The Academy of Economic Studies BucharestThe Faculty of Finance, Insurance, Banking and Stock Exchange
DOFIN - Doctoral School of Finance and Banking
MSc Student Dragomir Ioana
Supervisor Professor Moisă Altăr
Monetary policy through the “credit-cost channel”.
A VECM APROACH FOR ROMANIA
Topics
Introduction Literature review The model Data description Methodology Estimation results Conclusions
Introduction
The aim of this paper is to contribute to the analysis of the effects of interest-rate based monetary policy by means of a model that blends the credit and cost channels of monetary policy into a single, integrated "credit-cost channel" (CCC).
The purposes of the model is to demonstrate that firms reliance on bank loans (“credit channel”) could make aggregate supply sensitive to bank interest rates (“cost channel”), which are driven by the policy rate, controlled by the central bank and by a credit risk premium charged by banks on firms.
Literature review
Monetary policy impulses have persistent real effects in the economy: aggregate demand credit channelGetler and Gilchrist(1993);Bernake and Getler (1995);Trautwein (2000); aggregate supply cost channelBarth and Ramey (2001);Christiano and Eichenbaum (1997,2005);
Chowdhury(2006);
Ravenna and Walsh(2003,2006); aggregate demand and supply credit-cost channel Greenwald and Stiglitz (1988,1993);
Fiorentini and Tamborini (2002);
Passamani and Tamborini (2005,2006);
The model The economy: 3 markets: Labor, Credit and Output 3 classes of agents: firms, households
and banks and a central bank; The economy operates sequentially t, t+1, ..., production takes 1 period; Firms
- (t): plan production for sale at t+1;
- (t): face uncertainty about revenue from output sales;
- (t): hire workers in the labor market;
- (t): borrow the wage bill in the credit market. Households
- (t): sell labor and receive their income- is saved for consumption in t+1;
- (t): consumption is brought out of saving carried over from t-1; Banks
- offer deposits services to households at zero interest and standard debt contracts to firms;
- insure against credit risk by borrowing reserves from the central bank at the policy rate;
The model Firms
- output of firm j;
- labour force used by firm j;
- total output in the economy;
- market clearing price level;
),()( 1 jtjt NQTQ 0,0 NNN QQ
1)( ttQ
1tP
1)( jttQ
jtN
jttjte uPP 11 )1(111 jtttjt
e uPPP
jtu1jt
eP - price forecast for firm j;
- forecast error for firm j, i.i.d. random variable,
0),(1)(
jtit
jt
uuCovuE with unit expected value and
zero correlation across firms
The loans demanded by a firm at time t:
against which the firm is committed to paying in t+1:
,if solvency state is declared
,if default state is declared
- the gross nominal interest rate charged by banks;
- the nominal wage
The firm expected one-period profit:
The model Firms
jttjtd NWL
tjtd RL
11 )( jtt tQPtjt
djtt RLtQP 11 )(
)1( tt rR
tW
ttjtjtte
jte RNWtQPZ 111 )(
RLtQP jtd
jtt 11 )(
The first order condition for maximazing profit:
1 jte
tN RwQ
111 1/ te
tjte
jte PP
ttt PWw /
The model Firms
111 1/ te
jte
tjte rRR
- curent real wage
- expected real interest rate
- expected inflation rate
),,,( 1 jte
ttd
jtd rwNN
)),,(()( 11 jte
ttd
jt rwNQtQ
The labour demand function:
The output supply:
0,0,0 d
rd
wd NNN
At period t, each household h choses the sequence:
in order to maximize their utility:
Constraints:
The model Households
121 ,:,,.....)1(,)(: hththhthth NNNtCtCC),(max , hhhtNC NCUU
,)(,)1( 111 ththte
thtt DtCPDtCP
htthtttt NWtCPDD )1(1
1)( httC
1tD
1htePtP
- amount of consumption goods at t+1 for h- price forecast for household h- price of goods at t- deposit due at t
The labour supply function
),,( 1 hte
ts
hts wNN 0,0
sw
s NN
t
t
t
t
RKBR
- loans oferred equals - deposits collected by bank b;
The model Banks
btsL btD
The expected net profit of a bank:
- borrowed reserves from central bank;
- gross official interest rate;
- default probability:)1** )()(1 tjttt uuPuF
tttt
tt
t
tt kr
KtR
1
1log,
1
)1(
tbts
t LBR
1)u*jt – is the critical value of the forecast of the firms regarding the clearingmarket price noise and F is the cumulative function of ujt
)1( tt kK - gross bank interest rate; )1( tt rR - credit risk premium;
,0)1( bts
ttttbts LKBRRL
The ModelMacroeconomic equilibrium
ttt NWL
),(),,( 11 tts
tttd wNrwN
ttt kr
1//)),,((
11
11
ttt
tttttd
PPPDrwNQ
•Labour market
•Credit market
•Output market
If changes in have negative effects on
The ModelShocks from credit variables tt k,
)())1(1()(
))1(1()())1(1(1
)(1
1
wss
Nwd
tt
Nws
wd
sw
sNw
dN
sw
d
t
t
t
NNQNddk
QNNNNQNQNN
dtdQ
dw
wss
N
NNQ
1
11,)(, ttt tQw ttk ,
output - the industrial production index;
inflation rate - the consumer price index;
real wage rate - the total economy gross wage index / CPI;
monetary policy variable - the 3M interbank rate ROBOR;
credit risk premium - the average bank lending rate for the
private sector;
the foreign variable - the interbank rate 3M EURIBOR;
Data descriptionMonthly series covering the period 2000M01 - 2009M03:
stst
Q
t
tw
tk
tk *
All variables, excluding interest rates are log–transformed. All variables are seasonal-adjusted. The base year of indices-2005.The gestation time of output s =12
The transmission mechanism
ND Ns
Wt-1 A_
Wt B
Nt Nt-1
Response to an increase in the bank interest rate:The Labour Market The output-market
AD AS
t A
t+1 B
Q(t)t+1 Q(t-1)t
Data description
.0
.1
.2
.3
.4
.5
.6
.7
.8
2000 2001 2002 2003 2004 2005 2006 2007 2008
average lending rateEURIBOR_3M_SAROBOR_3M_SA
Data description
0.4
0.6
0.8
1.0
1.2
1.4
1.6
2000 2001 2002 2003 2004 2005 2006 2007 2008
real gross wage indexconsumer price indexindustrial production index
Test Augmented-Dickey
Fuller
Phillips- Perron
Series Sig level 1% level 5% level 10% level t-Statistics Prob. t-Statistics Prob.
Ln_ipi_fore Level -2.59 -1.94 -1.61 0.5099 0.8242 0.6225 0.8494
First dif -12.4653 0.0000 -12.3937 0.0000
Ln_w_r_g_i_cpi_sa Level -2.59 -1.94 -1.61 -0.5541 0.4750 -0.7669 0.3819
First dif -14.2194 0.0000 -14.2336 0.0000
Ln_cpi_fore_sa Level -2.59 -1.94 -1.61 -0.2475 0.5948 -3.5104 0.0006
First dif -2.3390 0.0194 -2.8214 0.0051
av_lend_rate_sa Level -4.04 -3.45 -3.15 -2.1972 0.4861 -1.3161 0.8787
First dif -7.8283 0.0000 -8.2725 0.0000
Euribor_3m_sa Level -4.04 -3.45 -3.15 -1.8818 0.6572 -1.4960 0.8253
First dif -4.0080 0.0111 -3.9281 0.0140
Robor_3m_sa Level -4.04 -3.45 -3.15 -3.0139 0.1332 -3.0089 0.1346
First dif -8.4819 0.0000 -8.7278 0.0000
Data descriptionResults of the unit root tests I(1)
twLN_W_R_G_I_CPI_SA LN_IPI_FORE LN_CPI_FORE_SA ROBOR_3M_SA AV_LEND_RATE_SA
LN_W_R_G_I_CPI_SA(-1) 0.5934 0.0586 0.0189 0.0348 0.0320LN_IPI_FORE(-1) 0.2772 0.8721 0.0024 -0.1147 -0.0339LN_CPI_FORE_SA(-1) 0.4214 -0.0614 0.9671 -0.0702 -0.0245ROBOR_3M_SA(-1) 0.0058 0.1112 0.0166 0.8830 0.1533AV_LEND_RATE_SA(-1) 0.4386 -0.2700 -0.0202 -0.0864 0.7312C -0.1226 0.0408 0.0080 0.0203 0.0157EURIBOR_3M_SA 0.5475 0.1751 -0.0123 0.8351 0.2767
R-squared 0.9854 0.9381 0.9998 0.9828 0.9964 Adj. R-squared 0.9846 0.9344 0.9998 0.9818 0.9962 Sum sq. resids 0.0541 0.0480 0.0015 0.0457 0.0056 S.E. equation 0.0229 0.0216 0.0038 0.0211 0.0073 F-statistic 1162.5320 259.9391 75348.1000 979.8470 4750.5470 Log likelihood 262.8832 269.4645 460.7814 272.1552 388.0116 Akaike AIC -4.6524 -4.7721 -8.2506 -4.8210 -6.9275 Schwarz SC -4.4806 -4.6002 -8.0787 -4.6492 -6.7556 Mean dependent -0.0133 0.0929 -0.0491 0.2248 0.2252 S.D. dependent 0.1847 0.0843 0.2442 0.1560 0.1190
1)( ttQ 1t tr tk
MethodologyVAR(p) ESTIMATION
VAR ESTIMATIONLag Length SelectionStability condition check
Lag LogL LR FPE AIC SC HQ0 858.1876 NA 0.0000 -16.4697 -16.2139 -16.36611 1622.0850 1423.9640 2.85E-20* -30.8171* -29.9218* -30.4545*2 1645.8110 41.9246* 0.0000 -30.7925 -29.2577 -30.17083 1665.3820 32.6817 0.0000 -30.6870 -28.5127 -29.80644 1686.4620 33.1547 0.0000 -30.6109 -27.7971 -29.47125 1709.6940 34.2834 0.0000 -30.5766 -27.1233 -29.17796 1723.4540 18.9713 0.0000 -30.3583 -26.2656 -28.70067 1743.2060 25.3126 0.0000 -30.2564 -25.5242 -28.33978 1764.6360 25.3829 0.0000 -30.1871 -24.8153 -28.0114
Root Modulus
0.9832 0.98320.9618 0.9618
0.79 - 0.09i 0.8034 0.79 + 0.092i 0.8034
0.5000 0.5058
p=1
the VAR satisfies the stability condition test
1
*1212
1011
1
10
,]['],,,,,[',',''
t
tttttttttt
tt
n
iitt
x
kzqkwyzyx
txxzy
- elements: adjustment coefficients of variables towards their
long-run relationships
MethodologyStructural cointegration method Johansen&Juselius•Objective: The identification of the long-run structural relationships•Re specification of the model:
- matrices of coefficients;
- error correction mechanism;
- columns: r cointegration vectors long-run relationships
,'
VECM ESTIMATIONJohansen Cointegration TestStability condition check
HypothesizedNo. of CE(s) Statistic 5 % Critical Val 1% Critical Val Statistic 5 % Critical Val 1% Critical ValNone ** 273.23 87.31 96.58 129.16 37.52 42.36At most 1 ** 144.06 62.99 70.05 76.84 31.46 36.65At most 2 ** 67.23 42.44 48.45 48.34 25.54 30.34At most 3 18.89 25.32 30.45 16.64 18.96 23.65At most 4 2.25 12.25 16.26 2.25 12.25 16.26
Trace Test Max-Eigen Test
*(**) denotes rejection of the hypothesis at the 5%(1%) level
VECM: with 5 variables vector y’t = [wt, kt, qt+12, t, t+12], 1 exogenous variable z’t =[k*t], 3 cointegrating relations and 0 lag.
Root Modulus1.00 1.001.00 1.000.95 0.950.69 0.690.33 0.33
the VEC satisfies the stability condition test
Estimation resultsThe unrestricted model
Cointegration vectors2) CointEq1 CointEq2 CointEq3LN_W_R_G_I_CPI_SA(-1) 1 0 0LN_IPI_FORE(-1) 0 1 0LN_CPI_FORE_SA(-1) 0 0 1ROBOR_3M_SA(-1) 1.73*** 2.82*** 0.67***AV_LEND_RATE_SA(-1) -2.52*** -2.09*** -1.11***@TREND(00:01) -0.01*** 0 0.00***C 0.54 -0.21 0.41(* significant at 10%, **significant at 5%, *** significant at 1%)
The coefficients of the inter-bank rate in all of the 3 cointegration equations is positive and significant, underlying the negative correlation between the policy rate and the key variables of the economy.
2) The Beta coefficients are estimated based on the normalization of ’* S11*,where S11 is defined in Johansen 1995
tw1)( ttQ
1t
trtk
The unrestricted model
Error Correction: D(LN_W_R_G_I_CPI_SA) D(LN_IPI_FORE) D(LN_CPI_FORE_SA) D(ROBOR_3M_SA) D(AV_LEND_RATE_SA)CointEq1 -0.556222 0.029921 0.013261 0.03645 0.082512t-statics [-7.15254] [ 0.37560] [ 0.98080] [ 0.47492] [ 3.18154]CointEq2 0.221122 -0.016269 0.002465 -0.079155 0.004033t-statics [ 7.07453] [-0.50812] [ 0.45355] [-2.56601] [ 0.38688]CointEq3 0.288707 -0.013893 -0.034869 -0.025133 0.038311t-statics [ 6.73715] [-0.31649] [-4.68013] [-0.59427] [ 2.68069]C -0.010355 0.00758 0.009788 -0.022671 -0.009163t-statics [-1.31760] [ 0.94158] [ 7.16365] [-2.92306] [-3.49618]EURIBOR_3M_SA 0.504033 -0.188102 -0.037112 0.510108 0.170395t-statics [ 2.23312] [-0.81354] [-0.94572] [ 2.28997] [ 2.26370]
tw 1)( ttQ 1t tr tk
the short dynamics of : wt, t+1, Qt+1 are not explosive. adjusts significantly and rapidly in the direction of all three long-term relations; hardly adjusts to any long-term equilibrium relation; adjusts slowly and significantly in the direction of the 3th coEq.
w
Q
The unrestricted modelThe impulse response functions
-.002
.000
.002
.004
.006
.008
.010
2 4 6 8 10 12 14 16 18 20 22 24
Response of LN_W_R_G_I_CPI_SA to ROBOR_3M_SA
-.00150
-.00125
-.00100
-.00075
-.00050
-.00025
.00000
2 4 6 8 10 12 14 16 18 20 22 24
Response of LN_IPI_FORE to ROBOR_3M_SA
.0000
.0005
.0010
.0015
.0020
.0025
2 4 6 8 10 12 14 16 18 20 22 24
Response of LN_CPI_FORE_SA to ROBOR_3M_SA
.004
.008
.012
.016
.020
.024
2 4 6 8 10 12 14 16 18 20 22 24
Response of ROBOR_3M_SA to ROBOR_3M_SA
Response to Cholesky One S.D. Innovations
.004
.005
.006
.007
.008
.009
.010
2 4 6 8 10 12 14 16 18 20 22 24
Response of AV_LEND_RATE_SA to CholeskyOne S.D. ROBOR_3M_SA Innovation
The unrestricted modelThe impulse response functions
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
2000 2001 2002 2003 2004 2005 2006 2007 2008
COINTEQ01COINTEQ02COINTEQ03
The cointegration graph
Production, wages and inflation small deviations from the long
term level.
Estimation resultsThe unrestricted model
The unrestricted model
R-squared 0.9854 0.9381 0.9998 0.9828 0.9964 Adj. R-squared 0.9846 0.9344 0.9998 0.9818 0.9962 Sum sq. resids 0.0541 0.0480 0.0015 0.0457 0.0056 S.E. equation 0.0229 0.0216 0.0038 0.0211 0.0073 F-statistic 1162.5320 259.9391 75348.1000 979.8470 4750.5470 Log likelihood 262.8832 269.4645 460.7814 272.1552 388.0116 Akaike AIC -4.6524 -4.7721 -8.2506 -4.8210 -6.9275 Schwarz SC -4.4806 -4.6002 -8.0787 -4.6492 -6.7556 Mean dependent -0.0133 0.0929 -0.0491 0.2248 0.2252 S.D. dependent 0.1847 0.0843 0.2442 0.1560 0.1190
Lags LM-Stat Prob1 32.44 0.152 20.85 0.703 28.65 0.284 31.56 0.175 28.10 0.306 18.31 0.837 18.51 0.828 13.98 0.969 22.39 0.61
10 39.13 0.0411 21.18 0.6812 33.36 0.12
Component Jarque-Bera df Prob.1 115.68 2 0.002 34.66 2 0.003 5.34 2 0.074 352.21 2 0.005 23.00 2 0.00
Joint 530.90 10 0.00
Estimation resultsThe restricted model
Cointegrating Eq: CointEq1 CointEq2 CointEq3LN_W_R_G_I_CPI_SA(-1) 1.000 -0.847 0.000LN_IPI_FORE(-1) 0.000 1.000 0.000LN_CPI_FORE_SA(-1) -1.319 0.336 1.000ROBOR_3M_SA(-1) 1.918 2.529 0.043AV_LEND_RATE_SA(-1) -2.280 -1.413 -0.386@TREND(00:01) 0.000 0.003 -0.005C 0.021 -0.515 0.395Error Correction: CointEq1 CointEq2 CointEq3D(LN_W_R_G_I_CPI_SA) -0.362453 0.225052 -0.263843D(LN_IPI_FORE) 0.000 0.000 0.000D(LN_CPI_FORE_SA) 0.000 0.000 -0.034894D(ROBOR_3M_SA) 0.000 -0.070136 0.000D(AV_LEND_RATE_SA) 0.090732 0.000 0.15274
B(1,1)=1 B(2,2)=1 B(3,3)=1 B(3,2)=0 B(1,2)=0 B(3,1)=0 A(2,1)=0 A(3,1)=0 A(4,1)=0 A(3,2)=0 A(2,2)=0 A(5,2)=0 A(2,3)=0 A(4,3)=0Chi-square(6) 1.879083Probability 0.930476
Cointegration Restrictions:
Conclusions
Empirical results show that, by way of the CCC transmission mechanism the inter-bank rate is a co-determinant with negative sign of the long-run stochastic equilibrium paths of the real wage rate, output and inflation. The results for the premium risk variable reject the same hypothesis, due to the lack of a better measure of risk.
