From November 2011 through May 2013, 30-year conventional mortgage rates dipped below 4 percent, reaching a historic low of 3.35 percent in late 2012. During this period, roughly 37 percent of all mortgages in the U.S. were refinanced. As mortgage rates rose during the summer of 2013, reaching 4.5 percent in September of that year, the pace of refinancing slowed sharply. Yet since that month, rates have again declined, officially falling below 4 percent in December 2014. This current dip represents another refinancing opportunity for homeowners, and it prompts some questions, such as: How many homeowners are well-positioned to refinance but haven’t done so yet? And are they concentrated in certain places where targeted outreach could raise their awareness about the opportunity? To explore those questions, the Community Development Department of the Minneapolis Fed analyzed a dataset called CRISM that combines consumers’ mortgage-performance data with their credit profile data. Our findings, described below, show that there are a few pockets of the Ninth Federal Reserve District where a relatively high percentage of homeowners may be missing an opportunity to save money on their mortgages.
To shape our analysis, we considered the perspectives of the two main parties involved in refinancing mortgage loans: homeowners and lenders. When determining whether refinancing makes sense financially, homeowners typically weigh a number of factors, including how long they plan to stay in their home and how much they will incur in closing costs, but usually the key consideration is how much lower a new interest rate will be compared to their current rate. A lower interest rate can help homeowners save many thousands of dollars over the lifetime of their loan. The greater the gap between the old and new rates, the greater the savings.
While the promise of reducing their monthly mortgage payments might motivate homeowners to want to refinance, lenders will likely consider several factors before agreeing to make a new loan. For instance, what is the value of the home compared to how much is currently owed? Or, looking at a borrower’s monthly expenses, how much other debt does the homeowner have? The answers to these and other questions about a borrower’s personal finances will ultimately determine whether a lender will agree to refinance a mortgage.
Thresholds and findings
To identify areas in the Ninth District where a large share of homeowners might be good candidates to refinance, we examined the CRISM database, which is a proprietary dataset produced by Equifax. As mentioned above, CRISM combines the performance data of homeowners’ mortgages with their broader credit profiles (excluding current employment status, assets, or income history).
The first step in our analysis was to use CRISM’s mortgage performance data to determine the interest rates of mortgages throughout the Ninth District, in order to identify areas where homeowners have the highest rates and thus the greatest potential to see long-term savings from refinancing to a lower rate. We found that as of December 2014, 17.5 percent of the mortgages in the Ninth District were above our “high interest” threshold, which we define here as mortgages whose interest rates are at least 1.5 percentage points higher than the prevailing available rate. At the end of last year, “high interest” meant a mortgage with a rate of 5.36 percent or higher (based off of a prevailing rate of 3.86 percent at the time). Compared to the 11 other Federal Reserve districts, the Minneapolis Fed’s district boasts the smallest share of high-interest loans in the country. But the shares of high-interest loans within the district’s 57 micro- and metropolitan statistical areas vary, sometimes significantly, from a low of 13 percent in Helena, Mont., to a high of 41 percent in Sault St. Marie, Mich.
Next, we used CRISM to identify which of those high-interest loans would be most likely to meet lenders’ criteria for refinancing. To do so, we applied additional thresholds similar to those that lenders themselves might apply when deciding whether they would agree to refinance, including whether the homeowner:
- Is current on his or her mortgage payments,
- Doesn’t have a second mortgage on the house,
- Is current on all other debts,
- Has a credit score of at least 680, and
- Currently has a mortgage balance of less than 90 percent of the value of the home.
(For more details about our analysis, see the sidebar “More about our methodology” below.)
If borrowers with high-interest mortgages also met our additional thresholds, we considered them reasonable candidates for refinancing. We calculated that across the Ninth District as a whole, only 8.2 percent of mortgage-holding homeowners met all of our thresholds. However, a few communities in the Upper Peninsula of Michigan, as well as one in Minnesota and one in South Dakota, contained a relatively high share of households that, by our definition, seem like reasonable refinancing candidates.
Leading the group is Houghton, Mich., where 17.0 percent of mortgage-holding homeowners meet our thresholds. Also in the Upper Peninsula are Escanaba (16.6 percent) and Sault Ste. Marie (15.6 percent). One potential cause of the increased rates in this area is the relatively elevated level of unemployment (and therefore decreased level of income), which our analysis doesn’t measure. The unemployment rate for the Upper Peninsula of Michigan in December 2014 was 6.5 percent, compared to a national rate of 5.4 percent at that time.
At 3.0 percent, unemployment doesn’t appear to be a problem in the Huron, S.D., metropolitan statistical area (MSA). In that community, 16.3 percent of mortgage-holding homeowners meet our thresholds for being refi-eligible. And in the Worthington, Minn., MSA, where unemployment is even lower, at 2.8 percent, 15.8 percent of mortgage-holding homeowners appear capable of refinancing, based on our thresholds.
A notable aspect about Worthington is that, of all the communities examined in this analysis, it contains the highest share (14.8 percent) of mortgage-holding homeowners who have delinquent debt that is not associated with their first mortgage (and therefore do not meet our definition of refinance-eligible homeowners) but are otherwise current on their mortgage payments. This means that if these delinquent debts, attributable to things such as credit cards or car loans, could be retired or at least kept up-to-date, the share of mortgage-holding homeowners seemingly capable of refinancing would climb even higher.
Information for helping homeowners
Identifying places that have concentrations of refi-eligible homeowners may have practical applications for nonprofit organizations and financial institutions that serve those communities. For example, Lisa Graphenteen, chief operating officer at Southwest Minnesota Housing Partnership—which supports sustainable homeownership opportunities through educational services, among other activities—says that news of Worthington’s relatively high percentage of homeowners with delinquent non-mortgage debt could help inform her organization’s work in the Worthington area.
“Despite this area having low unemployment rates, it also has census tracts with high poverty levels of 22.3 percent and 39.2 percent, so people may be relying on credit cards,” she speculates. “This information provides a great opportunity for our organization to examine how we can market some of our educational courses, like financial literacy, budgeting, and mortgage counseling.”
For more analyses of mortgages in the Ninth District, visit the Minneapolis Fed’s Housing Market and Mortgage Conditions web page.
More about our methodology
We conducted this analysis using Equifax’s Credit Risk Insights Servicing McDash (CRISM) dataset. CRISM links McDash mortgage performance data provided by Black Knight Financial Services to an individual’s anonymized consumer credit profile. McDash is made up of data from several of the nation’s largest mortgage servicers but its coverage of the mortgage market is not comprehensive. The overall representativeness of McDash data, and consequently CRISM data, varies by loan product and over time.
We focused on mortgages as of December 2014 that at the time of origination were both first liens and for owner occupied properties. For each of these loans, we determined whether the primary borrower had any additional mortgages on their credit record (such as closed-end second mortgages, home equity lines of credit, or first mortgages on other properties). Records for individuals with multiple first mortgages are excluded due to the difficulty in assigning any second-lien mortgages to a particular property. We also exclude records where the balances reported in the Equifax data are not within 3 percent of their post-merge McDash equivalents.
To calculate the appraised value of the property securing the mortgage, we took the value available at the time of origination and updated it using CoreLogic’s Home Price Index (HPI) to determine the value of the property as of December 2014. The HPI is reported at various geographic levels (state, metropolitan area, county, and ZIP Code) and we used the most geographically precise level available for each mortgage record to estimate the updated appraised value. As a general rule, more populous areas have HPIs available at more precise geographic levels.
Prevailing mortgage rates were taken from the 30-year conventional mortgage rate series on the Federal Reserve Bank of St. Louis’s FRED (Federal Reserve Economic Data) web page at research.stlouisfed.org/fred2/series/MORTG. We compared the December 2014 value of this series, 3.86 percent, to the current interest rate on each outstanding mortgage and classified loans whose interest rates were at least 1.5 points higher than 3.86 as “high-interest” mortgages.
Share of Mortgages that Are Both High-Interest and Reasonable Candidates for Refinancing
The table below lists the top three results for each state in the Ninth Federal Reserve District. (Results for Michigan and Wisconsin reflect only the portions of those states that lie in the District.)
Source: Authors’ calculations based on Equifax's CRISM (Credit Risk Insights Servicing McDash) dataset.
 Authors’ calculations using data from Black Knight Financial Services McDash.
 The Michigan unemployment figure is from the Michigan Department of Technology, Management, and Budget’s Bureau of Labor Market Information and Strategic Initiatives; the national figure is from the U.S. Bureau of Labor Statistics.
 For the entire district, 9.2 percent of mortgage-holding homeowners fall in this category.
 The figures Graphenteen cites are from Minnesota Housing Finance Agency Community Profiles that are generated in PolicyMap and based on 2007–2011 data from the U.S. Census Bureau’s American Community Survey. For more information, visit mnhousing.gov and enter “Community Profiles Interactive Map” in the search box.