Mortgage data shows that borrowers could save $100 a month (or more) by choosing cheaper lenders

In this blog we take a look at how mortgage rates paid by consumers vary across lenders. This phenomenon, called price dispersion, exists in virtually every segment of the mortgage market, including loans backed by Fannie Mae and Freddie Mac, Federal Housing Administration loans, U.S. Department of Veterans Affairs (Veterans Affairs) loans, as well as jumbo loans.1 We analyzed Home Mortgage Disclosure Act data from 2021 to quantify the magnitude of price dispersion.2

We found that price dispersion for mortgages is often around 50 basis points of the annual percentage rate. To put this number in context, the median loan amount in 2021 was close to $300,000 and the median interest rate was 3 percent.3 The monthly payment for such a 30-year fixed loan is $1,265. The monthly payment for a 3.5 percent interest rate loan on a loan of the same amount is $1,347 – a difference of $82 a month (a 6.5 percent higher payment). Interest rates have increased drastically since 2021, but the math remains similar in a higher-interest rate environment. Keeping the loan amount at $300,000, the monthly payments for a 30-year fixed loan with a 6.5 percent interest rate and a 7 percent interest rate are, respectively, $1,896 and $1,996 – a difference of $100 a month (a 5.3 percent higher payment). In a higher interest-rate environment, with monthly payments being much higher overall, this $100 a month difference might matter even more as borrowers potentially are more stretched to make ends meet.

For our analyses we use Home Mortgage Disclosure Act data from 2021. We only focus on and report the numbers for the 20 largest-volume lenders for each of the market segments.4 Several previous studies documented the existence of price dispersion in the mortgage market,5 but this is the first report to use new Home Mortgage Disclosure Act data – reported by the vast majority of mortgage originators throughout the country – with multiple variables necessary for our analysis.6 Our results are largely consistent with previous studies, despite previous studies often using either rate sheet data and/or selected data on originations from private providers without coverage comparable to the Home Mortgage Disclosure Act. It is particularly notable that earlier studies considered pre-pandemic data, which does not have the same historic volume of refinance loans. We apply a multitude of filters to the data, to make our sample as homogeneous as possible, in order to rule out many of the variables that could be responsible for price differences. We filter for:

Federal Housing Administration loans are arguably the most relevant segment for expanding homeownership as they are disproportionately used by lower-income, lower credit score borrowers, and first-time homebuyers. The Federal Housing Administration's insurance pricing also makes it easy to evaluate price dispersion while keeping everything else equal. In particular, the Federal Housing Administration's most popular loan of 3.5 percent down payment (loan-to-value ratio of 96.5) is insured by the Federal Housing Administration at the same insurance rate, regardless of the borrower's credit score or debt-to-income ratio if the borrower qualifies . Figure 1 is an interval showing the average annual percentage rate of an originated loan for each of the Federal Housing Administration's top 20 lenders across loans that meet our filters, with the cheapest lender normalized to 0. Figure 1 shows that the dispersion between the highest and the lowest top 20 Federal Housing Administration lenders is 53 basis points, and it is not driven by one outlier lender either on the low end or on the high end. Plotting origination volumes of the top 20 lenders against these average prices does not produce a clear relationship – it is not the case that the cheapest lenders systematically get much higher volumes than the pricier ones.7

Averages might work reasonably well for this Federal Housing Administration subsegment of combined loan-to-value ratio of 96.5, but even there we are mixing loans originated throughout the middle half of 2021. In contrast to FHA, the Government Sponsored Entities (Fannie Mae and Freddie Mac) use nuanced risk-based pricing (loan-level price adjustment matrix) to guarantee loans.8 Accordingly, we run pricing regressions that flexibly control for various factors, including credit score, combined loan-to-value ratio, and application date.9 We also include indicator variables for each lender which effectively capture that lender's average effect once we control for all the other variables in the regression.10

We show four figures for different segments of the market: Government Sponsored Entities, Federal Housing Administration, Veterans Affairs, and the jumbo market.11 As in the analysis above, we normalize the cheapest lender in each segment to 0 basis points. The figures show that regression-adjusted price dispersions for the four segments are, respectively, 41, 61, 64, and 57 basis points.

As noted, a 50 basis point difference amounts to over $1,000 a year for a typical mortgage. For comparison, the U.S. Department of Housing and Urban Development recently announced a 30-basis-point reduction in the Federal Housing Administration mortgage insurance premium, with the White House noting that it could save $800 a year for Federal Housing Administration consumers . Our figures suggest that borrowers choosing even the median-priced lender could save approximately the same amount if their mortgages were priced at the level of the lowest-priced large lender.12 As another comparison, 50 basis points yearly is comparable to the Government Sponsored Entities loan-level price adjustment differences between borrowers with credit scores below 620 and borrowers with credit scores above 740 – the Government Sponsored Entities adjustment difference between under 620 and over 740 borrowers is a 2.5 percent one-time charge, or $7,500 for a $300,000 loan.13 That difference is comparable with $100 a month savings (or 50 basis point interest rate difference) as long as the duration of the mortgage is around six years, which is likely an underestimate for loans originated shortly before or during the pandemic.

Why is there so much price dispersion, even though the underlying financial instrument is essentially the same (say, a 30-year fixed rate Fannie Mae loan)? There are several potential reasons:

The views expressed here are those of the authors and do not necessarily reflect the views of the Consumer Financial Protection Bureau. Links or citations in this post do not constitute an endorsement by the Bureau.