VA Lending Analysis

VA Loan Costs and Rankings
November 11, 2020

Solving Information Asymmetry

A Review of Own Up's VA Loan Data Study
To mark Veteran's Day, the company Own Up blogged today about a study they've done on VA mortgage lending. In their post, they state their wish to correct information asymmetry between sellers and buyers regarding mortgage loan pricing.

Since this is a shared goal of ours at Polygon Research, and since their post got wide exposure in an NPR feature this morning, we took a close look at Own Up's claims, and were surprised to see that not all of them hold up. This post explains what we found.

Readers of this blog know that our flagship product is the software-as-a-service app HMDAVision®. Since we believe HMDAVision® is the most rigorous, in-depth HMDA analysis tool on the market, that's where we went to evaluate Own Up's article.

The first thing that caught our eye is Own Up's statement that:
"...US Veterans...account for more than 20% of homebuyers"
While this may be true, it's important to clarify that not all veterans purchase homes with VA loans. Here's a screenshot from HMDAVision® with the Loan Purpose filter set to Home Purchase (as opposed to Refinancing, Home Improvement, etc.), where we see that the % of VA loans for home purchase in 2019 was 9.0% or 8.6%, measured by loan count or dollar volume respectively:
2019 Home purchase loans by Loan Type; Source: HMDAVision®
The next Own Up claim we checked was that VA loans:
"...close at significantly lower rates than conventional loans."
Own Up bases this claim on data from a major loan origination system vendor, where the vendor is citing statistics from a sample of their customers. But when we widen the lens to the entire HMDA data set, we in fact see that the opposite is true: VA loans have a higher closing rate - both when we consider all VA loans, as well as when we segment by Home Purchase VA loans:
Top: Loan Type by Year; Bottom: 2019 Loan type by Loan Purpose; Source: HMDAVision®
Closing Rate defined as originated loans divided by net applications.
Next, when we examined Own Up's claim regarding interest rates on VA loans:
"Lenders use this argument to justify charging higher interest rates, which results in a higher profit margin."
We found that the HMDA data doesn't back up this claim either. From HMDAVision®, we see that VA loans in fact on average had the lowest interest rates in 2019 - both for all VA loans, and for the home purchase subset:
2019 Avg Interest Rate by Loan Type - All Loans; Source: HMDAVision®
2019 Avg Interest Rate by Loan Type - Home Purchase Loans; Source: HMDAVision®
Two more things caught our eye in the Own Up article. First, they claimed:
"VA Lenders can charge some of the highest origination fees in the industry."
When looking at the entire HMDA data set for home purchases in 2019, we found that on average this does not happen:
Top: 2019 Avg Home Purchase Origination Charges by Loan Type; Bottom: Avg Home Purchase Origination Charge Basis Points by Loan Type; Source: HMDAVision®
" 20 VA lenders in 2019..."
Finally, Own Up presents a list of the top 20 VA lenders by volume from 2019 to provide an analysis of rate spread earned by these lenders. When we checked their list against HMDAVision®, we saw that their numbers are accurate, but the list itself has two errors: they included the #24 and #25 lenders, displacing 2 lenders which are in the actual top 20. Here is the complete, accurate list of top 2019 lenders by dollar volume strictly of their VA loans:
2019 Top 20 VA Lenders by Loan Volume of All VA Originations; Source: HMDAVision®
Our reason for this evaluation is to emphasize the need for a consistent, capable approach when analyzing mortgage origination activity. We know that HMDA data, which includes over 92% of US mortgage transactions each year, is the right data source for the job. But with over 77 million rows of data over the past 5 years, and over 100 fields to filter and interact with, identifying the right data source is not enough; you also need a powerful analytics engine to provide a single filter context for the data and to enable instantaneous visual drill-down and exploration…and this of course is what we've built with HMDAVision®.

For more information about HMDAVision®, please explore our website, and/or find us on YouTube.