190 E.P.

de Lara-Tuprio, et al
DLSU Business & Economics Review (2017) 27(1): 190-200

RESEARCH ARTICLES

Forecasting the Term Structure of Philippine Interest

Rates Using the Dynamic Nelson-Siegel Model
Elvira P. de Lara-Tuprio, Ramil T. Bataller, Allen Dominique D. Torres,
Emmanuel A. Cabral and Proceso L. Fernandez Jr.
Ateneo de Manila University, Quezon City, Philippines
edelara-tuprio@ateneo.edu

Abstract: The three-factor Nelson-Siegel model is a widely used model for forecasting the term structure of interest rates.
Several extensions have recently been proposed. Even for the original model, different methods of treating the parameters
have been shown. Ultimately, what works best depends on the data used to estimate the parameters. In this paper, the
original three-factor model with fixed shape parameter was applied to forecast the term structure using market data from
the Philippines. Instead of giving a pre-determined model for the latent factors, the best time series model for them was
searched using standard statistical tools. Based on the historical data, the best model for each latent factor is of the form
ARMA(p,q)+eGARCH(1,1). The dependence structure of these parameters was considered in generating their future values.
This was carried out by finding the joint distribution of the residuals via appropriate copula. Results show that forecast of
interest rates for different tenors is reliable up to the near future. For an active market, this is good enough since the models
for the parameters can be adjusted as new information comes in.

Keywords: Yield curve, Nelson-Siegel, forecasting, copula, time series

JEL Classification: G17, E47, C53

In recent years, the influx of new financial products
in the Philippines, including financial derivatives,
has become inevitable. Some local banks have been
granted limited authority by the Bangko Sentral ng
Pilipinas (BSP) to engage in specified derivatives
transactions, while some others are still in the stage of
preparing to apply for a license for such transactions.
The bond market is also on the rise with an increasing
participation of corporate bond issuers. Amidst these
developments is the need for sound risk management
structure and reliable models for the pricing and
valuation of financial products.
Financial risk management and valuation of
financial securities require sound and reliable
mathematical models. Loss estimates and pricing are
based on current and forecast of underlying economic
variables such as interest rates and currency exchange
rates. Currently, several financial institutions in the
Philippines are either using vendor-developed systems,
which are essentially “black box,” or have developed
their internal models which are mostly based on
popular or classical models. There is indeed a challenge
for financial institutions, particularly in the banking
industry, to develop their own models which are more
reliable, relevant, and suited to the market data.
Public debt management also requires sound
mathematical models for forecasting the evolution of
macroeconomic factors such as interest rates, exchange

Copyright © 2017 by De La Salle University

 TERM STRUCTURE FORECASTING TERM STRUCTURE
TERM STRUCTURE FORECASTING
FORECASTING 4
TERM STRUCTURE FORECASTING
TERM STRUCTURE FORECASTING
TERM STRUCTURE FORECASTING time series model will betools. 4 using standard statistica
searched
time series model will be searched using
time standard
series statistical The model that will be used
model will be searched using standard 191 statistical tools.
Philippine Interest Rates Using the Dynamic Nelson-Siegel Model
time series model will be searched using standard statistical tools. The model that will be u
time series model will of
time
be the form
searched
series model will beofsearched
ARMA(p,q)+eGARCH(1,1)
using standard statistical the form of using
the
tools. form
or
The
standard
ARMA
ARMA(p,q)+eGARCH(1,1)
model forstatistical
ARMA(p,q)+eGARCH(1,1)
thatthe will mean tools.
be used equationor The model
is ARMA and forthat
or Exponential
ARMA thewill forbe
mean the u
eq
rates, and primary budget of the form ARMA(p,q)+eGARCH(1,1)
deficit/surplus. Finding an tools. The model or ARMA that will for the be usedmeanisequation of the form and Exponential
optimal debt strategy
of the form ARMA(p,q)+eGARCH(1,1) means of the form
determining
GARCH for the variance ARMA(p,q)+eGARCH(1,1)
a mix
or ARMAof debtequation.
for the mean
GARCH GARCH
Moreover,
ARMA(p,q)+eGARCH(1,1)
for the equationor for
variance ARMAthe
forand for
variance
forecasting,
Exponential
equation. the
or mean
equation.
ARMAthe equation
dependence
Moreover, for the for and
Moreover, mean Exponential
for forecasti
structure
forecasting, ofthe thed
instruments that minimizesGARCH cost to thefor government,
the variance in equation. equation and Exponential
Moreover, for forecasting, GARCH the for dependence
the variance structure of
terms of interest
GARCH payment,
for the variance beta GARCH
subject
parameters
equation. for
to a the
Moreover,willvariance
prudent beforrisk equation.
considered
forecasting,
beta beta
equation.
by the
parameters Moreover,
usingparameters
Moreover, befor
an appropriate
dependence
will for forecasting,
will be considered
forecasting,
structure
considered copulaof
bythe the to dependence
the
using byanusing
dependence
obtain theiranjoint
appropriate structure
appropriate
copula of
level. It may then be used beta to guide the policy makers structure of the
parameters will be considered by using an appropriate copula to obtain their joint beta parameters will be considered
in creating
beta a short-term
parameters orbeta
will bedistribution.
considered parameters
medium-term by using will
borrowing be considered
an appropriate
distribution. by an
distribution.
by copula
using using
to appropriate
obtain an appropriate
theircopula joint to copula obtaintotheir obtain joint their joint
plan, which will hopefully distribution. improve the domestic debt distribution.
market due to enhanced predictability
distribution. distribution. and transparency The remainderThe of this paper
asremainder of this proceedspaper as follows:
proceeds as follows:
The remainder of this paperThe proceeds
remainder follows:
of this paper The nextproceeds section as discusses
follows: the
TheNelsonextT
related to public debt. The next section discusses the Nelson-Siegel equation
The remainder of this paper proceeds as follows: The next section discusses the Nel
Forecasting the termofstructure of interest
The remainder this paper
Siegel The
equation proceedsfor rates
remainder has
theasyield.
follows:
ofSiegelWefor
this The
paperthe
Siegel
then
equation
yield.
next
describe We then
section
proceedsequation
for the as
the describe
discusses
follows:
for
data
yield. the
andWe
the
theThe
yield.
the data
Nelson-
next
thengeneralWe and
describe
the
section
then general
thediscusses
describe
procedure datathat the the
anddatatheNel
will an
be
ge
created a huge literature spanning several decades. procedure that will be implemented, to be followed
Siegel equation for the yield. We then describe the data and the general procedure that will
Some of
Siegel these make
equation useyield.
for the ofSiegel
mean-reverting
Weequation
then describe stochastic
for theyield.
the data and with
We thegeneral
the
then
implemented,results
describe of to the
procedure the beimplementation
data that
followed and will the including
begeneral
with procedure
theofresults theof thethat implemwill
implemented,
models, such as the models of Vasicek (1977) and Cox,
to be followed with
implemented, the results to be of the
followed implementation
with the resultsincluding the the time series
implementation
time series found for the beta parameters, the copula
Ingersoll, and to Ross implemented, towhich
be followed with to theobtainresults ofjoint the implementation including the time serie
implemented, be (1985).
followed Another
with the
implemented, classresults
toparameters,has implementation
beoffollowed
the used
with the beta
found including
results
for their
the of beta thethe time seriesthe
parameters,
implementation
parameters, and copula the forecast
including used the tothetime
obtain serieth
found
gained popularity among government policy makers for the beta found the copula used to obtain their
performance of the model. Finally, we provide the their joint
for the parameters, the joint
copula parameters,
used to obtain and forecas
due to for
found its good fit toparameters,
the beta observed found term forcopula
the the beta
structure parameters,
includes
used to obtain conclusion.the joint
their copula used to obtain
parameters, and the their
forecastjoint parameters, and the fore
found
performance for the beta parameters, the copula
performance used ofto obtain
the Finally, wejoint
model. their Finally, parameters,
we provide and
the theconclusfore
Nelson and Siegel’s (1987) model andofitsthe model. Finally,
variations. performance we provide of the model. the conclusion. provide the conclusion.
This paperof
performance aims
thetomodel.
presentperformance
a method
Finally, wefor offorecasting
provide the model. Finally, we provide the conclusion.
the conclusion.
performance of the model. Finally, weNelson-Siegel
provide the conclusion. The Nelson-Siegel Model Mo
the term structure of interest rates that is applicable to TheThe Nelson-Siegel Model Model The Nelson-Siegel
the Philippine market data. It is based on The theNelson-Siegel
framework Model The Nelson-Siegel Model
The aLetNelson-Siegel (𝜏𝜏)pricebe the atModel
price atof time of a zero-coupon
1𝑡𝑡 peso 𝜏𝜏,bo
developed by Diebold and LiLet 𝑝𝑝𝑡𝑡 (𝜏𝜏)which
(2006), be the isprice a at time Let𝑡𝑡 𝑝𝑝of
Let 𝑡𝑡 (𝜏𝜏) bebe𝑝𝑝the
zero-coupon 𝑡𝑡the price bond
at time
time thatt 𝑡𝑡of pays
a azero-coupon
zero-coupon at time bond𝑡𝑡 +that t
reinterpretation
Let 𝑝𝑝𝑡𝑡 (𝜏𝜏)ofbethe
themodel
price at Let 𝑡𝑡𝑝𝑝𝑡𝑡of(𝜏𝜏)
introduced
time by be the pricebond
Nelson
a zero-coupon at time
bond that 𝑡𝑡 of
pays a
that pays 1 peso at timezero-coupon
1 peso at time bond
𝑡𝑡 + that
𝜏𝜏, that pays 1
is, t is the peso at time 𝑡𝑡 + 𝜏𝜏
and Siegel (1987). Although is, 𝜏𝜏 is several Let 𝑝𝑝
the timeextensions (𝜏𝜏) be
to maturity the price
of 𝜏𝜏the
areis, is at time
is,
bond 𝜏𝜏
thetotime 𝑡𝑡
is of
the
in years. a zero-coupon
time
to maturity ofLet to
the𝑦𝑦bondmaturity
(𝜏𝜏)the bond
beyears.of that
the
the corresponding pays
bond 1
in pesoyears. at
Letthe𝑦𝑦continuously time
Let 𝑦𝑦 𝑡𝑡 c𝜏𝜏
𝑡𝑡 +
(𝜏𝜏)
𝑡𝑡 time maturity 𝑡𝑡of in bond in years.
Let yt (t) be 𝑡𝑡 (𝜏𝜏) be the
available,
is, 𝜏𝜏 is thethe original
time three-factor
to maturity is,of𝜏𝜏 the model
is thebond time will
in tobematurity
years. used.
Let 𝑦𝑦𝑡𝑡of (𝜏𝜏) thebe bond
the corresponding
corresponding in years. continuously Let continuously be the corresponding
𝑦𝑦𝑡𝑡 (𝜏𝜏)compounded nominal continuous
The procedure, however, is,
will 𝜏𝜏 isbethe time
modified
compounded nominal yield.compounded to maturity
and the of the
Thisyield. bond
compounded
yieldThis in
isnominal
also years.
yieldreferred is also Let
nominal
yield. 𝑦𝑦
toThis
referred 𝑡𝑡 (𝜏𝜏)
yield. be
as ayield
tozero the
This
as ais corresponding
rate.
zeroyield Then
alsorate. isThen
referred also referred tocontinuous
as a zero to a
results of the implementation
compounded nominal yield. will be highly
This yield isnominal
compounded dependent
also referred
yield.toThis as ayield
zero rate. is also Then referred to as a zero rate. Then
on the data used. compounded nominal yield. This yield𝑝𝑝 is(𝜏𝜏) also = referred
𝑒𝑒 −𝜏𝜏𝜏𝜏𝑡𝑡(𝜏𝜏) to as a zero rate. Then = 𝑒𝑒 −𝜏𝜏𝜏𝜏𝑡𝑡(𝜏𝜏) (
(𝜏𝜏)𝑡𝑡(𝜏𝜏)
𝑝𝑝𝑒𝑒𝑡𝑡(1)
−𝜏𝜏𝜏𝜏
𝑡𝑡 𝑝𝑝 𝑡𝑡 (𝜏𝜏) =
The Nelson-Siegel model is a popular𝑝𝑝𝑡𝑡model (𝜏𝜏) = for (𝜏𝜏)
𝑒𝑒 −𝜏𝜏𝜏𝜏𝑡𝑡 𝑡𝑡 (𝜏𝜏) (1)
𝑝𝑝𝑡𝑡 (𝜏𝜏) = 𝑒𝑒 −𝜏𝜏𝜏𝜏
the term structure of interest rates. It was introduced Ifisrate 𝑝𝑝is𝑡𝑡is(𝜏𝜏)
the=
the −𝜏𝜏𝜏𝜏
𝑒𝑒 at
instantaneous
(𝜏𝜏)
forward rate withatmaturity
𝑓𝑓𝑡𝑡 (𝜏𝜏) 𝑡𝑡at+
𝑡𝑡
by 𝑓𝑓Nelson and Siegel
If 𝑓𝑓𝑡𝑡 (𝜏𝜏) is the instantaneousIf forward
(1987) as a class of parametric 𝑓𝑓𝑡𝑡 (𝜏𝜏)If the with
instantaneous maturity instantaneousforwardtime 𝑡𝑡forward + 𝜏𝜏 with
rate rate
contracted maturitywith time
at time 𝑡𝑡, then
If 𝑡𝑡 (𝜏𝜏) is the instantaneous If forward
𝑓𝑓𝑡𝑡 (𝜏𝜏) is rate with
the instantaneous maturity at time
maturity rate
forward 𝑡𝑡 + 𝜏𝜏
at time contracted
with t +maturity at
t contracted timeat time𝑡𝑡, then
at time 𝑡𝑡 +t, 𝜏𝜏then contracted at time 𝑡𝑡, t
functions to capture the If range of isshapes typically
the instantaneous forward rate with𝜕𝜕maturity at time contracted at𝜕𝜕time
𝑓𝑓𝑡𝑡 (𝜏𝜏)
associated with yield curves. Diebold and Li (2006) ln 𝑝𝑝 𝑡𝑡 (𝜏𝜏) 𝑡𝑡 + 𝜏𝜏 𝜕𝜕 ln=𝑝𝑝𝑡𝑡−(𝜏𝜏) ln 𝑝𝑝𝑡𝑡𝑡𝑡,(𝜏𝜏)(t
𝜕𝜕 ln 𝑝𝑝𝑡𝑡 (𝜏𝜏) 𝑓𝑓𝑡𝑡 (𝜏𝜏) = − 𝜕𝜕 ln 𝑝𝑝 (𝜏𝜏) 𝑓𝑓𝑡𝑡(2) (𝜏𝜏) = − 𝑓𝑓𝑡𝑡 (𝜏𝜏)
reformulated it as a dynamic factor model 𝑓𝑓𝑡𝑡 (𝜏𝜏)and=− used 𝜕𝜕𝜕𝜕 𝑡𝑡 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕
𝜕𝜕𝜕𝜕 𝑓𝑓𝑡𝑡 (𝜏𝜏) = − 𝜕𝜕 ln 𝑝𝑝𝑡𝑡 (𝜏𝜏) (2)
it to forecast the yield curve. Several extensions and 𝑓𝑓𝑡𝑡 (𝜏𝜏) = − 𝜕𝜕𝜕𝜕
It follows that It It follows
follows that that 𝜕𝜕𝜕𝜕
variations
It follows that of the model have been introduced with the
goal of improving the out-of-sample It followsforecasts that (see for
It follows
example Christensen, Diebold, & Rudebuscha, 2011; that It follows that 1 𝜏𝜏 1 𝜏𝜏 1 𝜏𝜏 (
1 𝜏𝜏 𝑦𝑦𝑡𝑡 (𝜏𝜏) = ∫1 𝑓𝑓𝑡𝑡 𝜏𝜏(𝑢𝑢)𝑑𝑑𝑑𝑑 (3) 𝑦𝑦𝑡𝑡 (𝜏𝜏) =(𝑢𝑢)𝑑𝑑𝑑𝑑 ∫ 𝑓𝑓𝑡𝑡 (𝑢𝑢)𝑑𝑑𝑑𝑑
𝑦𝑦 (𝜏𝜏) = ∫ 𝑓𝑓 (𝑢𝑢)𝑑𝑑𝑑𝑑 𝑦𝑦 (𝜏𝜏) = ∫ 𝑓𝑓
Exterkate, Dijk, Heij, & Groenen, 2013; 𝑡𝑡 Koopman, 𝑡𝑡
𝜏𝜏 0 𝑦𝑦𝑡𝑡 (𝜏𝜏) = 1
𝜏𝜏 0 𝜏𝜏 (𝑢𝑢)𝑑𝑑𝑑𝑑 𝑡𝑡
∫ 𝑓𝑓𝑡𝑡 𝜏𝜏 0 𝑡𝑡 𝜏𝜏 0
Mallee, & Van der Wel, 2010). 𝑦𝑦𝑡𝑡 (𝜏𝜏) = 𝜏𝜏 ∫0 𝑓𝑓𝑡𝑡 (𝑢𝑢)𝑑𝑑𝑑𝑑
This paper will focus on theYield original three-factor Yield 𝜏𝜏 a0 curve
curve (3)
Yield curve at a particular point curve in timeat a 𝑡𝑡particular
is a curve point
Yield incurve
that describes time at𝑡𝑡theis spotatinterest
a particular a particular
thatpoint describes
rates
in point
timethe 𝑡𝑡inisspot time
a curve 𝑡𝑡 is that
interest a curve rate
des
Nelson-Siegel model. Following the method in
Yield curve at a particular point in time 𝑡𝑡 is a curve that describes the spot interest r
Diebold and Li (2006), the latent factors, which
Yieldcurves curve will
at abe particular point in different
time 𝑡𝑡 is or a curve that describes theorspot interest
𝑦𝑦𝑡𝑡 (𝜏𝜏) for different maturities 𝑦𝑦𝑡𝑡 (𝜏𝜏) for different
𝜏𝜏. These
referred to throughout this paper as beta parameters,𝑦𝑦𝑡𝑡
maturities
are 𝜏𝜏. These
typically
(𝜏𝜏) (𝜏𝜏)
𝑦𝑦𝑡𝑡different
curves
monotonic,
forYield for
curve are a typically
humped
maturities
at particular maturities
𝜏𝜏. monotonic,
S-shaped
These
point These
in𝜏𝜏.curves
time humpedcurves
t isarea curvetypically are monotor
typically
S-shaped
𝑦𝑦𝑡𝑡 (𝜏𝜏) for different maturities that 𝜏𝜏. These
describes curves the spot are typically
interest rates monotonic,
yt (t) for different humped or S-shaped
will be estimated from historical 𝑦𝑦𝑡𝑡 (𝜏𝜏) for data by fitting
different the
maturities 𝜏𝜏. These curves are typically monotonic, humped or S-shaped
Nelson-Siegel equation to the yield curves over time maturities t. These curves are typically monotonic,
and assumed to follow a time series model. However, humped or S-shaped (Nelson & Siegel, 1987). To
instead of assuming independent univariate AR(1) generate this range of shapes, a parsimonious model
processes for all the beta parameters, the best time introduced by Nelson and Siegel assumes that the
series model will be searched using standard statistical forward rate follows the equation
 URE FORECASTING 0𝑡𝑡 0𝑡𝑡 TERM STRUCTURE FORECASTING
5 5
0𝑡𝑡 𝜏𝜏→+∞ 𝑡𝑡
&STRUCTURE
RE
M Siegel, 1987).
FORECASTING To generate this range
FORECASTING of shapes,
Nelson
TERM and a5parsimonious
Siegel
STRUCTURE assumes
5
FORECASTING model
that the introduced
forward rate by 5
follows the equation
𝜏𝜏→+∞

TERM
TERM STRUCTURE
STRUCTURE FORECASTING
FORECASTING yield curve shifts. The factor 𝛽𝛽1𝑡𝑡 is related to 𝛽𝛽the yield
0𝑡𝑡 and an curve
increaseslope. 0𝑡𝑡5If
in 𝛽𝛽0𝑡𝑡 the slope
5increases of theequally.
all yields yield curve
Thus, t
and an
𝛽𝛽0𝑡𝑡1987).
el, increase in 𝛽𝛽0𝑡𝑡 this
To generate increases
rangeallofyields equally.
shapes, Thus,
𝛽𝛽 this
andfactor
a parsimonious yield
in 𝛽𝛽 curve
is responsible
model
an increase introduced forshifts.
by allThe
0𝑡𝑡
parallel
increases factor
yields 𝛽𝛽1𝑡𝑡 isThus,
equally. related thistofactor
the yield curve slope.
is responsible for If
part
and Siegel assumes (Nelson that&the Siegel, forward 1987). rate To follows generate 0𝑡𝑡
the equation this rangemodel of shapes, 0𝑡𝑡
a parsimonious model
(Nelson introduced & −𝜏𝜏/𝜆𝜆 Siegel, by 1987). 𝑓𝑓𝑡𝑡 (𝜏𝜏)generate = 𝛽𝛽0𝑡𝑡 +
egel,
syield range 1987). of shapes, To generate a parsimonious this range model of shapes, is introduced
defined a parsimonious
as by lim 𝑦𝑦 (𝜏𝜏) − lim introduced
𝑦𝑦 (𝜏𝜏), then itbyyield
is equalcurve to shifts.−𝛽𝛽 The .
𝜏𝜏factor
Lastly, 𝛽𝛽
𝛽𝛽 1𝑡𝑡 isisrelated related to To thethe
to yield curve s
el,
lson
rate 1987). &
curve
this Siegel,
range To
shifts. generate
1987).
The
of shapes, factor To this generate
𝛽𝛽a range is
parsimonious related of this shapes,
to (Nelson range
the yield
model a ofparsimonious
& shapes,
curve introduced
Siegel,
yield slope. curve
a 5
1987). 𝑡𝑡 model
parsimonious
If the
by
shifts. slope
To is
The𝑓𝑓 introduced
(𝜏𝜏)
of
defined
generate factorthe
𝑡𝑡 model
= 𝛽𝛽
yield
as this by
+introduced
curve
lim
is range𝛽𝛽
related 𝑦𝑦 𝑒𝑒
(𝜏𝜏)
−𝜏𝜏/𝜆𝜆
of to − 𝑡𝑡by
shapes,
the lim + yield𝛽𝛽 1𝑡𝑡
𝑦𝑦 (𝜏𝜏),
acurve 𝑒𝑒then
parsimoniousslope.
1𝑡𝑡
it𝑡𝑡
2𝑡𝑡 is If equal
the model
slopeto −𝛽𝛽 of introduce
the . Lastly,
yield c
el assumes TERM
that
TERM the STRUCTURE forward
STRUCTURE rateFORECASTING
FORECASTING follows the equation 𝑡𝑡 + 𝛽𝛽 0𝑡𝑡 1𝑡𝑡 2𝑡𝑡 5
bylim++5𝑦𝑦𝑡𝑡𝑡𝑡 (𝜏𝜏), then
1𝑡𝑡 𝜏𝜏→+∞ 𝜏𝜏→0 𝑡𝑡 + E.P.𝜆𝜆de Lara-Tuprio,
𝑡𝑡 1𝑡𝑡
(Nelson & & Siegel, 1987). ToTo generate this range of shapes, aa parsimonious et al
is model introduced by
1𝑡𝑡𝜏𝜏→+∞
(Nelson 192
Nelson Siegel, and 1987).
Siegel assumes generate that this
the range
forward 𝜏𝜏5 −𝜏𝜏/𝜆𝜆 of rate shapes, follows parsimonious
the equation model
defined
(4) as introduced
𝜏𝜏→0+
lim 𝑦𝑦𝑡𝑡𝑡𝑡 (𝜏𝜏) 𝑡𝑡 − it is equalthe to −𝛽𝛽
iegel assumes that the forward rate follows 5 −𝜏𝜏/𝜆𝜆 the equation where Nelson 𝛽𝛽 , and 𝛽𝛽 Siegel
, 𝛽𝛽 , assumes
and 𝜆𝜆𝑡𝑡 areisthat constants for 11
ard rate follows the equation
𝑓𝑓 𝑡𝑡 (𝜏𝜏) = 𝛽𝛽 + 𝛽𝛽 𝑒𝑒 + 𝛽𝛽 𝑒𝑒 𝜏𝜏→+∞
𝜏𝜏→+∞ 𝜏𝜏→0
𝜏𝜏→0
gel
son s defined
he forward rate assumes
and Siegel
as lim that 𝑦𝑦the
assumes
follows 𝑡𝑡 (𝜏𝜏) forward
− that
lim
the+equation
𝑦𝑦 the
𝑡𝑡rate
(𝜏𝜏),
0𝑡𝑡 forwardfollows
then 1𝑡𝑡 it is
Nelson𝜏𝜏𝛽𝛽and−𝜏𝜏/𝜆𝜆 rate the
equal
curvature, 𝑡𝑡
equation
followsto −𝛽𝛽defined
2𝑡𝑡
is
Siegel defined the
. Lastly,
𝜆𝜆𝑡𝑡 𝑡𝑡, 𝛽𝛽assumes
1𝑡𝑡 equation
by as
𝑡𝑡
Diebold
𝛽𝛽 lim 2𝑡𝑡 is 𝑦𝑦
that
related
and
(𝜏𝜏)
curvature, the − Li to as
lim the
forward
defined
2𝑦𝑦𝑦𝑦 (24)
(𝜏𝜏), rate then −
by𝜆𝜆FORECASTING
𝑦𝑦it
follows
Diebold
(3)
is equal−
and
𝑦𝑦0𝑡𝑡(120),
theLithatto −𝛽𝛽
equation
1𝑡𝑡 where
as 1𝑡𝑡 . 2𝑡𝑡 maturity
Lastly, 𝛽𝛽 is related given to in
the
𝑡𝑡 (24) 𝑦𝑦𝑡𝑡 (3)is−a 𝑦𝑦solution 𝑡𝑡 (120),
𝑡𝑡 𝑡𝑡 𝑡𝑡 2𝑦𝑦
TERM STRUCTURE FORECASTING 𝑡𝑡 where , and TERM 𝑡𝑡 are
𝑡𝑡TERM STRUCTURE
constants 𝜏𝜏 (4)
STRUCTURE + 𝑡𝑡 FORECASTING
and 0. Note 5= −𝑓𝑓𝑡𝑡2𝑡𝑡
Li as (𝜏𝜏)
𝜏𝜏→+∞
𝑒𝑒𝜏𝜏, 𝛽𝛽this 𝜆𝜆the ≠ defined 𝑥𝑥(𝜏𝜏)
𝜏𝜏→0
e this range Nelson
Nelson of𝑓𝑓𝑡𝑡 (𝜏𝜏)
(Nelson
(Nelson shapes,
and
and=Siegel Siegel
& 𝛽𝛽
& aSiegel,
0𝑡𝑡
parsimonious
+assumes
Siegel, assumes 𝛽𝛽1𝑡𝑡1987). −𝜏𝜏/𝜆𝜆
𝑒𝑒1987). that
that +To To model
the
the
𝛽𝛽generate
2𝑡𝑡 forward
generateforward introduced
0𝑡𝑡 1𝑡𝑡
thisrate
rate range
range2𝑡𝑡 by
follows
follows 𝜏𝜏→+∞
of of −𝜏𝜏/𝜆𝜆 the
shapes,
shapes, – b equation
equation
. a a
Lastly,
𝜏𝜏→0
parsimonious
parsimonious b
−𝜏𝜏/𝜆𝜆 curvature,
is related 𝑡𝑡 model
model to the introduced
by
introduced Diebold
curvature, andby by (4)
defined 2𝑦𝑦by 𝑡𝑡𝑡𝑡(24) − 𝑦𝑦𝑡𝑡𝑡𝑡(3) − 𝑦𝑦
this range of shapes, 𝜏𝜏 a=parsimonious model
−𝜏𝜏/𝜆𝜆 𝑡𝑡 months. 𝑓𝑓
𝛽𝛽𝜆𝜆𝑡𝑡0.
introduced (𝜏𝜏)
𝑦𝑦𝜏𝜏
=𝑒𝑒−−𝜏𝜏/𝜆𝜆
𝛽𝛽(4) by 𝑡𝑡+𝜏𝜏𝛽𝛽1𝑡𝑡where 𝑒𝑒 maturity 𝑡𝑡 +1t 𝛽𝛽 𝑒𝑒(4)2t(4) 𝑡𝑡
second-order
(4)− 𝑦𝑦𝑡𝑡 (3) linear ordinary differential 𝑓𝑓𝑡𝑡 (𝜏𝜏)is= eq
𝛽𝛽
𝑡𝑡 (𝜏𝜏) 𝑦𝑦𝑡𝑡0𝑡𝑡
curvature, defined 𝑓𝑓by Diebold 𝛽𝛽constants
and + Li 𝛽𝛽= as 𝑒𝑒and + 𝑡𝑡 (3)−𝜏𝜏/𝜆𝜆 2𝑡𝑡 is given in
pes, a parsimonious
−𝜏𝜏/𝜆𝜆 model −𝜏𝜏/𝜆𝜆 𝜏𝜏𝑡𝑡(𝜏𝜏)
introduced 2𝑦𝑦 𝑡𝑡𝑡𝑡𝑡𝑡(24) by − (120), (4) −𝜏𝜏/𝜆𝜆𝑡𝑡 months. 𝜆𝜆 𝜏𝜏
𝛽𝛽)+ 0𝑡𝑡 =
𝛽𝛽
, 1𝑡𝑡
𝛽𝛽 𝛽𝛽 𝑒𝑒
1𝑡𝑡 ,
+ 𝛽𝛽 𝛽𝛽
𝑡𝑡 +
2𝑡𝑡 , 𝑓𝑓
𝑒𝑒and 𝛽𝛽
(𝜏𝜏)
−𝜏𝜏/𝜆𝜆
𝑡𝑡 2𝑡𝑡 𝜆𝜆 =
𝑡𝑡𝑡𝑡+ 𝛽𝛽
𝑒𝑒
are
𝛽𝛽 0𝑡𝑡 𝑓𝑓+
0𝑡𝑡 𝛽𝛽 𝑒𝑒1𝑡𝑡 𝑒𝑒
1𝑡𝑡
−𝜏𝜏/𝜆𝜆
−𝜏𝜏/𝜆𝜆
−𝜏𝜏/𝜆𝜆
𝛽𝛽 +
𝑡𝑡0𝑡𝑡 second-order + 𝜆𝜆 𝛽𝛽
𝑡𝑡 𝛽𝛽 ≠2𝑡𝑡 𝑒𝑒 −𝜏𝜏/𝜆𝜆
2𝑡𝑡 Note
𝑡𝑡 𝑡𝑡−𝜏𝜏/𝜆𝜆
𝑒𝑒
𝜆𝜆 curvature,
+ linear that
𝛽𝛽𝑡𝑡𝑡𝑡 𝑥𝑥(𝜏𝜏) defined
𝑒𝑒
ordinary (4) = 𝑓𝑓 by
𝑡𝑡
𝑓𝑓
Diebold
(𝜏𝜏)𝜏𝜏
Diebold
differential
𝜏𝜏
(𝜏𝜏) is = a and
𝛽𝛽
𝑡𝑡and
solution Li Li
equation
+
asas
months.
𝛽𝛽 to 2𝑦𝑦
𝑒𝑒 the𝑡𝑡
−𝜏𝜏/𝜆𝜆
(24)
+ 𝛽𝛽
−
𝑒𝑒
𝑦𝑦 𝑡𝑡
−𝜏𝜏/𝜆𝜆
(120),, where where maturity give
orward 0𝑡𝑡 rate1𝑡𝑡 follows 𝑡𝑡 𝑡𝑡
𝜆𝜆and the&Siegel 0𝑡𝑡equation
𝑡𝑡 1𝑡𝑡 𝑡𝑡 1𝑡𝑡 2𝑡𝑡 2𝑡𝑡
(4)
𝜆𝜆𝑡𝑡 (4)
𝑡𝑡 𝑡𝑡
Nelson
Nelson (Nelson 𝑡𝑡and 2𝑡𝑡 Siegel,
Siegel assumes 1987). that
To the
=generate 𝜆𝜆the 𝑡𝑡 forward
𝑡𝑡
this 𝜆𝜆rate 𝑡𝑡 𝑡𝑡𝑡𝑡follows
range (Nelson
of 𝑡𝑡(Nelson
shapes, the & equation
aSiegel, &
𝑡𝑡parsimonious Siegel, 1987). 1987). To model generate
To2𝑡𝑡generate introduced this this range byrange of shapes, of shapes, a parsi ap
𝑡𝑡 𝜆𝜆assumes that forward 𝑒𝑒−𝜏𝜏/𝜆𝜆 rate follows the
𝑒𝑒𝑒𝑒−𝜏𝜏/𝜆𝜆 equation
𝑓𝑓𝑓𝑓𝑡𝑡𝑡𝑡(𝜏𝜏) = 𝛽𝛽 maturity is𝑡𝑡0𝑡𝑡given 1𝑡𝑡 in months.
0𝑡𝑡 + 𝛽𝛽of 𝑒𝑒= + 𝛽𝛽𝛽𝛽a2𝑡𝑡
𝜆𝜆and (𝜏𝜏) −𝜏𝜏/𝜆𝜆 −𝜏𝜏/𝜆𝜆
months.
rward 𝛽𝛽2𝑡𝑡 , and rate 𝑡𝑡 are constants
𝜆𝜆follows the equation 𝑡𝑡 ≠ 0.
TERM Note EstimationSTRUCTURE 𝛽𝛽𝑡𝑡0𝑡𝑡
that +
𝑥𝑥(𝜏𝜏) 𝛽𝛽1𝑡𝑡 1𝑡𝑡 FORECASTING
Parameters 𝑓𝑓 𝑡𝑡 (𝜏𝜏) is 2𝑡𝑡
+ solution to the
ws the equation where 𝛽𝛽 , 𝛽𝛽 , 𝛽𝛽 , and 𝜆𝜆 are constants months. and 𝜆𝜆 ≠ 0.𝜆𝜆
𝜆𝜆
Estimation 𝑡𝑡 Note that of Parameters
𝑥𝑥(𝜏𝜏) Estimation = 𝑓𝑓 (𝜏𝜏) of is
Parameters a solution to the 𝑥𝑥 ′′
+
order
𝛽𝛽 𝛽𝛽linear ,≠ and ordinary
𝜆𝜆where
TERM
𝑡𝑡 are
differential
constants
STRUCTURE 0𝑡𝑡𝜏𝜏 FORECASTING 1𝑡𝑡andequation 2𝑡𝑡 , and l
𝑡𝑡Note arethat constants and 𝑡𝑡 is a solution to the
(𝜏𝜏) 𝑡𝑡
2 ′ 5to 1𝑡𝑡 where 𝛽𝛽0𝑡𝑡𝛽𝛽0𝑡𝑡 , 𝛽𝛽1𝑡𝑡 , 𝛽𝛽2𝑡𝑡 , and 𝜆𝜆𝑡𝑡 are consta𝜆𝜆
s,re
constants
1𝑡𝑡
𝛽𝛽 and
onstants𝛽𝛽, 2𝑡𝑡
𝛽𝛽 + 𝜆𝜆and
,2𝑡𝑡 ,𝛽𝛽
𝑡𝑡 𝛽𝛽
and 𝑒𝑒 𝜆𝜆0.
−𝜏𝜏/𝜆𝜆 Note
are
,𝑡𝑡𝑡𝑡Parameters
𝛽𝛽 Nelson𝑡𝑡 ,+ constants
and that
𝛽𝛽 . 𝜆𝜆𝑡𝑡𝑥𝑥(𝜏𝜏)
and
Note areSiegel
𝑒𝑒 that and
−𝜏𝜏/𝜆𝜆
that
= 𝑡𝑡𝜆𝜆𝑓𝑓𝑡𝑡𝑡𝑡assumes
constants 𝜆𝜆≠
(𝜏𝜏) 𝑡𝑡 ≠ 0. isand
where
0. a solution
Note t
that
(𝜏𝜏) 𝜆𝜆𝑡𝑡𝛽𝛽is that
is ≠the aa 0. toNote
𝑥𝑥(𝜏𝜏)
forward
solution
solution
𝑥𝑥(𝜏𝜏)
the=(4) that
to
=𝑓𝑓(7),
to
, ratethe
and
𝑓𝑓𝑡𝑡𝑥𝑥(𝜏𝜏)
(𝜏𝜏) the is
Nelson
follows aare= solution
Estimation
Nelson (𝜏𝜏)
𝑓𝑓constants
and the 𝜏𝜏𝜏𝜏 ′′and
𝑥𝑥 Siegel to
is−𝜏𝜏/𝜆𝜆
equation
+ aofthe solution
Parameters
Siegel
and assumes
𝑥𝑥 + assumes the
that
𝑥𝑥 =
Note the
that forward
the forward (4)(4)rate = 𝑓𝑓𝜆𝜆𝑡𝑡rate follows isthea the
follows equa
the e
𝜆𝜆 ≠ 0. 𝑥𝑥(𝜏𝜏) = 𝑓𝑓 2that (𝜏𝜏) solution
Estimation 2𝑡𝑡 0𝑡𝑡 1𝑡𝑡
of 2𝑡𝑡 𝑡𝑡 In , 𝛽𝛽
Equation , 𝛽𝛽 𝑡𝑡
𝑡𝑡 −𝜏𝜏/𝜆𝜆 the 𝜆𝜆 parameter 𝑡𝑡 𝜆𝜆 is referred 𝜆𝜆 to≠ as 0. the shape 𝑥𝑥(𝜏𝜏)
parameter. For each t,towhen
ear0𝑡𝑡ordinary −𝜏𝜏/𝜆𝜆differential
𝜏𝜏 ,𝜆𝜆𝛽𝛽 equation 𝑡𝑡𝑓𝑓𝑓𝑓 𝑡𝑡𝑡𝑡(𝜏𝜏) ==𝛽𝛽Estimation ++𝜆𝜆𝛽𝛽𝛽𝛽 𝑒𝑒𝑒𝑒0. 𝑡𝑡𝑡𝑡+ 𝛽𝛽𝛽𝛽2𝑡𝑡2𝑡𝑡In 𝑒𝑒𝑒𝑒𝑡𝑡 =𝜆𝜆𝑓𝑓𝑡𝑡 (𝜏𝜏)
(𝜏𝜏) −𝜏𝜏/𝜆𝜆 −𝜏𝜏/𝜆𝜆
𝜏𝜏
1𝑡𝑡
where
where
𝑡𝑡
𝛽𝛽𝛽𝛽 ,, 𝛽𝛽 𝛽𝛽
2𝑡𝑡
, 𝛽𝛽 , , and
and 𝜆𝜆
𝜆𝜆 are
(Nelsonare constants
constants &
0𝑡𝑡
Siegel, 𝛽𝛽
and 1𝑡𝑡
and
0𝑡𝑡0𝑡𝑡 (4) 𝜆𝜆
1987).
2𝑡𝑡1𝑡𝑡 ≠ ≠
1𝑡𝑡 of0. To Note
Note
𝑡𝑡
Parameters +
generate that
that 𝑥𝑥(𝜏𝜏)
𝑥𝑥(𝜏𝜏) Equation
this =
range
𝑡𝑡
𝑓𝑓 (7),
(𝜏𝜏) of
𝑡𝑡 In
is the
is𝜆𝜆 a
shapes,Equation
a2parameter
solution
solution a 𝜆𝜆 (7), to
parsimonious the
𝜆𝜆
to theparameter
is
the referred 𝑡𝑡𝑡𝑡to
model is as referred shape
introduced as the pas
𝛽𝛽0𝑡𝑡 + 𝛽𝛽1𝑡𝑡 𝑒𝑒−𝜏𝜏/𝜆𝜆second-order 𝑡𝑡
second-order
+ 0𝑡𝑡
0𝑡𝑡 𝛽𝛽 1𝑡𝑡
1𝑡𝑡 𝑒𝑒 −𝜏𝜏/𝜆𝜆
𝑡𝑡 linear
2𝑡𝑡
2𝑡𝑡 linear𝑡𝑡
ordinary 2ordinary
𝑡𝑡𝑡𝑡 (4) differential1 differential 𝛽𝛽 equation equation
𝑡𝑡𝑡𝑡 In 𝜆𝜆 Equation
𝜆𝜆 𝑡𝑡 (7),
𝑡𝑡 𝑡𝑡𝑡𝑡 the
(5) parameter
𝑡𝑡 second-order l
𝑡𝑡 is referred
𝑡𝑡
linear to as
ordinary the differenti
+ linear
equation 𝛽𝛽2𝑡𝑡ordinary ordinary
𝑒𝑒 linear (Nelson
𝑡𝑡 differential 2𝑡𝑡& Siegel, 1987). To
𝜆𝜆𝑡𝑡 differential equation generate this range 0𝑡𝑡of shapes, a parsimonious model
𝑡𝑡 introduced by
From
t
near
ond-order
rential 𝜆𝜆In𝑡𝑡equation
Equation differential
ordinary
(7), the equation
parameter 𝑥𝑥 ′′
+𝜆𝜆𝑡𝑡 is𝑥𝑥equation ′
referred
second-order+𝜆𝜆𝑡𝑡 is2to 𝑥𝑥 as=thelinear
specified, shape theInmodel parameter. becomes shape For each
linear 𝜏𝜏 5t,in−𝜏𝜏/𝜆𝜆
parameter. when the isFor
𝜆𝜆𝑡𝑡𝑡𝑡𝑡𝑡𝜆𝜆parameters
specified,each Equation t,the𝛽𝛽when model
,= 𝛽𝛽the (3),
lt,becomes
is and specified,𝛽𝛽𝛽𝛽(4) .𝑒𝑒Let
linear the
in𝑌𝑌 𝑡𝑡 parameters
𝑡𝑡the =For 𝑦𝑦 (𝜏𝜏), 𝜏𝜏 −𝜏𝜏 𝜏𝜏0
t,𝛽𝛽w
+ordinary 𝛽𝛽Equation 𝜆𝜆differential
(7), the parameter 𝑒𝑒equation 𝑡𝑡 is referred to0𝑡𝑡𝑓𝑓 as shape parameter. ,each
TERM STRUCTURE FORECASTING
stants and second-order
second-order 𝜆𝜆where ≠ 0. Note linear
linear that 2 ordinary
𝑥𝑥(𝜏𝜏)
ordinary 𝜆𝜆
= 1 𝑡𝑡 differential
𝑓𝑓 From
differential(𝜏𝜏) 𝜆𝜆𝛽𝛽is𝑓𝑓Equation
0𝑡𝑡 a(𝜏𝜏) solution =𝜆𝜆𝑡𝑡𝛽𝛽20𝑡𝑡
equation
equation (3), to the 𝑒𝑒 −𝜏𝜏/𝜆𝜆 𝑡𝑡 is
𝑡𝑡model +specified,
𝛽𝛽2𝑡𝑡 becomes (5) thelinear model in becomes
the 𝑡𝑡 (𝜏𝜏)
𝑓𝑓parameters 𝑡𝑡 (𝜏𝜏)
linear 𝛽𝛽0𝑡𝑡
1𝑡𝑡 = in
b +
𝛽𝛽the,0𝑡𝑡b
2𝑡𝑡+
1𝑡𝑡 and 𝛽𝛽−𝜏𝜏/𝜆𝜆
parameters 1𝑡𝑡b 𝑒𝑒𝜏𝜏.−𝜏𝜏/𝜆𝜆 + 𝛽𝛽
𝛽𝛽𝑡𝑡 𝑡𝑡+
2𝑡𝑡 𝛽𝛽
𝛽𝛽2𝑡𝑡1𝑡𝑡𝑒𝑒, a
𝑡𝑡where 𝛽𝛽and ′′ ,Siegel
,+𝛽𝛽𝛽𝛽1𝑡𝑡 1𝑡𝑡, ,𝑥𝑥2 𝛽𝛽𝛽𝛽2𝑡𝑡
′
2𝑡𝑡, ,𝑓𝑓 and
and Nelson 𝜆𝜆𝜆𝜆the 𝑡𝑡 are
𝑡𝑡and 𝑡𝑡
constants Siegel assumes
and 2 (5) 1𝑡𝑡
𝜆𝜆𝜆𝜆′𝑡𝑡𝑡𝑡equation ≠≠that 0.10.Note the forward that
𝛽𝛽 𝜆𝜆𝑡𝑡 𝑥𝑥(𝜏𝜏) rate = follows𝑓𝑓 𝑡𝑡(𝜏𝜏) isisthe aasolution equation to t the (5) 0𝑡𝑡
𝜆𝜆 𝜆𝜆𝑡𝑡
aare constants and Note that solution to the
Nelson 𝑥𝑥
𝛽𝛽0𝑡𝑡0𝑡𝑡 assumes + that
2 𝑥𝑥 𝑡𝑡= forward rate follows the 𝑥𝑥(𝜏𝜏) (𝜏𝜏)
(5)𝑡𝑡𝑡𝑡 = 𝑓𝑓𝑡𝑡1−𝑒𝑒
𝑡𝑡 0𝑡𝑡
𝜆𝜆ants 2is specified,and 𝜆𝜆1𝑡𝑡 ≠the 0. model Note
𝛽𝛽 that ′′ 𝜆𝜆𝑥𝑥(𝜏𝜏)
2 = ′ 𝜆𝜆𝑡𝑡𝑡𝑡1 (𝜏𝜏) 1the is 𝜆𝜆 𝛽𝛽
2𝛽𝛽
solution 𝑥𝑥 to
′′ the
(5) 𝑡𝑡 + 𝑥𝑥 + 𝑥𝑥 = 𝑡𝑡 −𝜏𝜏/𝜆𝜆𝑡𝑡𝑡𝑡
0−𝜏𝜏/𝜆𝜆 1t 2t
1 𝜏𝜏 𝑡𝑡
𝑡𝑡Note ′ that 2 2𝑥𝑥(𝜏𝜏)
(Nelson & =1 𝑓𝑓𝑡𝑡 𝑥𝑥 ′′′′𝑥𝑥+
0𝑡𝑡
Siegel, (𝜏𝜏) becomes
1987). is
𝛽𝛽0𝑡𝑡 + a solution
𝑡𝑡 To linear
′′𝑥𝑥 ′′+ to2the
generate in this 2 range parameters
𝑥𝑥𝑥𝑥1𝜏𝜏′=+
𝑡𝑡 𝑡𝑡 of
0𝑡𝑡
0𝑡𝑡 1
0𝑡𝑡
1−𝑒𝑒
shapes,
−𝜏𝜏/𝜆𝜆
𝜆𝜆 𝛽𝛽 is
a 0𝑡𝑡 , 𝛽𝛽
parsimonious 𝛽𝛽
specified,
0𝑡𝑡
𝜆𝜆1𝑡𝑡 , and
(5)
𝑡𝑡 𝛽𝛽
the 1−𝑒𝑒
model
2𝑡𝑡 . 2 Let
model −𝜏𝜏/𝜆𝜆 𝑡𝑡 𝜏𝜏 =2𝑦𝑦𝑡𝑡 (𝜏𝜏),
𝑌𝑌
introduced becomes −𝜏𝜏/𝜆𝜆by (5) ,linear
𝑡𝑡𝑡𝑡 𝑡𝑡 .2Then𝑋𝑋 = in
1−𝑒𝑒−𝜏𝜏/𝜆𝜆
the
1 (5) parameters
, and
and 𝛽𝛽 𝑋𝑋 𝑡𝑡𝑡𝑡
= 𝛽𝛽
1−𝑒𝑒−𝜏𝜏/𝜆𝜆
1−𝑒𝑒
, 𝛽𝛽
𝑡𝑡𝑡𝑡
,− and 𝑒𝑒 −𝜏𝜏/𝜆𝜆𝑡𝑡𝑡𝑡
−𝜏𝜏/𝜆𝜆
𝛽𝛽 . . ThenLet 𝑌𝑌 𝑡𝑡
Equation = 𝑦𝑦𝑡𝑡𝑥𝑥(𝜏𝜏′′(
quation
ntial 𝑥𝑥 + (3), 𝑥𝑥 = 𝜆𝜆𝑥𝑥 + 𝑥𝑥 + 𝑥𝑥 𝑋𝑋
= = 2 𝑥𝑥 𝑡𝑡 , and
= 𝑡𝑡 𝑋𝑋 𝜏𝜏 = 𝜆𝜆 −
1−𝑒𝑒𝜆𝜆 𝑒𝑒 −𝜏𝜏/𝜆𝜆
−𝜏𝜏/𝜆𝜆 1𝜏𝜏
1𝜏𝜏 Equation
(4)𝜏𝜏/𝜆𝜆 1−𝑒𝑒 −𝜏𝜏/𝜆𝜆
−𝜏𝜏/𝜆𝜆
(7) 𝑡𝑡𝑡𝑡 2𝜏𝜏
𝜏𝜏0𝑡𝑡 becomes
2𝜏𝜏 𝜏𝜏/𝜆𝜆𝑡𝑡𝑡𝑡1𝑡𝑡 𝑦𝑦 (𝜏𝜏)
0𝑡𝑡𝜏𝜏/𝜆𝜆 2𝑡𝑡
= ∫ 𝜏𝜏
[𝛽𝛽 0𝑡𝑡 +
𝜆𝜆𝑡𝑡𝑥𝑥 equation
𝑡𝑡 𝑡𝑡 𝜏𝜏/𝜆𝜆 −𝜏𝜏/𝜆𝜆
′′
+ −𝜏𝜏/𝜆𝜆 𝑥𝑥 ′
𝜆𝜆𝑡𝑡 𝑡𝑡second-order +
From 2
𝜆𝜆𝑡𝑡2Equation 𝑥𝑥 = ,linear 2(3), ordinary 𝜆𝜆
2 (𝜏𝜏) differential 2𝜆𝜆 𝜏𝜏/𝜆𝜆2 2 −𝜏𝜏/𝜆𝜆
equation 1 1 2𝜏𝜏
2𝑡𝑡 𝜆𝜆 where
−𝜏𝜏/𝜆𝜆 𝛽𝛽
𝑡𝑡
𝑋𝑋 𝛽𝛽
𝜏𝜏/𝜆𝜆𝑡𝑡 𝑡𝑡=
Then 1 𝜏𝜏 𝑡𝑡
Equation
𝑥𝑥 ′′
+ , and (7) 𝑥𝑥 𝑋𝑋
′ becomes
+ = 𝑡𝑡𝑡𝑡
𝑥𝑥 = − 𝑒𝑒
𝑢𝑢 (5)
(5) 𝑡𝑡
𝑡𝑡 . Then Equation
𝑡𝑡 (7) becom
2𝑡𝑡, ,−𝜏𝜏/𝜆𝜆 and are 𝑡𝑡 aconstants 𝑡𝑡 and 𝜆𝜆𝑡𝑡 ≠𝜆𝜆𝑡𝑡0.≠Note that t𝑥𝑥
2 2 where
where
second-order 𝛽𝛽assumes 𝜆𝜆𝛽𝛽
linear 𝑡𝑡
, 𝛽𝛽 ordinary
2𝑡𝑡𝜆𝜆forward
𝑓𝑓
,𝑡𝑡𝑡𝑡and2𝑡𝑡 𝑡𝑡𝜆𝜆 𝜆𝜆 = 𝛽𝛽 are
differential
𝑡𝑡 𝜆𝜆𝑥𝑥
2 +
𝑡𝑡 ′′𝜆𝜆+
0𝑡𝑡′′ 𝛽𝛽
𝑡𝑡constants1𝑡𝑡 𝑒𝑒
𝑡𝑡 𝑡𝑡 +
′equation
′𝜆𝜆 𝑡𝑡 and
𝛽𝛽 𝑒𝑒
=≠ 𝑓𝑓𝑡𝑡 (𝜏𝜏) 0.0𝑡𝑡 Note
0𝑡𝑡
1𝜏𝜏 𝛽𝛽
=0𝑡𝑡𝛽𝛽𝜏𝜏/𝜆𝜆 , 𝛽𝛽that𝛽𝛽𝑡𝑡0𝑡𝑡
+, ,+ 𝛽𝛽𝛽𝛽1𝑡𝑡𝛽𝛽
𝑥𝑥(𝜏𝜏)
𝛽𝛽 𝛽𝛽=
2𝜏𝜏 𝑓𝑓,𝑡𝑡𝑡𝑡and
𝜆𝜆(𝜏𝜏) is
𝜆𝜆𝑡𝑡𝑡𝑡𝛽𝛽 are solution constants
−𝜏𝜏/𝜆𝜆 to the and 𝜏𝜏 0.0 Note
𝑡𝑡 the𝑥𝑥𝑥𝑥 + 𝑥𝑥𝜆𝜆𝑡𝑡𝑥𝑥𝑡𝑡𝑡𝑡(𝜏𝜏) 𝑡𝑡1𝑡𝑡 𝑒𝑒−𝜏𝜏/𝜆𝜆 +𝑡𝑡𝑡𝑡𝜏𝜏/𝜆𝜆 𝛽𝛽 𝜆𝜆𝜆𝜆2𝑡𝑡𝑡𝑡 2𝑒𝑒 𝑒𝑒 −𝑢𝑢/𝜆𝜆𝑡𝑡 ]𝑌𝑌𝜏𝜏𝜏𝜏𝑡𝑡𝑑𝑑𝑑𝑑
al𝑡𝑡 equation 1−𝑒𝑒𝜆𝜆Nelson and𝜆𝜆𝑡𝑡Siegel 0𝑡𝑡𝜆𝜆 𝑡𝑡𝑡𝑡1𝑡𝑡 𝑥𝑥𝑡𝑡follows + + 𝑡𝑡= 𝑡𝑡 1𝑡𝑡 −𝑢𝑢/𝜆𝜆
2𝑡𝑡 2
0𝑡𝑡𝑡𝑡 𝜆𝜆
𝑡𝑡 rate
𝑡𝑡 that the equation 22𝑦𝑦 = 2𝑡𝑡2 ∫1−𝑒𝑒 [𝛽𝛽 1𝑡𝑡 𝑒𝑒 𝑡𝑡𝜆𝜆𝑡𝑡 +2𝑡𝑡
−𝜏𝜏/𝜆𝜆
𝑋𝑋3), 𝑡𝑡 , and 𝑡𝑡 1−𝑒𝑒 𝑡𝑡
−𝜏𝜏/𝜆𝜆 1−𝑒𝑒 −𝜏𝜏/𝜆𝜆 0𝑡𝑡 −𝜏𝜏/𝜆𝜆 𝑡𝑡 𝑡𝑡𝑡𝑡
𝑡𝑡 . Then Equation𝑡𝑡 (7) becomes
1𝜏𝜏 = 𝑋𝑋 = − 𝑒𝑒 𝜆𝜆
𝜆𝜆 𝜆𝜆 𝜆𝜆𝑡𝑡𝑡𝑡𝑡𝑡 , and𝜆𝜆𝑋𝑋𝑡𝑡𝑡𝑡2𝜏𝜏 =𝑌𝑌0𝑡𝑡 = 𝜆𝜆 𝜏𝜏 − 𝑒𝑒 𝑋𝑋 𝑡𝑡 . Then 𝜆𝜆 = 𝛽𝛽 + 𝛽𝛽 𝑋𝑋 + 𝛽𝛽2𝑡𝑡
2𝑡𝑡𝑋𝑋2
From Equation 1𝜏𝜏/𝜆𝜆, 𝑡𝑡𝛽𝛽𝜏𝜏 (3), 𝑋𝑋𝑡𝑡1𝜏𝜏
𝑡𝑡 = 𝛽𝛽2𝑡𝑡Equation 𝑡𝑡 𝑡𝑡 𝑡𝑡 (7) becomes
2𝜏𝜏 0𝑡𝑡
0𝑡𝑡 1𝑡𝑡
1𝑡𝑡 1𝜏𝜏
1𝜏𝜏
on (3), 2
𝜏𝜏/𝜆𝜆𝑡𝑡
1 𝛽𝛽 𝜏𝜏 𝑢𝑢 (5)𝜏𝜏/𝜆𝜆 𝜏𝜏 (𝜏𝜏) 𝜏𝜏/𝜆𝜆𝛽𝛽 𝑡𝑡 0𝑡𝑡 + 𝛽𝛽1𝑡𝑡(6) 1𝜏𝜏 the + From 𝑋𝑋2𝜏𝜏 Equation
𝑡𝑡 (3), 𝑡𝑡(8) 𝑡𝑡 (8)
where , and are constants and 𝑡𝑡 𝑒𝑒22 𝑒𝑒𝑥𝑥 ′𝑡𝑡 ′+𝑡𝑡 ]1second-order
Note that 𝑑𝑑𝑑𝑑𝑥𝑥 second-order 𝛽𝛽𝛽𝛽0𝑡𝑡𝑡𝑡0𝑡𝑡 linear is a linear
solution ordinary to Evaluating
ordinary differential 𝑌𝑌
differential =
𝜏𝜏the integral 𝛽𝛽 +
(5)
0𝑡𝑡equation 𝛽𝛽 equation𝑋𝑋
1𝑡𝑡at1𝜏𝜏the right, + 𝛽𝛽 𝑋𝑋
𝑡𝑡𝑡𝑡 2𝑡𝑡 2𝜏𝜏we obt
(3), 1 second-order 𝛽𝛽 = 1𝑡𝑡∫ linear
, 𝛽𝛽 =ordinary
𝜆𝜆 differential 𝜆𝜆 ≠ 0.equation 1 𝑥𝑥(𝜏𝜏) = 𝑓𝑓 (4) (5)
m ′′Equation 0𝑡𝑡 −𝑢𝑢/𝜆𝜆 −𝑢𝑢/𝜆𝜆
𝑥𝑥 ′ +
2 ′𝑥𝑥 ′ +(3), 𝑦𝑦𝑡𝑡 (𝜏𝜏)𝑥𝑥 = 𝛽𝛽 0𝑡𝑡
𝜏𝜏 0𝑡𝑡 [𝛽𝛽
2𝑡𝑡
0𝑡𝑡 +
𝑓𝑓𝑡𝑡 (𝜏𝜏) 𝑡𝑡𝛽𝛽𝛽𝛽0𝑡𝑡From 1𝑡𝑡+𝑒𝑒𝛽𝛽1𝑡𝑡 𝑒𝑒Equation
−𝜏𝜏/𝜆𝜆 𝑡𝑡 + 𝑡𝑡 +𝛽𝛽
𝑡𝑡 𝑢𝑢 𝑥𝑥𝑥𝑥 ′′+
2𝑡𝑡 (5)
𝛽𝛽2𝑡𝑡
′′ (3),
−𝜏𝜏/𝜆𝜆
== Specific value of 𝜏𝜏 gives specific values of 𝑋𝑋1𝜏𝜏 1𝜏𝜏 and 𝑋𝑋2𝜏𝜏
𝑡𝑡𝑡𝑡
2𝜏𝜏.
+ 𝛽𝛽0𝑡𝑡 𝜆𝜆𝑥𝑥𝑡𝑡From + Equation
𝜆𝜆 𝑥𝑥21
= 𝜆𝜆 𝜏𝜏
(3),2
𝑌𝑌 𝑡𝑡
= 𝛽𝛽 where + (5)
Evaluating 𝛽𝛽 𝑋𝑋𝛽𝛽 + , 𝛽𝛽
𝛽𝛽 𝑋𝑋,
the integral +
𝜆𝜆𝑡𝑡𝜆𝜆
𝛽𝛽 , 𝑥𝑥and(6) + 𝜆𝜆
at 𝑡𝑡the right, are 𝑥𝑥 constants (6) and(8) 𝜆𝜆 ≠ 0. Note that 𝑥𝑥(𝜏𝜏) = 𝑓𝑓 (𝜏𝜏) is yield curve𝜏𝜏 to
a solution
From 2Equation (3),+ 𝛽𝛽ordinary 𝜆𝜆𝑡𝑡𝑡𝑡we 𝑢𝑢obtain the Nelson-Siegel
+2𝜏𝜏𝛽𝛽. 1𝑡𝑡 𝑋𝑋1𝜏𝜏 + 𝛽𝛽2𝑡𝑡 𝑋𝑋model for the
𝑡𝑡 2 2
0 𝜏𝜏 𝑡𝑡
𝜆𝜆]𝜆𝜆𝑡𝑡𝑑𝑑𝑑𝑑 2 2 𝑡𝑡 𝑡𝑡 𝑡𝑡
1𝑒𝑒value of 𝑌𝑌−𝑢𝑢/𝜆𝜆𝑋𝑋𝜏𝜏1𝜏𝜏=and 𝛽𝛽0𝑡𝑡𝑋𝑋 𝑡𝑡 (6)
−𝑢𝑢/𝜆𝜆 −𝑢𝑢/𝜆𝜆 𝑡𝑡 𝑡𝑡
𝑡𝑡𝜏𝜏 gives𝜆𝜆𝜆𝜆specific 𝜆𝜆values
(𝜏𝜏)𝜆𝜆𝑡𝑡= 𝑡𝑡 +Specific 0𝑡𝑡 1𝑡𝑡 𝑡𝑡of 2𝑡𝑡 𝑡𝑡 𝑡𝑡 𝑡𝑡
= 𝜆𝜆𝑡𝑡 2𝑦𝑦𝑡𝑡where
𝑡𝑡 ∫ [𝛽𝛽
1𝑡𝑡 𝑒𝑒𝜏𝜏 𝛽𝛽2𝑡𝑡 equation
𝜏𝜏 0𝑡𝑡 1𝑡𝑡 1𝜏𝜏 2𝑡𝑡 2𝜏𝜏
𝜆𝜆𝑢𝑢𝜏𝜏𝜏𝜏𝑡𝑡 2𝜏𝜏𝑡𝑡0𝑡𝑡linear
𝛽𝛽10𝑡𝑡∫
second-order 1 differential 𝑢𝑢 𝑡𝑡 Specific
−𝑢𝑢/𝜆𝜆 Specific value
value of (6)
of 𝜏𝜏 t gives
gives specific
specific values 2𝜏𝜏
of
of 𝑋𝑋 and
and 𝑋𝑋 𝑡𝑡
.
.
(𝜏𝜏) 𝛽𝛽0𝑡𝑡1 𝜏𝜏, 𝛽𝛽01𝑡𝑡 , 𝛽𝛽2𝑡𝑡−𝑢𝑢/𝜆𝜆 , and 𝜆𝜆1𝑡𝑡 are constants −𝑢𝑢/𝜆𝜆 𝑦𝑦 (𝜏𝜏)
𝑡𝑡𝑡𝑡 and 𝑡𝑡 +𝜆𝜆𝑡𝑡𝛽𝛽≠𝑡𝑡𝑢𝑢 = 𝜆𝜆 𝑡𝑡0. Note∫ [𝛽𝛽
(6) that
−𝑢𝑢/𝜆𝜆 0𝑡𝑡 𝑥𝑥(𝜏𝜏) +𝑡𝑡2 𝛽𝛽
𝑢𝑢𝑑𝑑𝑑𝑑 =
1𝑡𝑡 𝑒𝑒
𝑓𝑓𝑡𝑡 (𝜏𝜏)1is a solution 𝑡𝑡 + 𝛽𝛽 𝛽𝛽𝜏𝜏 0𝑡𝑡to the
2𝑡𝑡 (6)𝑒𝑒 𝑡𝑡 ] 𝑑𝑑𝑑𝑑In practice,
(6) a record of 2 (5)′2 ′𝑦𝑦1𝑡𝑡 (𝜏𝜏)
daily
1𝜏𝜏 datasets 2𝜏𝜏 is 1 collected,
= 1𝛽𝛽each − [𝛽𝛽
+
𝜏𝜏 𝜆𝜆 −𝑢𝑢/𝜆𝜆
𝛽𝛽 𝑒𝑒 𝑦𝑦
(𝜏𝜏) + =
𝛽𝛽 ∫ 𝑒𝑒
(𝜏𝜏) [𝛽𝛽 𝑢𝑢 +] 𝑑𝑑𝑑𝑑𝛽𝛽 𝑒𝑒
−𝑢𝑢/𝜆𝜆
−𝑢𝑢/𝜆𝜆 𝜏𝜏
−𝑢𝑢/𝜆𝜆 𝑒𝑒
−𝑢𝑢/𝜆𝜆
𝑡𝑡 𝑥𝑥+ 𝛽𝛽
−𝑢𝑢/𝜆𝜆 ′′
+ ] 𝑥𝑥 (6) ′ −𝑢𝑢/𝜆𝜆
+ 𝑥𝑥 1 =
In practice, 𝜆𝜆 a record of daily datasets 𝑢𝑢
𝑥𝑥 ′′
+ 𝑥𝑥 ′′
is + 𝑥𝑥
collected, + 𝑥𝑥 + 𝑥𝑥 = 𝑥𝑥 =
0𝑡𝑡
ng
Specific
∫ [𝛽𝛽0𝑡𝑡 + the 𝑡𝑡
1𝑡𝑡 integral
𝑦𝑦
value 𝑡𝑡𝑡𝑡 𝛽𝛽 𝑒𝑒
𝑡𝑡 of
𝑡𝑡
= at
𝜏𝜏 gives
−𝑢𝑢/𝜆𝜆 the
2𝑡𝑡 ∫ 𝜏𝜏𝑦𝑦 𝑡𝑡𝑡𝑡𝑡𝑡+ right,
[𝛽𝛽
specific 𝛽𝛽
0𝑡𝑡 = +
0𝑡𝑡 we 𝑡𝑡
𝛽𝛽
values∫𝑒𝑒
1𝑡𝑡obtain 𝑒𝑒
−𝑢𝑢/𝜆𝜆1𝑡𝑡[𝛽𝛽 of 𝑋𝑋
second-order
𝑡𝑡 ] 𝑑𝑑𝑑𝑑 the
𝑡𝑡+
𝑡𝑡
1𝜏𝜏 1+1 and
𝛽𝛽 Nelson-Siegel
𝛽𝛽 𝜏𝜏𝜏𝜏
2𝑡𝑡𝑒𝑒 𝑋𝑋 2𝑡𝑡
2𝜏𝜏In . 𝜆𝜆𝑒𝑒0 linear
Specific
practice,
𝑡𝑡𝑡𝑡] 𝑑𝑑𝑑𝑑 model
ordinary
value
a 𝑒𝑒
record 0𝑡𝑡 (𝜏𝜏) = of for 𝜏𝜏 𝑡𝑡 ]the
differential
of gives 𝑑𝑑𝑑𝑑𝑢𝑢
daily𝑢𝑢 yieldspecific
datasets curve
equation
𝑡𝑡 values is−𝜏𝜏/𝜆𝜆 of
collected, 𝑋𝑋
−𝑢𝑢/𝜆𝜆 𝑡𝑡
and each 𝑋𝑋 𝑡𝑡
.
dataset (6)
𝑒𝑒(6) −𝑢𝑢/𝜆𝜆
−𝜏𝜏/𝜆𝜆 𝑦𝑦
indexed (𝜏𝜏) = by 𝛽𝛽 a 2 +
particular 𝛽𝛽 𝜏𝜏 ( 2
𝑡𝑡 𝑡𝑡 𝑦𝑦 ∫In [𝛽𝛽 −+ 𝛽𝛽a1𝑡𝑡number 𝑒𝑒 +daily 𝛽𝛽indicating 𝑒𝑒 the𝜆𝜆isinterest ]collected,
𝑑𝑑𝑑𝑑 2
second-order From
From
1𝑡𝑡 𝜏𝜏 Equation 𝜆𝜆 0 2𝑡𝑡 𝜆𝜆(3),
Equation
00 𝑡𝑡 ordinary
linear
0𝑡𝑡
𝜏𝜏(3), 1𝑡𝑡
0𝑡𝑡 (𝜏𝜏)
differential
𝑦𝑦𝑦𝑦 (𝜏𝜏)
0𝑡𝑡
𝑡𝑡
= = equation ∫ ∫
1𝑡𝑡 2𝑡𝑡
𝑥𝑥 ′′ [𝛽𝛽 +
[𝛽𝛽
2
𝜆𝜆𝑡𝑡𝑡𝑡 𝑥𝑥+ 𝑡𝑡
+
′
+𝛽𝛽 𝛽𝛽
1 2𝑡𝑡
𝑒𝑒 𝑒𝑒𝑥𝑥𝜆𝜆
−𝑢𝑢/𝜆𝜆𝜆𝜆=
−𝑢𝑢/𝜆𝜆
𝛽𝛽
𝑡𝑡𝑡𝑡𝑡𝑡+ 𝜆𝜆
+ 𝛽𝛽𝛽𝛽
2
𝑡𝑡2𝑡𝑡 each 𝜏𝜏 𝑒𝑒𝑒𝑒 𝜆𝜆−𝑢𝑢/𝜆𝜆
−𝑢𝑢/𝜆𝜆
dataset
2 1𝑡𝑡0𝑡𝑡
𝑡𝑡practice, 𝑡𝑡]]𝑑𝑑𝑑𝑑
𝑒𝑒 day
indexed
𝑑𝑑𝑑𝑑
𝑡𝑡
record
(5)
by
1𝜏𝜏 𝑡𝑡
of
taand particular
12𝑡𝑡−
2𝜏𝜏
datasets
𝜆𝜆 day
𝑡𝑡
𝑡𝑡 𝑡𝑡
𝑡𝑡number 𝜆𝜆𝑡𝑡−𝜏𝜏/𝜆𝜆 rates
0𝑡𝑡
𝜆𝜆𝑡𝑡𝑡𝑡t 𝑦𝑦𝜆𝜆each (𝜏𝜏) 𝜆𝜆
𝑡𝑡𝑡𝑡𝑡𝑡
1𝑡𝑡
dataset
for 𝜆𝜆𝑡𝑡𝜏𝜏2
0
differ
𝑡𝑡 differe
ntegralInatpractice,
0 the right, we obtain 𝑡𝑡 the 𝑡𝑡 Nelson-Siegel 𝜆𝜆𝑡𝑡 model
0𝑡𝑡
0𝑡𝑡 1𝑡𝑡 1𝑡𝑡 𝑦𝑦
𝑡𝑡 for
2
𝑡𝑡 the(𝜏𝜏) =
𝑡𝑡 yield
2 𝛽𝛽 0𝑡𝑡 curve
2𝑡𝑡 + 𝛽𝛽 0 ( ) + 𝛽𝛽 ( 𝑡𝑡 − 𝑒𝑒 )
is𝜏𝜏𝑡𝑡𝜏𝜏the 𝜆𝜆dataset 𝜆𝜆indexed 𝜆𝜆𝜆𝜆𝑡𝑡𝑡𝑡aindicating
1𝑡𝑡 2𝑡𝑡
a record ofthe
Evaluating
Evaluating daily
theintegral datasets
integral at at collected,
the 00 right, each
1 𝜏𝜏𝜏𝜏 we we obtain
obtain In the
practice,𝑡𝑡the Nelson-Siegel aand by
record particular
of (5) daily 𝜏𝜏/𝜆𝜆
model the
datasets for
interest isthe collected, yield
rates 𝜏𝜏/𝜆𝜆
curve
ymaturities
each (t) for
𝑡𝑡 dataset different
e
tain integral the at
Nelson-Siegel the right, we model 𝑢𝑢 obtain 1 for − the 𝑒𝑒
the
−𝜏𝜏/𝜆𝜆
Nelson-Siegel
yield 2day number
curve
1 1 −
𝛽𝛽tmodel
and 𝑒𝑒 −𝜏𝜏/𝜆𝜆
(6)
indicating for the yield
the interest curve 𝑢𝑢 rates 2 𝑦𝑦 𝑡𝑡 (𝜏𝜏)
parameters (7) 1for Evaluating
different 𝛽𝛽1𝑡𝑡1𝑡𝑡0𝑡𝑡, and t the
2𝑡𝑡 in(6)
integral
Equation 𝜏𝜏. indexed
For(8) at
each the
are
byt, right,
athe
estimated
particula we us
t, ntegral
luating
integral
[𝛽𝛽 we +obtain at
the
𝛽𝛽 𝑦𝑦 theintegral
𝑒𝑒(𝜏𝜏)−𝑢𝑢/𝜆𝜆
the From
right, =
Nelson-Siegel 𝛽𝛽
Nelson-Siegel
+ at Equation
we𝛽𝛽 the
+ 𝑢𝑢 obtain
𝛽𝛽 right,model
𝑒𝑒 ( −𝑢𝑢/𝜆𝜆 (3),
the we
model for
] Nelson-Siegel
obtain
𝑥𝑥
𝑑𝑑𝑑𝑑
′′
the +
for )
yield𝑥𝑥 ′
thethe
+ + 1 𝛽𝛽 Nelson-Siegel
curve
yield 𝑥𝑥 ( model
=
0𝑡𝑡
curve (6) for the model − From
yield
𝑒𝑒 day
−𝜏𝜏/𝜆𝜆 From
for
maturitiesEquation
curve
number
𝑡𝑡 ) the
𝑥𝑥 Equation
′′𝑢𝑢 yield
+ t
t. and (3),
For 𝑥𝑥curve each (3),
indicating
′ 𝑡𝑡
+ The
t, the 𝑥𝑥 𝛽𝛽 0𝑡𝑡, 𝛽𝛽interest
the
parameters
=parameters
0𝑡𝑡 𝛽𝛽rates
𝛽𝛽
2𝑡𝑡 (6)
, 𝑦𝑦𝛽𝛽 𝑡𝑡 (𝜏𝜏) , for
and different
𝛽𝛽 are matur
calle
𝛽𝛽 +
0𝑡𝑡 𝛽𝛽 𝑒𝑒
1𝑡𝑡 −𝑢𝑢/𝜆𝜆
𝑡𝑡 From 𝑡𝑡 + 𝛽𝛽
𝑡𝑡 Equation0𝑡𝑡 2𝑡𝑡
2𝑡𝑡 1integral − 𝑒𝑒 (3),
−𝑢𝑢/𝜆𝜆
1𝑡𝑡
𝜆𝜆𝑡𝑡𝑒𝑒 interest
−𝜏𝜏/𝜆𝜆 𝑡𝑡 ]𝑡𝑡𝑑𝑑𝑑𝑑𝑦𝑦
𝑡𝑡
𝑦𝑦 Evaluating
(𝜏𝜏)(𝜏𝜏) 𝜆𝜆𝑡𝑡 = = 1we 𝜆𝜆 ∫ 2
2𝑡𝑡
∫ the
[𝛽𝛽
−𝜏𝜏/𝜆𝜆[𝛽𝛽 𝜆𝜆 integral
2
+ + 𝛽𝛽𝛽𝛽 𝑒𝑒 𝑒𝑒at
−𝑢𝑢/𝜆𝜆−𝑢𝑢/𝜆𝜆 the 𝑡𝑡 right,
𝑡𝑡 + + 𝛽𝛽 𝛽𝛽 we 𝑒𝑒 obtain
𝑒𝑒 −𝑢𝑢/𝜆𝜆
(7) −𝑢𝑢/𝜆𝜆 𝑡𝑡 𝑡𝑡]]the𝑑𝑑𝑑𝑑
𝑑𝑑𝑑𝑑 Nelson-Siegel
2 2 model 0𝑡𝑡 for
1𝑡𝑡 the yield 2𝑡𝑡 curve
day 0𝑡𝑡 number t and indicating the 𝜏𝜏/𝜆𝜆 rates 𝑦𝑦𝑡𝑡 (𝜏𝜏) 𝜏𝜏− forobtain𝑒𝑒different 𝜏𝜏/𝜆𝜆 maturities For each t, the 𝜆𝜆−𝜏𝜏/𝜆𝜆 𝜆𝜆yield 𝜆𝜆for
𝑡𝑡
𝑢𝑢Evaluating
Evaluating the
the 𝜆𝜆integral atat) the
the 𝑡𝑡𝑡𝑡right,
𝑡𝑡right, (6) we 𝜏𝜏 0obtain
0𝑡𝑡
1the the 𝑒𝑒,𝑡𝑡Nelson-Siegel
1𝑡𝑡
Nelson-Siegel 𝜏𝜏. 2𝑡𝑡 model
model for the curve
(8)for 𝑡𝑡the yield curve
1𝑡𝑡 𝑡𝑡 𝑡𝑡
0𝑡𝑡number 1𝑡𝑡 t and 2𝑡𝑡
The (parameters day 1𝑡𝑡𝑡𝑡,𝑡𝑡)𝛽𝛽 and indicating are 𝜆𝜆the
1𝜆𝜆Equation 𝑒𝑒interest
𝑡𝑡 factors rates 𝑡𝑡 (𝜏𝜏) 𝑡𝑡 different maturities For each factor t, the
𝑡𝑡called
𝛽𝛽 −0𝑡𝑡 ,𝛽𝛽and
−𝜏𝜏/𝜆𝜆
−𝜏𝜏/𝜆𝜆
in𝛽𝛽2𝑡𝑡 in 𝑡𝑡− (8) are 𝑡𝑡(OLS) 𝑦𝑦and
estimated their coefficients
using, for example,
(7) areregression.
𝜏𝜏.called
+ 𝑡𝑡 (𝜏𝜏)
𝑦𝑦−𝜏𝜏/𝜆𝜆
𝛽𝛽 =𝑒𝑒𝛽𝛽 0𝑡𝑡 +
−𝑢𝑢/𝜆𝜆 𝑡𝑡 ]𝛽𝛽 𝑑𝑑𝑑𝑑1𝑡𝑡 ( −𝜏𝜏/𝜆𝜆 𝑡𝑡
1 − 𝑒𝑒 −𝜏𝜏/𝜆𝜆 +𝑡𝑡 𝛽𝛽2𝑡𝑡parameters 1 −0 𝑒𝑒𝜏𝜏𝛽𝛽−𝜏𝜏/𝜆𝜆 ,− 𝑡𝑡 𝑒𝑒
𝛽𝛽 2𝑡𝑡parameters Equation (7) are
squares
,
estimated
and in
using,
method
Equation
formultiple
𝜏𝜏for example,
𝜏𝜏 are linear
(8)
the
estimated
ordinary
using,
least 1
(7) is𝛽𝛽for 1,𝑒𝑒(ae−𝑢𝑢
0𝑡𝑡 1𝑡𝑡 𝛽𝛽 , 𝛽𝛽 𝛽𝛽
−𝜏𝜏/𝜆𝜆
− 𝑒𝑒 𝑦𝑦 𝜆𝜆(𝜏𝜏) 2𝑡𝑡 𝑡𝑡 From Equation 1 − (3), 𝑒𝑒 1(− 𝜏𝜏/𝜆𝜆 𝑒𝑒Equation
𝑡𝑡 −𝜏𝜏/𝜆𝜆
−𝜏𝜏/𝜆𝜆 𝑦𝑦 (𝜏𝜏)
1𝑡𝑡𝑡𝑡(8) − = 1
𝑒𝑒are𝛽𝛽 𝛽𝛽
−𝜏𝜏/𝜆𝜆 𝜏𝜏
+ 𝜏𝜏/𝜆𝜆
1(0𝑡𝑡−+𝑒𝑒1𝑡𝑡 1 𝛽𝛽 −𝜏𝜏/𝜆𝜆
−𝜏𝜏/𝜆𝜆 ( 1𝜏𝜏/𝜆𝜆
𝑡𝑡 −
−𝜏𝜏/𝜆𝜆
𝑒𝑒𝑒𝑒example, 𝑢𝑢 ) + 𝛽𝛽
the (
ordinary 𝑢𝑢 (7)
0𝑡𝑡 least 1𝑡𝑡 −
squares 𝑒𝑒 (6)
loadings. (7)
2𝑡𝑡 (OLS)
𝑡𝑡 ) 1 Themethod 1 factor
−𝜏𝜏/𝜆𝜆𝑡𝑡 loading
for (6) multiple
𝑦𝑦 (𝜏𝜏) = of 𝛽𝛽0𝑡𝑡 𝑢𝑢
(7) 𝑡𝑡𝛽𝛽+ 𝑡𝑡++
−𝜏𝜏/𝜆𝜆 −𝜏𝜏/𝜆𝜆 𝑡𝑡
𝑡𝑡𝑡𝑡(𝜏𝜏) 0𝑡𝑡 𝑡𝑡 𝑡𝑡
𝑡𝑡 −𝜏𝜏/𝜆𝜆 𝑡𝑡
= 𝛽𝛽 𝑡𝑡𝑡𝑡 ,+ 𝛽𝛽 𝑡𝑡 𝑦𝑦−𝜏𝜏/𝜆𝜆 ) + 𝑡𝑡 𝑒𝑒 −𝑢𝑢/𝜆𝜆 − )
𝑡𝑡−𝑢𝑢/𝜆𝜆 𝑡𝑡2𝑡𝑡 −𝜏𝜏/𝜆𝜆
parameters 1 − ) 𝑡𝑡 𝑒𝑒+ 𝛽𝛽 𝛽𝛽 , 𝛽𝛽 ( and 𝛽𝛽 1in − − 𝑒𝑒 𝑒𝑒 = )𝑦𝑦 ∫
(𝜏𝜏) [𝛽𝛽
estimated
= ∫ 𝛽𝛽 using,
[𝛽𝛽 +
for
+ 𝛽𝛽
𝛽𝛽
−𝜏𝜏/𝜆𝜆
−𝜏𝜏/𝜆𝜆 𝑒𝑒 𝑒𝑒
−𝑢𝑢/𝜆𝜆 𝑡𝑡the ]+ 𝑑𝑑𝑑𝑑
−𝜏𝜏/𝜆𝜆 ordinary
𝛽𝛽 1 𝜏𝜏/𝜆𝜆 − 𝑒𝑒 𝑒𝑒
−𝑢𝑢/𝜆𝜆
least 𝑦𝑦
𝑡𝑡 ]𝑡𝑡 𝑑𝑑𝑑𝑑 (𝜏𝜏) 𝑦𝑦 =(𝜏𝜏) 1 =∫ − 𝑒𝑒
[𝛽𝛽
∫ [𝛽𝛽+ 𝛽𝛽 +𝑡𝑡 𝑒𝑒𝛽𝛽 −𝑢𝑢/𝜆𝜆
𝑒𝑒 −𝑢𝑢/𝜆𝜆 0𝑡𝑡 𝛽𝛽 𝛽𝛽
1𝑡𝑡
(𝑡𝑡𝑡𝑡𝑡𝑡(𝜏𝜏)the
𝑦𝑦obtain 𝛽𝛽=0𝑡𝑡 𝛽𝛽 𝑦𝑦1𝑡𝑡+1𝑡𝑡(𝜏𝜏) 𝛽𝛽1𝑡𝑡 =(𝛽𝛽𝛽𝛽 + 𝛽𝛽called () for + 𝑡𝑡
𝛽𝛽 (Equation )11𝜏𝜏+ 𝛽𝛽 (method −𝑦𝑦𝑡𝑡𝑡𝑡 𝑒𝑒(𝜏𝜏) 𝜆𝜆, 𝑡𝑡)𝛽𝛽1𝑡𝑡1 − 𝑒𝑒𝛽𝛽 is()𝜆𝜆𝑡𝑡1,
𝑡𝑡 0𝑡𝑡 1𝑡𝑡 From 2𝑡𝑡 (3),
ameters
we the 0𝑡𝑡 2𝑡𝑡
Evaluating 𝑡𝑡 , +
𝛽𝛽Nelson-Siegel
,Nelson-Siegel and 2𝑡𝑡 are
the model
𝜏𝜏/𝜆𝜆
integral
𝑡𝑡
1 𝜏𝜏at
factors the 𝑒𝑒the
𝑡𝑡𝜏𝜏2𝑡𝑡 −𝜏𝜏/𝜆𝜆
0right,yield and − −
we their
curveparameters
1𝑡𝑡
𝜏𝜏/𝜆𝜆 𝑡𝑡 𝑢𝑢coefficients
−𝜏𝜏/𝜆𝜆
𝑒𝑒𝑒𝑒2𝑡𝑡 −𝜏𝜏/𝜆𝜆 0𝑡𝑡 2𝑡𝑡
𝑡𝑡𝛽𝛽1𝑡𝑡 𝑡𝑡
𝑡𝑡
are ,linear
1of and
− − called 2𝑡𝑡in
𝑒𝑒𝛽𝛽𝑒𝑒0𝑡𝑡𝛽𝛽−𝜏𝜏/𝜆𝜆
−𝜏𝜏/𝜆𝜆 𝑡𝑡
regression. 𝑡𝑡factor
Equation 𝑡𝑡 (8)Diebold
𝑡𝑡 are estimated 𝑡𝑡 and Li (2006) using,
𝜏𝜏(7)(7) 0𝑡𝑡 used for
0𝑡𝑡 1𝑡𝑡 example,
the same
𝑒𝑒 −𝜏𝜏/𝜆𝜆 1𝑡𝑡 𝑡𝑡 value the ordinary ofas 2𝑡𝑡𝜆𝜆𝑡𝑡𝑡𝑡 for
𝜆𝜆𝜏𝜏𝑡𝑡2𝑡𝑡
→a𝜆𝜆
l
0𝑡𝑡 obtain the 0𝑡𝑡Nelson-Siegel model for the yield
e𝜏𝜏/𝜆𝜆
𝛽𝛽 1𝑡𝑡obtain 0𝑡𝑡0𝑡𝑡
Evaluating ) 1𝑡𝑡 𝛽𝛽𝜏𝜏/𝜆𝜆 2𝑡𝑡 2𝑡𝑡 ( 0𝑡𝑡model
𝑡𝑡the 𝜏𝜏/𝜆𝜆 integral 1𝑡𝑡𝑡𝑡for the
−𝜏𝜏/𝜆𝜆
at loadings.
the yield
2𝑡𝑡
𝑡𝑡 right,
squares 𝑡𝑡𝑡𝑡curve
) 𝜏𝜏/𝜆𝜆 Thewe
(OLS) factor
obtain 𝑡𝑡 𝜏𝜏/𝜆𝜆 loading
the for= 𝛽𝛽multiple
𝑡𝑡 Nelson-Siegel 0𝑡𝑡
squares
+ 1𝑡𝑡 linear
2𝑡𝑡
(OLS) (6) model
𝑡𝑡− amethod
constant
regression.−𝜏𝜏/𝜆𝜆 for )for +that
the 𝛽𝛽
multiple 2𝑡𝑡does
yield (𝜏𝜏 curve curve0not
linear 0decay
regression.
−to zero )even
𝜏𝜏/𝜆𝜆 𝑦𝑦
𝑦𝑦 (𝜏𝜏)
𝜏𝜏/𝜆𝜆
(𝜏𝜏)
(𝜏𝜏)
𝑡𝑡𝑡𝑡𝑡𝑡 factors0𝑡𝑡
𝑡𝑡𝑡𝑡== = 𝛽𝛽𝛽𝛽 + + 𝛽𝛽𝛽𝛽 ( (
−𝑢𝑢/𝜆𝜆 𝑡𝑡 +𝑡𝑡 ) )
−𝑢𝑢/𝜆𝜆 + + 𝑡𝑡 ] 𝛽𝛽 𝛽𝛽 (( − 𝑒𝑒
𝜏𝜏/𝜆𝜆 𝑒𝑒 −𝜏𝜏/𝜆𝜆 𝑡𝑡𝑡𝑡)
) 𝜏𝜏/𝜆𝜆
𝛽𝛽0𝑡𝑡 , 𝛽𝛽1𝑡𝑡
on-Siegel
squares (OLS) , and 𝑡𝑡
model 𝛽𝛽2𝑡𝑡
method for arefor the called yield
multiple
𝑦𝑦
curve 𝑡𝑡
linear
∫
𝜏𝜏at the
[𝛽𝛽
and
0𝑡𝑡
0 regression.
0𝑡𝑡 their + 𝛽𝛽 1𝑡𝑡 𝑒𝑒
1𝑡𝑡 coefficients
1𝑡𝑡
𝜏𝜏/𝜆𝜆
𝜏𝜏/𝜆𝜆
𝛽𝛽 2𝑡𝑡
𝜆𝜆𝑡𝑡 (OLS)
𝑒𝑒 are
(7) called 𝑑𝑑𝑑𝑑 2𝑡𝑡2𝑡𝑡 factorDiebold and 𝑡𝑡Li (2006)
𝜏𝜏/𝜆𝜆
𝜏𝜏/𝜆𝜆 Thus, was called itused
obtained has the significantsame
based on this contribution
𝑡𝑡 value of
single value of 𝜆𝜆to the
Evaluating
The parameters the integral 𝛽𝛽 𝑡𝑡a , constant 𝛽𝛽 right, , and we 𝛽𝛽 obtain are squares
thecalled
𝑡𝑡𝑡𝑡Nelson-Siegel factors method model and 1 for
for 𝜏𝜏 𝑡𝑡multiple
their 𝑡𝑡the yield coefficients linear
curve 1𝑡𝑡 , 𝛽𝛽regression.
𝛽𝛽1𝑡𝑡 2𝑡𝑡 }are
𝑢𝑢 factor 𝑡𝑡𝑡𝑡 . Other
Othe
s.
ers 1 1 The
− −
𝛽𝛽 𝑒𝑒 𝑒𝑒factor
,
−𝜏𝜏/𝜆𝜆−𝜏𝜏/𝜆𝜆
𝛽𝛽 𝑡𝑡 loading
, and Evaluating𝛽𝛽 1 of
are− 1 𝛽𝛽
−
𝑒𝑒 called𝑒𝑒
−𝜏𝜏/𝜆𝜆
the is
−𝜏𝜏/𝜆𝜆 1,
integralfactors
0𝑡𝑡 1𝑡𝑡
Thus,
at and the that
their
itright, 2𝑡𝑡
has does
Diebold coefficients
significant
1we − not
(7) obtain
𝑒𝑒and (7)
decay
−𝜏𝜏/𝜆𝜆 Li 𝑡𝑡the are
(2006)
contribution to called
Evaluating zero
Nelson-Siegell used
Evaluating for1 evenfactor
− all
the to the
𝑒𝑒 as
t.
same
−𝜏𝜏/𝜆𝜆
the 𝜏𝜏
Thus,
integral
the →
𝑡𝑡value
yield
model +∞.
integrala forseries
2𝑡𝑡
of
forat 𝜆𝜆
the
any
the The
at for of
right,
the parameters
all
maturity;
yield estimatest.
right, weThus,
curve obtain
we
hence, a{
(7)(7)𝛽𝛽
series
obtain the,it 𝛽𝛽 of ,
Nelson-Siegel
isthe and
estimates
referred Nelson-Sieg 𝛽𝛽 to{𝛽𝛽are as,m ct
ed parameters
𝛽𝛽0𝑡𝑡
(are factors
, 𝛽𝛽1𝑡𝑡
0𝑡𝑡 𝑡𝑡
, and
1𝑡𝑡 and 𝛽𝛽0𝑡𝑡+
) 𝛽𝛽their
,2𝑡𝑡
𝛽𝛽 𝛽𝛽2𝑡𝑡are
2𝑡𝑡 coefficients
(, and called 0𝑡𝑡
𝛽𝛽2𝑡𝑡
𝑡𝑡
factors
are are
− called
−𝜏𝜏/𝜆𝜆 𝑒𝑒 called
and
−𝜏𝜏/𝜆𝜆 𝑡𝑡 ) factors their
𝑡𝑡 ) factor coefficients
and 1− their 𝑒𝑒model−𝜏𝜏/𝜆𝜆
are
coefficients
𝑦𝑦 𝑡𝑡
(𝜏𝜏) called = t
factor
∫ are 1Diebold−called
[𝛽𝛽 −𝜏𝜏/𝜆𝜆
𝑒𝑒 + and 𝛽𝛽 factor
𝑡𝑡
𝑒𝑒 −𝑢𝑢/𝜆𝜆
Li−𝜏𝜏/𝜆𝜆 (2006) 𝑡𝑡 𝑡𝑡
+ used
𝛽𝛽 −𝑢𝑢/𝜆𝜆
the𝑒𝑒 same ]value 𝑡𝑡 0𝑡𝑡
𝑑𝑑𝑑𝑑 1𝑡𝑡
of 𝜆𝜆𝑡𝑡 for all t. Thus 2𝑡𝑡 0𝑡𝑡
actor
0𝑡𝑡called
loading
1𝑡𝑡Evaluating
Diebold ) factors
−𝜏𝜏/𝜆𝜆of + and 𝛽𝛽 2𝑡𝑡
𝛽𝛽0𝑡𝑡the ( and
1𝑡𝑡integral
Li (2006) is 1, their at
a𝜏𝜏/𝜆𝜆constant
used thecoefficients
−
right,
𝑡𝑡𝑡𝑡,(𝜏𝜏)
𝑦𝑦𝑦𝑦the 𝑒𝑒
(𝜏𝜏) we
same ==The
that obtain 𝛽𝛽 𝛽𝛽does
value are parameters
the
++ called
1of𝛽𝛽
Nelson-Siegel
𝛽𝛽1𝑡𝑡
not 𝑒𝑒𝑡𝑡 (
𝜆𝜆1𝑡𝑡 factor
all𝛽𝛽t.0𝑡𝑡
( 𝑡𝑡factors
decay
for to , zero
Thus, 𝛽𝛽11𝑡𝑡)−𝑡𝑡for
)a, + and
the
𝑒𝑒+
even
series yield
𝛽𝛽𝛽𝛽2𝑡𝑡Li𝛽𝛽
𝑡𝑡𝜏𝜏as
of curve
((2006) 𝜏𝜏are
(0}estimates was
→ called
0𝑡𝑡
+∞. obtained{𝛽𝛽 factors
1𝑡𝑡
−−,same 𝑒𝑒𝑒𝑒 De based
−𝜏𝜏/𝜆𝜆
long-term and
𝑡𝑡 )) on their
2𝑡𝑡
𝜆𝜆this
𝜆𝜆factor. coefficients
single value aare ofcalled factor valt
𝑡𝑡 𝑡𝑡for all The factor 𝛽𝛽1𝑡𝑡aofis called
1𝜏𝜏/𝜆𝜆 𝑡𝑡 2𝑡𝑡 𝛽𝛽𝛽𝛽(7) − −𝜏𝜏/𝜆𝜆
Diebold and
−𝜏𝜏/𝜆𝜆 2𝑡𝑡 used theClaes, (7) value
𝑡𝑡
Ceuster, factor of & Zhang, 2012)
t. Thus, considered series rangeestimates of valu
− 𝑡𝑡The 𝑡𝑡𝑒𝑒 parameters ,=and are called (and their coefficients are
are called
0𝑡𝑡
−𝜏𝜏/𝜆𝜆𝑡𝑡𝛽𝛽 𝑡𝑡,,𝑦𝑦𝛽𝛽
𝜏𝜏/𝜆𝜆 The parameters
loadings. 𝜏𝜏/𝜆𝜆 𝛽𝛽The
0𝑡𝑡
0𝑡𝑡 𝑡𝑡𝛽𝛽
𝑡𝑡factor
(𝜏𝜏)
1𝑡𝑡
1𝑡𝑡 and 𝛽𝛽loading
0𝑡𝑡 +2𝑡𝑡 2𝑡𝑡
0𝑡𝑡
𝛽𝛽for are
(,of called
𝛽𝛽 is factors
)1, 𝜏𝜏/𝜆𝜆 𝜏𝜏/𝜆𝜆
+a−𝜏𝜏/𝜆𝜆 constant
𝛽𝛽𝑡𝑡2𝑡𝑡 and their that
2𝑡𝑡
− coefficients
does
𝑒𝑒 −𝜏𝜏/𝜆𝜆 𝜏𝜏/𝜆𝜆
𝜏𝜏/𝜆𝜆 not
𝑡𝑡 )
decay
0𝑡𝑡
called
to zero factor
even as 𝜏𝜏 → +∞.
has ( significant contribution
− 𝑒𝑒 ) to the yield 𝛽𝛽 any 𝛽𝛽 }maturity;
was obtained hence,
𝑡𝑡 based it is
lon . referred
this
Others single 𝑡𝑡to
(Nelson
𝑡𝑡 as
value the &of Siegel,
𝜆𝜆 . Others
loadings. 1987; (Nelson Annaert,
The &
factor Siegel, Claes,
loading 1987; Annaert,
of 𝛽𝛽 is−𝜏𝜏
a𝛽𝛽eactor factor loading of is 1, a𝛽𝛽constant −that does not 1decay 𝑡𝑡to zero even as
𝛽𝛽not 1𝑡𝑡 0𝑡𝑡 𝜏𝜏was → +∞.
1long-term 𝜏𝜏𝜏𝜏/𝜆𝜆 factor. The −𝜏𝜏/𝜆𝜆 factor 𝜏𝜏/𝜆𝜆 𝛽𝛽1𝑡𝑡 t, is called theas
−𝜏𝜏/𝜆𝜆
𝑒𝑒obtained short-term based factor because (7)
−𝜏𝜏/𝜆𝜆 its 𝑡𝑡factor
−𝜏𝜏/𝜆𝜆 loading,
to𝑡𝑡on this
grid single 1 −value 𝑒𝑒of 𝜆𝜆𝑡𝑡 .𝑡𝑡the Others 1(Nelso
1𝑡𝑡 2𝑡𝑡 𝑡𝑡
dings.2𝑡𝑡constant
loading
The𝜏𝜏/𝜆𝜆 that
factor of does 𝛽𝛽 loading is
0𝑡𝑡 1, decay a
of constant to+single
is zero 1, that a even constant
−𝜏𝜏/𝜆𝜆
𝑒𝑒 does as𝑡𝑡
not →that𝑡𝑡 +∞.
decay 1
− 𝑒𝑒−
does to 𝑒𝑒not zero decay 𝑡𝑡 𝑡𝑡𝛽𝛽
even 𝑡𝑡 )1𝑡𝑡 to as 𝛽𝛽zero 2𝑡𝑡𝜏𝜏 1}→ − (7)
even +∞. was
𝑡𝑡
𝜏𝜏 →
referred +∞. as search. 1𝑒𝑒In− this case, value of −0𝑡𝑡1
𝜆𝜆𝑒𝑒 −
is 1, a wasconstant obtained that based does on notthis decay value
to zero of even . Others as (Nelson & Siegel, De 1987;
Ceuster, Annaert, & Zhang, 2012) considered a range of →an
−𝜏𝜏/𝜆𝜆
0𝑡𝑡
ecalledificant
1𝑡𝑡 , 𝛽𝛽
calledfactors
}
factors
𝑡𝑡
2𝑡𝑡 contribution to the yield 𝑦𝑦
and
0𝑡𝑡
0𝑡𝑡
their
𝑦𝑦 𝑡𝑡 (𝜏𝜏) =
coefficients
𝛽𝛽 0𝑡𝑡
0𝑡𝑡
for
𝛽𝛽
𝑡𝑡 (𝜏𝜏)
1𝑡𝑡 (
Evaluating
any loadings.
are =
𝜏𝜏/𝜆𝜆 𝛽𝛽𝑡𝑡called
maturity;
) 𝜆𝜆
+
+1,𝑡𝑡 the 𝛽𝛽The
𝛽𝛽 2𝑡𝑡 (
(factor
integral
𝛽𝛽hence,
factor , 𝜏𝜏/𝜆𝜆𝜏𝜏 →
𝛽𝛽2𝑡𝑡𝑡𝑡 }itwas loading
+∞.is
−
at
𝑒𝑒
the
referred )
obtained + of𝛽𝛽2𝑡𝑡
right, 𝛽𝛽to
based0𝑡𝑡(
we as isobtain 1, this
on
the a constant 𝑡𝑡 (𝜏𝜏)
𝑦𝑦single
the 𝑒𝑒=value that
−𝜏𝜏/𝜆𝜆
(𝜏𝜏)
−𝑦𝑦Nelson-Siegel 𝛽𝛽
−𝜏𝜏/𝜆𝜆 = )does
of 𝑡𝑡+
𝛽𝛽𝜏𝜏𝜆𝜆0𝑡𝑡𝛽𝛽 +
𝑡𝑡 .1𝑡𝑡 not
Others 1𝑡𝑡decay
(𝛽𝛽model ((Nelson for to the zero
&) Siegel,
+yield even
)𝛽𝛽 +( 𝛽𝛽as
curve
1987; 2𝑡𝑡 ( 𝜏𝜏 Ann 𝑡𝑡
𝑡𝑡
loadings.
loadings. The and The
their
The
parameters factor
it1𝑡𝑡factorcoefficients loading
𝛽𝛽loading are ofof called 𝛽𝛽 𝛽𝛽Claes, 0𝑡𝑡isfactor
is 1, a 1𝑡𝑡 constant
aCeuster,
constant
1𝑡𝑡
factors that
𝑡𝑡that
does
does not
not decay
decay toto zero
zero 𝑡𝑡even
1−𝑒𝑒
even 0𝑡𝑡asas ,𝜏𝜏has → +∞.
→ofsignificant
factor +∞. 𝜏𝜏/𝜆𝜆 𝑡𝑡𝜏𝜏/𝜆𝜆
2𝑡𝑡
𝜏𝜏/𝜆𝜆
The Thus, parameters has significant
0𝑡𝑡𝛽𝛽
, , ,the 𝛽𝛽𝛽𝛽 1𝑡𝑡,short-term and
,𝛽𝛽,contribution
and 𝛽𝛽𝛽𝛽called are are
De called
to hence,
called the 𝜏𝜏/𝜆𝜆 yield
factors
&their Zhang, itsand
for and any
2012)
values their
theirmaturity; coefficients
𝜏𝜏/𝜆𝜆
coefficients
considered
of lt the for𝑡𝑡hence, eacha range are it
aret.𝜏𝜏/𝜆𝜆iscalled
The of referred
called valuesmethod itfactor to𝜆𝜆was𝑡𝑡as for the each
referred contribution
t.𝑡𝑡 22Thetoof method to the𝑡𝑡
The
The parameters
parameters and and 0𝑡𝑡
are are called
factors andfactors coefficients are called factor
m ignificant
eClaes, factor.
yield for The contribution
any factor
maturity; 𝛽𝛽 to is the
hence, called
𝛽𝛽 , 0𝑡𝑡 𝛽𝛽1𝑡𝑡
yield
0𝑡𝑡
it is
1𝑡𝑡
for referred 1−𝑒𝑒 any
2𝑡𝑡 0𝑡𝑡
−𝜏𝜏/𝜆𝜆 2𝑡𝑡
maturity;
2𝑡𝑡
𝑡𝑡
to as factorthe because itforis referredfactor
Claes, De loading,
to as
Ceuster, 𝛽𝛽2𝑡𝑡 &} Zhang, were Thus,
chosen 𝑡𝑡2012) based has
considered on significant
the highest a range 𝑅𝑅 contribution
. Thus, values different of 𝜆𝜆to da
onnificant
s,
d their itto the hasDe contribution
significant
Ceuster,
coefficients
The
yield and
parameters for & their Zhang,
any 𝛽𝛽 to
contribution
are0𝑡𝑡coefficients
, the
called
𝛽𝛽
maturity; 2012), yield
and considered
factor
𝛽𝛽 for
to the
are called
are
hence, any
called
Thus, yield
it amaturity;
factors
is range
factor for
,has
referred
it andof any
hasloadings. hence,
values
their
Claes,
significantmaturity;
significant
to of
coefficients
as Dethe it𝜆𝜆 The is
𝑡𝑡 contribution
Ceuster, referred
hence,
contributionare each called&asZhang, t. itTheto
grid to
factor is asreferred
method
to
search.
2012)thethethe−𝜏𝜏/𝜆𝜆 value
yield to
2𝑡𝑡
In𝑡𝑡 this
considered as for the
of case, 𝑦𝑦a𝑡𝑡 (𝜏𝜏)
any range the
maturity; atofvalue shortervalues
−𝜏𝜏/𝜆𝜆hence, of maturities
of lt𝜆𝜆and 𝑡𝑡 itforisthe eachreferred(smaller
t. The to metho valu
as
𝑡𝑡
t
.s1,The 1,a aconstant factor 𝛽𝛽 that is does called notnot the
1𝑡𝑡
decay short-term 2𝑡𝑡
toloading zero factor because its factor loading, 1 − 𝑒𝑒 1 − 𝑒𝑒
constant that does ofdecay to ofzero ,𝛽𝛽even asathe
𝑡𝑡even as 𝑡𝑡
Thus,
Thus, it1𝑡𝑡
loadings.
loadings. The has
itlong-term
has significant
significant
parameters The
The factor
factor.factor contribution
contribution
loading
The 𝛽𝛽factor
𝜏𝜏/𝜆𝜆
was
and of is to
1,to
referred
𝛽𝛽 is𝜏𝜏constant
the is →
are 𝜏𝜏1,+∞.
yield
yield
1,
called to →anot
called that
as for
+∞.
for
constant
the grid does anyany
factors
short-termsearch. not maturity;
maturity;
that
The decay and doesIn
parameters
The totheir
this
factor zerohence,
not
parametershence,
case, even
decay
coefficients
because 𝛽𝛽 it
as
the {is
itgrid 𝜏𝜏,is
value
to
𝛽𝛽 → referred
𝛽𝛽referred
its+∞.
zero
, , 𝛽𝛽 of
and
are
factor 𝜆𝜆even
even to
,called
and
𝛽𝛽 to
and as as
loading, 𝛽𝛽the
are the
the
factor set
called
are
𝜏𝜏𝜏𝜏case,
→ of
+∞. called estimates
factors factors {𝛽𝛽
and ,and
their𝛽𝛽1𝑡𝑡 their , coef
tor. The factor loadings.
factor
is loading The
called factor 𝛽𝛽 , is
loading 1, aconstant
0𝑡𝑡 of 𝛽𝛽
constant 𝛽𝛽𝛽𝛽0𝑡𝑡 is that a𝑦𝑦 doesconstant
(𝜏𝜏) not= 𝛽𝛽 that was + set does
𝛽𝛽 of
referred (estimates
not todecay as 0𝑡𝑡 to
)0𝑡𝑡 zero
+
search.
1𝑡𝑡 𝜏𝜏). 𝛽𝛽 long-term
Moreover,
1𝑡𝑡 𝑡𝑡( In 2𝑡𝑡 }as
this were
2𝑡𝑡 →
factor.
it chosen
+∞.
decreases −
the 𝑒𝑒
value
The based
−𝜏𝜏/𝜆𝜆
to )
𝑡𝑡of
factor 0 𝜆𝜆 0𝑡𝑡and
as 𝛽𝛽 𝜏𝜏 →the
is +∞ set
call
was
the referred loadings. to as 𝛽𝛽contribution
The
grid factor
search. loading Inthe thisof short-term
0𝑡𝑡
𝛽𝛽 case, is 1, 1𝑡𝑡 a
theloading, value factor
1𝑡𝑡
𝑦𝑦𝑡𝑡 of
2𝑡𝑡
0𝑡𝑡
that because
does and 𝑡𝑡
the decay
itsset its to
offactor zero 0𝑡𝑡
estimates even loading, as
1𝑡𝑡 𝜏𝜏 → +∞. 2𝑡𝑡
thas short-term
significant factor because to its the 𝜏𝜏factor
0𝑡𝑡 value of (𝜏𝜏) 𝑡𝑡at
𝜆𝜆was shorter maturities (smaller
{𝛽𝛽0𝑡𝑡 , In 𝛽𝛽 values
, 𝑡𝑡 2 case,of 𝑡𝑡
𝜏𝜏/𝜆𝜆 value of𝜏𝜏/𝜆𝜆 estimates1𝑡𝑡{𝛽𝛽0𝑡𝑡 , 𝛽𝛽1
sr.
g-term The factor
factor. 𝛽𝛽The isfactor called the
1𝑡𝑡 is
𝛽𝛽even short-term
called the factor
.short-term because factor factor
because loading,
grid its factor
search. loading,
1𝑡𝑡 referred to as 1𝑡𝑡this thedifferent 𝜆𝜆𝑡𝑡 and 𝑡𝑡 the set ofhave
does
called notthe decay decay
short-term 1𝑡𝑡 to
1𝑡𝑡 to zero zerofactor even as
because → long-term +∞.
its factor Thus, it
loading, has
factor. significantThe factor 𝛽𝛽 on the
is highest
called the Rto.short-term Thus, factor datasets because could its factor loading,
tothe theyield yield forfor Thus,
any any it
maturity; has
maturity; significant hence, hence, contribution
it to is itthe 𝜏𝜏).referred
isany Moreover, to the to yield as it
for
the decreases
any maturity; to 0
hence, as 𝜏𝜏
it →
is +∞.
referred Lastly, as the 𝛽𝛽 is referred to as the medium-term fa
isreferred to as the 1𝑡𝑡 2
nificant long-term
long-term
contribution Thus, factor.
factor.
it has to
𝑡𝑡 the The
The
significant value factor
factor of 𝛽𝛽 𝛽𝛽
𝑦𝑦
contribution (𝜏𝜏) is
𝛽𝛽 called
at }maturity;
called were
shorter the
to thechosen
the short-term
short-term
maturities yield based itfor iton isfactor
factor
(smaller
any the highest
because
maturity; because
tovalues 𝑅𝑅 The .its
Thus,
its
of
hence, factor different
it isisloading, 2𝑡𝑡 datasets could
loading,
referred have different shape
𝑡𝑡to 𝛽𝛽as
loadings. different
loadings. The shape factor parameters loading ltof . even
Furthermore, is 1, 2the aThus, the
constant aissue of
that does notcou
𝛽𝛽 } were
Thus,
chosen
contribution
loadings.
it1−𝑒𝑒
Thus, has
based
significant
−𝜏𝜏/𝜆𝜆
it has
on the
to
The the
contribution
significant
highest factor yield 𝑅𝑅 2 for
loading
contribution
.
1𝑡𝑡
𝑡𝑡1𝑡𝑡
Thus,
yield 2𝑡𝑡 of
for
different
any
𝛽𝛽 to is
maturity;thehence,
datasets 1, ayield constant
hence,
could
is
for have any 𝛽𝛽that
referred does
}maturity;
differentwere as the not
chosen
shape decay
hence, factor
based to
it on loading
zero referred
the 𝛽𝛽of
highest
−𝜏𝜏/𝜆𝜆 0𝑡𝑡 as
to as
𝑅𝑅
0𝑡𝑡𝜏𝜏 is
→.the 1, +∞. constant
different −𝜏𝜏/𝜆𝜆 that
datasets does codn
2𝑡𝑡 𝑡𝑡 , has significant The contribution
parameters
0𝑡𝑡
𝛽𝛽2𝑡𝑡 to𝛽𝛽} 0𝑡𝑡 the
were 𝛽𝛽value chosen ofbased 𝑡𝑡 (𝜏𝜏)
𝑦𝑦multicollinearity
2𝑡𝑡
itsonfactor
atcalled
the shorter highest factors maturities
2
𝑅𝑅 .because Thus, 1−𝑒𝑒 (smaller
different itsand , has
factor values
datasets significant
loading, couldof are have
1−𝑒𝑒
contribution different𝑡𝑡
factor − 𝑒𝑒 −𝜏𝜏/𝜆𝜆 shap to
ysignificant contribution to the value of 𝑦𝑦isvalue (𝜏𝜏) 𝛽𝛽𝑡𝑡2𝑡𝑡at isshorter ,maturities 1𝑡𝑡 ,its isand (smaller
𝛽𝛽 are values of was and their
considered coefficients ridge regression called
−𝜏𝜏/𝜆𝜆
over,
e maturity;
value
nificant itshort-term
decreases
of hence,
𝑦𝑦
contribution referred
long-term
(𝜏𝜏) at
𝜏𝜏/𝜆𝜆 it𝑡𝑡shorter
to is 0to
factor.
to as as
referred
the 𝜏𝜏the The
→
maturities
value long-term
+∞. toistoof
factor as 𝛽𝛽
Lastly,the (smaller factor.
called
at The
the
values
shorter factor
short-term
referred of
maturities factor
to as because
the
(smaller medium-term 𝑡𝑡 values
loading,
of factor
gnificant
ion ed the, has significant
long-term 𝑡𝑡 factor.
factor contribution
The
(𝜏𝜏) factor
because 𝛽𝛽 its called the
factor 𝑦𝑦 1𝑡𝑡
1−𝑒𝑒 the(𝜏𝜏) 𝑡𝑡−𝜏𝜏/𝜆𝜆
short-term
loading, of 𝑦𝑦 (𝜏𝜏)
factor at
because shorter factor maturities
1−𝑒𝑒
2𝑡𝑡
−𝜏𝜏/𝜆𝜆
loading, (smaller values of 𝜏𝜏/𝜆𝜆𝑡𝑡 𝜏𝜏/𝜆𝜆𝑡𝑡
alled𝜆𝜆on to the the value
short-term−𝜏𝜏/𝜆𝜆 𝑡𝑡𝑡𝑡 of 𝑦𝑦factor at because shorter maturities
its factor
𝑡𝑡𝑡𝑡 ,(smaller
loading, has 𝑡𝑡significant values of contribution to the
−𝜏𝜏/𝜆𝜆 value of 𝑦𝑦 (𝜏𝜏) at shorter maturities (smaller valu
1𝑡𝑡
1−𝑒𝑒 −𝜏𝜏/𝜆𝜆
called
long-term Thus, the it short-term
factor.
has significant The factorfactor becausebecause 𝛽𝛽𝛽𝛽 isisits its
called factor
factor loading,
loading, short-term Thus, was
Thus, factor implemented
itmaturity;
has −itbecause𝑒𝑒significant
has , starts
significant as
its a
itsfactor itatiscontribution
remedy
contribution 𝑡𝑡0, increases,
loading, (Annaert totothe then
et
to thedecays
al.,
yield 2012).for any
yield for to 0. maturity;
any More matur
𝛽𝛽2𝑡𝑡contribution to the yield for any hence, referred as the
1−𝑒𝑒 𝑡𝑡
𝑡𝑡
decreases1−𝑒𝑒 to−𝜏𝜏/𝜆𝜆long-term
01−𝑒𝑒 ,as has
,−𝜏𝜏/𝜆𝜆
has 𝜏𝜏𝑡𝑡 →
𝑡𝑡
significant
significant
+∞.factor. Lastly, The contribution
contribution factor is
𝜏𝜏/𝜆𝜆 𝑡𝑡1𝑡𝑡to
referred
1𝑡𝑡 to called
the
the tovalue
value as the the of short-term
of 𝑦𝑦𝑦𝑦𝑡𝑡𝑡𝑡(𝜏𝜏)
medium-term (𝜏𝜏) atat
𝜏𝜏/𝜆𝜆 factor
shorter
shorter
𝑡𝑡 factor because
maturities
maturities factor
(smaller
(smaller loading,
values
values ofof
m, itfactor decreases
𝜏𝜏/𝜆𝜆
because 𝜏𝜏/𝜆𝜆 𝑡𝑡𝑡𝑡 𝑡𝑡𝜏𝜏). Moreover,
, 𝜏𝜏/𝜆𝜆
has its , has has
factor loading,
significant −𝜏𝜏/𝜆𝜆 significant
significant
contribution it decreasescontribution loadings.
contribution
to the to
value to 0 as
theThe to
𝜏𝜏
value →
of 𝑦𝑦𝑡𝑡 (𝜏𝜏) at shorter the
factor +∞.
of value
𝑦𝑦 loading
(𝜏𝜏) Lastly, at
maturities of shorter of 𝛽𝛽 𝛽𝛽
maturities
(smaller is
2𝑡𝑡 0𝑡𝑡 valuesfactor referred
is 1, a
(smaller
of constantto as
values the that of
precisely, medium-term does
𝜏𝜏). Moreover, not
it approaches decay factor to
it decreases zero
0 as 𝜏𝜏 to even → 00 asand as + 𝜏𝜏 𝜏𝜏→→+ a
𝑡𝑡 to 0referred
𝑡𝑡 as 𝜏𝜏 → +∞. −𝜏𝜏/𝜆𝜆 Lastly, is referred to as the medium-term
1−𝑒𝑒 𝑡𝑡
𝛽𝛽2𝑡𝑡 𝑡𝑡
∞. otits
Moreover, the Lastly,
factor
decreases value loading,
𝛽𝛽𝜏𝜏/𝜆𝜆
it to
of2𝑡𝑡 0is
decreases
𝑦𝑦 as
(𝜏𝜏) 𝜏𝜏 at →
at to
+∞.
shorter
shorter to0 − as 𝑒𝑒
Lastly, the
𝜏𝜏 →
maturities
maturities medium-term
𝑡𝑡 , starts
+∞. 𝛽𝛽 is
Lastly,
(smaller(smaller at
referred 0, 𝛽𝛽factor
increases,
values to
values is as referred
of the of then medium-term
t). to decays as the to 0.
medium-term factor More factor
𝜏𝜏to→the +∞. value Lastly, long-term
−𝜏𝜏/𝜆𝜆
of 𝑦𝑦𝑡𝑡 ,,has 𝛽𝛽 is referred
factor. Theto as factor 2𝑡𝑡
𝜏𝜏).the medium-term
Moreover, 𝛽𝛽 is called
2𝑡𝑡 it decreasesfactor
the short-term long-term
to 0 long-term
as factor
𝜏𝜏 → factor. +∞. factor.
because The
Lastly, factor
The its factor
factor
𝛽𝛽 𝛽𝛽 is is
loading,
𝛽𝛽
referred called is called
to the as short-term
the short-term
medium-term factor factor bec fa
itat shorter maturities (smaller values of
1−𝑒𝑒 −𝜏𝜏/𝜆𝜆
(𝜏𝜏) 𝑡𝑡 2𝑡𝑡 +
0precisely, 1𝑡𝑡 itto approaches 0ofis as and as Its
1−𝑒𝑒 1−𝑒𝑒 𝑡𝑡
−𝜏𝜏/𝜆𝜆 𝑡𝑡
2𝑡𝑡 𝜏𝜏/𝜆𝜆 𝑡𝑡
significant contribution the value of 𝑦𝑦𝑦𝑦 𝜏𝜏(𝜏𝜏) → at0atshorter 𝜏𝜏 medium-term
maturities → +∞.2𝑡𝑡(smaller 1𝑡𝑡significant 1𝑡𝑡 valuescontribution of to the yield
𝑒𝑒has significant 00contribution isto the𝛽𝛽2𝑡𝑡 value shorter maturities (smaller values ofit is referred
𝑡𝑡
r loading, 𝜏𝜏).
𝜏𝜏). 𝜏𝜏).Moreover,
Moreover,Moreover,
𝜏𝜏).
Moreover, 𝜏𝜏/𝜆𝜆 Moreover, −it decreases itdecreases
it−𝜏𝜏/𝜆𝜆 decreases 𝑡𝑡 ,tostarts
itdecreases
decreases 0 as 𝜏𝜏to to atThus,
to
to
→ +∞. 0,as as 𝜏𝜏𝜏𝜏→
asincreases,
𝜏𝜏Lastly,
1−𝑒𝑒 it→ → has
+∞. +∞.
𝛽𝛽+∞.
−𝜏𝜏/𝜆𝜆 Lastly, .Lastly,
then
2𝑡𝑡𝑡𝑡significant Lastly,
Lastly,
referred decays
toisas 𝛽𝛽𝛽𝛽 2𝑡𝑡 is
referred
contribution
the
2𝑡𝑡 to referred
0.
𝑡𝑡to
medium-term (𝜏𝜏)
referred
𝑡𝑡Data More
as the and
to to
to
medium-term
the
factor as as the
Procedurethe
yield medium-term
factor
for anymedium maturity; factor
factor hence, to 𝑡𝑡as th
(𝜏𝜏) at shorter 𝜏𝜏/𝜆𝜆 𝜏𝜏/𝜆𝜆
because
maturities
1−𝑒𝑒 𝑡𝑡
𝑡𝑡−𝜏𝜏/𝜆𝜆
𝑡𝑡 𝑡𝑡 its factor
(smaller
−𝜏𝜏/𝜆𝜆 loading,
values of − 𝑒𝑒because −𝜏𝜏/𝜆𝜆 𝑡𝑡 ,decaysstarts at 0, increases, then decays the because to 0. More term
its factor to maturity.
loading,
1−𝑒𝑒 −𝜏𝜏/𝜆𝜆
− 𝑒𝑒
y, ctor itloading, loading,
approaches 1−𝑒𝑒
1−𝑒𝑒 is referred
−𝜏𝜏/𝜆𝜆
−𝜏𝜏/𝜆𝜆
0isas 𝑡𝑡𝑡𝑡 − to
1−𝑒𝑒 𝑒𝑒 as
−𝜏𝜏/𝜆𝜆
+ the 𝑡𝑡 , starts medium-term at 0, increases, factor then its to 0. More
,and as Itsatsignificant contribution 𝑡𝑡 to to the 0.yield is in
/𝜆𝜆 𝜏𝜏𝑡𝑡 𝑒𝑒→ then 0 to𝑡𝑡𝑡𝑡1−𝑒𝑒 𝜏𝜏medium-term
→ +∞.
or
ause 𝑡𝑡
𝑡𝑡+∞.
−
𝑡𝑡 , starts
𝑒𝑒itsLastly,
−𝜏𝜏/𝜆𝜆 factor at 0,
𝑡𝑡𝑡𝑡 , starts 𝛽𝛽𝜏𝜏/𝜆𝜆 increases,
loading,
1−𝑒𝑒 𝜏𝜏/𝜆𝜆
−𝜏𝜏/𝜆𝜆
at −
𝑡𝑡referred
0,
−𝜏𝜏/𝜆𝜆
−𝜏𝜏/𝜆𝜆
increases, as decays
starts −the 𝑡𝑡𝑒𝑒1−𝑒𝑒
then
−𝜏𝜏/𝜆𝜆
at to
the 0,
decays
𝑡𝑡 ,𝑡𝑡0.
increases,
starts
𝑡𝑡 ,medium
More
𝜏𝜏/𝜆𝜆 𝑡𝑡
to 0. 0,
factor
term then
More increases, to decays maturity. to
then
1−𝑒𝑒1−𝑒𝑒 −𝜏𝜏/𝜆𝜆
0. −𝜏𝜏/𝜆𝜆
1−𝑒𝑒 More
decays 𝑡𝑡 −𝜏𝜏/𝜆𝜆 𝑡𝑡
More 𝜏𝜏/𝜆𝜆𝑡𝑡
, has significant contribution to the value of , has
at , significant
has
shorter
−𝜏𝜏/𝜆𝜆 significant maturities contribution contribution
(smaller to the
values to value
the of value of 𝑦𝑦 of (𝜏𝜏) at
(𝜏𝜏) short
at s
−𝜏𝜏/𝜆𝜆
→ +∞. Lastly,
because 2𝑡𝑡
factor 𝛽𝛽
its factor
Moreover, is loading, referred
loading, itit𝜏𝜏decreases to as −
1−𝑒𝑒the 𝑒𝑒 because
−𝜏𝜏/𝜆𝜆
−𝜏𝜏/𝜆𝜆
−𝜏𝜏/𝜆𝜆
medium-term
−𝜏𝜏/𝜆𝜆
to , 00− starts
𝑡𝑡starts
as −𝜏𝜏/𝜆𝜆 its
at 𝑡𝑡at
0, factor0,
increases, factor
increases, loading,
then
Lastly, decays to 𝑦𝑦
Data
0.
is (𝜏𝜏)
More
referred − 𝑒𝑒 to
𝑡𝑡 , starts at 0, increases, then decays to 0.𝑦𝑦 More
inas asthe medium-term factor
because its factor 𝜏𝜏/𝜆𝜆 loading, 𝑡𝑡− , starts at 0, increases, then decays to 0. More
0 as𝜏𝜏).
𝑒𝑒as𝑒𝑒𝑒𝑒𝜏𝜏factor. → →𝑡𝑡+∞. 0, 𝛽𝛽
𝜏𝜏/𝜆𝜆 1−𝑒𝑒 long-term The factor
Moreover, decreases 𝑡𝑡 tosignificant Lastly, 𝑡𝑡is
is called
referred the to short-term theto medium-term factor because factorits factor loading,
𝜏𝜏). 2𝑡𝑡 𝜏𝜏−𝜏𝜏/𝜆𝜆
−𝜏𝜏/𝜆𝜆 𝑡𝑡+∞. 𝛽𝛽𝛽𝛽2𝑡𝑡 𝑡𝑡 𝑡𝑡 𝑡𝑡
roachesbecause its factor loading, ,,contribution
starts at 0, increases, 𝑡𝑡 then decays 0.
0. More
𝑡𝑡
𝑡𝑡 + 𝑡𝑡
because 𝜏𝜏then → its
𝜏𝜏/𝜆𝜆0decays 𝑡𝑡factorand to asloading, 0.
→𝜏𝜏/𝜆𝜆𝑡𝑡
More
+∞. 𝜏𝜏/𝜆𝜆
precisely,
Its − it approaches
starts 0
atas increases,
to
𝜏𝜏/𝜆𝜆𝜏𝜏/𝜆𝜆
1𝑡𝑡
2𝑡𝑡 the 𝑡𝑡𝜏𝜏/𝜆𝜆 yield then is decays to More
is referred to0to precisely,
as𝜏𝜏itthe + it approaches
medium-term 𝜏𝜏/𝜆𝜆
𝜏𝜏/𝜆𝜆 0𝑡𝑡 as 𝜏𝜏significant +
→ 0 and contribution as 𝜏𝜏 → +∞. to ItstheHistorical
significant data contributionof interest rates
to theityield were initially
is in 0 as 𝜏𝜏 → 0+ an
um term maturity. +factor
𝑡𝑡
pproaches as → +0 and as 𝜏𝜏→ → +∞. 𝜏𝜏Its yield is in precisely, approaches
𝑒𝑒as
proaches −𝜏𝜏/𝜆𝜆
isely, 𝜏𝜏 →
+−𝜏𝜏/𝜆𝜆 𝑡𝑡 , it +∞.
starts precisely,
0
approaches as Its
at 𝜏𝜏 0,
→
precisely, significant
increases,
0 +
approaches 0 and
as
and
it 𝜏𝜏as
as
approaches 0→ contribution
as
then 𝜏𝜏 𝜏𝜏→
0 +decays
0 0and
+∞.as and𝜏𝜏 .as → asto
Its
Its to
0𝜏𝜏 0.
→the
+significant
→
significant
and +∞.
More+∞. yield
as Its
𝜏𝜏 →Its is
significant
contribution
+∞. in
contribution
significant Its contribution
significant to
contribution to the
obtained the yield
contribution
𝛽𝛽𝜏𝜏Moreover, yield is to
from in
is
to
itthen in
the
thethe
decreases yield
website
yield is is
intoto in
of00. Philippine
as Dealing Lastly, and 𝛽𝛽2𝑡𝑡 is 𝛽𝛽2𝑡𝑡referred
m0𝑒𝑒toand
− 𝑡𝑡 ,as
maturity. starts 𝜏𝜏 because
→𝜏𝜏). +∞.
at 0,
TERM Moreover, Its
increases,
STRUCTURE
its significant
factor itloading,
6decreases
then
FORECASTING contribution
decays
1−𝑒𝑒 precisely,
1−𝑒𝑒−𝜏𝜏/𝜆𝜆
1−𝑒𝑒 to+to +,0and
−𝜏𝜏/𝜆𝜆
−𝜏𝜏/𝜆𝜆
𝑡𝑡
0.
has
as−
𝑡𝑡 𝑡𝑡 to
More 𝑒𝑒the
it𝜏𝜏significant
approaches
→ −𝜏𝜏/𝜆𝜆
−𝜏𝜏/𝜆𝜆+∞. yield𝑡𝑡𝑡𝑡, starts is 0in𝜏𝜏).
Lastly,
contribution as0,
at 𝜏𝜏).
2𝑡𝑡 →isMoreover,
increases, 0referred
+
to and the asvalue
to 𝜏𝜏it → asdecreases
decays
6
of
the
+∞. medium-term
(𝜏𝜏) Its to
at
0𝜏𝜏 as
significant
More
shorter
→ 𝜏𝜏+∞. →factor +∞.
contribution
maturities
Lastly,
(smaller to the is yield refe
value
precisely,
precisely,
the medium because
to the the itit
mediumyield
term to maturity.its
approaches
approaches factor
is in term the loading,
0 0
medium
to asas 𝜏𝜏
maturity.𝜏𝜏 → →
𝜏𝜏/𝜆𝜆𝑡𝑡𝜏𝜏/𝜆𝜆
0
term 0 and
to − asas
maturity. 𝑒𝑒 𝜏𝜏
𝜏𝜏 → → +∞.,
+∞. starts Its Its at 0,
significant
significant
Exchange increases, contribution
contribution
Corporation then decays or 𝑦𝑦toto
𝑡𝑡PDEx to
the
the 0. yield
yield More
(originally isis inin http://www.
erm
, increases, to maturity. then the The medium
decays three termtofactors 0. to More maturity. arearealso
𝜏𝜏/𝜆𝜆𝑡𝑡𝑡𝑡
interpreted inisterms of slope,pdex.com.ph,
the medium term to maturity.
mndto
medium asmaturity.
𝜏𝜏 →term +∞.toIts maturity.
significant The three contribution
factors thealso medium to the
interpreted
1−𝑒𝑒 −𝜏𝜏/𝜆𝜆yield 𝑡𝑡 in term terms inof𝑡𝑡maturity.
to level, and curvature ofbut the yield data now 1−𝑒𝑒 −𝜏𝜏/𝜆𝜆
available
1−𝑒𝑒𝑡𝑡−𝜏𝜏/𝜆𝜆𝑡𝑡 −𝜏𝜏/𝜆𝜆−𝜏𝜏/𝜆𝜆
thru the website
vel, and as the slope, and because
curvature Itsterm of
significantits the factor yield loading,
contribution −𝜏𝜏/𝜆𝜆
, starts because at because
0, its
increases, factorits factor loading,
then loading,
decays to 0. − More 𝑒𝑒 − as 𝑒𝑒is𝑡𝑡 ,the starts 𝑡𝑡 , starts at 0,atincreases, 0, increa
𝜏𝜏the →medium +∞.
medium level,
precisely,
precisely, slope, term to
to maturity.
and
ititapproaches
approaches curvature
maturity. 00as
𝜏𝜏). of
as the
Moreover,𝜏𝜏𝜏𝜏→ →to0𝑡𝑡0the
yield
𝜏𝜏/𝜆𝜆
+ +and −
curveityield
and 𝑒𝑒 as
decreasesas𝜏𝜏𝜏𝜏is→
(Diebold
→in+∞. +∞. to 0Its of
assignificant
Its the
𝜏𝜏 →parent
significant +∞. Lastly, company
contribution
contribution 𝛽𝛽2𝑡𝑡Philippine is𝜏𝜏/𝜆𝜆to to𝑡𝑡 the
referred the𝑡𝑡yield
𝜏𝜏/𝜆𝜆
Dealing
yield to isin Systems,
in medium-term fact
& Li, 2006). The factor b0t governs the 0𝑡𝑡level of the curve
curve (Diebold & Li, 2006). The factor 𝛽𝛽 governs the level of the http://www.pds.com.ph/
curve since lim 𝑦𝑦 (𝜏𝜏) = ). The rates for tenors at most
ts of
el significant
the curve since contribution
since lim 𝑦𝑦𝑡𝑡 (𝜏𝜏) to the = byield is in 𝜏𝜏→+∞
𝑡𝑡

and an increase in b increases one year were zero rates so0𝜏𝜏these

thethe medium
precisely,
medium
𝛽𝛽0𝑡𝑡 and an
𝜏𝜏→+∞ term itterm
increaseapproaches to toinmaturity.
maturity.
0t
increases 0 as 𝜏𝜏all→ 0+equally. and 0t asThus, 𝜏𝜏 →this precisely,
+∞.
1−𝑒𝑒 precisely,
−𝜏𝜏/𝜆𝜆𝑡𝑡
Itsis responsible it−𝜏𝜏/𝜆𝜆
significant approaches
it𝑡𝑡approaches contribution 0 as as→to 𝜏𝜏0were +
→theand +used. The data
0yield as 𝜏𝜏isas
and →in𝜏𝜏+∞. → +∞. Its significa Its signi
all yields equally. Thus, 𝛽𝛽0𝑡𝑡 because this factor itsyields isfactor responsible loading, for 𝜏𝜏/𝜆𝜆 factor for tenors longer than one year, however,then
− 𝑒𝑒 ,for starts parallel at 0, increases, weredecays not zeroto 0. More
𝑡𝑡
s, this factor is parallel responsible yieldfor curve parallel shifts. The
𝛽𝛽1𝑡𝑡 isfactor b1ttheisyield related curvetoslope. rates so the corresponding zero rates were obtained
the theyield yield medium curve
curveterm
shifts.
slope.
The tofactor Ifmaturity.
the slope of the yield curve
related to the Ifmedium the the medium slope of
term the
from+Bloomberg (2013). Note that the zero rates can
yield
term to maturity. curve
to maturity.
ve slope. If the slope of the as yield limcurve
precisely, it approaches 0 as 𝜏𝜏 → 0 and as 𝜏𝜏 → +∞. Its significant contribution to the yield
is defined
is defined as 𝜏𝜏→+∞
𝑦𝑦 (𝜏𝜏) − lim 𝑦𝑦 (𝜏𝜏), then it is equal to −𝛽𝛽 . Lastly, 𝛽𝛽 is related to the
𝑡𝑡
𝜏𝜏→0+
𝑡𝑡 , then it is equal to
1𝑡𝑡 also2𝑡𝑡be computed using a method called bootstrapping.

𝛽𝛽1𝑡𝑡 . Lastly, 𝛽𝛽2𝑡𝑡 is curvature,

related todefined the and
the by Diebold medium
Li as 2𝑦𝑦term to maturity.
𝑡𝑡 (24) − 𝑦𝑦𝑡𝑡 (3) − 𝑦𝑦𝑡𝑡 (120), where maturity is given in

months.
− 𝑦𝑦𝑡𝑡 (120), where maturity is given in
Estimation of Parameters
 Nelson
Nelson
From Equation andand
(3),
NelsonSiegel
Siegel
and assumes
assumes
Siegel that
that
assumes the
the
forward
that forward
the rate
rate
forward follows
follows
rate thethe
followsequation
equation
the equatio

1 𝜏𝜏 𝑓𝑓 (𝜏𝜏) −𝜏𝜏/𝜆𝜆 𝑢𝑢𝑡𝑡 +

−𝜏𝜏/𝜆𝜆
𝜏𝜏 𝜏𝜏−𝜏𝜏/𝜆𝜆 𝜏𝜏 𝑡𝑡 −𝜏𝜏/
𝑦𝑦𝑡𝑡 (𝜏𝜏) = ∫ [𝛽𝛽𝑡𝑡0𝑡𝑡𝑓𝑓𝑡𝑡+(𝜏𝜏)
𝑓𝑓= =
𝛽𝛽𝑒𝑒0𝑡𝑡−𝑢𝑢/𝜆𝜆
(𝜏𝜏)
𝛽𝛽𝑡𝑡1𝑡𝑡 𝛽𝛽=0𝑡𝑡
+𝛽𝛽𝛽𝛽
+ +𝛽𝛽+
1𝑡𝑡
𝑡𝑡0𝑡𝑡 𝑒𝑒1𝑡𝑡 𝑒𝑒1𝑡𝑡
𝛽𝛽
𝛽𝛽2𝑡𝑡 𝑒𝑒 −𝜏𝜏/𝜆𝜆
𝛽𝛽
𝑒𝑒193+2𝑡𝑡
𝑡𝑡𝛽𝛽
𝑡𝑡 −𝑢𝑢/𝜆𝜆
+ ]
𝑡𝑡2𝑡𝑡 𝑒𝑒2𝑡𝑡𝑒𝑒 −𝜏𝜏/𝜆𝜆
𝛽𝛽𝑑𝑑𝑑𝑑 𝑒𝑒 𝑡𝑡
Philippine Interest Rates Using the Dynamic Nelson-Siegel Model 𝜆𝜆 𝜆𝜆 𝜆𝜆
𝜏𝜏 0 𝜆𝜆𝑡𝑡 𝑡𝑡 𝑡𝑡 𝑡𝑡

Evaluating where where where𝛽𝛽0𝑡𝑡𝛽𝛽,0𝑡𝑡𝛽𝛽𝛽𝛽 , 1𝑡𝑡𝛽𝛽,1𝑡𝑡 𝛽𝛽𝛽𝛽

,the ,2𝑡𝑡
, 2𝑡𝑡1𝑡𝑡
𝛽𝛽 , and, and
𝛽𝛽 𝜆𝜆are
𝜆𝜆, 𝑡𝑡and 𝑡𝑡 are𝜆𝜆constants
𝑡𝑡 areconstantsconstants andand 𝜆𝜆𝑡𝑡and 𝜆𝜆≠𝑡𝑡 ≠ 0.
𝜆𝜆𝑡𝑡 0.Note
≠ Note 0. that
Note that𝑥𝑥(𝜏𝜏) 𝑥𝑥(𝜏𝜏)
that =𝑥𝑥(
PDEx is one of the two major exchanges in the the integral the other at
0𝑡𝑡
method right, 2𝑡𝑡we
that obtain
used a thefixed Nelson-Siegel
shape parameter model for the yiel
Philippines, the other being the Philippine Stock second-order across all
second-order linear datasets,
linear ordinary which
ordinary will
differential be referred
differential to for the rest
second-order linear ordinary 1 − 𝑒𝑒 −𝜏𝜏/𝜆𝜆 𝑡𝑡 equation
differential equationequation − 𝑒𝑒 −𝜏𝜏/𝜆𝜆
To(1
𝑡𝑡
Exchange (PSE). TERMPDEx STRUCTURE is a venue for trading fixed-
FORECASTING of𝑦𝑦the (𝜏𝜏) paper= 𝛽𝛽 as+fixed 𝛽𝛽 ( lambda method. ) + 𝛽𝛽 determine 5 − 𝑒𝑒 −𝜏𝜏/𝜆𝜆𝑡𝑡 )
𝑡𝑡 0𝑡𝑡 1𝑡𝑡 2𝑡𝑡
income and other securities, most of which are this fixed shape parameter, 𝜏𝜏/𝜆𝜆first, for2each lambda 𝜏𝜏/𝜆𝜆
2 ′ 1′ 1 1 𝛽𝛽0𝑡𝑡𝛽𝛽0𝑡𝑡𝛽𝛽0𝑡𝑡 in
′′ ′′ ′ 2
𝑡𝑡 𝑡𝑡
government securities. From daily trading, PDEx the same range used in grid search, 𝑥𝑥 ′′𝑥𝑥+ 𝑥𝑥+beta 𝑥𝑥+𝑥𝑥+parameters+ 𝑥𝑥 + 𝑥𝑥 = 𝑥𝑥 2=𝑥𝑥2=2 2
(Nelson & Siegel, 1987). To generate this range of shapes, a parsimonious model 𝜆𝜆𝑡𝑡 𝜆𝜆𝑡𝑡𝜆𝜆𝑡𝑡 2𝜆𝜆𝑡𝑡 2𝜆𝜆by
𝜆𝜆𝑡𝑡 introduced 𝑡𝑡 𝜆𝜆𝑡𝑡 𝜆𝜆𝑡𝑡 𝜆𝜆are
calculates and publishes Philippine Dealing TheSystem parameters 𝛽𝛽0𝑡𝑡 , 𝛽𝛽1𝑡𝑡 , and 𝛽𝛽2𝑡𝑡 are were estimated
called factors andand corresponding
their coefficients 𝑡𝑡 calle
Treasury Reference Rates such as PDS Treasury R were taken for each t. The average over all t was
2
Nelson From From Equation
From Equation Equation (3),(3), (3),
Reference Rate AM and Siegel assumes
(PDST-R1) and PDS that
loadings.the forward
Treasury The factor rate
then follows
loading
calculated. the
of 𝛽𝛽equation
0𝑡𝑡 is 1, a the
Finally, constant shapethat does not that
parameter decay to zero eve
Reference Rate PM (PDST-R2). Both PDST-R1 and produced the highest 𝜏𝜏 −𝜏𝜏/𝜆𝜆 average 1 1was 𝜏𝜏 𝜏𝜏 chosen. Because the 𝑢𝑢 𝑢𝑢
−𝜏𝜏/𝜆𝜆𝑡𝑡 1 𝜏𝜏 (4) 𝑡𝑡 it −𝑢𝑢/𝜆𝜆 𝑢𝑢 𝑡𝑡 −𝑢𝑢
PDST-R2 benchmarks are intended to become Thus,
𝑓𝑓𝑡𝑡 (𝜏𝜏) it has
=the𝛽𝛽0𝑡𝑡 significant
+ 𝛽𝛽results1𝑡𝑡 𝑒𝑒 ofcontribution
+the𝛽𝛽2𝑡𝑡 fixed 𝑦𝑦𝑡𝑡to(𝜏𝜏)
𝑦𝑦𝑡𝑡𝑒𝑒lambda
(𝜏𝜏) =
𝑦𝑦the
𝑡𝑡
=
(𝜏𝜏) yield
method
∫ = ∫ [𝛽𝛽for
[𝛽𝛽
∫
0𝑡𝑡 +any
were[𝛽𝛽
0𝑡𝑡 𝛽𝛽
+ maturity;
1𝑡𝑡more
𝛽𝛽𝑒𝑒1𝑡𝑡−𝑢𝑢/𝜆𝜆
+ 𝑒𝑒1𝑡𝑡
𝛽𝛽 +𝑡𝑡hence,
stable,
−𝑢𝑢/𝜆𝜆
𝑡𝑡 −𝑢𝑢/𝜆𝜆
𝑒𝑒 𝛽𝛽
+2𝑡𝑡𝛽𝛽+ 2𝑡𝑡 𝛽𝛽𝑒𝑒2𝑡𝑡is𝑒𝑒 −𝑢𝑢/𝜆𝜆
referr
𝑒𝑒] 𝑑𝑑𝑡𝑡
𝜆𝜆𝑡𝑡 𝑡𝑡
𝜏𝜏 𝜏𝜏 𝜏𝜏 0𝑡𝑡
𝜆𝜆 𝜆𝜆 𝜆𝜆
source of reference rates for the repricing of loans, the succeeding steps proceeded 0 0from 0 these. 𝑡𝑡 𝑡𝑡 𝑡𝑡
securities, derivative transactions, and other long-term
interestfactor. The It isfactor important 𝛽𝛽1𝑡𝑡 is called to notethe that short-term
the results factor because its factor lo
in either
where 𝛽𝛽0𝑡𝑡 , 𝛽𝛽1𝑡𝑡 , 𝛽𝛽2𝑡𝑡 , and 𝜆𝜆𝑡𝑡 are constants Evaluating and
Evaluating
Evaluating 𝜆𝜆𝑡𝑡 the ≠the 0. Note
integralintegral
the that
at the
integral ator𝑥𝑥(𝜏𝜏)
theright,
at = right,
right,
the 𝑡𝑡 (𝜏𝜏)
𝑓𝑓we we is
obtain obtain
we a obtain
solution
thethe toNelson-Siegel
Nelson-Siegel the
Nelson-Siegel
the model mode mo fo
rate sensitive instruments to be issued. They are also method, grid search fixed lambda, were dependent
−𝜏𝜏/𝜆𝜆𝑡𝑡
intended to become the bases for market1−𝑒𝑒 valuation , has equation on the choice
significant contribution of rangetoofthe values value ofofl𝑦𝑦t.−𝜏𝜏/𝜆𝜆
second-order linear ordinary differential (𝜏𝜏) 𝑡𝑡 at
−𝜏𝜏/𝜆𝜆 𝑡𝑡 shorter 1maturities −𝜏𝜏/𝜆𝜆 𝑡𝑡 (sm
−𝜏𝜏/𝜆𝜆
of Government Securities and other Philippine-peso- 𝜏𝜏/𝜆𝜆 𝑡𝑡 Having 𝑦𝑦chosen the fixed 1−
shape 1 𝑒𝑒−1 𝑡𝑡 𝑒𝑒
−
parameter 𝑒𝑒 −𝜏𝜏/𝜆𝜆 𝑡𝑡
and −1 𝑒𝑒−1 𝑒𝑒− 𝑒𝑒 −𝜏𝜏/
𝑡𝑡
(𝜏𝜏)
𝑦𝑦 (𝜏𝜏)
=
𝑡𝑡 𝑡𝑡 𝑦𝑦𝑡𝑡 0𝑡𝑡= =𝛽𝛽
(𝜏𝜏) 𝛽𝛽 + 𝛽𝛽
+
0𝑡𝑡𝛽𝛽0𝑡𝑡 𝛽𝛽
1𝑡𝑡 + ( (
1𝑡𝑡 𝛽𝛽1𝑡𝑡 ( ) + ) 𝛽𝛽
+ ) 𝛽𝛽
2𝑡𝑡 + ( (
2𝑡𝑡𝛽𝛽2𝑡𝑡 ( − 𝑒𝑒−
denominated fixed income securities. 𝜏𝜏/𝜆𝜆 𝜏𝜏/𝜆𝜆 𝑡𝑡the𝑡𝑡 second 𝜏𝜏/𝜆𝜆
𝑡𝑡 t,𝜏𝜏/𝜆𝜆 𝜏𝜏/𝜆𝜆
𝑡𝑡 𝜏𝜏/𝜆𝜆
𝜏𝜏). Moreover, ′′
2itthe decreases
′
resulting
1 to 0beta 𝛽𝛽as parameters for each
0𝑡𝑡 𝜏𝜏 → +∞. Lastly, 𝛽𝛽 is referred (5) to as the
𝑡𝑡 𝑡𝑡
medium
In this work, TERMthe STRUCTURE
PDST-R2 rates from January𝑥𝑥2, + step
FORECASTING 𝑥𝑥 +was2 𝑥𝑥to=fit a2 univariate time series model for 2𝑡𝑡 5
The The𝜆𝜆parameters
parameters 𝜆𝜆𝑡𝑡 { 𝛽𝛽 𝛽𝛽,},𝜆𝜆𝛽𝛽𝑡𝑡𝛽𝛽 𝛽𝛽,1𝑡𝑡 factorsfactors
2008 to December 25, 2013 were chosen. The PDST-R2 𝑡𝑡The
each parameters
of 0𝑡𝑡 0𝑡𝑡 ,{1𝑡𝑡 0𝑡𝑡 ,and , and
}𝛽𝛽 𝛽𝛽
, 2𝑡𝑡
1𝑡𝑡and and are
𝛽𝛽2𝑡𝑡{ 𝛽𝛽are called
2𝑡𝑡 }. called
are called
First, each and
factors and
series their
andtheircoefficients
their coefficiecoeffi
1−𝑒𝑒 −𝜏𝜏/𝜆𝜆𝑡𝑡
rate is the weighted average of the yields because from done its factor was loading, tested for stationarity − 𝑒𝑒 −𝜏𝜏/𝜆𝜆 𝑡𝑡 ,and
starts was at 0, found increases, to have then decays to 0
transactions ofFrom (Nelson Equation
of& (3), 1987).
Siegel, To generate this range of shapes, 𝜏𝜏/𝜆𝜆𝑡𝑡 a parsimonious model introduced by
the set benchmark securities for each loadings.
loadings.
loadings.
a unit The The
root.factor
The factor loading
factor
Hence, loading of of
loading
each is 𝛽𝛽1,
𝛽𝛽0𝑡𝑡𝛽𝛽of0𝑡𝑡
series is
0𝑡𝑡1,
was aisconstant
atransformed
1,constant
a constant that that does
by does
that not
does notdecay decay
not de to
tenor up to 4:15 p.m. For simplicity, weeklyprecisely, data were differencing to remove
𝜏𝜏 it approaches 0 as 𝜏𝜏 → 0 and as 𝜏𝜏 → +∞. Its significant contribution to + the unit root. Each series also
Nelson and Siegel assumes that1 the forward Thus, Thus, it rate
has
it has follows
itsignificant
significant 𝑢𝑢equation
thecontribution (6)
extracted by taking the rates Wednesday of each
𝑦𝑦𝑡𝑡 (𝜏𝜏) = week. ∫ [𝛽𝛽 Thus,
showed
0𝑡𝑡 + 𝛽𝛽1𝑡𝑡 𝑒𝑒
−𝑢𝑢/𝜆𝜆 has
strong
𝑡𝑡 + significant
𝛽𝛽2𝑡𝑡presence 𝑒𝑒contribution
contribution
−𝑢𝑢/𝜆𝜆 𝑡𝑡of
] 𝑑𝑑𝑑𝑑 ARCHto to thethe toyield
effect.yield
the for
yield for
The any any
for
bestmaturity;
maturity;
any maturity; hence, henc h
Index t is attached to each week. 𝜏𝜏 model obtained for 𝜆𝜆
𝜏𝜏 each differenced series followed
0 𝑡𝑡
the medium term to maturity. (4)factor
𝑓𝑓𝑡𝑡 (𝜏𝜏) =long-term long-term
𝛽𝛽0𝑡𝑡 𝛽𝛽1𝑡𝑡 factor.
+long-term factor.
𝑒𝑒 −𝜏𝜏/𝜆𝜆
ARMA(p,q)+eGARCH(1,1). The
𝑡𝑡factor.
+ The
𝛽𝛽 2𝑡𝑡factor
Thefactor is 𝛽𝛽called
𝛽𝛽1𝑡𝑡𝛽𝛽𝑡𝑡1𝑡𝑡
𝑒𝑒 −𝜏𝜏/𝜆𝜆
factor isThen
1𝑡𝑡called thethe
is called short-term
error short-term
the short-term
terms factor
of factor because
because beca it
Evaluating the integral at the right, we obtain the Nelson-Siegel 𝜆𝜆𝑡𝑡 model for the yield curve
Procedure the estimated models for the betas were considered as
−𝜏𝜏/𝜆𝜆 −𝜏𝜏/𝜆𝜆 −𝜏𝜏/𝜆𝜆
In using the Nelson-Siegel model to forecast the 1−𝑒𝑒 1−𝑒𝑒 a𝑡𝑡 ,random
1−𝑒𝑒 𝑡𝑡
has , has
𝑡𝑡
significant vector whose
significant contribution joint probability
contribution distribution
where 𝛽𝛽0𝑡𝑡 , 𝛽𝛽1𝑡𝑡 , 𝛽𝛽2𝑡𝑡 , and 𝜆𝜆𝑡𝑡 are constants 1 −
𝜏𝜏/𝜆𝜆𝑒𝑒 and
−𝜏𝜏/𝜆𝜆
𝜏𝜏/𝜆𝜆 𝜏𝜏/𝜆𝜆 𝑡𝑡𝜆𝜆𝑡𝑡 ,≠ has 0. significant
Note 1was −that 𝑒𝑒 −𝜏𝜏/𝜆𝜆
contribution
𝑥𝑥(𝜏𝜏)
𝑡𝑡 = 𝑓𝑓to𝑡𝑡 (𝜏𝜏) to
thethe to
isvalue value
athe solutionof of
value 𝑦𝑦𝑡𝑡 (𝜏𝜏)
𝑦𝑦of
to
(7) at𝑡𝑡 (𝜏𝜏)
𝑡𝑡 (𝜏𝜏)
𝑦𝑦
the shorter
at shorter
at shorte matu m
term structure, three major steps were implemented
𝑦𝑦𝑡𝑡 (𝜏𝜏) = 𝛽𝛽0𝑡𝑡 + 𝛽𝛽1𝑡𝑡 (
𝑡𝑡 function
𝑡𝑡 𝑡𝑡
) + 𝛽𝛽2𝑡𝑡 ( (pdf) to be determined
− 𝑒𝑒 −𝜏𝜏/𝜆𝜆 𝑡𝑡 ) by the copula
in this work. The first step was to run multiple linear 𝜏𝜏/𝜆𝜆method. 𝑡𝑡 Finally, random 𝜏𝜏/𝜆𝜆𝑡𝑡 numbers from the joint pdf
second-order linear ordinary differential 𝜏𝜏).𝜏𝜏). equation
Moreover,
Moreover, it decreases
it decreases to to 0 as 0toas 𝜏𝜏0→ 𝜏𝜏as→ +∞.
𝜏𝜏 +∞. Lastly,Lastly, 𝛽𝛽2𝑡𝑡𝛽𝛽2𝑡𝑡is 𝛽𝛽referred
is2𝑡𝑡referred to to as as th
regression in the form of Equation (8) for each dataset of Moreover,
𝜏𝜏). the error terms it decreases were generated. → These +∞. Lastly,
were then is referred to
t to obtain the The
beta parameters 𝛽𝛽0𝑡𝑡 , 𝛽𝛽1𝑡𝑡 , and 𝛽𝛽2𝑡𝑡 .are This called 2used factors in and the final
1 their 𝛽𝛽0𝑡𝑡step,
coefficients forecasting
−𝜏𝜏/𝜆𝜆 −𝜏𝜏/𝜆𝜆 are called
𝑡𝑡 −𝜏𝜏/𝜆𝜆
the term factor structure (5)
′′ ′ 1−𝑒𝑒1−𝑒𝑒 1−𝑒𝑒 𝑡𝑡 𝑡𝑡 −𝜏𝜏/𝜆𝜆
−𝜏𝜏/𝜆𝜆
step required a prior choice of value or values of the 𝑥𝑥because
because of𝑥𝑥its
+ because interestits
+ factor factor 𝑥𝑥loading,
its2 rates.
factor=loading, loading,
2 − 𝑒𝑒
− 𝑒𝑒− 𝑡𝑡 , −𝜏𝜏/𝜆𝜆
𝑒𝑒 starts
𝑡𝑡 , starts at 0,
𝑡𝑡 , starts at 0, increases,
atincreases,
0, increases, then then dt
𝜆𝜆𝑡𝑡 𝜆𝜆that
𝑡𝑡 does𝜆𝜆𝑡𝑡not decay to zero
𝜏𝜏/𝜆𝜆 𝜏𝜏/𝜆𝜆
𝑡𝑡 𝑡𝑡𝜏𝜏/𝜆𝜆𝑡𝑡
shape parameter loadings.lt. TheThe nextfactor
steploadingwas to of 𝛽𝛽0𝑡𝑡 is a1, a constant
assume even as 𝜏𝜏 → +∞.
model for the time series of beta parameters obtained. precisely,
precisely,
precisely, it approaches
it approaches
it approaches 0 as 0 as 𝜏𝜏0→ 𝜏𝜏as→0𝜏𝜏+0→ + +
and 0and asand as
𝜏𝜏 → 𝜏𝜏as→ +∞. → Its
𝜏𝜏 +∞. +∞. Its
significant
significant
Its significant contrco
From
Thus, itEquation
has to (3),
significant contribution to the yield for any maturity; hence, it is referred to as the
This model was then used forecast future values Implementation and Results
of these parameters. The last step was to plug in the
1 𝜏𝜏thethe medium medium
the medium term term to
term to
maturity.
maturity.
to𝑢𝑢maturity. (6)
future valueslong-term
of the beta factor. The factor
parameters 𝑦𝑦and
𝑡𝑡
𝛽𝛽
(𝜏𝜏) the
=
1𝑡𝑡 is called
same
∫ [𝛽𝛽 the
0𝑡𝑡 + short-term
Fitting𝛽𝛽 1𝑡𝑡 𝑒𝑒 the
−𝑢𝑢/𝜆𝜆 factor
𝑡𝑡Yield
+ 𝛽𝛽 2𝑡𝑡
because
Curve 𝑒𝑒 −𝑢𝑢/𝜆𝜆𝑡𝑡its ] 𝑑𝑑𝑑𝑑 factor loading,
shape parameter in Equation (7) to obtain forecast 𝜏𝜏 0of A total of 313 datasets 𝜆𝜆𝑡𝑡 of zero rates from January 2,
−𝜏𝜏/𝜆𝜆
interest rates y1−𝑒𝑒 (t). 𝑡𝑡
, has significant contribution to the 2008 of
value to 𝑦𝑦 December 25, 2013 were considered. Each
t 𝑡𝑡 (𝜏𝜏) at shorter maturities (smaller values of
For the firstEvaluating
step,
𝜏𝜏/𝜆𝜆𝑡𝑡 the integral
regression usingata the fixed right,
shape we obtain dataset the Nelson-Siegelgave the interest model for thein
rates yield
hundred curve basis
parameter across all datasets and using a grid search points for maturities 1 month (Mo), 3 Mo, 6 Mo, 1
𝜏𝜏). Moreover,
were both considered it decreases
initially. For𝑦𝑦 the grid to search,
0 as 𝜏𝜏 → the 1 − Lastly,
+∞. −𝜏𝜏/𝜆𝜆𝑡𝑡 𝛽𝛽 is referred −𝜏𝜏/𝜆𝜆
𝑒𝑒year 13Yr,− 𝑒𝑒 4Yr, to as𝑡𝑡 the −𝜏𝜏/𝜆𝜆 medium-term factor(7)
(Yr),
) + 𝛽𝛽2Yr,
2𝑡𝑡 5Yr,
− 𝑒𝑒 6Yr, 𝑡𝑡 ) 7Yr, 8Yr, 9Yr,
𝑡𝑡 (𝜏𝜏) = 𝛽𝛽0𝑡𝑡 + 𝛽𝛽1𝑡𝑡 ( 2𝑡𝑡 (
values of lt ranged from 2.000000 to 40.000000 with 𝜏𝜏/𝜆𝜆
10Yr, 𝑡𝑡 𝜏𝜏/𝜆𝜆
15Yr, 20Yr, and𝑡𝑡30Yr. Figure 1 shows sample
1−𝑒𝑒 −𝜏𝜏/𝜆𝜆𝑡𝑡
increments ofbecause0.000001 for each
its factor loading,t. The lt together − 𝑒𝑒 −𝜏𝜏/𝜆𝜆𝑡𝑡
, starts
yield at 0, increases, then decays to 0. More
curves.
𝜏𝜏/𝜆𝜆𝑡𝑡
with the set of betaThe parameters
parameters 𝛽𝛽0𝑡𝑡 , 𝛽𝛽1𝑡𝑡 , and 𝛽𝛽 2𝑡𝑡 that
are called factors
Before and their coefficients
performing linear regression, are called allfactorrates were
highest R2itfor
produced the precisely, each t were
approaches 0 as 𝜏𝜏 → 0+ and taken. It was as 𝜏𝜏first → converted
+∞. Its significant to continuously contribution compounding to the yield rates. is For in
noted that quiteloadings.
a number The factor loading
of datasets had veryofhigh 𝛽𝛽0𝑡𝑡 Ris2 1, a constant
the processing that does of data, not decay from estimation to zero even to as time 𝜏𝜏 → series+∞.
but the beta parameters
the mediumobtained were not realistic,
term to maturity. modelling of the beta parameters, only the rates from
that is, the values Thus, were it has
verysignificant
far from the contribution
trend in the to the Januaryyield for 2, any 2008 maturity; to December hence, it26, is referred
2012 were to asused. the
majority of datasets. This problem did not appear in Thus, 261 datasets were considered. The remaining
long-term factor. The factor 𝛽𝛽1𝑡𝑡 is called the short-term factor because its factor loading,

1−𝑒𝑒 −𝜏𝜏/𝜆𝜆𝑡𝑡
𝜏𝜏/𝜆𝜆𝑡𝑡
, has significant contribution to the value of 𝑦𝑦𝑡𝑡 (𝜏𝜏) at shorter maturities (smaller values of
 justified
justifiedbybyKPSS
KPSStesttest(for
(forthe
theactual
actualseries)
series)and
andARCH-LM
ARCH-LMtest test(for
(forthe
thedifferenced
difference
All statistical procedures were implemented using the software R. First, univariate time
giving all significant p-values. These are s
giving
giving all
allsignificant
significant p-values.
p-values. These
These
2 andare
arestandard
standard
ofprocedures
procedures inindetecting
detectingthese
thesefeat
fea
RM STRUCTUREseries models were fitted on the
FORECASTING estimated betas. As seen in Figures 3, presence non-
financialE.P.
timede series.
Lara-Tuprio, et al
10
194
stationarity and of conditionalfinancial
heteroscedasticity
financial time were apparent. These observations were
timeseries.
series.
justified by KPSS test (for the actual series) and ARCH-LM test (for the differenced series),

giving all significant p-values. These are standard procedures in detecting these features of most

financial time series.

Figure 2. Time serie

Figure
Figure2.2.Time
Timeseries
seriesplots
plotsofofthe
theestimated
estimatedbetas.
betas.

Figure 2. Time series plots of the estimated betas.

Figure 1.Figure
Actual1. Actual yield curves for selected dates.
yield curves for selected dates.
Table 1. Minimum, Maximum, Mean Values, and Standard Deviations of R2 and Beta Parameters

Min max mean standard deviation

Before performing linear
R
regression,
80.26%
all rates
99.72%
were first 3.04%
96.22%
converted to continuously
2

Figure 3. Time serie

β0 0.02242 1.49527 0.17384 0.18941
Figure
Figure3.3.Time
Timeseries
seriesplots
plotsofofthe
thedifferenced
differencedbetas
betas. .
mpounding rates. For theβprocessing
-1.41810 of 0.01044
data, from estimation0.17833
-0.13947 to time series modelling of the
1
Figure 3. Time series plots of the differenced betas.
β2 -1.80766 0.24361 -0.04411 0.25549
a parameters, only the rates from January 2, 2008 to December 26,The
2012
best were
models used. Thus,
for all three 2
differ
TheThebest bestmodels modelsfor forall
The allbest three
three models differenced
differenced for all three series
seriesdifferenced 𝑑𝑑𝛽𝛽
𝑑𝑑𝛽𝛽𝑖𝑖 ,𝑖𝑖 ,𝑖𝑖 𝑖𝑖==0,1,2, serieswere
0,1,2, wereARMA ARMA
historical data were to be used to check the reliability
The best models for all three differenced series 𝑑𝑑𝛽𝛽𝑖𝑖 , 𝑖𝑖 = 0,1,2,, were wereARMA equation
ARMA forfor and
the the mean Exponential
mean equation GARCH for the
of the forecast.
asets were considered.
Theequation
fixed and lambda The method remaining equation
equationand
described historical
andExponential
Exponential anddata GARCH
GARCH
Exponentialwerefor forto the
GARCH the be variance used
variance for the tovariance
equation.
equation. check In the reliability
Inparticular,
particular,
equation. based
basedon on
Exponential GARCH for theearliervariance equation. In particular,
In particular, based on the Bayesian Information basedBayesian on the Information Criterion (BIC), bot
produced lt = 9.807527. To summarize the results of
2 Bayesian
Bayesian Information
Information Criterion
Criterion
Criterion
TERM STRUCTURE (BIC),
(BIC),FORECASTING
(BIC), both
bothboth{𝑑𝑑𝛽𝛽 {𝑑𝑑𝛽𝛽00}}and and{𝑑𝑑𝛽𝛽
and {𝑑𝑑𝛽𝛽11}}followed followed
followed MA(1)+eGA MA(1)+eGA
of R and(BIC),
forecast. linear regression,
of Equation (8)TERM
the valuesCriterion
Bayesian Information
for the different datasets
STRUCTURE TERMare
FORECASTING
beta parameters
both {𝑑𝑑𝛽𝛽
STRUCTURE
TERM0 } and
describedFORECASTING
{𝑑𝑑𝛽𝛽
STRUCTURE 1 } followed
MA(1)+eGARCH(1,1) FORECASTING MA(1)+eGARCH(1,1)
models,
models, while
while {𝑑𝑑𝛽𝛽 2 } followed
followed 12 ARMA(1,1
in Table 1.
models, while {𝑑𝑑𝛽𝛽2 } followedmodels, models, while
while{𝑑𝑑𝛽𝛽
ARMA(1,1)+eGARCH(1,1). {𝑑𝑑𝛽𝛽22}}followed ARMA(1,1)+eGARCH(1,1).
followed ARMA(1,1)+eGARCH(1,1).
ARMA(1,1)+eGARCH(1,1).
TERM STRUCTURE FORECASTING The Theestimated Theestimated estimated
models models weremodels given were
were by the given
given by by
following the the following
12
equations. equatio
Let 𝑑𝑑𝛽𝛽
Forecasting of interest rates requires forecast of
The fixed lambda method
The estimated described
the beta parameters. Thus, the next step was to find models The were in
estimated
givenSection models
by the
following III
were
following produced
given
equations. by the following
equations. 𝜆𝜆
Let Let
𝑡𝑡 = 𝑑𝑑𝛽𝛽 9.807527.
equations.
𝑖𝑖,𝑡𝑡 = 𝛽𝛽 𝑖𝑖,𝑡𝑡 − Let
𝛽𝛽 𝑑𝑑𝛽𝛽
𝑖𝑖,𝑡𝑡−1
To
,
𝑖𝑖,𝑡𝑡 = 𝛽𝛽 𝑖𝑖,𝑡𝑡 − 𝛽𝛽𝑖𝑖,𝑡𝑡−
appropriate time series models for them. models were 𝑖𝑖 = 0,1,2, 𝑖𝑖 =𝑡𝑡 = 0,1,2,1, 2,𝑡𝑡…=, 𝑇𝑇, 1, where
2, … , 𝑇𝑇, T is where
wherethe sample TT is thesize. sample Thensize. size. Then
The estimated
𝑖𝑖 = 0,1,2,
𝑖𝑖 = 0,1,2, 𝑡𝑡 = 1, 2, … , 𝑇𝑇, where T is the sample 𝑡𝑡 = 1, 2, given
… , 𝑇𝑇, by
where
Thensize. Thenthe Tfollowing
is the equations.
sample size. Let
Then 𝑑𝑑𝛽𝛽 𝑖𝑖,𝑡𝑡 = 𝛽𝛽 𝑖𝑖,𝑡𝑡 − 𝛽𝛽 𝑖𝑖,𝑡𝑡−1 ,
2
mmarize the results of linear
Time Series Models for0,1,2,
the Beta
regression,
Parameters
the values of 𝑅𝑅 and𝑑𝑑𝛽𝛽beta 𝑖𝑖,𝑡𝑡 = 𝜇𝜇𝑑𝑑𝛽𝛽
parameters
𝑖𝑖 +𝑖𝑖,𝑡𝑡𝜑𝜑= 𝑖𝑖 𝑑𝑑𝛽𝛽𝜇𝜇𝑖𝑖,𝑡𝑡−1
𝑖𝑖 + 𝜑𝜑 +𝑖𝑖 𝑑𝑑𝛽𝛽 of
𝑎𝑎𝑖𝑖,𝑡𝑡 +Equation
𝑖𝑖,𝑡𝑡−1 𝜃𝜃+ 𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝜃𝜃𝑖𝑖 𝑎𝑎𝑖𝑖,𝑡𝑡
𝑖𝑖 𝑎𝑎𝑖𝑖,𝑡𝑡−1
(8
𝑖𝑖 = 𝑡𝑡 = 1, 2, … , 𝑇𝑇, where T is the sample 𝑑𝑑𝛽𝛽
size. = Then
𝜇𝜇 + 𝜑𝜑 𝑑𝑑𝛽𝛽 + 𝑎𝑎 + 𝜃𝜃 𝑎𝑎 (9) (9)
All statistical procedures were implemented 𝑑𝑑𝛽𝛽𝑖𝑖,𝑡𝑡 using
= 𝜇𝜇𝑖𝑖 + 𝜑𝜑𝑖𝑖 𝑑𝑑𝛽𝛽𝑖𝑖,𝑡𝑡−1 + 𝑎𝑎𝑖𝑖,𝑡𝑡 + 𝜃𝜃𝑖𝑖 𝑎𝑎𝑖𝑖,𝑡𝑡−1
𝑖𝑖,𝑡𝑡 𝑖𝑖 𝑖𝑖 𝑖𝑖,𝑡𝑡−1 𝑖𝑖,𝑡𝑡 𝑖𝑖 𝑖𝑖,𝑡𝑡−1
𝑎𝑎𝑖𝑖,𝑡𝑡 = 𝜎𝜎𝑖𝑖,𝑡𝑡 𝜀𝜀𝑖𝑖,𝑡𝑡 𝑎𝑎 = 𝜎𝜎 𝜀𝜀
𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡
𝑎𝑎 = 𝜎𝜎 𝜀𝜀 𝑎𝑎𝑖𝑖,𝑡𝑡 = 𝜎𝜎𝑖𝑖,𝑡𝑡 𝜀𝜀𝑖𝑖,𝑡𝑡 (10) (10)
the software R. First, univariate time series models
the different datasets are described in Table
𝑑𝑑𝛽𝛽𝑖𝑖,𝑡𝑡 = 1. 𝜇𝜇𝑖𝑖 +𝑖𝑖,𝑡𝑡𝜑𝜑𝑖𝑖 𝑑𝑑𝛽𝛽𝑖𝑖,𝑡𝑡−1
𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡+ 𝑎𝑎 + 𝜃𝜃 𝑎𝑎
ln(𝜎𝜎 2 𝑖𝑖,𝑡𝑡 2 𝑖𝑖 𝑖𝑖,𝑡𝑡−1
ln(𝜎𝜎 𝜔𝜔𝑖𝑖𝑖𝑖,𝑡𝑡+)|𝑎𝑎 2
𝛼𝛼=𝑖𝑖𝑖𝑖,𝑡𝑡−1
𝑎𝑎𝜔𝜔 𝑖𝑖 |++ 𝛼𝛼 2 |𝑎𝑎𝑖𝑖,𝑡𝑡−1 |
𝑖𝑖𝛾𝛾𝑎𝑎𝑖𝑖 𝑖𝑖,𝑡𝑡−1
(𝑎𝑎𝑖𝑖,𝑡𝑡−1 + 𝛾𝛾𝑖𝑖 −
|𝑎𝑎𝑖𝑖,𝑡𝑡−1
( 𝐸𝐸 (|
𝑎𝑎(9)
(11)
|
𝑖𝑖,𝑡𝑡−1
−2|))
𝑎𝑎𝑖𝑖,𝑡𝑡−1
𝐸𝐸 (|+ 𝜅𝜅𝑖𝑖 ln |
were fitted on the estimated betas. As seen in Figures2 𝑎𝑎
|𝑎𝑎 = 𝜎𝜎| 2 𝜀𝜀)𝑖𝑖,𝑡𝑡=
𝑖𝑖,𝑡𝑡 𝑎𝑎 𝑖𝑖,𝑡𝑡−1 (11) (10)
2 ln(𝜎𝜎
2 ) = 𝜔𝜔 + 𝑖𝑖,𝑡𝑡 𝛼𝛼𝑖𝑖 𝑎𝑎𝑖𝑖,𝑡𝑡−1 + 𝛾𝛾𝑖𝑖𝑖𝑖,𝑡𝑡−1
𝑖𝑖,𝑡𝑡−1 𝑖𝑖,𝑡𝑡 ( − 𝐸𝐸 (| 𝜎𝜎2𝑖𝑖,𝑡𝑡−1 |)) +𝜎𝜎𝜅𝜅𝑖𝑖,𝑡𝑡−1 𝜎𝜎𝑖𝑖,𝑡𝑡−1𝑖𝑖,𝑡𝑡 ) 𝜎𝜎𝑖𝑖,𝑡𝑡−1
𝑖𝑖 ln(𝜎𝜎
2 and 3, presence of non-stationarity ln(𝜎𝜎𝑖𝑖,𝑡𝑡 )and= 𝜔𝜔of𝑖𝑖 conditional
+ 𝛼𝛼𝑖𝑖 𝑎𝑎𝑖𝑖,𝑡𝑡−1 + 𝛾𝛾𝑖𝑖 ( 𝑖𝑖
𝑖𝑖,𝑡𝑡 − 𝐸𝐸 (| |))
𝜎𝜎𝑖𝑖,𝑡𝑡−1 + 𝜅𝜅𝑖𝑖 ln(𝜎𝜎 𝜎𝜎𝑖𝑖,𝑡𝑡−1
𝑖𝑖,𝑡𝑡 ) (11)
𝜎𝜎𝑖𝑖,𝑡𝑡−1
|𝑎𝑎𝑖𝑖,𝑡𝑡−1 |𝑎𝑎 𝜎𝜎𝑖𝑖,𝑡𝑡−1
𝑎𝑎𝑖𝑖,𝑡𝑡−1
Tablewere1 justified by KPSS test (for ln(𝜎𝜎
heteroscedasticity were apparent. These2 observations
𝑖𝑖,𝑡𝑡 ) = 𝜔𝜔𝑖𝑖 + 𝑎𝑎 𝛼𝛼𝑖𝑖 𝑎𝑎
2
𝑖𝑖,𝑡𝑡−1 +
where 𝐸𝐸where
𝑎𝑎𝑖𝑖,𝑡𝑡−1
𝛾𝛾(|𝑖𝑖 ( 𝐸𝐸|)(|= − 𝐸𝐸|)
𝐸𝐸(|𝜀𝜀
𝑖𝑖,𝑡𝑡−1 (|
𝑖𝑖,𝑡𝑡=|) =0
𝜈𝜈 𝐸𝐸(|𝜀𝜀 |)) when |)+= 𝜅𝜅√ 𝑖𝑖 0ln(𝜎𝜎
𝜈𝜈
when
2
𝜀𝜀𝑖𝑖,𝑡𝑡 )√2 . The
𝑖𝑖,𝑡𝑡 ~𝑡𝑡
𝜈𝜈
coefficients
𝜀𝜀𝑖𝑖,𝑡𝑡 ~𝑡𝑡 2 . The coeffi and
𝑎𝑎𝑖𝑖,𝑡𝑡−1 thewhere actual𝐸𝐸series) (|𝜎𝜎𝑖𝑖,𝑡𝑡−1and
|) = 𝐸𝐸(|𝜀𝜀 |)𝜎𝜎=
𝜎𝜎𝑖𝑖,𝑡𝑡−1
𝜈𝜈 𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡−10 when𝜎𝜎𝑖𝑖,𝑡𝑡−1 √ 𝜎𝜎𝑖𝑖,𝑡𝑡−1 𝜀𝜀 𝑖𝑖,𝑡𝑡
𝑖𝑖,𝑡𝑡
~𝑡𝑡 2 . The 𝜈𝜈−2 coefficients 𝜈𝜈−2 and corresponding
Minimum,
ARCH-LM Maximum,
where
test 𝐸𝐸 (|
(for the 𝜎𝜎Mean
differenced |) = 𝐸𝐸(|𝜀𝜀
𝑖𝑖,𝑡𝑡−1𝑎𝑎 Values,
series), 𝑖𝑖,𝑡𝑡 |)giving 0and
=𝑖𝑖,𝑡𝑡−1 all Standard
when √𝜈𝜈−2 𝜀𝜀𝑖𝑖,𝑡𝑡 ~𝑡𝑡2 . Deviations
𝜈𝜈
The coefficients 𝜈𝜈−2
ofandR2 correspondingand Beta Paramete
significant p-values.where These𝐸𝐸are 𝑖𝑖,𝑡𝑡−1
(| standard |) = 𝐸𝐸(|𝜀𝜀 procedures |) = standard
0inwhenstandard √errors𝜀𝜀𝑖𝑖,𝑡𝑡 are ~𝑡𝑡given
errors 2 . The arein coefficients
Table 2.
given in Table and 2. corresponding
𝜎𝜎𝑖𝑖,𝑡𝑡−1 standard 𝑖𝑖,𝑡𝑡 errors are given in Table 2.
detecting thesestandard Min
features of most financial
errors are given in Table 2.max time series. mean 𝜈𝜈−2
standard deviation
Table 2 Table 2
2 standard errors are givenTable in Table 2 96.22% 2.
R 80.26% Table 2 99.72% Estimated Parameters
3.04%
of ARMA +ofeGARCH
Estimated Parameters ofEstimated ARMA + eGARCH Parameters Models and ARMA +Models
Corresponding eGARCH andModels Corresponding
Standardand Errors Corres Sta
Table 2Parameters of ARMA + eGARCH Models and Corresponding Standard Errors
β0 0.02242Estimated 1.49527 0.17384 i=0 0.18941 i=1
i=0 i=1
i=2
i=2
Estimated Parameters of ARMA + eGARCH i=0 Models and i=1Corresponding i=2
Standard Errors
galalltime series. p-values. These are standard procedures in detecting these features of m
significant

ial time series.

Philippine Interest Rates Using the Dynamic Nelson-Siegel Model 195

Figure 2. Time series plots of the estimated betas.

Figure 2. Time series plots of the estimated betas.
Figure 2. Time series plots of the estimated betas.
TERM STRUCTURE FORECASTING

i=2 0.00144197 0.02802909

TERM STRUCTURE
TERM FORECASTING
STRUCTURE FORECASTING 1212
The assumed distribution of the error 𝜀𝜀𝑖𝑖,𝑡𝑡 was the t distribution with 2 degrees of
The
Theestimated models
estimated were
models given
were byby
given thethe
following equations.
following Let
equations. 𝑑𝑑𝛽𝛽𝑑𝑑𝛽𝛽
Let ==
𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝛽𝛽𝑖𝑖,𝑡𝑡𝛽𝛽𝑖𝑖,𝑡𝑡
−− 𝛽𝛽𝑖𝑖,𝑡𝑡−1 , ,
𝛽𝛽𝑖𝑖,𝑡𝑡−1
or 𝑡𝑡2 . Such distribution captures the heavy-tailedness of the series since it has infinite va
𝑖𝑖 = 0,1,2,
𝑖𝑖 = 𝑡𝑡 =
0,1,2, 1, 1,
𝑡𝑡 = 2, 2,
…… where
, 𝑇𝑇,, 𝑇𝑇, TT
where is is
thethe
sample size.
sample Then
size. Then
In generating future values of the beta parameters using the formulas in Equation
𝑑𝑑𝛽𝛽𝑑𝑑𝛽𝛽 ==
𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 𝜇𝜇𝑖𝑖𝜇𝜇+𝑖𝑖 +𝜑𝜑𝑖𝑖𝜑𝜑𝑑𝑑𝛽𝛽
𝑖𝑖 𝑑𝑑𝛽𝛽
𝑖𝑖,𝑡𝑡−1
𝑖𝑖,𝑡𝑡−1 ++ 𝑎𝑎𝑖𝑖,𝑡𝑡𝑎𝑎𝑖𝑖,𝑡𝑡
++ 𝜃𝜃𝑖𝑖 𝜃𝜃
𝑎𝑎𝑖𝑖𝑖𝑖,𝑡𝑡−1
𝑎𝑎𝑖𝑖,𝑡𝑡−1 (9)(9)
(11), their underlying 𝑎𝑎𝑖𝑖,𝑡𝑡𝑎𝑎 = = 𝜎𝜎𝑖𝑖,𝑡𝑡𝜎𝜎𝜀𝜀𝑖𝑖,𝑡𝑡𝜀𝜀dependence structure based on their (10)
observed values were consider
(10)
Figure 3. Time𝑖𝑖,𝑡𝑡series𝑖𝑖,𝑡𝑡plots 𝑖𝑖,𝑡𝑡
of the differenced betas. (11)
|𝑎𝑎|𝑎𝑎 | | 𝑎𝑎𝑖𝑖,𝑡𝑡−1 (11)
𝑖𝑖,𝑡𝑡−1
𝑖𝑖,𝑡𝑡−1 𝑎𝑎𝑖𝑖,𝑡𝑡−1
Figure 3. Time series plots of the differenced betas
2 2 2 2 2 2
ln(𝜎𝜎
ln(𝜎𝜎 )
𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡=) 𝜔𝜔
= 𝜔𝜔
𝑖𝑖 +𝑖𝑖 required
𝛼𝛼
+ 𝑖𝑖 𝑎𝑎
𝛼𝛼 𝑎𝑎
𝑖𝑖𝑖𝑖,𝑡𝑡−1 +
𝑖𝑖,𝑡𝑡−1 estimating
+𝛾𝛾𝑖𝑖 (
𝛾𝛾 (
𝑖𝑖 𝜎𝜎 the
− −𝐸𝐸distribution
(|
𝐸𝐸 (|
𝜎𝜎𝑖𝑖,𝑡𝑡−1
|)) |))of
+ +the
𝜅𝜅𝑖𝑖 𝜅𝜅𝑖𝑖 random
ln(𝜎𝜎ln(𝜎𝜎 )
𝑖𝑖,𝑡𝑡 𝑖𝑖,𝑡𝑡 ) vector (𝜀𝜀.0 , 𝜀𝜀1 , 𝜀𝜀2 ) consisting of the e
𝜎𝜎𝑖𝑖,𝑡𝑡−1
𝑖𝑖,𝑡𝑡−1 𝜎𝜎𝑖𝑖,𝑡𝑡−1

where
𝑎𝑎 𝑎𝑎𝑖𝑖,𝑡𝑡−1 variables.
0 when As𝜈𝜈assumed
𝜈𝜈 in thein time
the series
time series models,
models, the
thedistribution ofeach
distribution of eacherror
error variable w
where (|(|𝑖𝑖,𝑡𝑡−1
𝐸𝐸 𝐸𝐸 |)|)== 𝐸𝐸(|𝜀𝜀 𝑖𝑖,𝑡𝑡 |)
𝐸𝐸(|𝜀𝜀 ==
𝑖𝑖,𝑡𝑡 |) 0 when √ √𝜈𝜈−2 𝜀𝜀𝑖𝑖,𝑡𝑡𝜀𝜀~𝑡𝑡 2 . 2The
𝑖𝑖,𝑡𝑡 ~𝑡𝑡 . The coefficients
coefficients andandcorresponding
corresponding
where when
Figure 3. Time series plots of the differenced betas
𝜎𝜎𝑖𝑖,𝑡𝑡−1
𝜎𝜎𝑖𝑖,𝑡𝑡−1 𝜈𝜈−2
variable was t2. The dependence structure . of (e0 , e1 ,
The coefficients and corresponding standard errors
The dependence structure of (𝜀𝜀 , 𝜀𝜀 , 𝜀𝜀 ) was obtained using a copula function.
e2 ) was
0 obtained
1 2 using a copula function.
standard
are givenerrors
standard inerrors
Table areare
given
2. giveninin Table Table2. 2.
With T as the sample size, the Copula
estimatedfor initial
The best models for all three differenced series 𝑑𝑑𝛽𝛽 , 𝑖𝑖 = 0,1,2, were ARMA for the me
values to Table
Table
be used 2 2 in forecasting Using
are given in Table 3.
the Joint Density
Using
𝑖𝑖 of the
Copula Error
for the JointVariables
Density of the Error
Variables
TheEstimated
assumed distribution
Parameters ofof
of the
ARMA error ei,t was
+ eGARCH Models the and Corresponding Standard
Estimated Parameters ARMAFirst, pseudo-observations
+ eGARCH Models and defined
First,
Corresponding by Errors
pseudo-observations
Standard Errors defined by
t distribution with 2 degrees of freedom or t2. Such
onThe
andbest
Exponential GARCH
models for fordifferenced
all three the variance
distribution captures the i=0 equation.
series 𝑑𝑑𝛽𝛽 , 𝑖𝑖 In
= particular,
0,1,2, werebased
heavy-tailedness
i=0 ARMA on the
i=1for the me
ofi=1the series i=2i=2
𝑖𝑖 𝑅𝑅𝑖𝑖
since it has μi μinfinite variance.
-0.000385
-0.000385 0.000386
0.000386 -0.000247
-0.000247 𝑈𝑈𝑖𝑖 =
i
𝑇𝑇 + 1 (12)
In generating future(0.000480) values
(0.000480) of the beta(0.000426)
parameters
(0.000426) (0.000409)

(0.000409)
an
on Information Criterion
and Exponential GARCH(BIC),
forboth
using theφformulas {𝑑𝑑𝛽𝛽 } and
the variance
i φi
{𝑑𝑑𝛽𝛽 } followed
equation. MA(1)+eGARCH(1,1)
In particular,
in Equations based on the
0 0 (9) to (11), their 0 00 -0.152738 1
-0.152738
underlying dependence structure were created,
based whereon their T is the sample
were
(0.043782) size and 𝑅𝑅 is the rank of the 𝑖𝑖 𝑡𝑡ℎ observation. These
created, where T𝑖𝑖 is the sample size and Ri is the
(0.043782)
observedθivalues were-0.215126 considered. This required rank of the ith observation. These pseudo-observations
s,ian
while {𝑑𝑑𝛽𝛽 } followed
Information
estimating
2 ARMA(1,1)+eGARCH(1,1).
Criterion
θi
the distribution
-0.215126
(0.053564)
(0.053564)
-0.199544
-0.199544
(BIC), both {𝑑𝑑𝛽𝛽 } and {𝑑𝑑𝛽𝛽 } followed MA(1)+eGARCH(1,1)
ofobservations were used
the random(0.050340)
vector
(0.050340)
-0.099143
-0.099143
(e0 ,0 to (0.039429)
check
were the association
used
(0.039429) to1 check theamong the error
association among variables
the errorand in estimat
e1 , e2 ) consisting of the error variables. As assumed variables and in estimating the parameters of the
ωiω -0.095836
-0.095836 -0.105876
-0.105876 -0.079393
-0.079393
i
(0.036828)
parameters(0.049583)
of the copula model. The pairwise scatterplots of the pseudo-observations are
(0.051335)
(0.036828) (0.049583) (0.051335)
s, while {𝑑𝑑𝛽𝛽2 } αfollowed
α
ARMA(1,1)+eGARCH(1,1).
0.390405
0.390405 -0.276128
-0.276128 -0.316096
-0.316096
i i
(0.207581)
in Figure 4.(0.188705) (0.206423)
(0.207581) (0.188705) (0.206423)
γi γi 0.537282
0.537282 0.626099
0.626099 0.558905
0.558905
196 E.P. de Lara-Tuprio, et al

Table 2. Estimated Parameters of ARMA + eGARCH Models and Corresponding Standard Errors

i=0 i=1 i=2

-0.000385 0.000386 -0.000247
μi
(0.000480) (0.000426) (0.000409)
0 0 -0.152738
φi 
    (0.043782)
-0.215126 -0.199544 -0.099143
θi 
(0.053564) (0.050340) (0.039429)
-0.095836 -0.105876 -0.079393
ωi 
(0.036828) (0.049583) (0.051335)
0.390405 -0.276128 -0.316096
αi 
(0.207581) (0.188705) (0.206423)
0.537282 0.626099 0.558905
γi
(0.277760) (0.343980) (0.360170)
0.984873 0.983536 0.983487
κi 
(0.006315) (0.008814) (0.008566)

Table 3. Estimated Initial Values Based on the Fitted ARMA + eGARCH Models

  ai,T σi,T
i=0 -0.00492075 0.01729929
i=1 0.00121345 0.01571196
i=2 0.00144197 0.02802909

copula model. The pairwise scatterplots of the pseudo- was based on maximum pseudo-likelihood estimation
observations are shown in Figure 4. proposed by Genest, Ghoudi, and Rivest (1995) and
Based on the scatterplots of the pseudo- the goodness of fit test was based on the empirical
observations, the first error variable had strong copula proposed by Genest, Rémillard, and Beaudoin
negative association to the second and third error (2009). The results are shown in Table 4. Note that ri,j
variables, while the second and the third had strong is the Pearson’s correlation coefficient of the ith and jth
pseudo-observations.
positive correlation. Notice that the associations

among the error variables showed a symmetric- Table 4. Results of the Goodness-of-Fit Test Under
type dependence. That is, the dependence on the the Gaussian Copula Assumption
tails was almost the same as the distributions on
the middle part of the distributions. One model that Estimated parameter Sn p-value
captures such symmetry in association between
two uniform random variables, as depicted by the ρ0,1= -0.981 0.0047 0.9296
scatterplots, is the elliptical family. This family ρ0,2= -0.959    
contains the Gaussian and t copulas. ρ1,2= 0.909    
Estimation and goodness of fit test were performed
for the Gaussian copulas. The estimation procedure
s of the copula model. The pairwise scatterplots of the pseudo-observations
positive correlation. Notice that the associations among the error variables showed a

tric-type dependence. That is, the dependence on the tails was almost the same as the
4.tions on the middle
Philippine Interest Rates Using the Dynamic Nelson-Siegel Model
part of the distributions. One model that captures such symmetry in
197

tion between two uniform random variables, as depicted by the scatterplots, is the

al family. This family contains the Gaussian and t copulas.

Estimation and goodness of fit test were performed for the Gaussian copulas. The

ion procedure was based on maximum pseudo-likelihood estimation proposed by Genest,

, and Rivest (1995) and the goodness of fit test was based on the empirical copula

ed by Genest, Rémillard, and Beaudoin (2009). The results are shown in Table 4. Note

is the Pearson’s correlation coefficient of the 𝑖𝑖 𝑡𝑡ℎ and 𝑗𝑗 𝑡𝑡ℎ pseudo-observations.

Table 4
Results of the Goodness-of-Fit Test Under the Gaussian Copula Assumption

Estimated parameter Sn p-value

ρ0,1= -0.981 0.0047 0.9296
ρ0,2= -0.959
Figure 4 . Pairwise scatterplots of the pseudo-observations.

Figure 4 . Pairwise scatterplots of the pseudo-observations.

ρ1,2= 0.909
With p-value of 0.9296, it seemed that the Gaussian Forecasting
sed on the scatterplots
copula was adequate to captureof the
the pseudo-observations,
dependence of the thea fraction
To validate the model,
With p-value of 0.9296, it seemed that the Gaussian copula was adequate to capture the
first oferror variable ha
the available
pseudo-observations. The estimated correlation matrix data, from January 2 to December 25, 2013 (52
was given by
ence of the pseudo-observations. The estimated correlation matrix was given by Wednesdays), was taken out. The goal was to compare
ssociation to the second and third error variables, while the second and the
1 −0.981 −0.959
the forecasts and the actual data taken out.
First, 10,000 triples of random numbers (e0,1 , e1,2 ,
(−0.981 1 0.909 ) e2,1 ) from the joint distribution of the random vector
−0.959 0.909 1 (e0 , e1 , e2 ) of the residuals were produced. Each triple
ERM STRUCTURE FORECASTING was substituted into the time 15series model to obtain the
stimated parameters were then used to simulate trivariate data with 𝑡𝑡2 margin and normal
These estimated parameters were then used to simulate values of b01 , b11 , b21 . Thus, 10,000 triples (b01, b11 ,
trivariate data with t2 margin and normal copula to
to capture the dependence. b21 ) were obtained. Then the median of each beta was
From Equations
capture the(9) to (11) and the random numbers15generated
dependence. taken from the distribution
to obtain one triple (bof01 , b11 , b21 ). This median
From Equations (9) to (11) and the random numbers triple was then used in Equation (7) to get the yield
he random vector generated
(𝜀𝜀0 , from 2 ), future
𝜀𝜀1 , 𝜀𝜀the distributionvalues of ofthethe beta parameters
random vector wereofcomputed
curve and used
week 1 (January in
2, 2013), y1 (t), where t is
) and the random (e0 , enumbers
1 generated
, e2 ), future values of from the distribution
the beta parameters wereof the maturity.
he Nelson-Siegel computed
equation and used in the Nelson-Siegel equation For the second week (t = 2),10,000 triples of random
uture values of the beta parameters were computed and used in numbers (e0,2 , e1,2 , e2,2 ) from the joint distribution of
1 − 𝑒𝑒 −𝜏𝜏/𝜆𝜆𝑡𝑡 1 − 𝑒𝑒 −𝜏𝜏/𝜆𝜆the
𝑡𝑡 random vector (e0 , e1 , (13) e2 ) were again produced.
−𝜏𝜏/𝜆𝜆𝑡𝑡
𝑦𝑦𝑡𝑡 (𝜏𝜏) = 𝛽𝛽0𝑡𝑡 + 𝛽𝛽1𝑡𝑡 ( ) + 𝛽𝛽2𝑡𝑡 ( − 𝑒𝑒 )
𝜏𝜏/𝜆𝜆𝑡𝑡
𝜏𝜏/𝜆𝜆 These and the betas of previous weeks were plugged
(13) 𝑡𝑡 in the time series model to obtain 10,000 triples (b ,
02
1 − 𝑒𝑒 −𝜏𝜏/𝜆𝜆𝑡𝑡
1 − 𝑒𝑒 −𝜏𝜏/𝜆𝜆𝑡𝑡 (13) b12 , b22 ). Again, the median of each beta was taken to
with( 𝜆𝜆𝑡𝑡 = 9.807527. ) + 𝛽𝛽2𝑡𝑡 The( initial values − 𝑒𝑒were
−𝜏𝜏/𝜆𝜆𝑡𝑡 those obtained for the December 26, 2012 dataset:
) get one triple (b02 , b12 , b22 ). This median triple was
1𝑡𝑡
𝜏𝜏/𝜆𝜆𝑡𝑡 𝜏𝜏/𝜆𝜆𝑡𝑡
used in the Nelson-Siegel equation to get the yield
𝛽𝛽0,0 = 0.042670232 , 𝛽𝛽0,1 = 0.035735575, and 𝛽𝛽0,2 = 0.107586694. curve yThe (t)randomness of the 9, 2013). This process
for week 2 (January
2
values were those obtained for the December
with lt = 9.807527. The initial values were those 26, 2012 dataset: was continued to find the yield curve for week t, t >
eta parameters were due to the residuals 𝜀𝜀𝑖𝑖,𝑡𝑡 , 𝑖𝑖 = 0,1,2.
obtained for the December 26, 2012 dataset: b 0,0 2. Notice that medians of beta parameters were taken
35735575, and 𝛽𝛽0,2 = 0.107586694.
= 0.042670232, The randomness
b 0,1 = 0.035735575, and ofb 0,2the
= first before computing the yield yt (t). An alternative
Forecasting 0.0107586694. The randomness of the beta parameters approach is to compute for each t the yield yt (t) using
were due
esiduals 𝜀𝜀𝑖𝑖,𝑡𝑡 , 𝑖𝑖 = 0,1,2. to the residuals e i,t
., i = 0,1,2. each of the 10,000 triples (e0,t , e1,t , e2,t ) then take
To validate the model, a fraction of the available data, from January 2 to December 25,

013 (52 Wednesdays), was taken out. The goal was to compare the forecasts and the actual data
action of the available data, from January 2 to December 25,
he actual data 5fortocomparison.
Figures 8 present sample forecast of interest rates per tenor and per week along with
E.P. de Lara-Tuprio, et al
he actual data
198 for comparison.

Figure 5. Forecast for 3-month tenor.

Figure 5. Forecast for 3-month tenor.
Figure 5. Forecast for 3-month tenor.

Figure 6. Forecast for 5-year tenor.

ng Head: TERM STRUCTURE FORECASTING
Figure 6. Forecast for 5-year tenor.

Figure 6. Forecast for 5-year tenor.

Figure 7. Yield curve forecast for week 1.

Figure 7. Yield curve forecast for week 1.
ing Head:Philippine
TERMInterest
STRUCTURE FORECASTING
Rates Using the Dynamic Nelson-Siegel Model 199

FigureFigure
7. Yield curve
8. Yield forecast
curve forecast for
for week 9. week 1.

the median yt (t) for each t. This, however, will still Forecast of interest rates was based on the
resort to taking a single triple of beta parameters assumption that yield curve would follow the Nelson-
from previous weeks; otherwise, the process will be Siegel equation with the same shape parameter as
computationally expensive. produced from historical data and with beta parameters
Figures 5 to 8 present sample forecast of interest generated from the time series model. The dependence
rates per tenor and per week along with the actual data structure of the beta parameters was considered in
for comparison. generating their future values. This was carried out by
It was observed that the forecasting accuracy is finding the joint distribution of the error variables via
weakened as one goes farther into the future. This is appropriate copula.
a limitation of any time series models. Thus, instead Results showed that forecast of interest rates for
of 1-year forecast, only 12 weeks were considered. different tenors, short, medium or long, was relatively
The root mean square errors (RMSE) per tenor and good up to the next three months. From then on, the
per week are shown in Tables 5 and 6. In can be seen accuracy weakened. This showed that the model,
that the best forecast is for 5-year yield, with only 6 wherein parameters were based on historical data,
Figure 8. Yield curve forecast for week 9.
bps RMSE, and the least accurate is for 10-year yield could be reliable only for the near future. For an active
with 27 bps RMSE. market, this is good enough since the models for the
parameters can be adjusted every trading day.
As mentioned earlier, a model for forecasting the
It wasConclusion
observed and Recommendation
that the forecasting accuracy is weakened as one goes farther into
term structure of interest rates is essential in designing
In this paper, the three-factor Nelson-Siegel model an optimal debt strategy for the government. This paper
was applied to forecast the term structure of interest presents one such model and how it can be applied to
e. This is a limitation of any time series models. Thus, instead of 1-year forecast, only
rates using Philippine market data. Using a fixed shape the Philippine market data. Forecast of interest rates
will help the policy makers determine the appropriate
parameter, the equation for the yield became linear in
the beta parameters. Such equation was fitted to each mix of debt instruments that will minimize cost subject
to a prudent risk level.
ks were considered.
historical dataset of zero rates thereby producing a
series of beta parameters. An appropriate time series
model was then obtained for each beta parameter.
The root mean square errors (RMSE) per Acknowledgement
tenor and per week are shown in Tables 5
Based on the historical data, the best model for each
beta was of the form ARMA(p,q)+eGARCH(1,1). It
E. de Lara-Tuprio and E. Cabral are grateful for the
is important to note that a different set of historical
financial support provided by the National Research Council
can be seen that the best forecast is for 5-year yield, with only 6 bps RMSE, and the l
data may produce a different time series model for
the beta parameters.
of the Philippines.

rate is for 10-year yield with 27 bps RMSE.

200 E.P. de Lara-Tuprio, et al

Table 5. RMSE Per Tenor

  1 Mo 3 Mo 6 Mo 1 Yr 2 Yr 3 Yr 4 Yr 5 Yr
RMSE 0.1106% 0.1556% 0.1489% 0.1071% 0.1821% 0.1460% 0.1310% 0.0635%
  6 Yr 7 Yr 8 Yr 9 Yr 10 Yr 15 Yr 20 Yr 30 Yr
RMSE 0.0830% 0.1399% 0.1921% 0.2378% 0.2743% 0.2636% 0.1507% 0.1655%

Table 6. RMSE Per Week

Week 1 2 3 4 5 6
RMSE 0.1236% 0.1280% 0.1163% 0.1100% 0.1265% 0.1389%
Week 7 8 9 10 11 12
RMSE 0.1407% 0.1491% 0.1321% 0.1640% 0.1719% 0.2246%

References Genest, C., Ghoudi, K., & Rivest, L. P. (1995). A

Annaert, J., Claes, A., De Ceuster, M., & Zhang, H. (2012).
Estimating the yield curve using the Nelson-Siegel
model: A ridge regression approach. International
Review of Economics & Finance, 27, 482–496.
Bloomberg (2013). Bloomberg Professional. [Online].
Available at: Subscription Service (Accessed: May 7,
2013).
Christensen, J. H. E., Diebold, F. X., & Rudebuscha, G. D.
(2011). The affine arbitrage-free class of Nelson-Siegel
term structure models. Journal of Econometrics, 164(1),
4–20.
Cox, J. C., Ingersoll, J. E., & Ross, S. A. (1985). A theory
of the term structure of interest rates. Econometrica, 3,
385–407.
Diebold, F. X., & Li, C. (2006). Forecasting the term
structure of government bond yields. Journal of
Econometrics, 130, 337-364.
Exterkate, P., Dijk, D. V., Heij, C., & Groenen, P. J. F. (2013).
Forecasting the yield curve in a data-rich environment
using the factor-augmented Nelson-Siegel model.
Journal of Forecasting, 32, 193–214.
semiparametric estimation procedure of dependence
parameters in multivariate families of distributions.
Biometrika, 82, 543–552.
Genest, C., Rémillard, B., & Beaudoin, D. (2009). Goodness-
of-fit tests for copulas: A review and a power study.
Insurance: Mathematics and Economics, 44, 199–213.
Koopman, S. J., Mallee, M. I. P., & Van der Wel, M. (2010).
Analyzing the term structure of interest rates using
the dynamic Nelson-Siegel model with time-varying
parameters. Journal of Business and Economic Statistics,
28, 329–343.
Nelson, C. R., & Siegel, A. F. (1987). Parsimonious modeling
of yield curves. The Journal of Business, 60, 473–489.
Vasicek, O. A. (1977). An equilibrium characterisation of
the term structure. Journal of Financial Economics, 5,
177–188.