The Y-axis is the "after, what happened." If you think about the first firm, moving right along the X-axis: did it default? If it did, then the line is headed up straightly because X = 1 and Y = 1/total defaults. Then the second firm, did it default? If yes, the line is still going straight because X = 2 and Y = (2/total defaults) have defaulted. Then to the third, if it did not default, then line is flat here because x= 3 but y = still 2/total defaults. Model is now imperfect. At the end the model is only perfect all of the firms that actually defaulted were sorted to the top of the PD list.
Or in other words you can say that if your population has X% bads then you would expect that perfect model should help you sorting the population in such a manner that your first X% of the population is the total % bads in the population....
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.