Rating Models and Validation - Oesterreichische Nationalbank
Rating Models and Validation - Oesterreichische Nationalbank
Rating Models and Validation - Oesterreichische Nationalbank
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
<strong>Rating</strong> <strong>Models</strong> <strong>and</strong> <strong>Validation</strong><br />
smaller steps in the same direction for a given rating class, or where the probability<br />
of ending up in a certain rating class is more probable for more remote<br />
rating classes than for adjacent classes. In the transition matrix, inconsistencies<br />
manifest themselves as probabilities which do not decrease monotonically as<br />
they move away from the main diagonal of the matrix. Under the assumption<br />
that a valid rating model is used, this is not plausible.<br />
Inconsistencies can be removed by smoothing the transition matrix.<br />
Smoothing refers to optimizing the probabilities of individual cells without violating<br />
the constraint that the probabilities in a row must add up to 100%. As a<br />
rule, smoothing should only affect cell values at the edges of the transition<br />
matrix, which are not statistically significant due to their low absolute transition<br />
frequencies. In the process of smoothing the matrix, it is necessary to ensure<br />
that the resulting default probabilities in the individual classes match the default<br />
probabilities from the calibration.<br />
Chart 42 shows the smoothed matrix for the example given above. In this<br />
case, it is worth noting that the default probabilities are sometimes higher than<br />
the probabilities of transition to lower rating classes. These apparent inconsistencies<br />
can be explained by the fact that in individual cases the default event<br />
occurs earlier than rating deterioration. In fact, it is entirely conceivable that<br />
a customer with a very good current rating will default, because ratings only<br />
describe the average behavior of a group of similar customers over a fairly long<br />
time horizon, not each individual case.<br />
Chart 42: Smoothed Transition Matrix (Example)<br />
Due to the large number of parameters to be determined, the data requirements<br />
for calculating a valid transition matrix are very high. For a rating model<br />
with 15 classes plus one default class (as in the example above), it is necessary to<br />
compute a total of 225 transition probabilities plus 15 default probabilities. As<br />
statistically valid estimates of transition frequencies are only possible given a sufficient<br />
number of observations per matrix field, these requirements amount to<br />
several thous<strong>and</strong> observed rating transitions — assuming an even distribution of<br />
transitions across all matrix fields. Due to the generally observed concentration<br />
90 Guidelines on Credit Risk Management