| Summary: | This dissertation describes, implements and tests a numerical methodology to calibrate a reduced-form Credit Default Swap (CDS) pricing model to market spreads. For the intensity rate function, we independently test a set of six different functional forms. We also assume a constant recovery rate of 40%, in line with the market standard. In order to test the model accuracy, we consider a data sample comprised of daily quotes on 166 CDS curves (including the entire available structure of maturity points) on 153 reference entities across eleven sectors, over a time span of approximately twelve years, ranging from October 2004 to December 2015. The calibration results show that, overall, the Svensson functional form is the one that provides the best fit to market prices, consistently over the sample period and independently of the CDS spread magnitude. There is also a deterioration in the model accuracy during the 2007-2009 period, when CDS spreads reached extreme levels. In order to further analyze the impacts of the financial crisis, we study the historical evolution of two companies that went through great financial distress during this period. Finally, we design a simple methodology extension, in order to test if the assumption of a constant recovery rate is related with the decrease in model accuracy. This hypothesis is partially confirmed, with results confirming a better fit to market prices, at least under higher CDS spreads, with a recovery rate of around 50%, different from the market assumption of 40%.
|
|---|