The Role of Intermediary Competition in Government Interventions: The Case of HARP

Notes from the Vault

Vincent Yao
January 2016

The interest rate on 30-year conventional mortgages fell from around 6.5 percent in 2007–08 to below 5 percent in 2009. Yet many of the distressed borrowers who might have obtained substantial benefits from the fall in rates were unable to refinance their mortgages at the lower rates. The problem is that the drop in housing prices had made their loan to value (LTV) much higher and ineligible for regular refinancing.

In an effort to help these borrowers and reduce mortgage default rates, the federal government, working with Fannie Mae and Freddie Mac, developed the Home Affordable Refinance Program (HARP). The Obama administration originally estimated that up to 8 million borrowers could benefit from HARP. However, the program got off to a slow start, refinancing only about 300,000 loans during its first year.

This post begins with an explanation of HARP. It then discusses the findings of Agarwal et al. (2015) who provide a comprehensive analysis of HARP and analyze one reason why HARP may have failed to live up to initial expectations. Their analysis shows that as implemented, HARP gave a significant competitive advantage to current servicers over new lenders in refinancing distressed loans, which resulted in limited competition in executing HARP in the first few years.

HARP eligibility
The U.S. Treasury and the Federal Housing Finance Agency (FHFA), the regulator of the government-sponsored enterprises (GSEs), developed the Home Affordable Refinance Program (HARP) to expand the set of borrowers who could refinance their loans. Absent HARP, borrowers with a LTV ratio above 80 percent would not qualify for regular refinancing of their mortgages after 2008.1 HARP provided a mechanism for these borrowers to refinance at a lower rate. However, this program came with one important limitation: HARP is available only to prime conventional conforming loans active on GSEs' books as of March 2009.

The decision to limit HARP to prime conventional conforming mortgages had the effect of excluding some of the most distressed borrowers, including those who took out a subprime, Alt-A, or jumbo loan.2 These loans were generally not eligible to be guaranteed by the GSEs and, hence, were not on the GSEs' books.

The reason for limiting the loans to those active on the GSEs' books is that the GSEs already owned the credit risk on these mortgages. Thus, refinancing the current balance on these loans at a lower rate would not increase the GSEs' credit risk. Indeed, refinancing them at a lower rate would reduce the risk of credit losses by making it easier for borrowers to stay current on their loans and avoid default. However, if the program were extended to loans that were not on the GSEs' books, the GSEs would be guaranteeing loans to which they had no prior exposure, which would increase the GSEs' expected losses.

The requirement that the loans be conventional loans excluded loans guaranteed by federal agencies such as the Federal Housing Administration (FHA) and Veteran Affairs (VA). These loans were not included in HARP because their credit risk was being borne fully by FHA and VA rather than the GSEs.

Current servicers' advantage under HARP
In principle, borrowers wishing to refinance under HARP could go to any mortgage lender that participated in HARP. The ability to go to any lender created an important potential for competition under HARP between mortgage lenders that would maximize the share of the refinancing gains that went to the borrower. In contrast, if borrowers were required to use their current servicer, that servicer could decide if and on what terms individual borrowers could refinance based on what was best for the servicer.3

In practice, the competitive playing field favored the existing servicer for at least two reasons. First, current servicers have developed valuable "soft" information about the borrower, while new servicers could not strategically solicit borrowers for lack of that information. In addition to typical soft information to the originators at origination, current servicers also have knowledge about a borrower's detailed payment history and performance history of other assets such as credit cards. Soft information is an important determinant of the quality of the mortgage or any other lending. It is disadvantageous for a new lender to select and price the HARP borrowers without it. This superior access to borrower information allowed existing servicers to retain a market share of 28 percent to 33 percent (in different years) in a normal market.

The second advantage of existing servicers arises because lenders are required to recertify the truthfulness of information used in underwriting and pricing, such as borrower income, assets, and property value, known as representation and warranty (R&W), as well as to recertify the primary mortgage insurance. Any mortgage found to be in violation of its R&W can be put back to the originator, and the latter bears all the credit losses. The put-back risk became particularly pronounced in the wake of the financial crisis when mortgage investors and GSEs began conducting aggressive audits to recover losses. R&W liabilities of HARP loans for a new lender are very much like a new loan, while they are much reduced for existing servicers, creating preferential treatment to the incumbent lenders (see more details in Goodman 2011). In large part because of these advantages, existing servicers' market share of refinanced loans increased to 54 percent for HARP loans.

Results of reduced competition
The above analysis suggests that people refinancing under HARP may have been disadvantaged by a lack of competition in the mortgage market. A recent paper by me and several coauthors (Agarwal et al., 2015) provides evidence that the lack of competition resulted in a significant decrease in savings from mortgage refinancing to many who participated in HARP as well as a lower participation rate.

We start our analysis by quantifying the impact of HARP on mortgage refinancing activity and analyzing consumer spending and other economic outcomes among borrowers and regions exposed to the program. Borrowers received a reduction of around 140 basis points in interest rate, on average, due to HARP refinancing, amounting to about $3,500 in annual savings per borrower. There was a significant increase in the durable spending (new auto financing) by borrowers after refinancing, with a larger increase among more indebted borrowers. We also find that regions more exposed to the program saw a relative increase in nondurable and durable consumer spending (auto and credit card purchases), a decline in foreclosure rates, and faster recovery in house prices.

The principal issue our paper had to address is how to measure the increase in mortgage rates resulting from reduced competition. We start by focusing on the loan to value ratio. The LTV ratio was a major measure of credit risk for HARP loans, since lenders are not required to repull borrower credit reports for most HARP loans. The higher the LTV is, the less equity the borrower has and the more risky that individual will be in the future, implying a greater risk of loss from any violation of R&W. In particular, we compare HARP loans made for a LTV above 80 percent with a randomly assigned conforming refinance mortgage with LTV exactly at 80 percent in the same month, in the same location, and with similar credit score at the time of refinancing. In doing this comparison, we subtract differences in mortgage rates due to variation in guarantee fees charged for insurance of loans' credit risk. We call this the HARP-conforming refinance spread.

If there were perfect competition, we should expect such spread to be around zero, since lenders' competition will drive the excess return to near zero. Conversely, if the spread is significantly greater than zero, that would suggest that soft information and R&W concerns resulted in significantly less competition from potential new lenders. We found an average 16 bps (annually) of HARP-conforming refi spread over the course of 2009–12, which is substantial relative to an average 140 bps of rate savings on HARP loans. It suggests more than 10 percent of pass-through benefits of low interest rates are lost in market frictions, limiting the actual benefits to distressed homeowners who need the relief the most.

One way of checking our primary results is to see if increases in LTV are associated with higher market share for existing servicers and higher interest rates for borrowers. That's exactly what we find when we compare the riskiest borrowers with less risky borrowers based on LTV.

We define the riskiest borrowers as those with an LTV greater than 125 percent, that is, for borrowers whose houses were under water by 25 percent. The share of the same servicers is 78 percent among the borrowers. We also created a group of risky borrowers who were just below 90 percent (LTV between 80 and 90 percent). The existing servicer had a market share of 51 percent for these borrowers who were just below the borderline. Both of these rates in turn exceed the 33 percent share of existing servicers for regular refinancings in our sample.

The differences in competition are reflected in the price markups. The average price markup for those with an LTV greater than 125 percent is 34 bps, three times that of those with an LTV between 80 and 90 percent. Even after controlling for other relevant factors, the markup for HARP loans at a higher LTV or more risky borrowers is about 140 percent higher than for those less risky borrowers.

A final piece of evidence that R&W were a significant factor comes from a January 2013 change in HARP. This change provided clarity that R&W for all the HARP loans sunsets one year after its HARP date. This change helped to create a more level playing field between existing servicers and new servicers. The effect of the mitigating policy changes is very significant. The price markup in HARP loans dropped by 9 bps immediately after the new R&W policy took effect (more than 30 percent drop in relative terms), suggesting that the increased competition has helped to improve the pass-through of low interest rate to refinancing borrowers. There was also a sizable increase (6 percent relative increase) in the refinancing rate among eligible loans, implying that the enhanced competition also has helped improve HARP's reach to potential borrowers.

The U.S. Treasury and the FHFA worked with the GSEs to create the HARP program to help distressed borrowers refinance their mortgages at a lower interest rate. However, while HARP led to a substantial number of refinances, it initially proved much less effective than anticipated in helping these borrowers. While certainly the borrower specific factors or other institutional frictions (e.g., like servicer capacity constraints) may help account for part of this muted response, the Agarwal et al. (2015) study suggests that competitive frictions in the refinancing market played an important role in hampering the HARP program's overall impact. We estimate that these frictions reduced the take-up rate among eligible borrowers by between 10 and 20 percent and cut interest rate savings by between 16 and 33 basis points, amounting to $400 to $800 of annual foregone savings per borrower. By adversely affecting the most indebted borrowers, the competitive frictions in refinancing market may have substantially limited the overall impact of the program on aggregate demand and the broader economy.

A crucial implication of these results is the importance of understanding institutional detail when designing and modifying policies. Whether a program will work as intended, be less effective than intended, or possibly have unintended consequences depends on how the details of that policy interact with the incentives of important participants in the relevant market(s).

Vincent Yao is an associate professor of real estate in the Robinson College of Business at Georgia State University. The author thanks Larry Wall and Kris Gerardi for helpful comments on the paper. The views expressed here are the author's and not necessarily those of the Federal Reserve Bank of Atlanta or the Federal Reserve System. If you wish to comment on this post, please email or


Agarwal, S., G. Amromin, S. Chomsisengphet, T. Piskorski, A. Seru, and V. Yao (2015). "Mortgage Refinancing, Consumer Spending, and Competition: Evidence from Home Affordable Refinancing Program," NBER Working Paper 21512,

Goodman, L. (2011). "Topic: The Obama Administration's Response to the Housing Crisis," Testimony to the U.S. House of Representatives, Committee on Financial Services,


1 The FHFA had been appointed conservator for the GSEs and, hence, was effectively in control of them. The Treasury was involved in the process because the Treasury had guaranteed the GSEs would have positive capital after they were put into conservatorship.

2 Subprime loans are those made to borrowers with blemished credit (typically a credit score below 620). Alt-A mortgages are loans that are riskier than prime mortgages for reasons other than the credit quality of the borrower, typically they are loans with no or limited documentation. Jumbo loans are those that exceed the maximum loan amount for a conforming mortgage. That maximum has been $417,000 since 2006 for single-family one-unit properties.

3 Once the loan is originated, servicers are responsible for collecting payments from borrowers and disburse them to investors and other parties. They also actively monitor the payment history, the current home equity, and the current credit score of the borrower. They know a great deal about the borrower and the property.