You are here

Testing the Relationship between CRA Grading and Lending to LMI Borrowers

Adam Rust's picture

Posted October 15, 2014

Does the CRA exam process encourage banks to step up their lending to low-and-moderate income borrowers? 

To test that question, I collected data HMDA data from the Federal Reserve and the Federal Financial Institutions Examination Council ("FFIEC"). Every year, economists at the Fed release their findings on new HMDA data. The FFIEC disseminates the CRA exam grades. 

My hypothesis is that the grades given out in one year will influence the degree of CRA-qualifying activities in the following year. I used home mortgage lending to LMI borrowers as a yardstick for performance. Specifically, I drew from reports released by economists at the Federal Reserve which sort home purchase loan originations by the income category of borrowers. By using shares, the ability of fluctuations in the overall volume of lending is eliminated. Because I want to see if CRA grades influence the future behavior of banks (and not if grades are reflective of prior performance),I lagged the mortgage lending data by one year from the grading results; thus, grades in 2002 are correlated with the share of home purchase mortgage lending in 2003. 

About CRA Exams

Regulators give grades to their banks and thrifts for their lending, service, and investment to a variety of traditionally-underserved constituencies. There are four grades: outstanding, satisfactory, needs to improve, and substantial non-compliance. Some banks make it an institutional priority to earn an outstanding grade; Hugh McColl has said that he considered it a baseline expectation. But clearly there are other banks that view the whole process with indifference. Witness Reynolds State Bank, a small bank in Illinois, which has received seven consecutive "substantial non-compliance" grades going back to 2006. But in my opinion, Reynolds is an outlier. I like to believe that banks want to do well in their service, lending, and investing in their communities.

The next chart shows how the distribution of CRA grades has changed over time. 

Distribution of Community Reinvestment Exam Ratings, 1997 to 2014

As you can see, the examiners have fluctuated in their general tendencies. At the beginning of the period under review here, more than 22 percent of all ratings were "outstanding" and only about 2.2 percent came in as either "substantial non-compliance" or "needs to improve." Thereafter, there were fewer "outstanding" and more "satisfactory" ratings. There was a bump in in "outstandings" in 2005, but that was a high-water mark. From then on, bad scores (substantial non-compliance or needs to improve) have increased and "outstanding" scores have become more and more infrequent. The share of outstandings given in this year's exams is only one-third the share of the set of ratings recorded back in 1997. Of course, the key change is in the rate of satisfactory. Increasingly, almost all banks get the same satisfactory rating. To date this year, 91.5 percent of all grades have been "satisfactory." 

My conclusion

The behind-the-scenes work of CRA examiners appears to have a significant impact on the share of home purchase loans that are originated to low-to-moderate ("LMI") borrowers.

The next chart shows two things: fluctuation in types of CRA grades and fluctuations in the share of home purchase mortgage loans that went to LMI borrowers. It only covers a period of ten years, as the Fed's report only covered lending from 2004 to 2013. But, as I mentioned earlier, the home purchase data lags the grading data. A 2005 score is correlated against a 2006 home purchase LMI result.

Relationship Between Home Purchase Mortgage Lending and CRA Exam Scores

There is a strong correlation between LMI home purchase market share and grading. But the positive correlation is between the bad grades and share of LMI lending; in years after the share of bad grades is high, there tend to be more loans made to LMI borrowers. In turn, in years after the share of outstanding grades is high, there tends to be a lower share of loans made to LMI borrowers. The correlation for the former (2003-13) is 0.68. For the latter, it is -0.79. 

In other words, when grading becomes too friendly, banks step back in their outreach to LMI borrowers. On the other hand, when more of their brethren have recently received bad grades, financial institutions tend to be more aggressive about making loans to LMI borrowers.