Modelling Convergence Dynamics in Capital

This paper investigates a trading strategy which aims to exploit the relative mispricing between a firm's credit default swap (CDS) credit spread and stock price. An investor inves­tigates trading opportunities by estimating the fair value of a firm's CDS spread with a credit risk model, and enters a position in the firm's CDS and stock when there is a significant difference between model and market spreads. Problems arise as market spreads often do not converge to the fair values estimated by the credit risk model, causing large losses for investors. We use the CreditGrades model to estimate CDS spreads and investigate the dy­namic dependence between model and market spreads with a dynamic conditional correlation model, a linear cointegration model and a threshold cointegration model. In addition, we use these models to produce trading signals and incorporate these signals in trading strategies. We find that the CreditGrades model tends to underestimate CDS spreads, but do find ev­idence of a stable relation between model and market spreads, which reinforces the validity of the strategy. Our trading strategies produce lower returns than the benchmark, but also decrease the volatility of its returns and both the size and frequency of its drawdowns.