Financial stress and its impact on economic activity in Romania

Transcription

1 The Academy of Economic Sudies Docoral School of Finance and Banking Financial sress and is impac on economic aciviy in Romania MSc suden : Toma Alice Teodora Supervisor: Professor Moisă Alăr Buchares

2 Moivaion One of he objecives of consrucing a financial sress index is o help policymakers idenifying sress levels in he financial sysem ha can be a serious concern The imporance of financial sress arises from is poenial adverse effec on he real economy. Even if in some cases he high levels of financial sress are no necessarily followed by economic downurn i is sill possible hrea o he growh of he real economy Since he financial sress is no direcly observable i is necessary o build an index ha reflecs he sress componen of he financial markes variables The European Cenral Bank (ECB) and he Federal Reserve have creaed several indicaors wih he purpose of measuring he curren sae of insabiliy, i.e. he curren level of fricions, sresses and srains in he financial sysem

3 Objecives Consruc a Financial Sress Index for Romania including financial variables for he banking secor, securiies marke, sock marke and foreign exchange marke I follow he approach of Cardarelli, Elekdag and Lall (2011) Esimae a Vecor Auoregressive model o explore he effecs of financial sress on economic aciviy Assess he macroeconomic impac of: - a financial sress shock - a moneary policy shock

4 Lieraure review Only few sudies on he subjec prior o he financial crises; he lieraure was derived from previous sudies on early warning indicaors for banking crises & financial sress Bordo e al. (2001) pioneered he financial sress lieraure financial condiions index o deermine he frequency of financial crises in hisorical daa Iling M. and Ying L. - among he firs researchers who have developed a financial sress indicaor (developed FSI for Canada where he daa covered equiy, bond, foreign exchange and banking secor) Financial sress is defined as he force exered on economic agens by uncerainy and changing expecaions of loss in financial markes and insiuions. Financial sress is a coninuum, measured wih an index called he Financial Sress Index (FSI), where exreme values are called financial crises. Ineresing findings are brough by Cardarelli e. al. - periods of financial disurbance brough by banking disress are more likely o be associaed wih deep downurns han periods of sress mainly relaed o securiies or foreign exchange markes

5 Lieraure review Vecor auoregressive models have become an imporan economeric ool in order o appraise he effecs of moneary and fiscal policy shocks Based on VAR approach, Van Aarle (2003) esimaed he impac of fiscal and moneary policy for he members of Economic and Moneary Union highlighing differen reacions among various counries of he Euro Area The empirical evidence on cenral bank s reacions o financial insabiliy is raher scan. Baxa, Horvah and Vasicek (2010) sudied he reacion of cenral bank inflaion argeing o financial sress normally do no reac o financial sress bu heir behavior changes in imes of large and longer sress Mallick and Sousa (2011) use wo idenificaions in a Bayesian VAR and sign resricion VAR o examine he real effecs of financial sress

6 Mehodology Financial sress index: equal-variance-weighed average of seven variables, grouped ino hree caegories: 1. Banking relaed subindex formed by: - he bea of banking secor = correlaion beween reurn of banking secor sock index and he overall sock marke index - TED spread=difference beween inerbank raes and he yield on reasury bills - he slope of he yield curve = difference beween he shor and long erm yields on governmen issued securiies 2. Securiies marke relaed subindex: - corporae bond spreads = difference beween corporae bond yields and LT governmen bond yields - sock marke reurn - ime-varying sock marke volailiy 3. Foreign exchange relaed subindex: - ime-varying volailiy of real effecive exchange rae

7 A VAR model wih he FSI The baseline reduced VAR model can be wrien: p = c+ AY i 1 i= 1 Y + U Y is he vecor of n endogenous variable given by Y = π [ y i s ]' y is he GDP growh π is he inflaion i is he shor erm ineres rae s is he indicaor for financial marke condiions (FSI) FSI is ordered las which implies ha i reacs conemporaneously o all variable in he sysem The ordering also implies ha he moneary policy shocks do no have an impac, conemporaneously, on oupu or inflaion

8 Daa Monhly daa from 2000 M1 o 2011 M12 (IMF daabase, BVB) Variables in VAR: y p i s annual growh rae of he log of real GDP used: annual growh rae of log of price level used: money marke rae (monhly average of overnigh money) financial sress index compued as below p Y = log( Y ) log( Y 12) = log( P ) log( P 12) Financial sress variables: Bank sress: - normalized bea beween reurn of Transylvania Bank shares and BET-C β = cov( TLV, BET C) VAR( BET C) - covariance is esimaed hrough a Mulivariae Garch model: vech( H q p ' ) = c + A jvech( r jr j ) + B j= 1 j= 1 j vech( H j )

9 - normalized TED spread= ROBOR 3M Treasury bill yield (3M) - invered erm spread=difference beween deposi rae and lending rae Sock marke sress: - Monhly reurn of socks (BET-C index) compued as - Volailiy of monhly reurns for BET-C index derived from GARCH (1,1) Foreign exchange sress: - Volailiy of monhly changes in he REER derived from GARCH (1,1) specificaion = + + S S R 1 1 ln = = + + = p j j j q i i i σ β α ε ω σ

10 Preliminary analysis for he series of reurns of BET_C, TLV, REER specific ess o check he suiabiliy of models as GARCH were applied Descripive saisics BET-C reurn TLV reurn REER reurn Skewness Kurosis Jarque-Bera Probabiliy Tesing for saionariy: BET-C Prob.* TLV Prob.* REER Prob.* ADF es saisic % criical value % criical value % criical value

11 Tesing for serial correlaion: BET-C TLV REER

12 Tesing for heerosckedasiciy: Breusch-Godfrey Serial Correlaion LM Tes: BET-C F-saisic 1.37 Prob Obs*R-squared Prob. Chi-square 0.02 TLV F-saisic 2.10 Prob Obs*R-squared Prob. Chi-square 0.03 REER F-saisic 1.36 Prob Obs*R-squared Prob. Chi-square 0.01 Graph of BET-C, TLV, REER volailiy and covariance beween BET-C and TLV

13 Esimaion of GARCH (1,1) model for BET-C reurn and REER reurn GARCH(1,1) for BET-C reurn: GARCH(1,1) for REER reurn: Dependen Variable: L_BET_C Mehod: ML - ARCH Dae: 05/21/12 Time: 23:03 Sample (adjused): 2000M M12 Included observaions: 143 afer adjusmens Convergence achieved afer 10 ieraions Presample variance: backcas (parameer = 0.7) GARCH = C(2) + C(3)*RESID(-1)^2 + C(4)*GARCH(-1) Dependen Variable: L_REER Mehod: ML - ARCH (Marquard) - Normal disribuion Dae: 05/22/12 Time: 23:23 Sample (adjused): 2000M M12 Included observaions: 143 afer adjusmens Convergence achieved afer 22 ieraions Presample variance: backcas (parameer = 0.7) GARCH = C(2) + C(3)*RESID(-1)^2 + C(4)*GARCH(-1) Variable Coefficien Sd. Error z-saisic Prob. AR(1) Variance Equaion C RESID(-1)^ GARCH(-1) R-squared Mean dependen var Adjused R-squared S.D. dependen var S.E. of regression Akaike info crierion Sum squared resid Schwarz crierion Log likelihood Hannan-Quinn crier Durbin-Wason sa Invered AR Roos.41 Variable Coefficien Sd. Error z-saisic Prob. AR(1) Variance Equaion C 1.44E E RESID(-1)^ GARCH(-1) R-squared Mean dependen var Adjused R-squared S.D. dependen var S.E. of regression Akaike info crierion Sum squared resid Schwarz crierion Log likelihood Hannan-Quinn crier Durbin-Wason sa Invered AR Roos.49

14 Graph of financial sress index for Romania during : High sress periods are defined by periods when he FSI exceeds is mean by more han one sandard deviaion: 2008 M M6 The index capures he high sress episodes seen in he pas

15 Variables in levels are used in he VAR specificaion GDP growh Prob.* Inflaion rae Prob.* Ineres rae Prob.* FSI Prob.* ADF es saisic % criical value % criical value % criical value VAR sabiliy

16 Impulse response afer a shock o he Financial Sress

17 Variance decomposiion afer a financial sress shock

18 Impulse response afer a moneary policy shock

19 Resuls Following a posiive shock in financial sress, real GDP growh falls - afer 8 monhs he effec is he larges reducing real GDP growh by 0.8 p.p. The impac on ineres rae is less persisen and afer an iniial increase i converges back o is primary level Deerioraion of financial sress condiions negaively impacs he inflaion rae which increases hough only afer a few monhs. The effec is persisen A a horizon of 8 quarers shocks in financial sress explains abou: 30 percen of variaion in real GDP growh 5 percen of he variaion in ineres rae only a few percenage poins in inflaion rae

20 Resuls The ineres rae is used as he moneary policy insrumen Afer a conracionary moneary policy, real GDP growh falls even if no immediaely bu afer approximaely 3-4 monhs Inflaion begins o fall in response o a conracionary moneary policy shock The responses o a moneary policy shocks are no significan from saisical poin of view The moneary policy conracion explains only a small fracion of he variaion of he FSI, real GDP growh and inflaion rae (5% or less)

21 Conclusions This paper develops a financial marke sress indicaor for Romania during The aim of he indicaor is o provide a quick, clear and inuiive assessmen of he curren sae of he financial sysem The FSI buil capures he high sress periods seen in he pas in Romania The increase in financial sress has shown o have large effecs in economic aciviy, dampening he real GDP growh No significan impac of he moneary policy changes in he face of high financial sress nor a meaningful conribuion in explaining he flucuaions in he daa Deeced in early sages, he impac of financial sress can be miigaed by adequae fiscal and moneary measures.

22 Reference: Arle, B., Garresen, H. and Gobbin, N. (2003), Moneary and fiscal policy ransmission in he Euroarea:evidence from a srucural VAR analysis, Journal of Economics and Business 55, Afonso, A. and Sousa, R. (2009b), The macroeconomic effecs of fiscal policy. European Cenral Bank, Working Paper 991 Afonso, A., Baxa, J. and Slavik, M. (2011), Fiscal developmens and financial Sress. A Threshold VAR analysis, European Cenral Bank Working Paper 1319 Balakrishnan, R., Danninger, S., Elekdag, S. and Tyell, I. (2009), The ransmission of financial sress from advanced o emerging economies, IMF Working Paper WP/09/133 Baxa, J., Horvah, R. and Vasicek, B. (2011), Time-Varying Moneary-Policy Rules and Financial Sress: Does Moneary Insabiliy Maer for Moneary Policy?, Journal of Financial Sabiliy Bloom, N. (2009), The impac of uncerainy shocks, Economerica, 77 (3), Bollerslev, T. (1986), Generalized Auoregressive Condiional Heeroskedasiciy, Journal of Economerics, 31, Borio, C and Drehmann, M. (2009), Assessing he risk of banking crises revised, BIS Quarerly Review Caldara, D. and Kamps, C. (2008), Wha are he effecs of fiscal shocks? A VAR-based comparaive analysis, European Cenral Bank Working Paper 877 Cardarelli, R., Elekdag, S. and Lall, S. (2009), Financial Sress, Downurns, and Recoveries, IMF Working Papers 09/100 Cardarelli, R., Elekdag, S. and Lall, S. (2011), Financial sress and economic conracions, Journal of Financial Sabiliy,78-97 Deak, S. and Lenarcic, A. (2011), The Macroeconomic Effecs of Fiscal Policy Shocks in Good Times and Bad, Bocconi Deparmen of Economics Galvao, A., B. and Marcellino, M. (2011), An endogenous hreshold VAR model for he moneary ransmission mechanism, Queen Mary Universiy of London, Deparmen of Economics

23 Reference: Gadanecz, B. and Jayaram, K. Measures of financial insabiliy a review, Moneary and Economic Deparmen, BIS Grimaldi, M., B. (2010), Deecing and inerpreing financial sress in he euro area, European Cenral Bank Working Paper 1214 Hakkio, C. and Keeon, W. (2009), Financial Sress: wha is i, how can i be measured and why does i maer?, Federal Reserve Bank of Kansas Ciy Economic Review, Second Quarer, 5-50 Illing, M. and Liu, Y. (2006), Measuring financial sress in a developed counry: An applicaion o Canada, Journal of Financial Sabiliy, 2(3), Li, F. and S-Aman, P. (2008), Financial sress, moneary policy and economic aciviy, Mimeo, Bank of Canada Mallick, S. and Sousa, R., (2011), The real effecs of financial sress in he Euro zone, Universidade do Minho, NIPE WP 12/2011 Masen, B. and Coricelli, F., Masen I, Financial Inegraion and financial developmen in ransiion economies: Wha happens during financial crises? Oe, M., Eiben, R., Bianco, T., Gramlich, D. and Ong, S. (2011), The financial sress index: idenificaion of sysemic risk condiions, Federal Reserve Bank of Cleveland, Working Paper, Rahman, S. and Serleis, A. (2010), The asymmeric effecs of oil price and moneary policy shocks: A nonlinear VAR approach, Energy Economics, 32, Roye, B. (2011), Financial sress and economic aciviy in Germany and he Euro Area, Kiel Insiue for he world Economy, Working Papers Slingenberg, W. and Haan, J. (2011), Forecasing financial sress, De Nederlandsche Bank, Working Paper no. 292 Sorge, M. (2004), Sress-esing financial sysems: an overview of curren mehodologies, Moneary and Economic Deparmen, BIS Working Papers no 165

24 Thank you!