Inernal and Exernal Facors for Credi Growh in Macao Nicholas Cheang Research and Saisics Deparmen, Moneary Auhoriy of Macao Absrac Commercial banks are dominan eniies in he Macao financial secor. They rely on radiional lending business as loans accoun for a sizable proporion of heir asses. In recen years, he incremen in credi exended o he privae secor has even oupaced Macao s gross domesic produc growh. This paper aemps o apply a scienific model o weigh various poenial inernal and exernal facors for he credi growh. Based on our esimaion, he growh of loans exended o residens is largely induced by inernal facors such as real economic growh, domesic ineres rae and banks healh, while exernal moneary condiions play a more significan role in credi exended o non-residens. In paricular, our findings show ha he healh saus of he local banking secor, represened by he lagged non-performing loan raio, is an imporan deerminan of he incremen of bank lending, especially for he residen secor. For enhancemen o he overall financial sabiliy, our esimaion resuls advocae a forward-looking aiude in credi operaion by local banks and a srenghening in linkage beween loans and deposis. - 87 -
1. Inroducion Commercial banks are dominan eniies in he Macao financial secor. Their aggregae asse accouns for 98% of he oal asses of all local financial insiuions. In line wih he expanding economy and public needs, local banks have sepped up heir effor in providing a wide variey of financial producs and services, while radiional businesses including deposi-aking and loan exension mainain heir core posiion. Nowadays, he loan business sill represens a sizable proporion of bank asses. In paricular, oal loans in nominal value winessed consecuive and noable incremens, from MOP51.3 billion a he onse of 2003 o MOP689.7 billion a he end of 2014. Their share in oal asses of he whole banking secor rose from an approximae 30% o almos 60% (Char 1). Char 1: Privae Secor Loans and Loan-o-Asse Raio, 2003-2014 (10 9 paacas) (%) 700 65 600 500 400 300 200 100 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 60 55 50 45 40 35 30 25 20 Toal Privae Secor Loans (lef-axis) Loan-o-Asse Raio (righ-axis) Noe: End-of-period figures. Source: Moneary Auhoriy of Macao (AMCM). - 88 -
Afer discouning for price effecs wih he composie consumer price index (CPI), credi exended o he privae secor winessed a subsanial growh of almos 700% in 2014 when compared o he level in 2003 a pace ha was even faser han Macao s real gross domesic produc (GDP) growh. In real erms, bank loans marked consecuive incremens since end-2004. Real credi growh had oupaced real GDP growh in mos of he ime since mid-2007 hough heir movemens were largely consisen wih each oher. For insance, real bank credi and he real economy winessed shor-lived conracion in he second quarer of 2009 due o he financial sunami (Char 2). Char 2: Privae Secor Loans and GDP, 2003-2014 (Year-on-year change in %) 100 80 60 40 20 0-20 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Toal privae secor loans in real erms Real GDP Sources: Macao Saisics and Census Service (DSEC) and AMCM. The significan growh of privae secor loans was winessed in boh he residen and non-residen secors, wih faser growh of loans o non-residens, highlighing he rising imporance of offshore business in he local banking secor. As a resul, he share of non-residen credi in oal privae-secor credi moved up by a marked 17.3-89 -
percenage poins from end-2003 o 50.8% a end-2014 a he expense of residen credi (Char 3). Char 3: Srucure of Toal Privae Secor Loans, 2003-2014 (10 6 paacas; end-of-period) 800,000 700,000 600,000 500,000 400,000 300,000 200,000 100,000 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Residens Non-residens Source: AMCM. While fas credi expansion cerainly benefis bank profiabiliy, i simulaneously poses risk o he financial sysem. Classic examples in he family of financial acceleraor models sugges ha banks can amplify shocks o an economy hrough credi channel. Therefore, he lending behaviour of banks is paricularly relevan during imes of financial insabiliy or crisis. Indeed, recen sudies by inernaional organisaions, such as he Bank for Inernaional Selemens (BIS) and he Inernaional Moneary Fund (IMF), underline ha financial crises are usually preceded by privae-secor credi booms and hence credi developmen can be used o consruc early warning indicaors for sysemic crises. 1 In view of he noable expansion of Macao bank credi in recen years, i is of policy significance o analyse he driving forces behind his ype of asse expansion in order o faciliae he 1 See, for example, Drehmann (2013). - 90 -
consisen monioring of he soundness of he financial sysem and he design of proacive measures from he roo of concern. The effeciveness of ransmission o credi growh depends on a hos of facors, such as source of funds, macroeconomic condiion, financial marke condiion, and moneary condiion. They could be furher classified as inernal or exernal facors. The objecive of his paper is o assess hose inernal and exernal facors for he srong credi growh in he local banking secor by a scienific model. The remainder of his paper is srucured as follows. The second secion reviews relevan lieraure. The hird secion presens relaed daa of Macao for esimaion and he esimaion resuls. The las secion concludes. 2. Lieraure Review While he rise in household debs reflecs he deepening of financial markes, i could indicae rising vulnerabiliy of household balance shees. Filardo (2008) raises a policy quesion abou wheher such developmen represens a key source of risks o he macroeconomy and how bes hese issues could be modeled. In benchmark moneary policy models, household deb usually plays a minor role as i is ypically viewed as a passive endogenous variable reflecing iner-emporal consumpion and saving decisions. He, however, employs an opimal moneary policy model including household deb, which comprises several inerrelaed blocks of equaions ha explore some of he heoreical radeoffs of a cenral bank in an economy subjec o cyclical flucuaions and asse price dynamics. I is shown ha calibraing policy implicaions would require a clear undersanding of he relaive imporance of household debs role in differen policy responses and he likely oucomes, including he possibiliy of significan ail risks arising from financial insabiliies. Aisen and Franken (2010) conduc esimaion based on cross-secion economeric echniques and a daase covering over 80 advanced, emerging and developing economies in order o idenify he drivers of bank credi growh. They sudy he - 91 -
growh of bank loans in he era of pos-lehman Brohers bankrupcy and selec i as he dependen variable in heir economeric model. I is shown ha counries wih faser economic growh afer he crisis presen higher growh of real bank credi, while economic growh of heir rading parners also helps o bring in higher loan growh. In addiion, higher credi growh is observed in economies wih money marke raes dropping he mos afer he crisis, implying ha real bank credi grows faser in counries where couner-cyclical moneary policy is more aggressive. On he oher hand, credi boom prior o he crisis is an imporan deerminan of credi conracion in he afermah of he crisis. Guo and Sepanyan (2011) conduc a cross-counry sudy abou he deerminans of bank credi growh in emerging-marke economies. They employ quarerly daa series, which are mainly sourced from he Inernaional Financial Saisics, World Economic Oulook and Global Financial Sabiliy Repor of he IMF. The daa spans from he firs quarer of 2001 o he second quarer of 2010 wih a coverage of 38 emerging economies from Asia, Europe, he Middle Eas and Africa. Bank credi o he privae secor is chosen as he dependen variable, while cerain explanaory variables are seleced in benchmark and alernaive specificaions. In he benchmark model, hey selec variables including foreign deposis, domesic deposis, real GDP, inflaion, deposi rae, exchange rae and he US Fed funds rae. In he alernaive specificaion, exernal facors such as change in bilaeral exchange raes and US M2 growh (as an alernaive measure of he global moneary condiion), as well as inernal facors such as he non-performing loan raio, are added. I is concluded ha he Asian counry group is a clear conras o oher groups in he sample. Is credi growh in all periods is mosly deermined by inernal facors, such as local GDP expansion and domesic deposis. On he oher hand, for he counry group of he European Union, credi expansion is mainly driven by foreign funding, which proves o be vulnerable o exernal shocks. - 92 -
3. Esimaion of Macao 3.1 Daa We primarily adop he mehodology presened in Guo and Sepanyan (2011) o he case of Macao. The effecs of various poenial inernal and exernal deerminans of credi growh are invesigaed by sandard economeric echniques including Uni Roo Tess and Ordinary Leas Squares (OLS). Iniially, he benchmark specificaion wih he core se of explanaory variables is employed. Addiional variables are added o he model in he second sage as an alernaive specificaion. We esimae various inernal and exernal facors for loans growh in wo ime spans, i.e. he full sample period and he period prior o he global financial crisis of 2008/09, in order o explore if here is a srucural break in he pos-crisis era. Original daa series are quarerly and sar from he year 1997. Loans o he privae secor or cusomer loans, i.e. he sum of loans & advances and bank accepances & rade bills discouned as presened in Macao s official financial saisics is specified as he dependen variable. Core independen variables include (i) cusomer deposis, which can be broken down ino residen deposis and non-residen deposis, (ii) composie CPI, (iii) GDP, (iv) hree-monh MOP ime deposi rae as local ineres rae, and (v) Hong Kong s bes lending rae as foreign ineres rae. Alernaive independen variables include (i) he non-performing loan (NPL) raio, (ii) broad money supply (M2) and (iii) economic growh of Macao s major rading parners, namely Mainland China and Hong Kong. I is noed ha all seleced variables are in he form of year-on-year change, excep ineres raes and he NPL raio. Conducing he Uni Roo Tess on he variables is required o examine wheher he daa series are saionary in order o avoid spurious OLS regression resuls for furher esimaion. We employ he Augmened Dickey Fuller (ADF) es, and is resuls are presened in Table 1. Mos of he daa series are saionary and saisically significan a he 1% or 5% level as saed. - 93 -
Table 1: Augmened Dickey Fuller Tes for Uni Roo on Seleced Variables Variable Tes Saisic Remark Cusomer loans -2.68* H 0 is rejeced. Residen loans -4.66* H 0 is rejeced. Non-residen loans -2.50** H 0 is rejeced. Residen deposis -3.48* H 0 is rejeced. Non-residen deposis -2.77* H 0 is rejeced. 1 s difference of composie CPI -4.76* H 0 is rejeced. 1 s difference of real GDP -3.90* H 0 is rejeced. 3-monh MOP ime deposi rae -2.20** H 0 is rejeced. Hong Kong bes lending rae -4.56* H 0 is rejeced. M2-3.40* H 0 is rejeced. NPL raio -2.41** H 0 is rejeced. 1 s difference of Hong Kong real GDP -5.19* H 0 is rejeced. 1 s difference of Mainland China real GDP -4.75* H 0 is rejeced. Noes: 1. Null hypohesis γ=0 agains alernaive hypohesis γ<0. Null hypohesis is rejeced when he es saisic is less han he criical value, i.e. no uni roo is presened. 2. All variables excep he NPL raio and ineres raes are expressed in year-on-year change. 3. The composie CPI of Macao is used for inflaion adjusing o obain real values of variables; he composie CPI of Hong Kong is used for inflaion adjusmen in Hong Kong s bes lending rae. 4. * and ** represen criical values of 1% and 5% respecively. Sources: Underlying daa from he AMCM, DSEC, Hong Kong Moneary Auhoriy and Hong Kong Census and Saisics Deparmen. 3.2 Esimaion models and resuls Based on Guo and Sepanyan (2011), he benchmark specificaion of he model is as follows: RL RY 4 = 0 + 1( Shdepo 4 RD ) + 2 ( Shforlia 4 RnrD ) + 3 p 4 + (1) 5 + RLDR 5 1 + RHKR 6 + ε where RL = real cusomer loans Shdepo = raio of oal deposis o oal privae secor credi RD = real residen deposis Shforlia = raio of non-residen deposis o oal privae secor credi RnrD = real non-residen deposis - 94 -
P = 1 s differenced composie CPI RY = 1 s differenced real GDP RLDR = real 3-monh MOP ime deposi rae RHKR = real Hong Kong bes lending rae In order o conrol for he overall imporance of deposis as a funding source, residen and non-residen deposi growh are respecively weighed by he raios of residen deposis o privae-secor credi and non-residen deposis o privae secor credi wih a lag of four quarers. 2 Theoreically speaking, inflaion and GDP growh would have posiive influences on credi growh. Lasly, local hree-monh MOP ime deposi rae is seleced as a proxy for local moneary sance while Hong Kong s bes lending rae is used o represen exernal financial condiions. The esimaion resuls are presened in Table 2. Resuls of he benchmark specificaion are mixed even hough he goodness of fi of he model, wih an adjused R 2 of 0.65, is accepable. The esimaed coefficiens of deposis growh in he residen secor, inflaion and GDP growh do no have he heoreically expeced posiive signs. 3 Coefficiens of oher explanaory variables carry he expeced signs bu some of hem are saisically insignifican a 5% level. 2 This paper follows he assumpion ha more deposis lead o more credi as in Guo and Sepanyan (2011). In order o reduce confounding effecs of deposi growh on credi growh, we follow he pracice of Guo and Sepanyan (2011) o conrol for he overall imporance of deposis as a funding source. We would have o divide he respecive deposi by privae secor credi, which hus gives us he percenage of loans ha is sourced by deposis. The deposi-o-credi raio in lags reflecs he imporance of deposis as he funding source wihin a reasonably recen period. By weighing deposi growh wih he raio, he model would quaniaively provide a clear look of he rue effec of deposi growh on credi growh. 3 We have considered including public-secor deposis ino residen-secor deposis. However, he inclusion does no show informaive resuls for he esimaion. - 95 -
Table 2: Regression Resuls on Real Growh of Cusomer Loans Benchmark Specificaion (1) Dependen variable: Real cusomer loans Independen Esimaed Variable Coefficien P-value Lower 95% Upper 95% Inercep 26.1989 0.0002* 13.1828 39.2149 (6.5000) RD -0.4209 0.0096* -0.7356-0.1063 (0.1571) RnrD 0.3116 0.0218** 0.0471 0.5762 (0.1321) P -1.9065 0.0007* -2.9657-0.8474 (0.5289) RY -0.0344 0.6973-0.2105 0.1417 (0.0880) RLDR -1.1674 0.2001-2.9708 0.6360 (0.9006) RHKR -1.7088 0.0992*** -3.7500 0.3325 (1.10194) Adjused R 2 : 0.65 Observaions: 64 Noes: (i) Sandard errors are in parenheses. (ii) * Significan a 1%. (iii) ** Significan a 5%. (iv) *** Significan a 10%. In he second specificaion, real cusomer loans for he residen secor replace oal real cusomer loans as he dependen variable. For he explanaory variable, he lagged NPL raio is added o capure he effec of banks healh on domesic credi growh whereas he Hong Kong bes lending rae is removed. The alernaive specificaion is as follows: RLr RY 4 = 0 + 1( Shdepo 4 RD ) + 2( Shforlia 4 RnrD ) + 3 p 4 + (2) 5 + RLDR 5 1 + NPL 6 4 + ε where RLr = real cusomer loans for he residen secor - 96 -
NPL = NPL raio Table 3 shows he esimaion resuls of he second specificaion. Table 3: Regression Resuls on Real Growh of Cusomer Loans (Residens Only) Specificaion (2) Dependen variable: Real cusomer loans for he residen secor Independen Esimaed Variable Coefficien P-value Lower 95% Upper 95% Inercep 15.1132 0.0000* 12.1408 18.0856 (1.4844) RD -0.1264 0.0996** -0.2777 0.0248 (0.0755) RnrD 0.1210 0.0488** 0.0007 0.2414 (0.0601) P -1.1101 0.0000* -1.6153-0.6048 (0.2523) RY 0.1245 0.0052*** 0.0388 0.2103 (0.0428) RLDR -1.3856 0.0000* -1.7787-0.9924 (0.1963) NPL -0.6024 0.0000* -0.8153-0.3894 (0.1064) Adjused R 2 : 0.85 Observaions: 64 Noes: (i) Sandard errors are in parenheses. (ii) * Significan a 1%. (iii) ** Significan a 5%. (iv) *** Significan a 10%. Based on he resuls of specificaion (2), economic growh, local deposi rae and banks healh saus are saisically significan variables wih heoreically expeced signs. Moreover, he goodness of fi improves wih he inclusion of he addiional explanaory variable. The posiive sign carried by he esimaed coefficien of he firs differenced real GDP growh exhibis he posiive relaionship beween credi growh and economic expansion; he coefficien indicaes ha a one percenage poin rise in real GDP growh from he same period of he preceding year spurs he real growh of residen credi o rise by 0.1 percenage poin. - 97 -
The esimaed coefficien of local ime deposi rae, which is a proxy for local moneary sance, carries a heoreically correc negaive sign, indicaing ha igher moneary sance or higher deposi ineres rae resuls in slower credi growh by abou one-o-one proporion. However, i should be noed ha ineres rae movemens of he wo SARs are similar given he linked exchange rae arrangemen beween he MOP and HKD. Hence, he disincion of is role as an exernal or inernal facor migh no be sharp o he edge. 4 Our esimaed resuls also suppor he convenional hough ha worsening banks healh reduces credi expansion. In he capioned period, he local banking secor winessed remendous improvemen in asse qualiy as refleced in he persisen downrend of he NPL raio, which would encourage he banking secor o exend loans in recen years. The negaive sign of he NPL raio coefficien implies ha deerioraion in banks asse qualiy brings abou a negaive impac on he growh of cusomer loans. A one percenage poin rise in he NPL raio is likely o drag residen credi growh by 0.6 percenage poins. However, variables ha represen he radiional source of bank funds show mixed resuls. Growh in residen deposis does no show an expeced posiive relaionship wih growh in residen credi. On he oher hand, non-residen deposi growh has a posiive impac on credi growh and he variable s coefficien is saisically significan. In he hird specificaion, he dependen variable is he real growh of cusomer loans for he non-residen secor. We apply Hong Kong s bes lending rae o represen he exernal financial environmen in view of he close financial relaionship of Macao wih Hong Kong. Meanwhile, a weighed average of Mainland China and Hong Kong GDP is included o reflec he performance of Macao s major economic parners. 4 We have ried broad money supply M2, insead of local ineres rae, as a proxy for Macao s moneary condiion. However, he broad money series do no show informaive resuls in our model. Therefore, he local deposi rae series remain in use. - 98 -
Mainland China and Hong Kong are also radiionally leading desinaions for fund placemens of Macao banks. The specificaion (3) is as follows: NRLr RY 4 = 0 + 1( Shdepo 4 RD ) + 2 ( Shforlia 4 RnrD ) 5 + RHKR 5 1 + NPL 6 4 + CNHK 7 5 + ε + p 3 4 + (3) where NRLr = real cusomer loans for he non-residen secor RHKR = real Hong Kong s bes lending rae CNHK = 1 s differenced real GDP of Mainland China and Hong Kong Table 4 shows he esimaion resuls of he hird specificaion. The esimaion resuls for specificaion (3) also show ha igher exernal liquidiy condiions would reduce credi exension by he local banking secor, especially under an environmen of similar ineres rae movemens beween Macao and Hong Kong. The change of Hong Kong s bes lending rae, a variable o reflec exernal moneary condiions, is a saisically significan deerminan of real growh of credi o he nonresiden secor. A one percenage poin rise in he Hong Kong lending rae is likely o drag non-residen credi growh o reduce by 3.7 percenage poins. However, he low level of he adjused R 2 suggess ha he dependen variable in ha specificaion was no explained well by hose explanaory variables. - 99 -
Table 4: Regression Resuls on Real Growh of Cusomer Loans (Non-residens Only) Specificaion (3) Dependen variable: Real cusomer loans for he non-residen secor Independen Esimaed Variable Coefficien P-value Lower 95% Upper 95% Inercep 52.3037 8.82E-11* 39.1493 65.4582 (6.5666) RD -0.8934 0.0316** -1.7053-0.0815 (0.4053) RnrD 0.4421 0.1738-0.2007 1.0850 (0.3209) P -4.5806 0.0191** -8.3827-0.7786 (1.8979) RY -0.4771 0.0930*** -1.0364 0.0821 (0.2792) RHKR -3.7057 0.0019* -5.9853-1.4262 (1.1379) NPL 0.0422 0.9404-1.0831 1.1675 (0.5617) CNHK 1.6154 0.1440-0.5684 3.7993 (1.0902) Adjused R 2 : 0.36 Observaions: 64 Noes: (i) Sandard errors are in parenheses. (ii) * Significan a 1%. (iii) ** Significan a 5%. (iv) *** Significan a 10%. To observe he siuaion before he global financial crisis in 2008 or possible difference in pre-crisis era, we proceed o run he regression of specificaions (2) and (3) for he pre-crisis period. Esimaion resuls are shown in Table 5. As for he real growh of credi exended o residens in he pre-crisis sample period, he esimaed coefficien of lagged local ineres rae coninues o carry a heoreically expeced sign and is saisically significan a he 1% level, indicaing ha he rise in local ineres rae has a negaive impac on credi growh. The NPL raio, which reflecs he healh of he banking secor, also remains significan in he period. On he exernal fron, non-residen deposis becomes saisically significan in he pre-crisis - 100 -
sample period; he posiive-sign esimaed coefficien reflecs ha a one percenage poin rise in non-residen deposis growh pushes residen credi growh up by a moderae 0.2 percenage poins. Table 5: Regression Resuls on Real Growh of Cusomer Loans in Pre-crisis Period Specificaions (2) & (3) Pre-crisis Sample: 1999-2008Q2 Real Cusomer Loans (Residens) Real Cusomer Loans (Non-residens) Independen Variable Esimaed Coefficien P-value Lower 95% Upper 95% Esimaed Coefficien P-value Lower 95% Upper 95% Inercep 15.9331 4.3E-06* 10.0896 21.7766 49.9636 0.0037* 17.5375 82.3898 (2.8651) (15.8775) RD -0.1631 0.1250-0.3740 0.0479-0.9156 0.1855-2.2955 0.4643 (0.1034) (0.6757) RnrD 0.1730 0.0269* 0.0211 0.3249 0.1933 0.7020-0.8286 1.2152 (0.0745) (0.5004) P -1.4216 0.0171* -2.5722-0.2710-2.4680 0.5995-11.9635 7.0276 (0.5642) (4.6495) RY 0.0118 0.8323-0.1012 0.1248-0.0618 0.8672-0.8098 0.6863 (0.0554) (0.3663) RLDR -1.0077 0.0049* -1.6856-0.3298 (0.3324) NPL -0.6906 0.0002* -1.0239-0.3573-1.8246 0.0086* -3.1500-0.4992 (0.1634) (0.6490) RHKR -5.1260 0.0489* -10.2255-0.0266 (2.4970) CNHK 0.0880 0.9706-4.7462 4.9222 (2.3671) Adjused R 2 : 0.83 0.41 Observaions: 38 38 Noes: (i) Sandard errors are in parenheses. (ii) * Significan a 1%. Concurrenly, esimaed resuls for real growh of credi exended o non-residens in he pre-crisis sample period are largely in line wih he resuls for he full sample period. The exernal moneary condiions coninue o show saisically significan impac on he dependen variable. Similarly o he resuls for he full-sample period, - 101 -
he goodness of fi of he non-residen model in he pre-crisis sample period is no as saisfacory as ha of he residen model. 4. Conclusion: Inerpreaions of Resuls This paper assesses possible inernal and exernal facors for bank credi growh in Macao by a model originally developed by Guo and Sepanyan (2011). Separae OLS regressions are run for he residen and non-residen secors. Our esimaion shows ha loans exended o residens are beer presened by he economeric model in erms of he goodness of fi. Neverheless, esimaion resuls separaely for he residen and non-residen secors give differen insighs and resuls for full-sample period and pre-crisis sample period are generally consisen wih each oher, possibly reflecing he relaively mild impac of he financial sunami on he Macao financial sysem. Inernal facors as a whole play a more imporan role han exernal facors in credi expansion in Macao, especially for he residen secor or domesic credi. Saisically significan inernal deerminans include he healh saus of banks represened by he NPL raio, local economic performances represened by real GDP growh, and domesic moneary condiions by local ime deposi ineres rae. Our resuls are generally consisen wih he original lieraure s findings on he Asian counry group. The significance of he inernal facor, i.e. he NPL raio, is undersandable, as Macao banks would adop a more cauious credi sraegy when he NPL raio is high, as experienced in he Asian financial crisis. In view of he subsanial improvemen in banks asse qualiy, Macao banks have been winessed o become more aggressive in lending, resuling in a noable pick-up in bank credi in recen years. Neverheless, i should be noed ha he NPL raio is a lagging indicaor of bank asse qualiy, as presened in our model, and could deeriorae swifly in economically or financially difficul imes. Therefore, local banks should formulae proper expecaions for fuure asse qualiy, prudenly incorporaing a forward-looking aiude in credi operaion. - 102 -
In he meanime, our esimaion resuls sugges ha Macao s GDP growh has a moderae posiive impac on credi growh. Buoyan economic environmens encourage business invesmen and privae consumpion, which would require more bank finance. The srong economy of Macao, especially in he pas decade, was advanageous o banks o operae and mainain business healh, which also proves o be significan in our model. Looking ahead, he consolidaing economy of Macao in he shor erm could creae some cooling effec on credi growh. Banks should remain aenive o loans qualiy and he associaed credi risk in view of he changing economic environmen. Domesic moneary sance also shows a saisically significan impac on credi growh for he residen secor. On he fron of bank credi o he non-residen secor, exernal facors, paricularly exernal moneary condiions, appear o have more significan impacs on exernal credi growh. I should be noed ha he local ineres-rae rend largely follows ha in Hong Kong or he USA. The exremely low ineres-rae environmen in he pas few years has creaed relaxed moneary condiions, which have induced credi o grow. The US Federal Reserve has already sared is process of moneary policy normalisaion and is expecing o raise he benchmark Fed funds rae in he near fuure. The rae hike cycle could have a negaive impac on credi growh in general. Our esimaion resuls show ha deposis generally do no have significan effec on credi growh, excep for residen credi in he pre-crisis period. Tradiionally, deposis are an imporan source of funds in he local banking secor and mainaining a balance beween loans and deposis is imporan for he sabiliy of he financial secor. The weak linkage beween deposis and loans suggesed by our esimaion resuls heighens cerain concern for financial vulnerabiliy, especially when credi growh coninues o oupace ha of deposis and he loan-o-deposi raio is a relaively high level. The liquidiy posiion of banks is also likely o be weakened. - 103 -
Lasly, he model s goodness of fi in he specificaion for he residen secor is saisfacory in he full sample period and he pre-crisis sample period. Ye, he values of he adjused R 2 for he non-residen secor series in boh sample periods sand comparaively lower han is counerpar. Such a resul migh sugges ha some oher poenial facors for exernal credi growh, which are no included in exising lieraure, are missing in our model. I would require furher research in he fuure in view of he increasing imporance of exernal credi in Macao banks asse porfolio. - 104 -
References Aisen, A., and M. Franken (2010), Bank Credi during he 2008 Financial Crisis: A Cross-Counry Comparison, IMF Working Paper, No. 10/47. Cheang, N. and I. Choy (2011), Aggregae Financial Sabiliy Index for an Early Warning Sysem, Macao Moneary Research Bullein, Issue No. 21, 27-51. Drehmann, M. (2013), Toal Credi as an Early Warning Indicaor for Sysemic Banking Crises, BIS Quarerly Review, June. Filardo, A. (2008), Household Deb, Moneary Policy and Financial Sabiliy: Sill Searching for a Unifying Model, BIS Asian Office Research Research Papers, 22 April. Gigineishvil N. (2011), Deerminans of Ineres Rae Pass-Through: Do Macroeconomic Condiions and Financial Marke Srucure Maer?, IMF Working Paper, No. 11/176. Guo, K. and V. Sepanyan (2011), Deerminans of Bank Credi in Emerging Marke Economies, IMF Working Paper, No. 11/51. Liu, M. H. and M. Wang (2012), Deposi Rae Pass-hrough and Compeiion: Evidence of Macao, Macao Moneary Research Bullein, Issue No. 22, 83-109. McGuire, P., and N. Tarashev (2008), Bank Healh and Lending o Emerging Markes, BIS Quarerly Review, December, 67-80. Takas, E. (2010), Was i Credi Supply? Cross-border Bank Lending o Emerging Marke Economies during he Financial Crisis, BIS Quarerly Review, June, 49-56. - 105 -