Post on 01-Aug-2020
A Combination of Monetary Base Targeting and Interest Rate Targeting: Case of Thailand
Phai Phongthiengtham
Chulalongkorn University
Presentation Outline
Conclusion and Suggestion
Empirical Analysis
Theoretical Model
Introduction
Purpose of the study
Purpose of the study
To evaluate and compare the interest rate targeting and monetary base targeting policy.
Illustrate the combination of both interest rate and monetary base targeting.
Introduction
Introduction
John Maynard Keynes (1883-1946) favors interest-rate targeting.
Controversy among
Economists
Milton Friedman (1912-2006) argues for a quantitative policy.
Introduction
Interest RateLM
IS
Output
Introduction According to the IS-LM framework,
Interest Rate Targeting: Adjusting money supply such that the LM curve shifts to keep interest rate constant.
Monetary Aggregate Targeting: Keep money supply at a specific value.
Differences arise according to these factorsShocks in goods market (IS) and shocks in money market
(LM)Slope of IS and LM curves
Introduction
LM1
IS1
Interest Rate
IS2
LM2
Output
Negative shocks in goods market causes output to fall…
…For interest rate targeting policy; however, output falls even more…
…Interest rate targeting policy performs worse.
Introduction
LM1
IS1
Interest Rate
Output
LM2 Negative shocks in money market causes output to fall under monetary targeting regime…
…For interest rate targeting policy; however, output remain the same…
Introduction
Slope of IS and LM curveDifferent values of slope alter the result of monetary policy.
More concrete mathematical model can be derived to corporate all factors simultaneously.
There is no concise conclusion whether interest rate targeting policy is superior than monetary targeting policy.
Introduction
Combination Policy
The policy that lies between pure interest rate and monetary base targeting.
Monetary base is a function of the interest rate prevailed in the market.
For example, M(t) = c0 + c1R(t)» c1 = 0; pure monetary base targeting» c1 approach infinity; pure interest rate
targetingOptimum values of c1 and c0 can be calculated.
Theoretical Model
Theoretical Model
Mathematical model of IS and LM curve
IS Equation:
LM Equation:
M(t) = bo + b1R(t-1) + b2 Y(t-1) + b3M(t-1) + v(t)
Y(t) = ao + a1R(t-1) + a2 Y(t-1) + a3M(t-1) + u(t)
Theoretical Model
Y(t) = ao + a1R(t-1) + a2 Y(t-1) + a3M(t-1) + u(t)
M(t) = bo + b1R(t-1) + b2 Y(t-1) + b3M(t-1) + v(t)
where,Y = Real National Income (Output)M = Real Money stockR = Real Interest rateu = error term in IS equation where E(u) = 0, E(u2) = σu
2
v = error term in LM equation where E(v) = 0, E(v2) = σv2 and
E(uv) = ρu,v
Theoretical Model
Y(t) = ao + a1R(t-1) + a2 Y(t-1) + a3M(t-1) + u(t)
M(t) = bo + b1R(t-1) + b2 Y(t-1) + b3M(t-1) + v(t)
Expected sign of the coefficients are
Theoretical Model
For simplification, many features have been left out. 1) Price Rigidity2) Exchange International Trade and Capital Flow3) Government Sector
The models also assumes that monetary policy goal is to stabilize output of national income.
If price rigidity feature could be incorporated in the model , then we can build the model such that price stability is the goal for monetary policy.
Theoretical Model
More assumptions about the model
Central bank can control money supply completely.
Interest rate structure has been left out.
Diagram
1
Optimal value of the instrument in each policy is calculated.
2Each policy will be evaluated by comparing between their loss function of the deviation.
3
The loss function is
퐿=퐸 [푌−퐸(푌∗) ]2.
The analysis has the following procedure.
Theoretical Model
Interest Rate Policy
First, update IS equation by 1 period
Substitute LM equation into the updated IS equation to eliminate M(t)
Theoretical Model
Loss Function of interest rate targeting policy is
If monetary transmission only works though interest rate, the coefficient a3 would be zero. This reduces the loss function into σu
2 as in the traditional IS-LM model.
E{Y(t+1) - E[Y*(t+1)] }2 = 푎32휎푣2+ 2푎3휌푢,푣+ 휎푢2
Theoretical Model
Monetary Aggregate Policy
Update LM equation by 1 period
Rewrite the updated LM equation
Update IS equation by 1 period
Substitute R(t) is the updated IS equation to eliminate R(t)
Theoretical Model
Loss Function of interest rate targeting policy is
E{Y(t+1) - E[Y*(t+1)] }2 = 휎푢2 − 2(푎1/푏1)휌푢,푣 + (푎1/푏1)2휎푣2
Theoretical Model
Combination Policy
Assume linear function: M(t) = c0 + c1R(t)Rewrite as, R(t) = (M(t) - c0)/c1
Update the IS equation
Substitute R(t) in the updated equation
Substitute LM Equation to eliminate M(t)
Theoretical Model
Lose Function of combination targeting policy is
L = {(a1/c1)+a3}2휎v2 + 2{(a1/c1)+a3}휌u,v + 휎u
2
Minimize loss function by differentiating L with respect to c1
휕 L /휕c1 = - 2휎v2{ (a1/c1)+a3} (a1/c1
2) – 2휌u,v(a1/c12) = 0
vߪ2{ (a1/c1)+a3} (a1/c1
u,v(a1/c1ߩ + (22) = 0
Theoretical Model
Conclusion
Lose Function of interest rate targeting policy isa3
vߪ22 +2a3 uߪ+u,vߩ
2
Lose Function of monetary base targeting policy isuߪ
2 -2(a1/b1)ߩu,v+(a1/b1)2ߪv2
Lose Function of combination targeting policy is
{(a1/c1)+a3}2ߪv2 + 2{(a1/c1)+a3}ߩu,v uߪ +
2
Where c1 can be calculated as vߪ
2{ (a1/c1)+a3} (a1/c1u,v(a1/c1ߩ + (2
2) = 0
Empirical Analysis
Empirical Analysis
Estimate all the parameters using Vector Error Correction Model (VECM)
VECM model ensures that any deviation from equilibrium will be adjusted, as the IS-LM model illustrates
IS equation : Y(t) = ao + a1R(t-1) + a2 Y(t-1) + a3M(t-1) + u(t)LM equation : M(t) = bo + b1R(t-1) + b2 Y(t-1) + b3M(t-1) + v(t)
And also the variance and covariance of disturbances, which are 휎u
2 휎v2 and 휌u,v
Empirical Analysis
Description Source
Y Real GDP (1988 Baht) NESBD
R MLR BOT
Inflation CPI IMF
M M2A BOT
Data Description
1.) Variables R and M are converted in real terms using CPI.2.) The estimation has all variables in logarithmic scale.
Empirical Analysis
Econometrics Procedure
1.)Unit Root Test
2.)Co-integration Test
3.)VECM
Empirical Analysis
Econometrics Procedure
1.)Unit Root Test
2.)Co-integration Test
3.)VECM
Empirical Analysis
Unit Root Test
t-Statistic Prob.*
LMLR -0.90668 0.771
LGDP -1.86655 0.3421
LM2A -1.78337 0.6823
*MacKinnon (1996) one-sided p-values.
Empirical Analysis
Econometrics Procedure
1.)Unit Root Test
2.)Co-integration Test
3.)VECM
Empirical Analysis
Unrestricted Cointegration Rank Test (Trace) Series LM2A LGDP
Hypothesized Trace 0.05
No. of CE(s) Eigenvalue Statistic Critical Value Prob.**
None * 0.387709 13.58771 12.32090 0.0305**At most 1 0.012621 0.342937 4.129906 0.6208
Trace test indicates 1 cointegrating eqn(s) at the 0.05 level* denotes rejection of the hypothesis at the 0.05 level**MacKinnon-Haug-Michelis (1999) p-values
Empirical Analysis
Econometrics Procedure
1.)Unit Root Test
2.)Co-integration Test
3.)VECM
Empirical Analysis
Vector Error Correction Estimates
t-statistics in [ ]
Cointegrating Eq: CointEq1
LOG(GDP(-1)) 1.000000
LOG(M2AN(-1)) -0.44536
[-26.5829]
C -8.59015
[-23.9381]
Empirical Analysis
Error Correction: D(LOG(GDP)) D(LOG(M2AN))
CointEq1 -0.399071*** 0.194854*
D(LOG(GDP(-1))) 0.303786* 0.384474*
D(LOG(M2AN(-1))) 0.635298*** 0.393956*
LOG(MLR(-1)) -0.101852** 0.061683*
*** = 1% significant / ** = 5% significant /* = 10% significant
Empirical Analysis
The equations can be illustrated as follows,
ΔYt = -0.4[Yt-1 - 0.45Mt-1 - 8.6] + 0.3(ΔYt-1) + 0.64(ΔMt-1) – 0.1Rt-1
ΔMt = 0.2[Yt-1 - 0.45Mt-1 - 8.6] + 0.38(ΔYt-1) + 0.4(ΔMt-1) +0.06Rt-1
After rearranging equations, the estimated values of all parameters could be obtained
Empirical Analysis
a0 = 3.44, a1 = -0.1, a2 = 0.9, a3 = 0.82b0 = -1.72, b1 = 0.06, b2 = 0.58, b3 = 1.31
The estimation also provides residual terms. The variance and covariance of residual terms are as follows.
휎u2 = 0.001
휎v2 = 0.0014
휌u,v = 0.007
Empirical Analysis
Therefore, we can substitute the parameters and variance / covariance in the loss function to compare between interest rate targeting policy and monetary base targeting policy.
Loss function from interest rate targeting policy = a3
2휎v2 +2a3 휌u,v+휎u
2 = 0.00309
Loss function from monetary base targeting policy= 휎u
2 -2(a1/b1)휌u,v+(a1/b1)2휎v2 = 0.00632
Empirical Analysis
Interest rate targeting policy is preferred to monetary base targeting policy
Combination Policy Calculate c1 from vߪ
2{ (a1/c1)+a3} (a1/c1u,v(a1/c1ߩ + (2
2) = 0
Get c1 = 0.12185 Evaluate loss functionL = {(a1/c1)+a3}2휎v
2 + 2{(a1/c1)+a3}휌u,v + 휎u2 = 0.00099
Empirical Analysis
Loss function estimated
Interest rate targeting policy = 0.00309Monetary base targeting policy = 0.00632Optimal combination policy = 0.00099
Interest rate targeting policy seems to be superior to monetary base targeting. However, optimal combination policy performs better than both.
Conclusion and Suggestion
Conclusion and Suggestion
This paper illustrates the theoretical idea regarding a combination of monetary base targeting and interest rate targeting in the context of Thailand.
Empirical evidence suggested that interest rate targeting is superior to monetary base targeting.
Nevertheless, combination policy performs better.
Conclusion and Suggestion
The model could be modified to corporate more realistic features of the economy. Exchange Rate and international trade Government sector, private sector and consumers Price Rigidity
Also, loss function could also be modified. The ARCH model of the disturbance? Dynamic optimization?