the determinants of deposit euroization in european post-transition countries 29/06/2011 dubrovnik,...
TRANSCRIPT
29/06/2011Dubrovnik, Croatia
Marina Tkalec17th DEC - YES
The determinants of deposit euroization in European post-transition countries
The determinants of deposit euroization in European post-transition countries:
evidence from threshold VAR
Marina Tkalec
Institute of Economics, Zagreb
29/06/2011Dubrovnik, Croatia
Marina Tkalec17th DEC - YES
The determinants of deposit euroization in European post-transition countries
Content
• Financial euroization
• Contribution
• Data and methodology
• Results
• Policy recommendations
29/06/2011Dubrovnik, Croatia
Marina Tkalec17th DEC - YES
The determinants of deposit euroization in European post-transition countries
Financial euroization Contribution Data and methodology Results
• unofficial euroization is a result of voluntarily using foreign currency for different money functions (Feige and Dean, 2002)
• DE is a result of households, government and enterprises saving in foreign currency
• CE is a result of banks’ behaviour of granting loans in foreign currency or linked to foreign currency
UNOFFICIALEUROIZATION
CURRENCYSUBSTITUTION
ASSETSUBSTITUTION
FINANCIALEUROIZATION
(FE)
DEPOSITEUROIZATION
(DE)
CREDITEUROIZATION
(CE)
29/06/2011Dubrovnik, Croatia
Marina Tkalec17th DEC - YES
The determinants of deposit euroization in European post-transition countries
Financial euroization Contribution Data and methodology Results
• came with high inflation rates in 1980s’ and persisted with the exchange rate as the nominal anchor (Mishkin, 2000; Frankel, 2010)
• exchange rate based monetary regimes continued to persist (currency boards, pegs, fixed, managed or even dirty floating exchange rate regimes)
• “fear of floating” (Calvo and Reinhart, 2002) central banks’ reluctance to allow the exchange rate to adjust significantly and rapidly resulting in episodes of central bank interventions aimed at avoiding major devaluation shifts
• high levels of FE limit the choices for monetary policy makers large exchange rate depreciations increase the cost of servicing foreign currency denominated debt (Reinhart, Rogoff and Savastano, 2003)
• Chang and Velasco (2002) find that detaining depreciation eventually pushes output down
• Cabral (2010) warns of larger employment losses under “fear of floating”
• Tsangarides (2010) reports that pegs have been recovering much slower than floaters in the latest 2010-2011 recovery phase
• FE is one of the biggest weaknesses in European post-transition economies
29/06/2011Dubrovnik, Croatia
Marina Tkalec17th DEC - YES
The determinants of deposit euroization in European post-transition countries
• Theoretical determinants of FE (Levy-Yeyati, 2003.):
T1 portfolio view FE is an outcome of minimum variance portfolio choices, taking returns on various curencies into consideration
T2 market failure view FE is a result of moral hazard induced by aysmmetric information
T3 institutional view FE is a consequence of domestic market and legal framework imperfections; weak institutional framework and
low level of confidence in economic policy encourage policy makers to build confidence through exchange rate anchoring
Financial euroization Contribution Data and methodology Results
29/06/2011Dubrovnik, Croatia
Marina Tkalec17th DEC - YES
The determinants of deposit euroization in European post-transition countries
• Research on FE determinants:
PANEL DATA ANALYSIS• DE: real exchange rate (+), exchange rate volatility (-) (Kokeyne, Ley and Veyrune, 2010)• DE: interest rate differential (-), access to foreign funds (-) (Basso, Calvo-Gonzales and
Jurgilas, 2011)• DE: interest rate differential (+), exchange rate volatility (-) (Luca and Petrova, 2008)• FE: large depreciations have a negative affect on the pass-through coefficient with the
impact being higher the higher the level of euroization (Carranza, Cayo and Galdón-Sanchez, 2003)
• FE: increased access to global capital markets (Reinhart, Rogoff and Savastano, 2003), closeness to the European Union (ECB, 2010; Neanidis, 2010), country size (Rosenberg and Tirpák, 2008)
TIME SERIES ANALYSIS• FE: role of underdeveloped domestic financial markets (Feige, 2002; Levy Yeyati, 2003)• FE in Croatia: massive arbitrage opportunities Šošić (2007)
MICRO DATA ANALYSIS• FE: remittances and income from tourism and underdevelopment of domestic financial
markets (Stix, 2010)
Financial euroization Contribution Data and methodology Results
29/06/2011Dubrovnik, Croatia
Marina Tkalec17th DEC - YES
The determinants of deposit euroization in European post-transition countries
• FE decreases very slowly in periods of macroeconomic stability but increases swiftly in periods of economic uncertainty
• exchange rate depreciations seem to push FE strongly and quickly while the opposite exchange rate changes have a much more moderate impact regime/threshold dynamics transaction costs
NONLINEAR:• DE: positive short-run effects of depreciations decrease with the level of
euroization, interest rate differentials (-), they use an index of asymmetry of exchange rate movements (Neanidis and Savva, 2009)
• FE in Croatia: nominal exchange rate (-), they use threshold cointegration (Ivanov, Tkalec and Vizek, 2011); no possibility of diverse DE responses to exchange rate appreciations/depreciations
Financial euroization Contribution Data and methodology Results
29/06/2011Dubrovnik, Croatia
Marina Tkalec17th DEC - YES
The determinants of deposit euroization in European post-transition countries
DE in European post-transitionBelarus pegged
within horizontal
bands
57%
Bulgaria currency board
55%
Croatia stabilized arrangement
80%
Czech
Republic
free floating 11%
47%
54%
61%
68%
75%
Jan-04 Jan-10
50%
53%
56%
59%
Jan-03 Jan-09
64%
71%
78%
85%
Jul-95 Jul-01 Jul-07
8%
10%
12%
14%
16%
Jan-99 Jan-05
29/06/2011Dubrovnik, Croatia
Marina Tkalec17th DEC - YES
The determinants of deposit euroization in European post-transition countries
DE in European post-transitionHungary managed
floating 22%
Latvia pegged to euro
78%
Lithuania currency board
31%
Macedonia stabilized arrangement
51%
15%
18%
21%
24%
27%
Jan-99 Jan-05
70%
74%
78%
82%
Jan-02 Jan-08
20%
27%
34%
41%
Jan-99 Jan-05
43%
47%
51%
55%
59%
Jan-05
29/06/2011Dubrovnik, Croatia
Marina Tkalec17th DEC - YES
The determinants of deposit euroization in European post-transition countries
DE in European post-transitionPoland free floating 20%
Romania managed floating
37%
Serbia managed floating
67%
Turkey free floating 40%
17%
20%
23%
Jan-99 Jan-05
32%
35%
38%
41%
May-05
60%
64%
68%
72%
76%
Jan-04 Jan-10
28%
36%
44%
52%
Jan-99 Jan-05
29/06/2011Dubrovnik, Croatia
Marina Tkalec17th DEC - YES
The determinants of deposit euroization in European post-transition countries
• we investigate monetary determinants of deposit euroization in European post-transition economies
• DE determinants: exchange rates and differences between domestic and euro interest rates
• linear (cointegration) and threshold (TVAR) models (Koop, Pesaran and Potter, 1996; Balke, 2000)
• we test for the presence of threshold effects with respect to the level of DE
Q1 What kind of threshold effects characterize an economy with a high level of DE?
Q2 And if existing, how do these nonlinearities differ with respect to the prevailing exchange rate regime and/or the DE level?
Financial euroization Contribution Data and methodology Results
29/06/2011Dubrovnik, Croatia
Marina Tkalec17th DEC - YES
The determinants of deposit euroization in European post-transition countries
Financial euroization Contribution Data and methodology Results
C1 New insights into the origins, characteristics and consequences of DE in European post-
transition economies since we model monetary determinants of DE
C2 Scarce existing research on FE that tests for nonlinear or threshold effects
C3 We test whether the determinants of DE behave in a nonlinear fashion
29/06/2011Dubrovnik, Croatia
Marina Tkalec17th DEC - YES
The determinants of deposit euroization in European post-transition countries
Financial euroization Contribution Data and methodology Results
DATA:• deposit euroization (DE), nominal exchange rate (NER)/real
exchange rate (RER) and interest rate differential (IRD)• monthly observations, seasonally adjusted (X12ARIMA), DE and
NER/REER in logarithms• stationary in first differences (ADF)
LINEAR METHOD:• Johansen cointegration
NONLINEAR METHOD:• Threshold Vector Autoregression (TVAR)• Generalized impulse response functions
29/06/2011Dubrovnik, Croatia
Marina Tkalec17th DEC - YES
The determinants of deposit euroization in European post-transition countries
Financial euroization Contribution Data and methodology Results
• transition variable separates the baseline VAR into different regimes (Hansen 1996, 1997; Tsay 1998)
• VAR model adjusted for the threshold specification:
• gamma - coefficient matrices
• - error matrix
• - threshold variable with d being a possible time lag
1 2Γ Γ *t t t t d ty X X I z z u
1(1, ,..., )'t t t jX y y
tu
t dz
1 if *
0 if *
t d
t d
z
z
zI
z
29/06/2011Dubrovnik, Croatia
Marina Tkalec17th DEC - YES
The determinants of deposit euroization in European post-transition countries
Financial euroization Contribution Data and methodology Results
• Hansen linearity test (Hansen, 1996, 1997)
• Least Squares (LS) estimation:
• since the threshold value is not identified under the null of linearity, distribution is not standard (Hansen, 1996) approximation of the asymptotic distribution using a bootstrap procedure
1
1 1
ˆ( ) ( ( ) ( )') ( ( ))T T
t t t tt t
δ z X z X z X z y
2 2
2
sup ( )
ˆ ( )ˆ ( )
T Tz Z
T TT
T
F F z
σ σ zF T
σ z
ˆˆ ( ) ' ( )t t tu y X z δ z 2 2
1ˆ ˆ1 T
T ttσ ut
29/06/2011Dubrovnik, Croatia
Marina Tkalec17th DEC - YES
The determinants of deposit euroization in European post-transition countries
Financial euroization Contribution Data and methodology Results
• the nonlinear model requires impulse response functions that account for nonlinearity of the system:
1. history dependent (Gallant, Rossi and Tauchen, 1993; Koop 1996; Koop, Pesaran and Potter, 1996)
2. asymmetric (i.e. negative shocks are not exactly the opposite of positive shocks)
3. shocks not proportional to their size
• GIRF is the difference between two conditional expectations with a single exogenous shock:
• m - forecasting horizon• - history at time t -1
t t+1 t+m 1
t t+1 t+m 1
| , 0,..., 0,Ω
| 0, 0,..., 0,Ω
t m t
t m t
GIRF E X ε ε ε
E X ε ε ε
1Ωt
29/06/2011Dubrovnik, Croatia
Marina Tkalec17th DEC - YES
The determinants of deposit euroization in European post-transition countries
Financial euroization Contribution Data and methodology Results
Johansen cointegration
Country VariableCointegration
vector
Cointegratingvector
with restrictionsCountry Variable
Cointegrationvector
Cointegratingvectorwith
restrictions
Bulgaria
DE 1 1
Latvia
DE 1 1
RER 1.335 0.107 RER 0.105 0.136
IRD -0.199 -0.073 IRD -0.001 -0.001
Const. -2.716 0 Const. 0 -0.058
Croatia
DE 1 1
Lithuania
DE 1 1
NER -1.371 -1 RER -3.250 -1
IRD -0.055 -1 IRD -0.086 -0.080
Const. 1.431 3.397
Poland
DE 1
Czech R.
DE 1 1 NER 1.132
NER 0.911 1 IRD -0.001
IRD 0.955 1
Turkey
DE 1 1
Hungary
DE 1 NER -0.454 -1
NER -6.936 IRD -0.014 -0.026
IRD 0.018 Note: all coefficients are in vector notation.
29/06/2011Dubrovnik, Croatia
Marina Tkalec17th DEC - YES
The determinants of deposit euroization in European post-transition countries
Financial euroization Contribution Data and methodology Results
Estimation of TVAR and test of nonlinearity
CountryEstimated
thresholdSup F Bootstrapped p Chi-square p
Corresponding DE (in %)
Belarus -0.287 41.3653 0.174 0.000 -
Bulgaria -0.252 46.8602 0.008*** 0.000 56.1
Croatia -0.125 51.8103 0.007*** 0.000 74.4
Czech R. -1.011 45.5666 0.054 0.000 -
Hungary -0.718 47.8170 0.018** 0.000 18.8
Latvia -0.086 45.3061 0.033** 0.000 81.5
Lithuania -0.426 53.5303 0.002*** 0.000 37.2
Macedonia -0.266 37.2685 0.335 0.000 -
Poland -0.685 40.8365 0.240 0.000 -
Romania -0.433 41.7328 0.034** 0.000 37.0
Serbia -0.171 43.8639 0.040** 0.000 67.7
Turkey -0.383 59.9263 0.000*** 0.000 41.9
Note: *** null hypothesis about linearity rejected on 1 percent level of significance; ** hypothesis about linearity rejected on 5 percent level of significance.
29/06/2011Dubrovnik, Croatia
Marina Tkalec17th DEC - YES
The determinants of deposit euroization in European post-transition countries
Financial euroization Contribution Data and methodology Results
LINEAR FRAMEWORK
Czech Republic Poland
NER - -IRD - +
NONLINEAR FRAMEWORK
Bulgaria Croatia Hungary Latvia Lithuania Romania Serbia Turkey
depreciation
DE+ - + + + + + +
depreciation
IRD+ + + - + + + +
IRD widening
DE- + + / + + + +
Note: circles denote nonlinear behaviour
29/06/2011Dubrovnik, Croatia
Marina Tkalec17th DEC - YES
The determinants of deposit euroization in European post-transition countries
Financial euroization Contribution Data and methodology Results
• nonlinear behaviour depreciations have a stronger effect on DE and on IRD than appreciations
• rise in domestic interest rates relative to euro ones increases DE levels
• Czech Republic and Poland flexible ER regimes lowest FE levels
• Latvia, Lithuania and Bulgaria convergence official euroization
• Croatia, Hungary, Romania, Serbia and Turkey reform of macroeconomic regimes and institutions increase macroeconomic and institutional credibility
29/06/2011Dubrovnik, Croatia
Marina Tkalec17th DEC - YES
The determinants of deposit euroization in European post-transition countries
Thank you for your attention!