Do highly liquid banks insulate their lending behavior?

Transcription

1 UCD GEARY INSTITUTE FOR PUBLIC POLICY DISCUSSION PAPER SERIES Do highly liquid baks isulate their ledig behavior? Supriya Kapoor UCD Geary Istitute for Public Policy Geary WP2017/09 September 11, 2017 UCD Geary Istitute Discussio Papers ofte represet prelimiary work ad are circulated to ecourage discussio. Citatio of such a paper should accout for its provisioal character. A revised versio may be available directly from the author. Ay opiios expressed here are those of the author(s) ad ot those of UCD Geary Istitute. Research published i this series may iclude views o policy, but the istitute itself takes o istitutioal policy positios.

2 Do highly liquid baks isulate their ledig behavior? Supriya Kapoor * Uiversity College Dubli Abstract The role of baks i the trasmissio of moetary policy has bee of sigificace lately. We aim to aalyse the bak ledig behaviour durig chages i moetary policy. We test for loa supply shifts by segregatig baks based o their liquidity alog with size ad capital ratio. This paper employs uisured, o-reservable liabilities such as time deposits ad ivestigates whether baks are able to isulate themselves durig a moetary policy chage. We fid that the loa supply shock ca be eutralized post moetary policy chages. Furthermore, the less liquid ad small baks are uable to carry out such operatios ad are more affected by moetary shocks. This has importat implicatio i the workig of commercial baks ad effects of moetary policy. Keywords: bak ledig chael, time deposits, moetary policy, liquidity JEL Classificatio: E52, G21, E50 * PhD cadidate at the School of Ecoomics, Uiversity College Dubli, Irelad. supriyakapoor18@gmail.com

3 1. Itroductio The credit chael of moetary policy trasmissio has bee widely discussed at the aggregate level; 1 a tight moetary policy by the cetral bak reduces volume of the reserves leadig to a fall i the supply of loas. However, at the bak level, studies such as Kashyap ad Stei, (1995); Housto et al., (1997); Campello, (2002); Westerlud (2003); Cetorelli ad Goldberg (2008); suggests that the fiacial positios of idividual baks have sigificat effect o the way they react to policy chages. The studies such as Ashcraft (2003) ad Kisha ad Opiela (2009) 2 use time deposits to idetify isulatio mechaisms by the baks based o their asset size ad capitalizatio. This study builds o their work ad examies the relatioship betwee cetral bak s policy actios ad idividual bak s ledig behavior with respect to its liquidity, size ad capital. We ivestigate whether liquid baks reallocate fuds through uisured/o-reservable liabilities such as time deposits, isulatig their loa supply behavior. To the best of our kowledge, oe of the studies have focused o the role of bak liquidity i raisig time deposits to deal with moetary policy shocks. The existig literature o the isulatio mechaisms of baks icludes studies such as Kashyap ad Stei (1995), Gambacorta (2005), ad Cetorelli ad Goldberg (2008), to ame a few. Amogst them, Kashyap ad Stei (1995) provides empirical evidece o the theoretical model of Bak Ledig Chael (BLC). The study uses quarterly bak-level data from the Uited States of America to show that baks give out fewer loas if they ow less total assets. Gambacorta (2005) use the Italia micro bak level data from ad tests the crosssectioal differeces i the effectiveess of bak ledig chael. The fidigs suggest that the effects of moetary policy differ across baks. Further, Cetorelli ad Goldberg (2008) documet a systematic relatioship betwee moetary policy shocks ad globalizatio of baks through iteral capital markets. The study suggests that large ad global baks ca make use of their iteratioal, iteral capital markets to shield their ledig behavior as a respose to moetary policy chages. 1 See Berake ad Blider, (1988); Berake ad Blider, (1992); Berake ad Gertler, (1993); Hoshi et al.(1993);ramey (1993) ad Berake ad Gertler (1995). 2 The study by Kisha ad Opiela (2009) fids that small ad uder-capitalized baks are most resposive to moetary policy shocks ad are uable to raise time deposits to shield their ledig behavior.

4 The studies as metioed above, show that the bak-specific characteristics such as size, liquidity ad capital are importat factors towards the ledig behavior of baks. A importat policy implicatio is the likage betwee moetary policy ad the busiess cycle. This is critical because the fiacial positios of baks vary whe there are chages i ecoomic performace or stages of the busiess cycle. Sice each busiess cycle has its uique feature, baks fiacial positio will also vary across cycles. Hece, this lie of research may help policy makers judge the importace of the credit chael. Lastly, asymmetric bak resposes to moetary policy shocks result i differet effects for differet baks either based o their geographical locatio or their specific characteristics. The baks may also be extedig loas to differet types of busiesses ad hece, moetary policy implemetatio will result i asymmetric effects across regios ad idustries. This paper addresses the questio of whether liquid baks raise time deposits to shield their loa supply agaist moetary policy shocks. We use balace sheet pael-data o idividual baks i Idia from Liquidity serves as a importat bak-specific characteristic because liquid baks may attract customers cofidece durig ucertai times ad some depositors might move their deposits towards baks that are liquid i their balace sheets. Additioally, a liquid bak might be more flexible i offerig higher iterest rates o time deposits tha the oe offered o savig accouts. Hece, our study/aalysis o bak liquidity amidst chages i moetary policy is a importat cotributio to the existig literature. The fidigs suggest that liquid baks are less resposive to chages i moetary policy due to their ability to isulate their ledig behavior through time deposits. I additio, size is also foud to be a sigificat characteristic for ledig behavior of baks, much i lie with existig studies. The results suggest that the loa growth of small baks is sigificatly affected by policy chages. I the ext sectios, we discuss the previous literature ad outlie our data ad approach, followed by presetatio of results ad coclusio. 2. Literature Review There is ample evidece that empirically supports the bak ledig chael ad how chages i moetary policy alter the supply of loas. However, recet literature has started to focus o the isulatio mechaisms adopted by baks durig moetary shocks i order to maitai

5 their supply of loas. Previous evidece o moetary policy trasmissio is associated with the work of Berake ad Blider (1992); Kashyap, Stei ad Wilcox (1993). Berake ad Blider (1992) highlighted the existece of the BLC by estimatig reduced form equatios of credit supply. They employed iterest rate o Federal fuds as a idicator of future movemets i real macroecoomic activities ad fid a iverse relatioship betwee bak loas ad a cotractioary moetary policy. A primary drawback of usig aggregate data lies i the iability to distiguish betwee credit demad ad credit supply resposes, give that moetary shocks affect both (Romer ad Romer, 1990; Westerlud, 2003). Usig disaggregated data over aggregate measure is that the former is more iformative. A moetary cotractio results i credit reductio either due to a fall i loa supply or a reductio i borrowers ledig. With the use of aggregate data, it is challegig to idetify the BLC separately. Attempts i the literature have bee made to carry out aalysis by sub-structurig data at the bak-level. Few such studies are discussed below i the remaiig sectio. Studies by Kisha Opiela (2000); Altuvas et al. (2002); Ageloi et al (2003); Ehrma et al. (2003) i Frace, Germay, Italy ad Spai; Huag (2003) ad Gambacorta (2005) i the UK ad for the Euro area; Ashcraft (2006); Cetorelli ad Goldbeg (2012) i the US have ucovered the role of baks i moetary trasmissio. Kashyap ad Stei (1995) empirically supports the theoretical model of BLC, where they exploit quarterly U.S. bak-level data ad exhibit the respose of large baks o their ledig activity durig a cotractioary moetary policy. They foud that lesser the total assets a bak ows, fewer loas will it give out. Other researchers employ differet bak characteristics to differet coutries ad datasets. As a alterative, Altubas et al. (2002) applied the cocept of BLC o the Europea bak balace sheets from ad by categorizig the data o the basis of asset ad capital ratio, they witessed a weak respose of moetary policy o uder-capitalized baks. Gambacorta (2005) experimeted with the Italia micro bak level data from ad foud that stroger baks, i terms of liquidity are relatively less resposive to the moetary policy chages tha the illiquid baks. O the similar lies, Kashyap ad Stei (2005) agai study U.S. data from ad are able to demostrate the existece of BLC ad foud out results cosistet with Gambacorta (2005). Hosoo (2006) uses Japaese bak data for the time period ragig from ad cocludes that moetary policy has weaker effect o udercapitalized

6 baks. All these studies attempted to show the existece of bak ledig chael ad large, liquid ad well-capitalized baks are uresposive ad uaffected to that ed. Several recet papers focus o the roles played by baks to isulate their ledig behavior durig chages i the moetary policy. Workig with the aggregate data, Kashyap, Stei ad Wilcox (1993) show that usig a mix of bak loas ad commercial paper for fiacig compaies durig moetary policy cotractio ca serve as a alterate compositio for firms exteral fiace. They did fid a icrease i the issuace of commercial paper servig as a substitute to bak loas durig such shock, cocludig that a reductio i bak loas is ot a reflectio of a fall i credit demad due to a slow ecoomy but it s ideed due to the tight moetary policy. Olier ad Rudebusch (1996b) revisited the work by above authors ad brought about two developmets, first, usig a differet mix of exteral fiace ad secod, divided the data accordig to bak size. Their results were t cosistet with the BLC i particular, however the evidece was broadly relatig to the credit chael overall. Talkig about isulatio mechaisms, whe baks experiece a shock, they ca either raise o-reservable liabilities to make up for the lost reservable deposits, or sell off their assets ad securities i order to shrik their assets. This was empirically aalyzed by Kashyap ad Stei (2000). A large bak which ows more liquid assets will have a optio of sellig securities ad thus isulatig its ledig behavior, whereas a less liquid bak will be forced to deplete its loa portfolio. Ashcraft (2003) illustrates that stad-aloe baks face severe fiacial costraits tha affiliated baks whe measured by sesitivity of loa growth to isured deposit growth. Usig data o isured commercial baks from Call Reports of Icome ad Coditio, he first revealed that bak holdig compay reduces the fiacial costraits faced by its subsidiaries. Next, he explores the possibility where affiliated baks are able to shield their ledig behavior i times of a moetary shock by issuig large certificate of deposits (CDs) ad federal fuds. Other authors studies the bakig behavior o the same lies 3. Cetorelli ad Goldberg (2008) study the respose of bak ledig of global baks 4. The paper documets a systematic relatioship betwee moetary policy effectiveess ad globalizatio 3 Baks are also able to have access to o-reservable/ uisured liabilities if baks are affiliated with a multibak holdig compay as studied by Housto et al. (1997). The authors show that loa growth at subsidiary baks is more sesitive to bak holdig compay s cash flow tha to the baks ow cash flow. Campello (2002) studies the bakig behaviour o the same lies ad fids out that small baks that are affiliated with large multibak hold compaies, have a less sesitive ledig behaviour to moetary policy tighteig as compared to small stad-aloe baks. 4 Cetorelli ad Goldberg (2012) defie global baks as baks with global operatios, developig growig etworks of physical braches ad subsidiaries i foreig coutries.

7 of baks through iteral capital markets. Large, global baks ca make use of their iteratioal, iteral capital markets i respose to moetary policy. A chage i domestic moetary policy causes chages i baks loa portfolio ad is cosistet with iteratioal moetary policy trasmissio. The results suggest that a domestic policy shock impacts the domestic bak ledig ad the ecoomy as a whole, ad at the same time the shock is trasmitted to the foreig markets. Peek ad Rosegre (1997, 2000) also specifies chages i bakig behavior ad thus iteratioal trasmissio shocks, where the authors focus o the capital problems stemmig from the declie i Japaese stock prices. Kisha ad Opiela (2009) focus o isulatio mechaisms by baks durig chages i moetary policy. Usig data from , the authors use cross-sectioal differeces i bak fiacig ad ledig decisios ad divide the dataset ito six asset ad three capital leverage ratio categories. They use bak s capital-to-asset ratio as a proxy for bak s ability to raise time deposits. The paper cotributes positively to the existig literature ad fids that whe baks raise large time deposits i order to shield their ledig behavior, it appears that small ad udercapitalized are the oes that are most resposive to moetary policy. We build o the above work ad provide quatitative evidece by addig i the model, the measure of liquidity for all baks, costructed as the ratio of liquid assets to total assets. Baks are subject to differetial balace sheet characteristics. We discuss whether high liquid baks are likely to be more isulated durig chages i moetary policy ad is raisig time deposits a feasible mechaism for this. However, to reach the above coclusio it is a prerequisite that the baks must respod i their ledig with a policy shock. Hece, we begi by testig the existece of BLC. The literature above aims to provide a foudatio ad support for a ew isight that the curret research proposes to cotribute i the subsequet sectios. 3. Data Descriptio The sample comprises aually balace sheet iformatio for idividual baks take from yearly publicatio of Reserve Bak of Idia, Statistical Table relatig to Baks i Idia. A few baks operated cotiuously from 1995 ad other few operated cotiuously at a later stage i our data. Hece, we have a ubalaced pael that cosists of 138 commercial baks with 1805 bak-year observatios. We exclude the data o regioal rural baks because these baks exist for the developmet of rural ad backward areas focusig o fiacial iclusio rather tha moetary policy trasmissio. Sice these primarily focus o the basic bakig services i the rural areas, the ledig resposes of these baks might be differet from commercial

8 baks; hece it is difficult to get cosistet results with the iclusio of regioal baks ad cooperatives. Figure 1 Moetary Policy idicator ad loa To measure chages i moetary policy that iflueces baks the cost of fuds, we use repurchase or repo rate 5 suggested by a variety of researchers (Padit et al., 2006; ad Sumitra ad Toto, 2012). I regard to policy regime, the cetral bak alters this bechmark iterest rate to cotrol iflatio ad credit i the ecoomy (Aleem, 2010; Mohaty, 2012). The repo rate has bee a effective policy rate sice 2000, while the cetral bak used bak rate as the moetary policy variable for the earlier period. 6 Accordigly, we use bak rate (till 2000) ad repo rate (post 2000) as primary policy rates as the measure for moetary policy. 7 We use Reserve Bak of Idia (RBI) data to calculate our three bak-specific characteristics. 8 Size is calculated as the log of total assets. We measure liquidity as the ratio of liquid assets to total assets. 9 The tier 1 capital ratio 10 is used as a proxy for capitalizatio. The paper employs 5 The repo rate is the rate at which the cetral bak leds moey to commercial baks for short term. 6 Bak rate ca be defied as the iterest rate charged by cetral bak o loas to commercial baks. Whe cetral bak fids iflatioary pressures have started i the ecoomy, it raises the bak rate. Borrowig from cetral bak becomes costly ad commercial baks borrow less. This leads to baks raisig their ledig rates. Borrowers borrow less from baks ad result i cotractio of credit. I Idia, bak rate has bee used as the prime policy rate till Thereafter, it was fixed at 6% (Sumitra ad Toto, 2012). 7 See Sumitra ad Toto (2012) about a discussio o advatages of repo rate ad cetral bak policy rate prior to The literature by Black et al. (2010), Topi ad Vilmue (2003), Havrylchyk ad Jurzyk (2005) amog others, suggests these characteristics are importat 9 Liquid assets iclude cash, balaces with RBI, balaces with baks i Idia, balaces with baks outside Idia ad govermet ad other securities. 10 Basel III defies Tier 1 ratio as the ratio of bak s core equity capital to its total risk weighted assets. It is a measure of bak s fiacial stregth.

9 log of total advaces as a measure of baks ledig. Figure 1 shows the policy rate regime i Idia for our data. We observe a proouced iverse relatioship betwee total advaces ad the policy rate. However, this iverse relatio seems to have weakeed post A series of data-cleaig procedures are adopted such as reportig of repeated time values withi the pael ad dealig with mergers ad birth of ew baks. The bakig sector has witessed several mergers ad amalgamatios i the domestic commercial baks icludig private sector baks such as ICICI, HDFC ad UTI (ow Axis Bak). We treat such baks as ew idividual baks i our data. We deflate the values of all variables by usig the cosumer price idex (2010 as the base year). GDP growth rate, iflatio rate ad exchage rate (betwee USD ad Idia rupee) are used as the mai macroecoomic variables to cotrol for overall ecoomic health ad loa demad effects. The iclusio of these variables allows us to capture the cyclical movemets ad serves to separate out chages i total loas that are caused by movemets i loa demad. Table 1 below gives descriptive profile of the mai variables used i the aalysis. The values are expressed i Idia Rupee (INR). The descriptio is for all the baks used i the data; therefore we see high stadard deviatio ad wide differece betwee the miimum ad maximum values of each variable. We iclude all the commercial baks i Idia ad the data rages from very small to very big baks (State Bak of Idia, Axis Bak, Bak of Idia, Dea Bak). The table also reports features of some macroecoomic variables used. The aual average quatity of loas made by baks was INR 2,430 millio. Durig the sample period, the average rate of GDP growth was recorded as 6.8%. The mea value of exchage rate volatility was early 45, ad i stable times it was as low as

10 Table 1: Descriptive Statistics of key variables (INR millio uless specified) Variable Observatio Mea Stadard Miimum Maximum Deviatio Bakig ad Fiacial variables Advaces Time deposits Total Assets Liquidity Capital Adequacy ratio (%) Moetary policy Idicator Repo rate (%) Macroecoomic Factors GDP growth rate (%) Iflatio rate (%) Exchage rate (INR/$US) I order to empirically iterpret our research questio, the ext sectio discusses empirical specificatio to test the behavior of time deposit growth o bak ledig for differet baks based o their characteristics. However, a pre-requisite to the above is to test the existece of bak ledig chael for our data. 12 Ivestigatig the bak-ledig chael is also useful for moetary authority i desigig appropriate policies to achieve price stability or iflatio targetig. This serves as a potetial policy implicatio. 4. The ecoometric model We follow the approach of Kisha ad Opiela (2000) ad divide the baks o the basis of liquidity ad size ito two categories. Further, we also use capitalizatio to divide baks ito three categories. These categories are as follows; a bak that has more tha total assets INR500 millio (INR= Idia Rupees) is categorized a large bak, whereas a bak <INR 500millio is a small bak. The categories based o capitalizatio are; tier 1 capital ratio: < 10% (udercapitalized), 10% ad <15% (adequately capitalized) ad >15% (well-capitalized). We use time deposits as a measure of uisured/o- reservable liabilities to ivestigate 11 Sum of cash i had, balaces with RBI, with baks i Idia, with baks outside Idia ad moey call ad short otice 12 Advace it = α i + j=1 α j Advace it j + j=0 β j RRate i + j=0 δ j X it RRate i + j=0 γ j GDPgr t + j=0 φ j l(exch) t + j=0 ω j if t + ε it I this equatio, subscript i (cross sectio) deote a bak idicator ad t (time series) is a year idicator. Advace it j is the growth i loa supply captured by the log of total advaces. Amog the explaatory variables, we use lagged depedet variable (Advace it j ), moetary policy idicator (RRate i ) that is captured by policy rate. X it deote bak characteristics such as bak size, liquidity ad capitalizatio. To accout for the loa demad effects, the model icorporates the use of macroecoomic variables such as GDP growth (GDPgr t ), iflatio rate (if t ) ad log of exchage rate (l(exch) t ). I order to exhibit differetial respose to macroecoomic ad moetary developmets, the policy idicator iteracts with all the three bak characteristics. Thus the iteractio term ca be iterpreted as the effect of moetary policy idicator o bak specific characteristics of each bak. To test the direct effect of moetary policy, the primary iterest is o the coefficiet of moetary policy idicator, whereas to study the existece of bak ledig chael, the iteractio betwee moetary policy ad bak characteristics is aalyzed.

11 whether liquid baks isulate themselves durig domestic moetary chages. This behavior of the baks is examied by regressig large time deposits o the moetary policy idicator ad lagged values of GDP growth rate. TDeposits it = α 1 + β 1 TDeposits it j + β 2 RRate i + β 3 GDPgr t j + β 4 dliquid i RRate i j=1 j=0 j=0 j=0 + ε it (1) TDeposits it idicates time deposits for each bak, RRate i is the moetary policy idicator, i.e., repo rate, GDPgr t j deotes the lag of GDP growth rate ad dliquid i represets liquidity. It is a categorical variable, where 0 deotes low liquid baks ad 1 otherwise. It is to be oted that we are iterested i kowig the importace of liquidity durig baks isulatio mechaism. The more liquid a bak is, the less pressure it should have o its balace sheet durig ay moetary shock. 13 We iclude lags of the depedet variable ad estimate the model i lagged differeces i order to capture the pro-cyclical time deposit behavior of commercial baks i the dataset. Due to the iclusio of lagged depedet variable, the explaatory variables are correlated with the error terms. I such a case, the dyamic pael model suffers from Nickel (1981) bias ad eeds to be corrected. Arellao ad Bover (1995) ad Bludell ad Bod (1998) have itroduced ad exteded the geeralized methods of momets (GMM) estimators, which deals with the edogeeity problem. The bak liquidity used i the aalysis is importat because it deals with day-to-day aalysis of future cash iflows ad outflows ad hece affects the bak s balace sheet ad profitability. Also, we use time deposits, sice they are importat compoet of liabilities for fiacial itermediaries (See Ashcraft, 2003). Ituitively, moetary policy will have a small effect o ledig behavior of baks if they are easily able to replace the loss i ledig with exteral fiace, such as term deposits, which acts as a margial source of fuds durig cotractioary moetary policy. 13 A low liquid bak is defied as ay bak whose ratio of liquid assets to total assets lies below the media value. From the data, we fid that media for liquidity is 37%. We perform aalysis for differet quitiles of liquidity ad get similar results.

12 5. Results I this sectio we preset the empirical fidigs. We begi by presetig the results that idicate the existece of bak ledig chael, as show i Table 2, followed by aswerig whether high liquid baks isulate their loa growth by raisig time deposits The estimates i Table 2 demostrate the effect of moetary policy o loa supply alog with its differetial impact based o characteristics of the baks (liquidity, size, ad capital). The fidigs of the full sample regressio results are preseted i Colum IV ad suggest cosistecy with the existig literature; sice cotractioary moetary policy leads to declie i bak loa supply, as idicated by egative coefficiet o Repo rate. The egative coefficiet implies that baks reduce their supply of loas whe repo rate icreases such that a uit icrease i the repo rate leads to a expected reductio i loa supply by almost 55 percet. The positive iteractio betwee bak liquidity ad repo rate also implies the effect of moetary policy o idividual bak s ledig depedig o their balace sheet stregth as depicted by each bak s liquidity. Sice the iteractio terms tur out to be positive for all the three bak characteristics, it implies large, liquid ad well-capitalized baks are able to isulate their ledig behavior from moetary shocks. To test the effects of bak ledig o resposive to moetary policy, we ru regressio o differet sub-samples usig bak characteristics, i.e., size, liquidity ad capital ratio. The results are show i Colum I, II ad III i Table 2. For liquid ad high-capitalized baks (Table 2, Colum II ad III), all the variables are statistically sigificat ad those capturig key moetary policy effects are appropriately siged. The parameter estimate of iterest: the iteractio term betwee moetary policy ad bak characteristics is sigificat ad positive, cosistet with the theoretical expectatios. However, this is ot true for bak size (Colum I). The size variable o its ow as well as iteractio term) is carryig a wrog sig, but is statistically sigificat at covetioal levels. The fidigs from this estimatio suggest that moetary policy does ot sigificatly ifluece bak size o its ow. The test results suggest that the efficacy of moetary policy trasmissio i the bakig system is stroger for large, liquid ad well-capitalized baks. We exted this aalysis ad further discuss the empirical results for baks ability to raise time deposits ad shield their ledig behavior durig moetary shocks.

13 Table 2: Loa supply dyamics i the evet of Moetary Policy chages Variables ( I ) ( II ) ( III ) ( IV ) First lag of depedet variable.03 (.07) -.31*** (.115) -.28** (.148) -.27*** (.024) Repo rate.45*** (.152) -.59*** (.111) -.48*** (.077) -.55*** (.069) Reporate*size -.025*** (.007).009*** (.002) Reporate*liquid.51* (.217).43*** (.069) Reporate*capratio.35*** (.098).31*** (.022) Other cotrols YES YES YES YES Time Dummies YES YES YES YES Observatios Number of groups AR(2) p-value Hase test p-value Robust stadard errors i paretheses ***p<0.01, **p<0.05, *p<0.1 Table 3 presets empirical results for isulatio mechaisms of liquid baks through time deposits. We hypothesize that BLC is ieffective for some baks as they are able to maitai their loa supply eve at the time of moetary cotractios. This is because durig cotractioary policy by the cetral bak, the commercial baks must reduce their reservable deposits. However, if they are able to supplemet these deposits by raisig time deposits (TD), their loa supply will remai uchaged. As see from the previous results, a small, less liquid ad uder-capitalized bak is more resposive to moetary policy ad would be uable to raise fuds through alterative sources. Thus we would expect time deposits to be oresposive for these baks. Most of the variables icluded i the model are sigificat. The evidece of raisig time deposits by some baks should be reflected by a positive ad statistically sigificat coefficiet. The sum of coefficiets o policy idicator ad its iteractio with liquidity, i.e., β 2 + β 4 explais the effect of moetary policy o large time deposits for high liquid baks, whereas the coefficiet o moetary policy idicator shows the effect for low liquid baks. Most coefficiets for low liquid baks are statistically isigificat. This meas that low liquid baks are resposive to moetary shocks ad do ot or are uable to raise time deposits. However, large ad high liquid baks appear to raise time deposits. This is idicated by the positive ad statistically sigificat coefficiets (highlighted i bold i Table 3). This could be due to these baks easier access to other fudig opportuities. Most coefficiets o (β 2 + β 4 ) are statistically sigificat. However, the coefficiets with a egative ad statistically sigificat sig idicates that a bak ca access fuds through either sellig securities or iteral capital

14 market operatios or other opportuities, if ot by raisig time deposits. Liquidity is a importat bak characteristic ad its sigificace has bee further highlighted post the global fiacial crisis. Liquidity provides baks with a cushio to protect agaist capital losses or ay other uexpected expeses. Hece, a liquid bak is i a better positio to face moetary shocks. I additio, bak behavior differs accordig to bak size. As see from the results, large baks are less adversely affected by cotractioary policy due to their ability to raise time deposits ad maitai loa supply. Table 2: The effect of moetary policy o growth rate of time deposits Size Capitalizatio Number of observatios Co-efficiet o moetary policy idicator (β 2 ) Co-efficiet o iteractio of moetary policy variable ad liquidity (β 4 ) (β 2 + β 4 ) Small Udercapitalized Small Medium-capitalized Small Well-capitalized * Large Udercapitalized 99.19**.17.02*** Large Medium-capitalized ***.033*** Large Well-capitalized ***.51*** Robust stadard errors i paretheses ***p<0.01, **p<0.05, *p< Coclusio There is ample evidece that empirically supports the bak ledig chael ad how chages i moetary policy alter the supply of loas. However, recet literature has started to focus o the isulatio mechaisms adopted by baks durig moetary shocks i order to maitai their supply of loas. Kisha ad Opiela (2000) show that the small, udercapitalized baks are ot able to fiace their ledig behavior through large time deposits. We add o to their aalysis by icludig liquidity ad fid that the use of liquidity as a balace sheet stregth idicator of moetary policy describes baks loa respose to policy through raisig time deposits. The results are statistically sigificat ad ecoomically importat. The mechaism that we study implies that, liquidity is sigificat i bak portfolio ad should be cosidered wisely whe formulatig or amedig moetary policy. First, we fid that the ledig behaviors of large, liquid ad well-capitalized baks are less sesitive to chages i moetary policy. Secod, usig data o time deposits, we show that liquidity is a importat characteristic ad large ad liquid baks are able to raise alterative fuds such as time deposits to cotiue ledig durig tight moetary policy. Moetary policy will have a small effect o ledig behavior of baks if they are easily able to replace the loss i ledig with exteral fiace.

15 Refereces Aleem, Abdul [2010]. Trasmissio mechaism of moetary policy i Idia, Joural of Asia Ecoomics, 21, Ashcraft, A. (2006). New Evidece o the ledig chael. Joural of Moey, Credit ad Bakig 38, Berake, B. S., & Blider, A. S. (1988). Credit, moey ad aggregate demad. America Ecoomic Review (Papers ad Proceedigs), Berake, B.S., & Blider, A. S. (1992). The Federal Fuds Rate ad the Chages of Moetary Trasmissio. America Ecoomic Review 82, Berake, B. S., & Gertler, M. (1995). Iside the Black Box: The Credit Chael of Moetary Policy Trasmissio. Joural of Ecoomic Perspectives, Black, L.K., Hacock, D., Passmore, W. (2010). The bak ledig chael of moetary policy ad its effect o mortgage ledig. FEDS workig paper Federal Reserve Board of Goverors. Bludell R., & Bod S. (1998). Iitial Coditios ad momet restrictios i dyamic pael data models. Joural of Ecoometrics 87, Campello M. (2002). Iteral Capital Markets i Fiacial coglomerates: Evidece from small baks respose to moetary policy. Joural of Fiace, 57(6), Cetorelli, N., Goldberg, L. (2012). Bakig Globalizatio ad Moetary Trasmissio. Joural of Fiace Ehrma, M., Gambacorta L., Martiez Pagés, J., Sevestre, P., & Worms A. (2003b). The Effects of Moetary Policy i the Euro Area. Oxford Review of Ecoomic Policy, 19(1), Gambacorta, L. (2005). Iside the Bak Ledig Chael. Europea Ecoomic Review, Havrylchyk, Olea, Jurzyk, Emilia (2005). Does the bak ledig chael work i a trasitio ecoomy? A case of Polad. Mimeo, Europea Uiversity Viadria Housto, J., Marcus, D., & James C. (1997). Capital Market frictios ad the role of Iteral Capital Markets i bakig?. Joural of Fiacial Ecoomics 46, Kashyap, A., & Stei, J. (1995). The Impact of Moetary Policy o Bak Balace Sheets. Caregie- Rochester Coferece Series o Public Policy 42, Kashyap, A., & Stei, J. (2000). What Do a Millio Observatios o Baks Say About the Trasmissio of Moetary Policy. America Ecoomic Review, Kisha, Ruby, & Opiela, T. (2000). Bak Size, Bak Capital, ad the Bak Ledig Chael. Joural of Moey, Credit ad Bakig,

16 Mohaty, D. (2012). Evidece o iterest rate chael of moetary policy trasmissio i Idia.RBI Workig Paper 6. RBI: Mumbai. Padit, B.L., Mittal, A., Roy, M., & Ghosh, S. (2006). Trasmissio of Moetary Policy ad Bak Ledig Chael: Aalysis ad Evidece from Idia. Departmet of Ecoomic Aalysis ad Policy, Reserve Bak of Idia. Study No. 25 Prasad, A., & Ghosh S. (2005). Moetary policy ad corporate behavior i Idia, IMF Workig Paper No. 25, Washigto D.C. Reserve Bak of Idia (2003), Report o currecy ad fiace, RBI: Mumbai. Reserve Bak of Idia, Hadbook of statistics o Idia ecoomy , RBI: Mumbai. Reserve Bak of Idia, Statistical tables relatig to baks i Idia (various years), RBI: Mumbai. Topi, J. ad J. Vilmue (2003). Trasmissio of moetary policy shocks i Filad: evidece from bak level data o loas, Moetary Policy Trasmissio i the Euro Area, (Cambridge Uiversity Press), forthcomig.