Exhibit 1-8

The exhibit is divided into three boxes that describe steps used to create a confidence interval for the CLR:

Step one - Compare improved CLR to actual losses over time. This step is illustrated by a line graph that uses a solid line to show the CLR estimate calculated following McKinsey's recommended methodology and a dotted line to show actual losses from failures. For each quarter, you would determine the difference between the CLR projection and the actual loss.

Step two - Create distribution of CLR errors. This step is illustrated by a bar chart that plots the number of quarters with given size errors. Errors are defined as the difference between actual loss and the CLR as a percent of the CLR in a given quarter. In McKinsey's example, forty-two quarters are plotted. The chart shows the following:

Three quarters where the actual loss was 75 to 100 percent less than the CLR.
Three quarters where the actual loss was 50 to 75 percent less than the CLR.
Seven quarters where the actual loss was 25 to 50 percent less than the CLR.
Six quarters where the actual loss was 0 to 25 percent less than the CLR.
Eleven quarters where the actual loss was 0 to 300 percent greater than the CLR.
Eight quarters where the actual loss was 300 to 600 percent greater than the CLR.
Three quarters where the actual loss was 600 to 900 percent greater than the CLR.
One quarter where the actual loss was 900 percent greater or more than the CLR.

A box indicates that the four quarters where the actual loss was 600 percent greater or more than the CLR represent the largest 10 percent of errors. This largest 10 percent points to…

Step three - Report how far away the tail is. If the four quarters where the actual loss was 600 percent or more greater than the CLR represent the largest 10 percent of errors, one may assume that 90 percent of the time, the actual loss will be no more than 600 percent larger than the current CLR. If the current CLR is \$1 billion, then report \$6 billion as "additional loss that is possible but not probable in the coming year."

The source for this exhibit is FDIC and McKinsey analysis.