Social Trends from 2023 HMDA

​Regular readers of Recursion blogs will recall that the final version of the HMDA data[1] is generally released about mid-year and contains data of interest to those looking into important issues related to the role of housing in social trends. In this post, we will focus on both borrower and geographic breakdowns by loan class as HMDA 2023 is out. To perform the socioeconomic analysis, it is often very useful to link HMDA data with outside databases from other publicly available data sources such as FHFA Duty-to-Server, or FEMA National Risk Index, that provides various characteristics at the census tract level.
 
Low-income households
​
Ginnie Mae discloses low-to-moderate borrower exposure at the pool level.  Using its definition, Recursion constructed the flag in HMDA. We start by looking at the share that low-to-moderate income households obtained in purchase mortgage originations from 2014-2023[2].

RUN UNDERLYING QUERY

The interesting takeaway here is the narrowing of the LMI shares advantage over that has traditionally been held by Government programs (FHA, VA or RHS) compared to conventional loans. For second liens, the gap plunged by 12.5% to 7.2% in 2023, the lowest since 5.0% in 2016. The first lien gap narrowed by 6.3% to 5.2%, the lowest level obtained in over 10 years.
 
Here is the distribution by loan size:

RUN UNDERLYING QUERY

​After growing sharply from 2020-2022, the average first lien loan size for conventional loans dipped for both LMI (-4.8%) and non-LMI borrowers (-5.7%) in 2023 compared to the prior year. In comparison, the Government programs were essentially unchanged (+.025% for LMI borrowers and +0.03% for non-LMI borrowers). The ratio of loan size for non-LMI to LMI for conventional loans fell very slightly by 0.03% to 2.52 times while for Government it was unchanged at 2.66 times over the same period.
 
Finally, we look at loan rates:

RUN UNDERLYING QUERY

It’s interesting to note that the rates for conventional LMI borrowers to non-LMI borrowers edged up a bit more than Government borrowers last year.

RUN UNDERLYING QUERY

The spread of Conventional to Government Loans rose last year for LMI borrowers but reversed for non-LMI borrowers.
 
Low-income Areas
​ 
Next, we turn from a focus exclusively on borrowers to a geographic approach looking at low-income areas. These are census tracts in which the median income is at or below 80 percent of applicable AMI (area median income).[3] Ginnie Mae also discloses LMI-area share at the pool level, and Recursion is able to implement the flag at the loan level in HMDA.
 
Below, find the share of purchase loans originated in low-income areas by lien type for conventional and Government loans:

RUN UNDERLYING QUERY

​The low-income area share for government is persistently higher than that for conventional loans and, similar to the chart for LMI borrowers, the gap narrowed last year, but not in as pronounced a manner. For first-liens, the gap fell by 0.6% to a three-year low 7.0%. In the case of second liens, the gap plunged by 8.2% to a 12-year low of 8.0%.
 
We now turn to loan size:

RUN UNDERLYING QUERY

Interestingly, the average loan size for conventional mortgages in both low-income and non-low-income areas declined, while those for government mortgages increased in both categories in 2023 compared to the prior year.
 
And next, we have loan rates:

RUN UNDERLYING QUERY

This chart shows the increases in loan rates were substantially greater for conventional loans (1.79% and 1.88% respectively for low-income and non-low-income areas) compared to government loans (1.50% and 1.49%, respectively).
 
Rural Loans
​
These charts are based on the FHFA definition of rural areas, by linking census tract to FHFA Duty-to-Serve data.[4]

RUN UNDERLYING QUERY

Two observations jump out immediately. First is the strong convergence of the rural area share of conventional and government loans for both 1st and 2nd lien loans. Second is the remarkable hike of the rural share of all of the groups the 2019-2023 period.
 
As before we look next at loan sizes:

RUN UNDERLYING QUERY

In this instance, the notable development is the surge in loan sizes for first-lien conventional mortgage in both non-rural and rural areas from 2020 –2022, followed by a correction last year. By way of contrast, loan sizes for government mortgages have shown steady increases over the past five years.
 
And finally interest rates:

RUN UNDERLYING QUERY

Once again, we see bigger increases in mortgage rates in 2023 from the prior year for conventional mortgages (1.87% for both rural and non-rural areas) compared with government mortgages (1.53% and 1.48% respectively).
 
Conclusion
​
The purpose of this note is to demonstrate how combining large complex datasets can lead to new insights on recent developments in the mortgage market. Digging down into further levels of detail can lead to deeper understandings of the importance of various factors driving these complex trends. These factors include economic trends, policy changes and the ongoing fallout from the Covid shock. These simple charts represent a massive collection of borrower socioeconomic characteristics, geographies, the behavior of specific lenders and servicing firms across the spectrum of banks and nonbanks. Our HMDA Analyzer allows our clients to dig deep into these issues with a few clicks of the mouse. Let us know if you want to take it out for a spin.
 
Finally, is there more we can do with a complete HMDA dataset besides this kind of loan trend analysis? You bet, stay tuned.

---

[1] https://www.recursionco.com/blog/a-first-look-at-2023-hmda
[2] Recursion identifies a borrower as a low-to-moderate income borrower when their family income is less than 80% of the FFIEC MSA/MD median family income.
[3] FHFA Underserved Areas Data: https://www.fhfa.gov/data/underserved-areas-data
[4] FHFA Duty To Serve Eligibility Data: https://www.fhfa.gov/data/duty-to-serve/eligibility-data