Seller-Required Mortgage Preapprovals and the Homebuying Process

Abstract

Mortgage preapprovals have been commonly available for about 20 years. A buyer may benefit from a mortgage preapproval by increasing the likelihood of closing on the loan. A seller with an offer from a preapproved buyer is exposed to less risk of a transaction not closing and perhaps a quicker average time to closing. The findings show that commercial sellers are more likely to require preapprovals, especially for REO transactions. Time until the closing of a sale (TUS) is about 4.2% quicker for transactions with seller-required preapprovals. Time until the signing of a sales contract (TUC) is not less for seller-required preapprovals, but it is 15.3% quicker for REO preapproval sales. The selling price discount for preapproval properties averages 1.7%–3% for non-REOs and 3%–4% for REO properties.

Keywords:

• Mortgage preapproval; mortgage prequalification; homebuying process

Notes

1 Mortgage preapprovals commonly state the maximum loan amount for which a borrower qualifies; as such this may put the borrower at a disadvantage if that information were disclosed to the seller.

2 It is important to differentiate the buyer’s and seller’s perspectives based on an individual property. The individual buyer with a preapproval may have a reduction in closing time because the buyer’s lender already has much of the needed information to process a loan. The seller’s perspective is different because some buyers making offers are denied credit, requiring the property to be placed back on the market, and therefore lengthening the expected TOM.

3 REO is standard industry abbreviation for “real estate owned.” When a business or government entity obtains ownership of a foreclosed property, the asset is recorded on its balance sheet in the category of “real estate owned” or “other real estate owned.”

4 Lenders have different mortgage prequalification and preapproval processes, and larger lenders are more likely to have more formalized the processes and forms. For example, Quicken Loans has a Prequalified Approval that is based on “an initial interview of the income, credit and asset information you provided over the phone” according to a form letter provided to potential borrowers. The letter also suggests that the potential borrower obtain a Verified Approval, which is offered after allowing Quicken Loans to “verify more of your information,” and that a Verified Approval is “the next best thing to paying cash.” The prequalification and preapproval information in this section is condensed from numerous sources including bankrate.com, and Investopedia.com.

5 However, the seller should be aware that a mortgage preapproval is just a conditional agreement to lend, and not an absolute guarantee. Changes in a borrower’s financial situation, changes in interest rates, an unsatisfactory property appraisal, or problems with a title search can result in a lender not approving the loan request. Nonetheless, mortgage preapprovals can reduce seller uncertainty about closing the transaction.

6 Some lenders may also verify income and assets, performing a more rigorous analysis. This type of preapproval may take more time and add cost to the preapproval, but sellers and their agents may attach more credibility to a buyer’s offer.

7 Buyers benefiting the most from preapprovals may (a) have a more complicated financial history, (b) have a high probability of wanting to complete a house purchase in the time frame stipulated by the preapproval letter, (c) be pushing the limits of the loan amount, (d) desire a larger pool of available properties sold by most non-individual sellers, and (e) be price sensitive and therefore attracted to a property with a price discount.

8 Preapproval may also be advantageous for buyers that consider the purchase of Federal Housing Administration (FHA), Veterans Administration (VA), or other eligible programs with stricter property condition requirements. If a preapproved buyer wants an FHA loan, for example, and the seller knows that the property is unlikely to meet the standards required by FHA and is unwilling to improve the property to FHA standards, the preapproval avoids the time wasted seeking the loan.

9 The degree of overpricing is measured many ways such as the difference in the actual listing price and predicted listing price, the difference in the actual listing price and the predicted selling price, the difference in the actual listing price and the actual listing price, and by dummy variables indicating changes upward or downward in in the listing price. The complexity increases as some researchers use the original listing price while others use the final listing price. However, the commonality is that all of these measures seek to determine if a property has an attractive price to lower TOM, or if a property is overpriced, which lengthens TOM.

10 Some of these expenses would be incurred anyway once a non-preapproved buyer arranges for a loan after signing a contract to purchase a house. However, if a preapproved buyer is unable to find a suitable property, the buyer that has arranged for a preapproval will have incurred the costs; the buyer without preapproval will have avoided the preapproval costs.

11 The types of sellers in this study are government-sponsored enterprises (GSEs), the VA, commercial banks, other financial firms, asset management firms, builders, realty, community development, investors, other commercial, and individuals. Commercial entities refer to sellers other than individuals and GSEs (Fannie Mae and Freddie Mac).

12 Equation (1) assumes a simplified model with one expected reduction in the selling price occurring $t_m$ days after preapproval status is granted. However, there is a probability associated with each day until the preapproved buyer decides to end the search or the preapproval status expires, whichever one occurs first. If $K$= the number of days for the search, subject to the maximum days until preapproval status expires, and the probability of closing on day $k$ is $P_k,$ then $V_{pb}={\sum }_{k=1}^K{V}_{pb,k}{*P}_k.$

13 It should be noted that the expected discount ($R_b$) reflects the discount and the probability of the discount occurring. While a preapproved buyer has a greater chance of receiving the mortgage loan than a buyer without loan approval, personal financial and market conditions could change the lender’s approval decision. Also, the expected discount to the buyer is different than the expected discount offered by the seller, because some sellers may not require preapproval but offer a discount to the buyer because of the greater probability of closing on the property and reduced TOM.

14 Equation (2) assumes a simplified model with one expected reduction in the selling price occurring $t_n$ days after property is listed with the preapproval requirement. However, there is a probability associated with each day until the seller closes on the property with a preapproved buyer or the listing ends. If $N$= the number of property listing days in the contract, and the probability of closing on day $n$ is $P_n,$ then $V_{ps}={\sum }_{n=1}^N{V}_{ps,n}*P_n.$

15 Real Estate Owned (REO) properties are held by lenders, asset management companies, or government-sponsored entities such as Fannie Mae and Freddie Mac. REO properties may be sold in-house, although REO specialists often choose to sell REO properties through real estate agents who list them on the MLS. Commercial and GSE entities may sell REOs at a discount and more quickly to eliminate the costs of holding the properties. REOs are more likely to need repairs, and REO sellers often sell properties “as is” leading REOs to sell at a discount.

16 An additional complexity is that if TOM is measured by time-under-contract, some Multiple Listing Services do not include the time when a seller makes an offer that is accepted by a buyer which subsequently falls through. During this time, the property is technically “off the market” because it cannot be sold. Subtracting this time has the effect of lowering an agent’s time-on-the-market performance measure. Also, another measure of time called cumulative TOM adds the time when a property has been taken off the market and relisted in a short time. Except in a small percentage of cases, TOM and cumulative TOM are the same. This study uses cumulative TOM (or TUC in this context) and TUS as measures of duration.

17 Although TUS may be theoretically superior because sellers should base their decision on the entire time until closing, problems encountered after a sales contract is signed are often beyond the control of the agent, and largely unknown in advance.

18 In addition, the probit model is helpful for examining for potential sample selection bias. If preapproval properties have unobservable characteristics that make them different from non-preapproval properties, and if these latent characteristics significantly influence the selling price, failure to correct for this problem would result in biased regression coefficients. Sample selection methodology corrects for this bias. Using a two-step procedure described by Greene (1997), the preapproval binary variable and the inverse Mills ratio (IMR) variable (which essentially measures the probability of a preapproval) from the first-stage probit model enters the full sample regression.

19 The Weibull AFT model is commonly used in TOM duration models in real estate. The Weibull is appropriate for modeling data with monotone hazard rates that increase or decrease exponentially over time, whereas the exponential distribution assumes a constant hazard. Manning, Basu, and Mullahy (2005) indicate that the lognormal TOM specification might be favored if kappa (κ) = 0 while the Weibull model might be preferred if kappa (κ) = 1. If κ = 1 with the additional constraint that sigma (σ) = 0, the exponential is preferred.

20 The degree of overpricing is measured by the natural logarithm of the listing price, ln(${LP}_{it}),$ minus the predicted selling price, ln(${\widehat{SP}}_i$), and expressed as a percentage; ${\mathit{DOP}}_i$ $\equiv$ [${ln(LP}_i)-\ln ({\widehat{SP}}_i$)] × 100.

21 The two-stage least squares (2SLS) and instrumental variables regression approaches provide identical coefficient estimates if all variables are correctly included in the first-stage regression; however, the 2SLS standard errors need adjustment.

22 The TUS variable decreases the number of observations to 32,141 because of missing or bad data, and the removal of observations with seller-financing and lease-to-own provision contracts. Mean (standard deviation) for the reduced TUS sample is 139.3198 (113.47), with a mean (standard deviation) of 116.71 (99.15) for the 2,258 REOs and 141.03 (114.30) for the 29,883 non-REOs.

23 Regulation C, Section 203. Accessed April 2, 2021 at https://www.ffiec.gov/hmda/pdf/regulationc2004.pdf#203.2(b)(2)

24 As a robustness check to using fixed effects census blocks, a distance and neighborhood variables model was tested. Property locations were geocoded, enabling distance computations to major shopping areas, schools, parks, major roads, sports complexes, parks, streams, and bodies of water. Alternative models tested include ten distance variables and three neighborhood (census block) variables (median age, % white ethnicity, and % vacancy rate), and all the variables for the property characteristics, brokerage characteristics, and year and month fixed effects. The findings are almost identical; however, the census block fixed effects model is selected based on the higher log-likelihood statistic, and the lower AIC and BIC statistics.

25 The sample selection coefficient is not statistically significant in either model.

26 Following Halvorsen and Palmquist (1980), binary variable percentage changes are transformed by y = (e^x − 1) × 100, where x is the binary variable coefficient, and y is the resulting transformed percentage change. When including the main preapproval, REO, and REO interaction effects, the percentage discount is y = (e^{(−0.0171 − 0.1421 − 0.0314)} − 1) × 100, or 17.1%; the preapproval effect alone on REOs is y = (e^{(−0.0171 − 0.0314)} − 1) × 100, or 4.7% discount.

27 The test for underidentification is whether the excluded instruments are sufficiently correlated with endogenous regressors. The Kleibergen-Paap statistic reported in the table is heteroscedastic-robust, and rejection of the null hypothesis indicates that the model is not underidentified. The Stock-Yogo weak identification test shows that instruments are strongly correlated with the TOM endogenous variable. The Hansen J statistic is a test that instruments are valid, and it is heteroscedastic robust. The joint test is that the instruments are uncorrelated with the error term and that these excluded instruments are correctly excluded from the regression. The Hansen J statistic does not reject the null hypothesis, indicating that regression is not overidentified.

28 Adding the Preapproval, REO, and Preapproval × REO coefficients in Table 4, we get −0.187; therefore exp(−0.187) − 1 is −0.171. The same calculation, using the coefficients in Table 5, we get −0.145 therefore exp(−0.145) − 1 is −0.135. In total, the effect of preapprovals drops from −0.171 to −0.135, or by .036. So TOM decreases the discount by 3.6%, or said differently, decreases the effect on the selling price by 3.6%.

29 The pattern for preapproval growth and decline as shown in Figure 1 corresponds closely to the financial crisis as preapprovals increased rapidly for REOs during, and shortly after, the 2008–9 financial crisis.

30 The average transaction discount is $4,445 for all preapprovals, and $6,336 for REO preapprovals, using the average selling prices in Table 1 and the price discounts in Table 4.