Page 45 - 臺大管理論叢第32卷第1期

NTU Management Review Vol. 32 No. 1 Apr. 2022

4. Contribution

This study makes two main contributions to the CDS related literature. First, we
provide a better estimation technique to conduct the empirical work about liquidity risk

factor. We use an uncented Kalman filter to estimate the hazard rates under a one-factor
square root stochastic process specification while previous studies such as Chen et al.
(2010) or Lin, Liu, and Wu (2011) use a two-factor model to estimate the hazard rate
and the liquidity factor separately. We believe the empirical approaches they employ to

estimate the hazard rate and the liquidity factor are questionable because both stuides
cannot necessarily update the hazard rate estimation. In addition, estimation results
conducted by the uncented Kalman filter are reported to be more efficient and powerful
than the Maximum Likelihood Estimation (MLE) used by Chen et al. (2010), or the

Generalized Method of Moments (GMM) employed by Lin et al. (2011).
Second, this paper is the first to propose a joint estimation which combines CDS
market quotations with corporate bond yield rates to estimate hazard rates. The empirical
results demonstrate that the liquidity risk factor extracted from CDS market quotations

combined with corporate bond yield rates has more goodness of fit than the other factor
extracted purely from CDS market quotations when fitting traditional liquidity measure.
We thus conclude that risk factors generated by joint estimation can be more representative
of the market than the ones generated by separate estimation.

37