Discussion of Reserve Requiremens for Price and Financial Sabiliy: When Are They Effecive? Carl E. Walsh Deparmen of Economics, Universiy of California, Sana Cruz Since he onse of he 2008 financial crisis, many economiss have sudied he effecs of unconvenional moneary policies such as credi-easing policies or acions aimed a alering he mauriy srucure of he privae secor s asse holdings. Bu surprisingly, here has been relaively lile work on one of he radiional policy ools of a cenral bank required reserve raios and how required reserves migh be used as a cyclical policy insrumen. Chrisian Glocker and Peer Towbin fill his gap by employing a modern dynamic sochasic general equilibrium (DSGE) model of a small open economy o analyze he use of reserve requiremens as a ool for cyclical sabilizaion. A one ime, we all learned ha cenral banks had hree policy insrumens he quaniy of non-borrowed reserves, he ineres rae charged on discoun window borrowing, and he required reserve raio. The firs wo insrumens operaed by affecing he supply of bank reserves; he hird worked by affecing he demand for reserves. Then, many cenral banks adoped a shor-erm inerbank rae as heir primary policy insrumen. Doing so made he quaniy of non-borrowed reserves an endogenous variable, adjusing o ensure reserve supply and demand were consisen wih he policy ineres rae. Under such a policy regime, he discoun rae and reserve requiremen became irrelevan. Of course, his wasn (and isn ) rue of developing economies. There, cenral banks coninued o Prepared for he Fall 2011 IJCB Conference on Moneary Policy Issues in Open Economics, hosed by he Bank of Canada, Oawa, Canada, Sepember 29 30, 2011. Auhor e-mail: walshc@ucsc.edu. 115
116 Inernaional Journal of Cenral Banking March 2012 employ reserve aggregaes and reserve requiremens as acive insrumens of moneary policy. However, mos policy-relaed research offered lile guidance o hese economies, as sandard DSGE models for policy analysis relied exclusively on ineres rae rules o represen policy. So an invesigaion ino he role of reserve requiremens in a modern modeling framework is a useful addiion o he sudy of moneary policy. In my commens, I firs give a brief overview of he paper and describe wha he auhors do. I hen focus on a simplified version of he financial secor of heir model o highligh how reserve requiremens affec he economy and why using hem, ogeher wih an ineres rae insrumen, can expand he se of oucomes achievable by moneary policy. Finally, I conclude by highlighing he parallels beween moneary policy ools and ax insrumens and offer some suggesed exensions o he Glocker-Towbin analysis. 1. Overview of Paper The core model used by Glocker and Towbin (henceforh, G-T) is a sandard small open-economy DSGE model wih nominal fricions. They add o his srucure a banking secor and financial fricions. Since he real side of he model is similar o oher small openeconomy DSGE models, l will focus on he srucure of he financial side of he model, as his srucure is criical for he ransmission of changes in reserve requiremen o he real economy. Three financial fricions characerize he economy. Firs, here is marke segmenaion; households he ulimae savers in he model canno lend direcly o enrepreneurs. Insead, households save by accumulaing deposis wih he banking secor or holding foreign bonds, and enrepreneurs are forced o obain credi from banks. This marke segmenaion is imporan because i means he imposiion of reserve requiremens on banks a form of ax on inermediaed credi canno be avoided, as i could be if households were able o lend direcly o enrepreneurs. Second, G-T assume here is a direc real resource cos associaed wih deposi banking. This cos depends on he banking secor s holdings of excess reserves. Specifically, deposi-aking insiuions face coss given by
Vol. 8 No. 1 Discussion: Walsh 117 G ς = ψ 1 ( ς ς MP ) + ( ψ2 2 ) (ς ς MP ) 2, (1) where ς is he raio of reserve holdings o deposis and ς MP is he required reserve raio. The deposi-aking banks hen lend a he inerbank rae i IB o loan-making banks who in urn lend direcly o enrepreneurs. Finally, a hird financial fricion arises due o he presence of agency coss associaed wih bank lending o enrepreneurs (Bernanke, Gerler, and Gilchris 1999). These agency coss drive a wedge beween he ineres rae on loans and he loan-making banks opporuniy cos of funds. A final complicaion inroduced ino he model is o compare cases in which loans o enrepreneurs are denominaed in foreign currency raher han in domesic currency. Given his srucure, here are hree ineres raes in he model: he rae paid o households on bank deposis, he rae deposi banks charge lending banks (he inerbank rae), and he lending rae. The spreads beween hese raes will depend on he srucure of compeiion for deposis, he coss associaed wih banking, and he agency coss associaed wih lending. Various policy regimes are considered. These are (i) a sandard regime in which only he inerbank rae is used as a policy insrumen; (ii) a regime in which boh he inerbank rae and reserve requiremens adjus o inflaion, oupu, and a financial variable (loan quaniy); (iii) a regime in which he nominal exchange rae is fixed bu reserve requiremens adjus o inflaion, oupu, and he quaniy of loans; and (iv) a regime in which he policy rae adjuss o inflaion and he oupu gap and reserve requiremens adjus o he quaniy of loans. G-T conduc hree primary experimens using a calibraed version of heir model. Firs, hey repor impulse responses o an increase in he required reserve raio for differen degrees of nominal price rigidiy, for differen policy regimes, and wih and wihou a financial acceleraor mechanism. Second, hey repor opimal coefficiens in policy rules for he inerbank rae and he reserve raio. Oucomes are evaluaed using ad hoc quadraic loss funcions involving inflaion, oupu, and, in some cases, he quaniy of loans as a proxy for financial sabiliy objecives. Third, hey show how
118 Inernaional Journal of Cenral Banking March 2012 he response of he economy o a echnology shock differs under alernaive policies. The mos ineresing conclusions emerging from his analysis are, I believe, he insighs i gives o he insrumen assignmen problem. G-T find ha he opimal coefficiens in a basic Taylor rule for he inerbank rae are very lile affeced when he reserve requiremen ς MP is allowed o respond opimally o inflaion and oupu (G-T s able 3). Nor is he loss funcion reduced very much by using ς MP as an insrumen. However, when policy also responds o he quaniy of loans, G-T obain some ineresing new resuls. Firs, even when he inerbank rae i IB is he only insrumen, reacing o loan quaniies significanly reduces he loss funcion (G-T s able 4). This resul is perhaps no surprising, as he model has five shocks (echnology, cos push, governmen expendiures, foreign ineres rae, and expor demand), so responding o more han inflaion and oupu should provide a beer approximaion o an opimal policy which would reac o each of he five fundamenal shocks. Second, for he case in which he financial acceleraor is absen, essenially he same loss is achieved if he inerbank rae reacs o inflaion, oupu, and loans as when he inerbank rae reacs only o inflaion and oupu and he required reserve raio reacs only o he quaniy of loans (heir regime IV). Tha is, here is a separaion of policy insrumens in ha he reserve raio can be moved in response o flucuaions in loans o achieve he cenral bank s financial sabiliy objecive while he inerbank rae is employed o achieve inflaion and oupu objecives. This second resul is srenghened when he financial acceleraor is added back in. In his case, he lowes value for he loss funcion is achieved when boh i IB and ς MP reac o inflaion, oupu, and loan quaniy, bu here is almos no deerioraion in loss if i IB responds only o inflaion and oupu while ς MP responds only o loan quaniy. This is an ineresing and novel resul. 2. Reserve Requiremens and Financial Equilibrium In his secion, I wan o focus on he financial secor of he G-T model o highligh how reserve requiremens work o affec he economy and why hey can provide a useful supplemen o an ineres
Vol. 8 No. 1 Discussion: Walsh 119 rae insrumen, a leas in principle. 1 Doing so helps one beer undersand Glocker and Towbin s resuls. Deposi-aking banks earn i R on heir reserve holdings and i IB on asses len. Given he cos funcion G in (1), he equilibrium spread beween he inerbank rae and he rae paid on reserves mus be such ha he ineres earned on reserve ne of he marginal cos of addiional reserve holding mus equal he rae earning on lending: i R Gς = i R [ ( ψ 1 + ψ 2 ς ς MP )] = i IB ς. Thus, reserve demand is given by ( i ς = ς MP R + i IB ψ 1 ψ 2 Define Δ as he spread beween he inerbank rae and he rae paid on reserve balances: Δ i IB i R. G-T assume he cenral bank mainains Δ equal o a consan Δ, wih Δ 0. Doing so requires ha he cenral bank adjus reserve supply so ha ( ) ς = ς MP Δ+ψ1, (2) where he second erm on he righ is a consan. While he spread beween he rae on reserves and he inerbank rae is consan, he spread beween he rae paid on deposis and he inerbank rae is deermined by he zero-profi condiion for deposi-aking banks: ψ 2 ). i D =(1 ς )i IB + ς i R G ς = i IB ς MP Δ G ς, (3) where G ς is he consan cos of servicing deposis. Using (2) in (1), ( ) ( )( ) 2 G ς Δ+ψ1 ψ2 Δ+ψ1 = ψ 1 + ψ 2 2 ψ 2 is a consan (given Δ). 1 A reduced-form version of he G-T financial secor is similar in many ways o he framework employed by Romer (1985) o sudy reserve requiremens.
120 Inernaional Journal of Cenral Banking March 2012 The equilibrium relaionship (3) is key o undersanding he way in which he reserve raio affecs he economy. Wih Δ 0, a rise in ς MP forces banks o hold more reserves and acs as a ax on deposis, reducing he ineres rae households receive on bank deposis. The deposi ineres rae appears in he household secor s Euler condiion, so as i D falls, he demand for deposis by households falls as households increase curren consumpion. Deposi-aking banks lend o loan-making banks a he inerbank rae. Thus, he opporuniy cos of funds o lenders is i IB in nominal erms. Le i L denoe he nominal ineres rae on loans o enrepreneurs. Then he Bernanke, Gerler, and Gilchris (1999) model delivers an equilibrium relaionship beween he spread beween hese wo raes as a funcion of enrepreneur ne worh, he supply price of capial, and he demand for loans by enrepreneurs. Wriing loan demand as L d (i L,i IB,Z ), where Z is a vecor of hese oher variables (including expeced inflaion), and leing D(i D ), D > 0 denoe he reduced-form household demand for deposis, he supply of loans will equal deposis ne of reserves (1 ς )D(i D ), and equilibrium in he loan marke can be summarized by (1 ς )D ( i D ) = L d ( i L,i IB,Z ). (4) Finally, reserve demand is ς D(i D ), so if H denoes reserve supply, equilibrium in he reserve marke requires ha H = ς D ( i D ). (5) Equaions (2) (5) consiue a sylized represenaion of he financial marke in a closed-economy version of Glocker and Towbin s model. These equaions consis of four equilibrium condiions involving i D, i IB, i L,Δ,H, ς, and ς MP, implying ha here are choices ha can be made wih respec o which variables he cenral bank uses as insrumens. There are four poenial insrumens: he inerbank rae i IB, he required reserve raio ς MP, he spread beween he inerbank rae and he rae charged on reserve borrowing Δ, and he quaniy of reserves H. Combining (2), (3), and (5) yields H = [ ς MP ( Δ+ψ1 ψ 2 )] D ( i IB ς MP Δ G ς),
Vol. 8 No. 1 Discussion: Walsh 121 which illusraes how, in he radiional analysis wih ς MP and Δ fixed, H mus adjus endogenously, once he cenral bank has se i IB, o ensure reserve supply (he lef side) equals reserve demand (he righ side). However, an alernaive would be o fix H and le he reserve requiremen adjus o be consisen wih he chosen values of he inerbank rae and he spread Δ. In he open economy, he nominal exchange rae becomes an addiional poenial insrumen (combined wih a new equilibrium condiion given by uncovered ineres pariy). Equaions (3) and (4) are key o undersanding he ransmission of reserve requiremens o he real economy under an ineres rae policy ha deermines i IB. From (3), i D = i IB ς MP Δ G ς, and so wih Δ fixed, i D ς MP = Δ 0. A rise in ς MP acs as a ax on he banking secor when he ineres paid on reserves is less han he inerbank rae (i.e., when Δ > 0). This ax is passed on o households in he form of lower ineres on bank deposis. A rise in ς MP lowers he reurn available o households and, from he household s Euler equaion, increases curren consumpion. Thus, one channel hrough which reserve requiremens affec he economy is via he reurn earned by households on heir holdings of bank deposis. How big is his consumpion channel likely o be? In he calibraion exercise performed by Glocker and Towbin, Δ = 0.015 (150 basis poins). Thus, a 50 percen rise in he reserve requiremen rae from 10 percen o 15 percen, for example would decrease he deposi ineres rae by Δ(.15.10)=0.00075, or 7.5 basis poins. Hence, a fairly large change in ς MP produces a very small change in he deposi rae and, correspondingly, is likely o have a small effec on consumpion. Even his small effec would disappear under a Friedman-rule policy which would se Δ = 0 o eliminae he ax on reserves.
122 Inernaional Journal of Cenral Banking March 2012 In addiion o he ax channel affecing consumpion, changes in he reserve raio also affec he real economy hrough he loan marke. Here, here is boh a direc and an indirec effec of ς MP on loan supply. Since he supply of loans is (1 ς )D(i D ), for a given level of deposis a rise in ς MP increases ς (see (2)) and reduces lending by he banking sysem. This is he direc effec on loan supply. The indirec effec occurs via he fall in i D caused by a rise in ς MP. This reduces he demand for deposis, shrinking he banking secor and furher reducing loan supply. As previously argued, he direc effec on i D is small, so his indirec effec on loan supply is also likely o be small. Boh direc and indirec effecs work in he same direcion, reducing loan supply. For a given policy rae i IB, he rae on loans mus rise and he quaniy of loans mus fall o mainain equilibrium in he loan marke (4). This ranslaes ino a decline in invesmen spending. This discussion of he channels hrough which changes in ς MP affec he economy serves o illusrae why ς MP has effecs ha differ from hose of he policy rae. A rise in i IB reduces boh consumpion and invesmen; a rise in ς MP reduces invesmen bu increases consumpion. Because ς MP and i IB affec consumpion differenly, using hem in combinaion expands he effecive number of insrumens available o he cenral bank. My discussion of he G-T financial secor ignored open-economy consideraion. The effecs of ς MP on he exchange rae and ne expors will depend on he policy regime adoped by he cenral as he pri- produced by a rise in he reserve raio leads o a depreciaion of he exchange rae and a rise in ne expors. Under a fixed exchange rae regime, i IB would have o rise o preven a depreciaion. Table 1 summarizes how macro variables are affeced by a change in he reserve requiremen. 2 As a baseline, column 1 gives he effecs of a rise in he inerbank rae, holding he reserve raio and he spread Δ fixed. Columns 2 and 3 show he effecs of a rise in he bank. Under a flexible exchange rae sysem, wih i IB mary policy insrumen, he fall in i D 2 In he able, S is he local currency price of foreign currency, so a rise denoes a depreciaion. The effecs are for he baseline calibraion; figure 1 of G-T shows ha he oupu response o ς MP under an i IB policy can change signs if prices are relaively flexible.
Vol. 8 No. 1 Discussion: Walsh 123 Table 1. Impac Effecs (1) (2) (3) i IB ς MP Policy Regime Variable i IB S Deposi Rae: i D Loan Rae: i L Consumpion: C Invesmen: I Exchange Rae: S Ne Expors: NE Oupu: Y required reserve raio under differen policy regimes. Oucomes when i IB is he policy insrumen and he exchange rae is flexible are shown in column 2. The case of a fixed exchange rae policy is shown in column 3. ς MP expands he feasible allocaions achievable by he cenral bank because i affecs consumpion and invesmen differenially. Having an exra policy insrumen should improve he rade-offs he cenral bank faces. To gain some inuiion for how using ς MP can improve policy rade-offs, i is useful o consider he expression for real marginal cos common o basic New Keynesian models. Wih flexible wages, real marginal cos is proporional o he marginal rae of subsiuion beween leisure and consumpion and he marginal produc of labor. When log-linearized around he seady sae, his yields an expression of he form ln mc = σ ln c + η ln y (1 + η)z + μ, where mc is real marginal cos, c is real consumpion, y is oupu, z is a produciviy shock, and μ is a markup (cos) shock. The coefficien of relaive risk aversion is denoed by σ, and η is he inverse Frisch elasiciy of labor supply wih respec o he real wage. In he face of a posiive markup shock μ, he cenral bank can moderae he rise in inflaion by raising i IB, hereby lowering c and y. A reducion in ς MP reduces consumpion relaive o oupu.
124 Inernaional Journal of Cenral Banking March 2012 So if ς MP is cu as i IB is increased, a given decline in σ ln c + η ln y can be achieved wih a smaller fall in oupu. Tha is, using he reserve raio allows he cenral bank o gain he same movemen of real marginal cos wih a smaller decline in y. Thus, given ha G-T evaluae oucomes using a quadraic loss funcion in inflaion and oupu volailiy, he rade-off beween inflaion sabilizaion and oupu sabilizaion is improved when ς MP is used. Bu, a leas in he calibraed version of he G-T model, his effec is small, so he gains from using reserve requiremens acively are generally small, and he opimal coefficiens in he simple Taylor-ype rule for i IB are lile changed if reserve requiremens become an acive ool of moneary policy (see able 3 of G-T). While G-T consider a sandard quadraic loss funcion involving oupu and inflaion, hey also consider one expanded o include a loan sabilizaion objecive so ha loss is given by L = π 2 + λ Y Ŷ 2 + λ L ˆL2, (6) where inflaion, oupu, and he quaniy of loans are expressed as deviaions from heir seady-sae values. The fac ha using ς MP helps very lile in achieving inflaion and oupu goals implies ha i can insead focus on sabilizing loans. Hence, G-T obain he resul ha i IB should respond o inflaion and oupu while ς MP should respond solely o loans. 3. Cenral Bank and Tax Policies Glocker and Towbin s analysis is a useful reminder ha he ools of moneary policy can have ax-like effecs by alering he allocaion of resources among compeing uses. Of course, axes give rise o axavoidance behavior, and he effecs of changing a ax will be mued if agens are able o shif ou of he axed aciviies. This is a concern of paricular relevance in hinking abou reserve requiremens. By axing deposis a one class of financial insiuions, funds will flow ou of he banking secor ino oher financial insiuions no subjec o reserve requiremens. In Glocker and Towbin s model, his ype of acion is ruled ou by assumpion bank deposis are he only source of funds ha can be len o enrepreneurs. While his is fine in a heoreical exercise, i does sugges ha he quaniaive
Vol. 8 No. 1 Discussion: Walsh 125 magniude of G-T s findings may provide an upper bound for he acual effecs of changing he required reserve raio. If non-reservable asses were also available o households, he effecs of he required reserve raio on consumpion would be smaller han G-T find. However, he impac of a rise in he reserve raio in reducing bank loan supply would be larger as funds move ino asses no subjec o he reserve ax. Wheher his also increases he impac on invesmen would depend on he exen o which bank lending is special. If borrowers can access oher sources of funds, a rise in he ax on banking would have lile ne effec on eiher consumpion or invesmen. However, he issue of wheher bank lending is special is he subjec of an old and inconclusive debae. The analogy wih axes is helpful in hinking abou moneary policy in general (Walsh 1984). We are rained o hink of seadysae inflaion as a ax, bu in New Keynesian models, deviaions from price sabiliy affec markups in ways ha are similar o wha would occur wih a cyclical ax (or subsidy) on he inpus used o produce final goods (Ravenna and Walsh, forhcoming). Thus, deviaions from price sabiliy have ax-like effecs. Taxes creae disorions; heir usefulness as ools o improve allocaions arises in he conex of he second bes. There mus be some oher disorion in he economy wihou axes ha can be reduced if a ax is inroduced. The financial fricions and nominal rigidiies in he G-T model sugges he allocaion wihou axes is no opimal. The quesion hen is: How can reserve requiremens offse exising disorions? Unforunaely, his is where he limiaions of he ad hoc objecive funcions G-T use become clear, as a loss funcion such as (6) is no adequae for assessing welfare. For example, no all flucuaions in oupu should be sabilized in he presence of echnology shocks. And presumably no all flucuaions in he quaniy of loans should be sabilized, as some will reflec he efficien movemen of loans in he face of produciviy shocks. In general, he quaniy of loans should rise in he face of echnology shocks, so a policy ha acs o sabilize he level of loans would be misguided. And in he face of financial fricions, i could even be he case ha loans are oo sable and opimal policy should increase he volailiy of loans. For example, Faia and Monacelli (2007) find ha financial fricions preven asse prices from moving sufficienly in response o produciviy shocks, and opimal policy calls for ineres raes o be cu as
126 Inernaional Journal of Cenral Banking March 2012 asse prices rise. Thus, while he ad hoc loss funcions employed by G-T provide a useful saring poin for analyzing he role of reserve requiremens, i would be ineresing o see his policy insrumen inegraed ino a welfare-based analysis of opimal policy. To conclude, Glocker and Towbin have provided an imporan exension o he exising lieraure by inegraing reserve requiremens ino a DSGE framework. In building on heir work, i will be imporan o expand he financial secor of he model o allow for non-bank sources of credi. This exension may reduce he effecs reserve requiremens can have on he economy. Anoher exension would be o link he analysis of Glocker and Towbin o he discussion of procyclical bank capial requiremens, as hese would seem o have similar effecs o a procyclical reserve requiremen. And finally, i will be ineresing o see wha role reserve requiremens migh play in a welfare-based evaluaion of policy. References Bernanke, B. S., M. Gerler, and S. Gilchris. 1999. The Financial Acceleraor in a Quaniaive Business Cycle Framework. In Handbook of Macroeconomics, Vol. 1C, ed. J. B. Taylor and M. Woodford, 1341 93. Amserdam: Elsevier Norh-Holland. Faia, E., and T. Monacelli. 2007. Opimal Ineres Rae Rules, Asse Prices, and Credi Fricions. Journal of Economic Dynamics and Conrol 31 (10): 3228 54. Ravenna, F., and C. E. Walsh. Forhcoming. Moneary Policy and Labor Fricions: A Tax Inerpreaion. Journal of Moneary Economics. Romer, D. 1985. Financial Inermediaion, Reserve Requiremens, and Inside Money: A General Equilibrium Analysis. Journal of Moneary Economics 16 (2): 175 94. Walsh, C. E. 1984. Opimal Taxaion by he Moneary Auhoriy. NBER Working Paper No. 1375 (June).