No, because you're comparing apples and oranges. If the banks were forced to take on loans to the groups in question due to political pressure despite having been previously correct in demanding higher interest to compensate them for the loans, and also simultaneously engaged in an epic miscalculation about the safety of loans to other groups based on naive models, then you could produce this inversion.

Only if the loans were in different asset classes. Because these are all housing loans, the underlying model of the housing market is an input to both group's loans, with minority status as an additional adjustment.

Anyway, I don't think looking at CRA and pre-CRA loans to minorities you can answer discrimination questions with any kind of precision. The data seems to lean in the direction of former discrimination, but massive structural changes to finance dwarf the tiny changes created by the CRA.