Capital Flows and Money Supply: The Degree of Sterilisation in Pakistan

Transcription

1 The Pakisan Developmen Review 42 : 4 Par II (Winer 23) pp Capial Flows and Money Supply: The Degree of Serilisaion in Pakisan ABDUL QAYYUM and MUHAMMAD ARSHAD KHAN * 1. INTRODUCTION Under he curren managed floa exchange rae sysem; he cenral bank may respond o an exchange marke disequilibria by changing eiher he inernaional reserves or he exchange raes. Under such a regime, a major policy difficuly is he ineracion beween exchange rae policies and moneary policies. The moneary auhoriies inervene in he exchange marke in response o undesired flucuaions in exchange raes, 1 could adversely affec moneary conrol and move he economy away from inernal arge such as price sabiliy. Under such a policy dilemma, fully serilised inervenion 2 involves a pure swap of foreign and domesic asses, which have no effec on he money supply, received greaer aenion by he policy-makers in early 198s, paricularly, hrough he experience of Wes Germany [Obsfeld (1983)]. Ideally, i provides an independen policy ool o deal wih he exchange rae wihou affecing he inernal policy arges. Moreover, i is argues ha fully serilised inervenion insulae domesic policies compleely from balance of paymens consideraions. Furher, he effecs of inervenion on exchange raes are close o zero if inervenion is compleely serilised. Given his convicion, i is hard o see why he cenral bank would inervene in he foreign exchange marke and serilised compleely a he same ime [Neumann (1984)]. I is furher argued ha serilisaion is capable o move exchange raes hrough eiher a porfolio or signaling channel. In developing counries, an inervenion may no be used purely o sabilise exchange rae bu o reduce is impacs of volaile exchange raes on price level. Abdul Qayyum is Associae Professor, Pakisan Insiue of Developmen Economics, Islamabad. Muhammad Arshad Khan is Associae Professor, Governmen Inermediae College, Nar (Koli), Azad Jammu and Kashmir. 1 Official inervenion in foreign exchange marke occurs when he auhoriies buy or sell foreign currency agains heir own currency in order o affec he exchange rae. 2 Serilised inervenion occurs when he auhoriies simulaneously ake acion o offse or serilise he effec of he resuling change in official foreign asses holdings on he domesic moneary base.

2 976 Qayyum and Khan In his paper, we specifically invesigae how Sae Bank of Pakisan s inervenions respond o exchange rae shocks in erms of he degree of serilisaion and is effeciveness on limiing he moneary impac of shocks and how foreign exchange reserves respond o changes in domesic credi. Previous empirical evidence, iner alia, Kouri and Porer (1974); Argy and Kouri (1974); Ujiie (1978); Connolly and Taylor (1989) and Ki Pasula (1994) measured he degree of serilisaion by uilising he simple moneary model of he balance of paymens under he fixed exchange rae sysem. Since 1973 he lieraure has sressed he imporance of including he exchange raes and foreign ineres raes as deerminans of money demand. Tower (1975); Arango and Nadiri (1981); 3 McNown and Wallace (1992); Bahmani-Oskooe and Rhree (1994); Bahmani-Oskooe (1996) and Bahmani-Oskooe and Techaraanachai (21) have shown he exisence of a relaionship beween money demand and exchange raes. 4 Similarly, Arango and Nadiri (1981) and Booh and Chawdhury (1992) have sressed he imporance of foreign ineres raes as a deerminan of money demand because of is effecs on he desired sock of real cash balances and exchange rae expecaions. The objecive of his paper is o esimae he domesic credi policy reacion funcion o deermine he degree of serilisaion by combining he moneary model of he balance of paymens and he recen developmens in he money demand lieraure 5 by employing Johansen (1988) mulivariae coinegraion echnique for Pakisan. 6 The reminder of he paper is organised as follows: Secion 2 briefly describes he Sae Bank of Pakisan (SBP) serilisaion policy. Secion 3 conains he model specificaion. Secion 4 discusses he daa descripion, economeric mehodology and empirical resuls while summary and some concluding remarks are given in he final secion. 2. STERILISATION POLICY OF THE STATE BANK OF PAKISTAN In recen years, 7 here is a sharp improvemen in he SBP foreign exchange inflows. This increase in foreign exchange inflows placed upward pressure on he 3 They argued ha a depreciaion of he domesic currency increases he value of foreign securiies held by domesic individuals. If his increase is perceived as an increase in wealh, he demand for domesic cash balances may increases. Furher, when he exchange rae is expeced o depreciae he expeced reurn from holding foreign money increases and he demand for domesic currency falls (as individuals subsiue foreign money for domesic currency). 4 For example, Bahmani-Oskooe and Techaraanachai (21) argued ha he exchange rae in addiion o income and ineres rae is included as oher deerminan of money holdings. 5 Which incorporaes real exchange raes and foreign ineres raes as addiional argumens of he money demand funcion. 6 Using quarerly daa covering he period 1982Q3 o 21Q2. 7 Ocober 21 onwards. For furher deail see SBP Annual Repor for FY2.

3 Capial Flows, Money Supply, and Serilisaion 977 value of he Rupee/Dollar exchange rae, which appreciaed over 8 percen. The appreciaion in he rupee value has been considerably larger if he SBP had no inervened in he inerbank foreign exchange marke, purchasing US dollar o preven rupee from appreciaing. 8 This massive injecion of he rupee liquidiy in he banking sysem by SBP has serious repercussions for he moneary policy, banking secor sabiliy and inflaionary expecaions in he economy. Thus, i is necessary for he cenral bank o sabilise changes in foreign exchange reserves and inervene in he foreign exchange marke simulaneously. On pos Sepember 11, 21, he foreign exchange reserves of he SBP sharply increased. Under his siuaion, he cenral bank aemp o preven changes in ne foreign asses from affecing reserve money by implemening offseing changes in ne domesic asses. During FY2, he reiremen of Rs 287 billion worh of governmen securiies wih SBP more han offse he impac of SBP inervenion in he foreign exchange marke. Consequenly, he growh of reserve money was held down o 9.6 percen despie sizeable foreign exchange purchases of SBP 9. During he firs quarer of FY3, SBP purchased 1.28 billion US dollar from foreign exchange marke ha led o an equivalen injecion of Rs 67 billion. Meanwhile, he governmen raised Rs 127 billion from commercial banks and used hese funds o reire SBP deb. In overall erms, SBP holding of Governmen of Pakisan (GOP) securiies fell by Rs 78 billion (see Table 1). This includes a oal of Rs 1.7 billion worh T-bills ha were off-loaded via open marke operaions. Table 1 Serilisaion during Third Quarer of Fiscal Year 23 (Rs Billion) Impac on Ne Foreign Asses Ne Domesic Asses Inerbank US Dollar Purchases (Including Expor of Currencies) 76 SBP Holding GOP Securiies 78 Ne Impac Ne Impac on Reserve Money 2 Source: Sae of Pakisan s Economy: Firs Quarerly Repor for fiscal year 23. The immediae cos of serilisaion will be he difference beween falling SBP earnings from lower T-bills holdings and he reurns on SBP invesmen of foreign currency reserves due o he falling T-bills rae. The fall in SBP profi led lower ransfers o he governmen. Hence, he shorfall in governmen non-ax revenue 8 SBP increased is official reserves hrough he purchase of US Dollar in order o mainain exchange rae. 9 See Sae Bank of Pakisan, Annual Repor for Fiscal Year 22.

4 978 Qayyum and Khan receips by a leas Rs 2 billion. The SBP T-bill holdings declined for Rs 485 billion in June FY1 o Rs 119 billion in Sepember 22. This massive decline of Rs 366 billion is no enirely aribued o serilisaion. In fac, T-bills worh of Rs 197 billion were adjused agains special deb repaymen accoun wih SBP. This implies ha T-bill holdings due o serilisaion alone acually fell by Rs 169 billion. Thus we may conclude ha given he naure of exernal shock ha promped abrup foreign exchange inflows, here was a lile opion for SBP o inervene in he forex marke, and serilise is impac on moneary policy, banking sysem and inflaionary expecaions. Moreover, serilisaion policy always involves direc coss, bu he choice of serilisaion insrumen simply shifs hese coss on he concerned economic agens including he commercial banks, he SBP and he governmen. 3. MODEL SPECIFICATION To examine he degree of serilisaion, i is necessary o derive a domesic credi policy reacion funcion. The reacion funcion can be derived from wo equaions, which summarise he moneary approach o he balance of paymens. d β β 1 2 M = AY Q exp( β3i + β4i + u ) (1) s M = κ( D + R)... (2) In Equaion (1), money demand (M d ) is a funcion of real income (Y), real exchange rae (Q), domesic ineres rae (i) and foreign ineres rae (i * ). The parameers β 1, β 2, β 3, β 4 are assumed o be consan elasiciies and A a scale facor. In Equaion (2), money supply (M s ) is composed of money muliplier (k) of domesic (D) and foreign componens (R) of moneary base. From Equaion (2) we 1 s obain D = ( M ) R which demonsraes he link beween domesic credi and κ foreign exchange reserves. When he cenral bank buys foreign currency from he foreign exchange marke, R goes up. The rise in R can be serilised by reducing he cenral bank s credi holdings (D). If he change in R is fully mached by an offseing change in D, here is no change in M s. Assuming moneary equilibrium where M d =M s, we have 1 d D = ( M ) R (3) κ Taking logarihms of Equaion (3) we obain d ln D = ln M ln κ ln R (4) Using logarihms of Equaion (1) and subsiuing in Equaion (4) we ge he following equaion

5 Capial Flows, Money Supply, and Serilisaion 979 ln D = ln A + β1 lny + β2 lnq + β3i + β4i ln κ ln R + u (5) We can wrie Equaion (5) in he following esimaable form + β1 y + β2q + β3i + β4i + β5m + β6r u (6) d = β + Where lower case leer represens he logarihms of variables involved in Equaion (5), β = lna and m is he logarihm of money muliplier and u is he error erm. The coefficien of r i.e. β 6 is called he serilisaion coefficien. I measures he hrus of moneary policy o serilise he impac of inernaional reserve flows on moneary base. The serilisaion coefficien ranging in value from zero (no serilisaion) o 1 (full serilisaion). If he serilisaion coefficien (β 6 ) is zero, hen no serilisaion akes place. If β 6 = 1, serilisaion is complee. Bu when he resuls of regression are such ha 1<β 6 <, serilisaion is conduced incompleely by he SBP. Equaion (6) may be subjec o simulaneous bias. However, using mulivariae coinegraion echniques, we can obain unbiased long-run elasiciies of he deerminans of money demand and moneary base. 4. DATA DESCRIPTION, ECONOMETRIC METHODOLOGY, AND EMPIRICAL RESULTS For he empirical analysis we employed quarerly daa ranging from 1982Q3 o 21Q2 Quarerly daa of GDP (y) was no available, hence was generaed quarerly observaions from annual observaions. 1 Inernaional reserve (r) excluded gold are defined in domesic currency. Domesic credi (d) is calculaed by aking he difference beween he cenral bank reserve money and gross foreign asses. For domesic (foreign) ineres rae i (i * ), inerbank call money rae and Unied Saes Federal Fund Rae is used respecively. Money muliplier (m) is obained from dividing reserve money by M 2. Real exchange rae is calculaed as q = s+p * p, where s is he annual average exchange rae (IFS line rf) and p * (p) are respecively he US and Pak WPI. 11 All he daa are aken from various issues of Inernaional Financial Saisics (IFS). Excep domesic (foreign) ineres rae all oher are expressed d in logarihmic form. Time series daa on mos of he macroeconomic variables ends o be nonsaionary in levels. For meaningful resuls, he difference beween he variable and is lag should be saionary. Generally, a nonsaionay series becomes saionary when difference d, ime is said o be inegraed of order d; represened as X ~ I (d), 1 Quarerly observaions for GDP are generaed from he annual GDP daa, following he mehodology developed by Goldsein and Khan (1976). 11 Whole Sale Price Index ( 1995=1).

6 98 Qayyum and Khan and combinaion of hese variables are coinegraed when one or more linear combinaions are deeced. In his sudy, Augmened Dickey-Fuller (ADF) es is used o deermine he saionariy and degree of inegraion of he variables. Lagrange Muliplier (LM) es is used for he whie noisiness of he residuals. The resuls of ADF es are repored in Table 2. Table 2 Uni Roo Tes Variable Trend Lags ADF Saisics Diagnosis D no I(1) d no * y no I(1) y no * q no I(1) q no i no I(1) i no * i * no I(1) i * no * m no I(1) m no * R yes I(1) r yes * *,** Indicae significan a 1 percen and 5 percen level respecively. The criical values are given by Mackinnon (1991). I appears from he Table 2 ha, in levels all he variables are non-saionary while he firs difference of hese variables become saionary, indicaing ha X(= d, y, q, i, i*, m, r) ~ I(1). To check for he linear combinaion beween he variables he Jahonsen (1988) and Johansen and Juselius (199) mulivariae coinegraion echnique is used. To his end, wo likelihood raio ess based on race of he sochasic marix (λ-race) and based on maximal eigenvalue of he sochasic marix (λ-max) are used. In he likelihood raio ess, he null hypohesis ha here are a mos r (where r=, 1.7) coinegraing vecors is esed agains he alernaive of r+1 coinegraing vecors. The resuls of Table 3 indicae ha he null hypohesis of no coinegraion is rejeced and here exi a mos 6 coinegraing vecors based on λ- race and a mos 2 significan coinegraing vecor based on λ-max. This implies ha here exis a long-run relaionship beween domesic credi, real income, real exchange rae, domesic ineres rae, foreign ineres rae, money muliplier and inernaional reserves.

7 Capial Flows, Money Supply, and Serilisaion 981 Table 3 Resuls of Coinegraing Analysis Series X (= d, y, q, i, i*, m, r) and lag 4 Hypohesis Eigenvalue λrace 95% CV Alernaive λmax 95% CV r = * r = * r < * r = * 4.3 r < * 76.7 r = r < * r = r < * r = r < ** r = r = r = Criical values come from Oserwald-Lenum (1992). Also see Banerjee, e al. (1993), Table 8.7, p *, ** Indicae significan a 1 percen and 5 percen respecively. CV= criical values. To deermine he sign and magniude of he long-run elasiciies in Equaion 6, he coinegraing vecors have been normalised on d. The normalised vecor corresponds wih he maximum eigenvalue and he resuls are repored in Table 4. These resul show ha apar from real exchange rae all oher variables are saisically significan a he 1 percen level of significance and possessed expeced signs. The coefficien of inernaional reserves (r) is posiive and significan, indicaing some serilisaion. Table 4 Significan Coinegraing Vecor Normalised on d d y q i i* m r Consan (.479) (.9721) (.2478) (.1488) (.38488) (.4329) ( ) ( ) (.9623) ( ) (4.3611) (4.8138) (6.4116) (1.3417) Noe: Figures in parenheses represen he sandard errors and -values respecively. The degree o which he SBP serilises he foreign exchange reserves is equal o (1 β 6 ), which is approximaely equal o.72. This implies ha he SBP serilises 72 percen of capial inflows during he period of sudy. Esimaion of Error-correcion Model The performance of he domesic credi policy reacion funcion can be improved by inroducing shor-run dynamic ino he saic model. According o Granger (1987) represenaion heorem, if a coinegraing relaionship exiss beween a series of I (1) variables, hen a dynamic error-correcion represenaion also exiss. The general form of he error-correcion model can be wrien as

8 982 Qayyum and Khan d = α + β d i 1 + τ m i i i + γ y i + λ r i i i + δ q + φε 1 where he error-correcion erm ε is given by i i + µ + θ i i i + η ii i ε = d dˆ φ reflecs he speed of adjusmen owards he long-run equilibrium value, which is deermined by he normalised coinegraing coefficiens given in he Table 4. All he variables in regression Equaion (7) are I(1) saionary. Hence he convenional - saisics based on OLS mehod can be used o derive inference on he magniudes of he esimaed coefficiens. Using Hendry s general o specific mehodology (saring wih 4 lags) Table 5 represens he parsimonious error-correcion model, where he convenional - saisics are given in parenheses. Table 5 (i = 1,2.4) (7) Error-correcion Model for Domesic Credi Reacion Funcion Independen Variables Coefficiens -saisics Consan d y * q * i * i ** m * m ** r * ε * R 2 =.7773 R 2 adj = S.E = F-sa(9, 66) = * FF[F (1,65)] = NO[χ 2 (2)] = HS[F(1,74)] =.3693 SC[F(4,62)] =.1394 DW. sa = Predicive Failure (1,66) = ** and * represens significance a he 5 percen and 1 percen level of significance respecively. FF, NO, He and SC represens Ramsey s RESET es of funcional form, Jarque-Bera s normaliy es, es of heeroscedasiciy based on ARCH es and serial correlaion es based on Lagrange Muliplier es. These ess are disribued as F-sa and χ 2 wih d..f is given in parenheses.

9 Capial Flows, Money Supply, and Serilisaion 983 The resuls from he esimaed error-correcion model are presened in Table 5. All he esimaed coefficiens are significanly differen from zero a he 5 percen level of significance excep he coefficien of changes in domesic credi lagged by wo periods ( d 2 ). The error-correcion erm possessed expeced sign and saisically significan. However, he size of he error-correcion erm is.1924, which implies ha abou 11 percen adjusmen owards long-run equilibrium akes place in one quarer. The shor-run elasiciies are smaller han he long-run elasiciies. Furhermore, he coefficien of foreign exchange reserves is.12486, which sugges ha he SBP serilises 88 percen of capial inflows in he shor-run. Moreover, when he esimaed values of d are fied agains he acual values, i perform well in erms of racking he cyclical naure of he movemens in Pakisan represened by he Figure 1. Fig. 1. Plo of Acual and Fied Values Q2 1985Q4 1988Q2 199Q4 1993Q2 1995Q4 1998Q2 2Q4 22Q1 DLDC4 Fied Quarers The esimaed error-correcion model was found o be sable no only in erms of Chow forecas es bu also in erms CUSUM of squares es. Figure 2 plos he CUSUM of squares of recursive residuals Fig. 2. Plo of Cumulaive Sum of Squares of Recursive Residuals Q1 1985Q3 1988Q1 199Q3 1993Q1 1995Q3 1998Q1 2Q3 The sraigh lines represen criical bounds a 5 percen significance level 21Q4

10 984 Qayyum and Khan CUSUM of squares pah lies wihin he bands, indicaing no srucural change a he 5 percen level. 5. SUMMARY AND CONCLUSIONS In his sudy an aemp has been made o develop and esimae he domesic credi policy reacion funcion o analyse he moneary implicaions of inervenions and serilisaion policy in Pakisan using quarerly daa ranging from 1982Q3 hrough 21Q2. By employing Johansen mulivariae coinegraion echnique, his paper has considered he degree of serilisaion ha he Pakisan has used in conrolling capial flows. The degree o which a change in inernaional reserves affecs domesic credi is measured by he coefficien β 6. The rae a which he SBP serilises inernaional reserve flows is measured by (1-β 6 ). The evidence suggess ha Pakisan serilises around 72 percen of inernaional reserve inflows in he long-run while 88 percen in he shor-run during he period of sudy. REFERENCES Arango, S., and M. Nadiri (1981) Demand for Money in Open Economies. Journal of Moneary Economics 7, Argy, V., and P. Kouri (1974) Serilisaion Policies and he Volailiy in Inernaional Reserves. In R. Z. Aliber (ed.) Naional Moneary Policies and he Inernaional Financial Sysem. Chicago: Universiy of Chicago Press Bahmani-Oskooee, M., and H. J. Rhree (1994) Long-run Elasiciies of he Demand for Money in Korea: Evidence from Coinegraion Analysis. Inernaional Economic Journal 8:2, Bahmani-Oskooe, M. (1996) The Black Marke Exchange Rae and Demand for Money in Iran. Journal of Macroeconomics 18, Bahmani-Oskooe, M., and A. Techaraanachai (21) Currency Subsiuion in Thailand. Journal of Policy Modelling 23, Booh, G., and M. Chowdhury (1992) Canadian Foreign Exchange Policies: Inervenion, Conrol, Coinegraion. Welwirschafliches Archv 128:1, Banerjee, A., J. J. Dolado, J. W. Galbraih, and D. F. Hendry (1993) Coinegraion, Error-Correcion, and he Economeric Analysis of Non-saionary Daa. Advanced Texs in Economerics. Oxford: Oxford Universiy Press, UK. Connolly, M., and D. Taylor (1989) Adjusmen o Ineres Rae Shocks in Lain America and he Caribbean. In P. Brock, M. Connolly and C. Gonzalez-Vega (eds.) Lain American Deb and Adjusmen: Exernal Shocks and Macroeconomic Policies. New York: Praeger Goldsein, M., and M. S. Khan (1976) Large Versus Small Price Changes in he Demand for Impors. IMF Saff Papers 23, Inernaional Moneary Fund (Various Issues) Inernaional Financial Saisics (IFS).

11 Capial Flows, Money Supply, and Serilisaion 985 Inoguchi, M. ( 23) How did he Bank of Thailand Respond o Capial Inflows Before a Crisis? Serilisaion and Base Money in The 199s. The Singapore Economic Review 48:1, Johansen, S. (1988) Saisical Analysis of Coinegraion Vecors. Journal of Economic Dynamics and Conrol 12, Johansen, S., and K. Juselius (199) Maximum Likelihood Esimaion and Inference on Coinegraion wih Applicaions o he Demand for Money. Oxford Bullein of Economics and Saisics 52, Kouri, P., and M. Porer (1974) Inernaional Capial Flows and Porfolio Equilibrium. Journal of Poliical Economy 82, McNown, R., and M. Wallace ( 1992) Coinegraion Tess of a Long-run Relaion beween Money Demand and he Effecive Exchange Rae. Journal of Inernaional Money and Finance 11, Mackinnon, J. (1991) Criical Values for Coinegraion Tess. In R. F. Engle and C. W. J. Granger (eds.) Long-run Economic Relaionships: Reading in Coinegraion. Chaper 13. Oxford: Oxford Universiy Press, UK. Naumann, M. J. M. (1984) Inervenion in he Mark-Dollar Marke: The Auhoriies Reacion Funcion. Journal of Inernaional Money and Finance 3, Obsfeld, M. (1983) Exchange Rae, Inflaion and he Serilisaion Problem: Germany, European Economic Review 21, Oserwald-Lemun, M. (1992) A Noe wih Quaniles of he Asympoic Disribuion of he ML Coinegraion Rank Tes Saisics. Oxford Bullein of Economics and Saisics 54, Pasula, K. (1994) Serilisaion, Ricardian Equivalence and Srucural and Reduced Form Esimaes of he Offse Coefficien. Journal of Macroeconomics 111:4, Sae Bank of Pakisan (23) Sae of Pakisan s Economy: Firs Quarerly Repor for FY3. Sae Bank of Pakisan (Various Issues) Annual Repors for FY1, FY2 and FY3. Tower, E. (1975) Money Demand and he Term of Trade. Welwirschafliches Archiv 111:4, Ujiie, J. (1978) A Sock Adjusmen Approach o Moneary Policy and he Balance of Paymens in he Economics of Exchange Raes. In Frenkel and H. Johnson (eds.) Seleced Sudies. Longmand: Addison-Wesley