There is a strong relationship between Estimated Loss and Total Deposits. However, there are several large deposit values that correspond to a zero loss for the FDIC. Since these observations are randomly located throughout the US and are in different time periods I'm assuming that there is nothing special about these points. (Hopefully they were entered in correctly and the zero values shouldn't NA's). Therefore, using this data the fitted values of the regression is:

Log(Estimated Loss) = -0.06 + 0.852* log(Total. Deposits) + Error

(.18) (.016)

RSE = 1.26

R^2=.54

From this regression I impute the values that are NA for Estimated Loss (there were 35 values that had no Deposit information so they are excluded). Below is the graphs through time of FDIC losses, log losses, and number of troubled banks.

One sees most losses are in late 1980's and lat 2000's. Surprisingly, There were no losses in 2005 and 2006. Since this period is right before the 'great recession' it perhaps shows a "This time is different mentality".

Next post I'll compare the $ value FDIC Loss with the $ value of checkable and time deposits. This should compare the actual losses of the FDIC with the potential liabilities of the FDIC.

Next post I'll compare the $ value FDIC Loss with the $ value of checkable and time deposits. This should compare the actual losses of the FDIC with the potential liabilities of the FDIC.