Cityscape: A Journal of Policy Development and Research - National Survey of Mortgage Originations

by Tim Critchﬁeld, Jaya Dey, Nuno Mota, and Saty Patrabansh

Cityscape: A Journal of Policy Development and Research • Volume 21, Number 2 • 2019

U.S. Department of Housing and Urban Development • Ofﬁce of Policy Development and Research

Cityscape

Mortgage Experiences of Rural

Borrowers in the United States:

Insights from the National Survey

of Mortgage Originations

Tim Critchﬁeld

Consumer Financial Protection Bureau

Jaya Dey

Freddie Mac

Nuno Mota

Fannie Mae

Saty Patrabansh

Federal Housing Finance Agency

Disclaimer

The views expressed in this article are those of the authors and are not necessarily those of

the Consumer Financial Protection Bureau, Freddie Mac, Fannie Mae, or the Federal Housing

Finance Agency.

Abstract

To date, research on rural mortgage markets in the United States has been limited by a lack of data on

the speciﬁc mortgage experiences of borrowers living in rural areas. To ﬁll this data gap, the National

Survey of Mortgage Originations (NSMO) conducted a survey that oversampled people who took out

mortgages in completely rural counties in 2014. This article shows results from this survey, contrasting

the characteristics, experiences, and loan terms of mortgage borrowers in completely rural counties

to those of borrowers in metropolitan and other non-metropolitan areas. Completely rural counties

are those with no urban cluster or an urban population less than 2,500. We ﬁnd that borrowers in

completely rural counties paid slightly higher interest rates on average and were less satisﬁed that their

mortgage was the one with the best terms to ﬁt their needs than borrowers in other areas. These results

persist even after controlling for income, credit quality, and other borrower characteristics. Completely

rural borrowers were less likely than other borrowers to be satisﬁed with the mortgage closing process,

the timeliness of disclosures, and the disclosure documents themselves. Finally, compared with borrowers

in more urban areas, borrowers in completely rural areas tend to be less conﬁdent or less knowledgeable

about some details of mortgages, and they are more likely to initiate contact with their lender.

National Survey of Mortgage Originations

Critchﬁeld, Dey, Mota, and Patrabansh

Introduction

There is a widespread belief that lenders and credit markets in rural areas of the United States differ

from those in urban areas. The literature on “relationship lending” argues that lending in rural

areas differs fundamentally from lending in other areas because rural lenders have greater personal

knowledge about their borrowers and local economic conditions.

The literature on community

banking shows that lenders in rural areas tend to have fewer assets and generally have smaller

geographic markets than larger financial institutions (Critchfield et al., 2004). The smaller scale of

lending in rural areas could potentially constrain the supply of mortgages, make mortgages more

costly to originate, and could adversely affect mortgages taken out by borrowers.

To some degree, federal housing and mortgage policies reflect the distinctive features and challenges

of mortgage lending in rural areas. For example, the Housing and Economic Recovery Act of 2008

(HERA) assigned Fannie Mae and Freddie Mac a Duty to Serve (DTS) three underserved markets

by increasing the liquidity of mortgage investments and improving the distribution of mortgage

investment capital to those markets.

One of the underserved markets specifically mentioned is the

rural market, and another is manufactured housing—which is much more prevalent in rural areas

(CFPB, 2014). Similarly, the Consumer Financial Protection Bureau (CFPB) provided exceptions for

loans made by small creditors that operate predominantly in rural or underserved areas in several of

its mortgage rules.

This article explores the underlying premise of these government policies: that mortgage borrowers

in rural areas are potentially “less well-served” than those in other areas. Such an article has been

difficult to write in the past because of a lack of data.

There are few available data sets containing

detailed information on the characteristics of borrowers and their mortgages that are both

representative and contain enough borrowers in rural areas to make meaningful comparisons. For

example, data reported under the Home Mortgage Disclosure Act (HMDA) provide data on millions

of mortgages and thus include many rural loans, but HMDA exempts small lenders and lenders with

branches exclusively outside metropolitan areas, making the coverage of rural loans incomplete

and potentially unrepresentative.

To fill the need for data on mortgages and mortgage borrowers in rural areas, the Federal Housing

Finance Agency (FHFA) and CFPB conducted a special supplemental survey of 1,000 mortgages

as part of the National Survey of Mortgage Originations (NSMO) which is based on the National

Mortgage Database (NMDB

The supplemental sample that is the cornerstone of this article covers

mortgages originated at any time in 2014 in counties defined as “completely rural.” This article uses

data for 267 borrowers from the special sample of completely rural mortgages and 6,273 respondents

to the regular NSMO survey who took out a mortgage in 2014.

See, for example, https://www.stlouisfed.org/bank-supervision/2013-community-banking-conference/videos/small-

business-lending-and-social-capital-are-rural-relationships-dierent or https://www.stlouisfed.org/Publications/

Regional-Economist/April-2002/Community-Ties-Does-Relationship-Lending-Protect-Small-Banks-When-the-Local-

Economy-Stumbles.

See HAC (2012) for a description how mortgage ﬁnance in rural communities has evolved over the last decade.

See 12 U.S.C. 4565(a)(1).

11Cityscape

Mortgage Experiences of Rural Borrowers in the United States:

Insights from the National Survey of Mortgage Originations

NSMO data allow us to compare the expectations, perceptions, knowledge, experience, and

satisfaction of respondents in completely rural areas with those respondents in other rural areas, as

well as urban areas. The remaining sections of the article summarize the relevant literature; define the

three geographical groups used in the analysis; and describe the geographic differences in property,

loan, and borrower characteristics, borrowers’ experience and knowledge, and borrower’s lender and

mortgage choice.

Literature

This article bridges two largely distinct literatures, namely, (1) studies of rural housing and mortgage

markets, and (2) the literature on borrower perceptions and experiences within the mortgage market.

The first set of studies have described the trends in employment, incomes, and housing market

characteristics across the urban-rural divide (HAC, 2012; Mota, 2016; USDA, 2016) along with the

effect and prevalence of government-sponsored enterprises in rural markets (Ambrose and Buttimer,

2005; MacDonald, 2001; Vandell, 1997). These studies compared rural and urban housing and

mortgage markets at an aggregate level but did not consider differences in the characteristics of

borrowers across rural and urban areas.

Because these studies did not consider the differences in characteristics between rural and urban

borrowers, they generally find differences in borrower perceptions and experience by income,

race and ethnicity, and prior experience with mortgages, rather than across area distinctions.

Bucks and Pence (2008) showed that consumers were aware of the general terms of their mortgage

(for example, adjustable-rate versus fixed-rate mortgages, number of months of required payment,

and required monthly payment amount); they were less aware of details about potential interest rate

changes in their adjustable-rate mortgages. Huang et al. (2017) reported that there were common

misperceptions among borrowers about mortgage qualification criteria; specifically, borrowers

tended to overestimate minimum credit score and down payment requirements. Cai and Shahdad

(2015) found that one-third of homebuyers in their sample did not obtain quotes from multiple

lenders when shopping for a mortgage. In addition, Cai and Shahdad concluded that borrowers with

higher incomes, younger borrowers, and minority borrowers were more likely to obtain multiple

lender quotes compared with other borrowers; while lower-income and first-time homebuyers were

particularly likely to use family, friends, and co-workers when selecting a lender.

This lack of data has limited the academic and policy research on rural mortgage markets, and both Fannie Mae and

Freddie Mac cite lack of data and research on rural mortgage lending as one of the critical issues in rural mortgage

markets. See Freddie Mac’s “Freddie Mac Duty to Serve Underserved Markets Plan” from December 2017 and Fannie

Mae’s “Introduction of the Duty to Serve Underserved Markets Plan for the Manufactured Housing, Aordable Housing

Preservation, and Rural Housing Markets” from December 2017.

NSMO and NMDB are described in the guest editor’s introduction.

While the NSMO data for the regular sample is publicly available, the data for the supplemental NSMO sample is not

publicly available for privacy considerations because membership in the supplemental sample conveys information

about the location of the borrowers, which is suppressed in the public use NSMO. Out of the 1,000 surveys mailed in

the supplemental survey, only 267 were useable records.

National Survey of Mortgage Originations

Critchﬁeld, Dey, Mota, and Patrabansh

Deﬁnition of Urban and Rural Areas

We classify U.S. counties into three groups based on the U.S. Department of Agriculture’s (USDA)

county-level Rural Urban Continuum Code (RUCC) classification from 2013.

The first group,

which we refer to as “metro” counties, consists of 1,167 counties in Metropolitan Statistical Areas

(MSA) (RUCC codes 1, 2, or 3).

The second group includes 1,332 counties that are not in

metropolitan areas, but have at least one urban cluster of 2,500 or more people (RUCC codes

4, 5, 6, or 7)—many of these counties, which we refer to as “non-metro” counties, are in

Micropolitan Statistical Areas.

The last group, which we refer to as “completely rural” counties,

comprises 644 counties that are designated as “completely rural or less than 2,500 urban

population” (RUCC codes 8 or 9). The supplemental NSMO sample used in this article was

drawn from mortgages in completely rural counties.

Using this classification, 37 percent are metro counties, 42 percent are non-metro counties, and

21 percent are completely rural counties. Exhibit 1 shows that most counties along the coasts are

metro counties.

Non-metro counties are spread throughout the country, but the coasts have the

lowest share ranging from 33 percent of counties in the Middle Atlantic Census division to

35 percent in the Pacific division. Completely rural counties are primarily located in parts of

the Midwestern, Mountain, and Southern states.

Exhibit 2 shows the share of the 2014 population and housing units in each of the three county types

according to the Census Bureau’s American Community Survey (ACS) 2010–2014 5-Year Estimates.

While there are many deﬁnitions of “rural” areas in the literature and government programs, including DTS, this article

arises out of the supplemental NSMO sample and is restricted to how the sample was drawn. See USDA documentation

at https://www.ers.usda.gov/data-products/rural-urban-continuum-codes/documentation/.

MSAs have at least one urbanized area of 50,000 or more population, plus adjacent territory that has a high degree of

social and economic integration with the core as measured by commuting patterns.

Micropolitan Statistical Areas have at least one urban cluster of at least 10,000 but less than 50,000 population, plus

adjacent territory that has a high degree of social and economic integration with the core as measured by commuting

patterns. For the USDA, “non-metro” would include RUCC codes 8 and 9, but we have broken them out for the

purpose of this study.

The Census Bureau divides the country into four regions and nine divisions.

13Cityscape

Mortgage Experiences of Rural Borrowers in the United States:

Insights from the National Survey of Mortgage Originations

Exhibit 1

Geographic Classication of Counties and County Equivalents

Source: USDA RUCC data, 2013

Most people and housing units in the United States—86 percent of people and 85 percent of housing

units—were in metro counties in 2014. The remaining population and housing units were primarily

located in non-metro counties. Completely rural counties accounted for only 2 percent (4.6 million

people) of the U.S. population and 2 percent (2.6 million housing units) of housing units in 2014.

The final bars in exhibit 2 show the share of mortgage originations by geographic location. The

share of mortgage originations in metro counties is 88.4 percent, a few percentage points higher

than the share of people and housing units in these areas. Furthermore, just 1 percent of originations,

or 55,000 loans, were in completely rural counties.

This highlights the importance of drawing a

supplemental

sample of mortgages in completely rural counties in order to obtain more accurate

estimates.

In particular, about 5 percent of our sample (345 out of 6,540) is for mortgages in

completely rural areas, a sample size that is five times larger than a simple random sample of the

country would be expected to yield for these areas.

County-type shares of NMDB 2014 originations in the NMDB data are similar to those in HMDA 2014 data, where

metro accounts for 89.4 percent of originations, non-metro 9.7 percent, and completely rural 1.0 percent of originations.

Analysis in this article, including the regressions, is based on analytic weights that account for both the sampling

weight and the non-response adjustment. For each survey response, the sample weight adjusted for non-response

was computed by multiplying the sampling weight and the non-response adjustment. Then, the analytic weight was

computed separately for each of the following three groups: (1) mortgages in completely rural counties included in

the special supplementary sample (267 mortgages), (2) mortgages in completely rural counties included in the regular

sample (78 mortgages), and (3) mortgages in non-metro and metro counties included in the regular sample (6,273

mortgages). The analytic weight for a survey response was computed by multiplying the non-response-adjusted sample

weight of that survey response by the total sample size of the group and dividing it by the sum of the non-response-

adjusted sample weight of that group. The total number of completely rural mortgages in the analysis sample is 345 (267+78).

National Survey of Mortgage Originations

Critchﬁeld, Dey, Mota, and Patrabansh

The supplemental sample uses the same questionnaire as the regular sample for 2014 mortgages with a few exceptions

noted later.

Single-family attached dwellings include townhouses, row houses, villas, apartments, and multi-unit dwellings. In the

NSMO survey, mobile and manufactured homes are identiﬁed as manufactured housing because all such homes built

after 1976 are deﬁned by the HUD as manufactured housing.

Geographic Differences in Property, Loan, and Borrower

Characteristics

The primary data we use are survey responses from the regular quarterly NSMO samples and

a special supplemental sample of borrowers in completely rural areas.

NSMO data cover a

representative sample of first-lien residential mortgages taken out since 2013.

Exhibit 3 shows property and mortgage characteristics for our sample. Because purchasers and

refinancers may have different expectations, knowledge, or experience, we present results for all

originations, purchase mortgages, and refinance mortgages, separately.

Compared with those in metro areas, properties associated with mortgage originations in

completely rural and non-metro areas were less likely to be single-family attached dwellings,

and were more likely to be manufactured homes.

Manufactured housing is often titled as chattel

(personal property) even though about three-fifths of manufactured-housing residents own the

Metro Counties Non-Remote Non-Metro Counties Remote Non-Metro Counties

POPULATION

HOUSING

UNITS

MORTGAGE

ORIGINATIONS

STUDY

SAMPLE

PERCENT

12.9

1.5

2.0

1.0 5.3

85.6

83.4

88.4

84.7

14.7

10.6

10.0

Exhibit 2

Population, Housing Unit, Mortgage Originations, and Sample Size by County Type, 2014

Note: Study sample includes a special oversample of mortgages originations in completely rural counties.

Sources: Census Bureau ACS 2010–2014 5-year estimates data for population and housing units by county; NMDB

data for 2014 ﬁrst-lien mortgage

originations; NSMO 2014 data for sample sizes by county

15Cityscape

Mortgage Experiences of Rural Borrowers in the United States:

Insights from the National Survey of Mortgage Originations

Characteristics

All Mortgages Purchase Mortgages Reﬁnance Mortgages

Metro

(%)

(n=5,541)

Non-

Metro

(%)

(n=654)

Completely

Rural

(%)

(n=345)

Metro

(%)

(n=3,044)

Non-

Metro

(%)

(n=348)

Completely

Rural

(%)

(n=161)

Metro

(%)

(n=2,497)

Non-

Metro

(%)

(n=306)

Completely

Rural

(%)

(n=184)

Purchase Mortgage Share

55 52 53 NA NA NA NA NA NA

Property Type

Single-Family Detached 83 86 86 82 88 89 84 85 83

Attached 16 7 2 17 8 2 15 5 2

Mobile or Manufactured 1 6 9 1 2 5 1 10 13

Land 0 1 3 0 2 4 0 0 2

Loan Amount

Less than $50,000 3 10 14 2 8 10 4 12 17

$50,000 to $149,999 35 60 55 34 60 60 38 59 50

$150,000 to $299,999 39 25 27 41 27 29 37 24 25

$300,000 or More 22 5 4 23 5 1 22 5 8

Mortgage Term to Maturity

30 Years or More 76 65 59 89 82 78 60 47 38

Median Terms (in Years) 30 30

30 30 30 30 30 20 15

Loan-to-Value (LTV) Ratio (Median) 80 80 80 88 90 95 73 74 73

Household Income

Less than $35,000 6 12 10 6 12 8 7 11 13

$35,000 to $49,999 10 15 21 11 16 20 9 13 21

$50,000 to $74,999 18 23 28 20 21 30 16 24 25

$75,000 to $99,999 18 25 16 18 21 14 19 28 17

$100,000 to $174,999 30 20 20 29 21 21 31 18 20

$175,000 or More 17 7 5 17 8 7 17 5 3

Household Employment

One or More Full-Time 87 83 81 90 86 86 84 79 75

None Full-Time 13 17 19 10 14 14 16 21 26

Household Type

Couple 76 78 75 76 75 76 75 81 74

Single 24 22 25 24 25 24 25 19 26

Respondent Race/Ethnicity

Non-Hispanic White 77 90 92 77 89 94 77 91 91

Hispanic and Non-White 23 10 8 23 11 7 23 9 9

Exhibit 3

Property, Mortgage, and Borrower Characteristics by Mortgage and Market Type (1 of 2)

National Survey of Mortgage Originations

Critchﬁeld, Dey, Mota, and Patrabansh

Characteristics

All Mortgages Purchase Mortgages Reﬁnance Mortgages

Metro

(%)

(n=5,541)

Non-

Metro

(%)

(n=654)

Completely

Rural

(%)

(n=345)

Metro

(%)

(n=3,044)

Non-

Metro

(%)

(n=348)

Completely

Rural

(%)

(n=161)

Metro

(%)

(n=2,497)

Non-

Metro

(%)

(n=306)

Completely

Rural

(%)

(n=184)

Respondent Age

35 or Younger 28 29 30 40 39 45 13 17 13

36 to 45 23 24 23 23 24 25 23 24 20

46 to 55 23 19 17 17 16 10 29 23 24

56 to 65 16 17 19 12 14 12 22 21 25

66 or Older 10 11 12 7 8 8 13 15 17

Median 45 45 43 39 39 37 51 49 52

Respondent Education

Some School 1 2 3 1 2 3 2 1 2

High School 10 21 19 8 13 14 11 29 26

Technical School 5 9 8 4 8 8 5 9 8

Partial College 20 23 21 19 24 21 22 22 22

College Degree 36 29 35

38 33 40 34 25 30

Postgraduate 28 17 14 29 20 14 26 13 13

Respondent Credit Score

Lower Than 620 5 10 9 4 7 9 6 14 8

620 to 639 4 5 6 4 6 4 4 3 8

640 to 659 6 5 9 5 6 12 7 4 5

660 to 679 6 8 8 6 7 7 7 9 10

680 to 699 8 10 9 8 10 9 8 10 9

700 to 719 8 9 12 8 10 11 7 9 14

720 to 739 10 10 12 11 9 17 10 11 7

740 or Higher 52 43 35 54 46 31 50 40 39

Median 744 725 716 747 729 717 739 720 714

Exhibit 3

Property, Mortgage, and Borrower Characteristics by Mortgage and Market Type (2 of 2)

NA = data not available.

Notes: Attached properties include townhouses, row houses, villas, apartments, and multi-unit dwellings. VantageScore

3.0 credit score ranges from 300 to 850.

Source: NSMO, 2014

17Cityscape

Mortgage Experiences of Rural Borrowers in the United States:

Insights from the National Survey of Mortgage Originations

land it is sited on (CFPB, 2014). Because chattel loans generally are not identifiable as mortgages

in the NMDB credit files, the NSMO sample misses most chattel loans and thus substantially

undercounts loans for manufactured housing. We found the percentage of initial purchases for

manufactured housing much lower than for resale or any refinancing transactions for the same.

Underrepresentation of manufactured housing may not be as severe for refinance mortgages or

purchase mortgages by repeat buyers. Homeowners may be able to obtain mortgage refinancing on

manufactured housing originally titled as chattel after a title change if the house is on a permanent

foundation and especially if they also own the underlying land.

Generally, mortgage loan amount was lowest in completely rural counties and highest in metro

counties. In addition, loans with terms of less than 30 years were more common in completely

rural areas. This partially reflects the greater share in completely rural areas of manufactured

housing loans, which tend to have shorter terms. Nonetheless, these differences in loan terms

persist even when manufactured housing are excluded from the samples. Purchase loans in

completely rural counties had a higher median loan-to-value (LTV) ratio.

This is partly explained by the slightly higher share of purchasers in completely rural counties

that were first-time homebuyers, who tend to have loans with higher LTVs.

NSMO is a representative sample of mortgages, but the survey is answered by a single respondent

who may be one of multiple borrowers. Seventy-five percent of mortgage loans involve multiple

borrowers. For simplicity, we refer to the respondents as “borrowers.” In characterizing borrowers,

we focus on age, household type, number of co-borrowers, race and ethnicity, education,

employment, income, and credit scores.

Overall, borrowers in completely rural areas had lower incomes, were less likely to be employed

full-time, and were more likely to identify as non-Hispanic White. Age differences across areas

were small, though purchasers were slightly younger and refinancers were slightly older in

completely rural counties than in metro and non-metro areas. Educational levels differed more

noticeably with metro borrowers more likely than others to have a graduate degree. Although

median credit scores were only slightly lower in completely rural counties than in more populous

areas, the share of borrowers with a credit score of at least 740 was much lower in completely

rural counties (35 percent) than in non-metro (43 percent) and metro (52 percent) counties.

According to the Manufactured Housing Resource Guide by the National Consumer Law Center, approximately

three-quarters of the states have statutes that set forth a procedure to convert a manufactured home from personal to real

property and document that conversion. See https://www.nclc.org/images/pdf/manufactured_housing/cfed-titling-homes.pdf.

The LTV ratio is obtained by dividing the mortgage loan amount by the property value, and the dierence between

100 percent and the LTV indicates the borrower’s share of equity on the property. A typical mortgage will have an LTV

of 80 percent, which indicates that the borrower has 20 percent equity in the property.

First-time purchases are purchase mortgages taken out by borrowers who were younger than 55 years of age,

who did not have any record of having a prior mortgage in the NMDB data, and who were buying a house they will

primarily live in.

VantageScore

3.0 is the credit scoring model developed jointly by Equifax, Experian, and TransUnion. The NMDB

contains VantageScore

3.0 credit scores that range from 300 to 850. Generally, the higher a borrower’s credit score is,

the less risky the borrower is assumed to be. See https://your.vantagescore.com/resource/52/understanding-credit-scores.

National Survey of Mortgage Originations

Critchﬁeld, Dey, Mota, and Patrabansh

The smaller share of completely rural borrowers with scores of 740 or greater suggests that loans

in completely rural areas were likely to have higher interest rates because the best rates are typically

offered to borrowers with scores above 740. To test this conjecture, we examine whether there

were statistically significant differences in Freddie Mac’s Primary Mortgage Market Survey (PMMS

)

spread across the analysis groups (metro, non-metro, and completely rural).

We consider both the

observed difference in PMMS spreads as well as an adjusted difference based on a regression model

that accounts for borrowers, property, and loan attributes:

PMMS_Spread

i,c

= α +

+ βX

+ ε

i,c

The equation models the PMMS spread of the loan originated to borrower i in analysis group c

(PMMS_Spread

i,c

) on indicator variables of whether analysis group c is completely rural

(CR

) or non-metro (NM

), as well as a vector of borrower, property, and loan attributes (X

The coefficient estimates for

and

indicate the average difference in the PMMS

spread for loans originated in completely rural and non-metro counties relative to metro counties.

Exhibit 4 displays the coefficient estimates and indicators of statistical significance for the sample of

all mortgages, and then separately for purchase and refinance mortgages. This exhibit also provides

the borrower, property, and loan attributes that are used as control variables for estimating the

differences between counties.

Borrowers in completely rural counties indeed paid a slightly higher

interest rate than borrowers in metro areas,

and the spread over the PMMS rate for borrowers in

completely rural counties was 14 basis points higher than metro areas with a statistically significant

difference. The results imply that completely rural borrowers paid 16 basis points more than

non-metro borrowers. While 14 basis points appears small for the average of completely rural

counties, a mortgage of $100,000 with a 4.00 percent rate that moved to 4.14 percent would cause

the monthly payment to rise $8.10, or $97.20 per year, and cost the borrower $2,916.00 over the

life of the loan. For purchase mortgages, the differences between completely rural and metro areas

was 24 basis points and statistically significant, but the refinance mortgages difference was not

significant. Notably, in this regression framework, the PMMS spread does not vary significantly

with most borrower characteristics, whereas the purchase flag, loan amount, property type, and

credit score are statistically significant predictors of the interest rate spread.

Geographic Differences in Borrowers’ Experiences

and Knowledge

In examining differences in mortgage borrowers’ experience and knowledge by geography, we

use the framework of the PMMS equation and exhibit 4 to examine borrowers’ self-reported

satisfaction, knowledge, and lender selection. More specifically, exhibit 5 shows the results of the

NSMO asking borrowers how satisfied they were with several aspects of their mortgage and the

Freddie Mac publishes the average PMMS rate by mortgage term on a weekly basis. The PMMS spread is calculated as

the dierence between the actual note rate of a mortgage and Freddie Mac’s PMMS average prime oer note rate for that

term at that time. This spread indicates how expensive a mortgage is compared to the average mortgage of similar term

taken out in that week. The spread used in this article is unbounded while the spread in the NSMO public use ﬁle is

bounded for privacy considerations.

Some of the controls, such as race, ethnicity, age and gender, are characteristics that lenders do not or cannot use in

loan pricing models. We include them to account indirectly for unmeasured characteristics that may be correlated with

these controls.

The R-squared for the models ranged from 0.06070 to 0.08412.

19Cityscape

Mortgage Experiences of Rural Borrowers in the United States:

Insights from the National Survey of Mortgage Originations

NA = data not available.

* denotes signiﬁcance at 10-percent level. ** denotes signiﬁcance at 5-percent level. *** denotes signiﬁcance at 1-percent level.

Notes: The controlled difference model has the following covariates for the respondent: age, race/ethnicity, credit score, and education; and household: household

type, annual income, mortgage loan amount, and property type. The controlled difference model intercept reﬂects the percentages for respondents who were 36

to 45 years old, non-Hispanic White, and in a coupled household; who had a college degree or higher, credit score of 740 or higher, annual household income

from $75,000 to $99,999, and mortgage loan amount from $150,000 to $299,999; and who lived in a single-family detached house.

Survey question: “What is the interest rate on this mortgage?”

Source: NSMO, 2014

All Mortgages

Purchase

Mortgages

Reﬁnance

Mortgages

Average for Metro (M) (Percentage Points) 0.22 0.19 0.27

Controlled Model

Parameter Categories Estimate Estimate Estimate

Intercept 0.22***(0.03) 0.15***(0.03) 0.23***(0.03)

Non-Metro (NM) – 0.03 (0.03) – 0.04 (0.04) – 0.02 (0.04)

Completely Rural (CR) 0.14***(0.04) 0.23***(0.05) 0.05 (0.05)

Purchase Mortgage – 0.06***(0.02) NA NA

Property Type: Mobile or Manufactured 0.10*(0.06) 0.16*(0.10) 0.06 (0.07)

Attached 0.06**(0.02) 0.02 (0.03) 0.11***(0.03)

Land 0.06 (0.12) 0.03 (0.03) NA

Loan Amount: Less Than $50,000 0.31***(0.04) 0.37***(0.06) 0.27***(0.06)

$50,000 to $149,999 0.12***(0.02) 0.13***(0.03) 0.11***(0.03)

$300,000 or More – 0.15***(0.02) – 0.11***(0.03) – 0.19***(0.03)

Reported Household

Income:

Less Than $35,000 0.01 (0.04) 0.008 (0.05) 0.02 (0.05)

$35,000 to $49,999 – 0.03 (0.03) – 0.05 (0.04) 0.001 (0.05)

$50,000 to $74,999 – 0.01 (0.03) – 0.04 (0.03) 0.02 (0.04)

$100,000 to $174,999 – 0.01 (0.02) – 0.004 (0.03) – 0.03 (0.03)

$175,000 or More – 0.07**(0.03) – 0.11***(0.04) – 0.04 (0.04)

Household Type: Single – 0.001 (0.02) – 0.007 (0.03) 0.006 (0.03)

Respondent

Race/Ethnicity:

Hispanic and Non-White – 0.03 (0.02) – 0.03 (0.03) – 0.02 (0.03)

Respondent Age: 35 or Younger – 0.01 (0.02) – 0.002 (0.03) 0.008 (0.04)

46 to 55 – 0.01 (0.02) 0.05 (0.03) – 0.08**(0.03)

56 to 65 – 0.05*(0.03) – 0.02 (0.03) – 0.08**(0.03)

66 or Older – 0.04 (0.03) 0.03 (0.05) – 0.10**(0.04)

Respondent Education: Some School 0.03 (0.07) 0.07 (0.10) 0.008 (0.10)

High School 0.06**(0.03) 0.06 (0.04) 0.06 (0.04)

Technical School 0.05 (0.04) 0.05 (0.05) 0.06 (0.05)

Partial College 0.05**(0.02) 0.07**(0.03) 0.04 (0.03)

Postgraduate – 0.02 (0.02) – 0.01 (0.03) – 0.02 (0.03)

Respondent Credit Score: Lower Than 620 0.16***(0.03) 0.05 (0.05) 0.24***(0.05)

620 to 639 0.06 (0.04) – 0.05 (0.06) 0.19***(0.06)

640 to 659

0.12***(0.03) 0.04 (0.05) 0.20***(0.05)

660 to 679 0.08**(0.03) 0.08*(0.05) 0.08*(0.04)

680 to 699 0.11***(0.03) 0.06 (0.04) 0.18***(0.04)

700 to 719 0.01 (0.03) – 0.02 (0.04) 0.03 (0.04)

720 to 739 0.04 (0.03) 0.04 (0.04) 0.04 (0.04)

Number of Observations 6,540 3,553 2,987

R-Squared 0.07 0.06 0.08

Exhibit 4

Spread Regression

National Survey of Mortgage Originations

Critchﬁeld, Dey, Mota, and Patrabansh

CR = completely rural counties. M = metro counties. NM = non-metro counties.

* denotes signiﬁcance at 10-percent level. ** denotes signiﬁcance at 5-percent level. *** denotes signiﬁcance at 1-percent level.

Notes: The controlled difference model has the following covariates for the respondent: age, race/ethnicity, credit score, and education; and household:

household type, annual income, mortgage loan amount, and property type. The controlled difference model intercept reﬂects the percentages for respondents

who were 36 to 45 years old, non-Hispanic White, and in a coupled household; who had a college degree or higher, credit score of 740 or higher, annual

household income from $75,000 to $99,999, and mortgage loan amount from $150,000 to $299,999; and who lived in a single-family detached house.

Survey questions: “Overall, how satisﬁed are you that the mortgage you got was one with the following? Overall, how satisﬁed are you with the following?”

Source: NSMO, 2014

Share “Very Satisfied”

Percent

Controlled Difference Model

County Type Difference

(Percentage Point)

M NM – M CR – M CR – NM

All Mortgages

Best Terms to Fit Needs 78 2 – 8*** – 10***

Lowest Interest Rates Qualified 70 3 – 3 – 6*

Lowest Closing Costs 58 1 – 6** – 6*

Lender 76 1 0 – 1

Settlement Agent 70 1 – 4 – 5

Application Process 62 0 – 3 – 3

Loan Closing Process 66 0 – 8*** – 9***

Disclosure Documents 65 1 – 8*** – 8***

Timeliness of Documents 64 1 – 9*** – 11***

Purchase Mortgages

Best Terms to Fit Needs 78 2 – 7** – 9**

Lowest Interest Rates Qualified 70 – 1 – 8** – 7*

Lowest Closing Costs 55 – 2 – 5 – 3

Lender 76 – 3 2 5

Settlement Agent 69 2 – 4 – 6

Application Process 61 – 3 – 5 – 2

Loan Closing Process 64 – 1 – 9** – 8

Disclosure Documents 63 – 1 – 8** – 7

Timeliness of Documents 63 1 – 10*** – 12***

Refinance Mortgages

Best Terms to Fit Needs 78 3 – 10*** – 13

***

Lowest Interest Rates Qualified 70 7** 2 – 5

Lowest Closing Costs 61 3 – 7

– 10

Lender 76 5** – 2 – 8

Settlement Agent 71 0 – 3 – 4

Application Process 63 3 1 – 4

Loan Closing Process 69 1 – 7

– 9

Disclosure Documents 66 3 – 7

– 10

Timeliness of Documents 66 1 – 8

– 10

Exhibit 5

Satisfaction of Borrowers with Mortgage and Mortgage Process

mortgage process.

Most borrowers were “very satisfied” with their mortgage and the mortgage

process, but borrowers in completely rural counties were less likely be “very satisfied,” as most

of the completely rural–metro differences in exhibit 5 are negative and statistically significant.

The survey asked borrowers if satisfaction with the mortgage they got was: (1) the best terms to ﬁt their needs,

(2) the lowest interest rate for which they qualiﬁed, and (3) the lowest closing costs. The survey asked the borrowers

about satisfaction with: (1) their lender or broker, (2) their settlement agent, (3) the application process, (4) the loan

closing process, (5) the information in the disclosure documents, and (6) timeliness of mortgage disclosure documents.

21Cityscape

Mortgage Experiences of Rural Borrowers in the United States:

Insights from the National Survey of Mortgage Originations

CR = completely rural counties. M = metro counties. NM = non-metro counties.

* denotes signiﬁcance at 10-percent level. ** denotes signiﬁcance at 5-percent level.

Notes: The controlled difference model has the following covariates for the respondent: age, race/ethnicity, credit score, and education; and household:

household type, annual income, mortgage loan amount, and property type. The controlled difference model intercept reﬂects the percentages for respondents

who were 36 to 45 years old, non-Hispanic White, and in a coupled household; who had a college degree or higher, credit score of 740 or higher, annual

household income from $75,000 to $99,999, and mortgage loan amount from $150,000 to $299,999; and who lived in a single-family detached house.

Survey questions: “When you began the process of getting this mortgage, how concerned were you about qualifying for a mortgage? How ﬁrm an idea did

you have about the mortgage you wanted?”

Source: NSMO, 2014

The share of borrowers in completely rural counties who were “very satisfied” that they received

a mortgage with the best terms to fit their needs was 8 and 10 percentage points lower than the

shares in metro counties and non-metro counties, respectively. Completely rural borrowers were

also 6 percentage points less likely than those in metro areas to report that they were very satisfied

they received the lowest closing costs. Completely rural borrowers were 8 to 9 percentage points

less likely to be very satisfied with the closing process, disclosure documents, and timeliness of

mortgage documents. This pattern largely holds for refinance mortgages, and for both completely

rural/metro and completely rural/non–metro comparisons. For purchase mortgages, these

differences are less likely to be significant.

The NSMO data offer several measures of borrowers’ expectations at the start of the mortgage

process and familiarity with aspects of mortgage lending. These include borrowers’ indications

of how concerned they were about qualifying for a mortgage and how firm an idea they had

about the type of mortgage they wanted. The difference in concern about qualifying for a mortgage

between metro, non-metro, and completely rural borrowers was virtually nil. Exhibit 6, however,

shows that completely rural borrowers were less likely to have a firm idea about the type of

mortgage they wanted than borrowers in the other two areas.

The survey measured borrowers’ understanding of the mortgage process by asking about their

familiarity with their own credit history; available interest rates and mortgage products; and

requirements—such as income and down-payment requirements—to obtain a mortgage.

Percent

Controlled Difference Model

County Type Difference

(Percentage Point)

NM – M CR – M CR – NM

All Mortgages

Not at All Concerned 52 – 1 – 1 – 1

Have Firm Idea 59 0 – 6** – 6*

Purchase Mortgages

Not at All Concerned 47 3 1 – 3

Have Firm Idea 54 – 1 – 4 – 3

Refinance Mortgages

Not at All Concerned 57 – 5* – 3 2

Have Firm Idea

1 – 8** – 9*

Exhibit 6

Borrower Concern about Qualifying and Idea about Mortgage Wanted

National Survey of Mortgage Originations

Critchﬁeld, Dey, Mota, and Patrabansh

Generally, borrowers in completely rural counties were less familiar with aspects of mortgage

lending than borrowers in non-metro or metro counties, as shown in exhibit 7. Borrowers were

typically very familiar with their credit history with little differences among the areas. Completely

rural counties had a smaller fraction of borrowers who were very familiar with interest rates

available, the types of mortgages available, the mortgage process, down-payment requirements,

and income requirements. Completely rural borrowers and refinancers among them were also

less likely to be very familiar with the money required for closing. These results were consistent

with the notion that borrowers in completely rural counties have less information or fewer

lenders to choose from than borrowers in metro areas.

CR = completely rural counties. M = metro counties. NM = non-metro counties.

* denotes signiﬁcance at 10-percent level. ** denotes signiﬁcance at 5-percent level. *** denotes signiﬁcance at 1-percent level.

Notes: The controlled difference model has the following covariates for the respondent: age, race/ethnicity, credit score, and education; and household:

household type, annual income, mortgage loan amount, and property type. The controlled difference model intercept reﬂects the percentages for respondents

who were 36 to 45 years old, non-Hispanic White, and in a coupled household; who had a college degree or higher, credit score of 740 or higher, annual

household income from $75,000 to $99,999, and mortgage loan amount from $150,000 to $299,999; and who lived in a single-family detached house.

Survey question: “When you began the process of getting this mortgage, how familiar were you with each of the following?”

Source: NSMO, 2014

Share “Very Familiar”

Percent

Controlled Difference Model

County Type Difference

(Percentage Point)

NM – M CR – M CR – NM

All Mortgages

Credit History or Score 76 2 – 1 – 3

Interest Rates Available 58 1 – 8*** – 9***

Mortgage Types Available 48 – 2 – 9*** – 7**

Mortgage Process 55 0 – 10*** – 11***

Down Payment to Qualify 59 1 – 9*** – 9***

Income Needed to Qualify 57 0 – 8*** – 8**

Money Needed for Closing 51 – 1 – 9*** – 8**

Purchase Mortgages

Credit History or Score 75 – 1 0 1

Interest Rates Available 54 0 – 8** – 9*

Mortgage Types Available 45 – 4 – 10*** – 6

Mortgage Process 49 1 – 9** – 10**

Down Payment to Qualify 60 – 1 – 8** – 7*

Income Needed to Qualify 54 2 – 6 – 8*

Money Needed for Closing 48 – 4 – 5 – 2

Refinance Mortgages

Credit History or Score 77 4 – 3 – 7*

Interest Rates Available 64 1 – 8** – 10**

Mortgage Types Available 51 0 – 8** – 8*

Mortgage Process 64 – 3 – 13*** – 10**

Down Payment to Qualify 59 0 – 10** – 10

Income Needed to Qualify 60 – 3 – 10** – 7

Money Needed for Closing 55 0

– 13** – 13**

Exhibit 7

Borrower Familiarity with Aspects of Mortgage Lending

23Cityscape

Mortgage Experiences of Rural Borrowers in the United States:

Insights from the National Survey of Mortgage Originations

The NSMO also probed borrowers on their knowledge about mortgage concepts by asking

them how well they could explain the concepts to someone else. Based on borrowers’ responses,

the mortgage concepts from the least to the most challenging are:

(1) difference between fixed and adjustable rates,

(2) consequences of not making required payments,

(3) difference between interest rate and annual percentage rate (APR),

(4) amortization of a loan, and

(5) difference between prime and subprime loans.

We classify the first two as simple concepts and the last three as complex concepts.

Overall, borrowers were more knowledgeable about simple concepts than they were about

complex concepts as shown in exhibit 8. For the simple concept of adjustable versus fixed-rate

mortgages, borrowers in completely rural counties were less likely to say they could explain the

APR = annual percentage rate. CR = completely rural counties. M = metro counties. NM = non-metro counties.

* denotes signiﬁcance at 10-percent level. ** denotes signiﬁcance at 5-percent level. *** denotes signiﬁcance at 1-percent level.

Notes: The controlled difference model has the following covariates for the respondent: age, race/ethnicity, credit score, and education; and household:

household type, annual income, mortgage loan amount, and property type. The controlled difference model intercept reﬂects the percentages for respondents

who were 36 to 45 years old, non-Hispanic White, and in a coupled household; who had a college degree or higher, credit score of 740 or higher, annual

household income from $75,000 to $99,999, and mortgage loan amount from $150,000 to $299,999; and who lived in a single-family detached house.

Survey question: “How well could you explain to someone the following?”

Source: NSMO, 2014

Share Able to Explain

“Very Well”

Percent

Controlled Difference Model

County Type Difference

(Percentage Point)

NM – M CR – M CR – NM

All Mortgages

Fixed Versus Adjustable Rate 70 – 1 – 8*** – 7**

Consequence of Not Paying 67 1 – 3 – 3

Amortization of Loan

39 0 – 6** – 5*

Interest Rate Versus APR

28 0 – 4 – 4

Prime Versus Subprime

22 – 1 – 2 – 1

Purchase Mortgages

Fixed Versus Adjustable Rate 67 1 – 1 – 2

Consequence of Not Paying 66 2 1 0

Amortization of Loan 36

– 3 – 5 – 3

Interest Rate Versus APR 26

– 3 – 4 – 1

Prime Versus Subprime 20

– 2 0 2

Refinance Mortgages

Fixed Versus Adjustable Rate 72

– 4 – 17*** – 13***

Consequence of Not Paying 68

– 1 – 9** – 7

Amortization of Loan 41

2 – 5 – 7

Interest Rate Versus APR 31

4 – 3 – 7*

Prime Versus Subprime 24 2 – 4 – 5

Exhibit 8

Borrowers’ Abilities to Explain Aspects of Mortgages

National Survey of Mortgage Originations

Critchﬁeld, Dey, Mota, and Patrabansh

differences very well. After adjusting for differences in characteristics of borrowers and loans, the

share of borrowers in completely rural counties who could explain the differences very well was

8 percentage points lower than those in metro areas, and 7 percentage points lower than in non-

metro counties. Refinancing transactions drove the overall differences, as the share of refinancers

in completely rural counties able to explain the difference between adjustable and fixed rates very

well was 17 percentage points lower than those in metro areas and 13 percentage points lower

than those in non-metro counties. Fewer borrowers reported familiarity with complex mortgage

concepts and there were no significant differences among the geographic areas.

The NSMO asked several questions about the shopping and the mortgage application process,

namely, whether the borrower:

(1) picked the lender or broker before the loan;

(2) applied directly to a lender (as opposed to through a broker or a builder);

(3) initiated contact with the lender or broker (as opposed to those who were contacted by

the lender or broker first, or those were who were put in touch with the lender or broker

by a third party);

(4) seriously considered multiple lenders and brokers; and

(5) applied to multiple lenders or brokers.

We interpret the first three items as reflecting how proactive borrowers were with lender selection.

The last two items reflect how much borrowers shopped across multiple lenders.

Exhibit 9 shows that most borrowers were proactive with lender selection—they picked the lender

before the loan, initiated contact, and applied directly with a lender. With respect to shopping

behavior, more than half of borrowers seriously considered applying to multiple lenders, but only

one-fourth or fewer applied to more than one lender.

Outside of metro areas, borrowers were more proactive in initiating contact and applying directly to

a lender. Completely rural borrowers seeking a loan for home purchase were more proactive than

their home-purchase counterparts in metro areas, as they were 13 and 16 percentage points more

likely to have applied directly or to have initiated contact, respectively. These borrowers may have

fewer options for lenders in their area as solicitation by lenders may be less common in completely

rural areas or lenders in those areas may have less competition.

These differences, however, do not translate into differences in the likelihood that a completely

rural borrower considered or applied to multiple lenders. Borrowers who refinanced in non-metro

counties were more proactive than those in metro areas, but they did not show a different level of

engagement with multiple lenders than other areas.

The NSMO asked borrowers how likely they were to: sell their property, move but keep the

property, refinance, or pay off the mortgage to have a mortgage-free property in the next couple of

years. Borrowers’ responses demonstrated their expectations about their property and mortgage.

In our analysis, we combined borrowers who were very likely to take an action together with those

who were somewhat likely to take that action and referred to these borrowers as generally likely to

take that action.

Borrowers were generally very unlikely to take any of these actions, as seen in exhibit 10.

Borrowers who refinanced were the most likely to report any of these anticipated actions, with 40

25Cityscape

Mortgage Experiences of Rural Borrowers in the United States:

Insights from the National Survey of Mortgage Originations

CR = completely rural counties. M = metro counties. NM = non-metro counties.

* denotes signiﬁcance at 10-percent level. ** denotes signiﬁcance at 5-percent level.

Notes: The controlled difference model has the following covariates for the respondent: age, race/ethnicity, credit score, and education; and household:

household type, annual income, mortgage loan amount, and property type. The controlled difference model intercept reﬂects the percentages for respondents

who were 36 to 45 years old, non-Hispanic White, and in a coupled household; who had a college degree or higher, credit score of 740 or higher, annual

household income from $75,000 to $99,999, and mortgage loan amount from $150,000 to $299,999; and who lived in a single-family detached house.

Survey questions: “Which of the following best describes your shopping process? How did you apply for the mortgage? Who initiated ﬁrst contact between

you and the lender/mortgage broker you used for the mortgage you took out? How many different lenders/mortgage brokers did you seriously consider before

choosing where to apply for this mortgage? How many different lenders/mortgage brokers did you end up applying to?”

Source: NSMO, 2014

Percent

Controlled Difference Model

County Type Difference

(Percentage Point)

NM – M CR – M CR – NM

All Mortgages

Picked Lender Before Loan 71 0 4 4

Applied Directly to a Lender 63 9*** 10*** 1

Borrower Initiated Contact

66 8*** 12*** 4

Considered Multiple Lenders

52 – 1 0 1

Applied to Multiple Lenders

23 – 2 0 3

Purchase Mortgages

Picked Lender Before Loan 73 0 3 3

Applied Directly to a Lender 58 9*** 13*** 4

Borrower Initiated Contact 64

10*** 16*** 6

Considered Multiple Lenders 55

– 1 3 4

Applied to Multiple Lenders 26

– 5* 1 6

Refinance Mortgages

Picked Lender Before Loan 68

1 5 4

Applied Directly to a Lender 69

10*** 9** – 1

Borrower Initiated Contact 69

5* 7** 3

Considered Multiple Lenders 49

– 2 – 4 – 2

Applied to Multiple Lenders 19 0 – 1 – 1

Exhibit 9

Borrowers’ Mortgage Shopping and Application Steps

Share Stating

“Likely” to

Percent

Controlled Difference Model

County Type Difference

(Percentage Point)

NM – M CR – M CR – NM

All Mortgages

Sell Property 34 – 3* – 2 2

Move but Keep Property 23 – 2 – 1 1

Refinance Mortgage

31 – 3* – 2 1

Have Mortgage-Free Property

22 2 2 0

Purchase Mortgages

Sell Property 30 0 5 5

Move but Keep Property 23 – 2 4 6

Refinance Mortgage 34

– 1 – 5 – 4

Have Mortgage-Free Property 21

0 2 2

Exhibit 10

Borrowers’ Mortgage and Property Expectations (1 of 2)

National Survey of Mortgage Originations

Critchﬁeld, Dey, Mota, and Patrabansh

Geographic Differences in Borrowers’ Lender and

Mortgage Choice

The detail of the question on how borrowers chose their lenders is an advantage of the NSMO

survey. NSMO asked specifically about the importance of seven factors:

(1) reputation of the lender or broker,

(2) having an established banking relationship with the lender or broker,

(3) having a local office or branch of the lender or broker nearby,

(4) recommendation from a friend, relative or co-worker,

(5) recommendation from a real estate agent or builder, and

(6) whether the lender or broker was a friend.

Unfortunately, the survey questions changed from a three-point scale of “very,” “somewhat,”

and “not at all” to a two-point scale of “important” and “not important” in the seventh wave of

the survey when the supplemental sample was administered along with mostly 2015 mortgages

from the regular sample. Consequently, we examined lender and mortgage choice with a different

comparison group than that used in the earlier part of the article. Supplemental sample borrowers

(267) were combined with regular sample borrowers who took out their mortgage in 2015 (6,199)

to create an analysis file of borrowers who answered the question in the two-point scale. This led to

a sample size of 6,466 mortgages with 5,508 (85 percent) in metro, 618 (10 percent) in non-metro,

and 340 in completely rural counties shown in exhibits 11 and 12.

percent reporting they might sell the property in the next few years. Refinancers in completely rural

counties were 9 percentage points less likely to sell their property in the next few years. Refinance

borrowers in non-metro counties were 7 percentage points less likely than borrowers in metro areas

to anticipate selling their homes, and were also less likely to report that they expect to refinance

again in the next few years compared to both metro and completely rural refinancers.

Share Stating

“Likely” to

Percent

Controlled Difference Model

County Type Difference

(Percentage Point)

NM – M CR – M CR – NM

Refinance Mortgages

Sell Property 40

– 7** – 9** – 2

Move but Keep Property 23

– 3 – 6* – 3

Refinance Mortgage 28

– 6** 0 6

Have Mortgage-Free Property 22 4 2 – 1

Exhibit 10

Borrowers’ Mortgage and Property Expectations (2 of 2)

CR = completely rural counties. M = metro counties. NM = non-metro counties.

* denotes signiﬁcance at 10-percent level. ** denotes signiﬁcance at 5-percent level. *** denotes signiﬁcance at 1-percent level.

Notes: The controlled difference model has the following covariates for the respondent: age, race/ethnicity, credit score, and education; and household:

household type, annual income, mortgage loan amount, and property type. The controlled difference model intercept reﬂects the percentages for respondents

who were 36 to 45 years old, non-Hispanic White, and in a coupled household; who had a college degree or higher, credit score of 740 or higher, annual

household income from $75,000 to $99,999, and mortgage loan amount from $150,000 to $299,999; and who lived in a single-family detached house.

Survey question: “How likely is it that in the next couple of years you will do the following?”

Source: NSMO, 2014

These county shares are comparable to the ones for the 2014 sample, reported in exhibit 2.

27Cityscape

Mortgage Experiences of Rural Borrowers in the United States:

Insights from the National Survey of Mortgage Originations

CR = completely rural counties. M = metro counties. NM = non-metro counties.

* denotes signiﬁcance at 10-percent level. ** denotes signiﬁcance at 5-percent level. *** denotes signiﬁcance at 1-percent level.

Notes: Given inclusion of 2015 NSMO observations, sample sizes for exhibits 8 and 9 differ from those in remaining tables. Sample sizes for this exhibit are:

5,508 for M, 618 for NM, and 340 for CR. The controlled difference model has the following covariates for the respondent: age, race/ethnicity, credit score,

and education; and household: household type, annual income, mortgage loan amount, and property type. The controlled difference model intercept reﬂects the

percentages for respondents who were 36 to 45 years old, non-Hispanic White, and in a coupled household; who had a college degree or higher, credit score of

740 or higher, annual household income from $75,000 to $99,999, and mortgage loan amount from $150,000 to $299,999; and who lived in a single-family

detached house.

Survey question: “How important were each of the following in choosing the lender/mortgage broker you used for the mortgage you took out?”

Source: NSMO, 2014-2015

Borrowers most often identified lender reputation as an important factor for lender selection,

followed by having an established banking relationship and a local office or branch. Having an

established banking relationship was more important to borrowers in non-metro and completely

rural counties than in metro areas, shown in exhibit 11. In contrast, metro borrowers felt the

agent or builder recommendation was more important. Differences were even more stark in the

importance of an established banking relationship and agent or builder recommendation for

purchase mortgages. Furthermore, non-metro borrowers also stated that having a local office

or branch was important more frequently than metro borrowers, particularly for refinancers.

Share Stating Factor

“Important”

Percent

Controlled Difference Model

County Type Difference

(Percentage Point)

NM – M CR – M CR – NM

All Mortgages

Reputation 71 – 4* 0 3

Established Bank Relationship 55 9*** 9*** 0

Local Office or Branch 49 8*** 5 – 3

Friend/Relative Recommended 40 – 2 1 3

Agent/Builder Recommended 36 – 6*** – 12*** – 6**

Was a Friend or Relative 14 0 – 3 – 3

Purchase Mortgages

Reputation 72 – 4 – 4 0

Established Bank Relationship 50 12*** 19*** 7

Local Office or Branch 55 5* 0 – 5

Friend/Relative Recommended 48 – 1 – 5 – 3

Agent/Builder Recommended 55 – 11*** – 21*** – 10**

Was a Friend or Relative 15 – 1 – 4 – 3

Refinance Mortgages

Reputation 71 – 3 2 5

Established Bank Relationship 60 7** 0 – 7

Local Office or Branch 44 10*** 7* – 4

Friend/Relative Recommended 32 – 3 6* 9**

Agent/Builder Recommended 17 – 2 – 5* – 3

Was a Friend or Relative 14 1

– 1 – 3

Exhibit 11

Importance of Factors in Choosing a Lender

National Survey of Mortgage Originations

Critchﬁeld, Dey, Mota, and Patrabansh

Exhibit 12

Importance of Factors in Choosing a Mortgage (1 of 2)

NSMO also asked about the importance of seven factors in deciding on a mortgage:

(1) lower interest rate,

(2) fixed-interest rate for the life of the loan,

(3) lower annual percentage rate (APR),

(4) lower closing fees,

(5) lower monthly payment,

(6) a term of 30 years, and

(7) no mortgage insurance.

Exhibit 12 shows the factors for selecting a mortgage are listed in the order of importance with

near universal agreement that getting a lower interest rate was very important. For all mortgages,

one difference between borrowers in non-metro areas relative to metro borrowers was their ranking

of importance of having a lower monthly payment and not having mortgage insurance. The shares

reporting these features as being important for mortgage selection in non-metro counties was

between 4 and 6 percentage points lower than in metro counties. Completely rural purchasers

were 7 percentage points less likely to indicate that having lower monthly payment was important

than metro purchasers. Completely rural purchasers were also 8 to 10 percentage points less

likely to indicate that a mortgage term of 30 years was important than non-metro and metro

purchasers, respectively.

Share “Very Familiar”

Percent

Controlled Difference Model

County Type Difference

(Percentage Point)

NM – M CR – M CR – NM

All Mortgages

Lower Interest Rate 98 0 0 0

Fixed-Interest Rate 89 1 – 1 – 3

Lower APR 88 – 1 2 3

Lower Closing Fees 84 0 0 1

Lower Monthly Payment 80 – 4** – 3 1

30-Year Term 61 – 2 – 3 0

No Mortgage Insurance 58 – 6*** – 2 4

Purchase Mortgages

Lower Interest Rate 97 – 1 0 1

Fixed-Interest Rate 89 0 – 1 – 1

Lower APR 87 0 1 1

Lower Closing Fees 82 1 – 2 – 3

Lower Monthly Payment 81 – 3 – 7** – 4

30-Year Term 69 – 2 – 10** – 8*

No Mortgage Insurance 54 – 6* – 2 5

29Cityscape

Mortgage Experiences of Rural Borrowers in the United States:

Insights from the National Survey of Mortgage Originations

Conclusions

Rural credit markets are commonly viewed as differing from those in more populous areas for

several reasons, including a greater share of smaller and locally focused lenders. This article

offers, to our knowledge, the first comprehensive analysis that contrasts mortgage borrowers’

expectations, knowledge, and outcomes in completely rural areas to those of borrowers in metro

and non-metro areas. To do so, we take advantage of survey data from the NSMO, including a

special sample of mortgage borrowers in completely rural counties.

The comparisons we provide do not paint a simple picture of how borrowers and mortgage

markets differ by geographic location, as completely rural borrowers differ from other borrowers

on some dimensions but not others. Nonetheless, the results provide some suggestive evidence

that completely rural borrowers may have been initially less familiar with the mortgage market

conditions, products, and requirements. Completely rural borrowers were significantly less

likely than borrowers in metro areas to say they were very familiar with elements of the mortgage

process—such as the down-payment requirements—when they began it. There were few

significant differences by geographic location, however, in borrowers’ self-reported level of concern

in qualifying for a mortgage, or in their ability to explain specific mortgage features to someone

else at the time of the survey. It is not clear how to reconcile these findings, but one possibility

is that completely rural borrowers were less familiar at the outset because they expected to or

relied more heavily on the lender to qualify for a loan and to become familiar with the details

of getting a mortgage.

APR = annual percentage rate. CR = completely rural counties. M = metro counties. NM = non-metro counties.

* denotes signiﬁcance at 10-percent level. ** denotes signiﬁcance at 5-percent level. *** denotes signiﬁcance at 1-percent level.

Notes: Given inclusion of 2015 NSMO observations, sample sizes for exhibits 8 and 9 differ from those in remaining tables. Sample sizes for this exhibit are: 5,508

for M, 618 for NM, and 340 for CR. The controlled difference model has the following covariates for the respondent: age, race/ethnicity, credit score, and education;

and household: household type, annual income, mortgage loan amount, and property type. The controlled difference model intercept reﬂects the percentages for

respondents who were 36 to 45 years old, non-Hispanic White, and in a coupled household; who had a college degree or higher, credit score of 740 or higher, annual

household income from $75,000 to $99,999, and mortgage loan amount from $150,000 to $299,999; and who lived in a single-family detached house.

Survey question: “How important were each of the following in determining the mortgage you took out?”

Sources: NSMO, 2014-2015

Share “Very Familiar”

Percent

Controlled Difference Model

County Type Difference

(Percentage Point)

NM – M CR – M CR – NM

Refinance Mortgages

Lower Interest Rate 98 1 0 – 1

Fixed-Interest Rate 90 2 – 1 – 3

Lower APR 88 – 2 3 6*

Lower Closing Fees 85 – 1 3 4

Lower Monthly Payment 80 – 4 1 5

30-Year Term 52 – 3 3 6

No Mortgage Insurance 62 – 6* – 1 5

Exhibit 12

Importance of Factors in Choosing a Mortgage (2 of 2)

National Survey of Mortgage Originations

Critchﬁeld, Dey, Mota, and Patrabansh

Our results may also point to a relatively greater importance of borrower-lender relationships

outside of metro areas. For example, borrowers in completely rural and non-metro areas were

more likely than those in metro areas to rate having an established relationship as important in

choosing their lender. Additionally, completely rural borrowers were similarly satisfied with the

lender, settlement agent, and the application process as were borrowers in more populous areas,

even though they paid somewhat higher interest rates and were less likely to be very satisfied

with the mortgage they obtained, or with the closing process.

We conclude this article with a note of caution. Results presented in this article are for three

groups of counties as a whole. Because counties are very heterogeneous in each of the three

groups, the results do not necessarily apply to any individual county.

Acknowledgements

The authors thank Robert B. Avery, Ron Borzekowski, Kenneth P. Brevoort, Lariece Brown,

Brian Bucks, Daniel E. Coates, Thomas Daula, Michael Eriksen, Samuel Frumkin, Ian H. Keith,

Michael LaCour-Little, Forrest W. Pafenberg, Shiv Rawal, David Sanchez, Susan Singer,

and Rebecca Sullivan for providing comments and Julia Nguyen for assisting with data.

Authors

Tim Critchfield is an interdisciplinary statistician in the Office of Research at the Consumer

Financial Protection Bureau.

Jaya Dey is a senior economist with the Affordable Lending Analytics and Research group at

Freddie Mac.

Nuno Mota is an economist with the Economic and Strategic Research group at Fannie Mae.

Saty Patrabansh is the manager of the National Mortgage Database Program at the Federal

Housing Finance Agency.

References

Ambrose, Brent, and Richard Buttimer. 2005. “GSE Impact on Rural Mortgage Markets,” Regional

Science and Urban Economics 35 (4): 417–443.

Avery, Robert B., Kenneth P. Brevoort, Tim Critchfield, Ian H. Keith, Ismail E. Mohamed, Forrest W.

Pafenberg, Saty Patrabansh, Jay D. Schultz, and Claudia E. Wood. 2017. National Mortgage Database

Technical Report 1.2. Federal Housing Finance Agency and the Consumer Financial Protection

Bureau.

https://www.fhfa.gov/PolicyProgramsResearch/Programs/Documents/NMDB-Technical-

Report_1.2_10302017.pdf

31Cityscape

Mortgage Experiences of Rural Borrowers in the United States:

Insights from the National Survey of Mortgage Originations

Avery, Robert B., Mary F. Bilinski, Audrey Clement, Tim Critchfield, Ian H. Keith, Ismail E.

Mohamed, Saty Patrabansh, Chad Redmer, Jay D. Schultz, and Rebecca Sullivan. 2018. National

Survey of Mortgage Originations Technical Documentation: Technical Report 2. Federal Housing Finance

Agency and the Consumer Financial Protection Bureau, National Mortgage Database Program.

https://www.fhfa.gov/DataTools/Downloads/Documents/NSMO-Public-Use-Files/NSMO-

Technical-Documentation-20181108.pdf

Bucks, Brian, and Karen Pence. 2008. “Do Borrowers Know Their Mortgage Terms?” Journal of

Urban Economics 64 (2): 218–233.

Cai, Qiang, and Sarah Shahdad. 2015. What is the Mortgage Shopping Experience of Today’s

Homebuyer? Lessons from Recent Fannie Mae Acquisitions, Fannie Mae: Perspectives.

http://

fanniemae.com/portal/research-insights/perspectives/041315-cai-shahdad.html

Consumer Financial Protection Bureau (CFPB). 2014. Manufactured-Housing Consumer Finance

in the United States. Washington, DC: Consumer Financial Protection Bureau.

http://files.

consumerfinance.gov/f/201409_cfpb_report_manufactured-housing.pdf

Critchfield, Tim, Tyler Davis, Lee Davison, Heather Gratton, George Hanc, and Katherine Samolyk.

2004. “The Future of Banking in America – Community Banks: Their Recent Past, Current

Performance, and Future Prospects” [Vol. 16, No. 3], FDIC Banking Review Series.

https://www.

fdic.gov/bank/analytical/banking/2005jan/article1.pdf

Huang, Li-Ning, Michael LaCour-Little, Mark Palim, and Ying Pan. 2017. “How Much do U.S.

Households Know About Qualifying for a Mortgage?” Working Paper, May 31 2017. Washington,

DC: Fannie Mae.

http://fanniemae.com/resources/file/research/datanotes/pdf/consumer-

knowledge-working-paper.pdf.

Housing Assistance Council (HAC). 2012. Taking Stock: Rural People, Poverty, and Housing in the

21st Century. Washington, DC: Housing Assistance Council.

http://www.ruralhome.org/storage/

documents/ts2010/ts_full_report.pdf

MacDonald, Heather I. 2001. “Fannie Mae and Freddie Mac in Nonmetropolitan Housing Markets:

Does Space Matter?” Cityscape 5 (3): 219–264.

Mota, Nuno. 2016. Rural Mortgage Lending Over the Last Decade. Working paper 10.25.2016.

Washington, DC: Fannie Mae.

http://fanniemae.com/resources/file/research/datanotes/pdf/

working-paper-102716.pdf

U.S. Department of Agriculture (USDA). 2016. “Rural America at a Glance, 2016 Edition.”

Economic Information Bulletin 162. Washington, DC: U.S Department of Agriculture, Economic

Research Division.

https://www.ers.usda.gov/webdocs/publications/80894/eib-162.pdf

Vandell, Kerry D. 1997. “Improving Secondary Markets in Rural America.” In Proceedings –

Rural and Agricultural Conferences: Financing Rural America, April: 85–120. Kansas City, MO:

Federal Reserve Bank of Kansas City.

National Survey of Mortgage Originations32