Modelling Central Bank Intervention Activity under Inflation Targeting

Transcription

1 MPRA Munich Personal RePEc Archive Modelling Cenral Bank Inervenion Aciviy under Inflaion Targeing Horvah, Roman Czech Naional Bank 0. May 2006 Online a hp://mpra.ub.uni-muenchen.de/4/ MPRA Paper No. 4, posed 07. November 2007 / 0:23

2 Modelling Cenral Bank Inervenion Aciviy under Inflaion Targeing Roman Horváh * Czech Naional Bank and Insiue of Economic Sudies, Charles Universiy ABSTRACT Using daily daa from he Czech Republic in //8-3/2/2002, we find ha foreign exchange inervenion aciviy is deermined by he degree of exchange rae misalignmen and lagged inervenion. Addiionally, inflaion argeing regime is a binding consrain of inervenion aciviy. JEL Classificaion: F3; F33 Keywords: Foreign exchange inervenion, cenral bank, inflaion argeing * Conacs: Roman Horváh, Czech Naional Bank, Na Prikope 28, 03 Prague, Czech Republic, Tel.: , fax: , roman.horvah@cnb.cz or roman.horvah@gmail.com I hank Rober F. Engle for helpful discussions. I also hank Ian Babeskii, Jarko Fidrmuc and Luboš Komárek for he help in obaining he daa used in his paper. All remaining errors are enirely my own. The views expressed here do no necessarily represen hose of he Czech Naional Bank.

3 . Inroducion Several sudies have recenly analyzed he deerminans of cenral bank s foreign exchange (FX) inervenion aciviy under flexible exchange raes (Almekinders and Eijffinger, 4, 6; Baillie and Oserberg, 7; Neely, 2002; Kim and Sheen, 2002; Frenkel, Pierdzioch and Sadmann, 2004; and Io and Yabu, 2006). Typically, hey find ha cenral bank (CB hereafer) inervenes in he FX marke, when he exchange rae is largely misaligned and/or excessively volaile. To our knowledge, CB inervenion lieraure does no explicily link FX inervenion behavior o CB s main arges, which is low inflaion and/or oupu close o is poenial. This is sriking, as CBs ypically do no formulae heir FX inervenion policy in an ad hoc manner (reacing only o informaion in exchange rae), bu raher in line wih heir moneary policy regime oriened owards achieving he arges. In his paper, we focus on assessing how moneary policy regime (inflaion argeing in our case) affecs CB s FX inervenion aciviy. We build on anecdoal evidence by Holub (2004), who proposes he crieria for assessing when FX inervenions are consisen wih he inflaion argeing regime. The crierion we analyze in his paper is simply ha he CB should ighen he moneary condiions when inflaion and oupu forecas is heading above is argeed value and vice versa. If CB inervenions are in line wih he aforemenioned crierion, we label i as consisen wih inflaion argeing (IT consisency hereafer). 2 See Sarno and Taylor (200) for comprehensive survey of FX inervenion lieraure. 2 See Holub (2004, p. 4) for evidence when CBs likely violaed his crierion. 2

4 As a resul, his paper aims o bridge he gap in evidence on he linkages beween FX inervenion aciviy and inflaion argeing regime using he daily daa from he Czech Republic. Firs, we consruc a simple indicaor of IT consisency based on deviaion of inflaion from is arge and oupu gap. Second, we esimae he deerminans of FX inervenion aciviy employing he measures of exchange rae misalignmen and volailiy and IT consisency. The paper is srucured as follows. Daa and empirical mehodology are described in secion 2. Resuls are presened in secion 3. Conclusion follows in secion Daa and Mehodology The sample period runs from //8 o 3/2/2002. We use daily daa on acual FX inervenions and CZK/EUR exchange rae (prior o CZK/EUR is rerieved from CZK/DEM and irrevocable fixing rae of DEM/EUR). We also employ monhly daa on ne inflaion, as he inflaion arge of he Czech Naional Bank (CNB) has been defined in erms of ne inflaion during he sample period. 3 Real GDP growh daa are available on a quarerly basis (HP filer is used o esimae poenial growh). 3 CNB sared argeing inflaion in 8. Excep inflaion developmens, he bank reacion funcion includes also oupu gap. Given he openness of Czech economy, he CNB may also reac o exchange rae developmens in case hey seriously jeopardize he inflaion prospecs. See Kolan and Navrail (2003) on inflaion argeing design in he Czech Republic. 3

5 As menioned above, we define he inflaion argeing consisency variable ( IT ) using he difference beween ne inflaion and inflaion arge and oupu gap esimae as follows: IT * * ( π π ) ( y y ) = α β (2) where π is he yearly ne inflaion, * π is he inflaion arge, y is acual GDP growh and * y is an esimae of poenial growh. α and β is deermined by principal componens. The raionale for using he curren values of inflaion and oupu gap insead of is forecass is uncerainy in fuure developmens in an economy undergoing massive srucural changes during he sample period. 4 Generally, we link FX inervenion aciviy o is lagged value (o address he persisence of inervenions), exchange rae developmens and he IT consisency. In line wih Io and Yabu (2006), we firs esimae he following probi model: MA ( INT = ) = Φ( + α ( s s ) + α ( s s ) + α ( s s ) α () P INT denoes a binary variable aking value of one, when he CNB inervenes in day ; zero oherwise. s is he CZK/EUR exchange rae. MA s represens he moving average of s using 36 days window. As a resul, α, α 2 and α 3 capure he shor-erm, medium-erm and long-erm movemens of exchange rae, respecively. The sign of 4 IT is inerpolaed linearly o daily values. Inerpolaion using differen procedures had infiniesimal effec on he resuls in he secion 4. Alernaively, we esimae shor-erm exchange rae volailiy using a GARCH model of and ake he esimaes of medium-erm exchange rae misalignmen from Babeskii and Eger (200) and Crespo-Cuaresma, Fidrmuc 4

6 α 2 and α 3 is expeced o be negaive, i.e. inervenion aciviy is inversely relaed o he deviaion of he exchange rae from is las monh value or is moving average. The sign of α is unconsrained (see Sarno and Taylor, 200), hough we expec i o be posiive, as he CNB sells he domesic currency, when he currency appreciaes. I is noeworhy ha he CNB conduced inervenion almos only agains he appreciaion of domesic currency (CNB inervened in 8 days selling he CZK, only 4 days buying he CZK ou of 242 sample days) 6. To invesigae if he inflaion argeing regime served as imporan consrain of CNB s FX inervenion aciviy, we nex esimae regression of he form: MA ( INT ) = Φ( β + β ( s s ) + β ( s s ) + β ( s s ) + β IT ) P = (2) Given our definiion of IT, we expec β 0 ; i.e. CNB is less likely o inervene 4 < agains he appreciaion of currency, when he difference beween he inflaion and is arge rises (analogously for oupu gap). To address he sensiiviy of resuls, we proceed in wo seps. Firs, following Frenkel e al. (2004), we define INT as he number of consecuive inervenion days. Using his alernaive indicaor, we esimae he idenical se of FX inervenion deerminans as in () and (2) by he negaive binomial model. Second, we esimae he following equaion by OLS: and MacDonald (200). The resuls, using hese alernaive specificaions, are similar o hose presened in his paper and are available on reques. 6 Given he fac ha he inervenion has been primarily oriened agains appreciaion of CZK, we do no op for he ordered probi model.

7 INT MA ( s s 2 ) + γ 2 ( s s 2 ) + γ 2 ( s s ) + γ 4 IT + γ INT ε γ 0 + γ + (3) = INT is defined as in he equaion () and alernaively using he acual daily FX inervenion volume. 4. Resuls The monhly inervenion volumes and exchange rae developmens are presened in Figure. As apparen from Figure, here has been hree major inervenion episodes associaed primarily wih he sharp CZK/EUR appreciaion (he firs half of 8, end of, and mid 2002). Figure CZK/EUR exchange rae and inervenion volumes CZK/EUR mil. EUR /8 / /00 /0 /02 Time Source: Czech Naional Bank Noe: The lef scale measures he CZK/EUR exchange rae; he righ scale measures he monhly FX inervenion volume in he millions of euro. Table repors our resuls on he deerminans of FX inervenion aciviy. Typically, he CNB does no reac o day-o-day exchange rae volailiy and raher has been 6

8 concerned wih he medium o long-erm exchange rae developmens. This is in line wih he official bank saemens expressing ha inervenion have been conduced only when he CZK/EUR appreciaed excessively (see Holub, 2004). The resuls also sugges ha inflaion argeing regime is imporan consrain for FX inervenion aciviy. Concreely, he greaer is inflaion and oupu in comparison o heir argeed values, he less likely he CNB will inervene agains domesic currency appreciaion. 7 In addiion, here is also evidence ha FX inervenion aciviy is persisen, o a cerain exen (hough inervenion volumes are much less persisen). Table Esimaion Resuls Neg. Binomial Probi OLS - indicaor OLS - volume s s * (0.62) (0.) (0.44) (0.43) (0.06) (0.06) (0.) (0.) s s *** -0.8*** -0.8*** -0.4*** -0.0*** -0.24** -0.20*** -0.8*** MA s s (0.2) (0.8) (0.0) (0.0) (0.0) (0.0) (0.06) (0.06) -0.7*** -0.30*** -0.08* -0.3*** (0.07) (0.0) (0.04) (0.04) (0.0) (0.0) (0.02) (0.02) IT -0.6*** -0.36*** -0.0*** -0.03*** -0.07* -0.07* (0.4) (0.06) (0.0) (0.0) (0.04) (0.04) INT 0.2*** 0.08** (0.0) (0.03) No. of obs R-sqr Noe: McFadden s R-squared for he negaive binomial and probi model. Inervenion volume is in he billions of CZK. Robus sandard errors in parenhesis. ***, **, and * - denoes significance a percen, percen, and 0 percen, respecively. 7 The resuls using a monhly lag of IT o address he poenial endogeneiy are similar and available on reques. 7

9 . Conclusions In his paper, we sudy he deerminans of FX inervenion aciviy of he Czech Naional Bank. The novely of our approach is he examinaion of he role of inflaion argeing regime on his aciviy. Subjec o various sensiiviy ess, we find ha inflaion argeing regime is imporan consrain for he cenral bank inervenions. Besides, cenral bank reacs o he medium-erm exchange rae misalignmen. We also find ha inervenion aciviy is persisen, o a cerain exen. References Almekinders, G.J., Eijffinger, S.C.W., 4. Daily Bundesbank and Federal Reserve Inervenions: Are They a Reacion o Changes in he Level and Volailiy of he DM/$ Rae?, Empirical Economics, : -30. Almekinders, G.J., Eijffinger, S.C.W., 6. A Fricion Model of Daily Bundesbank and Federal Reserve Inervenion, Journal of Banking and Finance, 20: Babeski, I., Eger, B., 200. Equilibrium Exchange Raes in he Czech Republic: How Good Is Czech BEER?, Finance a uver - Czech Journal of Economics and Finance, : Baillie, R.T., W.P. Oserberg, 7. Why Do Cenral Banks Inervene?, Journal of Inernaional Money and Finance, 6: 0-. Crespo-Cuaresma, J., Fidrmuc, Jarko, MacDonald, R., 200. The Moneary Approach o Exchange Raes in he CEECs, Economics of Transiion, 2: Holub, T., Foreign Exchange Inervenions under Inflaion Targeing: The Czech Experience, Czech Naional Bank Research and Policy Noe, No./

10 Frenkel, M., Pierdzioch, Ch., Sadmann, G., Modelling he Inensiy of Foreign Exchange Inervenion Aciviy: an Exension, Economics Leers, 8: Io, T., Yabu, T., Wha Promoes Japan o Inervene in he Forex Marke? A New Approach o a Reacion Funcion, Journal of Inernaional Money and Finance, forhcoming. Kim, S., Sheen, J, The Deerminans of Foreign Exchange Inervenion by Cenral Banks: Evidence from Ausralia, Journal of Inernaional Money and Finance, 2: Kolan, V., Navrail, D., Inflaion Targeing as a Sabilizaion Rule: Is Design and Performance in he Czech Republic. Finance a uver - Czech Journal of Economics and Finance, 3: Neely, C.J., The Temporal Paern of Trading Rule Reurns and Cenral Bank Inervenion: Inervenion Does No Generae Technical Trading Rule Profis, Journal of Inernaional Economics, 8: Sarno, L., Taylor, M.P., 200. Official Inervenion in he Foreign Exchange Marke: Is I Effecive and, If So, How Does I Work?, Journal of Economic Lieraure, 3: