The Role of Household Misreporting on Mortgage Applications in the Financial Crisis

Misreporting of household income is an as-of-yet-understudied driver of the Financial Crisis. The basic argument is that some borrowers were able to obtain loans that they could not pay back, because they exaggerated their incomes on their loan applications. Two recent papers published in the Journal of Finance (1) document that some borrowers indeed misreported their incomes before the crisis, (2) find that borrowers who misreported were more likely to default post-crisis, even after controlling for observable borrower risk characteristics, and (3) explore why people might misreport.

Using loan applications from an unnamed U.S. Bank, Garmais (2015) first shows that loan applicants tended to round their incomes up to the nearest multiple of $100,000 in order to obtain larger loans. Then, he finds that (1) counties in which a greater fraction of loan applicants inflate their incomes have higher mortgage default rates, and (2) misreporting is more common in counties with a greater fraction of non-U.S. citizens, non-English speakers, and homeowners that probably have experienced negative home equity. (Misreporting is not correlated with the fraction of married or college-educated people.) From his second result, he deduces that misreporting is more common in areas with lower financial literacy and social capital. Overall, he concludes that financial illiteracy drives both misreporting (applicants misreport because they don’t know the value of their assets) and default.

Using loan origination documents from New Century Financial Corporation (New Century), Ambrose, Conklin, and Yoshida (2016) (ACY) largely corroborate Garmais’s findings; they find that applicants who could but did not verify income (1)  exaggerate their incomes by a greater amount and (2) are more likely to default. (Income exaggeration is estimated by (1) regressing actual income on borrower characteristics for those with verified income; (2) predicting income for non-verified borrowers using the estimates from (1); (3) taking the difference.) Importantly, these results hold only for non-verified borrowers that submitted W-2 forms, not for self-employed borrowers. ACY argue that misreporting is borrower-specific, not common to people in the same industry.

Pros and Cons of Regulation

  • Garmais: Limiting loans to applicants who report incomes right above multiples of $100,000 can exclude borrowers that misreport their incomes at the expense of excluding some creditworthy borrowers.
  • ACY: Banning low-documentation loans (loans originated to borrowers with little documentation to verify, for example, income) can mitigate adverse selection at the expense of excluding creditworthy self-employed people, the target market for these loans. (Banks developed low-documentation (low-doc) loans in order to serve self-employed people, for whom income verification is costly and who are less likely to misreport, because poor reputations limit future credit availability. )
The Role of Household Misreporting on Mortgage Applications in the Financial Crisis

How to Write an Introduction

First paragraph – The main question

  • The first sentence should answer, “what is the key economic friction this paper addresses?”
    • Zoom out as much as possible. You want your first line to appeal to the the widest audience possible.
  • (Explain why the implications of the friction X on outcome Y are not immediately obvious in three sentences:
    1. “The implications of X on Y are not immediately obvious.”
    2. “On the one hand, <bright side>”
    3. “On the other hand, <dark side>”)
  • State why the friction is important
    • Academically, is this a question academics are still stumped on or debating over? (If so, refer to above bullet point) Is this a nascent field on which we lack intuition?
    • Normatively, was a major law just passed or revoked? Was there a major event or troubling trend in an industry?
  • (If it fits here, state why the friction is urgent.)

Second paragraph – The setting

  • What is your setting?
    • “To address this issue, I look at …”
  • Why is it a reasonable, if not the ideal, laboratory to address the question you posed in the first paragraph?
  • (If it fits here, state why the friction is urgent.)

Data and Identification paragraphs

  • Use two or three paragraphs for these two topics.
  • If the data is novel, state that first, then describe how you use it to construct your identification strategy. Ideally, identification in previous papers was limited by their data, and, in contrast, your data allows you pursue the ideal (or close to ideal) identification strategy.
  • If the identification is novel, state that first, and describe what data you use to pursue it.

Results and Implications

  • What are the key results you want your readers to remember? State both the direction and magnitude in the simplest terms possible.
    • “A $y (y%) increase in Y leads to a $x (%x) decrease in X).
  • What are the implications for the issue in your first paragraph, i.e. what is so important about your findings?
    • Is the magnitude or direction surprising, i.e. do they overturn or corroborate intuition?
    • Do your results favor one argument/mechanism over another? Better yet, do they reconcile contradicting predictions?
  • Ideally, your results are simple enough to remember, even if they are rich enough to qualify different arguments.
How to Write an Introduction

Financing Spurs Employment Growth

Finance and Growth at the Firm Level: Evidence from SBA Loans
Brown and Earle, JF 2017

Main Question (General and Specific) and Results
This paper studies the real effects of finance. Specifically, it looks at how loans partially guaranteed by the Small Business Administration (SBA loans) generate jobs. It finds that jobs increase by 3-3.5 at the firm-level for each $1 million in SBA loans, which translates to a taxpayer cost of $21-25K per job.

The main contribution is the data: the authors have micro-level loan data with a long times-series and a sizeable cross-section. Previous studies relied on coarser approximations, such as the effect of SBA credit supply on county-level employment, or used micro-data that is limited cross-sectionally or in the times-series.

The authors match firms that received SBA loans with those that did not based on age, industry, size, year, and growth history, made possible by the richness of their micro-data.

To address selection into receiving an SBA loan and the amount obtained, the authors rely on institutional details. Lenders that obtain preferred lender status (PLP) in the Preferred Lender Program are granted lower administrative costs when they obtain partial loan guarantees through the SBA program.  This gives rise to an all-or-nothing participation strategy at the bank level. Indeed, “a large share” of all SBA loans are generated by PLP lenders. SBA loans account for only 1% of total bank business, making it unlikely that SBA demand drives banks’ locational decisions. This is important because SBA lending is highly local — 96.8% of SBA loans are given to borrowers in counties in which there was at least one PLP lender. PLP lenders are dispersed unevenly throughout the county and change over time, strengthening identification.

The conclusion, that financing has real effects, has been found in previous papers (e.g. Chodorow-Reich, QJE 2014; Mondragon, 2015). This paper corroborates the existing literature by looking into a specific institutional setting, SBA loans.

Financing Spurs Employment Growth