| Abstrak/Abstract |
In this paper, we consider the problem of modelling the yield c
urve using Nelson-Siegel model classes.
Nelson-Siegel model classes discussed here are NS model, BL mod
el, NSS model, RF model, and our
proposed NSSE models. NSSE model is a model which extends the s
tandard NS model as Nelson-Siegel
model class by adding some linear
and non-linear parameters in
which form the fourth hump of the model
class. The purpose of adding the hump is to accommodate the pos
sibility of having the following cases: the
first, the condition when the sho
rt term and the medium term yi
elds are higher than the long term yield. The
second, the condition when the upper-value short term yields ar
e higher than both the short term yields on
average and the long term yields. The third, the case when the
upper-value medium term yields are higher
than both the medium term yields on average and the long term y
ields. These considered cases make the
yield curve more likely to have
minimum locals and therefore, t
he Nelson-Siegel model classes become
more difficult to be estimated. To overcome this problem, in th
is paper we estimate the model using the
hybrid-genetic algorithm approach and compare it with the stand
ard estimation based on NLS method. We
provide an empirical study using Indonesian Government-Bond Yie
ld Curve (IGYC) data, and found that
the best model for IGYC is 6-factors model. |
