Identifying the Effects of Monetary Policy Shock on Output and Prices in Thailand

Transcription

1 MPRA Munich Personal RePEc Archive Idenifying he Effecs of Moneary Policy Shock on Oupu and Prices in Thailand Komain Jiranyakul Naional Insiue of Developmen Adminisraion December 2016 Online a hps://mpra.ub.uni-muenchen.de/75708/ MPRA Paper No , posed 21 December :46 UTC

2 Idenifying he Effecs of Moneary Policy Shock on Oupu and Prices in Thailand Komain Jiranyakul School of Developmen Economics Naional Insiue of Developmen Adminisraion Bangkok, Thailand Absrac This paper aemps o idenify he effecs of moneary policy shock on oupu and price level in Thailand during 2005Q1 and 2016Q2. Recenly available policy rae is used as a moneary policy variable. The srucural VAR mehodology is employed o idenify he moneary policy shock. To enhance he precision of he model specificaion, he shor-run resricions are imposed on he specified srucural model of coinegraed variables o allow he levels of variables o inerac simulaneously wih each oher. The resuls from he analysis of he srucural model reveal ha a shock o moneary policy drives cycles for boh real GDP and he inflaion rae. Keywords: Moneary policy shock, srucural VAR, impulse response JEL Classificaion: C32, E52 1. Inroducion Moneary policy can affec boh real aciviy and price level. The efficacy of moneary policy migh be enhanced by an implemenaion of inflaion argeing. Clarida e al. (1998) give evidence indicaing ha some form of inflaion argeing can be a nominal anchor for moneary policy raher han fixing exchange raes. Recenly, researchers employ srucural vecor auoregressive (SVAR) models o esimae dynamic responses of macroeconomic variables due o moneary policy shocks. Canova and de Nicolo (2002) find ha moneary shocks significanly drive oupu and inflaion cycles in all G-7 counries. In addiion, moneary shocks are he dominan sources of macroeconomic flucuaions in hree of he seven counries. Mounford (2005) finds conradicory evidence indicaing ha moneary policy shocks play only a small role in he oal variaions of macroeconomic variables. Only supply shocks impose permanen effec on oupu, which is consisen wih previous finding by Blanchard and Quah (1989). Using he US daa, Uhlig (2005) finds ha conracionary policy shocks have no clear effec on oupu. Furhermore, prices fall gradually in response o a conracionary moneary shock. For some developing economies, Bhuiyan (2012) finds ha moneary policy shocks are no he dominan source of flucuaions in indusrial producion in Banglades. Furhermore, indusrial producion responds o moneary shocks wih a lag of over half a year while inflaion responds wih a lag of more han one year. The lagged responses of oupu and price level are consisen wih he noion proposed by Obsfeld and Rogoff (1995). Using monhly daa for he period from 1978 o 2005 o examine he effecs of ineres rae, money growh and he movemens in nominal exchange rae on growh and inflaion in Sri Lanka, Amrasekar (2008) 1

3 finds ha inflaion does no decline following a conracionary moneary policy shock. Furhermore, his shock also leads o exchange rae appreciaion, which always leads o a rise in growh. Using annual daa for a sample of 105 developing counries o idenify he impacs of moneary policy shocks on oupu growh and inflaion, Kandil (2014) finds ha moneary policy shocks cause oupu growh and inflaion o increase. This sudy makes use of policy rae and real money supply o reflec he exercises of moneary policy ools. 1 The specified SVAR model is used o idenify he impacs of moneary policy shocks on real GDP and price level. Moneary policy shocks from he esimaed SVAR model of coinegraed variables in Secion 2 are shown in Figure Moneary Policy Shock Fig. 1 Moneary Policy Shocks from he Esimaed SVAR model. The plo of moneary policy shock wih he policy rae as he moneary policy variable in Fig. 1 can be defined as he unanicipaed moneary policy. The sizes of boh posiive and negaive shocks are quie small wih he highes absolue size of shock occurred in he hird quarer of The plo of moneary policy shock using real money supply as he policy variable is similar. However, real money supply is no coinegraed wih real GDP and price level. The remainder of he paper is organized as follows: Secion 2 describes he daa and he SVAR model used in he analysis, Secion 3 presens empirical resuls, and he las secion draws conclusion. 2. Daa and Mehodology 2.1 Daa For empirical invesigaion, his paper rerieves he daase from various sources. The policy rae as a measure of moneary policy is obained from he Bank of Thailand websie. Real GDP series is obained from he Office of Naional Economic and Social Developmen Board while he consumer price index series is obained from he websie of Minisry of Commerce. The period of invesigaion is during 2005Q1 and 2016Q2, which is period of implemening inflaion 1 According o Inernaional Moneary Fund, he policy rae is he rae used by he cenral bank o implemen or signal is moneary policy sance. In Thailand, he decisions on he policy rae are aken by he Bank of Thailand moneary policy commiee. The commiee will lower he policy rae when he growh rae of he counry is low and inflaion ends o fall. The commiee will leave he policy rae unchanged when i considers ha he curren moneary policy is appropriae o suppor he economic recovery and does no pose risks o financial sabiliy. 2

4 argeing. The number of observaion is The ADF es wih consan only is used o es for uni roo in all series, which are expressed in logarihmic series. The resuls of uni roo ess are repored in Table 1. Table 1 Resuls of Uni Roo Tess. Variable ADF saisic a. Level y (real GDP) [6] p (consumer price index) [2] r (policy rae) [1] b. Firs Difference y ***[5] p ***[1] r ***[0] Noe: The number in bracke is he opimal lag lengh deermined by Akaike Informaion Crierion (AIC). *** indicaes significance a he 1% level. The resuls in Table 1 show ha all variables are inegraed of order 1 or hey are I(1) series because hey conain uni roo in level, bu conain no uni roo in firs differences Srucural VAR Model The srucural vecor auoregressive (SVAR) model o coinegraion is used o examine he impac of moneary policy shock o oupu and price level. This model is esimaed by he approach proposed by Blanchard and Quah (1989). However, Sims (1980) and Sims e al. (1990) sugges ha all variables in level should ener ino he VAR model in order o preserve he degree of freedom when he sample size is relaively small. 3 Furhermore, he variables in heir firs differences can hrow away informaion from he daa and hus make he inerrelaionships among variables inaccurae. The srucural form VAR model of order p can be expressed as: AX ( + Be (1) = C L) X 1 where X is he 3 dimensional vecor of endogenous variables, namely he log of real GDP, he log of CPI and he log of policy rae a ime, e is he 3 dimensional vecor of srucural innovaions, which is idenically and independenly disribued wih mean of zero and posiive variance. The marix A is he 3 x 3 marix of srucural coefficiens, and C(L) is he polynomial operaor L of order p. An infinie se of differen values of parameers of marices A and C(L) for a given daase canno be obained wihou imposing idenifying resricions because differen srucural forms are able o give he same reduced form VAR model. From Eq. (1), he reduced form model can be expressed as: = D L) X 1 X ( + Eu (2) 2 The policy rae series is available from 2005Q1 o 2016Q2, and hus he number of observaions is limied by his series. 3 For a large srucural model, he Bayesian analysis should be used (see for example, Kociecki e al., 2012). 3

5 1 1 where D( L) = A C( L), and Eu = A Be. The error erm u in Eq. (2) is idenically and independenly disribued wih zero mean and posiive variance, which is he innovaions or shocks ha have no direc economic inerpreaion. This reduced form can be esimaed by he ordinary leas squares (OLS) mehod. According o Enders (2004), an esimaor from OLS mehod is asympoically unbiased and efficien. By imposing resricions on he shor-run behavior of he VAR sysem in Eq. (2), he random sochasic residual Eu can be esimaed from he residual e of he esimaed unresriced VAR model by he following expression: 4 Eu = e (3) Since 1 E = A B, reformulaing Eq. (3) will give 1 ' ' ' 1 A Buu B A = e e. Since u u ' I, hus ' = 1 ' ' 1 A BB A = ' e e (4) Le k be he number of variables in he sysem. The symmery propery of Eq. (4) requires imposing k(k+1)/2 resricions on he 2k 2 unknown elemens in marices A and B. Therefore, addiional k(3k-1)/2 resricions mus be imposed. The resricion scheme is expressed as: Ae = Bu (5) The shor-run resricions are imposed in erms of he vecor of residuals, e, which is obained from he esimae of an unresriced VAR model. Also, he vecor of he fundamenal random error, u, is obained from he srucural sysem. The resuls from he esimaed SVAR model can be used o analyze he impulse responses. 3. Empirical Resuls Since all variables are I(1) series as shown by he resuls of uni roo es repored in Table 1, Johansen and Juselius (1990) and Johansen (1991) coninegraion ess wihou deerminisic rend are performed wih he lag of 4 deermined by AIC and he resuls are shown in Table 2. Table 2 Resuls of Coinegraion Tess. A: Trace Tes Coin. Rank Trace saisic 5% Criical value p-value None* Almos Almos B: Max Eigenvalue Tes Coin. Rank Max-Eigen saisic 5% Criical value p-value None* Almos Almos Noe: * denoes rejecion of he hull hypohesis of no coinegraion a he 5% level, p-values are from MacKinnon e al. (1999). 4 Sims (1986) and Leeper e al. (1996) also use shor-run resricions o idenify shocks. 4

6 The resuls in Table 2 show ha here is one coinegraing relaionship among he log of real GDP, he log of price level and he log of policy rae. Therefore, i should be more meaningful for he esimaed SVAR model wih coinegraing relaion of he variables. 5 In performing he esimaion of he SVAR model, all variables in heir levels are used. The resuls of he analysis of impulse responses are shown in Figure 1. Response o Srucural One S.D. Innovaions ± 2 S.E. Response of GDP o Moneary Policy Shock Response of CPI o Moneary Policy Shock Fig. 2 Impulse Responses of Oupu and Prices due o Moneary Policy Shock. In Fig. 2, he solid lines are poin esimaes of he impulse responses of oupu and prices due o a one-sandard-deviaion shock o he policy rae. The dashed lines conain he 95% error bands. The response horizon is in quarers. The response of oupu due o moneary policy 5 Coinegraion among real GDP, prices and real money supply (M2) is also esed. The resuls of he race es shows one coinegraing relaion, bu he maximum eigenvalue es indicaes no coinegraion. Therefore, i canno be concluded ha coinegraion among hese hree variables exiss. However, he resuls of impulse response analysis are somewha similar. 5

7 shock shows ha oupu increases wih a lag of abou a half quarer. The rising of oupu does no las long enough because he oupu sars falling in jus one quarer and becoming negaive. The oupu ends o flucuae during he horizon of 10 quarers. The response of price level due o moneary policy shock shows ha price level also increases wih a log of abou a half quarer. The price level sars o decline wihin hree quarers and he paern of flucuaions is evidenced. The findings are in line wih he findings by Canova and de Nicolo (2002) in ha moneary policy shock drives oupu and inflaion cycles. However, he findings is conradicory o he findings by Kandil (2004) because moneary shock does no always cause oupu and inflaion o increase. I should be noed ha he implemenaion of inflaion argeing by he Thai moneary auhoriy migh be a nominal anchor for moneary policy as menion by Clarida e al. (1998). Oherwise, he oupu and inflaion cycles migh be more urbulen. 4. Conclusion This paper employs a srucural VAR model o idenify he impacs of moneary policy shock on oupu and price level in Thailand during he period ha inflaion argeing has been implemening. The policy rae and real money supply (M2) are used as moneary policy variables. Since he main purpose of his sudy is o idenify srucural shocks of coinegraed variables, he sysem analysis should use he policy rae as he moneary policy variable. The impulse responses generaed in his paper indicae ha moneary policy shock drives oupu and inflaion cycles. Wihou inflaion argeing, he variaions of oupu and prices migh be larger. Even hough moneary policy ools can have sabilizaion effecs on he economy, here are oher imporan facors ha policymakers should ake ino accoun for economic sabiliy. The exchange rae and oil prices are no included in his sudy due o a relaively small sample size. Therefore his sudy has some limiaions. The analysis of he srucural model wih a larger number of variables is lef for fuure research in he case of his small open economy. References Bhuiyan, R., The Effecs of Moneary Policy Shocks in Bangladesh: A Bayesian Srucural VAR Approach, Inernaional Economic Journal, 26(2), Blachard, O. J. and Quah, D. T., The Dynamic Effecs of Aggregae Demand and Supply Disurbances, American Economic Review, 79(4), Canova, F. and de Nicolo, G., Moneary Disurbances Maer for Business Flucuaions in he G-7, Journal of Moneary Economics, 49(6), Clarida, R. H., Gali, J. and Gerler, M. L., Moneary Policy Rules in Pracice: Some Inernaional Evidence, European Economic Review, 42(6), Enders, W., Applied Economeric Time Series, New York, John Wiley & Sons, Inc. Johansen, S., Esimaion and Hypohesis Tesing for Coinegraion Vecors in Gaussian Vecor Auoregressive Models, Economerica, 59(6),

8 Johansen, S. and Juselius, K., Maximum Likelihood Esimaion and Inference on Coinegraion wih Applicaions o he Demand for Money, Oxford Bullein of Economics and Saisics, 52(2), Kandil, M., On he Effec of Moneary Policy Shocks in Developing Counries, Borsa Isanbul Review, 14(2), Kociecki, A., Rubaszek, M. and Ca Zorzi, M., Bayesian Analysis of Recursive SVAR Models wih Overidenifying Resricions, Working Paper No. 1492, European Cenral Bank, Germany. Leeper, E. M., Sims, C. A., Zha, T., Hall, R. E. and Bernanke, S., Wha does Moneary Policy do?, Brooking Papers on Economic Aciviy, 2, MacKinnon, J. G., Haug, A. A. and Michelis, L., Numerical Disribuion Funcions of Likelihood Raio Tess for Coinegraion, Journal of Applied Economerics, 14(5), Mounford, A., Leaning o he Wind: A Srucural VAR Invesigaion of UK Moneary Policy, Oxford Bullein of Economics and Saisics, 67(5), Obsfeld, M. and Rogoff, K., Exchange Rae Dynamics Redux, Journal of Poliical Economies, 103(3), Sims, C. A., Macroeconomics and Realiy, Economerica, 48(1), Sims, C. A., Are Forecasing Models usable for Policy Analysis, Federal Reserve Bank of Minneapolis Quarerly, 10(1), Sims, C. A., Sock, J. H. and Wason, M. W., Inference in Linear Time Series Models wih Some Uni Roos, Economerica, 58(1), Uhlig, H., Wha are he Effecs of Moneary Policy on Oupu? Resuls from an agnosic Idenificaion Procedure, Journal of Moneary Economics, 52(2),