the forward spread in thailand
TRANSCRIPT
Thitirat Sittakaradej
The Forward Spread in Thailand: Its Predictive Power and Key Influencing Factors
Term structure of interest rate
Studied by
Goodfriend(1998) and Kim
and Orphanides (2007),
the term structure of
interest rate serves a
vital role as a precise
source of information
on public expectations
and responses to the
shifts in the economy.
1. Introduction
Expectation Hypothesis (EH)
Expectation hypothesis (EH) under the term structure
of interest rate explains the connection between the
interest rates and the maturities of investment though
the forward rates.
where is the spot interest rate on n-year to maturity of zero-coupon bond at time t
is the forward interest rate on m-year to maturity of zero-coupon bond at time t
Therefore, with EH, the term structure of interest
rate provides the best forecast of future spot rates.
Case of two time periods :
Prior studies
The term structure of
interest rate contains
little information
about future spot
rates and the study
leads to a conclusion
of weak predictive
power overtime.
(See Fama and Bliss, 1987; Mishkin, 1988; Davis, 1996).
from Germany from U.S.A.
The forward spreads are
very powerful predictors
of short-term interest rate
changes. With less
predictive power for
long-term rates, both the
signs and the coefficient
estimates are still
coherent with EH.
(See Boero and Torricelli, 1999).
The term structure of
interest rate contains
little information
about future spot
rates and the study
leads to a conclusion
of weak predictive
power overtime.
(See Fama and Bliss, 1987; Mishkin, 1988; Davis, 1996).
The term structure of
interest rate contains
little information
about future spot
rates and the study
leads to a conclusion
of weak predictive
power overtime.
(See Fama and Bliss, 1987; Mishkin, 1988; Davis, 1996).
The forward spreads are
very powerful predictors
of short-term interest rate
changes. With less
predictive power for
long-term rates, both the
signs and the coefficient
estimates are still
coherent with EH.
(See Boero and Torricelli, 1999).
2. Objective
This paper aims to study the
predictive power of the forward
rate from the expectation hypothesis
(EH) under the term structure of
interest rate in forecasting future
spot rate from the market, and
examine the key factors influencing
this forward spread.
3. Forward Spread in Thailand
Data set • The time period is 2002-2012.
• Focus on yields of zero-coupon government bonds
with 1 year, 3 years and 5 years time to maturity.
• Other monthly data include policy rate, Manufacturing
Production Index (MPI), export value, amount of capital inflow,
number of working hour, NEER and inflation rate.
Methodology Step 1 : Calculate the forward spread from expectation
hypothesis under the term structure of interest rate
Step 2 : (Model1) Test the predictive power of the forward spread
Step 3 : (Model2) Determine the key factors influencing the forward spread
Model 1 : Predictive power of the forward spread
The forward spread would be a powerful predictor of future spot rate if the value of β is 1.
where is the spot interest rate on n-year to maturity of zero-coupon bond at time t is the forward interest rate on m-year to maturity of zero-coupon bond at time t and is ith exogenous variable including policy rate, inflation and export value at time t
Case of two time periods at time = 0 :
Yield spread (ysp(n))
Forward spread (fsp(n))
Model1 : Result (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
OLS OLS FGLS ARCH(1) GARCH(2,1) OLS FGLS ARCH(1) OLS FGLS ARCH(1)
VARIABLES ysp1 ysp1 ysp1 ysp1 ysp1 ysp3 ysp3 ysp3 ysp5 ysp5 ysp5
fsp1 1.159*** 1.174*** 1.174*** 1.034*** 0.748***
(0.223) (0.231) (0.222) (0.0944) (0.0365)
fsp3 -1.135*** -1.135*** -1.251***
(0.257) (0.205) (0.141)
fsp5 -2.741*** -2.741*** -3.001***
(0.340) (0.321) (0.169)
dpolrate 0.942 0.942 -0.385 -0.497** 1.101* 1.101** -0.147 0.616 0.616 -0.0270
(0.697) (0.625) (0.271) (0.238) (0.625) (0.525) (0.304) (0.528) (0.429) (0.253)
infla -0.170*** -0.170** 0.0448* 0.00409 -0.0613 -0.0613 -0.101*** 0.00460 0.00460 -0.0345
(0.0535) (0.0736) (0.0260) (0.0127) (0.0546) (0.0468) (0.0355) (0.0436) (0.0353) (0.0249)
export -1.22e-06 -1.22e-06 -8.64e-08 -4.24e-08 -1.45e-06 -1.45e-06 1.60e-06** -4.06e-06*** -4.06e-06*** -6.19e-07
(1.02e-06) (9.10e-07) (4.06e-07) (2.69e-07) (9.64e-07) (8.77e-07) (6.29e-07) (8.66e-07) (7.77e-07) (5.02e-07)
Constant -0.611*** 0.348 0.348 -0.538*** -0.427*** 1.575*** 1.575*** 0.0123 3.185*** 3.185*** 1.598***
(0.170) (0.414) (0.433) (0.207) (0.121) (0.394) (0.432) (0.256) (0.426) (0.412) (0.246)
Observations 120 119 119 119 119 119 119 119 119 119 119
Adjusted
R-squared 0.180 0.262 0.262 . . 0.216 0.216 . 0.410 0.410 .
fwdsp=1 0.477 0.452 0.434 0.720 0 0 0 0 0 0 0
CI(lb) 0.718 0.717 0.734 0.849 0.676 -1.644 -1.541 -1.528 -3.413 -3.376 -3.333
CI(ub) 1.600 1.632 1.614 1.219 0.819 -0.625 -0.729 -0.974 -2.068 -2.106 -2.669
BIC 377.4 373.6 373.6 307.1 306.7 360.3 360.3 302.4 315.3 315.3 251.8
Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Time to maturity : 1 -4
-20
2
0 .5 1 1.5 2
ysp1 Fitted values
Slope = 1
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
OLS OLS FGLS ARCH(1) GARCH(2,1) OLS FGLS ARCH(1) OLS FGLS ARCH(1)
VARIABLES ysp1 ysp1 ysp1 ysp1 ysp1 ysp3 ysp3 ysp3 ysp5 ysp5 ysp5
fsp1 1.159*** 1.174*** 1.174*** 1.034*** 0.748***
(0.223) (0.231) (0.222) (0.0944) (0.0365)
fsp3
fsp5
dpolrate 0.942 0.942 -0.385 -0.497**
(0.697) (0.625) (0.271) (0.238)
infla -0.170*** -0.170** 0.0448* 0.00409
(0.0535) (0.0736) (0.0260) (0.0127)
export -1.22e-06 -1.22e-06 -8.64e-08 -4.24e-08
(1.02e-06) (9.10e-07) (4.06e-07) (2.69e-07)
Constant -0.611*** 0.348 0.348 -0.538*** -0.427***
(0.170) (0.414) (0.433) (0.207) (0.121)
Observations 120 119 119 119 119
Adjusted R-squared 0.180 0.262 0.262 . .
fwdsp=1 0.477 0.452 0.434 0.720 0
CI(lb) 0.718 0.717 0.734 0.849 0.676
CI(ub) 1.600 1.632 1.614 1.219 0.819
BIC 377.4 373.6 373.6 307.1 306.7
Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
β is 1 and close to 1.
Time to maturity : 1 Time to maturity : 1
For bond with 1 year
time to maturity,
forward spread
predicts future yield spread well .
Time to maturity : 3
-2-1
01
23
0 .5 1 1.5 2
ysp3 Fitted values
Slope = 1
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
OLS OLS FGLS ARCH(1) GARCH(2,1) OLS FGLS ARCH(1) OLS FGLS ARCH(1)
VARIABLES ysp1 ysp1 ysp1 ysp1 ysp1 ysp3 ysp3 ysp3 ysp5 ysp5 ysp5
fsp1 1.159*** 1.098*** 1.098*** 1.004*** 0.740***
(0.223) (0.224) (0.214) (0.0881) (0.0449)
fsp3 -1.135*** -1.135*** -1.251***
(0.257) (0.205) (0.141)
fsp5 -2.770*** -2.770*** -2.999***
(0.331) (0.315) (0.150)
gpolrate 4.341** 4.341** -0.812 -1.229*** 1.101* 1.101** -0.147 2.485* 2.485** -0.276
(1.714) (1.806) (0.614) (0.379) (0.625) (0.525) (0.304) (1.307) (1.198) (0.669)
infla -0.187*** -0.187*** 0.0417* 0.00168 -0.0613 -0.0613 -0.101*** -0.00222 -0.00222 -0.0346
(0.0511) (0.0703) (0.0242) (0.0133) (0.0546) (0.0468) (0.0355) (0.0426) (0.0360) (0.0245)
export -1.37e-06 -1.37e-06 -1.47e-07 -9.67e-08 -1.45e-06 -1.45e-06 1.60e-06**
-4.18e-
06***
-4.18e-
06*** -5.47e-07
(1.00e-
06)
(9.15e-
07)
(4.25e-
07) (3.21e-07) (9.64e-07) (8.77e-07) (6.29e-07) (8.61e-07) (7.91e-07) (4.91e-07)
Constant -0.611*** 0.484 0.484 -0.484** -0.391*** 1.575*** 1.575*** 0.0123 3.262*** 3.262*** 1.565***
(0.170) (0.407) (0.431) (0.206) (0.133) (0.394) (0.432) (0.256) (0.419) (0.409) (0.228)
Observations 120 119 119 119 119 119 119 119 119 119 119
Adjusted R-squared 0.180 0.290 0.290 . . 0.216 0.216 . 0.421 0.421 .
fwdsp=1 0.477 0.661 0.647 0.960 0 0 0 0 0 0 0
CI(lb) 0.718 0.655 0.674 0.832 0.652 -1.644 -1.541 -1.528 -3.426 -3.393 -3.292
CI(ub) 1.600 1.542 1.522 1.177 0.828 -0.625 -0.729 -0.974 -2.113 -2.146 -2.705
BIC 377.4 369.0 369.0 308.1 307.6 360.3 360.3 302.4 313.0 313.0 251.6
Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
β is less than 1.
Time to maturity : 3 Time to maturity : 3
Time to maturity : 5 -2
-10
12
0 .5 1 1.5 2
ysp5 Fitted values
Slope = 1
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
OLS OLS FGLS ARCH(1) GARCH(2,1) OLS FGLS ARCH(1) OLS FGLS ARCH(1)
VARIABLES ysp1 ysp1 ysp1 ysp1 ysp1 ysp3 ysp3 ysp3 ysp5 ysp5 ysp5
fsp1 1.159*** 1.098*** 1.098*** 1.004*** 0.740***
(0.223) (0.224) (0.214) (0.0881) (0.0449)
fsp3 -1.138*** -1.138*** -1.315***
(0.251) (0.201) (0.133)
fsp5 -2.741*** -2.741*** -3.001***
(0.340) (0.321) (0.169)
gpolrate 4.341** 4.341** -0.812 -1.229*** 4.073*** 4.073*** -0.926 0.616 0.616 -0.0270
(1.714) (1.806) (0.614) (0.379) (1.546) (1.509) (0.684) (0.528) (0.429) (0.253)
infla -0.187*** -0.187*** 0.0417* 0.00168 -0.0713 -0.0713 -0.0837** 0.00460 0.00460 -0.0345
(0.0511) (0.0703) (0.0242) (0.0133) (0.0532) (0.0473) (0.0336) (0.0436) (0.0353) (0.0249)
export -1.37e-06 -1.37e-06 -1.47e-07 -9.67e-08
-1.60e-
06*
-1.60e-
06*
2.06e-
06*** -4.06e-06*** -4.06e-06*** -6.19e-07
(1.00e-
06)
(9.15e-
07)
(4.25e-
07) (3.21e-07) (9.51e-07) (8.99e-07) (5.82e-07) (8.66e-07) (7.77e-07) (5.02e-07)
Constant -0.611*** 0.484 0.484 -0.484** -0.391*** 1.654*** 1.654*** -0.168 3.185*** 3.185*** 1.598***
(0.170) (0.407) (0.431) (0.206) (0.133) (0.388) (0.435) (0.234) (0.426) (0.412) (0.246)
Observations 120 119 119 119 119 119 119 119 119 119 119
Adjusted R-squared 0.180 0.290 0.290 . . 0.241 0.241 . 0.410 0.410 .
fwdsp=1 0.477 0.661 0.647 0.960 0 0 0 0 0 0 0
CI(lb) 0.718 0.655 0.674 0.832 0.652 -1.634 -1.536 -1.576 -3.413 -3.376 -3.333
CI(ub) 1.600 1.542 1.522 1.177 0.828 -0.641 -0.739 -1.055 -2.068 -2.106 -2.669
BIC 377.4 369.0 369.0 308.1 307.6 356.5 356.5 301.4 315.3 315.3 251.8
Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
β is a lot less than 1.
Time to maturity : 5 Time to maturity : 5
For bonds with higher time
to maturity, both signs and
coefficient estimates of the
impacts show low or no predictive power.
where is the spot interest rate on n-year to maturity of zero-coupon bond at time t is the forward interest rate on m-year to maturity of zero-coupon bond at time t is policy rate at time t and is ith variable including MPI, export value, amount of capital inflow, working hours,
NEER and inflation rate at time t
Model 2 : Key Influencing Factors of the Forward Spread
The changes in factors affect the forward spread
when the coefficient estimates are significant.
Forward spread (fsp(n))
Model2 : Result
(1) (2) (3) (4) (5) (6) (7) (8)
VARIABLES fsp1 fsp1 fsp1 fsp1 fsp1 fsp1 fsp3 fsp5
1. Monetary Policy
dpolrate 1.047*** 0.882*** 0.865*** 0.756*** 0.764*** 0.742*** 0.761*** 0.398***
(0.230) (0.243) (0.258) (0.283) (0.283) (0.281) (0.250) (0.151)
2. Business Cycle
ympi 0.768** 0.716* 0.765* 0.882* 0.864* -0.288 0.221
(0.346) (0.428) (0.430) (0.448) (0.444) (0.396) (0.239)
yexport 0.0950 0.0464 -0.0764 -0.188 0.579 0.194
(0.460) (0.461) (0.479) (0.480) (0.427) (0.258)
ykinflow 0.0510 0.0445 0.0402 0.0355 0.0100
(0.0374) (0.0381) (0.0378) (0.0337) (0.0204)
ywhour -0.591 -0.382 -0.662 1.537 1.125
(2.791) (2.801) (2.784) (2.480) (1.499)
yneer -1.070 -0.998 3.784*** 0.923
(1.124) (1.116) (0.994) (0.601)
3. Inflation
dinfla 0.0896 -0.0208 0.000447
(0.0551) (0.0491) (0.0296)
Constant 0.600*** 0.570*** 0.563*** 0.543*** 0.564*** 0.577*** 0.577*** 0.496***
(0.0397) (0.0465) (0.0575) (0.0594) (0.0631) (0.0631) (0.0562) (0.0340)
Adjusted R-squared 0.144 0.171 0.163 0.164 0.163 0.176 0.211 0.146
Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Under OLS, growth of the
policy rate becomes an only influencing factor.
(1) (2) (3)
VARIABLES fsp1 fsp3 fsp5
1. Monetary Poicy
dpolrate 12.247*** 35.592*** -1.679
(4.284) (11.464) (1.477)
2. Business Cycle
ympi 7.741 -4.189 2.426
(7.169) (19.109) (2.353)
yexport 2.133 32.247 0.152
(7.408) (19.861) (2.506)
ykinflow -0.884* -3.320** 0.379*
(0.514) (1.365) (0.198)
ywhour 499.924*** -1233.362*** 141.304***
(54.819) (142.741) (14.523)
yneer 6.926 27.521 -8.034
(14.640) (38.928) (5.840)
3. Inflation
dinfla -1.843* -3.873 -2.560***
(0.998) (2.711) (0.288)
Constant 0.440 0.054 0.171
Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Also, under VECM, an immediate effect is shown.
An increase in a policy
rate would also increase the forward spread.
Effects of the growth
rate of capital inflow,
working hours and
inflation rate on different maturities of
the forward spreads vary a lot.
1) The predictive power of the forward spread in
Thailand varies across different maturities. While the
Expectation Hypothesis(EH) explains most of the
change in future interest rates for yields with 1-year
time to maturity, the forecasts for yields with 3-year
and 5-year time to maturity are biased.
2) Only key factor influencing forward spread is growth
of policy rate. The impact is stronger in yields with
shorter time to maturities.
4. Conclusion