 8 The posiive response of he ineres rae o inflaion, however, canno be so large ha i prevens he needed real ineres rae adjusmen. 25

27 value. Finally, we also consider a rule ha reacs only o inflaion. In his case, he response o inflaion has o be negaive and sufficienly srong o sabilize real governmen deb. We now se.75, and. The dynamic response of inflaion and oupu under his rule is large as r rb in he case of he firs rule (see par (c) of he figure). As hese examples sugges, ineres rae rules ha sabilizes real governmen deb can lead o greaer inflaion variabiliy. In sochasic simulaions where he economy faces shocks o produciviy and impor prices as well as o governmen expendiures, we find ha inflaion variabiliy under all hree rules discussed above is much higher as compared o rules ha do no arge governmen deb. 9 Successful implemenaion of moneary policy requires, moreover, ha i is well undersood and is credible. These condiions would be more difficul o mee for moneary rules designed o sabilize governmen deb, which require large flucuaions in inflaion and real ineres raes. 5. Conclusions The paper develops a Dynamic Sochasic General Equilibrium Model o analyze moneary policy in Pakisan. A his ime, moneary policy is consrained by he needs of he governmen o borrow from he Sae Bank. Facing his consrain, he Sae Bank canno independenly deermine an inflaion arge. However, if he long-erm needs of he governmen are clearly esablished and fiscal policy akes seps o sabilize governmen deb o GDP raio around a feasible arge level, moneary policy can follow an ineres rae rule o sabilize inflaion (around a long-run rae deermined by fiscal needs) and oupu. For such policy regime, 9 The sandard deviaions of inflaion for he firs, second and hird rule are, respecively,.725,.82 and.93. The corresponding sandard deviaions for oupu are.128,.91 and

28 we use he model o examine macroeconomic adjusmen o various shocks and compare he macroeconomic performance of alernaive moneary policy rules. The evidence in Pakisan suggess ha wages and prices, especially he laer, are changed more frequenly han developed economies. Even under he assumpion of less wage and price sickiness, our analysis shows ha moneary policy exers significan real effecs hrough ineres rae changes. We also examine he macroeconomic effecs of changes in governmen expendiures, and find ha hey lead o significan crowding ou of privae invesmen. We find, moreover, ha supply shocks conribue lile o inflaion, bu hey do have an imporan impac on oupu. The model is also used o explore how ineres raes should respond o inflaion and oupu. Esimaion of he ineres rae rule based on pas daa for Pakisan suggess a weak response o inflaion (an inflaion coefficien close o.1). Our resuls show ha a sronger response (an inflaion coefficien of.5 or more), would significanly reduce inflaion variabiliy and improve he welfare of boh low- and high-income households. We also examine he appropriae ineres rae response o oupu. Our analysis suggess a rade off beween inflaion and oupu variabiliy, which has implicaions for he welfare of households a differen income levels. For example, a posiive response o oupu gap causes a decrease in oupu variabiliy, bu also leads o an increase in inflaion variabiliy. This policy also affecs households a high and low income levels differenly: high-income households gain while low-income households loose. The paper also considers an alernaive policy regime in which fiscal policy does no aemp o sabilize governmen deb. In his case, moneary policy is consrained furher by he need o use ineres raes o conrol he growh of governmen deb. Feasible moneary rules 27

29 under his consrain are found o produce much larger variabiliy in inflaion. These rules may also be more difficul o implemen because hey involve ineres rae adjusmens ha migh be misundersood and no considered credible. The model in he paper can be exended and modified o invesigae a wide range of policy issues. Some valuable heoreical exensions would include developing a model of ne capial inflows, inroducing furher financial fricions, and exploring differen models of price expecaions. There is also a need for empirical work o idenify imporan empirical regulariies in Pakisan and verify how well he model explains hese regulariies. Such work could lead o modificaions of he model ha would improve he model and make i more useful for policy analysis. 28

30 References Ahmad, S., Ahmed, W., Pasha, F., Khan, S., and Rehman, M., 212, "Pakisan Economy DSGE Model wih informaliy," SBP Working Paper No. 47. Ahmed, Waqas, Adnan Haider, and Javed Iqbal, 212, "A Guide for Developing Counries on Esimaion of Discoun Facor and Coefficien of Relaive Risk Aversion," mimeo, Sae Bank of Pakisan. Bernanke, Ben, Mark Gerler and Simon Gilchris, 1999, The Financial Acceleraor in a Quaniaive Business Cycle Framework, in Taylor, John B. and Michael Woodford (eds.), Handbook of Macroeconomics (Amserdam, New York and Oxford: Elsevier Science, Norh-Holland), vol. I C, Canzoneri, Mahew, Rober Cumby, Behzad Diba, and David Lopez-Salido, 28, Moneary Aggregaes and Liquidiy in a Neo-Wicksellian Framework, Journal of Money, Credi and Banking, 4, Casro, M. R., Gouvea, S. N., Minella, A., Sanos, R.C., and Souza-Sibrinho, N. F., 211, "SAMBA: Sochasic Analyical Model wih a Bayesian Approach," Banco de Brazil, Working Paper, April 211. Clarida, R., J. Gali, and M. Gerler, 1999, The Science of Moneary Policy: A New Keynesian Perspecive, Journal of Economic Lieraure, 37, Gabriel, V., Levine, P., Pearlman, J., and Yang, B., 21, "An Esimaed DSGE Model for Indian Economy," Working Paper, Universiy of Surrey, DP No. 12/1. Gali, J., Lopez-Salido, J. D., Valles, J., 27, "Undersanding he effecs of governmen spending on consumpion," Journal of he European Economic Associaion 5 (1), Goodfriend, Marvin, and Benne T. McCallum, 27, Banking and Ineres Raes in Moneary Policy Analysis: A Quaniaive Exploraion, Journal of Moneary Economics 54: Kumhof, M., R. Nunes, and I. Yakadina, 28, "Simple Moneary Rules under Fiscal Dominance," Board of Governors of he Federal Reserve Sysem, Inernaional Finance Discussion Papers No Laxon, Douglas and Paolo Peseni, 23, "Moneary rules for small, open, emerging economies," Journal of Moneary Economics, 5, Roemberg, Julio J., 1982, Sicky Prices in he Unied Saes, Journal of Poliical Economy, Vol. 9, pp

31 Sargen, T. J. and Wallace, N., 1981, "Some Unpleasan Moneary Arihmeic," Federal Reserve Bank of Minneapolis, Quarerly Review, Fall, Shaari, M. H., 28, An esimaed DSGE Model for Malysian Economy: Full Sample Period 1975Q1:25Q2, PhD Disseraion, ANU Woodford, M., 21, Fiscal Requiremens for Price Sabiliy, NBER Working Paper, No Woodford, Michael, 23, Ineres and Prices: Foundaions of a Theory of Moneary Policy (Princeon: Princeon Universiy Press). 3

32 Appendix A Real Model Noe: The model is ransformed o a real model by convering nominal values in he home economy o real values in erms of he home composie good. Real values are denoed by lower case leers. The gross real ineres rae is relaed o he gross nominal ineres rae as 1 r (1 R ) P / E P 1, and P / P 1. A sar denoes foreign value. Foreign real values are expressed in erms of he foreign composie good. s S P P denoes he real exchange rae. * ( / ) Households c (1 r ) E c (A1) H, H, 1 d H, ch, r rd, HD, 1 r (A2) cu H, ch, r 11/ E 1 HC, 1 r (A3) c w n cu / cu (A4) L, L, L, L, 1 L, L, cu L, c E c / E L, L, LC, cl, (A5) Producion and invesmen y ( n ) ( n ) ( k ) (A6) H L 1 H L F, HY, L, k (1 ) y mc / re (A7) H L n y mc / w (A8) HY, H H, 31

33 n y mc / w (A9) L, L L, k 1 i (1 ) k (A1) E re (1 r ) q (1 ) E q (A11) 1 L. 1 q i 1 I k (A12) Foreign secor y z z (A13) D, X, M, p D, 1 p (1 ) (A14) z z p (A15) M, M, z (1 ) z p (A16) D, D, z z ( p ) (A17) * * X, X, p s p (A18) * M, M, p s p (A19) * X, X, p z p z fc (A2) M, M, X, X, Banking secor 32

34 cr ( r r ) w d r E w L, D, H, BL, H, (1 L, 1/ 1 ) H, / BL, (A21) b B, ( rl, rd, ) wh, / BL, (1 ) dh, ( rl, r ) wh, / BL, (A22) d cr b 1 H, BD, ( ) ( B, ) (A23) cr b l d (A24) B, B, H, l n (A25) B, BL, HB, Price and wage seing (1 AC ) ( 1) p mc p ( p mc ) dac zd, 1 pd, ( pd, 1 mc 1 ) dac1 PD, (1 r) z PD, s D, D, D, PD, D, (A26) dac AC P P D, PD, ( PD, / D, ) P 1 pd, 1 1 pd, 1 1 p (A27) dac1 ( AC / P p ) P p 1 D, 1 1 D, 1 1 PD, PD, 1 D, P 2 pd, pd, (A28) AC PD, P 2 p p D, D, (A29) (1 AC ) ( 1) p mc p ( p mc ) dac zx, 1 px, ( px, 1 mc 1 ) dac1 PX, (1 r) z PX, s X, X, X, PD, X, (A3) 33

35 dac AC P P X, PX, ( PX, / X, ) P 1 px, 1 1 px, 1 1 p (A31) dac1 ( AC / P p ) P p 1 X, 1 1 X, 1 1 PX, PX, 1 X, P 2 px, px, (A32) AC PX, P 2 p p X, X, (A33) 1 n (1 AC )( 1) w n c w dac w w dac1 (A34) 2 H, 1 WH, H, HN H, H, H, WH, H, H, 1 WH, 1 r nh, dac AC P w 1 WH, H, WH, H W H, wh, 1 1 wh, 1 1 (A35) dac1 AC w P w 1 WH, 1 H, 1 1 H, 1 1 WH, H 2 W H, 1 wh, wh, (A36) AC WH, H 2 w w H, H, (A37) 1 n (1 AC )( 1) w n c w dac w w dac1 (A38) 2 L, 1 WL, L, LN L, L L, WL, L, L, 1 WL, 1 r nl, dac AC P w 1 WL, L, WL, L W L, wl, 1 1 wl, 1 1 (A39) dac1 AC w P w 1 WL, 1 L, 1 1 L, 1 1 WL, L 2 W L, 1 wl, wl, (A4) 34

36 AC WL, w L 2 w L, L, (A41) Equilibrium condiions q i l l (A42) B, n n n (A43) H, HY, HB, z c c i g (A44) H, L, bp g H, L, ( mb mb 1 / ) (1 r 1 ) bp, 1 (A45) mb cu cu cr H, L, (A46) Basic moneary and fiscal rule Regime 1 y ln(1 R) ln 1 ln ln ln y R, r ry R (A47) 1 R E (1 r ) (A48) 1 H, H b bp, 1 bp. (A49) Regime 2 35

37 Replace (A47) by (A47') and drop (A49). y ln(1 R ) ln 1 R ln ln b b ln y r ry rb P, 1 P. R, (A47') Shocks ln g (1 g )ln g g ln g 1 xg, (A5) ln Y, (1 Y )lny YY, 1 xy, (A51) ln r, (1 r )lnr rr, 1 xr, (A52) * * * ln pm, (1 PM )ln pm PM pm, 1 xpm, (A53) Endogenous variables: AC, AC, AC, AC, b, b, c c, cr, cu, cu, dac, dac1, dac, dac1, PD PX WH WL B P H, L H L PD PD PX PX dac, dac1, dac, dac1, d, g, i, k, l, mb, mc, n, n, n, n, WH WH WL WL H B H HB HY L r r r p p p p p re q R w w y s * *, D, L, D, M, M, X, x,,,, H, L,,, z, z, z, z,,,, D M X H y r Exogenous variables: b cf g l z z x x x x x * P,,,,, L, H, ( * *), g, Y, r, pm, cf Parameers:,,,,,,,,,,,,,,, H L y BD BL,,,,,,,,,, HC HD HN F LC I H L P 36

38 Figure 1. Dynamic Effecs of a Temporary Reducion in he Ineres Rae 17 nominal ineres rae (%) 6 real ineres rae (%) inflaion rae (%) 15 1 real loan rae spread (%) L-H consumpion raio oupu (log dev.) loan-deposi raio L-H employmen raio Noe: The figure shows he response over 15 quarers (wih quarer 1 showing values for iniial seady sae). The ineres raes are convered o a percenage annual value and defined as: nominal ineres rae = R 4, real ineres rae = r 4, real loan rae spread = ( rl r) 4. Oher variables are defined as: oupu = ln( y/ y ), loan-deposi raio = lb / dh, L-H consumpion raio = cl / ch, and L-H employmen raio = nl / nh. 37

39 Figure 2. Dynamic Effecs of a Temporary Increase in Governmen Expendiures.5 gov. expendiures (log dev.) 16.3 nominal ineres rae (%) inflaion rae (%) invesmen (log dev.) L-H consumpion raio x 1-3 oupu (log dev.) -5 loan-deposi raio L-H employmen raio Noe: Governmen expendiures and invesmen variables are defined as: gov. expendiures = ln( g/ g ), and invesmen = ln( i/ i ). Oher variables are defined in he noe o Figure 1. 38

40 Figure 3. Dynamic Effecs of a Temporary Decline in Produciviy produciviy (log dev.) 16.3 nominal ineres rae (%) inflaion rae (%) invesmen (log dev.) L-H consumpion raio oupu (log dev.) loan-deposi raio L-H employmen raio Noe: The produciviy variable is defined as ln( / ). Oher variables are defined in he noe o Figure 1. y y 39

41 Figure 4. Dynamic Effecs of a Temporary Increase in Foreign Impor Price Index.1 impor price (log dev.) 16.4 nominal ineres rae (%) inflaion rae (%) invesmen (log dev.) L-H consumpion raio oupu (log dev.) loan-deposi raio L-H employmen raio Noe: The impor price index is defined as ln( m / m). Oher variables are defined in he noe o Figure 1. 4

42 Figure 5 Dynamic Effecs of a Temporary Increase in Gov. Expendiures under Fiscal Dominance (a) Ineres rae responds only o deb (,.1) r rb.5 gov. expendiures (log dev.).2 real deb (dev.) 2 17 nominal ineres rae (%) real ineres rae (%) inflaion rae (%) 2.1 oupu (log dev.)

43 Figure 5 (Coninued) (b) Ineres rae responds o boh inflaion and deb (.5,.1) r rb.5 gov. expendiures (log dev.).4 real deb (dev.) nominal ineres rae (%) 2 6 real ineres rae (%) inflaion rae (%) oupu (log dev.)

44 Figure 5 (Coninued) (c) Ineres rae responds only o inflaion (.75, ) r rb.5 gov. expendiures (log dev.).1 real deb (dev.) 2 2 nominal ineres rae (%) real ineres rae (%) inflaion rae (%) oupu (log dev.)

45 Table 1. Daa for Calibraion Descripion Average Annual Value Bank Deposi o GDP Raio.263 Currency o Deposi Raio.389 Cash Reserves o Deposis Raio.52 Governmen Securiies o Deposi Raio for Banks.61 Gov. Expendiures as Share of GDP.198 Invesmen Expendiures as a share of GDP.188 Rae of Capial Depreciaion.84 Share of Impors in GDP

46 Table 2. Performance of Differen Moneary Rules Inflaion Gap Oupu Gap Expeced Uiliy (Snd. Dev.) (Snd. Dev.) L Household H Household Baseline Rule (.9,.5 ) rr r Less Ani-Inflaion (.9,.15 ) rr r More Ani-Inflaion (.9,.85 ) rr r Less Ineria (.6,.5 ) rr r Reac o Oupu (.9,.5,.25 ) rr r ry Sronger Oupu Reacion (.9,.5,.5 ) rr r ry Noe: Inflaion and oupu gaps are defined as ln( / ) 4 and ln( y/ y ), respecively. Uiliy indexes for L and H households equal period uiliy levels given in (7) and (1). 45