• The biggest factors affecting borrowers’ access to mortgage credit are their credit scores and the type of loan requested (conforming or jumbo). But individual state foreclosure laws also have a big impact on a borrower’s odds of getting a mortgage quote.
• Independent of all other factors, judicial foreclosure laws can decrease a borrower’s odds of getting a mortgage quote by 13 percent compared to a similar borrower in a non-judicial state.
• Similarly, a borrower’s odds of getting a mortgage quote can increase by as much as 15 percent if they live in a state that allows recourse, compared to a similar borrower in states that prohibit it.

Sometimes, laws meant to protect consumers instead end up hurting those they’re designed to help, in part by limiting options and reducing market power. A powerful example includes laws aimed at protecting mortgage borrowers that can have the unintended consequence of making it less likely for borrowers to get a mortgage quote in some states, but not others.

Given the wide range of laws in the United States aimed at protecting the rights of mortgage borrowers, Zillow examined the relationship between borrower protection laws and mortgage credit access.[i] We found the likelihood of getting a mortgage quote in a given state can take as much as a 13 percent hit or receive a 15 percent boost, depending on what state regulations are in place.

Some mortgage policies are federal law, including mortgage insurance provided through Government Sponsored Enterprises (GSEs) Fannie Mae and Freddie Mac, or, more recently, protections enacted by the Dodd-Frank Wall Street Reform Act. Other protections vary across states, but can all impact foreclosure duration, cost and who bears risk.

Examples of these state laws can be broadly classified into several types, with each offering pros and cons to lenders and borrowers alike:

• Judicial foreclosure, in which a foreclosure sale must be approved by the courts.
• Power-of-sale foreclosure, in which lenders are granted the right to sell a property in the event of default.
• Recourse, which allows lenders to pursue other assets in case a borrower owes more than the resale value of the home.
• Right-of-redemption, gives a defaulter the option to repurchase the foreclosed home within a specified time period.

No matter what source you look at, the length of the foreclosure process is longer on average in judicial states than it is non-judicial states. And in the foreclosure process (as in many things in life), time is money.

Laws prohibiting recourse increase lenders’ risk of loss, since the lender bears all the risk of a foreclosure sale failing to recover the total outstanding amount on the mortgage.

Right-of-redemption laws impose additional uncertainty on the lender by potentially delaying revenues from the sale of a foreclosed home as the lender waits for the defaulter to possibly repurchase the home.

GSEs also play a critical role in shifting risk by setting the cutoff amounts for loans they are willing to purchase. A loan amount below the cutoff is considered a conforming loan and is eligible to be purchased by Fannie or Freddie, taking some risk off of the lender. If the loan amount is above this cutoff, it is considered a jumbo loan and is not eligible for GSE purchase, shifting more risk onto the individual lender.

Using a unique data set from Zillow on borrower requests for mortgages and lender responses, we can estimate the effect that these various borrower protection laws have on access to credit across states[ii].

Crossing State Lines

Our analysis revealed that where a borrower lives can play an important role in determining access to credit, consistent with accepted wisdom. But the biggest factors affecting borrowers’ access to credit are their credit scores and whether they are requesting a loan that falls below the conforming loan limit.

Maintaining strong credit helps maximize a borrower’s odds of receiving a mortgage quote. A FICO credit score in the “good” range of 640 to 719 gives a borrower 95 percent better odds[iii] of being quoted than a borrower with a “bad” credit score (below 640). With a credit score in the “very good” range of 720 to 850, odds increase an additional 27 percent over the merely “good” borrower (table 1). Having been foreclosed on in the past reduces a borrower’s odds by 33 percent; being a first-time buyer reduces the odds by 10 percent; and having never declared bankruptcy increases the odds by 34 percent. Having a higher income doesn’t hurt, either. Bumping income by 10 percent increases a borrower’s odds of being quoted by 2.5 percent.

	Table 1: Percent Change in Odds: Creditworthiness
	Good Credit (640-719)	Very Good Credit (720-850)	Foreclosed	Never Declared Bankruptcy	Income
Percent Change in Odds of Being Quoted	95%	147%	-33%	34%	2.5%
*Note: Odds are compared to a borrower with poor credit (<640), never been foreclosed, declared bankruptcy, and for a 10% increase in income respectively.

The conforming loan cutoff represents the most significant regulation affecting borrowers’ access to credit, and whether or not a requested loan qualifies for federal mortgage insurance appears to outweigh the positive and negative effects of state-level foreclosure regulations. Borrowers requesting loans above the conforming loan limit, i.e. a jumbo loan, have 50 percent lower odds of receiving a quote from a lender compared with borrowers requesting loans under the limit. And recourse laws do not appear to help borrowers requesting jumbo loans. Potential lender losses associated with jumbo loans are significant enough to overcome any benefits accompanying the lender’s ability to pursue recourse.

Beyond individual factors and federal law, state-level regulations have a large impact on borrowers’ odds of being quoted (table 2). Simply by living in a state with a judicial foreclosure process, a borrower’s odds of receiving a quote are reduced by 13 percent. And the longer the foreclosure process takes, the more those odds take a hit compared to living in a state with a faster foreclosure process. There’s also evidence that allowing recourse can improve access to credit by reducing the risk borne by lenders. Borrowers’ odds increase by as much as 15 percent if they live in a state that allows recourse, compared to states that prohibit it.

	Table 2: Percent Change in Odds: Regulations
	Judicial	Recourse	Jumbo
Percent Change in Odds of Being Quoted	-13%	15%	-50%
*Note: Odds are compared to the non-judicial, non-recourse, and conforming cases respectively

Methodology

To determine the effect borrower protection has on access to credit, we use Zillow’s database of real-time mortgage quotes to gauge lenders’ willingness to quote loan requests from June 2011 through April 2015, across states with different foreclosure laws.

We limited the data set to control for as many confounding factors as possible, following the methodology developed by Jihad Dragher and Yangfan Sun of the International Monetary Fund (IMF). We examined only those inquiries requesting 30-year, fixed-rate mortgages used to purchase a single-family primary residence, excluding VA- and FHA-insured inquiries in metro areas that cross state borders to limit demand variation across states. To determine which metros satisfied this criteria, we used the group of BEA defined metros containing multiple states in their name. Using this criteria, we identify a total of 47 metros that cross into 38 different states. Our dataset includes about 1.7 million unique loan inquiries in all 47 metros covering 35 of those states.

We measure mortgage credit access by lenders’ willingness to quote a loan inquiry, defined as the fraction of lenders that quote an inquiry in a given month, metro and state out of the total potential lenders active on Zillow during that month and in that state. The number of lenders across states is not uniform and varies from month to month. We consider a lender to be active in a state and month if it quotes at least one inquiry in that state during that month.

If a lender only offers a quote in one state of the metro, we count it as being a potential lender in all other states in the metro as well. A lender could choose to be active in only one state for a number of reasons, including costs of doing business or licensing considerations in other states. Borrower protection laws could also influence either of those decisions, and represent a loss of credit. Thus, we want the number of potential lenders in each state of a metro to reflect this.

The final step in building the dataset is to classify borrower protection regulations across states. We used data from RealtyTrac to classify states with judicial foreclosure processes or having the right-of-redemption. However, this classification is not always straightforward. The time it takes to foreclose on a property differs by state and across judicial and non-judicial states. To address this limitation, and to improve the precision of the regression results, Dragher and Sun supplemented their data with two semi-continuous measures of foreclosure length: Freddie Mac State Foreclosure Time Lines and the RealtyTrac Process Period. We did this as well.

Finally, to determine which states allow lenders to pursue recourse, we follow the classification developed by Andra Ghent and Marianna Kudlyak (table 3).

	Table 3: Classification of States by Foreclosure Laws
	Judicial	Recourse	Right-of-Redemption
Number of States in the Sample	17	28	15
Percent of Inquiries	67%	87%	46%
Total Number of States	24	40	22

Our estimation strategy used regression specifications from Dagher and Sun. But because our response variable is bounded between zero and one, we used a binomial generalized linear model with logit link, as developed by Leslie Papke and Jeffrey Wooldridge, instead of a linear probability model. This method transforms the response variable into an odds ratio p/(1 – p), where p is the fraction of lenders who quote a given inquiry.

Some advantages of this approach include:

• Limiting the predicted values to also fall between zero and one.
• Making it easier to recover the original regression specification.
• Improving the interpretation of the end points – zero and one – without requiring further transformations.

The first regression specification is:

1. F_i,m,s,y = β_m + ρ_y + δJ + ΘX_i,m,s,y + γReg_s + ε_i,m,s,y

where,

• F_i,m,s,y is the fraction of lenders that quote an inquiry (i) out of the total number of potential lenders active on Zillow during the month in metro (m) and state (s).
• β_m accounts for metro level fixed effects, and accounts for variation in credit across years y.
• J is a dummy variable capturing whether or not an inquiry is for a jumbo loan. We also supplement the judicial dummy with two other specifications to include the Freddie Mac Foreclosure Time Lines and the RealtyTrac Process Period.
• X_i,m,s,y includes all borrower characteristics, including the credit score bin dummy (good credit or very good credit), the log of income, a first-time buyer dummy, a foreclosed dummy, a bankruptcy dummy, the loan-to-value ratio and the loan-to-income ratio (borrower characteristics can be viewed in more detail in table 4).
• Reg_s includes all regulations in state (s), including judicial, right-of-redemption and recourse dummies.

	Table 4: Inquirer Characteristics
	Jumbo	Good Credit (640-719)	Very Good Credit (720-850)	Foreclosed	Never Declared Bankruptcy	First Time Buyer
Percent of Inquiries	13%	17%	82%	1%	98%	28%

The second regression specification is identical to the first, but adds interaction terms between state foreclosure regulations and jumbo inquiries. This helps capture any additional effect foreclosure laws have on credit supplied to jumbo loan inquiries above conforming loan inquiries. In the equation below is the variable capturing this interaction.

2. F_i,m,s,y = β_m + ρ_y + δJ + ΘX_i,m,s,y + γReg_s + ΦReg_sJ + ε_i,m,s,y

[i] This analysis was largely inspired by, is based on and extends a recent study by Jihad Dagher and Yangfan San of the International Monetary Fund (IMF).

[ii] It is important to understand how Zillow works: Borrowers interested in getting a mortgage submit information including their income, the type of loan they are seeking, the amount of the loan, their credit score, their down payment, other recurring debts they may have and several other variables. In response, a range of lenders offer quotes. There may be some lenders unable or unwilling to offer a quote. By looking at the fraction of lenders quoting a given loan inquiry out of the total number of potential lenders, we are able to provide real-time information about credit conditions across states with different foreclosure laws.

[iii] A borrower’s odds of being quoted are defined as the number of lenders who quote an inquiry over the total number of potential lenders who do not. So, if 10 lenders quote an inquiry and 100 do not, the odds ratio is 10/100. If one more lender quoted the inquiry, the odds ratio would be 11/99 or about an 11 percent increase in the odds of being quoted.