Ideally, we like to examine the historical default performance for all deals by issuer in order to determine a general pattern of defaults. We also look at the default patterns by specific vintages to determine whether vintage-specific default patterns serve as a more appropriate basis for forecasting default rates. If specific vintages deviate greatly from the overall default pattern for a particular issuer, we will use the vintage-specific pattern to forecast default rates. Once we have selected the default pattern for an issuer, often with the use of regression analysis, we then compute a peak CDR that equates the actual cumulative defaults-to-date with forecast cumulative defaults. For example, Figure 1 below shows the regression line we fit to the actual historical default rates for a particular issuer across all deals. One can see that the general default pattern rises from zero to a peak level over 24 months, levels off for another 24 months and then eventually declines to half the peak default rate by month 108 months. In applying this general pattern, we computed a peak CDR of 9.3% for the specific deal, which equates actual cumulative defaults with forecast cumulative defaults. The actual 3-month historical CDR for this deal was 14.2%.

>>>