I’ve been intrigued by the Home Mortgage Disclosure Act ever since the Atlanta Journal-Constitution won a Pulitzer Prize by using its data to demonstrate a pervasive lack of lending in some of Atlanta’s inner city zipcodes. I’ve always had a big interest in community reinvestment and CRA issues, so I was glad to see a way of quantifying the numbers beyond just the anecdotal.

I’m delighted to be appearing in this blog three times a month to share with you some of the enormous number of interesting things there are in this yearly “mortgage census,” as some have called it.

I also expect to be sharing with you some of the learning curve I’m experiencing as I use ComplianceTech’s LendingPatterns™ software to navigate the millions of items of data in HMDA. In previous years, I have used ComplianceTech as a resource in my HMDA reporting, and the company has run many reports I found useful as a journalist covering the mortgage industry. This year, they have granted me access into the program so I can run mortgage reports by myself. I hope you will benefit from my voyages of discovery!

I have been analyzing HMDA data for the past 15 years or so. Some of the areas I have looked at include total mortgage originations, funded versus unfunded mortgage applications, lending to minorities, especially American Indians, credit union mortgage stats, same-gender couples lending, and minority denials by the government sponsored enterprises. I have a feeling, though, that these inquiries just scratch the HMDA surface.

It is easy to get lost in the weeds of HMDA, though, so I think it will be valuable for me to tote up some lessons learned over the years. HMDA measures not only mortgage originations, but community investments and reinvestments as well. So “mortgages purchased” is measured along with originations, denials, etc. at individual lenders. The idea is that if a lender buys mortgages in addition to the ones it makes, this is additional mortgage investment in its footprint. However, it is easy to add the originations and purchases together and come up with an inflated originations number. I’ll admit to making this mistake at least once. So if you want only the origination number, you should exclude the purchase number (it is easy to do this in LendingPatterns™). If you want total community investment, add the two together.

Related to this, mortgage sales to investors like Fannie Mae, Freddie Mac and Ginnie Mae are also toted up separately from purchases, so a secondary market analysis should look at both of those things.

Similarly, the credit union mortgage universe isn’t hard to put together, but it can’t be done in one step. There is a category on LendingPatterns™ for “NCUA Lenders.” Those are all credit unions, but they aren’t quite all the credit unions in HMDA. Mortgage lenders with assets above $10 billion are in a different category, “CFPB Lenders.” Luckily, there are only four credit unions that big (Navy and Pentagon, State Employees of North Carolina and Boeing Employees), so with just a minimum of calculator work you can put the whole thing together.

Your list of credit union mortgage lenders is now at about 2,000. But wait, aren’t there thousands more credit unions than that? There are. But by law, mortgage lenders with assets less than $43 million aren’t required to file HMDA reports, and they don’t. (This year, by the way, that cutoff number raises a tick, to $44 million.)

Now you’re ready to go with an analysis of credit unions. And, I plan to share credit union mortgage data in my next blog.

(Mark Fogarty is a journalist and analyst who has been covering the mortgage industry for more than 30 years.)