Paying Too Much? Price Dispersion in the US Mortgage Market∗

Paying Too Much? Price Dispersion in the US Mortgage Market

∗

Neil Bhutta Andreas Fuster Aurel Hizmo

†

March 21, 2019

Abstract

We document wide dispersion in the mortgage rates that households get, and assess the role

of ﬁnancial knowledge and shopping in the rates obtained. To study dispersion, we draw on new

data from a mortgage industry pricing platform where we observe the “best” rates being oﬀered

by lenders, the mortgage rates actually obtained or “locked in” by consumers, and key rate

covariates including discount points, rate-lock date, geographic location, and all underwriting

information. Looking ﬁrst at locked rates (conditional on covariates), we estimate a gap between

the 10th and 90th percentile of over 50 basis points — equivalent to about $7,500 in upfront

costs (points) for the average loan. Much of this dispersion occurs within lender, and low-FICO

and low-wealth borrowers experience the most within-lender dispersion, suggesting an important

role for price discrimination. Comparing locked rates to the median b est oﬀer rate for the same

borrower in the same market on the same day, we ﬁnd that this lock-oﬀer spread is widest for

low-FICO and low-wealth borrowers, implying that such borrowers pay more not just because

of credit risk, but also because of less eﬀective se arch and negotiation. However, this spread

compresses when Treasury rates rise, suggesting that a rising level of borrowing costs encourages

more search and negotiation. Finally, we turn to survey data for direct measures of ﬁnancial

knowledge and shopping. First, using the new National Survey of Mortgage Originations, we

provide novel evidence that mortgage rates decline with mortgage knowledge and shopping;

that knowledge and shopping increase with FICO score and income; and that shopping activity

intensiﬁes in higher interest rate environments. Second, using the Survey of Consumer Finances

we document a strong negative relationship between mortgage rates and the Lusardi-Mitchell

metric of ﬁnancial literacy.

∗

We thank Jason Allen, Seraﬁn Grundl, Katherine Guthrie, Michael Haliassos, John Mondragon, David Zhang,

as well as seminar participants at the Federal Reserve Board, the Federal Reserve System Microeconomics meeting,

the CEPR European Conference on H ousehold Finance (Ortygia), the FDIC Consumer Research Conference and the

Arizona State University for helpful comments. The views expressed are those of the authors and do not necessarily

reﬂect those of the Federal Reserve Board, the Federal Reserve System, or the Swiss National Bank.

†

Bhutta and Hizmo are at the Federal Reserve Board; Fuster is at the Swiss National Bank. Emails:

neil.bh[email protected]; andreas.fuster@snb.ch; [email protected]v.

1 Introduction

According to the National Survey of Mortgage Originations (NSMO), half of the borrowers taking

out a mortgage in the US in 2016 on ly seriously considered one lender, and only three percent of

the borrowers considered more than three lenders.

Ninety-six percent of the respondents reported

that they were satisﬁed that they received the lowest interest rate for which they could qualify.

Taking these facts at face value, one might be led to conclude that either there is not much price

dispersion in the mortgage market, or that borrowers are very eﬃcient at searching and ﬁnding the

most competitive lenders. This might seem a reasonable conclusion especially when considering

that the mortgage market appears highly competitive: the majority of mortgages in the US are

very standardized and guaranteed by the government (through the GSEs and FHA/VA), and in

our data there are over one hundred diﬀerent lenders oﬀering mortgages in a local m arket in any

given day. However, in contrast to borrowers’ perceptions, in this paper we document a striking

amount of variation in the prices consumers pay for mortgages, especially among borrowers who

are not likely to be ﬁnancially sophisticated.

Identifying dispersion in mortgage rates that arises from market ineﬃciencies is challenging.

Even in an eﬃcient market, we would expect to see some variation across consumers in their

interest rates due to several factors such as diﬀerences in credit risk, day-to-day ﬂuctuations in

market rates, and heterogeneity in risk and time preferences that can aﬀect borrowers’ choices

of various contract terms. To address this challenge, we draw on a unique source of data —

an online platform used by lenders to price mortgages, initiate rate locks, manage pipeline risk,

and sell mortgages to investors. The platform provides data on both the terms being oﬀered for

speciﬁc mortgages in each market and each day, and data on the mortgages locked, or obtained, by

consumers. The data on locked mortgages include key variables for evaluating m ortgage pricing,

including several that are unavailable in any other dataset, such as “discount points”, exact time of

rate lock (as opposed to the closing date), and the lock period (e.g. 30, 45, 60 days, etc.).

Turning ﬁrst to the data on mortgage interest rate locks, we document a large amount of interest

rate dispersion. We ﬁnd that the diﬀerence between the 90th and 10th percentile interest rate that

identical borrowers pay in the same market, on the same day, and paying the same points, is over

50 basis points. Given the average point-rate trade-oﬀ in our data, 50 basis points is equivalent

to paying about 3 points more at closing or $7,500 for a average loan of $250,000. Moreover, a

substantial amount of dispersion exists even within lender, especially among borrowers who are

likely to be the least ﬁnancially sophisticated (low FICO, low wealth, low in come, or unexperienced

home buyers). Thus, getting a low rate is not simply about “going to the right lender.” Instead, it

appears that in order to get a low rate, borrowers must be knowledgeable and able to negotiate no

matter which lender they end up at.

The National Survey of Mortgage Originations is conducted jointly by the Federal Housing Finance Agency

(FHFA) and the Consumer Financial Protection Bureau (CFPB).

Notably, including lender ﬁxed eﬀects arguably accounts for diﬀerences in lender quality that might have helped

explain residual price dispersion.

Next, we draw on the real-time distribution of the “best” interest rates lenders could oﬀer to

borrowers with particular characteristics (LTV, DTI, FICO, loan amount, points, etc.) in a given

market on a given day. For a given consumer, we compute the diﬀerence between the rate they

locked and the median of these best oﬀer rates available to that consumer on the same day in the

same market. This locked-oﬀered rate gap is positive on average, meaning that borrowers tend

to get mortgage rates that are higher than what the median lender could oﬀer for an identical

mortgage.

More importantly, the locked-oﬀered rate gap varies substantially across borrower types. For

example, while borrowers getting the largest loans (“jumbo” borrowers) typically pay rates that

are 25 basis points below the median of their oﬀer distribution, FHA borrowers, who tend to have

lower income, wealth, and credit scores, typically pay 25 basis points more than what the median

lender could oﬀer. Overall, our results imply that low-FICO and less-wealthy borrowers pay more

for mortgages not simply because they present more credit risk, but also because they search and

negotiate less eﬀectively.

We also explore how changes in overall market interest rates aﬀect how borrowers fare relative

to what is oﬀered in the market. First, we show that when the level of market interest rates (as

measured by Treasury yields) is higher, borrowers are more likely to lock rates from the cheaper

end of the oﬀer distribution. Similarly, when Treasury yields increase, the price dispersion of

locked interest rates falls. These results are somewhat stronger for low-FICO borrowers, who

tend to overpay the most (relative to the median best oﬀer). This may partly reﬂect aﬀordability

constraints becoming more binding as rates rise; however, we show that for low-FICO borrowers,

even t hose that appear unconstrained in that dimension exhibit the same re lationship. Thus, we

conclude that behavioral factors, such as feeling less of a need to shop or negotiate when rates are

already low, likely contribute to these patterns.

Finally, in order to more directly assess the role of borrower search and sophistication on mort -

gage interest rates, we turn to two survey data sources. The ﬁrst is the National Survey of Mortgage

Originations (NSMO). The NSMO is a new dataset with a unique design that combines detailed ad-

ministrative records on recent mortgage originations with survey data on the individuals who took

out those mortgages. The survey component of the dataset asks several questions about borrowers’

shopping behavior and their knowledge of mortgages and interest rates, while the administrative

component provides many details on mortgage and underwriting characteristics (interest rate, LTV,

FICO, etc.). Using these data, we provide novel evidence that knowledge and shopping are strongly

related to lower mortgage rates, controlling for an array of credit risk variables. Moreover, we ﬁnd

that low-FICO and low-income borrowers get higher rates due to less shopping and mortgage knowl-

edge. And, lastly, we show that shopping activity is elevated in higher interest rate environments,

consistent with our conjecture that a rise in rates encourages p eople to shop more.

Second, we draw on the 2016 Survey of Consumer Finances (SCF). We take advantage of new

questions aimed at gauging ﬁnancial literacy that were added to the 2016 SCF. In particular, the

SCF added the “big three” questions designed by Annamaria Lusardi and Olivia Mitchell to assess

individuals’ understanding of basic ﬁnancial concepts related to saving, borrowing, and investing.

We ﬁnd that recent mortgage borrowers who answered all three questions correctly had mortgage

rates that are about 25 basis points lower than borrowers who did not answer any questions correctly,

even after including many controls for loan characteristics, credit risk, and demographics.

Overall, our empirical results provide evidence that a large fraction of the borrower population

in the US seems to overpay for mortgages likely because of lack of shopping/negotiating and lack of

ﬁnancial sophistication. As most of these borrowers have government guaranteed loans, our results

suggest that the GSEs and the FHA could consider policies aimed at limiting price dispersion. Our

ﬁndings also suggest that search frictions are important for the pass-through of monetary policy to

the mortgage market. When treasury rates increase, the amount of overpaying falls as borrowers

search and negotiate more eﬀectively, which has a dampening eﬀect on average mortgage rates.

Given the data challenges mentioned earlier, existing work on price dispersion in the US mort-

gage market is rather sparse, especially relative to the importance of the market.

Woodward and

Hall

(2012) use data on 1,500 FHA loans from 2001 and document wide dispersion in the fees paid to

mortgage brokers and argue that this reﬂects suboptimal shopping and consumer confusion around

discount points. Gurun et al. (2016) show substantial dispersion in the reset rates of privately-

securitized adjustable-rate mortgages during the housing boom and ﬁnd that these rates correlate

positively with lenders’ advertising expenditures. We build on their work by studying dispersion in

interest rates, which are more salient than reset rates, on a broader swath of the mortgage market.

Perhaps closest to our work is the paper by Alexandrov and Koulayev (2017), who document

substantial dispersion in oﬀers based on lenders’ rate sheets. We conﬁrm wide dispersion in oﬀer

rates, and add to this an analysis of actual rate locks. Our rate-lock data allow us to document

within-lender dispersion in transaction prices that widens for certain groups, and study how well

diﬀerent types of borrowers fare relative to available oﬀers.

Also, to our knowledge this is the ﬁrst

paper documenting how dispersion in contracted rates, and the locked-oﬀered gap, changes with

market rate over ti me. Fin ally, we provide novel direct evidence from the new NSMO data on how

borrowers’ mortgage knowledge and shopping relates to the interest rates they obtained.

Other related work comes from diﬀerent countries or other household ﬁnancial markets.

Allen

et al.

(2014) study the Canadian market, where there is no dispersion in posted rates, but large

dispersion in contracted rates, which they argue arises due to diﬀerences bargaining leverage across

consumers.

Damen and Buyst (2017) provide evidence that mortgage borrowers in Belgium who

shop more achieve substantial savings. Turning to other markets,

Stango and Zinman (2016) and

Argyle et al. (2017) show large dispersion in rates for credit cards and auto loans, respectively,

again suggesting limited shopping or negotiation.

The rest of the paper is organized as follows. Section 2 describes the Optimal Blue data on rate

locks and mortgage oﬀers. Section

3 documents price dispersion in the rate lock data. Section 4

explores how locked rates on average compare to the oﬀer distribution, and how this varies across

While we focus on how dispersion and the locked-oﬀered gap vary with borrower ﬁnancial characteristics such as

the FICO score, other work has instead looked at diﬀerences in contracted mortgage rates by race or ethnicity (e.g.

Bayer et al., 2018; Bhutta and Hizmo, 2018).

borrowers with diﬀerent characteristics. Section 5 studies how these patterns evolve over time

as market rates change. Section

6 introduces survey data from the NSMO and the SCF and

presents direct evidence on th e connection between shopping, mortgage knowledge, and interest

rate outcomes. Finally, Section 7 discusses potential policy implications.

2 Optimal Blue Data

The data comes from an online platform called Optimal Blue that connects over 600 mortgage

lenders with more than 200 whole loan investors. Through the platform, mortgage originators

can gather information on mortgage pricing, initiate rate locks, manage pipeline risk, and sell

mortgages to investors. Over forty thousand unique users access the system each month to search

loan programs and lock in consumer mortgages. There is a variety of lenders using the platform

such as community banks, mortgage banks, credit unions etc. Many institutions on this platform

act as correspondent lenders, meaning that t hey originate loans intended to be on-sold to other

ﬁnancial institutions such as a large bank like JP Morgan or Wells Fargo (referred to as “investors”

in the market). More than $500bn of mortgages were processed through this system in 2017, thus

accounting for about 25% of the loan originations nationally.

For this project we use two components of the data generated by the platform: a) data on

mortgage products and mortgage prices actually accepted by consumers, and b) data on mortgage

products available and mortgage prices oﬀered by lenders in each market.

2.1 Mortgage Rate Lock Data

The ﬁrst source of data is the universe of “rate lock”agreements for the mortgages processed through

the Optimal Blue platform. A mortgage rate lock is a guarantee that the borrower will be issued a

mortgage with a speciﬁc combination of interest rate and points if the mortgage closes by a speciﬁc

date. Borrowers typically lock their mortgage rates as a protection against rate increases between

the time of the lock and the time when the mortgage closes. A lock can occur at the same time

a borrower submits a loan application with a lender, but can also happen at a later time. Not all

rate locks ultimately lead to originated mortgages, since the loan application can still be rejected

afterwards (e.g. because the appraisal of the home comes in lower than expected) or the borrower

could renege.

We have access to all the mortgage locks generated by the platform since late 2013. Since the

market coverage increases over the course of 2013-2014, we start using the data from January 2015.

The data has a wide geographical coverage of about 280 metropolitan areas as well as rural areas.

All of the standard loan characteristics used for underwriting are included: loan-to-value (LTV)

ratio, debt-to-income (DTI) ratio, FICO score, loan amount, loan program, purpose (purchase or

reﬁnancing), asset documentation, income documentation, employment status, occupancy status,

house type, zip code location etc. There are a number of unique features of the data relative to

servicing data that is typically used in mortgage research. First, it includes not only the contracted

mortgage rate, but also the discount points or credits associated with that rate (meaning additional

upfront payments made or received by the borrower). Second, we see the exact time-stamp of when

the lock occurred, while in most other datasets, only the day or month of origination is recorded,

which can diﬀer from the pricing-relevant lock date by several months. Finally, we have unique

identiﬁers for the lender, the branch and the loan oﬃcer that processes each mortgage.

We restrict the sample in various ways to ensure that we study a relatively uniform set of

loans that is representative of the type of mortgages originated in recent years. For instance,

we only keep 30-year ﬁxed-rate mortgages on single-unit properties, with full documentation of

assets and income. We also drop small loans, and those with implausible values for LTV, DTI,

or points/credits. Finally, we restrict the sample to purchase mortgages and regular rate/term

reﬁnances, meaning that we drop cashout or streamline reﬁnances (which are a relatively small

part of the sample but are priced somewhat diﬀerently). This leaves us with just over 2.5 million

observations.

Table

1 presents some summary statistics from the lock data sample that we use for the analysis

in this paper, separating between the four loan programs in the data, since they diﬀer substantially

in terms of borrower and loan characteristics. The four programs are: conforming (with loan

amounts below the national conforming loan limit, so they are typically securitized through Fannie

Mae or Freddie Mac), super-conforming (with loan amounts above the national conforming limit

but below the local limit, so that Fannie Mae or Freddie Mac can still securitize the loan, but

at slightly worse prices), jumbo (loan amount above the local conforming limit, meaning the loan

cannot be securitized through the government-backed entities), and FHA loans (which require

additional mortgage insurance). The table shows that FHA loans are most likely to go to ﬁrst-time

homebuyers with low FICO scores and high LTV and DTI.

2.2 Mortgage Oﬀers Data

Second, we collect data on the menu of mortgage products available and mortgage rates that

lenders oﬀer through the platform’s pricing engine. Optimal Blue’s Pricing Insight allows users to

retrieve the real-time distribution of oﬀers for a loan with certain characteristics in a given local

market (where an oﬀer consists of a combination of a note rate and upfront fees and points that

the borrower pays or receives with this rate). The data is used primarily by mortgage banks to

compare their pricing against that of peers. We observe each institution’s set of best oﬀers (in terms

of lowest required borrower upfront payment for a given interest rate) for a given combination of

day, location, and loan characteristics.

A key advantage of these data is that the oﬀers from the Optimal Blue platform are “customer

facing,” meaning they are the interest rates and fees that would actually be paid by a borrower.

The rates and fees data come from lenders who use th e Optimal Blue Pricing Engine for their own

originations. The rates and fees for these mortgage lenders are determined primarily by: a) the

rate sheets of the “investor” who ultimately holds the mortgage, which could be the originator, or

any other secondary market investor. These rate sheets are up dated at least daily in the platform

by investors directly; and b) the markups and fees that the originator charges. The resulting rates

and fees are the best oﬀer each lender could make to a consumer who would request a mortgage

from them.

We conduct daily searches in one local market (Los Angeles), twice-weekly searches in four

markets, and weekly searches for 15 additional markets.

We collect oﬀer rate distributions for

100 diﬀerent loan types, diﬀering across the following dimensions: FICO score, loan-to-value ratio,

loan type (conforming, FHA, jumbo), loan purpose (purchase or cash-out reﬁnance), occupancy

(owner-occupied or investor), rate typ e (30-year ﬁxed or 5/1 adjustable), and loan amount. The

mortgages require full income, asset and employment documentation, and are used to ﬁnance single

unit homes.

Two limitations of the oﬀe rs data are: (i) we are not able to track institutions over time or

match them directly to the lenders in the lock data, si nce there is no ﬁxed lender identiﬁer; (ii) the

time series so far is relatively short: we started systematically tracking oﬀers in April 2016.

In the main analysis, we primarily use the oﬀered rates as a benchmark for the rates that

borrowers lock. However, in the online appendix we present a separate analysis documenting price

dispersion in oﬀers only, which is also substantial and of independent interest.

3 Dispersion in Locked Mortgage Rates

In this section we document the magnitude of price dispersion in mortgage rates that borrowers

lock. Dispersion in mortgage rates can arise for multiple reasons, some of which are diﬀerences

in borrower characteristics, mortgage characteristics and lender characteristics. We are interested

in investigating whether identical borrowers who choose the same mortgage product, in the same

market, at the same time pay diﬀerent prices. To investigate this, we regress locked mortgage rates

on borrower and loan characteristics, as well as time eﬀects, and then add an increasingly ﬁne set of

ﬁxed eﬀects. Our outcome of interest is the remaining dispersion in the residual, which we measure

in terms of standard deviations, as well as the gap b e tween 75th-25th or 90th-10th percentiles.

Table

2 shows the results from various speciﬁcations, estimated on the same set of 1.94 million

locked loans over the 2015-2018 period.

Across all columns, we control for a basic set of variables,

which consist of fully interacted bins of values for FICO, LTV, DTI, loan amount, as well as loan

program (conforming, super-conforming, jumbo, FHA). The resulting “grid” takes 7,680 unique

values. To allow for variation within t he grids, we furthermore linearly control of each of the four

continuous variables. In addition, we add a ﬁxed eﬀect for whether a locked loan is a reﬁnance,

and for the length of the lock period.

The markets with twice-weekly searches are New York City, Chicago, Denver, and Miami. The markets with

weekly searches are Atlanta, Boston, Charlotte, Cleveland, Dallas, Detroit, Las Vegas, Minneapolis, Phoenix, Port-

land, San Diego, San Francisco, Seattle, Tampa, and Washington DC.

The estimation drops “singleton” observations that are completely determined by the set of ﬁxed eﬀect. There

are more such singletons as we add more ﬁxed eﬀects; to ensure that our results are not driven by changing samples,

we use the remaining sample from the most restrictive speciﬁcation (7) in all speciﬁcations.

The lock period typically varies from 15 to 90 days, with 30 and 45 days being the most common choices. A

Column (1) is our baseline speciﬁcation, where we only add lock month-by-MSA ﬁxed eﬀects.

This is supposed to mimic the regressions one could typically run with a mortgage servicing dataset.

We see that the controls explain a sizable share of the raw variation in interest rates—the ad justed

R-squared is 0.68—but that substantial dispersion remains: the standard deviation in residuals is

0.27, and the borrower at the 90th percentile of the residual distribution pays 60 basis points (bp)

more than the borrower at the 10th percentile.

Columns (2) and (3) add bins for the points paid or received by the borrower to the grid, as well

as controlling for the exact day of the lock (rather than just the month). These (usually unobserved)

variables indeed explain some of the rate diﬀerences across borrowers, but substantial dispersion

remains—e.g. the 90th-10th percentile diﬀerence is still 54bp. Based on the regression coeﬃcient

on discount points (not shown in the table), we can translate interest rates to upfront points. This

coeﬃcient implies that 1 discount point changes the interest rate by about 16bp. Therefore, 54bp in

rate terms is approximately equivalent to 3 upfront discount points or 3% of the mortgage balance.

A borrower with a $250k mortgage borrowing at the 90

percentile interest rate would thus save

the equivalent of $7500 in upfront points and fees by borrowing at the 10

percentile interest rate.

In column (4), we add lender ﬁxed eﬀects, to allow for the possibility that some of the price

diﬀerences may reﬂect diﬀerences in lender quality as well as diﬀerences in lenders’ costs (e.g.

convenience of the oﬃce location, or service quality). Indeed, the residual dispersion in rates further

decreases, but remains substantial. In the remaining columns, we further interact the lender ﬁxed

eﬀects with other controls, to allow for the possibility that lenders may diﬀer in how they price

certain loan features, or that their (relative) pricing may change over time or across locations.

The ﬁnal two columns of the table suggest that indeed such within-lender pricing variation may

be important, since the remaining dispersion is roughly 30 percent lower in column (7) than in

column (4). Nevertheless, even conditional on these very ﬁne interacted ﬁxed eﬀects, which should

come close to looking at nearly-identical borrowers getting a loan from the same lender in the

same location at the same time, the 90th-10th percentile diﬀerence remains at 32bp, and the

interquartile range at 14bp. The remaining dispersion is further illustrated in Figure

1, which

compares the distribution of the residualized interest rates from speciﬁcation (7) of Table

2 with

the one from speciﬁcation (3), which does n ot feature any lender ﬁxed eﬀects. Adding the lender

eﬀects narrows the distribution, but it remains wide. The ﬁgure also shows that the distributions

are quite symmetric and bell-shaped.

Table

3 shows how the residual dispersion in interest rates from speciﬁcation (7) of Table 2

varies across diﬀerent loan programs and characteristics. What stands out is that the dispersion is

substantially larger for loan types and borrower characteristics that are associated with being more

ﬁnancially constrained and potentially less sophisticated. For instance, the 90th-10th percentile

diﬀerence is 44 basis points for borrowers with a FICO below 630, versus only 27 basis points

longer lock period leads to a slight increase in the fee (or equivalently the interest rate).

It is already somewhat more precise, since here we observe the month in which a loan is locked, along with

the length of the lock period, while in typical dataset loans originated in the same month may have been locked in

diﬀerent months.

for borrowers with FICO above 750, and the dispersion falls monotonically in between. Similarly,

for high-LTV loans, the dispersion is higher than for LTVs b e low 80. Since most of these high-

LTV loans are in the FHA program, it is also not surprising that residual price dispersion there is

larger than for other programs. Finally, the last section of the table shows that rates for ﬁrst-time

homebuyers also exhibit larger dispersion than for experienced borrowers.

It is interesting to note that the dispersion in oﬀered rates, which is studied in detail in the

online appendix, is also substantial, but does not vary much with borrower characteristics. This

is illustrated in Figure

2, which plots the interquartile ranges in residualized locked rates (from

speciﬁcation (3), i.e. without lender eﬀects) and in the oﬀer rates for identical mortgages across

lenders. This implies that the diﬀerences across FICO scores and LTVs in the locked data arise

from diﬀerential shopping or negotiating, but not from the supply side per se.

The ﬁndings so far have illustrated that there is a large amount of dispersion in the rates

that diﬀerent mortgage borrowers pay. While some of it is explained by diﬀerent timing, upfront

payments, or lender ﬁxed eﬀects, substantial dispersion remains once we control ﬁnely for variation

in diﬀerent lenders’ pricing over time, across locations, or across loan programs. This implies that

two observably identical borrowers may get quite diﬀerent deals even from the exact same lender

at the same time. Furthermore, this appears to be more pronounced for ﬁnancially l ess well-oﬀ

borrowers or those that are inexperienced in the market.

The analysis above has focused on dispersion, or “second moments.” We next turn to the

question of whether diﬀerent types of borrowers get good or bad deals on average (i.e. the ﬁrst

moment), relative to what is available in the market at the time they lock their mortgage.

4 Comparing the Locked Rates to Lenders’ Best Oﬀer

For the analysis in this section, we merge actual transaction interest rates from the mortgage

rate lock dataset with the data on lenders’ best oﬀer rates (described in Section

2.2). For each

observation of a rate lock in our data, we compute the median of the best oﬀer rates in the same

market, on the same day for an identical mortgage. We then study the diﬀerence between the rate

obtained by consumers and the median rate available in the market for an identical mortgage—the

locked-oﬀered rate gap.

4.1 Summary Statistics

Figure

3 shows the distribution of the locked-oﬀered rate gap for all mortgages in our data. The

thick black line denotes the mean of the distribution. The locked-oﬀered rate gap is positive on

average, meaning that borrowers end up with mortgage rates that are more expensive than what the

We use the rate at which the median lender oﬀers a loan with zero points and fees from the oﬀers data. To

compare to this oﬀer, we adjust the locked rate for points paid or received by the borrower based on the regression

coe ﬃcient from the previous analysis.

median lender could oﬀer for identical mortgages.

As shown in Table 4, the ave rage locked-oﬀered

rate gap is +11 basis points.

Figure 4 shows the distribution of the locked-oﬀered rate gap for various sub-segments of the

mortgage market. The ﬁgure shows that the locked-oﬀered rate gap distributions are centered to

the right of zero for conventional conforming and FHA loans, meaning that the average borrowers in

these segments pays more than th e median best oﬀer. The summary statistics for these distributions

are given in Table

4. We have fewer observations than in the previous analysis based on lock data

only, since here we need to observe the oﬀer side, which is only available for a subset of loan

types/characteristics, 20 MSAs, and a shorter time period.

The locked-oﬀered rate gap is largest for FHA loans, with an average of +25bp. This amounts to

almost 2% of the m ortgage balance in upfront points/fees, which for a typical FHA loan of $200k

amounts to $4000. On the other hand, the market for super-conforming mortgages and jumbo

mortgages looks very diﬀerent: the locked-oﬀered rate gap is on average slightly negative at -6bp

for super-conforming mortgages, and even more negative at -25bp for jumbo mortgages. Thus, in

these two market segments, borrowers pay less than what the median lender in their market could

oﬀer them.

Table

4 further shows summary statistics of the locked-oﬀered rate gap distribution by splitting

the sample by FICO scores, LTV ratios, and whether the borrower is a ﬁrst-time home-buyer.

On average, borrowers with a FICO larger than 740 lock in mortgage rates that are close to the

median oﬀer, while borrowers with lower FICO scores lock in rates well above the median oﬀer.

For instance, borrowers with FICO scores between 640 and 680 on average pay 22bp more than

what the median lender would oﬀer for identical mortgages. What this means is that low-FICO

borrowers on average tend to pay substantially higher rates not just due to additional r isk premia

embedded in lender oﬀers, but to a large extent due to the fact that they end up with worse rates

relative to what is in principle available in the market.

A similar pattern is evident when splitting the sample by LTV: borrowers with LTV l ess than

90% tend to obtain rates close to the median of the oﬀer distribution, while higher LTV borrowers do

worse relative to the median oﬀer. First-time homebuyers also tend to fare worse: on average, ﬁrst-

time buyers pay 14bp more than what the median lender could oﬀer them, while repeat homebuyers

pay only 7bp more.

It is worth noting that within each of the groups in Table

4, there is still substantial dispersion

in the locked-oﬀered rate gap, as shown in the table’s ﬁnal three column. Thus, even for high-FICO

or low-LTV borr owers, which on average have a gap close to zero, there are still a lot of borrowers

that lock rates well above w hat the median lender could oﬀer them. The dispersion tends to be

largest for the groups that on average do worst, meaning they have the most positive average gap.

In Figure

A-4 in the appendix, we validate that the median rate we use is close to the daily rate that is quoted

on Mortgage News Daily, an industry website.

4.2 Regression analysis

Next, we turn to a regression analysis to investigate whether similar patters emerge when con-

trolling for borrower, product, and lender ﬁxed eﬀects simultaneously. We estimate the f ollowing

speciﬁcation:

rate

lock

imt

− rate

off er

= α + f(X

) + µ

+ λ

+ ξ

+ ε

imt

(1)

where rate

lock

imt

is the interest rate locked by borrower i in market m on date t, and

rate

off er

is the

median oﬀer in market m at time t for a mortgage with characteristics X

. X

denotes observable

borrower and mortgage characteristics such as FICO, LTV, or loan amount. Finally, we control

for time ﬁxed eﬀects µ

, metropolitan area ﬁxed eﬀects ξ

, and lender ﬁxed eﬀects λ

, and in some

cases the interactions of these ﬁxed eﬀects. We estimate a ﬂexible function f by discretizing each

characteristic X and including each group separately. Standard errors are clustered at the month

and lender level.

The results from the estimation of equation (

1) are shown in Table 5. Column (1) controls for

MSA and month ﬁxed eﬀects. In line with the summary statistics above, borrowers with higher

FICO scores tend to choose lower rates from the oﬀer distribution available to them, even controlling

for other observab les. The estimated coeﬃcient, also shown graphically in Figure

5, implies that

the locked-oﬀered rate gap is about 8bp lower for borrowers with a FICO score of 740 or above than

for those with a FICO of between 640 and 680. Similarly, the locked-oﬀered rate gap is about 17bp

lower for borrowers with a LTV of less then 80% than for those with LTV of 96% or higher. Loan

amount is another statistically and economically signiﬁcant determinant of the gap: the largest

loans of $800k or more have a locked-oﬀered gap of 33bp lower than loans below $200k. Finally,

locked-oﬀered gaps tend to be slightly larger for ﬁrst-time homebuyers.

The key takeaway from the results in column (1) is that borrower characteristics that are

associated with being more ﬁnancially constrained or less sophisticated strongly correlate with

obtaining a mortgage rate that is worse relative to what we know lenders in the market to oﬀer

for borrowers with such characteristics. The 8bp premium that we observe for an otherwise similar

borrower with FICO below 680 relative to one with FICO above 740 is thus in addition to any

premium for higher default risk that is already embedded in lender oﬀers.

One possible explanation is that these coeﬃcients arise from sorting into cheap or expensive

lenders. Borrowers might choose expensive lenders because they oﬀer better service or simply

because they spend more on marketing and are more visible. To investigate whether this explains

our previous results, we include lender ﬁxed eﬀ ects in the remaining columns of Table

5. Column

(2) shows that just adding constant lender ﬁxed eﬀects increases the R-squared from 23 percent to

40 percent, meaning that lender-speciﬁc pricing diﬀerences do explain a fair amount of variation in

our data. However, most of the coeﬃcients of interest remain unchanged.

The same remains true when we allow for lenders to price diﬀerently across MSAs (column

3), or over time within an MSA (column 4). The last speciﬁcation allows for the possibility that

lenders’ relative pricing evolves over time (e.g. a new lender may price cheaply to gain market

share, but then become more exp e nsive). Adding these interacted ﬁxed eﬀects further increases

the explanatory power of the regression, but still leaves the co eﬃcients on borrower characteristics

essentially unchanged. Thus, it does not appear that e.g. lower-FICO borrowers end up with higher

locked-oﬀered gaps just because they get their loans from (temporarily or constantly) expensive

lenders. Instead, the result that they end up with relatively worse deals holds even conditioning on

going to the same lender at the same time as an otherwise identical high-FICO borrower.

Next we investigate if these patterns also hold within loan program. In column (4) Table

5 we

also include indicators for the loan program. Jumbo status is essentially collinear with loan amount,

so that including program indicators makes it diﬃcult to independently estimate the eﬀect of loan

amount. However, the other co eﬃcients (e.g. on FICO or LTV bins) remain similar if program

indicators are added to the regression. The patters we observed in Table 4 is also present here:

Jumbo borrowers have the lowest premium in terms of LORG and FHA borrowers have the highest.

In column (5) we restrict the sample to FHA loans only and we ﬁnd similar patterns to the overall

sample.

Robustness. One concern is that most of the lenders making oﬀers in our dataset may be small

and hard to ﬁnd. If that was the case, it would not be surprising that most borrowers pay more

than what the median lender is oﬀering. To rule out this potential explanation, we replicate the

same ﬁndings in our main regression using only oﬀers from high-volume lenders, as designated on

the Optimal Blue platform. Our results remain unchanged even for this sub-sample of lenders.

5 Time-series Movements in the Locked-Oﬀered Rate Gap and

Price Dispersion

The previous two sections explored the cross-sectional patterns in the dispersion of locked rates,

and in mean and dispersion of the locked-oﬀered rate gap. In this section, we instead focus on how

these measures move over time, with a particular interest in how they respond to changes in market

interest rates. Are borr owers more likely to end up with worse rates (relative to what the median

lender could oﬀer) when market rates are low, and more likely to get a good deal as rates increase?

Does price dispersion change with market interest rates?

Figure

6 plots the average locked-oﬀered rate gap against the 10-year Treasury yield.

Between

mid-2016 and mid-2018, 10-year Treasury yields increased by almost 1.5 percentage points, but with

substantial ﬂuctuations in between. There have also been signiﬁcant movements in the locked-

oﬀered rate gap, which almost mirror movements in the Treasury yield s. The locked-oﬀered rate

gap is largest when Treasury yields are low and the gap is low when yields are high.

We conﬁrm the statistical signiﬁcance of the relationship between the locked-oﬀered rate gap

and market rates in Table

6. The ﬁrst two columns add the 10-year Treasury yields as controls to

We use the 10-year Treasury yield since it is strongly correlated with the 30-year ﬁxed mortgage rate, but

avoids potential endogeneity issues due to the measurement of the latter. However, using the mortgage rate or the

current-coupon MBS yield instead leaves our conclusions unchanged.

the same regression estimated in Table 5. The coeﬃcient in column (2) implies that as the 10-year

Treasury yield increases by 1 percentage point, the average locked-oﬀered gap fall by about 12bp.

This is sizable, given that we saw earlier that over our sample as a whole, the gap averaged 11bp

with a standard deviation of 31bp.

The remaining columns of the table study the strength of this relationship for diﬀerent subsam-

ples, which may help us shed light on the underlying drivers. One possibility is that the relationship

is driven purely by aﬀordability constraints: as market rates increase, the implied monthly mort-

gage payments increase, and more borrowers may come up against DTI constraints embedded in

mortgage underwriting.

To study whether this is likely to be an important factor in the data,

we separate borrowers into those with a DTI up to 36 percent (which are likely unconstrained by

the payment burden) and those with higher DTI (for whom a higher rate may mean that they run

up against u nderwriting constraints). Another possibility is that the relationship is driven more

by “behavioral” factors: for instance, when the level of rates is already low, borrowers may feel less

compelled to search for a good deal or negotiate hard than when rates are higher, even though in

dollar terms the consequences are the same. This might be the case particularly after a recent drop

in rates, as borrowers might compare their oﬀer to a higher reference level.

While we do not have

a good individual measure of b eing subject to behavioral biases in our data, we use FICO score as

a proxy for ﬁnancial sophistication, and check whether the relationship is stronger for borrowers

with FICO below 680 than for those with higher FICOs.

We thus form four groups (FICO below/above 680 crossed with DTI below/above 36 percent)

and repeat the same regression. In colu mn (4), which includes MSA, month, and lender ﬁxed

eﬀects, we see that the relationship between treasury yield changes and locked-oﬀered gap tends to

be stronger for low-FICO borrowers than for high-FICO borrowers. This might be consistent with

a behavioral explanation, though of course it is d iﬃcult to rule out other factors. However, at least

for low-FICO borrowers, coming up against DTI constraints does not seem to drive the strength of

the relationship: the point estimates in columns (2) and (3) are almost identical. In contrast, for

higher-FICO borrowers, it is the case that the relationship is stronger for the higher-DTI group,

though it also remains signiﬁcant for the low-DTI (unconstrained) group. Overall, these results

suggest a potential rol e for both behavioral factors and underwriting constraints in driving the

strong negative correlation between locked-oﬀered gaps and market rates.

Next, we turn to investigating whether price dispersion also moves with market interest rates.

Table

7 regresses the monthly changes in the standard deviation of the residualized locked rate

(from speciﬁcation (4) in Table

2) on changes in market interest rates. We ﬁnd that dispersion

The relevant debt- to-income ratio in the US is usually the so-called “back-end” ratio, which divides the required

monthly payments on all debts (not just the mortgage) by the monthly income. Under the “qualiﬁed mortgage” rule

that has been in eﬀect in the US since 2014, this back-end DTI ratio is supposed to be below 43 percent (see e.g.

DeFusco et al., 2017).However, conforming mortgages guaranteed by Fannie Mae and Freddie Mac are exempt from

this requirement; these entities therefore impose their own requirements, which in some cases can be higher.

There are also behavioral factors that might push in the opposite direction: for instance, “relative thinking”

would make a 20bp rate saving appear larger when compared against a 3 percent bas e rate than compared against a

4 percent base rate, and might thus lead borrowers to shop more in the former case.

in residualized locked rates falls as interest rates increase. Again, this relationship is stronger for

low-FICO borrowers, but within this group, there is little diﬀerence based on borrowers’ DTI. For

these borrowers, as the market rate increases by 1 percent, the standard deviation in residualized

rates falls by about 6.5bp (relative to a mean over the sample period of about 23bp). Again, the

relationship is strong, as indicated by the R-squared values around 0.4.

Note that when we repeat the same regressions using dispersion in the oﬀer rate distribution

(not shown here), we ﬁnd almost no relationship betwee n price dispersion and market rates. The

coeﬃcients are both statistically and economically close to zero and the R-squared is very low.

Also, the standard deviation of rates in the oﬀer data changes very little over time.

To summarize, when interest rates increase, borrowers obtain mortgage rates that are lower

relative to the mean of t he oﬀer distribution, i.e. the locked-oﬀered rate gap decreases. The price

dispersion in the lock data also drops. These eﬀects are stronger for borrowers with low FICO scores;

DTI does not matter for the strength of the relationship within the low-FICO group, though it does

matter somewhat for higher-FICO borrowers (where the more constrained ones respond more to

market rates). It appears likely that behavioral factors play at least some role behind these patterns,

although aﬀordability constraints may also matter at least for some borrowers.

6 Shopping, Financial Knowledge, and Mortgage R ates

6.1 Evidence from the NSMO on the Eﬀects of Shopping and Mortgage Knowl-

edge

In this section, we use the National Survey of Mortgage Originations (NSMO) to document how

diﬀerent measures of borrower shopping and ﬁnancial literacy (in particular knowledge about mort-

gages) correlate with the mortgage rate a borrower obtains. We also document which borrower types

appear to overpay due to lack of shopping and knowledge, and how shopping eﬀort varies with the

level of market interest rates. In both cases, our ﬁndings align well with our earlier results.

The NSMO is a joint initiative of FHFA and CFPB as part of the “National Mortgage Database”

program. It surveys a nationally representative sample of borrowers with newly originated closed-

end ﬁrst-lien residential mortgages in the US, focusing in particular on borrowers’ experiences

getting a mortgage, their perceptions of the mortgage market, and their future expectations. In

November 2018, micro level data for the ﬁrst 15 survey waves were for the ﬁrst time made public

on the FHFA website, covering originations from January 2013 to December 2016.

The NSMO

contains a large number of questions, some of which were not asked in all waves, along with admin-

istrative information (from matched mortgage servicing and credit records) on borrower character-

istics such as FICO credit score at the time of origination, or the spread between a loan’s interest

rate and the market mortgage interest rate.

The full NSMO dataset contains 24,847 loans. For our analysis, we impose a number of sample

See

https://www.fhfa.gov/DataTools/Downloads/Pages/National-Survey-of-Mortgage-Originations-Public-Use-File.

aspx. We use the data version as of February 12, 2019.

restrictions. The main ones are that we only consider mortgages on a household’s primary residence

and drop mobile/manufactured homes as well as 2-4 unit dwellings. In addition, we require the loan

term to be either 10, 15, 20, or 30 years, and drop construction loans or those obtained through a

builder, mortgages with an associated additional lien, and those with more than two borrowers on

the loan. Finally, we drop a few observations where the survey respondent was not a borrower on

the loan. This leaves us with 19,906 mortgages for the analysis.

Our analysis in this section will proceed in three parts: ﬁrst, we estimate the relationship

between measures of borrower shopping or knowledge about the mortgage market and the rate

they obtain on their loan, controlling for a rich set of borrower and loan characteristics. Second, we

study which borrower and loan attributes correlate with lower rate spreads solely due to shopping

and knowledge about the mortgage market. Third, we show that shopping eﬀort increases when

market interest rates are higher.

6.1.1 The Relationship between Shopping, Knowledge, and Contract Rates

We estimate OLS regressions of the form

RateSpread

ijtw

= βX

+ ΓZ

+ α

+ δ

+ ǫ

ijtw

(2)

where RateSpread

ijtw

is the spread between the contract rate and the market mortgage rate prior

to origination, for borrower i with loan characteristics j, loan origination month t and responding to

survey wave w.

are diﬀerent measures of borrower i’s shopping eﬀort or knowledge about the

mortgage market, as described below. Z

is a rich set of borrower and mortgage characteristics that

could inﬂuence the pricing of the loan. The full list of controls is provided in the note to Table 8; it

contains for instance ﬂexible controls for FICO and LTV, loan term, program (e.g. GSE or FHA)

and purpose (purchase or reﬁnance), as well as borrower income, education, age, and race. We

further include origination month ﬁxed eﬀects α

and survey wave ﬁxed eﬀects δ

(since there were

a few small changes to the wording of questions across waves). In all our NSMO analyses, we use

the provided analysis weights, which are based on sampling weights and non-response adjustments.

We consider the following X

variables:

1. The answer to the question “How many diﬀerent lenders/mortgage brokers did you seriously

consider b efore choosing where to apply for this mortgage?” 49.5% of res pondents answer 1,

35.0% 2, 12.7% 3, 1.8% 4, and 1.0% 5 or more. We combine the last three groups into “3+”.

2. The answer to “How many diﬀerent lenders/mortgage brokers did you end up ap plying to?”

Here, 77.8% answer 1, 18.0% 2, 3.3% 3, 0.6% 4, and 0.3% 5 or more. We combine the last

four groups into “2+”.

The market mortgage rate is measured through the Freddie Mac Primary Mortgage Market Survey (PMMS),

lagged by two weeks relative to the time of loan origination. The gap is truncated at -1.5 and +1.5 percentage points.

The interest rate of individual mortgages is not contained in the public dataset.

3. Those who indicated that they applied to two or more lenders are asked which of four non-

exclusive reasons were driving the multiple applications. We create an indicator for those

who indicate that “Searching for better loan terms” was a reason.

4. A series of questions are asked about nine diﬀerent possible information sources the borrower

could use to get information about mortgages or mortgage lenders. For each of them, a

respondent can say they used a source “a lot”, “a little”, or “not at all”. We use the following,

which we think of as the best proxies for genuine search eﬀort: “Other lenders or brokers”

(32.5% a little, 9.2% a lot); “Websites that provide information on getting a mortgage” (31.5%

a little, 20.1% a lot); and “Friends/relatives/co-workers” (30.0% a little, 13.5% a lot).

5. The answer to the question “When you began the process of getting this mortgage, how

familiar were you (and any co-signers) with [t]he mortgage i nterest rates available at that

time?” 63.2% respond “Very”, 32.1% “Somewhat”, and 4.7% “Not at all”.

6. An index of “mortgage knowledge” based on 6 responses to the questions “How well could you

explain to someone the... Process of taking out a mortgage / Diﬀerence between a ﬁxed- and

an adjustable-rate mortgage / Diﬀerence between a prime and subprime loan / Diﬀerence

between a mortgage’s interest rate and its APR / Amortization of a loan / Consequences of

not making required mortgage payments”. In each case, the respondent picked from a three

point scale from “Not at all” (which we code as 1) to “Very” (3). We take the sum of the 6

responses and standardize it to have mean 0 and standard deviation 1.

7. An indicator for whether a borrower agreed with the state ment “Most mortgage lenders would

oﬀer me roughly the same rates and fees.” This question was only added in Wave 7 and so we

only have responses for roughly half of the sample. Of those, 67.8% agree with the statement.

We think of the ﬁrst four items as capturing shopping eﬀort, while the remaining three capture

mortgage market knowledge. We ﬁrst add these measures to the regression one at a time, and then

in a ﬁnal speciﬁcation jointly. The results are presented in Table

8. We see that most proxies for

intense shopping and better mortgage market knowledge are associated with lower mortgage rates:

for instance, considering 3+ lenders rather than just one lender is associated with a 8 basis point

(bp) lower rate; while applying to more than one lender in search of better loan terms is associated

with a 6bp lower rate. Similarly, more intense use of other lenders/brokers and the web as info

sources predicts lower rates, while relying on friends, relatives and co-workers seems to have little

eﬀect. A particularly strong predictor is familiarity with available mortgage r ates at the beginning

of the process of getting the mortgage: those who state they were very familiar on average pay

16bp less than those who say they were not at all familiar. A one-standard-deviation higher value

in the mortgage knowledge index is associated with an almost 5bp lower rate, while believing that

all lenders oﬀer roughly the same rate is associated with a higher rate.

The ﬁnal column controls for all X

jointly. As one might expect, some of the coeﬃcients

are attenuated relative to the earlier columns, but many of them remain individually signiﬁcant,

suggesting that there are diﬀerent dimensions to shopping and knowledge that can contribute to

a borrower obtaining a lower rate.

For instance, a borrower who is very familiar with market

conditions may not need to consider more than one lender, if they can negotiate a good rate purely

based on their knowledge. Again, it is important to remember that all of these regressions control

ﬁnely for other factors that likely inﬂuence loan pricing, in order to rule out to the extent possible

that these correlations reﬂect omitted variables that aﬀect loan pricing due to default or prepayment

risk.

6.1.2 Who Pays More Because of Lack of Shopping/Knowledge?

The previous subsection strongly suggests that more intense mortgage shopping and more knowledge

about the mortgage market is associated with lower contracted rates. We next ask which observable

borrower and loan characteristics are associated with stronger reported shopping intensity and

mortgage knowledge, and as a result pay lower interest rates. To do this, we ﬁrst isolate the part

of the interest rate spread that can only attributed to shopping and knowledge about the mortgage

market. Then, we study how this measure varies with observable characteristics.

We compute the predicted interest rate spread for each borrower speciﬁcation using a regression

almost identical to the one in speciﬁcation (10) of Table 8. The only changes we make are that we

omit the indicator for whether a borrower believed that most mortgage lenders would oﬀer roughly

the same rates and fees (since that question is only asked in later waves), and that instead of

the “knowledge index” we use each of the six underlying questions individually. All shopping and

knowledge variables are thus categorical, and for each of them we use as baseline/omitted value

the one that corresponds to the lowest level of shopping or knowledge. We thus compute for each

borrower the predicted rate diﬀerence relative to a hypothetical borrower that indicates th at they

did not engage in any shopping-related activities and have a poor understanding of the mortgage

market.

We summarize this predicted rate diﬀerence in Table

9. Due to shopping/knowledge, the average

borrower pays 27bp less than the hypothetical non-shopping, completely clueless borrower. Perhaps

more interesting is the magnitude of the diﬀerence between the 10th and 90th percentile, which is

21bp. This implies that there are large diﬀerences across borrowers in the rate spread that can be

explained by shopping behavior and mortgage knowledge.

Because shopping and knowledge about the mortgage market is correlated with various borrower

and loan characteristics, the interest savings diﬀer by group. Starting with loan characteristics,

borrowers in the jumbo market pay on average about 6bp less in rates than FHA b orrowers due

to shopping/knowledge. The highest FICO borrowers pay 4bp less than low FICO borrowers, and

high LTV borrowers pay more than low LTV borrowers. Also, borrowers with high loan amount

pay lower rates than those with low loan amounts.

It is interesting to note that the coeﬃcient on “applied to 2+ lenders” ﬂips sign if we simultaneously control for

having applied to 2+ lenders in search of better loan terms. This likely reﬂects that those who applied to multiple

lenders but not in search of better terms got turned down on their previous application (or learned negative news in

the process).

Turning to borrower characteristics, borrowers with income of $175k or higher pay 7bp lower

interest rates than borrowers with income of less than $35k due to their shopping and knowledge

about the mortgage market. More educated borrowers on average pay less than their less educated

counterparts, and ﬁrst time homebuyers pay more than repeat homebuyers.

The magnitudes of the diﬀerences in the group-speciﬁc means may appear relatively small.

However, it bears remembering that the right-hand-side variables of the underlying regression are

responses to qualitative survey questions, which are likely measured with substantial individual-

speciﬁc noise, leading to attenuation of the resulting coeﬃcients.

Furthermore, we note that the

cross-group diﬀerences in the 90th percentiles (the “worst” borrowers within each group) are often

larger than the diﬀerences in means.

Overall, the ﬁndings here corroborate the mechanism we p ostulated in our previous analysis

using the rate locks and oﬀers data. All of the evidence points to the hypothesis that borrowers

that are more likely to be less ﬁnancially sophisticated p ay more for mortgages for reasons that are

unrelated to credit risk.

6.2 Time-series Variation in Shopping Intensity

Earlier, we saw that the locked-oﬀered rate gap in the Optimal Blue data decreases when market

interest rates are higher, even for borrowers who do not appear constrained, and speculated that

this may partly be driven by increased shopping intensity when interest rates are higher. The

NSMO enables us to test this hypothesis directly. We estimate linear probability models of the

form:

Shopping

ijtw

= β · P M MS

+ ΓZ

+ δ

+ ǫ

ijtw

(3)

where Shopping

ijtw

is a binary measure of shopping intensity (discussed below) by borrower i with

loan characteristics j, loan origination month t and responding to survey wave w. P M MS

is our

main variable of interest, the market mortgage rate two weeks prior to loan origination. Z

are

borrower and mortgage characteristics, including the measures of borrowers’ mortgage kn owledge

discussed above. Finally, δ

are survey wave ﬁxed eﬀects.

As dependent variable, we use binary versions of the four main shopping variables that were

associated with lower contract interest rates in Table

8: (i) whether a borrower seriously considered

at least two lenders; (ii) whether a borrower applied to at least two lenders in search of better

terms; (iii) whether a borrower used other lender s/brokers to get information “a little” or “a lot”;

and (iv) whether a borrower used websites that provide information on getting a mortgage “a little”

or “a lot”. For each of these variables, we report regressions without other covariates (e xcept for

survey wave ﬁxed eﬀects) and with the same covariates as in Table

8, except for some variables

that seem likely endogenous to the shopping eﬀort itself.

Furthermore, we add the knowledge

For instance, respondents likely diﬀer in what they view as using an information source “a lot” vs. “a little”, or

b eing “very” vs. “somewhat” familiar with a topic.

These variables are whether a borrower obtained their mortgage through a broker, the term of the loan, and

whether it has an adjustable rate.

variables used in Table 8 as well.

Panel A of Table

10 reports the results of these regressions for the full sample. We see that

across the diﬀerent measures, a higher level of market mortgage rates is associated with more

shopping eﬀort, in most cases in statistically signiﬁcant way. For instance, column (1) implies that

a 1 percentage point increase in market mortgage rates increases the probability that a borrower

considered more than one lender by 4.5 percentage p oints, relative to a sample average of 51

percent.

Column (2) shows that this coeﬃcient is unaﬀected by the addition of ﬁne borrower-

and loan-level control variables, which alleviates c oncerns that the relationship is driven by variation

in the type of b or rower that applies at diﬀerent points in time (and at diﬀerent levels of market

rates).

The eﬀect on the probability of applying to multiple lenders is even substantially larger, es-

pecially compared to the sample mean (which is only 19 percent). A higher PMMS rate is also

signiﬁcantly associated with borrowers reporting that they obtained information from other lenders

or brokers. The association with using websites to provide information on getting a mortgage is

also positive, but not statistically signiﬁcant.

Panels B to D assess the robustness of these ﬁndings in diﬀerent subsamples. First, panel

B shows that the estimated coeﬃcients remain very similar if we restrict the sample to purchase

mortgages; this should alleviate concerns that the ﬁnding is driven by changing composition between

purchase and reﬁnance mortgages as market rates change. Panels C and D then restrict the sam ple

to borrowers that are objectively or subjectively unconstrained by payment-to-income constraints

(which, if binding, could “force” borrowers to shop more). In panel C, we only use borrowers whose

debt-to-income ratio ends up below 36 percent, suggesting that they had additional room to make

larger payments. In panel D, we restrict the sample to those borrowers who responded“not at all” to

the question “when you began the process of getting this mortgage, how concerned were you about

qualifying for a mortgage?” In both of these subsamples, the estimated coeﬃcients remain positive,

and for the ﬁrst two shopping measures statistically signiﬁcant. Thus, it does not appear that the

positive relationship between market interest rates and shopping is mainly driven by aﬀordability

constraints.

In the Appendix, we further complement this analysis by documenting univariate and multivari-

ate correlations between the shopping and knowledge measures, as well as between these measures

and various borrower and loan characteristics.

6.3 Evidence from the SCF on the Eﬀects of Financial Literacy

In this section, we draw on data from the longstanding and widely-used Survey of Consumer Fi-

nances (SCF). The SCF is a triennial, nationally representative survey of households sponsored by

the Federal Reserve Board that broadly covers U.S. families’ ﬁnancial circumstances. It collects de-

tailed information on families’ debts, assets, income, expenses, demographics, ﬁnancial institutions,

credit history, and ﬁnancial decision-making. Notably, for the ﬁrst time in 2016, the SCF added

Over our sample period, the market mortgage rate as measured by PMMS varied from 3.31% to 4.58%.

three questions designed by Annamaria Lusardi and Olivia Mitchell to gauge individuals’ general

ﬁnancial literacy.

The three questions assess understanding of basic concepts related to saving,

borrowing, and investing:

1. Suppose you had $100 in a savings account and the interest rate was 2% per year. After 5

years, how much do you think you would have in the account if you left the money to grow:

more than $102, exactly $102, or less than $102?

2. Imagine that the interest rate on your savings account was 1% per year and inﬂation was 2%

per year. After 1 year, would you be able to buy more than today, exactly the same as today,

or less than today with the money in this account?

3. Do you think that the following statement is true or false: buying a single company’s stock

usually provides a safer return than a stock mutual fund?

For each question, interviewees have the option to respond “do not know,” or can refuse to answer.

For each respondent, we compute the fraction of questions answered correctly, including “don’t

know” and “refuse” as not having answered correctly. Across all SCF respondents in 2016, 43%

answered all three correctly, 36% answered two correctly, 16% answered one correctly, and 4%

answered none correctly.

For our analysis here, we focus on a subsample of SCF households that own their home and

recently took out the mortgage on their home (either to reﬁnance or to purchase the property)

between 2013 and 2016. In this subsample, 56% answered all three ﬁnancial literacy questions

correctly, 31% answered two correctly, 11% answered one correctly, and 2% answered none correctly.

In Table

11, we provide estimates of the relationship between ﬁnancial literacy and the interest

rate respondents pay on their mortgage (interest rates are self-reported, and we subtract out the

average prime rate for the month when the loan was taken out). Column 1 indicates that moving

from none correct to getting all three questions correct is associated with a lower interest rate of

25 basis points. This magnitude is largely robust to adding controls. It drops a little in column

2 after controlling for credit history

, loan characteristics, race, income, age, and education, but

then rises back to about 25 basis points in column 3 after controlling for state ﬁxed e ﬀects.

In addition to this measure of ﬁnancial literacy, the SCF also asks respondents about h ow much

they shop when trying to get a loan: “When making major decisions about borrowing money or

obtaining credit, some people search for the very best terms while others don’t. On a scale from

zero to ten, where zero is no searching and ten is a great deal of searching, what numb er would you

(and your husband/wife/partner) be on the scale?”

A growing literature has explored the relationship between various ﬁnancial outcomes and this and other metrics

of ﬁnancial literacy. For a review, see

Lusardi and Mitchell (2014).

Note that these statistics and all other results reported in this section use the SCF sampling weights to adjust

for the sampling design of the SCF, which oversamples high wealth hous eholds.

Unlike the NSMO, we do not observe credit scores in the SC F. However, we control for any late payment in the

past year, bankruptcy in the last 4 years, and foreclosure in the last 5 years.

Table 11 shows how shopping relates to mortgage rates in the SCF, where we have divided

the numerical responses by 10 so that the shopping variable ranges from zero to one. The results

indicate that those who report shopping the most intensely have mortgage rates that are about 25

basis points lower than those who do no shopping. And, again, this result is robust to including a

number of controls that help explain a considerable amount of the variation in reported rates.

In sum, data from the 2016 SCF are c onsistent with the message from the NSMO data: bor-

rowers with higher ﬁnancial knowledge and those who shop more tend to obtain better mortgage

rates.

7 Discussion of Potential Policy Implications

Our empirical results provide evidence that many borrowers from the most vulnerable part of the

borrower population in the US seem to overpay for mortgages: those that are most likely to be

relatively low income, low net worth, and more likely to be ﬁrst-time homebuyers. These are the

exact borrowers that various government programs eﬀectively subsidize. If they were to obtain

mortgages from the lower end of the oﬀer distribution, this would make their mortgage payments

more aﬀordable and leave them with more disposable income. Alternatively, the FHA and the GSEs

could aﬀord to raise their guarantee fees substantially without aﬀecting ﬁnal cost to borrowers.

Thus, it might be worth at least considering policies that would help borrowers search and

negotiate more eﬀectively. This could take the form of required information disclosure to borrowers

of the rates available to them across diﬀerent lenders in the same market (for instance at the time

they lock their rate). We recognize that this is not a straightforward endeavor given the multi-

dimensional nature of mortgage pricing in the US, but advances in technology may make this more

feasible than in the past. Alternatively, the guaranteeing agencies could impose requirements on

the maximum locked-oﬀered gap they allow for loans to be securitized. Of course, one would also

want to consider general equilibrium eﬀects on the oﬀers that lenders make (see also

Alexandrov

and Koulayev

, 2017).

The negative relationships between average locked-oﬀered rate gap and rate dispersion with the

level of market rates that we document in Section 5 also matter for monetary policy transmission.

Our ﬁndings imply that as rates fall (e.g. in response to central bank actions), borrowers tend to

do worse relative to the distribution of oﬀered rates, perhaps due to less shopping or negotiation. It

follows that the contract rates they end up with do not fall as much as they could, based on lenders

oﬀers, adding another friction to the pass-through of monetary policy to the mortgage market.

Furthermore, this relationship is stronger for low-FICO borrowers, whose spending and default

hazard might respond most strongly to a larger drop in their mortgage rate (e.g.

Abel and Fuster,

2018).

Existing work has shown that oﬀers (as measured from investor rate sheets) respond less to increases in MBS

prices than to decreases, and less so when borrower demand is already high, which happens after falls in rates (

Fuster

et al., 2017). Limited competition may also limit pass-through (Agarwal et al., 2017; Scharfstein and Sunderam,

2016). Finally, many borrowers fail to reﬁnance when it is in their ﬁnancial interest to do so (e.g., Campbell, 2006;

Andersen et al., 2015; Keys et al., 2016).

References

Abel, Joshua and Andreas Fuster, “How Do Mortgage Reﬁnances Aﬀect Debt, Default, and

Spending? Evidence from HARP,” Staﬀ Rep ort 841, Federal Reserve Bank of New York 2018.

Agarwal, Sumit, Gene Amromin, Souphala Chomsisengphet, Tim Landvoigt, Tomasz

Piskorski, Amit Seru, and Vincent Yao, “Mortgage Reﬁnancing, Consumer Spending, and

Competition: Evidence From the Home Aﬀordable Reﬁnance Program,” Working Paper 21512,

National Bureau of Economic Research 2017.

Alexandrov, Alexei and Sergei Koulayev, “No Shopping in the U.S. Mortgage Market: Direct

and Strategic Eﬀects of Providing Information,” Working Paper, CFPB 2017.

Allen, Jason, Robert Clark, and Jean-Francois Houde, “Price Dispersion in Mortgage Mar-

kets,” The Journal of Industrial Economics, 2014, 62 (3), 377–416.

Andersen, Steﬀen, John Y. Campbell, Kasper Meisner Nielsen, and Tarun Ramadorai,

“Inattention and Inertia in Household Finance: Evidence from the Danish Mortgage Market,”

Working Paper 21386, National Bureau of Economic Research July 2015.

Argyle, Bronson, Taylor Nadauld, and Christopher Palmer, “Real Eﬀects of Search Fric-

tions in Consumer Credit Markets,” Working Paper 2017.

Bayer, Patrick, Fernando Ferreira, and Stephen L. R oss, “What Drives Racial and Eth-

nic Diﬀerences in High-Cost Mortgages? The Role of High-Risk Lenders,” Review of Financial

Studies, 2018, 31 (1), 175–205.

Bhutta, Neil and Aurel Hizmo, “Do Minorities Pay More for Mortgages?,” Working Paper

2018.

Campbell, John Y, “Household Finance,” Journal of Finance, 2006, 61 (4), 1553 – 1604.

Damen, Sven and Erik Buyst, “Mortgage Shoppers: How Much Do They Save?,” Real Estate

Economics, 2017, 45 (4), 898–929.

DeFusco, Anthony A., Stephanie Johnson, and John Mondragon, “Regulating Household

Leverage,” Working Paper, Northwestern University 2017.

Fuster, Andreas, Stephanie H. Lo, and Paul S. Willen, “The Time-Varying Price of Financial

Intermediation in the Mortgage Market,” Staﬀ Report 805, Federal Reserve Bank of New York

2017.

Gurun, Umit G., Gregor Matvos, and Amit Seru, “Advertising Expensive Mortgages,” The

Journal of Finance, 2016, 71 (5), 2371–2416.

Keys, Benjamin, D evin Pope, and Jaren Pope, “Failure to Reﬁnance,” Journal of Financial

Economics, 2016, 122 (3), 482 – 499.

Lusardi, Annamaria and Olivia S. Mitchell, “The Economic Importance of Financial Literacy:

Theory and Evidence,” Journal of Economic Literature, March 2014, 52 (1), 5–44.

Scharfstein, David and Adi Sunderam, “Market Power in Mortgage Lending and the Trans-

mission of Monetary Policy,” Working Paper, Harvard University 2016.

Stango, Victor and Jonathan Zinman, “Borrowing High versus Borrowing Higher: Price Dis-

persion and Shopping Behavior in the U.S. Credit Card Market,” The Review of Financial Studies,

2016, 29 (4), 979–1006.

Woodward, Susan E. and Robert E. Hall, “Diagnosing Consumer Confusion and Sub-optimal

Shopping Eﬀort: Theory and Mortgage-Market Evidence,” American Economic Review, 2012,

102 (7), 3249–76.

Table 1: Summary Statistics of the Rate Lock Data

Conforming Sup er-Conforming Jumbo FHA

Mean St. Dev. Mean St. Dev. Mean St. Dev. Mean St. Dev.

Loan Amount ($000) 256 92 528 65 715 242 222 93

Interest Rate 4.28 0.45 4.27 0.41 4.15 0.44 4.21 0.56

Discout Points Paid -0.08 1.02 0.08 0.99 -0.04 0.86 -0.19 1.18

FICO 738 43 748 36 762 28 674 44

LTV 83 13 80 12 77 9 96 5

DTI 35 9 36 9 31 9 42 9

First-time Homebuyer % 23 20 8 51

Reﬁnance Share % 18 25 25 5

N. Observations 1371329 82814 50048 776587

Table 2: Dispersion in Locked Interest Rates After Controlling for Borrower and Loan Characteristics

(1) (2) (3) (4) (5) (6) (7)

Residual Dispersion

St. Deviation 0.27 0.26 0.25 0.23 0.22 0.19 0.16

75-25th percentile 0.29 0.27 0.25 0.22 0.20 0.17 0.14

90-10th percentile 0.60 0.57 0.54 0.48 0.44 0.38 0.32

FICO x LTV x DTI x Loan Amount x Program grid Yes Yes Yes Yes Yes Yes Yes

Lock month x MSA F.E. Yes Yes

Discount points/credits added to grid Yes Yes Yes Yes Yes Yes

Lock date x MSA F.E. Yes Yes Yes Yes Yes

Lender F.E. Yes

Lender x FICO x LTV x Program F.E. Yes Yes

Lender x Points F.E. Yes Yes

Lender x MSA x Week F.E. Yes Yes

Lender x FICO x LTV x Program x Week F.E. Yes

Lender x Points x Week F.E. Yes

Observations 1939237 1939237 1939237 1939237 1939237 1939237 1939237

Adusted R-squared 0.68 0.71 0.72 0.76 0.79 0.82 0.84

Notes: The dependent variable is the mortgage interest rate locked. The data covers mortgage rates locked for 277 metropolitan areas during the

p eriod b etween 2015-2018. We focus on 30 year, ﬁxed rate, fully documented mortgages. All speciﬁcations include Lock period f.e., reﬁnance f.e., as well

linear controls for all the variables in the grid.

Table 3: Summary Statistics of the Residualized Locked Rate

Observations St. Deviation

Percentile Diﬀerences

− 25

− 10

All Mortgages 1,939,237 0.16 0.14 0.32

Program

Conforming 1,182,667 0.14 0.13 0.29

Super-Conforming 62,404 0.12 0.13 0.29

Jumbo 35,437 0.14 0.14 0.29

FHA 658,729 0.20 0.17 0.38

FICO

<630 102,797 0.20 0.21 0.44

630-660 222,588 0.19 0.18 0.39

660-690 289,280 0.19 0.16 0.36

690-720 326,162 0.17 0.15 0.34

720-750 340,604 0.15 0.13 0.30

>750 657,806 0.13 0.12 0.27

LTV (%)

<80 317,010 0.12 0.12 0.26

80-90 416,425 0.13 0.13 0.27

90-94 220,893 0.15 0.14 0.32

94-96 337,412 0.16 0.15 0.33

>96 647,497 0.20 0.17 0.39

First-Time Homebuyer

No 1,301,344 0.15 0.13 0.30

Yes 637,770 0.19 0.17 0.37

Note - This table summarizes the residualized locked mortgage rate from speciﬁcation (7) of table (2).

Table 4: Summary Statistics of the Rate Locked Minus the Median Oﬀered Rate for

Identical Mortgages

Observations Mean St. Deviation

Percentiles

All Mortgages 66,719 0.11 0.31 -0.07 0.25

Program

Conforming 37,754 0.08 0.24 -0.06 0.19

Super-Conforming 7,337 -0.06 0.25 -0.21 0.07

Jumbo 2,061 -0.25 0.42 -0.39 -0.07

FHA 19,567 0.25 0.35 0.03 0.44

FICO

640-679 12,680 0.22 0.36 0.00 0.42

680-719 16,256 0.14 0.33 -0.05 0.31

720-739 7,931 0.10 0.28 -0.07 0.24

740+ 29,852 0.04 0.25 -0.10 0.16

LTV

60-80 21,204 0.01 0.25 -0.12 0.14

81-90 10,480 0.03 0.30 -0.11 0.18

91-95 14,368 0.09 0.27 -0.07 0.22

96-97 20,667 0.25 0.34 0.03 0.43

First-Time Homebuyer

No 33,635 0.07 0.28 -0.09 0.20

Yes 33,083 0.14 0.32 -0.05 0.30

Note - For each mortgage rate locked by borrowers in our data, we compute the median rate

oﬀered by lenders in the same market on the same day for an identical mortgage. This table

summarizes the diﬀerence between each locked rate and the median oﬀered rate.

Table 5: Explaining the gap be tween the rates consumers lock and the rates oﬀered by the median

lender for identical mortgages

FHA Only

(1) (2) (3) (4) (5)

FICO groups

680≤F ICO<720

-0.041*** -0.040*** -0.039*** -0.029*** -0.037***

(0.007) (0.007) (0.007) (0.007) (0.008)

720≤F ICO<740

-0.057*** -0.050*** -0.049*** -0.031*** -0.056***

(0.008) (0.008) (0.008) (0.007) (0.009)

F ICO≥740

-0.082*** -0.070*** -0.067*** -0.042*** -0.065***

(0.008) (0.008) (0.008) (0.008) (0.012)

LTV groups

80<LT V ≤90

0.006 -0.001 -0.001 -0.007 -0.007

(0.006) (0.005) (0.006) (0.006) (0.021)

90<LT V ≤95

0.056*** 0.048*** 0.049*** 0.036*** -0.003

(0.007) (0.006) (0.006) (0.006) (0.017)

LT V >95

0.169*** 0.153*** 0.151*** 0.109*** 0.061***

(0.012) (0.011) (0.012) (0.008) (0.018)

Loan Amount

$200k≤Loan<$400k

-0.063*** -0.059*** -0.054*** -0.057*** -0.066***

(0.006) (0.005) (0.005) (0.005) (0.009)

$400k≤Loan<$600k

-0.090*** -0.083*** -0.082*** -0.033*** -0.010

(0.013) (0.011) (0.012) (0.011) (0.029)

$600k≤Loan<$800k

-0.165*** -0.154*** -0.156*** -0.012 0.008

(0.021) (0.019) (0.020) (0.017) (0.043)

Loan≥$800k

-0.334*** -0.305*** -0.330*** -0.049

(0.032) (0.030) (0.033) (0.036)

First Time Homebuyer 0.014** 0.010*** 0.003 0.002 0.009

(0.006) (0.004) (0.004) (0.004) (0.006)

Program

Sup e r-Conforming -0.146***

(0.016)

Jumbo -0.303***

(0.033)

FHA 0.037**

(0.017)

MSA F.E. Yes Yes

Month F.E. Yes Yes

Lender F.E. Yes

Lender x MSA x Month F.E. Yes Yes Yes

Observations 66718 66666 61520 61520 15841

R-squared 0.185 0.360 0.498 0.519 0.595

Note - The dependent variable is the mortgage interest rate locked minus the median oﬀer rate in the same

market and day for an identical mortgage. The data covers mortgage rates for 20 metropolitan areas during the

period between 2016-2018. We focus on 30 year, ﬁxed rate, fully documented mortgages. The standard errors are

clustered at the MSA, month, and lender level.

Table 6: The Relationship Between the Lock-Oﬀered Gap and Treasury Yields

(1) (2) (3) (4)

Treasury Yield -0.086*** -0.117***

(0.010) (0.017)

Treasury Yield ×

F ICO ≤ 680, DT I ≤ 36% -0.103*** -0.134***

(0.016) (0.021)

F ICO ≤ 680, DT I > 36% -0.122*** -0.154***

(0.017) (0.020)

F ICO > 680, DT I ≤ 36% -0.067*** -0.099***

(0.011) (0.020)

F ICO > 680, DT I > 36% -0.084*** -0.116***

(0.011) (0.017)

Borrower Controls Yes Yes Yes Yes

MSA F.E. Yes Yes Yes Yes

Lender F.E. Yes Yes Yes Yes

Month F.E. Yes Yes

Observations 66271 66271 66271 66271

R-squared 0.381 0.382 0.381 0.383

Note - The dependent variable is the mortgage interest rate locked minus the median oﬀer

rate in the same market and day for an identical mortgage. All speciﬁcations include con-

trols for FICO, LTV, loan amount and loan program. The data covers mortgage rates for 20

metropolitan areas during the period between 2016-2019. We focus on 30 year, ﬁxed rate, fully

documented purchase mortgages. The standard errors are clustered at the month and lender

level.

Table 7: Relationship between Changes in the Dispersion of Residualized Locked Rates and Changes in Treasury

Yields

Dep. Var.: F ICO ≤ 680 F ICO ≤ 680 F ICO > 680 F ICO > 680

∆Std. Residual of All Data DT I ≤ .36 DT I > .36 DT I ≤ .36 DT I > .36

Lo cked Rate

(1) (2) (3) (4) (5)

∆10 Year Treasury Yield

-0.036*** -0.064*** -0.067*** -0.016** -0.039***

(0.01) (0.01) (0.01) (0.01) (0.01)

Observations 43 43 43 43 43

R-Squared 0.28 0.40 0.35 0.09 0.33

Note - The dep e ndent variable is the month to month change in standard deviation of the residualized locked rates. Huber/White

robust standard errors shown in parentheses.

Table 8: Relationship between mortgage rates and measures of s hopping and knowledge. Dependent variable: spread between a borrower’s

mortgage interest rate and the market mortgage rate prior to origination, in percentage points (censored at -1.5 and +1.5)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

Seriously considered 2 lenders -0.039*** -0.020**

(0.009) (0.010)

Seriously considered 3+ lenders -0.078*** -0.043***

(0.012) (0.014)

Applied to 2+ lenders -0.039*** 0.051**

(0.010) (0.022)

Applied to 2+ lenders in search of better loan terms -0.057*** -0.079***

(0.010) (0.023)

Used other lenders/brokers to get info? A little -0.024*** 0.000

(0.009) (0.010)

Used other lenders/brokers to get info? A lot -0.061*** -0.021

(0.014) (0.015)

Used web to get info? A little -0.042*** -0.033***

(0.010) (0.010)

Used web to get info? A lot -0.065*** -0.036***

(0.011) (0.012)

Used friends/relatives to get info? A little -0.001 0.001

(0.010) (0.010)

Used friends/relatives to get info? A lot 0.010 0.013

(0.013) (0.013)

Familiar with mortgage rates? Somewhat -0.075*** -0.060**

(0.024) (0.024)

Familiar with mortgage rates? Very -0.159*** -0.122***

(0.024) (0.024)

Index of mortgage knowledge (Std) -0.046*** -0.031***

(0.005) (0.005)

Most lenders oﬀer same rate? Yes 0.033*** 0.026**

(0.012) (0.012)

Adj. R2 0.18 0.18 0.18 0.18 0.18 0.17 0.18 0.18 0.18 0.19

Obs. 19906 19906 19906 19906 19906 19906 19906 19906 19906 19906

Sample restricted to ﬁrst-lien loans (without a junior lien) for single-family principal residence properties, with no more than two borrowers, and a loan term of 10,

15, 20 or 30 years. Observations weighted by NSMO sample weights. All regressions control for origination month ﬁxed eﬀects, survey wave ﬁxed eﬀ ects, FICO

score (linear term plus dummies for 11 FICO bins), LTV (linear term plus dummies for each percentage po int from 79-98), indicators for loan purpose (purchase,

reﬁnance, or cash-out reﬁnance), 9 loan amount categories, l oan program (Freddie, Fannie, FHA, VA, FSA/RHS, other), loan term, ﬁrst-time homebuyer status,

single borrowers, using a mortgage broker, whether the loan has an adjustable rate, jumbo status, 6 borrower income categories, 6 borrower education categories,

whether the household owns 4 diﬀerent types of ﬁnancial assets, race and ethnicity, metropolitan CRA low-to-moderate income tract status, borrower age and

gender, and self-assessed creditworthiness, likelihood of moving, selling, or reﬁnancing, and risk aversion. Robust standard errors in parentheses. * p<0.1, **

p<0.05, *** p<0.01.

Table 9: Summary statistics of the interest rate spread that can be attributed to shopping and

mortgage knowledge

Observations Mean

Percentiles

All Mortgages 19,906 -0.27 -0.37 -0.16

Program

Conforming 11,103 -0.28 -0.37 -0.17

Jumbo 679 -0.31 -0.38 -0.22

FHA 2,734 -0.25 -0.35 -0.13

FICO

≤ 600 411 -0.24 -0.35 -0.14

601-640 1,089 -0.25 -0.35 -0.14

641-680 2,195 -0.26 -0.36 -0.15

681-740 4,784 -0.26 -0.36 -0.16

> 740 11,427 -0.28 -0.37 -0.18

LTV

≤ 75 8,216 -0.28 -0.37 -0.18

76-80 3,551 -0.28 -0.37 -0.18

81-95 4,551 -0.27 -0.37 -0.16

96-97 1,805 -0.24 -0.35 -0.12

Loan Amount

<100k 3,011 -0.25 -0.35 -0.13

[100k, 200k) 7,736 -0.27 -0.36 -0.16

[200k, 300k) 4,656 -0.28 -0.37 -0.17

[300k, 400k) 2,405 -0.29 -0.38 -0.19

≥ 400k 2,098 -0.30 -0.38 -0.21

First-Time Homebuyer

No 16,717 -0.28 -0.37 -0.18

Yes 3,189 -0.24 -0.36 -0.12

Income

<35k 1,189 -0.23 -0.34 -0.11

[35k, 75k) 6,014 -0.25 -0.35 -0.14

[75k, 175k) 9,752 -0.28 -0.37 -0.18

≥ 175k 2,951 -0.30 -0.38 -0.21

Education

Less than college 3,322 -0.24 -0.34 -0.13

Some college 3,975 -0.26 -0.36 -0.16

College grad 7,017 -0.28 -0.37 -0.17

Postgrad 5,592 -0.29 -0.38 -0.19

The variable we are summarizing here is the interest rate spread that is only due to shopping and knowledge ab out

the mortgage market. We compute the predicted value of the interest rate spread using only the displayed variables

on shopping behavior and knowledge about mortgages, in a way similar to sp eciﬁcation (10) of Table

8. (see text for

details)

Table 10: Relationship between various binary measures of mortgage shopping and mortage market

interest rates (PMMS).

Considered 2+ lenders Applied to 2+ lenders Used other lenders Used web

for better terms to get info to get info

A. Full sample (1) (2) (3) (4) (5) (6) (7) (8)

PMMS rate 0.045** 0.045** 0.069*** 0.062*** 0.048*** 0.050*** 0.019 0.026

(0.018) (0.018) (0.014) (0.014) (0.018) (0.018) (0.018) (0.018)

Controls? No Yes No Yes No Yes No Yes

Mean of Dependent Variable 0.510 0.510 0.190 0.190 0.418 0.418 0.533 0.533

Obs. 19906 19906 19906 19906 19906 19906 19906 19906

B. Purchase loans Considered 2+ lenders Ap plied to 2+ lenders Used other lenders to get info

PMMS rate 0.060** 0.054* 0.077*** 0.072*** 0.049* 0.041 0.014 0.009

(0.029) (0.029) (0.024) (0.024) (0.029) (0.028) (0.029) (0.027)

Controls? No Yes No Yes No Yes No Yes

Mean of Dependent Variable 0.534 0.534 0.223 0.223 0.430 0.430 0.550 0.550

Obs. 9254 9254 9254 9254 9254 9254 9254 9254

C. DTI ≤ 36 Considered 2+ lenders Applied to 2+ lenders Used other lenders to get info

PMMS rate 0.039 0.045* 0.081*** 0.074*** 0.029 0.036 0.003 0.016

(0.025) (0.025) (0.019) (0.019) (0.025) (0.025) (0.025) (0.024)

Controls? No Yes No Yes No Yes No Yes

Mean of Dependent Variable 0.503 0.503 0.176 0.176 0.411 0.411 0.541 0.541

Obs. 10590 10590 10590 10590 10590 10590 10590 10590

D. Not concerned about qualif. Considered 2+ lenders Applied to 2+ lenders Used other lenders to get info

PMMS rate 0.041* 0.045* 0.060*** 0.052*** 0.023 0.031 -0.005 0.014

(0.024) (0.024) (0.018) (0.017) (0.024) (0.023) (0.024) (0.023)

Controls? No Yes No Yes No Yes No Yes

Mean of Dependent Variable 0.488 0.488 0.165 0.165 0.387 0.387 0.499 0.499

Obs. 11203 11203 11203 11203 11203 11203 11203 11203

Sample restricted to ﬁrst-lien loans (without a junior lien) for single-family principal residence properties, with no

more than two borrowers, and a loan term of 10, 15, 20 or 30 years. All four dependent variables are binary. All

regressions control for survey wave ﬁxed eﬀects and use NSMO analysis weights. The multivariate regressions (even

columns) further control for FICO score (linear term plus dummies for 11 FICO bins), LTV (linear term plus dummies

for each percentage point from 79-98), indicators for loan purpose (purchase, reﬁnance, or cash-out reﬁnance), 9 loan

amount categories, loan program (Freddie, Fannie, FHA, VA, FSA/RHS, other), ﬁrst-time homebuyer status, single

b orrowers, jumbo status, 6 borrower income categories, 6 borrower education categories, whether the household

owns 4 diﬀerent types of ﬁnancial assets, race and ethnicity, metropolitan CRA low-to-moderate income tract status,

b orrower age and gender, and self-assessed creditworthiness, likelihood of moving, selling, or reﬁnancing, and risk

aversion. Robust standard errors in parentheses. * p<0.1, ** p<0.05, *** p<0.01.

Table 11: Relationship between Interest Rate Spreads and Measures of Financial Literacy and Shopping in the Survey

of Counsumer Finances

(1) (2) (3) (4) (5) (6) (7)

Financial Literacy (Fraction Correct) -0.247 * -0.198 * -0.245 * -0.243 *

(0.110) (0.095) (0.096) (0.099)

Shops Around for Credit -0.262 ** -0.262 ** -0.236 ** -0.228 *

(0.090) (0.086) (0.086) (0.089)

Loan Characteristics Yes Yes Yes Yes Yes

Borrower Characteristics Yes Yes Yes Yes Yes

State Fixed Eﬀects Yes Yes Yes

Observations 820 816 816 821 817 817 816

R-squared 0.011 0.15 0.225 0.009 0.151 0.222 0.229

Data source: 2016 Survey of Consumer Finances (SCF)

Notes: Sample comprised of households that took out a 15 year or 30 year ﬁxed-rate home purchase or reﬁnance mortgage in 2013-2016

for their principal residence. Outcome variable is the interest rate (self-reported) on the ﬁrst lien mortgage relative to the average Freddie

Mac PMMS prime rate for a loan of the same term in the month the mortgage was taken out. The Financial Literacy variable refers to

the fraction correct on three questions designed by Lusardi and Mitchell and asked in the 2016 SCF. The Shopping Around variable is a

self-rep orted value between 0 and 10 gauging the degree to which respondents shop for credit; we divide responses by 10 so that the range

is 0 to 1. The loan characteristics we control for in speciﬁcations (2), (3) and (5)-(7) include loan program, loan term, and loan purpose

(purchase, reﬁnance or cash out). Borrower controls include indicators of whether they were late on any payment in the past year, had a

bankruptcy in the last 4 years, had a foreclosure in the last 5 years, as well as controls for income, education, age and race/ethnicity.

-1 -.9 -.8 -.7 -.6 -.5 -.4 -.3 -.2 -.1 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1

Residual Interest Rate

With Lender F.E.

Without Lender F.E.

Figure 1: Re sidualized Locked Mortgage Rates Controlling for Borrower and Loan Characteristics,

and Allowing for Diﬀerential Pricing of Loans by Lender-Location-Loan Characteristics-Time

Note: This ﬁgure plots ﬁtted distributions of the residuals from the regression in columns (3) and (7) of Table

0.15

0.2

0.25

0.3

0.35

0.4

640 680 720 750

Interquartile Range

FICO

Interquartile Range by FICO

Locked

Offer

0.15

0.2

0.25

0.3

0.35

0.4

80 90 95 96

Interquartile Range

LTV

Interquartile range by LTV

Locked

Offer

Figure 2: Comparing the Mortgage Rate Dispersion in Oﬀer and Rate Lock Data

.05

.15

Fraction

-1 -.8 -.6 -.4 -.2 0 .2 .4 .6 .8 1 1.2 1.4 1.6 1.8

Locked Rate minus Median Offer Rate

All Locked Mortgages

Figure 3: Distribution of Rate Locked Minus the Median Best Oﬀered Rate for Identical Mortgages

Note: For each mortgage rate locked by borrowers in our data, we compute the median best oﬀer by lenders in the

same market on the same day for an identical mortgage. This ﬁgure shows the distribution of the diﬀerence between

each locked rate and the median oﬀered rate. The solid black line de notes the mean of the distribution.

.05

.15

Fraction

-1 -.8 -.6 -.4 -.2 0 .2 .4 .6 .8 1 1.2 1.4 1.6 1.8

Locked Rate minus Median Offer Rate

All Conventional Conforming Mortgages

.02

.04

.06

.08

Fraction

-1 -.8 -.6 -.4 -.2 0 .2 .4 .6 .8 1 1.2 1.4 1.6 1.8

Locked Rate minus Median Offer Rate

All FHA Mortgages

.05

.15

Fraction

-1 -.8 -.6 -.4 -.2 0 .2 .4 .6 .8 1 1.2 1.4 1.6 1.8

Locked Rate minus Median Offer Rate

All Super-Conforming Mortgages

.05

.15

Fraction

-1 -.8 -.6 -.4 -.2 0 .2 .4 .6 .8 1 1.2 1.4 1.6 1.8

Locked Rate minus Median Offer Rate

All Jumbo Mortgages

Figure 4: Distribution of Rate Locked Minus the Average Oﬀered Rate for Identical Mortgages

Note: For each mortgage rate locked by borrowers in our data, we compute the average rate oﬀered by lenders in the

same market on the same day for an identical mortgage. This ﬁgure shows the distribution of the diﬀerence between

each locked rate and the average oﬀered rate. The solid black line denotes the mean of the distribution.

0.02

0.04

0.06

0.08

0.1

0.12

640 680 720 >=740

Locked-Offer Rate Gap

FICO Score

FICO coefficients and 90% CI

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

100 200 400 600 >800

Locked-Offer Rate Gap

Loan amount ($000)

Loan amount coefficients and 90% CI

-0.05

0.05

0.1

0.15

0.2

0.25

<=80 90 95 >96

Locked-Offer Rate Gap

LTV

LTV coefficients and 90% CI

Figure 5: The Eﬀects of Observables on the Locked-Oﬀered Rate Gap

Note: These are plots of the coeﬃcients from speciﬁcation (1) in Table

1.5

2.5

3.5

10 Yr Treasury Yield

.05

.15

.25

Locked-Offer Rate

2016m1 2017m1 2018m1 2019m1

Date

Locked-Offer Rate

10 Yr Treasury Yield

Figure 6: The Evolution of Rate Locked Minus the Average Oﬀered Rate and Treasury Yields

Note: For each mortgage rate locked by borrowers in our data, we compute the average rate oﬀered by lenders in the

same market on the same day for an identical mortgage. The solid blue line is t he average diﬀerence between each

lo cked rate and the average oﬀered rate. The dashed line is the 10 year treasury yield.

Online Appendix for

“Paying Too Much? Price Dispersion in the US Mortgage Market”

A.1 Price Dispersion in Mortgage Oﬀers

In this app endix, we study price dispersion in oﬀered mortgage rates across diﬀerent lenders oﬀering

the same mortgage product in the same location at the same time, as observed on the Optimal

Blue Insights platform. We show that, similar to the large dispersion in locked rates documented

in the main text, there is also large dispersion in oﬀered rates.

There are two things to c onsider when thinking about what the price of a mortgage means in

this context. First of all, lenders do not oﬀer a single mortgage rate to borrowers but rather a

menu with diﬀerent combinations of mortgage rates and discount points to choose from. Borrowers

can pay discount points, each equal to one percent of their mortgage balance, in order to lower

their mortgage interest rate by roughly 15bp per discount point paid. Borrowers can also choose

negative points, known as lender credits, in return for a higher mortgage rate of roughly 15bp per

point. In this case, borrowers receive cash from the lender which can be used t oward closing costs.

Secondly, lenders also charge origination fees. While fees are not typically considered as part of

the price of the mortgage, they are part for the total cost of securing the mortgage. In a way we

can think of lender fees and discount points as interchangeable. From the borrowers perspective,

a lender that charges an origination fee of one percent to originate a mortgage at 4% interest is

equivalent to a lender that charges no fees but requires the borrower to pay one discount point for

a mortgage rate of 4%.

In light of the above considerations, there are two ways in which we quantify price dispersion.

First, we look at the dispersion in mortgage rates for identical mortgages oﬀered with no points

and fees. While most lenders charge fees, the platform reports the rate at which lender credits (or

negative discount points) would be equal to lender fees. A borrower would not have to pay the

lender anything to lock this mortgage rate. Computing the dispersion in oﬀer rates with no points

and fees is not possible with any oth er data set we are aware of since one would need to know both

the rate/point trade oﬀs and the lender fees.

The second way we quantify price dispersion is by looking at the total points and fees a borrower

would have to pay at diﬀerent lenders in order to borrow at a given median interest rate. Since

points and fees are paid by the borrower upfront, this makes it possible to quantify the price

dispersion in terms of dollars an d cents without having to engage in any present value calculations.

A.1.1 Dispersion in Oﬀered Rates

We start by do cume nting the dispersion in mortgage rates available from diﬀerent lenders for

identical mortgages in Los Angeles. We only compare identical real-time mortgage oﬀers with no

points or fees, with exactly the same FICO, loan-to-value ratio, debt-to-income ratio, loan amount

and location. We also focus on ﬁxed rate mortgages with a 30 year term, with fully documented

income, assets and employment, and mortgages secured by a single property for this analysis. The

ﬁrst panel of Figure

A-1 shows the distribution of rates oﬀered by d iﬀerent lenders for conforming

mortgages with an amount of $300k, FICO=750, LTV=80 and DTI=36 in Los Angeles. There are

about 120 diﬀerent lenders oﬀering this mortgage in Los Angeles in any given day. The histogram

shows the daily oﬀered rates after subtracting the median (for the same day) over the period of April

2016 to April 2018. Figure

A-1 uncovers a wide distribution of mortgage rates oﬀered by diﬀerent

lenders even for the same mortgage product in the same location, at the same time. There is almost

a full percentage point diﬀerence between the cheapest and the most expensive lend er. Moreover,

even though much of the mass is in the middle of the distribution, the tails of the distribution are

rather fat. These patterns can also be seen in the other two panels of Figure A-1, which plot the

dispersion in a typical FHA mortgage and a Jumbo mortgage. The exact shape of the distribution

does look diﬀerent across these diﬀerent mortgages, however the amount of dispersion is similar.

Figure A-2 shows the dispersion in mortgage rates available from diﬀerent lenders in all of the

20 metropolitan areas. To make the distribution comparable across time and locations, we demean

the oﬀered rates for each mortgage type in each market and day. Even in ou r pooled data, the

amount of price dispersion seems similar to that in Los Angeles.

Table

A-1 shows more detailed summary statistics of the rate dispersion in our oﬀer data, broken

down by mortgage types. There are typically about 120 unique lenders in any given day m aking

oﬀers for each mortgage type in each location. The median mortgage rate is higher for jumbo

loans than for conforming loans reﬂecting in part the fact that conforming loans are gu aranteed by

Fannie or Freddie in exchange for a low guarantee fee, which is rolled into the mortgage rate. FHA

mortgages have lower interest rates than other products since borrowers also have to pay upfront

(175bp) and ongoing mortgage insurance premia (85bp) which are not part of the quoted mortgage

rate. The price dispersion appears similar across diﬀerent programs, FIC O scores and loan-to-value

ratios. Generally, the price dispersion is a bit higher for mortgages with low FICO scores, high

LTVs and FHA mortgages. Overall, there is about a 75 basis point diﬀerence in mortgage rates

between the 1

percentile lender and the 99

percentile lender.

Table

A-2 compares the rate dispersion for a “plain vanilla” conforming mortgage with LTV

of 80 and FICO of 750 across MSAs. We see that, while there are some diﬀerences in the exact

amount of dispersion across MSAs, the qualitative points from above generalize across all of the

cities, and Los Angeles is not an outlier.

A.1.2 Dispersion in Oﬀered Points and Fees

In this subsection we focus on the points and fees charged by lenders to originate a mortgage with

a median interest rate. The median interest rate for each mortgage type is deﬁned exactly as in the

previous subsection: it is the median interest rate across lenders for an identical mortgage oﬀered

with no points and fees. Figure

A-3 shows the distribution of points and fees charged by diﬀerent

lenders to originate this median interest rate mortgage, with discount points and fees measured as a

percent of the mortgage balance. Table

A-3 summarizes this dispersion for diﬀerent mortgage types.

The diﬀerences in the upfront costs of a mortgage across lenders seems staggering. The diﬀerence

between the 99

percentile and 1

percentile lender is close to 4% of the mortgage balance. For

a typical conforming loan of $250K that amounts to a $9000 diﬀerence in upfront costs between

these lenders. Even going from the 75

percentile to the 25

percentile lender would save about

$3000 for a typical borrower with a $250k loan.

A.2 Correlates of Shopping Intensity and Knowledge

Section

6 strongly suggests that more intense mortgage shopping and better knowledge of the

mortgage market are associated with lower contracted rates. In this appendix, we document how

diﬀerent shopping and knowledge measures are correlated with one another, and also study which

observable borrower and loan characteristics are assoc iated with stronger reported shopping inten-

sity and higher knowledge.

In Table

A-4, we report results from regressions of the four b inary shopping measures already

used in S ection 6.2 on the three mortgage knowledge measures introduced in Section 6.1.1, as well as

various other loan and borrower characteristics, most of which we turn into binary variables for ease

of interpretation. We run regressions with one covariate of interest at a time (with survey wave ﬁxed

eﬀects as the only additional control), or controlling for all of them jointly and further controlling

for other factors that may also aﬀect shopping intensity (for instance, a stronger expectation of

selling the prop erty soon). The former type of regression is called “univar.” in Table

A-4 while the

latter type is called “multivar.”

In Table A-5, we report similar regressions but with the knowledge measures as dependent

variables (and only the borrower and loan characteristics as independent variables). Note that for

the ﬁrst two of the three outcomes in that table, higher values correspond to more knowledge, while

for the last one, the opposite is true. We discuss the results from both tables jointly, since in some

cases they contrast in interesting ways.

The ﬁrst three rows of Table A-4 indicate that borrowers that are more knowledgable also shop

more. Of course, in this case it is diﬃcult to rule out reverse causality, namely that the additional

shopping made them more knowledgable (f or instance, about price diﬀerences across lenders). The

fourth coeﬃcient shows that people who say that they were “not at all concerned about qualifying

for a mortgage when they began the process of getting this mortgage” also report shopping less.

This suggests that less conﬁdence in one’s ability to qualify for a loan can have the beneﬁcial side

eﬀect of inducing additional shopping.

Next, we reproduce the positive relationship between PMMS and shopping measures docu-

mented in Table 10.

We further see that mortgage knowledge tends to be slightly lower when

PMMS is higher, although the relationship is no longer signiﬁcant once other variables are con-

trolled for.

Turning to borrower and loan characteristics, we see that borrowers with higher FICO scores

are more likely to have seriously considered more than on e lender, although for the other shopping

measures the evidence is more mixed. Howe ver, high-FICO borrowers tend to be substantially more

knowledgable, e specially when considering the univariate correlations with mortgage-rate familiarity

and the knowledge index. There is no signiﬁcant relation between FICO and the propensity to think

that all lenders oﬀer similar terms.

Borrowers with higher LTVs tend to shop more, but are less knowledgable. Similarly, FHA

borrowers do not appear to shop less, but tend to be signiﬁcantly less knowledgable than other

borrowers (except that they do have a slightly higher propensity to believe in price dispersion).

Given that our earlier Optimal Blue analysis found that these groups see substantially higher

locked-oﬀered rate gaps, these patterns suggest that knowledge may be the key diﬀerential driver

of those patterns. Similarly, we also see that borrowers with purchase loans, and especially ﬁrst-

time homebuyers, report higher shopping intensity, but are substantially less knowledgable than

reﬁnancers (which makes sense, since the latter likely have more experience with the process).

Borrowers with larger loan amounts, and especially jumbo borrowers, both shop more and are

more knowledgable – in line with their lower rate spreads.

Finally, in terms of borrower demographics, more educated respondents are much more likely to

shop, and have better mortgage knowledge. Income appears to have little eﬀect on shopping once

other factors are controlled for, but still correlates signiﬁcantly with knowledge. Finally, we see

that minorities appear to shop more that Non-Hispanic White borrowers (the omitted category),

but were less familiar with mortgage rates and have a lower knowledge index. However, they are

more likely to believe in price dispersion.

This self-assessed creditworthiness was also used as a control variable in Table

The coeﬃcients diﬀer slightly because in this section, we use less ﬁne control variables.

Table A-1: The real-time interest rate dispersion for oﬀered mortgage products with no points

and fees

Median Median Standard Percentile Diﬀerences

No. Oﬀers Rate Deviation 75

− 25

− 10

− 1

Program

Conforming 115 4.24 0.17 0.23 0.44 0.73

Super-Conforming 143 4.47 0.18 0.25 0.47 0.75

Jumbo 97 4.59 0.17 0.22 0.45 0.77

FHA 112 3.79 0.19 0.29 0.53 0.77

FICO

640 113 4.62 0.18 0.25 0.48 0.76

680 110 4.34 0.17 0.24 0.46 0.75

720 118 4.20 0.17 0.23 0.45 0.75

750 118 4.13 0.17 0.23 0.45 0.75

LTV (%)

70 118 4.24 0.17 0.23 0.45 0.75

80 116 4.31 0.18 0.24 0.47 0.76

90 105 4.51 0.17 0.23 0.45 0.75

95 125 4.34 0.17 0.23 0.45 0.73

96 112 4.01 0.18 0.26 0.49 0.76

Note - This table compares real-time interest rates for identical oﬀered mortgages (same FICO, LTV, DTI, loan

amount, location, time etc.) with no points and fees. Column 1 shows the median number of lenders oﬀering each

mortgage product in a location on a speciﬁc day. Columns 3-5 show the diﬀerence between various percentiles of

the oﬀer distribution.

Table A-2: The real-time interest rate dispersion for oﬀered conforming mortgages with no points and fees

Median Median Percentile Diﬀerences

No. Oﬀers Rate 75

− 25

− 10

− 1

Atlanta, GA 104 4.20 0.20 0.39 0.64

Boston-Worcester-Lawrence, MA-NH-ME-CT 68 4.15 0.22 0.47 0.77

Charlotte-Gastonia-Ro ck Hill, NC-SC 81 4.22 0.19 0.39 0.70

Chicago-Gary-Kenosha, IL-IN-WI 109 4.14 0.23 0.42 0.69

Cleveland-Akron, OH 47 4.24 0.24 0.46 0.70

Dallas-Fort Worth, TX 129 4.19 0.22 0.40 0.68

Denver-Boulder-Greeley, CO 122 4.20 0.19 0.37 0.62

Detroit-Ann Arbor-Flint, MI 70 4.14 0.22 0.42 0.75

Las Vegas, NV 77 4.33 0.21 0.42 0.72

Los Angeles-Riverside-Orange County, CA 154 4.20 0.23 0.44 0.72

Miami-Fort Lauderdale, FL 102 4.21 0.27 0.44 0.73

Minneap oli s-St. Paul, MN 72 4.18 0.19 0.37 0.71

New York-Northern New Jersey-Long Island 97 4.17 0.25 0.47 0.75

Pho enix-Mesa, AZ 107 4.23 0.22 0.40 0.69

Portland-Salem, OR 87 4.22 0.21 0.39 0.71

San Diego, CA 113 4.20 0.21 0.40 0.66

San Francisco-Oakland-San Jose, CA 120 4.20 0.21 0.40 0.69

Seattle-Tacoma-Bremerton, WA 103 4.20 0.21 0.35 0.66

Tampa-St. Petersburg-Clearwater, FL 116 4.22 0.22 0.41 0.68

Washington-Baltimore, DC-MD-VA 114 4.18 0.21 0.40 0.68

Note - This table compares real-time interest rates for 30 year ﬁxed rate conforming mortgages with a LTV=80, FICO=750,

DTI=36, and with no points and fees. Column 1 shows the median number of lenders oﬀering mortgages in a location on a s peciﬁc

day. Columns 3-5 show the diﬀerence between various percentiles of the oﬀer distribution.

Table A-3: Dispersion in points and fees that lenders charge

to originate at the median interest rate

Percentile Diﬀerences

− 25

− 10

− 1

Program

Conforming 1.15 2.19 3.63

Super-Conforming 1.24 2.37 3.76

Jumbo 1.08 2.24 3.86

FHA 1.43 2.63 3.85

FICO

640 1.23 2.39 3.82

680 1.19 2.32 3.76

720 1.16 2.26 3.76

750 1.17 2.27 3.75

LTV

70 1.14 2.24 3.75

80 1.21 2.34 3.80

90 1.15 2.24 3.76

95 1.17 2.25 3.67

96 1.32 2.46 3.80

Note - This table compares real-time points and fees charged by

diﬀerent lenders to originate identical mortgages at the median interest

rate. Points and fees are given as percent of the mortgage balance. The

median interest rate is the same as in Table 1, and the average lender

charges no points and fees at this interest rat e.

Table A-4: Relationship between various binary measures of mortgage shopping and characteristics

of borrower and loan.

Considered 2+ lenders Applied to 2+ lenders Used other lenders Used web

for better terms to get info to get info

Univar. Multivar. Univar. Multivar. Univar. Multivar. Univar. Multivar.

(1) (2) (3) (4) (5) (6) (7) (8)

Very familiar with mortgage rates 0.057*** 0.045*** -0.007 0.009 0.022*** 0.010 0.003 0.008

(0.008) (0.009) (0.007) (0.007) (0.008) (0.009) (0.008) (0.009)

Index of mortgage knowledge (Std) 0.047*** 0.033*** 0.005* 0.006 0.030*** 0.022*** 0.038*** 0.041***

(0.004) (0.004) (0.003) (0.004) (0.004) (0.004) (0.004) (0.004)

Most lenders oﬀer same rate? Yes - 0.085*** -0.076*** -0.052*** -0.051*** -0.071*** -0.063*** -0.017 -0.014

(0.012) (0.012) (0.010) (0.010) (0.012) (0.012) (0.012) (0.011)

Not concerned about qualifying for mtg. -0.047*** -0.076*** -0.052*** -0.044*** -0.068*** -0.095*** -0.073*** -0.092***

(0.008) (0.009) (0.006) (0.007) (0.008) (0.009) (0.008) (0.009)

Market mortgage rate (PMMS) 0.045** 0.046** 0.069*** 0.063*** 0.048*** 0.050*** 0.019 0.027

(0.018) (0.018) (0.014) (0.014) (0.018) (0.018) (0.018) (0.018)

FICO/100 0.015** 0.017** -0.015*** -0.002 0.008 0.013* -0.005 0.017**

(0.006) (0.007) (0.005) (0.006) (0.006) (0.007) (0.006) (0.007)

LTV/100 0.051** 0.007 0.130*** 0.052*** 0.049** 0.045* 0.187*** 0.088***

(0.020) (0.025) (0.015) (0.019) (0.020) (0.025) (0.020) (0.025)

Loan amount > 200k 0.081*** 0.034*** 0.029*** 0.018** 0.083*** 0.049*** 0.061*** 0.009

(0.008) (0.009) (0.006) (0.008) (0.008) (0.009) (0.008) (0.009)

Jumbo 0.116*** 0.042** 0.017 0.000 0.116*** 0.047** -0.018 -0.073***

(0.020) (0.020) (0.016) (0.017) (0.020) (0.021) (0.020) (0.020)

FHA -0.004 -0.000 0.031*** -0.007 -0.010 -0.005 0.031*** 0.005

(0.011) (0.013) (0.010) (0.011) (0.011) (0.013) (0.011) (0.013)

VA/FSA -0.005 0.002 0.005 -0.013 0.009 0.014 0.003 0.019

(0.012) (0.014) (0.010) (0.011) (0.012) (0.014) (0.012) (0.014)

Purpose = home purchase 0.045*** 0.037*** 0.058*** 0.041*** 0.023*** 0.013 0.030*** -0.064***

(0.008) (0.010) (0.006) (0.008) (0.008) (0.010) (0.008) (0.010)

First-time homebuyer 0.048*** 0.023* 0.067*** 0.016 0.019* 0.003 0.148*** 0.110***

(0.011) (0.013) (0.009) (0.012) (0.011) (0.013) (0.010) (0.013)

At least college degree 0.087*** 0.053*** 0.028*** 0.018** 0.076*** 0.053*** 0.133*** 0.090***

(0.008) (0.009) (0.006) (0.007) (0.008) (0.009) (0.008) (0.009)

Household income > 100k 0.060*** 0.004 0.004 -0.010 0.050*** -0.000 0.057*** 0.015

(0.008) (0.010) (0.006) (0.008) (0.008) (0.010) (0.008) (0.010)

White Hispanic 0.033** 0.032** 0.057*** 0.042*** 0.014 0.010 0.052*** 0.043***

(0.016) (0.016) (0.013) (0.014) (0.016) (0.016) (0.016) (0.016)

Black 0.061*** 0.067*** 0.060*** 0.052*** -0.000 -0.010 0.055*** 0.052***

(0.017) (0.017) (0.014) (0.015) (0.016) (0.017) (0.017) (0.016)

Asian 0.115*** 0.061*** 0.030** 0.003 0.119*** 0.071*** 0.149*** 0.088***

(0.017) (0.017) (0.014) (0.015) (0.017) (0.017) (0.016) (0.016)

Other race 0.063*** 0.055** 0.046** 0.034* 0.038 0.028 0.057** 0.041*

(0.024) (0.024) (0.020) (0.020) (0.024) (0.024) (0.024) (0.022)

Mean of Dependent Variable 0.510 0.190 0.418 0.533

Adj. R2 0.04 0.03 0.03 0.07

Obs. 19906 19906 19906 19906

Sample restricted to ﬁrst-lien loans (without a junior lien) for single-family principal residence properties, with

no more than two borrowers, and a loan term of 10, 15, 20 or 30 years. All four dependent variables are binary.

Observations weighted by NSMO sample weights. The univariate regressions (odd columns) only feature one of

the covariates in the table, along with survey wave ﬁxed eﬀects. The multivariate regressions (even columns)

simultaneously control for all the variables listed in the table, survey wave ﬁxed eﬀects, and the following additonal

variables: indicators for single borrowers, cash-out reﬁnances, whether the household owns 4 diﬀerent types of

ﬁnancial assets, metropolitan CRA low-to-moderate income tract status, borrower age and gender, and self-assessed

likelihood of moving, selling, or reﬁnancing, as well as risk aversion. Robust standard errors in parentheses. * p<0.1,

** p<0.05, *** p<0.01.

Table A-5: Relationship between various measures of mortgage knowledge and characteristics of

borrower and loan.

Very familiar with Knowledge Index Thinks all lenders

mortgage rates (std) oﬀer same terms

Univar. Multivar. Univar. Multivar. Univar. Multivar.

(1) (2) (3) (4) (5) (6)

Market mortgage rate (PMMS) -0.061*** -0.022 -0.074** -0.003 -0.017 -0.024

(0.018) (0.017) (0.037) (0.034) (0.040) (0.039)

FICO/100 0.113*** 0.046*** 0.179*** 0.018 0.002 0.001

(0.006) (0.007) (0.013) (0.014) (0.008) (0.009)

LTV/100 -0.398*** -0.049** -0.658*** -0.096** 0.117*** 0.081**

(0.019) (0.023) (0.041) (0.049) (0.027) (0.035)

Loan amount > 200k 0.117*** 0.023*** 0.331*** 0.079*** -0.020** -0.015

(0.008) (0.009) (0.016) (0.018) (0.010) (0.012)

Jumbo 0.173*** 0.023 0.501*** 0.103*** -0.124*** -0.121***

(0.017) (0.017) (0.037) (0.037) (0.027) (0.028)

FHA -0.189*** -0.031** -0.344*** -0.063** -0.022 -0.040**

(0.011) (0.013) (0.023) (0.025) (0.015) (0.017)

VA/FSA -0.055*** 0.001 -0.116*** -0.047* 0.021 0.009

(0.012) (0.013) (0.025) (0.026) (0.015) (0.017)

Purpose = home purchase -0.168*** -0.051*** -0.181*** 0.009 0.044*** 0.043***

(0.008) (0.009) (0.016) (0.019) (0.010) (0.014)

First-time homebuyer -0.322*** -0.206*** -0.413*** -0.156*** 0.012 -0.043**

(0.010) (0.013) (0.021) (0.025) (0.014) (0.017)

At least college degree 0.067*** 0.014* 0.285*** 0.147*** 0.006 -0.000

(0.008) (0.008) (0.016) (0.017) (0.011) (0.012)

Household income > 100k 0.180*** 0.067*** 0.457*** 0.174*** -0.010 0.001

(0.008) (0.009) (0.015) (0.018) (0.010) (0.013)

White Hispanic -0.104*** -0.021 -0.224*** -0.061** -0.075*** -0.066***

(0.016) (0.015) (0.032) (0.030) (0.021) (0.021)

Black -0.102*** -0.027 -0.074** 0.059* -0.131*** -0.116***

(0.017) (0.017) (0.032) (0.032) (0.022) (0.023)

Asian -0.042** -0.070*** -0.086** -0.230*** -0.102*** -0.079***

(0.017) (0.016) (0.035) (0.034) (0.022) (0.023)

Other race -0.076*** -0.029 -0.070 -0.004 -0.115*** -0.110***

(0.024) (0.023) (0.051) (0.048) (0.033) (0.032)

Mean of Dependent Variable 0.617 -0.025 0.682

Adj. R2 0.14 0.16 0.02

Obs. 19906 19906 10275

Sample restricted to ﬁrst-lien loans (without a junior lien) for single-family principal residence properties, with no

more than two borrowers, and a loan term of 10, 15, 20 or 30 years. The dependent variables are binary except in

columns (3)-(4), where the knowledge index is standardized to have mean 0 and standard deviation 1 (in unweighted

sample). Observations weighted by NSMO sample weights. The univariate regressions ( odd columns) only feature

one of the covariates in the table, along with survey wave ﬁxed eﬀects. The multivariate regressions (even columns)

simultaneously control for all the variables listed in the table, survey wave ﬁxed eﬀects, and the following additonal

variables: indicators for single borrowers, cash-out reﬁnances, whether the household owns 4 diﬀerent types of ﬁnancial

assets, metropolitan CRA low-to-moderate income tract status, borrower age and gender, and self-assessed likelihood

of moving, selling, or reﬁnancing, as well as risk aversion. Robust standard errors in parentheses. * p<0.1, ** p<0.05,

*** p<0.01.

.01

.02

.03

.04

.05

Fraction

-.6 -.5 -.4 -.3 -.2 -.1 0 .1 .2 .3 .4 .5 .6

Spread in Offer Rates

Conforming, $300k, FICO=750, LTV=80, DTI=36, No Points/Fees

.01

.02

.03

.04

Fraction

-.6 -.5 -.4 -.3 -.2 -.1 0 .1 .2 .3 .4 .5 .6

Spread in Offer Rates

FHA, $300k, FICO=680, LTV=96, DTI=36, No Points/Fees

.01

.02

.03

.04

.05

Fraction

-.6 -.5 -.4 -.3 -.2 -.1 0 .1 .2 .3 .4 .5 .6

Spread in Offer Rates

Jumbo, $700k, FICO=750, LTV=80, DTI=36, No Points/Fees

Figure A-1: Interest Rate Oﬀer Dispersion for Identical Mortgages in Los Angeles

Note: The spread is deﬁned as the diﬀerence between real-time mortgage rate oﬀers and the median oﬀered rate for identical

mortgage products. The histogram includes daily data between April 2016 and August 2018.

.01

.02

.03

.04

Fraction

-.6 -.5 -.4 -.3 -.2 -.1 0 .1 .2 .3 .4 .5 .6

Spread in Offer Rates

All Mortgages with no Points/Fees

Figure A-2: Interest Rate Oﬀer Dispersion for Identical Mortgages in 20 Metropolitan Areas

Note: The spread is deﬁned as the diﬀerence between real-time mortgage rate oﬀers and the median oﬀered rate for identical

mortgage products. The histogram includes data between April 2016 and August 2018.

.01

.02

.03

.04

.05

Fraction

-3 -2.5 -2 -1.5 -1 -.5 0 .5 1 1.5 2 2.5 3

Points and Fees (% of mortgage balance)

Distribution in Points/Fees Charged for Mortgages at the Median Interest Rate

Figure A-3: Dispersion in Points and Fees Lenders Charge for Identical Mortgages at the Median

Interest Rate

Note: Points and fees are gi ven as percent of the mortgage balance. The median interest rate is calculated as the

rate at which the average lender charges no points and fees.

3.1

3.2

3.3

3.4

3.5

3.6

3.7

3.8

3.9

4.1

4.2

4.3

4.4

4.5

4.6

4.7

4.8

01jul2017 01oct2017 01jan2018 01apr2018

date

Mortgage News Daily

Optimal Blue Insight

Conforming Mortgages

3.1

3.2

3.3

3.4

3.5

3.6

3.7

3.8

3.9

4.1

4.2

4.3

4.4

4.5

4.6

4.7

4.8

01jul2017 01oct2017 01jan2018 01apr2018

date

Mortgage News Daily

Optimal Blue Insight

FHA Mortgages

3.1

3.2

3.3

3.4

3.5

3.6

3.7

3.8

3.9

4.1

4.2

4.3

4.4

4.5

4.6

4.7

4.8

01jul2017 01oct2017 01jan2018 01apr2018

date

Mortgage News Daily

Optimal Blue Insight

Jumbo Mortgages

Figure A-4: Comparison of average oﬀer rate from Optimal Blue with Mortgage News Daily data

Note: The Optimal Blue Data is for borrowers with LTV=80, FICO=750, DTI=36, with no points/fees. The

Mortgage News Daily data is a survey that does not control for loan characteristics and points and fees. To make

the data comparable we assume that MND data is quoted for 0.5% points and fees.