Screening on Loan Terms: Evidence from Maturity Choice in Consumer Credit

Screening on Loan Terms: Evidence from Maturity Choice in Consumer Credit Abstract We exploit a natural experiment in the largest online consumer lending platform to provide the first evidence that loan terms, in particular maturity choice, can be used to screen borrowers based on their private information. We compare two groups of observationally equivalent borrowers who took identical unsecured 36-month loans; for only one of the groups, a 60-month loan was also available. When a long-maturity option is available, fewer borrowers take the short-term loan, and those who do default less. Additional findings suggest borrowers self-select on private information about their future ability to repay. Received December 27, 2016; editorial decision December 12, 2017 by Editor Philip Strahan. Authors have furnished an Internet Appendix, which is available on the Oxford University Press Web Site next to the link to the final published paper online Received December 27, 2016; editorial decision December 12, 2017 by Editor Philip Strahan. Authors have furnished an Internet Appendix, which is available on the Oxford University Press Web Site next to the link to the final published paper online Asymmetric information between borrowers and lenders may induce inefficiencies in credit markets (e.g., Jaffee and Russell 1976; Stiglitz and Weiss 1981). In theory, lenders can partially mitigate these inefficiencies by using a menu of loans with different rates and contract terms to screen borrowers. Screening is achieved because borrowers with a higher unobservable probability of default self-select high-cost loans that contain features that are relatively more valuable to them, such as low collateral (Bester 1985), long maturity (Flannery 1986), or lenient covenants (Levine and Hughes 2005).1 Although the screening role of contract terms is well established in theory, empirical evidence of its relevance remains elusive. Most stylized facts consistent with screening are derived from the correlation between borrower contract choice and observable information (e.g., proxies for creditworthiness or the extent of private information).2 This evidence is circumstantial at best, because, by definition, screening implies that borrowers select loan terms based on information that is not observable by the lender (or the econometrician). In an attempt to document selection on unobservables, recent work has turned to the correlation between borrowers’ contract choices and ex post measures of their creditworthiness (e.g., default).3 However, even ex ante identical borrowers will exhibit different default probabilities ex post if they face different contract terms, for example, due to moral hazard. Thus, contract choice and default may be correlated even in the absence of selection. In this paper we provide the first direct evidence of the screening role of debt contract terms. This requires showing that (1) borrowers self-select into different loans based on their private information, and (2) the option chosen by borrowers with a higher unobserved probability of default has a higher rate. We first argue that, in order to empirically disentangle selection from the causal effect of contract terms, the econometrician must compare the repayment behavior of selected and unselected borrower samples who take the same loan contract.4 We then illustrate and apply this approach in the context of consumer credit in the United States, exploiting the staggered rollout of long-maturity loans by an online lending platform, Lending Club (hereafter, LC). This allows us to compare the ex post repayment behavior of ex ante identical borrowers facing the same short-term contract, but who chose their contract facing different menu of options, and who were thus differentially selected on maturity. Maturity serves as a screening device because long maturity reduces the need to roll over debt at a higher price in the future. Higher-risk borrowers, with an uncertain future observable creditworthiness, are willing to pay higher interest rates to secure this insurance.5 Consistent with this intuition, we find that LC borrowers who take a low-rate short-term loan when a long-term option is unavailable default substantially more than observationally identical borrowers who take the same short-term loan when LC also offers a higher-priced long-term loan. Thus, maturity can be effectively used as a screening device in credit markets: offering low-rate short-maturity and high-rate long-maturity loans induces borrowers of higher unobservable risk to self-select on the high-price contract. In the empirical setting, LC borrowers choose from a menu of loan amount, maturity, and price combinations. LC offers unsecured loans for amounts between $${\$}$$1,000 and $${\$}$$35,000 in either short—36 months—or long maturities—60 months. Loan price, set according to a proprietary algorithm, is increasing in amount, borrower risk, and maturity. Before 2013, long maturity loans were available only for amounts above $${\$}$$16,000. During 2013, the available menu of long-term loan options expanded twice: (1) to loans amounts between $${\$}$$12,000 and $${\$}$$16,000 in March 2013, and (2) to loan amounts between $${\$}$$10,000 and $${\$}$$12,000 in July 2013. Crucially for our analysis, during this period LC did not change the terms of any other borrowing option nor the criteria to qualify for a loan. Our empirical strategy compares the default rate of short-term loans for amounts between $${\$}$$10,000 and $${\$}$$16,000 issued before and after the availability of the long-maturity option at the corresponding amount (i.e., before and after the borrowers were screened on maturity). By comparing across borrowers who took identical loans (same short maturity, same price) we eliminate, by construction, the possibility that differences in repayment behavior are due to the causal impact of different contractual terms (due to, for example, moral hazard or the burden of repayment). To account for changes over time in the composition of borrowers on the LC platform we estimate a difference-in-differences specification that exploits the staggered roll-out of the long-term menu options, and uses short-term loans of amounts just above and just below the $${\$}$$10,000 to $${\$}$$16,000 interval to construct counterfactuals. Intuitively, our main test compares, amongst borrowers who appear ex ante identical in all observable dimensions and who took the exact same loans, the default rate of loans between $${\$}$$10,000 and $${\$}$$16,000 that were issued before and after the long-maturity loan became available at these amounts, relative to the same change in the default rate of loans between $${\$}$$5,000 and $${\$}$$10,000 or between $${\$}$$16,000 and $${\$}$$20,000 issued during the same period. The identification assumption is that any change in the composition of borrowers within a risk category that occurs for reasons other than the menu expansion, for example, due to changes in the supply of credit by other lenders, did not affect differentially loans between $${\$}$$12,000 and $${\$}$$16,000 in March 2013, and between $${\$}$$10,000 and $${\$}$$12,000 in July 2013, relative to other amounts in the analysis sample at those dates. To further ensure that all comparisons are done across observationally equivalent borrowers, we include in our specifications controls for all the borrower characteristics recorded by LC at origination, including month of origination, four-point FICO score range, and state fixed effects, among others. We begin by documenting that self-selection into long-maturity loans occurs among borrowers who would have borrowed between $${\$}$$10,000 and $${\$}$$16,000 had the long-maturity option not been added to the menu. We find that the number of short-maturity loans between $${\$}$$10,000 and $${\$}$$16,000 drops by 14.5% after the long-maturity loans become available, relative to loans issued at amounts just above and below this interval. Further, the decline was permanent and occurred on the same month the 60-month loan appeared in the menu for the corresponding amount. Then we explore how selection on maturity relates to ex post performance. We find that the average default rate of short-maturity loans decreases by 0.8 percentage points when a long-maturity loan is available at origination relative to when it is not. This implies that borrowers who look identical ex ante from the investors’ perspective but who have a higher default risk self-select out of short-term loans and into long-term ones. Assuming that the difference in short-term loan performance is due to the 14.5% of borrowers who self-select into long maturity, these self-selected borrowers would have had a default rate 5.5 percentage points higher (0.8/14.5) than the average 36-month borrower in our sample (9.2%). The findings are thus consistent with the joint hypotheses that LC borrowers have private information related to their future repayment probability, and that this private information affects loan maturity choice. Moreover, the large economic magnitude suggests that screening on maturity provides a powerful device for identifying, among a pool of observationally identical borrowers, those with the poorest repayment prospects. For maturity to be an effective screening device, long-term loans must be costlier than short-term loans. Indeed, we find that holding borrower characteristics and loan amount constant, the APR for 60-month LC loans was on average 3.3% higher than the APR for 36-month LC loans during our sample period. This represents a large maturity premium relative to the contemporaneous yield curve (0.2 percentage points) and can be fully explained by the 5.5 percentage point higher default rate of those borrowers who select into the long-maturity option.6 Consistent with a screening interpretation, only borrowers who are more exposed to repricing risk and most value the insurance provided by the long-maturity loan are willing to pay this higher maturity premium. Having established that borrowers select maturity based on private information that correlates with their repayment prospects, we turn to understanding the economic nature of this private information. Definitively characterizing the specific content of a borrower’s private information is by definition a difficult exercise because many underlying sources of information could explain differences in repayment behavior. Therefore, relative to our precise measurement of screening itself, we rely on the suggestive evidence our empirical setting permits. In theory, borrowers who are privately informed about their own high risk aversion will select the higher insurance against repricing risk provided by longer maturity loans (De Meza and Webb 2001). However, if risk averse borrowers are also expected to default less, self-selection on risk aversion is inconsistent with the higher default rate exhibited by long maturity borrowers. In addition, it is unlikely that borrowers are privately informed (relative to LC’s investors) about interest rate risk, the probability of credit supply shocks, or other macro determinants of the future cost of borrowing. It follows that borrowers who select long-maturity loans privately place higher value on the insurance it provides either because (1) they are less likely to have sufficient funds to make loan repayments in the future (e.g., they face a higher probability of job loss or illness or they will have a lower discretionary income), or (2) they are more exposed to rollover risk due to privately observed differences in the timing of their income. The two explanations have different predictions regarding the timing and level of default by borrowers who self-select into long maturity. Regarding the timing of default, borrowers who self-select into long maturity because their income arrives later will tend to default less over time, as their income realizes. In contrast, borrowers who self-select into long-maturity loans because they are more exposed to future shocks to their ability to repay default more over time, as the negative shocks realize. We find that selection does not significantly affect repayment during the first 12 months after origination, even though, unconditionally, more than a third of the loans that default do so during this period. In other words, we reject the hypothesis that the propensity to default of borrowers who self-select into long maturity loans decreases over time (relative to borrowers who self-select into short maturity loans). Regarding the level of default across maturities, if borrowers prefer a long- over a short-maturity loan because their income arrives in the future, their default probability should be lower under a long-term loan that aligns payments better with the timing of income. In our setting, however, the average default probability of 60-month loans is 3 percentage points higher than that of 36-month loans (conditioning on loan amount, month of origination, and FICO). This evidence is inconsistent with borrowers self-selecting on the basis of the timing of their income, and consistent with them self-selecting on private information about the exposure to shocks to their ability to repay. We find additional evidence in support of the interpretation that borrowers select maturity based on private information about the mean or the volatility of their discretionary income, for example, about the probability of losing a job or the need to take care of an elderly parent. We find that, on average, borrowers in the selected group—borrowers who chose the short maturity when the loan maturity was available— have higher future FICO scores and less time-series volatility of FICO scores relative to the unselected group. Thus, borrowers in the selected group are both observably more creditworthy, as measured by their FICO scores, and less exposed to shocks to their creditworthiness. Moreover, we find that the propensity for borrowers to prepay the short-term loan is lower in the selected group relative to the unselected group. Although this result is not statistically significant, it is inconsistent with the hypothesis that short-term loans are selected by borrowers based on private information that their income arrives sooner. In theory, our results could also be driven by borrowers who have a preference for long-term loans for behavioral reasons (e.g., borrowers may evaluate the price of a loan by the installment amount instead of by the interest rate and fees) and who, at the same time, are more likely to default. However, 87% of LC borrowers claim to use the LC loan proceeds to repay credit card debt. Since credit card debt is essentially very long-term debt, most borrowers in our sample are actively choosing to lower the maturity profile of their debt and to increase, not decrease, the monthly installment amounts.7 This provides suggestive evidence that LC borrowers seem to be unconstrained enough to commit to increase their minimum monthly payments relative to those imposed by their existing credit card debt and sufficiently sophisticated to understand the difference between price and monthly payment amounts. Moreover, it is important to note that, for unconstrained sophisticated borrowers, loan maturity (a contractual feature of the loan) is distinct from the actual timing of loan repayments (a choice variable). Insofar as the borrower has ongoing access to credit markets, an impatient borrower who has a short-term loan can lower the effective out-of-pocket payments by undertaking additional borrowing each period. Our paper is related to but distinct from the theory of Diamond (1991), who uses a framework with asymmetric information to predict a link between observable creditworthiness and the type of maturity that all borrowers will pool on in equilibrium. By isolating screening on private information, our paper is also distinct from theories of maturity choice that are unrelated to ex ante asymmetric information such as asset maturity matching (e.g., Myers 1977; Hart and Moore 1994), agency problems (e.g., Hart and Moore 1995), market conditions (e.g., Bosworth 1971; Taggart 1977), minimize rollover risk (e.g., Graham and Harvey 2001), predictable violations of the expectations hypothesis (e.g., Baker et al. 2003), and government behavior (e.g., Greenwood et al. 2010). Our paper contributes to this literature by relating maturity choice to a borrower’s private, that is, unobservable, information. Our paper also contributes to a relatively small empirical literature that has measured adverse selection in credit markets. Karlan and Zinman (2009) use an experiment in South Africa that isolates adverse selection on loan interest rates by randomizing the offered loan interest rate but resetting all loan terms after selection occurs. A different approach is taken by Adams et al. (2009) and Dobbie and Skiba (2013) estimate adverse selection on loan amount among subprime borrowers as a residual, given by the correlation between default and loan size that cannot be explained by the direct effect of loan size on default. Our results not only constitute evidence of adverse selection on a novel contract term (maturity), but also demonstrate that screening on maturity allows the lender to charge prices that are commensurate with borrowers’ unobserved default risk. Finally, our results suggest that the screening role of maturity may extend to other settings where long-term contracts provide insurance against repricing risk, such as labor (Holmstrom 1983) and health insurance markets (Cochrane 1995; Finkelstein et al. 2005). 1. Setting LC is the largest online lending platform in the United States. In 2014 alone, LC originated $${\$}$$4.4B in consumer loans across 45 states. By comparison, Prosper Marketplace, its nearest rival, originated $${\$}$$1.6B in the same year.8 LC loans are unsecured amortizing loans for amounts between $${\$}$$1,000 and $${\$}$$35,000 (in $${\$}$$25 intervals). LC loans are available in two maturities: 36 months, which is available for all amounts, and 60 months, which is available for different amounts at different points in time. Loans are funded directly by institutional and retail investors (LC holds no financial stake in the loans), and 80% of the total funds are provided by institutional investors (Morse 2015). Since each loan is considered an individual security by the Securities and Exchange Commission, the agency that regulates online loan marketplaces in the United States, LC is required to reveal publicly all the information used to evaluate the risk of each loan. This is an ideal institutional setting for the purposes of studying screening on borrowers’ private information, since we have all the borrower information that lenders observe at the time of origination. When a borrower applies for a loan with LC, she first enters her yearly individual income and sufficient personal information to allow LC to obtain the borrower’s credit report. In most cases (e.g., 71% of all loans issued in 2013) LC verifies the yearly income that a borrower enters using pay stubs, W2 tax records, or by calling the employer. Every loan application is processed in two steps. First, LC decides whether a borrower is eligible for a loan on the platform. The eligibility decision is made mechanically based purely on hard borrower information observable at the time of origination. For example, during 2013 LC issued loans only to borrowers with FICO scores over 660, nonmortgage debt payments to income ratios below 35%, and credit histories of at least 36 months. If LC determines that a borrower is eligible for a loan in the first step, she is then assigned to one of 25 risk categories (labeled by LC as risk “subgrades”). This assignment is made using a proprietary credit-risk assessment algorithm that uses the hard information in a borrower’s credit report (e.g., FICO score, outstanding debt, repayment status) and income. The assignment to risk category is made prior to the borrower selecting a loan amount or maturity and is therefore independent of both choices. The risk category determines the entire menu of interest rates faced by the borrower, for all loan amounts and for the two available maturities. That is, two borrowers assigned to the same risk category at the same time will face the same menu of interest rates for all amounts and for the two maturities. Interest rates for each subgrade are weakly increasing in amount and strictly increasing in maturity (ceteris paribus). The terms of all loans, other than interest rate, amount, and maturity, are identical. Once a borrower selects a loan from the menu, it is listed on LC’s website for investors’ consideration. Investors cannot affect any of the terms of the loan: they decide only whether or not to fund it. According to LC, over 99% of all listed loans are funded.9 Thus, we ignore the supply side of funds in the analysis. As of 2013, LC charges an origination fee, subtracted at origination, that varies between 1.1% and 5% of the loan amount depending on credit score and a further 1% fee from all loan payments made to investors. 1.1 Lending Club LC is the largest online lending platform in the United States. In 2014 alone, LC originated $${\$}$$4.4B in consumer loans across 45 states. By comparison, Prosper Marketplace, its nearest rival, originated $${\$}$$1.6B in the same year.10 LC loans are unsecured amortizing loans for amounts between $${\$}$$1,000 and $${\$}$$35,000 (in $${\$}$$25 intervals). LC loans are available in two maturities: 36 months, which is available for all amounts, and 60 months, which is available for different amounts at different points in time. Loans are funded directly by institutional and retail investors (LC holds no financial stake in the loans), and 80% of the total funds are provided by institutional investors (Morse 2015). Since each loan is considered an individual security by the Securities and Exchange Commission, the agency that regulates online loan marketplaces in the United States, LC is required to reveal publicly all the information used to evaluate the risk of each loan. This is an ideal institutional setting for the purposes of studying screening borrowers’ private information, since we have all the borrower information that lenders observe at the time of origination. When a borrower applies for a loan with LC, she first enters her yearly individual income and sufficient personal information to allow LC to obtain the borrower’s credit report. In most cases (e.g., 71% of all loans issued in 2013) LC verifies the yearly income that a borrower enters using pay stubs, W2 tax records, or by calling the employer. Every loan application is processed in two steps. First, LC decides whether a borrower is eligible for a loan on the platform. The eligibility decision is made mechanically based purely on hard borrower information observable at the time of origination. For example, during 2013 LC issued loans only to borrowers with FICO scores over 660, nonmortgage debt payments to income ratios below 35%, and credit histories of at least 36 months. If LC determines that a borrower is eligible for a loan in the first step, she is then assigned to one of 25 risk categories (labeled by LC as risk “subgrades”). This assignment is made using a proprietary credit-risk assessment algorithm that uses the hard information in a borrower’s credit report (e.g., FICO score, outstanding debt, repayment status) and income. The assignment to risk category is made prior to the borrower selecting a loan amount or maturity and is therefore independent of both choices. The risk category determines the entire menu of interest rates faced by the borrower, for all loan amounts and for the two available maturities. That is, two borrowers assigned to the same risk category at the same time will face the same menu of interest rates for all amounts and for the two maturities. Interest rates for each subgrade are weakly increasing in amount and strictly increasing in maturity (ceteris paribus). The terms of all loans, other than interest rate, amount, and maturity, are identical. Once a borrower selects a loan from the menu, it is listed on LC’s website for investors’ consideration. Investors cannot affect any of the terms of the loan: they decide only whether or not to fund it. According to LC, over 99% of all listed loans are funded.11 Thus, we ignore the supply side of funds in the analysis. As of 2013, LC charges an origination fee, subtracted at origination, that varies between 1.1% and 5% of the loan amount depending on credit score and a further 1% fee from all loan payments made to investors. 1.2 Staggered expansion of 60-month loans Before March 2013, 60-month loans were available only for loans of $${\$}$$16,000 and above. A borrower could not synthetically create a 60-month loan for an amount less than $${\$}$$10,000 using prepayment, because prepayment reduces the number of installments without changing their amount, effectively reducing the maturity of the loan. In March 2013, LC introduced 60-month loans between $${\$}$$12,000 and $${\$}$$16,000 to the menu. And in July 2013, it further expanded the available 60-month loans to include amounts between $${\$}$$10,000 and $${\$}$$12,000. The consequences of the menu expansion can be seen in Figure 1, where we plot the fraction of loans originated every month that have a 60-month maturity, grouped by loan amount. On December 2012, the first month of the analysis sample period, around 40% of loans between $${\$}$$16,000 and $${\$}$$20,000 are 60-month loans. This fraction remains relatively constant throughout the sample period, until October 2013. The fraction of 60-month loans is zero for loan amounts below $${\$}$$16,000 in December 2012, and jumps up for $${\$}$$12,000 to $${\$}$$16,000 loans in March 2013, and then for $${\$}$$10,000 to $${\$}$$12,000 loans on July 2013. By the end of the sample the fraction of 60-month loans stabilizes at around 30% for $${\$}$$12,000 to $${\$}$$16,000 loans and around 25% for $${\$}$$10,000 to $${\$}$$12,000 loans. The fraction of 60-month $${\$}$$5,000 to $${\$}$$10,000 loans remains at zero throughout the sample period. As we discuss in detail in Section 2, our empirical strategy exploits the fact that loan amounts between $${\$}$$10,000 and $${\$}$$16,000 were affected by the expansion of a long-maturity option, and that loan amounts outside this range were not. Figure 1 View largeDownload slide Staggered expansion of 60-month loans This figure shows the time series of the number of 60-month loans by listing month for $${\$}$$10,000 to $${\$}$$12,000 and $${\$}$$12,000 to $${\$}$$16,000. Figure 1 View largeDownload slide Staggered expansion of 60-month loans This figure shows the time series of the number of 60-month loans by listing month for $${\$}$$10,000 to $${\$}$$12,000 and $${\$}$$12,000 to $${\$}$$16,000. 1.3 Summary statistics LC makes publicly available on its website all the information used to assign borrowers to risk categories, the assigned risk category, and the loan performance of all funded loans. Our main analysis is conducted using data downloaded from LC’s website as of April 2015. The data is a cross section of all loans originated at LC. Variables are measured either at the time of origination (e.g., date of loan, loan terms, borrower income and credit report data, state of residence) or at the time of the performance data download (e.g., loan status, time of last payment, current FICO score of borrower). We complement our main outcomes, which are measured as of April 2015, with measures of FICO scores obtained from two previous loan performance updates, August 2014 and December 2014.12 We use the origination date of each loan to restrict the sample period of the analysis to meet two criteria: (1) that it contains the dates in which the 60-month loan menu was expanded (March 2013 and June 2013) and that are the basis of our empirical analysis and (2) that the interest rate assigned to each amount-maturity combination remained constant within each risk category (in other words, that all menu options other than the added long-term option remained constant). Thus, the beginning and ending months of our analysis sample are determined by two dates, surrounding the menu expansion events, on which we observe that LC repriced menu options (December 2012 and October 2013). We verify empirically that the interest rates of all risk category-amount pairs for 36-month loans are unchanged between these dates.13 We further limit the sample of loans to include those for amounts between $${\$}$$5,000 (closed) and $${\$}$$20,000 (open) because the interest rate schedule jumps discretely at $${\$}$$5,000 and $${\$}$$20,000 for all credit risk categories.14 This interval includes all 36-month loans issued at amounts affected by the 60-month borrowing threshold reduction ($${\$}$$10,000 to $${\$}$$16,000), as well as amounts above and below this interval that allow us to control for any time-of-origination changes in unobserved borrower creditworthiness or credit demand. Finally, we further limit our sample to those loans for which we can uniquely match the loan that a borrower chose to the menu associated with the risk category she was assigned to based on her publicly available data. We obtain this unique match for 98.6% of all loans in the sample period (we drop observations for which this matching does not yield a unique value). Our final sample has 60,514 loans.15 Table 1, panel A, presents summary statistics for the subset of our sample corresponding to the 12,091 36-month loans with amounts between $${\$}$$5,000 and $${\$}$$20,000 issued between December 2012 and February 2013 (prior to the menu expansion). On average, loans for this subsample have a 16.3% APR and a monthly installment of $${\$}$$380. Borrowers self-report that 87% of all loans were issued to refinance existing debt (this includes “credit card” and “debt consolidation”). We define a loan to be in default if it is late by more than 120 days.16 According to this definition, 9.2% of the loans in the subsample are in default as of April 2015. Figure 2 shows the default hazard rate by months-since-origination for loans issued before the menu expansion.17 The hazard rate exhibits the typical hump shape and peaks between 13 and 15 months. Figure 2 View largeDownload slide Hazard rate of default This figure shows the hazard rate of default by month since origination for 36-month loans issued by LC in amounts between $${\$}$$5,000 and $${\$}$$20,000 between December 2012 and February 2013 (pre-period). A loan is in default if payments are 120 or more days late in April 2015. The timing of default is the month, measured as time since origination in which payments were first missed. The hazard rate at horizon $$t$$ is the number of loans that enter default at that horizon as a fraction of the number of loans that are in good standing at $$t-1$$. Figure 2 View largeDownload slide Hazard rate of default This figure shows the hazard rate of default by month since origination for 36-month loans issued by LC in amounts between $${\$}$$5,000 and $${\$}$$20,000 between December 2012 and February 2013 (pre-period). A loan is in default if payments are 120 or more days late in April 2015. The timing of default is the month, measured as time since origination in which payments were first missed. The hazard rate at horizon $$t$$ is the number of loans that enter default at that horizon as a fraction of the number of loans that are in good standing at $$t-1$$. Table 1 Pre-period summary statistics Mean P50 SD A. Loan characteristics APR (%) 16.3 16.0 4.1 Installment ($ ${\$}$ $) 379.9 360.9 125.1 For refinancing (%) 87.0 Default (%) 9.2 Fully paid (%) 37.6 B. Borrower characteristics Annual income ($ ${\$}$ $) 65,745 57,500 74,401 Debt payments / Income (%) 17.4 16.9 7.7 FICO at origination (high range of 4-point bin) 695 689 26 FICO at latest data pull (high range of 4-point bin) 685 699 70 Home ownership (%) 55.5 Total debt excl mortgage ($ ${\$}$ $) 38,153 29,507 33,805 Revolving balance ($ ${\$}$ $) 14,549 11,592 12,719 Revolving utilization (%) 60.7 62.7 21.9 Months of credit history 182 164 84 N 12,091 Mean P50 SD A. Loan characteristics APR (%) 16.3 16.0 4.1 Installment ($ ${\$}$ $) 379.9 360.9 125.1 For refinancing (%) 87.0 Default (%) 9.2 Fully paid (%) 37.6 B. Borrower characteristics Annual income ($ ${\$}$ $) 65,745 57,500 74,401 Debt payments / Income (%) 17.4 16.9 7.7 FICO at origination (high range of 4-point bin) 695 689 26 FICO at latest data pull (high range of 4-point bin) 685 699 70 Home ownership (%) 55.5 Total debt excl mortgage ($ ${\$}$ $) 38,153 29,507 33,805 Revolving balance ($ ${\$}$ $) 14,549 11,592 12,719 Revolving utilization (%) 60.7 62.7 21.9 Months of credit history 182 164 84 N 12,091 This table shows summary statistics of the main sample of Lending Club borrowers for pre-expansion months. The main sample includes all 36-month loans with a list date between December 2012 and March 2013, and an amount between $ ${\$}$ $5,000 and $ ${\$}$ $20,000 and for which we estimate an initial risk category based on LC’s publicly available information. Table 1 Pre-period summary statistics Mean P50 SD A. Loan characteristics APR (%) 16.3 16.0 4.1 Installment ($ ${\$}$ $) 379.9 360.9 125.1 For refinancing (%) 87.0 Default (%) 9.2 Fully paid (%) 37.6 B. Borrower characteristics Annual income ($ ${\$}$ $) 65,745 57,500 74,401 Debt payments / Income (%) 17.4 16.9 7.7 FICO at origination (high range of 4-point bin) 695 689 26 FICO at latest data pull (high range of 4-point bin) 685 699 70 Home ownership (%) 55.5 Total debt excl mortgage ($ ${\$}$ $) 38,153 29,507 33,805 Revolving balance ($ ${\$}$ $) 14,549 11,592 12,719 Revolving utilization (%) 60.7 62.7 21.9 Months of credit history 182 164 84 N 12,091 Mean P50 SD A. Loan characteristics APR (%) 16.3 16.0 4.1 Installment ($ ${\$}$ $) 379.9 360.9 125.1 For refinancing (%) 87.0 Default (%) 9.2 Fully paid (%) 37.6 B. Borrower characteristics Annual income ($ ${\$}$ $) 65,745 57,500 74,401 Debt payments / Income (%) 17.4 16.9 7.7 FICO at origination (high range of 4-point bin) 695 689 26 FICO at latest data pull (high range of 4-point bin) 685 699 70 Home ownership (%) 55.5 Total debt excl mortgage ($ ${\$}$ $) 38,153 29,507 33,805 Revolving balance ($ ${\$}$ $) 14,549 11,592 12,719 Revolving utilization (%) 60.7 62.7 21.9 Months of credit history 182 164 84 N 12,091 This table shows summary statistics of the main sample of Lending Club borrowers for pre-expansion months. The main sample includes all 36-month loans with a list date between December 2012 and March 2013, and an amount between $ ${\$}$ $5,000 and $ ${\$}$ $20,000 and for which we estimate an initial risk category based on LC’s publicly available information. Table 1, panel B, shows borrower-level statistics of this sample. On average, LC borrowers in our sample have an annual income of $ ${\$}$ $65,745 and use 17.4% of their monthly income to pay debts excluding mortgages. The average FICO score at origination is 695, and credit report pulls show that the FICO score has on average decreased to 685 approximately 1 year later. LC borrowers have access to credit markets: 56% report that they own a house or have an outstanding mortgage. The average borrower has $ ${\$}$ $38,153 in debt excluding mortgage debt and $ ${\$}$ $14,549 in revolving debt, which represents a 61% revolving line utilization rate (the average revolving credit limit is $ ${\$}$ $27,464). LC borrowers have on average approximately 15 years of credit history. We compare our summary statistics to the credit card user statistics from Agarwal et al. (2015) to obtain a sense of how representative LC borrowers are of the average U.S. consumer credit user within the same FICO range. Using the average credit card limit in the subsample of borrowers with FICO scores between 660 and 719 ($ ${\$}$ $7,781) and assuming the average number of credit cards held by the average card-holder is 3.7 (according to Gallup 2014 survey) implies that the representative U.S. user of consumer credit has a revolving credit limit of $ ${\$}$ $28,789, very close to the $ ${\$}$ $27,464 average revolving credit limit of the LC borrowers in our sample. Thus, LC’s selection criteria imply that the analysis sample is drawn exclusively from prime U.S. consumer credit users (as measured by FICO scores), but LC borrowers do not seem to be different in their revolving credit availability from the average U.S. consumer credit user in the same FICO range. 2. Measuring Screening on Maturity 2.1 Empirical strategy We exploit the staggered menu expansion of 60-month loans during 2013 to identify selection on maturity. As described above, LC offered new loan options at longer maturities for amounts already offered on short-term contracts prior to the expansion. Crucially, the pricing of all loan options available prior to the expansion was unchanged after the expansion for all 25 risk categories during our sample period. This ensures that the only difference in the menu of borrowing options offered to borrowers assigned to the same risk category before and after the expansion is the availability of 60-month loans in lower amounts.18 We compare the outcomes of borrowers who took the short-term loan before the menu expanded with those who were assigned to the same risk category and took it after the expansion. We develop a research design that accounts for any other changes over time in the composition of borrowers within a risk category that are not driven by the menu expansion. The LC setting provides two sources of variation that allow us to construct a counterfactual using a difference-in-differences approach: (1) the menu expansion was staggered over time for different loan amounts (eventually selected amounts) and (2) some loan amounts were never affected by the menu expansion (never-selected and always-selected amounts). The three groups of loans defined this way by the loan amount and the time of origination are represented in Figure 3. Loans of amounts between $ ${\$}$ $10,000 and $ ${\$}$ $16,000 are eventually selected, in the sense that they are unselected at the beginning of the sample (no long-term option is available at the time of origination) and selected (long-term option is available) at the end of the sample. Since the menu expansion was staggered, loan amounts between $ ${\$}$ $10,000 and $ ${\$}$ $12,000 serve as a control group for loan amounts between $ ${\$}$ $12,000 and $ ${\$}$ $16,000 that were affected by the March expansion and the reverse applies for the expansion in July. We build two additional control groups with loan amounts whose selection status was not affected by the menu expansion. The always selected, for which the long-term loan was always available at the time of origination during the sample period ($ ${\$}$ $16,000 to $ ${\$}$ $20,000), and the never selected, for which the long-term option never became available ($ ${\$}$ $5,000 to $ ${\$}$ $10,000). Our identification assumption is that any change in the composition of borrowers within a risk category, for example, due to changes in the economic environment, changes in the borrowing options outside of LC, or changes in how LC assigns borrowers to risk categories, does not affect differentially borrowers opting to take loans between $ ${\$}$ $12,000 and $ ${\$}$ $16,000 in March and borrowers opting to take loans between $ ${\$}$ $10,000 and $ ${\$}$ $12,000 in July relative to loans issued at control amounts. Under this assumption, comparing the change in performance of eventually selected amounts before and after the menu expansion at those amounts with the change in performance of the control amounts in the same risk category isolates the effect of maturity selection induced by the menu expansion. We further include a comprehensive set of granular borrower controls, which ensures that the estimations come from comparing borrowers who took loans at selected amounts to observationally equivalent borrowers taking loans at nonselected amounts. Figure 3 View largeDownload slide Stylized depiction of identification strategy This figure shows a stylized depiction of our difference-in-differences strategy using the expansion of the menu of borrowing options. Figure 3 View largeDownload slide Stylized depiction of identification strategy This figure shows a stylized depiction of our difference-in-differences strategy using the expansion of the menu of borrowing options. Before providing evidence to support the identification assumption (see section 2.2.5 below), we discuss here its plausibility. First, even though it is unlikely that changes in economic conditions may have affected the demand for loans between $ ${\$}$ $10,000 and $ ${\$}$ $16,000 exactly at the same month of the menu expansion, to check whether there were any aggregate changes in the demand for LC loans, we plot in Figure 4 the total dollar amount of LC loans issued by month. There is no indication that the growth rate of LC lending changed around the dates of the two 60-month loan expansions. Second, in web searches we found no evidence of a change in the outside borrowing options that exclusively targeted the eventually selected loan amounts ($ ${\$}$ $10,000 to $ ${\$}$ $16,000) in a manner that corresponds with the staggered expansion of the menu. Third, we found no evidence that LC released advertising targeted at 60-month loans between $ ${\$}$ $10,000 to $ ${\$}$ $16,000 during the analysis sample. On the contrary, according to the information reported in the website Internet Archive, LC continued to advertise that 60-month loans were available only for amounts above $ ${\$}$ $16,000 until November 2013, after our analysis period ends.19 Fourth, any change in LC’s loan qualification criteria or assignment to risk categories cannot, by construction, affect borrower selection across different amounts within a risk category. The reason is that both eligibility for an LC loan and the assignment to risk categories are determined using borrowers’ observable information before the borrower selects a loan amount from the menu. Nevertheless, we verify that the criteria used to determine eligibility for an LC loan (the minimum FICO score of 660, minimum credit history length of 36 months, and maximum nonmortgage debt to income threshold of 35%) remain constant over the sample period. Figure 4 View largeDownload slide Total dollar amount issued by LC by month of listing This figure shows the time series of the total dollar amount of LC loans (of both maturities) by listing month since 2012. The vertical dashed lines show the 2 months in which the 60-month loan minimum amount was reduced. Figure 4 View largeDownload slide Total dollar amount issued by LC by month of listing This figure shows the time series of the total dollar amount of LC loans (of both maturities) by listing month since 2012. The vertical dashed lines show the 2 months in which the 60-month loan minimum amount was reduced. It is important to emphasize that there is no appropriate counterfactual for borrower selection on the 60-month loans. This is why our empirical strategy relies exclusively on a comparison of 36-month loans taken before and after the expansion, and ignores any changes in the composition of borrowers who take 60-month loans. The mix of borrowers taking a 60-month loan could have changed, for example, because some borrowers who take the 60-month loan would have not borrowed at all before this option became available. Since we are unable to account for such selection on the extensive margin for 60-month loans, we are limited in how much we can infer about the determinants of the performance of the 60-month loans. The focus on 36-month loans also implies that our approach for measuring the effect of selection is based on a revealed-preference argument, which relies on the axiom of independence of irrelevant alternatives. Specifically, we assume that a borrower who prefers not to borrow from LC over taking a 36-month loan when there is no 60-month option will not prefer to take the 36-month loan once the 60-month loan becomes available. Finally, we note that the empirical approach is aimed at estimating the effect of selection on maturity of LC loans. If LC borrowers have access to 60-month loans between $ ${\$}$ $10,000 and $ ${\$}$ $16,000 at a similar price elsewhere during the analysis period, we should fail to reject the null hypothesis and conclude that there is no screening on maturity in LC (since borrowers who wish to select long-term loans would already be taking them elsewhere). In effect, any impact of the menu expansion at LC can also be interpreted as indirect evidence that consumer credit markets are imperfectly competitive. This might be true because some intermediaries have a technology advantage over others that generates some market power or because there are search frictions in the market.20 2.2 Evidence of selection We start by measuring the amount of selection induced by the menu expansion: how does the number of borrowers who take the short-term loan at any given amount change after the long-term option becomes available at that amount? To do so we collapse the data and count the number of loans $ $N_{jkt}$ $ at the month of origination ($ $t$ $)$ $\times$ $risk category ($ $j$ $)$ $\times$ $$ ${\$}$ $1,000 loan amount bin ($ $k$ $) level for all 36-month loans issued during our sample period (amount bins measured starting from $ ${\$}$ $10,000, e.g., $ ${\$}$ $10,000 to $ ${\$}$ $11,000, $ ${\$}$ $11,000 to $ ${\$}$ $12,000, and so on). We define a “selected” dummy variable $ $D_{kt}$ $ equal to one for those loan amount bin-month pairs where a 60-month option was available, and zero otherwise. That is, \[D_{kt}=\begin{cases}1 & \mathit{if}\,\, \$16,000>\mathit{Loan}\ \mathit{Amounts}\geq\,\, \$12,000\;\&\;t\geq \mathit{March}\ 2013\\1 & \mathit{if}\,\, \$12,000>\mathit{Loan}\ \mathit{Amounts}\geq\,\, \$10,000\;\&\;t\geq \mathit{July}\ 2013\\0 & \,\mathit{otherwise}\end{cases}\] Then we estimate the following difference-in-differences regression: \begin{equation}\mathit{log}(N_{jkt})=\beta'_{k}+\delta'_{jt}+\gamma'\times D_{kt}+\epsilon_{jkt}.\label{eq:log_number_regression}\end{equation} (1) The coefficient of interest is $ $\gamma'$ $, the average percent change in the number of short-maturity loans originated for eventually selected amounts (i.e., amounts in which a long-maturity loan was not available at the beginning of the sample and became available due to the menu expansion) relative to control amounts. We include amount bin fixed effects $ $\beta'_{k}$ $, which control for level differences in the number of loans in each $ ${\$}$ $1,000 bin. In turn, risk category$ $\times$ $month fixed effects $ $\delta'_{jt}$ $ control for any changes over time in the number of borrowers who are approved at each of the 25 different risk categories. Table 2, Column 1, shows the results of regression (1), estimated on the full sample of borrowers who took a 36-month loan between $ ${\$}$ $5,000 and $ ${\$}$ $20,000 during the sample period (December 2012 to October 2013). The point estimate of $ $\gamma'$ $ is negative and significant, and implies that the number of borrowers who took a short-term loan is 14.5% lower once the new long-term loan option for the same amount becomes available. This estimate provides us with a magnitude for the number of borrowers who would have taken a short-term loan if the long term option had not been available.21 Table 2 Regression results: Selection into long-maturity loans (1) (2) (3) (4) (5) $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $D_{amount1000,t}$ $ –0.1451*** –0.0825 0.0586 –0.0355 –0.0441 (0.033) (0.065) (0.064) (0.048) (0.028) Main 60m 36m 36m Sample 16k - 24k 16k - 24k 1k - 10k Placebo Observations 3,663 1,738 1,637 2,374 3,861 $ $R^{2}$ $ 0.817 0.724 0.802 0.761 0.862 (1) (2) (3) (4) (5) $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $D_{amount1000,t}$ $ –0.1451*** –0.0825 0.0586 –0.0355 –0.0441 (0.033) (0.065) (0.064) (0.048) (0.028) Main 60m 36m 36m Sample 16k - 24k 16k - 24k 1k - 10k Placebo Observations 3,663 1,738 1,637 2,374 3,861 $ $R^{2}$ $ 0.817 0.724 0.802 0.761 0.862 This table shows that selection into the new 60-month options was higher among borrowers who would have selected a 36-month loan in the same range of amounts as the new 60-month options. The sample corresponds to loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $20,000 with a list date between December 2012 and October 2013. Column 1 shows the coefficient of the regression of log(N), the logarithm of the number of loans at each month, the credit-risk category, and the $ ${\$}$ $1,000 loan amount interval level, on a dummy that equals one for loan amounts at which the 60-month loan was first not available and then made available, and zero otherwise. Columns 2, 3, and 4 show the regression results on different samples where we redefine $ $D_{amount1000,t}$ $ in an ad hoc manner for each column. Column 2 restricts the sample to 60-month loans issued in the main sample period for amounts between $ ${\$}$ $16,000 and $ ${\$}$ $24,000; $ $D_{amount1000,t}$ $ is defined as one for loan amounts between $ ${\$}$ $16,000 and $ ${\$}$ $20,000 on and after March 2013, and zero in other cases. Column 3 restricts the sample to 36-month loans issued in the main sample period for amounts between $ ${\$}$ $16,000 and $ ${\$}$ $24,000; $ $D_{amount1000,t}$ $ is defined as one for loan amounts between $ ${\$}$ $16,000 and $ ${\$}$ $20,000 on and after March 2013, and zero in other cases. Column 4 restricts the sample to 36-month loans issued in the main sample period for amounts between $ ${\$}$ $1,000 and $ ${\$}$ $10,000; $ $D_{amount1000,t}$ $ is defined as one for loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $10,000 in and after July 2013 and zero in other cases. Column 5 reports the tests of a placebo sample that includes loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $20,000 issued between July 2013 and May 2014. Standard errors are robust to heteroscedasticity. *, **, and *** represent significance at the 10%, 5%, and 1% level, respectively. Table 2 Regression results: Selection into long-maturity loans (1) (2) (3) (4) (5) $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $D_{amount1000,t}$ $ –0.1451*** –0.0825 0.0586 –0.0355 –0.0441 (0.033) (0.065) (0.064) (0.048) (0.028) Main 60m 36m 36m Sample 16k - 24k 16k - 24k 1k - 10k Placebo Observations 3,663 1,738 1,637 2,374 3,861 $ $R^{2}$ $ 0.817 0.724 0.802 0.761 0.862 (1) (2) (3) (4) (5) $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $D_{amount1000,t}$ $ –0.1451*** –0.0825 0.0586 –0.0355 –0.0441 (0.033) (0.065) (0.064) (0.048) (0.028) Main 60m 36m 36m Sample 16k - 24k 16k - 24k 1k - 10k Placebo Observations 3,663 1,738 1,637 2,374 3,861 $ $R^{2}$ $ 0.817 0.724 0.802 0.761 0.862 This table shows that selection into the new 60-month options was higher among borrowers who would have selected a 36-month loan in the same range of amounts as the new 60-month options. The sample corresponds to loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $20,000 with a list date between December 2012 and October 2013. Column 1 shows the coefficient of the regression of log(N), the logarithm of the number of loans at each month, the credit-risk category, and the $ ${\$}$ $1,000 loan amount interval level, on a dummy that equals one for loan amounts at which the 60-month loan was first not available and then made available, and zero otherwise. Columns 2, 3, and 4 show the regression results on different samples where we redefine $ $D_{amount1000,t}$ $ in an ad hoc manner for each column. Column 2 restricts the sample to 60-month loans issued in the main sample period for amounts between $ ${\$}$ $16,000 and $ ${\$}$ $24,000; $ $D_{amount1000,t}$ $ is defined as one for loan amounts between $ ${\$}$ $16,000 and $ ${\$}$ $20,000 on and after March 2013, and zero in other cases. Column 3 restricts the sample to 36-month loans issued in the main sample period for amounts between $ ${\$}$ $16,000 and $ ${\$}$ $24,000; $ $D_{amount1000,t}$ $ is defined as one for loan amounts between $ ${\$}$ $16,000 and $ ${\$}$ $20,000 on and after March 2013, and zero in other cases. Column 4 restricts the sample to 36-month loans issued in the main sample period for amounts between $ ${\$}$ $1,000 and $ ${\$}$ $10,000; $ $D_{amount1000,t}$ $ is defined as one for loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $10,000 in and after July 2013 and zero in other cases. Column 5 reports the tests of a placebo sample that includes loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $20,000 issued between July 2013 and May 2014. Standard errors are robust to heteroscedasticity. *, **, and *** represent significance at the 10%, 5%, and 1% level, respectively. In the Internet Appendix Table A1 we conduct robustness tests where we vary the dimensions along which we collapse the loan-level data to count the number of loans. There we show that the selection result is slightly smaller in magnitude, ranging from 6.3% to 10%, but statistically significant across all specifications when we collapse the data in month of origination$ $\times$ $risk category$ $\times$ $$ ${\$}$ $1,000 loan amount$ $\times$ $ four-point FICO score bins (Column 1 in panel A), month of origination$ $\times$ $risk category$ $\times$ $$ ${\$}$ $100 loan amount $ $\times$ $ four-point FICO score bins (Column 1 in panel B), and month of origination$ $\times$ $ four-point FICO score$ $\times$ $five-point debt-to-income bins (Column 1 in panel C). We emphazise the estimated coefficient of 14.5% shown in Column 1 of Table 2 as our baseline result because it implies the smallest difference in default rates for individuals who choose the short term loan. We note that, qualitatively or quantitatively, none of our results, except the magnitude of the average difference in default rates, depend on this choice. 2.3 Selection and repayment Having shown that the expansion of the menu of borrowing options induced a significant amount of self-selection from short-term to long-term loans, we run our main test to uncover the unobserved quality of the borrowers who selected into the new long-term contract. We estimate the following difference-in-differences specification on the sample of 36-month loans: \begin{equation}Default_{i}=\beta_{i}^{1000bin}+\delta_{i}^{jt}+\gamma\times D_{i}+X_{i}+\epsilon_{i},\label{eq:selection_regression}\end{equation} (2) where data is at the loan level $ $i$ $. The outcome variable, $ $Default_{i}$ $, is defined as a dummy that equals one if the loan is late by more than 120 days measured as of April 2015. Standard errors are clustered at the state level (45 clusters). The main explanatory variable of interest, $ $D_{i}$ $, is a dummy equal to one if the 36-month loan $ $i$ $ is issued at a time when a 60-month loan of the same amount is also available, and zero otherwise: \[D_{i}=\begin{cases}1 & \mathit{if}\ \$16,000>\mathit{Loan}\ \mathit{Amount}_{i}\geq \$12,000\;\&\;t\geq \mathit{March}\ 2013\\1 & \mathit{if}\ \$12,000>\mathit{Loan}\ \mathit{Amount}_{i}\geq \$10,000\;\&\;t\geq \mathit{July}\ 2013\\0 & \,\mathit{otherwise}\end{cases}\] The coefficient of interest, $ $\gamma,$ $ measures the change in the default rate of 36-month loans for eventually selected amounts before and after the expansion of the menu options, relative to the change of the default rate for never-selected and always-selected amounts, which were not affected by the menu expansion. We include granular month of origination $ $t \times {\rm risk}$ $ category $ $j$ $ fixed effects, $ $\delta_{i}^{jt}$ $, which ensure we compare borrowers who took a loan on the same month with the same contract terms and with similar observed measures of credit risk (same risk category). We also include a vector of control variables observable at origination, $ $X_{i}$ $. In our baseline specification, $ $X_{i}$ $ includes four-point FICO score at-origination bin and state fixed effects. The rich set of fixed effects ensures that we perform the difference-in-differences estimation by comparing borrowers who are observationally equivalent. We also report results including as controls the full set of variables that LC reports and that investors observe at origination. These variables (61 in total) include annual income, a dummy for home ownership, stated purpose of the loan, length of employment, length of credit history, total debt balance excluding mortgage, revolving balance, and monthly debt payments to income, among others. Table 3, Columns 1 and 2, reports results of regression (2). The negative point estimate for $ $\gamma$ $ indicates that borrowers who take a 36-month loan once a 60-month option is available are significantly less likely to default than borrowers who take the same 36-month loan when the long-term option is not available. The point estimate of $ $-$ $0.0081 means that the default rate of the borrowers that are selected on maturity is 0.8 percentage points lower than the default rate of the nonselected borrowers (Column 1), and the magnitude is unchanged when we include as additional controls every single variable observable at origination in LC’s data set (Column 2). That our estimate is virtually unaffected by including this full suite of additional controls demonstrates that the granular fixed effects structure in our baseline regression is sufficiently comprehensive to absorb any changes in the composition of observed borrower characteristics. Table 3 Regression results: Screening with maturity (1) (2) $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $D_{i,t}$ $ –0.0081** –0.0080** (0.004) (0.004) Sample MAIN MAIN Observations 60,511 57,263 $ $R^{2}$ $ 0.035 0.047 # clusters 45 45 (1) (2) $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $D_{i,t}$ $ –0.0081** –0.0080** (0.004) (0.004) Sample MAIN MAIN Observations 60,511 57,263 $ $R^{2}$ $ 0.035 0.047 # clusters 45 45 This table shows that the default rate of borrowers who selected into a short-term loan when they have taken take a long-term loan is higher than for borrowers who could not take a long-term loan. The table shows the output of the regression of each outcome on a dummy for the staggered reduction of the minimum amount threshold for long-maturity loans in March 2013 (to $ ${\$}$ $12,000) and July 2013 (to $ ${\$}$ $10,000). The outcome is $ $Default$ $, a dummy that equals one if a borrower is late by more than 120 days, measured as of April 2015. The sample corresponds to loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $20,000 with a list date between December 2012 and October 2013. All regressions include risk category $ $\times$ $month and four-point FICO score bin, state, and $ ${\$}$ $1,000 loan amount bin fixed effects. Column 2 includes all borrower-level variables observed by investors at the time of origination as controls. Standard errors are clustered at the state level. *, **, and *** represent significance at the 10%, 5%, and 1% level, respectively. Table 3 Regression results: Screening with maturity (1) (2) $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $D_{i,t}$ $ –0.0081** –0.0080** (0.004) (0.004) Sample MAIN MAIN Observations 60,511 57,263 $ $R^{2}$ $ 0.035 0.047 # clusters 45 45 (1) (2) $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $D_{i,t}$ $ –0.0081** –0.0080** (0.004) (0.004) Sample MAIN MAIN Observations 60,511 57,263 $ $R^{2}$ $ 0.035 0.047 # clusters 45 45 This table shows that the default rate of borrowers who selected into a short-term loan when they have taken take a long-term loan is higher than for borrowers who could not take a long-term loan. The table shows the output of the regression of each outcome on a dummy for the staggered reduction of the minimum amount threshold for long-maturity loans in March 2013 (to $ ${\$}$ $12,000) and July 2013 (to $ ${\$}$ $10,000). The outcome is $ $Default$ $, a dummy that equals one if a borrower is late by more than 120 days, measured as of April 2015. The sample corresponds to loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $20,000 with a list date between December 2012 and October 2013. All regressions include risk category $ $\times$ $month and four-point FICO score bin, state, and $ ${\$}$ $1,000 loan amount bin fixed effects. Column 2 includes all borrower-level variables observed by investors at the time of origination as controls. Standard errors are clustered at the state level. *, **, and *** represent significance at the 10%, 5%, and 1% level, respectively. This decline in the default probability is due to the borrowers who self-select into the long-term loan, which we estimated to be 14.5% of the borrowers in the nonselected sample (Table 2, Column 1). Combining the two results allows us to obtain an estimate of the default probability of the borrowers that self-selected into the 60-month loan: it is $ $^{0.8\%}/{14.5\%}=5.5\%$ $ higher than for those who self-select into the 36-month loan when the long-term loan is available (significant at a 10% level, based on bootstrapping with 1,000 repetitions). This is an estimate of the counterfactual probability we are after: it is the default rate that borrowers who self-selected into the 60-month loan would have had if they had taken the 36-month loan. The economic magnitude of this difference is large compared with the average default rate of 36-month loans issued before the menu expansion, 9.2% (Table 1). The comparison implies that among observationally equivalent borrowers, those who self-select into a long-maturity contract are 59% more likely to default than those borrowers who self-select into the short-term contract, ceteris paribus (e.g., holding constant the contract characteristics). The results suggest that maturity choice reveals unobserved heterogeneity among borrowers. The lower default rate of borrowers who self-select into a short-maturity loan cannot be predicted by variables available to investors at the time of origination, as attested by the comparison between the estimates with and without controls for observables. Although we do not control for the exact FICO score but for scores within each four-point FICO bin, the predictive power of FICO on default in our sample is too small for selection within four-point FICO bins to account for our results. Indeed, a regression of $ $Default_{i}$ $ on the high end of the FICO four-point range at origination, including risk category by $ ${\$}$ $1,000 loan amount bin by month fixed effects, gives a coefficient of $ $-$ $0.0000362. That is, a one-point increase in FICO score at origination is correlated with a 0.004% decline in default rate, not statistically significant. Thus, variation in default rates within FICO score bins can at most account for a 0.012% difference in default rates (0.004% $ $\times$ $3), quantitatively irrelevant next to our estimated effect of 0.8% reduction in default. 2.4 The APR premium for 60-month loans For maturity to operate as a screening device, it must be that unobservably high-risk borrowers are self-selecting into loans with a higher APR. We estimate the difference in the long- and short-term loan by running a regression of $ $APR$ $ on $ $Long$ $, a dummy that equals one for long-term loans, controlling by credit-risk grade by month by $ ${\$}$ $1,000 loan amount by four-point FICO range fixed effects. As Internet Appendix Table C1 shows, LC charges a 3.3% higher APR for 60-month loans, holding all borrower and loan characteristics constant.22 Importantly, this APR differential cannot be explained by the upward sloping yield curve during our sample period. In Internet Appendix Section B we show that the upward sloping Treasury yield curve on March 1, 2013, would imply an APR premium between the 36- and 60-month loan of 0.2 percentage points. This implies that 3.1% of the 3.3% differential in our sample (94%) cannot be explained by the yield curve. Thus, borrowers who selected into the 60-month loan option were required to pay a premium for the insurance provided by this contract, consistent with any screening mechanism. Further, the APR difference between short- and long-maturity loans can be more than fully accounted for by the 5.5% higher expected default rate of those borrowers who self-select into the long-maturity option.23 2.5 Identification tests 2.5.1 Evidence to support the identifying assumption Our empirical strategy rests on the identifying assumption that there were no changes in unobserved borrower creditworthiness that differentially affected borrowers taking loans between $ ${\$}$ $12,000 and $ ${\$}$ $16,000 in March and borrowers taking loans between $ ${\$}$ $10,000 and $ ${\$}$ $12,000 in July. One potential threat to this assumption is the possibility that there is a gradual shift in the composition of borrowers over 2013 that approximately matched the pattern of the staggered expansion. We test for this possibility by running an amended version of (1) using a series of dummies that become active $ $\tau$ $ months after a 60-month loan is offered at each amount. Formally, we define: \[D\left(\tau\right)_{kt}=\begin{cases}1 & \mathit{if}\ \$16,000>\mathit{Loan}\ \mathit{Amount}\geq \$12,000\;\&\;t= \mathit{March}\ 2013+\tau\\1 & \mathit{if}\ \$12,000>\mathit{Loan}\ \mathit{Amount}\geq \$10,000\;\&\;t= \mathit{July}\ 2013+\tau\\0 & \,\mathit{otherwise}\end{cases},\] and we run the following regression:24 \begin{equation}log(N_{jkt})=\beta_{k}+\delta_{jt}+\sum_{\tau=-3}^{3}\gamma_{\tau}\times D\left(\tau\right)_{kt}+\epsilon_{jkt}.\label{eq:pretrends_regression-1}\end{equation} (3) Figure 6 shows the results of regression 3. The results show no differential pre-trends in the 3 months leading up to the expansion and then show a discontinuous fall in the number of loans made in these amounts exactly at the time of the expansion. This rules out that our results are coming from pre-existing trends in borrower demand or composition unrelated to the menu expansion. Figure 6 View largeDownload slide Pre-trends on number of loans originated This figure shows the regression coefficients ($ $\gamma_{\tau}$ $) and 90% confidence interval for the following regression: \[log(N_{j,t,amount1000})=\beta_{amount1000}+\delta_{j,t}+\sum_{\tau=-3}^{3}\gamma_{\tau}\times D\left(\tau\right)_{amount1000,t}+\epsilon_{i,t},\] which measures the difference in the number of loans issued between eventually selected and control amounts $ $\tau$ $ months after the threshold expansion. Standard errors are robust to heteroscedasticity. Figure 6 View largeDownload slide Pre-trends on number of loans originated This figure shows the regression coefficients ($ $\gamma_{\tau}$ $) and 90% confidence interval for the following regression: \[log(N_{j,t,amount1000})=\beta_{amount1000}+\delta_{j,t}+\sum_{\tau=-3}^{3}\gamma_{\tau}\times D\left(\tau\right)_{amount1000,t}+\epsilon_{i,t},\] which measures the difference in the number of loans issued between eventually selected and control amounts $ $\tau$ $ months after the threshold expansion. Standard errors are robust to heteroscedasticity. To further ensure that our results are not driven by differential trends in the demand for loans of varying amounts, we run regression (1) on a sample shifted forward to start when the 60-month loan option is available for any amount above $ ${\$}$ $10,000 (after the expansion in menus is complete). That is, we shift the definition of $ $D_{kt}$ $ forward by 7 months and run the regression on the sample of loans originated between July 2013 and May 2014. Column 5 of Table 2 shows the results. The coefficient on $ $D_{kt}$ $ equals $ $-$ $4.4% and is insignificant, and given the confidence interval we can reject the null that this coefficient equals our main estimate. 2.5.2 Simultaneous choice of maturity and loan amount A second identifying assumption behind our empirical approach is that the choice of loan amount is sufficiently inelastic to loan maturity. If this is not the case, the difference-in-differences estimate will be biased toward zero. This could be the case either because borrowers in what we classify as eventually selected amounts may be already selected on maturity before the menu expansion (e.g., if some borrowers who wanted to take a long-term loan at a treated amount before the expansion took a long-maturity loan at larger amount instead) or because borrowers in what we classify as never-selected amounts may be a selected group after the menu expansion (e.g., because some borrowers who wanted to take a long-term loan at a control amount after the menu expansion took a long-maturity loan at a treated amount instead). We first consider the possibility that eventually selected amounts are selected before the menu expansion. As an example, consider borrowers who would like to take a $ ${\$}$ $10,000 60-month loan, which is not available before the menu expansion. The closest feasible alternatives are (1) a $ ${\$}$ $10,000 36-month loan and (2) a $ ${\$}$ $16,000 60-month loan.25 Our empirical strategy will estimate the effect of maturity on selection if borrowers choose the first option; for example, they take a loan for the amount they prefer at a shorter maturity—36-months—when the 60-month option is not available. The reason is that these borrowers select out of the 36-month loan when the 60-month option is available, after the menu expansion.26 If, on the contrary, borrowers choose the second option, for example, they take a 60-month maturity loan but for a larger amount, then our difference-in-differences estimate will be zero. Indeed, these borrowers will not be in the eventually selected group of loans before or after the expansion because our estimation is exclusively based on the outcomes of 36-month loans. Thus, selection from one long-term loan to another will not affect our estimates.27 Figure 5 View largeDownload slide Pre-period loan amount histogram This top panel shows the number of 36-month loans issued by LC by loan amount in $ ${\$}$ $25 increments between $ ${\$}$ $5,000 and $ ${\$}$ $25,000 between December 2012 and February 2013. The bottom panel shows the same histogram for the same period of time for 60-month loans. Figure 5 View largeDownload slide Pre-period loan amount histogram This top panel shows the number of 36-month loans issued by LC by loan amount in $ ${\$}$ $25 increments between $ ${\$}$ $5,000 and $ ${\$}$ $25,000 between December 2012 and February 2013. The bottom panel shows the same histogram for the same period of time for 60-month loans. Now consider the second case, where the never-selected amounts are treated after the menu expansion. Take for example borrowers who would like to take a $ ${\$}$ $5,000 60-month loan, but since this option is not available before the menu expansion, they take a $ ${\$}$ $5,000 36-month loan instead. Although these borrowers are in the control group in our estimation, it is possible that they choose a $ ${\$}$ $10,000 60-month loan when this option becomes available in the menu. If this is the case, then the menu expansion will also cause self-selection into long maturity among the control group of loans, and the comparison between eventually selected and control loans will be biased toward zero. We investigate formally whether eventually selected loans amounts were affected prior to the expansion or if control loan amounts were impacted after the expansion. To do this we exploit the same setup as regression 1, which measures the change in the number of short-term loans issued at eventually selected and control amounts, before and after the menu expansion, and compare the evolution of the number of 60-month (36-month) loans in the $ ${\$}$ $16,000 to $ ${\$}$ $20,000 ($ ${\$}$ $5,000 and $ ${\$}$ $10,000) range relative to the evolution of the number of loans in the $ ${\$}$ $20,000 to $ ${\$}$ $24,000 ($ ${\$}$ $1,000 and $ ${\$}$ $5,000) range around the menu expansion. We estimate the same difference-in-differences regressions with a modified definition of the “selected” dummy $ $D_{kt}$ $ to equal 1 one after March 2013 or July 2013, for different loan amounts according to the timing of the menu expansion, as explained below. First, using the subsample of 60-month loans for amounts between $ ${\$}$ $16,000 and $ ${\$}$ $24,000, we define $ $D_{kt}$ $ to be equal to one after March 2013 for all amounts between $ ${\$}$ $16,000 and $ ${\$}$ $20,000. The coefficient on this dummy tells us whether the number of loans of amounts close to the $ ${\$}$ $16,000 expansion threshold declined relative to those further from the threshold. If so, it would be an indication that eventually selected loan amounts experienced selection to 60-month loans prior to the expansion. The coefficient on the interaction term is $ $-$ $8.25% and is not significantly different from zero (Table 2, Column 2). This suggests weak evidence that our estimates may understate the degree of selection because some borrowers in eventually selected amounts may have opted for 60-month loans above $ ${\$}$ $16,000 prior to the expansion.28 We repeat the exercise at the $ ${\$}$ $10,000 amount threshold using 36-month loans. We restrict the analysis to the sample of loan amounts between $ ${\$}$ $1,000 and $ ${\$}$ $10,000, and define $ $D_{kt}$ $ equal to one after July 2013 for amounts between $ ${\$}$ $5,000 and $ ${\$}$ $10,000 and zero otherwise. The coefficient on the interaction term is $ $-$ $3.6% and, again, not significantly different from zero (Column 4). Thus, there is no evidence that borrowers who in the pre-period selected a short-maturity loan below $ ${\$}$ $10,000 would have taken a larger long-maturity loan above the $ ${\$}$ $10,000 threshold when they became available in July. In other words, we find no evidence that the control group of loans in our main empirical design were affected by the menu expansion. Taken together the results in Table 2 confirm our conjecture that the bulk of any selection to longer-maturity loans induced by the expansion of the menu was in the eventually selected amounts.29 2.6 Robustness We present in Table 4 several tests that demonstrate the robustness of our results. First, Column 1 of Table 4 presents a counterpart to our main result in Column 1 of Table 2 but limiting the sample to loan amounts between $ ${\$}$ $6,000 and $ ${\$}$ $19,000 (a $ ${\$}$ $1,000 narrower window than our main sample, which uses loans from $ ${\$}$ $5,000 to $ ${\$}$ $20,000). The results are qualitatively similar, although the estimate is noisier and significant only at a 10% level. Table 4 Robustness (1) (2) (3) $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $D_{i,t}$ $ –0.0066* 0.0018 –0.0106 (0.004) (0.010) (0.007) Sample 6k–19k 36m, 16k–24k 36m, 1k–10k Observations 54,689 14,652 33,493 $ $R^{2}$ $ 0.037 0.061 0.035 # clusters 45 45 46 (1) (2) (3) $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $D_{i,t}$ $ –0.0066* 0.0018 –0.0106 (0.004) (0.010) (0.007) Sample 6k–19k 36m, 16k–24k 36m, 1k–10k Observations 54,689 14,652 33,493 $ $R^{2}$ $ 0.037 0.061 0.035 # clusters 45 45 46 The table shows the output of several robustness tests. Column 1 replicates Column 1 in Table 3 on a sample of loans listed between December 2012 and October 2013 and issued for amounts between $ ${\$}$ $6,000 and $ ${\$}$ $19,000 (an interval $ ${\$}$ $1,000 narrower than that in main sample). Columns 2 and 3 report the output for regressions run on a sample of loans listed between December 2012 and October 2013 for different loan amounts, where the independent variable is defined in an ad hoc manner using $ $\mathit{default}$ $ as outcome. Column 2 restricts the sample to 36-month loans issued in the main sample period for amounts between $ ${\$}$ $16,000 and $ ${\$}$ $24,000, and $ $D_{i,t}$ $ is equal to one for loan amounts between $ ${\$}$ $16,000 and $ ${\$}$ $20,000 listed on or after March 2013, and zero otherwise. Column 3 restricts the sample to 36-month loans issued in the main sample period for amounts between $ ${\$}$ $1,000 and $ ${\$}$ $10,000; $ $D_{i,t}$ $ is equal to one for loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $10,000 listed in or after July 2013, and zero otherwise. All regressions include risk category$ $\times$ $month and four-point FICO score bin, state, and $ ${\$}$ $1,000 loan amount bin fixed effects. Standard errors are clustered at the state level. *, **, and *** represent significance at the 10%, 5%, and 1% level, respectively. Table 4 Robustness (1) (2) (3) $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $D_{i,t}$ $ –0.0066* 0.0018 –0.0106 (0.004) (0.010) (0.007) Sample 6k–19k 36m, 16k–24k 36m, 1k–10k Observations 54,689 14,652 33,493 $ $R^{2}$ $ 0.037 0.061 0.035 # clusters 45 45 46 (1) (2) (3) $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $D_{i,t}$ $ –0.0066* 0.0018 –0.0106 (0.004) (0.010) (0.007) Sample 6k–19k 36m, 16k–24k 36m, 1k–10k Observations 54,689 14,652 33,493 $ $R^{2}$ $ 0.037 0.061 0.035 # clusters 45 45 46 The table shows the output of several robustness tests. Column 1 replicates Column 1 in Table 3 on a sample of loans listed between December 2012 and October 2013 and issued for amounts between $ ${\$}$ $6,000 and $ ${\$}$ $19,000 (an interval $ ${\$}$ $1,000 narrower than that in main sample). Columns 2 and 3 report the output for regressions run on a sample of loans listed between December 2012 and October 2013 for different loan amounts, where the independent variable is defined in an ad hoc manner using $ $\mathit{default}$ $ as outcome. Column 2 restricts the sample to 36-month loans issued in the main sample period for amounts between $ ${\$}$ $16,000 and $ ${\$}$ $24,000, and $ $D_{i,t}$ $ is equal to one for loan amounts between $ ${\$}$ $16,000 and $ ${\$}$ $20,000 listed on or after March 2013, and zero otherwise. Column 3 restricts the sample to 36-month loans issued in the main sample period for amounts between $ ${\$}$ $1,000 and $ ${\$}$ $10,000; $ $D_{i,t}$ $ is equal to one for loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $10,000 listed in or after July 2013, and zero otherwise. All regressions include risk category$ $\times$ $month and four-point FICO score bin, state, and $ ${\$}$ $1,000 loan amount bin fixed effects. Standard errors are clustered at the state level. *, **, and *** represent significance at the 10%, 5%, and 1% level, respectively. As we mention above when describing our empirical strategy, the expansion in the menu of borrowing options may have induced selection in the unaffected or control group of amounts, above and below the $ ${\$}$ $10,000 to $ ${\$}$ $16,000 interval. In Table 2 above we show that the number of loans issued at the control amounts did not change, which suggests that no such selection occurred. However, it is important to independently verify that there is no change in the credit quality of loans issued at control amounts induced by the menu expansion. Here, we test for this possibility. Column 3 of Table 4 restricts the sample to loans issued between December 2012 and October 2013, between $ ${\$}$ $16,000 and $ ${\$}$ $24,000. The independent variable of interest equals one for loans between $ ${\$}$ $16,000 and $ ${\$}$ $20,000 after March 2013. The coefficient is positive and insignificant. Column 4 of Table 4 repeats the exercise for loans between $ ${\$}$ $1,000 and $ ${\$}$ $10,000 issued between December 2012 and October 2013. Here, the independent variable of interest equals one for loans between $ ${\$}$ $5,000 and $ ${\$}$ $10,000 issued after July 2013, and the coefficient is negative and insignificant. Thus, we find no significant differences in the default rate of loans issued at amounts bordering the interval of eventually selected amounts in both cases. The results in Columns 2 and 3 of Table 4 also serve as placebo tests and confirm that our results are not spuriously driven by shifting creditworthiness at different loan amounts. Overall, these tests point to a robust conclusion: borrowers who self-select into long-maturity loans are unobservably more likely to default, holding the loan contract characteristics constant. 3. Interpretation: Private Information About What? 3.1 Time structure of private information So far, our empirical results show that borrowers who select into a longer-maturity loan with a higher rate are privately informed about their increased propensity to default on a short-term loan with a lower rate, and that, therefore, maturity is an effective screening device. We turn to understanding what is the specific private information that borrowers are selecting on. Since maturity provides insurance against future changes in the price of credit, then the private information must relate to how borrowers value this protection. It is theoretically possible that borrowers who are privately informed about their own high degree of risk aversion select into longer maturity loans (De Meza and Webb 2001). If more risk averse individuals also default less (e.g., because they endogenously select less risky income streams), selection on risk aversion is inconsistent with the higher default rate that these borrowers exhibit. It follows that borrowers who select a long-maturity loan privately place a higher value on the insurance it provides either because (1) they are less likely to have sufficient funds to make loan repayments in the future (e.g., they face a higher the probability of job loss or illness or have less discretionary income), or (2) they have a higher exposure to rollover risk due to privately observed differences in the timing of their income (the cash flow timing hypothesis). These two explanations differ in the horizon after origination at which borrowers become risky. Borrowers who self-select into long maturity because they are more exposed to shocks to their observable ability to repay will tend to default more the longer the horizon after origination, as the negative shocks are realized. In contrast, borrowers who self-select into long maturity because their income arrives later will tend to default less with time after origination, as their income is realized. Therefore, the two selection mechanisms can be distinguished by their predictions of the time structure of default implied by the private information. We exploit the fact that we observe when a borrower in our sample enters default to differentiate between these two accounts. To do this, we redefine our baseline measure of default and create two variables for default at different horizons: borrowers who missed their first payment within the first 12 and 24 months of loan origination (for loans that are 120 days past due in April 2015). We label these variables $ $Default12m$ $ and $ $Default24m$ $, respectively, and use them as dependent variables in regressions that are otherwise identical to the one we estimated in Column 1 of Table 3. The results are presented in Columns 1 and 2 of Table 5. Column 1 shows that borrowers who self-select into long-term loans have no differential propensity to default within the first year of the loan. Since the hazard rate of default in our sample peaks at 13 months (Figure 2), this result is not mechanically driven by lack of statistical power due to a low frequency of default early in the life of the average loan (unconditionally, loans are as likely to default in the first 12 months after origination than later). Column 2 shows that the differential propensity to default is present at the 24-month horizon from origination. Table 5 Interpretation of results (1) (2) (3) (4) (5) (6) $ $\mathit{Default}12m$ $ $ $\mathit{Default}24m$ $ $ $\mathit{FICO}$ $ $ $\mathit{FICO}$ $ $ $\mathit{SD(FICO)}$ $ $ $\mathit{Prepayment}$ $ $ $D_{i,t}$ $ –0.0039 –0.0082* 2.7464** 2.6705** –0.5764** –0.0063 (0.003) (0.004) (1.032) (0.999) (0.266) (0.008) Sample Main Main Main Main Main Main Obs 60,511 60,511 60,511 57,263 60,511 60,511 $ $R^{2}$ $ 0.024 0.032 0.192 0.215 0.027 0.023 Clusters 45 45 45 45 45 45 (1) (2) (3) (4) (5) (6) $ $\mathit{Default}12m$ $ $ $\mathit{Default}24m$ $ $ $\mathit{FICO}$ $ $ $\mathit{FICO}$ $ $ $\mathit{SD(FICO)}$ $ $ $\mathit{Prepayment}$ $ $ $D_{i,t}$ $ –0.0039 –0.0082* 2.7464** 2.6705** –0.5764** –0.0063 (0.003) (0.004) (1.032) (0.999) (0.266) (0.008) Sample Main Main Main Main Main Main Obs 60,511 60,511 60,511 57,263 60,511 60,511 $ $R^{2}$ $ 0.024 0.032 0.192 0.215 0.027 0.023 Clusters 45 45 45 45 45 45 This table shows the output of the regression of each outcome on a dummy for the staggered reduction of the minimum amount threshold for long-maturity loans in March 2013 (to $ ${\$}$ $12,000) and July 2013 (to $ ${\$}$ $10,000). Outcomes include $ $\mathit{Default}12m$ $ and $ $\mathit{Default}24m$ $, dummies that equal one if a borrower is late by more than 120 days as of April 2015 and whose last payment occurred within 12 and 24 months after origination, respectively. FICO is the (the high end of the four-point bin) FICO score measured as of April 2015; SD(FICO) is the time-series standard deviation of (the high end of the four-point bin) FICO scores within an individual, using four observations per individual at origination, as of August 2014, December 2014, and April 2015; and $ $\mathit{Prepayment}$ $, a dummy that equals one if the loan is fully paid as of April 2015. In Column 4 we include all variables that are observable by investors at origination as controls, like in Column 2 of Table 3. The sample corresponds to loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $20,000 whose listing date is between December 2012 and October 2013. All regressions include risk category$ $\times$ $month and four-point FICO score bin, state, and $ ${\$}$ $1,000 loan amount bin fixed effects. Standard errors are clustered at the state level. *, **, and *** represent significance at the 10%, 5%, and 1% level, respectively. Table 5 Interpretation of results (1) (2) (3) (4) (5) (6) $ $\mathit{Default}12m$ $ $ $\mathit{Default}24m$ $ $ $\mathit{FICO}$ $ $ $\mathit{FICO}$ $ $ $\mathit{SD(FICO)}$ $ $ $\mathit{Prepayment}$ $ $ $D_{i,t}$ $ –0.0039 –0.0082* 2.7464** 2.6705** –0.5764** –0.0063 (0.003) (0.004) (1.032) (0.999) (0.266) (0.008) Sample Main Main Main Main Main Main Obs 60,511 60,511 60,511 57,263 60,511 60,511 $ $R^{2}$ $ 0.024 0.032 0.192 0.215 0.027 0.023 Clusters 45 45 45 45 45 45 (1) (2) (3) (4) (5) (6) $ $\mathit{Default}12m$ $ $ $\mathit{Default}24m$ $ $ $\mathit{FICO}$ $ $ $\mathit{FICO}$ $ $ $\mathit{SD(FICO)}$ $ $ $\mathit{Prepayment}$ $ $ $D_{i,t}$ $ –0.0039 –0.0082* 2.7464** 2.6705** –0.5764** –0.0063 (0.003) (0.004) (1.032) (0.999) (0.266) (0.008) Sample Main Main Main Main Main Main Obs 60,511 60,511 60,511 57,263 60,511 60,511 $ $R^{2}$ $ 0.024 0.032 0.192 0.215 0.027 0.023 Clusters 45 45 45 45 45 45 This table shows the output of the regression of each outcome on a dummy for the staggered reduction of the minimum amount threshold for long-maturity loans in March 2013 (to $ ${\$}$ $12,000) and July 2013 (to $ ${\$}$ $10,000). Outcomes include $ $\mathit{Default}12m$ $ and $ $\mathit{Default}24m$ $, dummies that equal one if a borrower is late by more than 120 days as of April 2015 and whose last payment occurred within 12 and 24 months after origination, respectively. FICO is the (the high end of the four-point bin) FICO score measured as of April 2015; SD(FICO) is the time-series standard deviation of (the high end of the four-point bin) FICO scores within an individual, using four observations per individual at origination, as of August 2014, December 2014, and April 2015; and $ $\mathit{Prepayment}$ $, a dummy that equals one if the loan is fully paid as of April 2015. In Column 4 we include all variables that are observable by investors at origination as controls, like in Column 2 of Table 3. The sample corresponds to loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $20,000 whose listing date is between December 2012 and October 2013. All regressions include risk category$ $\times$ $month and four-point FICO score bin, state, and $ ${\$}$ $1,000 loan amount bin fixed effects. Standard errors are clustered at the state level. *, **, and *** represent significance at the 10%, 5%, and 1% level, respectively. We present in Figure 7 the coefficients from estimating our main specification using as the dependent variable an indicator for whether the first missed payment occurred before 1, 2, and so on, up to 24 months after origination.30 The figure indicates that the cumulative default probability differential between the two groups of borrowers increases linearly with the months after origination. That is, borrowers who select into the 60-month loan have a propensity to default on the 36-month loan that is increasing in the time since origination of their loan.31 This evidence indicates that the source of private information that is driving maturity selection is borrowers’ exposure to shocks to their own future observable creditworthiness. Note that Figure 2 demonstrates that the hazard rate of default for 36-month loans peaks at 16 months.32 This indicates that the bulk of default at either maturity occurs well before the 24-month horizon possible in our analysis, thereby ruling out the concern that our results are too near to origination to account for default behavior for either type of loan. Figure 7 View largeDownload slide Default rate coefficient by number of months since origination This figure shows the estimated coefficient and 90% confidence interval for the following regression: \[default\left(\Delta t\right)=\beta_{amount1000}+\delta_{j,\overline{FICO},t}+\gamma\times D_{amount100,t}+X_{i,t}+\epsilon_{i},\] where the outcome is $ $default\left(\Delta t\right)$ $, a dummy that equals one if a loan is late by more than 120 days as of April 2015 and if the last payment on these loan occurred $ $\Delta t$ $ months after origination, on $ $D_{amount1000,t}$ $, a dummy that captures the staggered expansion of the 60-month loans for amounts above $ ${\$}$ $12,000 and $ ${\$}$ $10,000 in March and July 2013, respectively. Standard errors are clustered at the state level. Sample includes loans issued between December 2012 and October 2013 for loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $20,000. Figure 7 View largeDownload slide Default rate coefficient by number of months since origination This figure shows the estimated coefficient and 90% confidence interval for the following regression: \[default\left(\Delta t\right)=\beta_{amount1000}+\delta_{j,\overline{FICO},t}+\gamma\times D_{amount100,t}+X_{i,t}+\epsilon_{i},\] where the outcome is $ $default\left(\Delta t\right)$ $, a dummy that equals one if a loan is late by more than 120 days as of April 2015 and if the last payment on these loan occurred $ $\Delta t$ $ months after origination, on $ $D_{amount1000,t}$ $, a dummy that captures the staggered expansion of the 60-month loans for amounts above $ ${\$}$ $12,000 and $ ${\$}$ $10,000 in March and July 2013, respectively. Standard errors are clustered at the state level. Sample includes loans issued between December 2012 and October 2013 for loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $20,000. 3.2 Private information about future observable creditworthiness We provide additional evidence in support of our preferred interpretation. Specifically, we observe the realized observable creditworthiness measured by each borrower’s credit score (FICO score) as of April 2015, roughly 2 years after origination. In Table 5, Columns 3 and 4, we run our main regression model but replace the main outcome $ $Default_{i}$ $ with $ $FICO_{i}$ $, the borrower’s FICO score as reported in the latest LC data pull. In Column 4 we include as controls all variables that are observable by investors at origination, like in Column 2 of Table 3. The results imply a statistically significant increase of future FICO scores of approximately 2.7 points among selected short-term borrowers relative to unselected ones. In economic terms this means that the average future FICO score of the 14.5% of borrowers who self-select into the long-maturity loans is $ $^{2.7}/{14.5\%}=18.6$ $ points lower than for the average borrower who selects the 36-month loan. To further demonstrate that borrowers are selecting maturity based on private information about their exposure to shocks to their future observable creditworthiness, we use the volatility of a borrower’s future credit rating as a measure of the reclassification risk she is exposed to. If borrowers have private information about this reclassification risk, we expect borrowers who self-select into the 36-month loan to have less volatile future FICO scores. To test this hypothesis we present in Column 5 of Table 5 the results of estimating our main specification using the within-individual standard deviation of the FICO score as the outcome variable, using FICO scores obtained from 4 different pulls of the LC loan performance data: at origination, as of August 2014, as of December 2014, and as of April 2015, which is the same outcome variable used in Table 2.33 The cross-sectional average and standard deviation of this measure for loans in our sample that were issued in the three pre-period months are 24.5 and 19.1, respectively. The point estimate in Column 5 of Table 5 is $ $-$ $0.57 and statistically significant at the 5% level. This implies that borrowers who select the 36-month loan have a future FICO score that is 2.3% (equal to 0.57/24.5) less volatile when the 60-month loan is available than when it is not. This pattern is strongly consistent with the insurance rationale for the screening mechanism: borrowers who select long-maturity loans are (unobservably) more exposed to reclassification risk. Note that $ $FICO_{i}$ $ measures borrowers’ repayment status on all of their debts. In particular, it considers a borrower’s performance not only on the 36-month loan with LC, but on loans of different maturities as well. Thus it is unlikely that this result is driven by the incompatibility between the short-term LC loan and the time profile of borrower’s future income. Instead, this directly shows that borrowers who select long-maturity loans have private information that directly relates to shocks to their observed creditworthiness and the impact that this will have on the price at which future lending will occur. Finally, we study how maturity choice relates to a borrower’s unobserved propensity to prepay her loan prior to maturity. The LC data record loans that have been fully prepaid as of April 2015, which we code in $ $Prepayment$ $, a dummy variable. If borrowers select maturity based on private information about the timing of their income, we would expect that those borrowers who select into a short-term loan would prepay at a higher rate than borrowers in an unselected group. If this were the case, the main coefficient in regression model (2) where we replace the outcome variable $ $Default$ $ with $ $Prepayment$ $ should be positive. We document the output of this regression in Column 6 of Table 5. The point estimate is negative but insignificant (p-value is .44). Although this result is not conclusive, it does suggest that maturity choice does not seem to be driven by private information about the timing of borrowers’ income shocks. It is difficult to believe that selection based on private information about the timing of income would simultaneously generate a statistically significant reduction in default but would produce a change in loan prepayment that is statistically undetectable, when the prediction about the timing of payment is most directly tied to the hypothesis itself. On the other hand, this finding is fully consistent with the interpretation that borrowers who are privately informed of their increased exposure to shocks to their ability to repay select into long-maturity loans: positive realized shocks lead to early prepayment, while negative shocks lead to default. 3.3 60-month loan performance Further evidence about the underlying private information that is driving maturity choice can be provided by looking at the default rate of borrowers who took 60-month loans. If, as we hypothesize, these borrowers are more exposed to shocks to their ability to repay then, after controlling for observables, the default rate should be higher at the longer-maturity loans. In contrast, if borrowers are selecting to match the privately observed horizon of their income, then the default rate should be no higher. Before presenting this evidence, an important caveat that stems from our core empirical challenge is required. Our analysis has so far focused on the propensity to default holding the terms of the contract constant, that is, focusing exclusively on a sample of 36-month loans. Thus, our analysis tells us what the default probability of borrowers who self-select into 60-month loans would have been had they selected a 36-month loan. We cannot empirically identify what their default probability is for a a 60-month loan. This is because the default rate of 60-month loans is also driven by selection in the extensive margin: there are some borrowers who would have chosen not to take a loan at all in the absence of a 60-month option, but do so when it becomes available, and we cannot independently isolate the repayment propensity of these extensive margin borrowers. Notwithstanding this problem, we can provide suggestive evidence by comparing the average default rate of 36-month and 60-month loans that have the same measured expected default risk (initial risk category and four-point FICO score bin), are issued the same month, and are the same size ($ ${\$}$ $1,000 loan amount bin). The propensity to enter default by April 2015, which holds the repayment horizon equal across the two loan contracts, is 3% higher for the 60-month than for the 36-month loans. This is commensurate with the 3.1% APR risk premium for the 60-month loan that we documented in Section 2.4. This provides further evidence that selection is based on private information about exposure to shocks to creditworthiness. If, alternatively, borrowers were selecting maturity based on the time horizon of their income rather than their future creditworthiness, then we should not expect to see higher default or interest rates for the longer-maturity loan. A different but related question is whether increased maturity impacts a borrower’s propensity to repay a loan. The answer also hinges on the average ability to repay of borrowers who select 60-month loans on the extensive margin, which we cannot measure in our setting. If we make the stark assumption that their ability to repay is the same as for borrowers who are selected away from the 36-month loan, then our results suggest that 2 more years of maturity reduces the propensity to default by 2.5% over the horizon for which we observe these loans.34 If borrowers who take the 60-month loan on the extensive margin have a lower (higher) ability to repay, then this will under- (over-) state the effect. This unmeasured margin could reconcile our results with Dobbie and Song (2017), who use a randomized experiment on U.S. household credit card borrowers to show that increased maturity does not causally change a borrower’s propensity to default, or with Field et al. (2013), who find that increased maturity induces entrepreneurs to undertake risky projects and leads to higher default. 3.4 Price reaction to screening Our empirical analysis benefits from the natural experiment created by LC’s decision to expand the availability of long-term loan contracts without changing any of the characteristics or terms of the short-maturity contract. This implies that, within the window of the natural experiment, the default probability of 36-month loans between $ ${\$}$ $10,000 and $ ${\$}$ $16,000 dropped while the interest rate did not change. If LC was earning a competitive return on these loans before the menu expansion, then it must have been earning rents after the expansion. In theory, competitive pressures should eventually drive the interest rate on the short-maturity loan down to reflect the lower risk of the borrowers that self-select into short maturity. Indeed, after our analysis sample period (during which all lending terms were held constant), LC adjusted the APR of the 36-month loan in a way that is consistent with this conjecture. We show this in Figure 8 which plots the average APR charged to borrowers on 36-month loans in each month controlling for loan amount and borrower characteristics.35 Consistent with our conjecture, we see that the APR fell by roughly 0.8% for short-term loans after long-term loans were added to the menu. This number is of the same order of magnitude as our estimate in Column 1 of Table 2 that showed that the expected default rate of the 36-month loans fell by 0.8% as a result of the selection into long-maturity loans. Figure 8 View largeDownload slide Reduction in APR This figure shows the time series of the residual of a regression of loan APR on $ ${\$}$ $1,000 loan amount dummies, FICO score bin dummies, annual income, and address state dummies by month of origination for 36-month loans issued between $ ${\$}$ $10,000 and $ ${\$}$ $16,000. Figure 8 View largeDownload slide Reduction in APR This figure shows the time series of the residual of a regression of loan APR on $ ${\$}$ $1,000 loan amount dummies, FICO score bin dummies, annual income, and address state dummies by month of origination for 36-month loans issued between $ ${\$}$ $10,000 and $ ${\$}$ $16,000. 4. Conclusion We document that loan terms, in particular maturity, can be used to screen borrowers based on unobserved creditworthiness in U.S. consumer credit markets. Borrowers who are unobservably more exposed to shocks to their ability to repay self-select into longer maturity loans with higher APRs. Extrapolating from the results in our paper may help understand the broader unsecured consumer credit market that platforms like LC and Prosper operate in. Relative to their main competition – credit card debt– these platforms offer loans at significantly shorter maturities, allowing them, through the mechanism we document in this paper, to skim off low-risk borrowers from the credit card market. This may explain how these platforms offer investors competitive returns while offering APRs to borrowers below that offered on credit-card debt. As these options grow in size, it is possible that they will eventually impact the credit card market, which will be left with an increasingly screened pool of high-risk borrowers. It also remains an open question, from both an empirical and a theoretical perspective, whether screening maturity is also a first-order determinant of equilibrium loan prices in consumer credit markets in which lenders may screen using other dimensions of the contract, such as collateral in mortgage markets. Providing concrete evidence of these broader implications is left for future work. We thank Sumit Agarwal, Asaf Bernstein, Emily Breza, Tony Cookson, Anthony DeFusco, Theresa Kuchler, Adair Morse, Holger Mueller, Christopher Palmer, Mitchell Petersen, Philipp Schnabl, Antoinette Schoar, Amit Seru, Felipe Severino, and Johannes Stroebel and numerous seminar and conference participants for helpful comments. We thank Siddharth Vij for outstanding research assistance. A previous version of this paper was circulated under the title “Adverse Selection on Maturity: Evidence from Online Consumer Credit.” Supplementary data can be found on The Review of Financial Studies web site. Footnotes 1 Examples of other contractual terms that have been shown in theory to have a screening role are inside ownership (Leland and Pyle 1977), managerial incentives and capital structure (Ross 1977), mortgage points (Stanton and Wallace (1998)), and prepayment penalties (Bian and Yavas 2013). 2 For the relationship between observable creditworthiness and maturity, see Barclay and Smith (1995), Guedes and Opler (1996), and Johnson (2003), and, for that between observable creditworthiness and collateral choice, see Leeth and Scott (1989), Berger and Udell (1990), Booth (1992), Degryse and Van Cayseele (2000), and Jimenez et al. (2006). For the relationship between observable proxies for the degree of private information and maturity, see Berger et al. (2005), and, for that between observable proxies for the degree of private information and collateral choice, see Berger and Udell (1995) and Berger et al. (2011). 3 For examples of this approach, see Goyal and Wang (2013) and Gopalan et al. (2014). Since contract terms are an endogenous choice (either by the borrower or the lender), controlling for contract characteristics in a regression estimation heavily relies on functional form assumptions and is likely to yield biased estimates due to reverse causality (for an example of this approach, see Kawai et al. (2014)). 4Karlan and Zinman (2009) make a parallel argument for the identification conditions to isolate empirically adverse selection on loan prices. 5 Following the logic in Rothschild and Stiglitz (1976), when borrowers have private information about the value they place on this insurance, the market for loan maturity may not be characterized by a single price at which borrowers can buy all the insurance—maturity—they require. 6 That the higher default rate and APR at the long-maturity option are lower than the 5.5 percentage points suggests that the causal effect of longer maturity is to reduce the probability of default. 7 For comparison, the monthly installments on a $ ${\$}$ $10,000 5-year 10% APR LC loan would be $ ${\$}$ $210, whereas the minimum repayment per month in a credit card with the same balance and APR would be $ ${\$}$ $93. If the credit card APR were 20%, the minimum monthly payments would be $ ${\$}$ $157, which is still lower than the monthly installments on the LC loan. 8 Figures reported in the firms’ 2014 10K reports. 9 See http://kb.lendingclub.com/borrower/articles/Borrower/What-if-my-loan-isn-t-fully-funded-when-my-listing-ends/?l=en_US &fs=RelatedArticle. 10 Figures reported in the firms’ 2014 10K reports. 11 See http://kb.lendingclub.com/borrower/articles/Borrower/What-if-my-loan-isn-t-fully-funded-when-my-listing-ends/?l=en_US &fs=RelatedArticle. 12 This allows us to estimate a measure of time-series volatility in FICO scores for each individual. 13 The exact dates correspond to loans listed as of December 4, 2012, and October 25, 2013. Even though we refer to months as the borders of the interval, all our analyses consider these two dates as the starting and end points of the sample period, respectively. We verify empirically that the interest rates of all risk category-loan quantities pairs are unchanged over this period. For example, Figure E2 in the Internet Appendix shows supply schedules (rate versus amount) before and after the expansion of the menu of borrowing options for borrowers assigned to risk categories B1 through B5: the graphs are identical. We establish the same point in general in a tractable way in Internet Appendix E by regressing the interest rate of all 36-month loans in our sample on fixed effects for loan amount by risk category. The regression yields an $ $R^{2}$ $ of 99.7%, which confirms that the pricing of each menu was constant throughout the sample period for all 25 risk categories. 14 We exclude loans with the “policy code” variable equals 2, because they have no publicly available information, and, according to the LC Data Dictionary, they are “new products not publicly available.” In robustness tests, we limit the sample to loan amounts between $ ${\$}$ $6,000 and $ ${\$}$ $19,000, a $ ${\$}$ $1,000 narrower interval. Also, in some placebo tests we shift our sample to loans issued between July 2013 and May 2014. 15 See Internet Appendix D for details on this reverse-engineering procedure. The error in matching loans to their subgrade does not vary systematically over the same period or by loan amount. 16 We also define borrowers as being in default if they are reported to be on a “payment plan.” Our results are robust to not considering these borrowers as in default. 17 The date of default is determined by the last payment date, a variable available in the LC data. 18 Note that because of the upfront origination fee, borrowers who took short-maturity loans prior to the expansion could not costlessly swap them for long maturity ones after the expansion. This ensures that the pool of borrowers who select the short-maturity loan prior to the expansion is not changed ex post by the expansion itself. 19 Indeed, we found no evidence of any change in outside borrowing options or advertising campaigns at all. 20 For evidence of search frictions in consumer credit markets, see Stango and Zinman (2015). 21 Standard errors for estimates of equation (1) are robust to heteroscedasticity, but other alternatives, for example, clustering in any dimension, are irrelevant in terms of statistical significance. For example, when clustering at the risk category level (25 clusters), the standard error of the coefficient $ $\gamma'$ $ in Column 1 of Table 2 is 0.028. 22 The distribution of the long-short spread is shown in the Internet Appendix Figure C1, where we plot the median long- and short-term APRs for loans issued between $ ${\$}$ $15,000 and $ ${\$}$ $20,000 in the post-period by subgrade (top panel) and initial four-point FICO range (bottom panel). 23 As we discuss below, the fact that the APR gap is less than 5.5% suggests that longer maturity may causally lower the expected default rate on a loan. 24 The final 60-month threshold reduction takes place in July 2013. This leaves 3 more months in our sample period up to October 2013. Similarly, the first 60-month threshold reduction occurs in March 2013. This leaves 3 months in the pre-period (from December 2012). 25 These borrowers may also choose not to borrow at all when their preferred option is not available in the menu, and take the 60-month loan when it becomes available. This extensive margin will not affect our estimates, since our results are exclusively based on the behavior of 36-month loans before and after the menu expansion. 26 One way to test for whether borrowers at control amounts are selected before the menu expansion is to look for evidence of bunching at the borders of the treated interval. The top panel in Figure 5 presents the pre-period loan amount histogram at the short maturity. The histogram suggests that borrowers choose “round” numbers like $ ${\$}$ $10,000 and $ ${\$}$ $12,000 much more frequently than other intermediate amounts. In turn, this makes it very difficult to find evidence of bunching at specific amounts. 27 The bottom panel in Figure 5 presents the pre-period loan amount histogram at the long maturity. The histogram has the same pattern as the top panel. Evidence of bunching is, again, very difficult to establish because of borrowers’ preference for round numbers. 28 In an analogous test we also check whether borrowers of 36-month loans in control amounts above the $ ${\$}$ $16,000 threshold were affected by the expansion. The coefficient on the interaction term is 5.9% and is not significantly different from zero (Table 2, Column 3), indicating that the expansion of the menu did not induce selection away from short-term loans above $ ${\$}$ $16,000. Given that long-maturity loans were always available for these amounts, this is not a surprising result. 29 In the Internet Appendix Table A1 we conduct robustness tests that mimic the results in Table 2 where we vary the dimensions along which we collapse the loan-level data and count the number of loans. These robustness tests consistently show that selection along the margins of the interval of treated amounts is not significantly different from zero, aside from one case in which it is significant at the 10% level. 30 At horizons of 19 months and longer, the sample used to run the regression is right censored because loans issued late in our sample do not have sufficient time to enter default at these horizons. This affects loans in the eventually selected and control amounts in the same way and does not affect the identification strategy. 31 The finding that information asymmetries grow with the horizon from origination is itself new and potentially important in its own right. For example, this supports the assumed time structure of information asymmetry in Milbradt and Oehmke (2014). 32 The hazard rate of default on 60-month loans issued at the same time is similarly shaped and peaks at 17 months. This indicates that repayment over the first 24 months of a loan is the crucial determinant of default at either maturity. 33 The standard deviation is calculated as $ $SD(FICO_{i})=\sqrt{\frac{1}{4}\times\sum_{t=1}^{4}\left(FICO_{i,t}-\overline{FICO_{i}}\right)^{2}}.$ $ 34 We obtain this number as 5.5%-3%=2.5%, where 5.5% is the default probability on the short-term loan for the 14.5% of borrowers who chose the long-term loan when it became available in the menu expansion, measured in Section 2, and 3% is the average excess default rate of long-term loans relative to short-term loans. 35 These characteristics are FICO score bin, annual income, and state of residence. Note that variation in APR before November 2013 in this graph is entirely accounted for by the fact that we do not control for the borrower’s initial risk category, which we cannot estimate after October 2013. This also implies that we are unable to simply compare the APR for the 36-month loan at each menu. References Adams, W., Einav, L. and Levin. J. 2009 . Liquidity constraints and imperfect information in subprime lending. American Economic Review 99 : 49 – 84 . Google Scholar CrossRef Search ADS Agarwal, S., Chomsisengphet, S. Mahoney, N. and Strobel. J. 2015 . Regulating consumer financial products: Evidence from credit cards. Quarterly Journal of Economics 130 : 111 – 64 . Google Scholar CrossRef Search ADS Baker, M., Greenwood, R. and Wurgler. J. 2003 . The maturity of debt issues and predictable variation in bond returns. Journal of Financial Economics 70 : 261 – 91 . Google Scholar CrossRef Search ADS Barclay, M. J., and Smith. C. W. 1995 . The maturity structure of corporate debt. Journal of Finance 50 : 609 – 31 . Google Scholar CrossRef Search ADS Berger, A. N., Espinosa-Vega, M. A. Frame, W. S. and Miller. N. H. 2005 . Debt maturity, risk, and asymmetric information. Journal of Finance 60 : 2895 – 923 . Google Scholar CrossRef Search ADS Berger, A. N., Espinosa-Vega, M. A. Frame, W. S. and Miller. N. H. 2011 . Why do borrowers pledge collateral? New empirical evidence on the role of asymmetric information. Journal of Financial Intermediation 20 : 55 – 70 . Google Scholar CrossRef Search ADS Berger, A. N., and Udell. G. F. 1990 . Collateral, loan quality and bank risk. Journal of Monetary Economics 25 : 21 – 42 . Google Scholar CrossRef Search ADS Berger, A. N., and Udell. G. F. 1995 . Relationship lending and lines of credit in small firm finance. Journal of Business 68 : 351 – 81 . Google Scholar CrossRef Search ADS Bester, H. 1985 . Screening vs. rationing in credit markets with imperfect information. American Economic Review 75 : 850 – 55 . Bian, X., and Yavas. A. 2013 . Prepayment penalty as a screening mechanism for default and prepayment risks. Real Estate Economics 41 : 193 – 224 . Google Scholar CrossRef Search ADS Booth, J. R. 1992 . Contract costs, bank loans, and the cross-monitoring hypothesis. Journal of Financial Economics 31 : 25 – 41 . Google Scholar CrossRef Search ADS Bosworth, B. 1971 . Patterns of corporate external financing. Brookings Papers on Economic Activity 1971 : 253 – 84 . Google Scholar CrossRef Search ADS Cochrane, J. H. 1995 . Time-consistent health insurance. Journal of Political Economy 103 : 445 – 73 . Google Scholar CrossRef Search ADS De Meza, D., and Webb. D. 2001 . Advantageous selection in insurance markets. RAND Journal of Economics 32 : 249 – 62 . Google Scholar CrossRef Search ADS Degryse, H., and Van Cayseele. P. 2000 . Relationship lending within a bank-based system: Evidence from european small business data. Journal of Financial Intermediation 9 : 90 – 109 . Google Scholar CrossRef Search ADS Diamond, D. W. 1991 . Debt maturity structure and liquidity risk. Quarterly Journal of Economics 106 : 709 – 37 . Google Scholar CrossRef Search ADS Dobbie, W., and Skiba. P. M. 2013 . Information asymmetries in consumer credit markets: Evidence from payday lending. American Economic Journal: Applied Economics 5 : 256 – 82 . Google Scholar CrossRef Search ADS Dobbie, W., and Song. J. 2017 . Targeted debt relief and the origins of financial distress: Experimental evidence from distressed credit card borrowers. Working Paper , NBER . Field, E., Pande, R. Papp, J. and Rigol. N. 2013 . Does the classic microfinance model discourage entrepreneurship among the poor? Experimental evidence from India. American Economic Review 103 : 2196 – 226 . Google Scholar CrossRef Search ADS Finkelstein, A., McGarry, K. and Sufi. A. 2005 . Dynamic inefficiencies in insurance markets: Evidence from long-term care insurance. American Economic Review 95 : 224 – 8 . Google Scholar CrossRef Search ADS PubMed Flannery, M. J. 1986 . Asymmetric information and risky debt maturity choice. Journal of Finance 41 : 19 – 37 . Google Scholar CrossRef Search ADS Gopalan, R., Song, F. and Yerramilli. V. 2014 . Debt maturity structure and credit quality. Journal of Financial and Quantitative Analysis 49 : 817 – 42 . Google Scholar CrossRef Search ADS Goyal, V. K., and Wang. W. 2013 . Debt maturity and asymmetric information: Evidence from default risk changes. Journal of Financial and Quantitative Analysis 48 : 789 – 817 . Google Scholar CrossRef Search ADS Graham, J. R., and Harvey. C. R. 2001 . The theory and practice of corporate finance: Evidence from the field. Journal of Financial Economics 60 : 187 – 243 . Google Scholar CrossRef Search ADS Greenwood, R., Hanson, S. and Stein. J. C. 2010 . A gap-filling theory of corporate debt maturity choice. Journal of Finance 65 : 993 – 1028 . Google Scholar CrossRef Search ADS Guedes, J., and Opler. T. 1996 . The determinants of the maturity of corporate debt issues. Journal of Finance 51 : 1809 – 33 . Google Scholar CrossRef Search ADS Hart, O., and Moore. J. 1994 . A theory of debt based on the inalienability of human capital. Quarterly Journal of Economics 109 : 841 – 79 . Google Scholar CrossRef Search ADS Hart, O., and Moore. J. 1995 . Debt and seniority: An analysis of the role of hard claims in constraining management. American Economic Review 85 : 567 – 85 . Holmstrom, B. 1983 . Equilibrium long-term labor contracts. Quarterly Journal of Economics 98 : 23 – 54 . Google Scholar CrossRef Search ADS Jaffee, D. M., and Russell. T. 1976 . Imperfect information, uncertainty, and credit rationing. Quarterly Journal of Economics 90 : 651 – 66 . Google Scholar CrossRef Search ADS Jimenez, G., Salas, V. and Saurina. J. 2006 . Determinants of collateral. Journal of Financial Economics 81 : 255 – 81 . Google Scholar CrossRef Search ADS Johnson, S. A. 2003 . Debt maturity and the effects of growth opportunities and liquidity risk on leverage. Review of Financial Studies 16 : 209 – 36 . Google Scholar CrossRef Search ADS Karlan, D., and Zinman. J. 2009 . Observing unobservables: Identifying information asymmetries with a consumer credit field experiment. Econometrica 77 : 1993 – 2008 . Google Scholar CrossRef Search ADS Kawai, K., Onishi, K. and Uetake. K. 2014 . Signaling in online credit markets. Working Paper . Leeth, J. D., and Scott. J. A. 1989 . The incidence of secured debt: evidence from the small business community. Journal of Financial and Quantitative Analysis 24 : 379 – 94 . Google Scholar CrossRef Search ADS Leland, H. E., and Pyle. D. H. 1977 . Informational asymmetries, financial structure, and financial intermediation. Journal of Finance 32 : 371 – 87 . Google Scholar CrossRef Search ADS Levine, C. B., and Hughes. J. S. 2005 . Management compensation and earnings-based covenants as signaling devices in credit markets. Journal of Corporate Finance 11 : 832 – 50 . Google Scholar CrossRef Search ADS Milbradt, K., and Oehmke. M. 2014 . Maturity rationing and collective short-termism. Journal of Financial Economics 118 : 553 – 70 . Google Scholar CrossRef Search ADS Morse, A. 2015 . Peer-to-peer crowdfunding: Information and the potential for disruption in consumer lending. Annual Review of Financial Economics 7 : 463 – 82 . Google Scholar CrossRef Search ADS Myers, S. C. 1977 . Determinants of corporate borrowing. Journal of Financial Economics 5 : 147 – 75 . Google Scholar CrossRef Search ADS Ross, S. A. 1977 . The determination of financial structure: The incentive-signalling approach. Bell Journal of Economics 8 (1) : 23 – 40 . Google Scholar CrossRef Search ADS Rothschild, M., and Stiglitz. J. E. 1976 . Equilibrium in competitive insurance markets: An essay on the economics of imperfect information. The Quarterly Journal of Economics 90 : 630 – 49 . Google Scholar CrossRef Search ADS Stango, V., and Zinman. J. 2015 . Borrowing high versus borrowing higher: Price dispersion and shopping behavior in the U.S. credit card market. Review of Financial Studies 29 : 979 – 1006 . Google Scholar CrossRef Search ADS Stanton, R., and Wallace. N. 1998 . Mortgage choice: What’s the point? Real Estate Economics 26 : 173 – 205 . Google Scholar CrossRef Search ADS Stiglitz, J., and Weiss. A. 1981 . Credit rationing in markets with imperfect information. American Economic Review 71 : 393 – 410 . Taggart, R. A. 1977 . A model of corporate financing decisions. Journal of Finance 32 : 1467 – 84 . Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of The Society for Financial Studies. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png The Review of Financial Studies Oxford University Press

Screening on Loan Terms: Evidence from Maturity Choice in Consumer Credit

Loading next page...
 
/lp/ou_press/screening-on-loan-terms-evidence-from-maturity-choice-in-consumer-ZeeeclpzsY
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of The Society for Financial Studies. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
ISSN
0893-9454
eISSN
1465-7368
D.O.I.
10.1093/rfs/hhy024
Publisher site
See Article on Publisher Site

Abstract

Abstract We exploit a natural experiment in the largest online consumer lending platform to provide the first evidence that loan terms, in particular maturity choice, can be used to screen borrowers based on their private information. We compare two groups of observationally equivalent borrowers who took identical unsecured 36-month loans; for only one of the groups, a 60-month loan was also available. When a long-maturity option is available, fewer borrowers take the short-term loan, and those who do default less. Additional findings suggest borrowers self-select on private information about their future ability to repay. Received December 27, 2016; editorial decision December 12, 2017 by Editor Philip Strahan. Authors have furnished an Internet Appendix, which is available on the Oxford University Press Web Site next to the link to the final published paper online Received December 27, 2016; editorial decision December 12, 2017 by Editor Philip Strahan. Authors have furnished an Internet Appendix, which is available on the Oxford University Press Web Site next to the link to the final published paper online Asymmetric information between borrowers and lenders may induce inefficiencies in credit markets (e.g., Jaffee and Russell 1976; Stiglitz and Weiss 1981). In theory, lenders can partially mitigate these inefficiencies by using a menu of loans with different rates and contract terms to screen borrowers. Screening is achieved because borrowers with a higher unobservable probability of default self-select high-cost loans that contain features that are relatively more valuable to them, such as low collateral (Bester 1985), long maturity (Flannery 1986), or lenient covenants (Levine and Hughes 2005).1 Although the screening role of contract terms is well established in theory, empirical evidence of its relevance remains elusive. Most stylized facts consistent with screening are derived from the correlation between borrower contract choice and observable information (e.g., proxies for creditworthiness or the extent of private information).2 This evidence is circumstantial at best, because, by definition, screening implies that borrowers select loan terms based on information that is not observable by the lender (or the econometrician). In an attempt to document selection on unobservables, recent work has turned to the correlation between borrowers’ contract choices and ex post measures of their creditworthiness (e.g., default).3 However, even ex ante identical borrowers will exhibit different default probabilities ex post if they face different contract terms, for example, due to moral hazard. Thus, contract choice and default may be correlated even in the absence of selection. In this paper we provide the first direct evidence of the screening role of debt contract terms. This requires showing that (1) borrowers self-select into different loans based on their private information, and (2) the option chosen by borrowers with a higher unobserved probability of default has a higher rate. We first argue that, in order to empirically disentangle selection from the causal effect of contract terms, the econometrician must compare the repayment behavior of selected and unselected borrower samples who take the same loan contract.4 We then illustrate and apply this approach in the context of consumer credit in the United States, exploiting the staggered rollout of long-maturity loans by an online lending platform, Lending Club (hereafter, LC). This allows us to compare the ex post repayment behavior of ex ante identical borrowers facing the same short-term contract, but who chose their contract facing different menu of options, and who were thus differentially selected on maturity. Maturity serves as a screening device because long maturity reduces the need to roll over debt at a higher price in the future. Higher-risk borrowers, with an uncertain future observable creditworthiness, are willing to pay higher interest rates to secure this insurance.5 Consistent with this intuition, we find that LC borrowers who take a low-rate short-term loan when a long-term option is unavailable default substantially more than observationally identical borrowers who take the same short-term loan when LC also offers a higher-priced long-term loan. Thus, maturity can be effectively used as a screening device in credit markets: offering low-rate short-maturity and high-rate long-maturity loans induces borrowers of higher unobservable risk to self-select on the high-price contract. In the empirical setting, LC borrowers choose from a menu of loan amount, maturity, and price combinations. LC offers unsecured loans for amounts between $${\$}$$1,000 and $${\$}$$35,000 in either short—36 months—or long maturities—60 months. Loan price, set according to a proprietary algorithm, is increasing in amount, borrower risk, and maturity. Before 2013, long maturity loans were available only for amounts above $${\$}$$16,000. During 2013, the available menu of long-term loan options expanded twice: (1) to loans amounts between $${\$}$$12,000 and $${\$}$$16,000 in March 2013, and (2) to loan amounts between $${\$}$$10,000 and $${\$}$$12,000 in July 2013. Crucially for our analysis, during this period LC did not change the terms of any other borrowing option nor the criteria to qualify for a loan. Our empirical strategy compares the default rate of short-term loans for amounts between $${\$}$$10,000 and $${\$}$$16,000 issued before and after the availability of the long-maturity option at the corresponding amount (i.e., before and after the borrowers were screened on maturity). By comparing across borrowers who took identical loans (same short maturity, same price) we eliminate, by construction, the possibility that differences in repayment behavior are due to the causal impact of different contractual terms (due to, for example, moral hazard or the burden of repayment). To account for changes over time in the composition of borrowers on the LC platform we estimate a difference-in-differences specification that exploits the staggered roll-out of the long-term menu options, and uses short-term loans of amounts just above and just below the $${\$}$$10,000 to $${\$}$$16,000 interval to construct counterfactuals. Intuitively, our main test compares, amongst borrowers who appear ex ante identical in all observable dimensions and who took the exact same loans, the default rate of loans between $${\$}$$10,000 and $${\$}$$16,000 that were issued before and after the long-maturity loan became available at these amounts, relative to the same change in the default rate of loans between $${\$}$$5,000 and $${\$}$$10,000 or between $${\$}$$16,000 and $${\$}$$20,000 issued during the same period. The identification assumption is that any change in the composition of borrowers within a risk category that occurs for reasons other than the menu expansion, for example, due to changes in the supply of credit by other lenders, did not affect differentially loans between $${\$}$$12,000 and $${\$}$$16,000 in March 2013, and between $${\$}$$10,000 and $${\$}$$12,000 in July 2013, relative to other amounts in the analysis sample at those dates. To further ensure that all comparisons are done across observationally equivalent borrowers, we include in our specifications controls for all the borrower characteristics recorded by LC at origination, including month of origination, four-point FICO score range, and state fixed effects, among others. We begin by documenting that self-selection into long-maturity loans occurs among borrowers who would have borrowed between $${\$}$$10,000 and $${\$}$$16,000 had the long-maturity option not been added to the menu. We find that the number of short-maturity loans between $${\$}$$10,000 and $${\$}$$16,000 drops by 14.5% after the long-maturity loans become available, relative to loans issued at amounts just above and below this interval. Further, the decline was permanent and occurred on the same month the 60-month loan appeared in the menu for the corresponding amount. Then we explore how selection on maturity relates to ex post performance. We find that the average default rate of short-maturity loans decreases by 0.8 percentage points when a long-maturity loan is available at origination relative to when it is not. This implies that borrowers who look identical ex ante from the investors’ perspective but who have a higher default risk self-select out of short-term loans and into long-term ones. Assuming that the difference in short-term loan performance is due to the 14.5% of borrowers who self-select into long maturity, these self-selected borrowers would have had a default rate 5.5 percentage points higher (0.8/14.5) than the average 36-month borrower in our sample (9.2%). The findings are thus consistent with the joint hypotheses that LC borrowers have private information related to their future repayment probability, and that this private information affects loan maturity choice. Moreover, the large economic magnitude suggests that screening on maturity provides a powerful device for identifying, among a pool of observationally identical borrowers, those with the poorest repayment prospects. For maturity to be an effective screening device, long-term loans must be costlier than short-term loans. Indeed, we find that holding borrower characteristics and loan amount constant, the APR for 60-month LC loans was on average 3.3% higher than the APR for 36-month LC loans during our sample period. This represents a large maturity premium relative to the contemporaneous yield curve (0.2 percentage points) and can be fully explained by the 5.5 percentage point higher default rate of those borrowers who select into the long-maturity option.6 Consistent with a screening interpretation, only borrowers who are more exposed to repricing risk and most value the insurance provided by the long-maturity loan are willing to pay this higher maturity premium. Having established that borrowers select maturity based on private information that correlates with their repayment prospects, we turn to understanding the economic nature of this private information. Definitively characterizing the specific content of a borrower’s private information is by definition a difficult exercise because many underlying sources of information could explain differences in repayment behavior. Therefore, relative to our precise measurement of screening itself, we rely on the suggestive evidence our empirical setting permits. In theory, borrowers who are privately informed about their own high risk aversion will select the higher insurance against repricing risk provided by longer maturity loans (De Meza and Webb 2001). However, if risk averse borrowers are also expected to default less, self-selection on risk aversion is inconsistent with the higher default rate exhibited by long maturity borrowers. In addition, it is unlikely that borrowers are privately informed (relative to LC’s investors) about interest rate risk, the probability of credit supply shocks, or other macro determinants of the future cost of borrowing. It follows that borrowers who select long-maturity loans privately place higher value on the insurance it provides either because (1) they are less likely to have sufficient funds to make loan repayments in the future (e.g., they face a higher probability of job loss or illness or they will have a lower discretionary income), or (2) they are more exposed to rollover risk due to privately observed differences in the timing of their income. The two explanations have different predictions regarding the timing and level of default by borrowers who self-select into long maturity. Regarding the timing of default, borrowers who self-select into long maturity because their income arrives later will tend to default less over time, as their income realizes. In contrast, borrowers who self-select into long-maturity loans because they are more exposed to future shocks to their ability to repay default more over time, as the negative shocks realize. We find that selection does not significantly affect repayment during the first 12 months after origination, even though, unconditionally, more than a third of the loans that default do so during this period. In other words, we reject the hypothesis that the propensity to default of borrowers who self-select into long maturity loans decreases over time (relative to borrowers who self-select into short maturity loans). Regarding the level of default across maturities, if borrowers prefer a long- over a short-maturity loan because their income arrives in the future, their default probability should be lower under a long-term loan that aligns payments better with the timing of income. In our setting, however, the average default probability of 60-month loans is 3 percentage points higher than that of 36-month loans (conditioning on loan amount, month of origination, and FICO). This evidence is inconsistent with borrowers self-selecting on the basis of the timing of their income, and consistent with them self-selecting on private information about the exposure to shocks to their ability to repay. We find additional evidence in support of the interpretation that borrowers select maturity based on private information about the mean or the volatility of their discretionary income, for example, about the probability of losing a job or the need to take care of an elderly parent. We find that, on average, borrowers in the selected group—borrowers who chose the short maturity when the loan maturity was available— have higher future FICO scores and less time-series volatility of FICO scores relative to the unselected group. Thus, borrowers in the selected group are both observably more creditworthy, as measured by their FICO scores, and less exposed to shocks to their creditworthiness. Moreover, we find that the propensity for borrowers to prepay the short-term loan is lower in the selected group relative to the unselected group. Although this result is not statistically significant, it is inconsistent with the hypothesis that short-term loans are selected by borrowers based on private information that their income arrives sooner. In theory, our results could also be driven by borrowers who have a preference for long-term loans for behavioral reasons (e.g., borrowers may evaluate the price of a loan by the installment amount instead of by the interest rate and fees) and who, at the same time, are more likely to default. However, 87% of LC borrowers claim to use the LC loan proceeds to repay credit card debt. Since credit card debt is essentially very long-term debt, most borrowers in our sample are actively choosing to lower the maturity profile of their debt and to increase, not decrease, the monthly installment amounts.7 This provides suggestive evidence that LC borrowers seem to be unconstrained enough to commit to increase their minimum monthly payments relative to those imposed by their existing credit card debt and sufficiently sophisticated to understand the difference between price and monthly payment amounts. Moreover, it is important to note that, for unconstrained sophisticated borrowers, loan maturity (a contractual feature of the loan) is distinct from the actual timing of loan repayments (a choice variable). Insofar as the borrower has ongoing access to credit markets, an impatient borrower who has a short-term loan can lower the effective out-of-pocket payments by undertaking additional borrowing each period. Our paper is related to but distinct from the theory of Diamond (1991), who uses a framework with asymmetric information to predict a link between observable creditworthiness and the type of maturity that all borrowers will pool on in equilibrium. By isolating screening on private information, our paper is also distinct from theories of maturity choice that are unrelated to ex ante asymmetric information such as asset maturity matching (e.g., Myers 1977; Hart and Moore 1994), agency problems (e.g., Hart and Moore 1995), market conditions (e.g., Bosworth 1971; Taggart 1977), minimize rollover risk (e.g., Graham and Harvey 2001), predictable violations of the expectations hypothesis (e.g., Baker et al. 2003), and government behavior (e.g., Greenwood et al. 2010). Our paper contributes to this literature by relating maturity choice to a borrower’s private, that is, unobservable, information. Our paper also contributes to a relatively small empirical literature that has measured adverse selection in credit markets. Karlan and Zinman (2009) use an experiment in South Africa that isolates adverse selection on loan interest rates by randomizing the offered loan interest rate but resetting all loan terms after selection occurs. A different approach is taken by Adams et al. (2009) and Dobbie and Skiba (2013) estimate adverse selection on loan amount among subprime borrowers as a residual, given by the correlation between default and loan size that cannot be explained by the direct effect of loan size on default. Our results not only constitute evidence of adverse selection on a novel contract term (maturity), but also demonstrate that screening on maturity allows the lender to charge prices that are commensurate with borrowers’ unobserved default risk. Finally, our results suggest that the screening role of maturity may extend to other settings where long-term contracts provide insurance against repricing risk, such as labor (Holmstrom 1983) and health insurance markets (Cochrane 1995; Finkelstein et al. 2005). 1. Setting LC is the largest online lending platform in the United States. In 2014 alone, LC originated $${\$}$$4.4B in consumer loans across 45 states. By comparison, Prosper Marketplace, its nearest rival, originated $${\$}$$1.6B in the same year.8 LC loans are unsecured amortizing loans for amounts between $${\$}$$1,000 and $${\$}$$35,000 (in $${\$}$$25 intervals). LC loans are available in two maturities: 36 months, which is available for all amounts, and 60 months, which is available for different amounts at different points in time. Loans are funded directly by institutional and retail investors (LC holds no financial stake in the loans), and 80% of the total funds are provided by institutional investors (Morse 2015). Since each loan is considered an individual security by the Securities and Exchange Commission, the agency that regulates online loan marketplaces in the United States, LC is required to reveal publicly all the information used to evaluate the risk of each loan. This is an ideal institutional setting for the purposes of studying screening on borrowers’ private information, since we have all the borrower information that lenders observe at the time of origination. When a borrower applies for a loan with LC, she first enters her yearly individual income and sufficient personal information to allow LC to obtain the borrower’s credit report. In most cases (e.g., 71% of all loans issued in 2013) LC verifies the yearly income that a borrower enters using pay stubs, W2 tax records, or by calling the employer. Every loan application is processed in two steps. First, LC decides whether a borrower is eligible for a loan on the platform. The eligibility decision is made mechanically based purely on hard borrower information observable at the time of origination. For example, during 2013 LC issued loans only to borrowers with FICO scores over 660, nonmortgage debt payments to income ratios below 35%, and credit histories of at least 36 months. If LC determines that a borrower is eligible for a loan in the first step, she is then assigned to one of 25 risk categories (labeled by LC as risk “subgrades”). This assignment is made using a proprietary credit-risk assessment algorithm that uses the hard information in a borrower’s credit report (e.g., FICO score, outstanding debt, repayment status) and income. The assignment to risk category is made prior to the borrower selecting a loan amount or maturity and is therefore independent of both choices. The risk category determines the entire menu of interest rates faced by the borrower, for all loan amounts and for the two available maturities. That is, two borrowers assigned to the same risk category at the same time will face the same menu of interest rates for all amounts and for the two maturities. Interest rates for each subgrade are weakly increasing in amount and strictly increasing in maturity (ceteris paribus). The terms of all loans, other than interest rate, amount, and maturity, are identical. Once a borrower selects a loan from the menu, it is listed on LC’s website for investors’ consideration. Investors cannot affect any of the terms of the loan: they decide only whether or not to fund it. According to LC, over 99% of all listed loans are funded.9 Thus, we ignore the supply side of funds in the analysis. As of 2013, LC charges an origination fee, subtracted at origination, that varies between 1.1% and 5% of the loan amount depending on credit score and a further 1% fee from all loan payments made to investors. 1.1 Lending Club LC is the largest online lending platform in the United States. In 2014 alone, LC originated $${\$}$$4.4B in consumer loans across 45 states. By comparison, Prosper Marketplace, its nearest rival, originated $${\$}$$1.6B in the same year.10 LC loans are unsecured amortizing loans for amounts between $${\$}$$1,000 and $${\$}$$35,000 (in $${\$}$$25 intervals). LC loans are available in two maturities: 36 months, which is available for all amounts, and 60 months, which is available for different amounts at different points in time. Loans are funded directly by institutional and retail investors (LC holds no financial stake in the loans), and 80% of the total funds are provided by institutional investors (Morse 2015). Since each loan is considered an individual security by the Securities and Exchange Commission, the agency that regulates online loan marketplaces in the United States, LC is required to reveal publicly all the information used to evaluate the risk of each loan. This is an ideal institutional setting for the purposes of studying screening borrowers’ private information, since we have all the borrower information that lenders observe at the time of origination. When a borrower applies for a loan with LC, she first enters her yearly individual income and sufficient personal information to allow LC to obtain the borrower’s credit report. In most cases (e.g., 71% of all loans issued in 2013) LC verifies the yearly income that a borrower enters using pay stubs, W2 tax records, or by calling the employer. Every loan application is processed in two steps. First, LC decides whether a borrower is eligible for a loan on the platform. The eligibility decision is made mechanically based purely on hard borrower information observable at the time of origination. For example, during 2013 LC issued loans only to borrowers with FICO scores over 660, nonmortgage debt payments to income ratios below 35%, and credit histories of at least 36 months. If LC determines that a borrower is eligible for a loan in the first step, she is then assigned to one of 25 risk categories (labeled by LC as risk “subgrades”). This assignment is made using a proprietary credit-risk assessment algorithm that uses the hard information in a borrower’s credit report (e.g., FICO score, outstanding debt, repayment status) and income. The assignment to risk category is made prior to the borrower selecting a loan amount or maturity and is therefore independent of both choices. The risk category determines the entire menu of interest rates faced by the borrower, for all loan amounts and for the two available maturities. That is, two borrowers assigned to the same risk category at the same time will face the same menu of interest rates for all amounts and for the two maturities. Interest rates for each subgrade are weakly increasing in amount and strictly increasing in maturity (ceteris paribus). The terms of all loans, other than interest rate, amount, and maturity, are identical. Once a borrower selects a loan from the menu, it is listed on LC’s website for investors’ consideration. Investors cannot affect any of the terms of the loan: they decide only whether or not to fund it. According to LC, over 99% of all listed loans are funded.11 Thus, we ignore the supply side of funds in the analysis. As of 2013, LC charges an origination fee, subtracted at origination, that varies between 1.1% and 5% of the loan amount depending on credit score and a further 1% fee from all loan payments made to investors. 1.2 Staggered expansion of 60-month loans Before March 2013, 60-month loans were available only for loans of $${\$}$$16,000 and above. A borrower could not synthetically create a 60-month loan for an amount less than $${\$}$$10,000 using prepayment, because prepayment reduces the number of installments without changing their amount, effectively reducing the maturity of the loan. In March 2013, LC introduced 60-month loans between $${\$}$$12,000 and $${\$}$$16,000 to the menu. And in July 2013, it further expanded the available 60-month loans to include amounts between $${\$}$$10,000 and $${\$}$$12,000. The consequences of the menu expansion can be seen in Figure 1, where we plot the fraction of loans originated every month that have a 60-month maturity, grouped by loan amount. On December 2012, the first month of the analysis sample period, around 40% of loans between $${\$}$$16,000 and $${\$}$$20,000 are 60-month loans. This fraction remains relatively constant throughout the sample period, until October 2013. The fraction of 60-month loans is zero for loan amounts below $${\$}$$16,000 in December 2012, and jumps up for $${\$}$$12,000 to $${\$}$$16,000 loans in March 2013, and then for $${\$}$$10,000 to $${\$}$$12,000 loans on July 2013. By the end of the sample the fraction of 60-month loans stabilizes at around 30% for $${\$}$$12,000 to $${\$}$$16,000 loans and around 25% for $${\$}$$10,000 to $${\$}$$12,000 loans. The fraction of 60-month $${\$}$$5,000 to $${\$}$$10,000 loans remains at zero throughout the sample period. As we discuss in detail in Section 2, our empirical strategy exploits the fact that loan amounts between $${\$}$$10,000 and $${\$}$$16,000 were affected by the expansion of a long-maturity option, and that loan amounts outside this range were not. Figure 1 View largeDownload slide Staggered expansion of 60-month loans This figure shows the time series of the number of 60-month loans by listing month for $${\$}$$10,000 to $${\$}$$12,000 and $${\$}$$12,000 to $${\$}$$16,000. Figure 1 View largeDownload slide Staggered expansion of 60-month loans This figure shows the time series of the number of 60-month loans by listing month for $${\$}$$10,000 to $${\$}$$12,000 and $${\$}$$12,000 to $${\$}$$16,000. 1.3 Summary statistics LC makes publicly available on its website all the information used to assign borrowers to risk categories, the assigned risk category, and the loan performance of all funded loans. Our main analysis is conducted using data downloaded from LC’s website as of April 2015. The data is a cross section of all loans originated at LC. Variables are measured either at the time of origination (e.g., date of loan, loan terms, borrower income and credit report data, state of residence) or at the time of the performance data download (e.g., loan status, time of last payment, current FICO score of borrower). We complement our main outcomes, which are measured as of April 2015, with measures of FICO scores obtained from two previous loan performance updates, August 2014 and December 2014.12 We use the origination date of each loan to restrict the sample period of the analysis to meet two criteria: (1) that it contains the dates in which the 60-month loan menu was expanded (March 2013 and June 2013) and that are the basis of our empirical analysis and (2) that the interest rate assigned to each amount-maturity combination remained constant within each risk category (in other words, that all menu options other than the added long-term option remained constant). Thus, the beginning and ending months of our analysis sample are determined by two dates, surrounding the menu expansion events, on which we observe that LC repriced menu options (December 2012 and October 2013). We verify empirically that the interest rates of all risk category-amount pairs for 36-month loans are unchanged between these dates.13 We further limit the sample of loans to include those for amounts between $${\$}$$5,000 (closed) and $${\$}$$20,000 (open) because the interest rate schedule jumps discretely at $${\$}$$5,000 and $${\$}$$20,000 for all credit risk categories.14 This interval includes all 36-month loans issued at amounts affected by the 60-month borrowing threshold reduction ($${\$}$$10,000 to $${\$}$$16,000), as well as amounts above and below this interval that allow us to control for any time-of-origination changes in unobserved borrower creditworthiness or credit demand. Finally, we further limit our sample to those loans for which we can uniquely match the loan that a borrower chose to the menu associated with the risk category she was assigned to based on her publicly available data. We obtain this unique match for 98.6% of all loans in the sample period (we drop observations for which this matching does not yield a unique value). Our final sample has 60,514 loans.15 Table 1, panel A, presents summary statistics for the subset of our sample corresponding to the 12,091 36-month loans with amounts between $${\$}$$5,000 and $${\$}$$20,000 issued between December 2012 and February 2013 (prior to the menu expansion). On average, loans for this subsample have a 16.3% APR and a monthly installment of $${\$}$$380. Borrowers self-report that 87% of all loans were issued to refinance existing debt (this includes “credit card” and “debt consolidation”). We define a loan to be in default if it is late by more than 120 days.16 According to this definition, 9.2% of the loans in the subsample are in default as of April 2015. Figure 2 shows the default hazard rate by months-since-origination for loans issued before the menu expansion.17 The hazard rate exhibits the typical hump shape and peaks between 13 and 15 months. Figure 2 View largeDownload slide Hazard rate of default This figure shows the hazard rate of default by month since origination for 36-month loans issued by LC in amounts between $${\$}$$5,000 and $${\$}$$20,000 between December 2012 and February 2013 (pre-period). A loan is in default if payments are 120 or more days late in April 2015. The timing of default is the month, measured as time since origination in which payments were first missed. The hazard rate at horizon $$t$$ is the number of loans that enter default at that horizon as a fraction of the number of loans that are in good standing at $$t-1$$. Figure 2 View largeDownload slide Hazard rate of default This figure shows the hazard rate of default by month since origination for 36-month loans issued by LC in amounts between $${\$}$$5,000 and $${\$}$$20,000 between December 2012 and February 2013 (pre-period). A loan is in default if payments are 120 or more days late in April 2015. The timing of default is the month, measured as time since origination in which payments were first missed. The hazard rate at horizon $$t$$ is the number of loans that enter default at that horizon as a fraction of the number of loans that are in good standing at $$t-1$$. Table 1 Pre-period summary statistics Mean P50 SD A. Loan characteristics APR (%) 16.3 16.0 4.1 Installment ($ ${\$}$ $) 379.9 360.9 125.1 For refinancing (%) 87.0 Default (%) 9.2 Fully paid (%) 37.6 B. Borrower characteristics Annual income ($ ${\$}$ $) 65,745 57,500 74,401 Debt payments / Income (%) 17.4 16.9 7.7 FICO at origination (high range of 4-point bin) 695 689 26 FICO at latest data pull (high range of 4-point bin) 685 699 70 Home ownership (%) 55.5 Total debt excl mortgage ($ ${\$}$ $) 38,153 29,507 33,805 Revolving balance ($ ${\$}$ $) 14,549 11,592 12,719 Revolving utilization (%) 60.7 62.7 21.9 Months of credit history 182 164 84 N 12,091 Mean P50 SD A. Loan characteristics APR (%) 16.3 16.0 4.1 Installment ($ ${\$}$ $) 379.9 360.9 125.1 For refinancing (%) 87.0 Default (%) 9.2 Fully paid (%) 37.6 B. Borrower characteristics Annual income ($ ${\$}$ $) 65,745 57,500 74,401 Debt payments / Income (%) 17.4 16.9 7.7 FICO at origination (high range of 4-point bin) 695 689 26 FICO at latest data pull (high range of 4-point bin) 685 699 70 Home ownership (%) 55.5 Total debt excl mortgage ($ ${\$}$ $) 38,153 29,507 33,805 Revolving balance ($ ${\$}$ $) 14,549 11,592 12,719 Revolving utilization (%) 60.7 62.7 21.9 Months of credit history 182 164 84 N 12,091 This table shows summary statistics of the main sample of Lending Club borrowers for pre-expansion months. The main sample includes all 36-month loans with a list date between December 2012 and March 2013, and an amount between $ ${\$}$ $5,000 and $ ${\$}$ $20,000 and for which we estimate an initial risk category based on LC’s publicly available information. Table 1 Pre-period summary statistics Mean P50 SD A. Loan characteristics APR (%) 16.3 16.0 4.1 Installment ($ ${\$}$ $) 379.9 360.9 125.1 For refinancing (%) 87.0 Default (%) 9.2 Fully paid (%) 37.6 B. Borrower characteristics Annual income ($ ${\$}$ $) 65,745 57,500 74,401 Debt payments / Income (%) 17.4 16.9 7.7 FICO at origination (high range of 4-point bin) 695 689 26 FICO at latest data pull (high range of 4-point bin) 685 699 70 Home ownership (%) 55.5 Total debt excl mortgage ($ ${\$}$ $) 38,153 29,507 33,805 Revolving balance ($ ${\$}$ $) 14,549 11,592 12,719 Revolving utilization (%) 60.7 62.7 21.9 Months of credit history 182 164 84 N 12,091 Mean P50 SD A. Loan characteristics APR (%) 16.3 16.0 4.1 Installment ($ ${\$}$ $) 379.9 360.9 125.1 For refinancing (%) 87.0 Default (%) 9.2 Fully paid (%) 37.6 B. Borrower characteristics Annual income ($ ${\$}$ $) 65,745 57,500 74,401 Debt payments / Income (%) 17.4 16.9 7.7 FICO at origination (high range of 4-point bin) 695 689 26 FICO at latest data pull (high range of 4-point bin) 685 699 70 Home ownership (%) 55.5 Total debt excl mortgage ($ ${\$}$ $) 38,153 29,507 33,805 Revolving balance ($ ${\$}$ $) 14,549 11,592 12,719 Revolving utilization (%) 60.7 62.7 21.9 Months of credit history 182 164 84 N 12,091 This table shows summary statistics of the main sample of Lending Club borrowers for pre-expansion months. The main sample includes all 36-month loans with a list date between December 2012 and March 2013, and an amount between $ ${\$}$ $5,000 and $ ${\$}$ $20,000 and for which we estimate an initial risk category based on LC’s publicly available information. Table 1, panel B, shows borrower-level statistics of this sample. On average, LC borrowers in our sample have an annual income of $ ${\$}$ $65,745 and use 17.4% of their monthly income to pay debts excluding mortgages. The average FICO score at origination is 695, and credit report pulls show that the FICO score has on average decreased to 685 approximately 1 year later. LC borrowers have access to credit markets: 56% report that they own a house or have an outstanding mortgage. The average borrower has $ ${\$}$ $38,153 in debt excluding mortgage debt and $ ${\$}$ $14,549 in revolving debt, which represents a 61% revolving line utilization rate (the average revolving credit limit is $ ${\$}$ $27,464). LC borrowers have on average approximately 15 years of credit history. We compare our summary statistics to the credit card user statistics from Agarwal et al. (2015) to obtain a sense of how representative LC borrowers are of the average U.S. consumer credit user within the same FICO range. Using the average credit card limit in the subsample of borrowers with FICO scores between 660 and 719 ($ ${\$}$ $7,781) and assuming the average number of credit cards held by the average card-holder is 3.7 (according to Gallup 2014 survey) implies that the representative U.S. user of consumer credit has a revolving credit limit of $ ${\$}$ $28,789, very close to the $ ${\$}$ $27,464 average revolving credit limit of the LC borrowers in our sample. Thus, LC’s selection criteria imply that the analysis sample is drawn exclusively from prime U.S. consumer credit users (as measured by FICO scores), but LC borrowers do not seem to be different in their revolving credit availability from the average U.S. consumer credit user in the same FICO range. 2. Measuring Screening on Maturity 2.1 Empirical strategy We exploit the staggered menu expansion of 60-month loans during 2013 to identify selection on maturity. As described above, LC offered new loan options at longer maturities for amounts already offered on short-term contracts prior to the expansion. Crucially, the pricing of all loan options available prior to the expansion was unchanged after the expansion for all 25 risk categories during our sample period. This ensures that the only difference in the menu of borrowing options offered to borrowers assigned to the same risk category before and after the expansion is the availability of 60-month loans in lower amounts.18 We compare the outcomes of borrowers who took the short-term loan before the menu expanded with those who were assigned to the same risk category and took it after the expansion. We develop a research design that accounts for any other changes over time in the composition of borrowers within a risk category that are not driven by the menu expansion. The LC setting provides two sources of variation that allow us to construct a counterfactual using a difference-in-differences approach: (1) the menu expansion was staggered over time for different loan amounts (eventually selected amounts) and (2) some loan amounts were never affected by the menu expansion (never-selected and always-selected amounts). The three groups of loans defined this way by the loan amount and the time of origination are represented in Figure 3. Loans of amounts between $ ${\$}$ $10,000 and $ ${\$}$ $16,000 are eventually selected, in the sense that they are unselected at the beginning of the sample (no long-term option is available at the time of origination) and selected (long-term option is available) at the end of the sample. Since the menu expansion was staggered, loan amounts between $ ${\$}$ $10,000 and $ ${\$}$ $12,000 serve as a control group for loan amounts between $ ${\$}$ $12,000 and $ ${\$}$ $16,000 that were affected by the March expansion and the reverse applies for the expansion in July. We build two additional control groups with loan amounts whose selection status was not affected by the menu expansion. The always selected, for which the long-term loan was always available at the time of origination during the sample period ($ ${\$}$ $16,000 to $ ${\$}$ $20,000), and the never selected, for which the long-term option never became available ($ ${\$}$ $5,000 to $ ${\$}$ $10,000). Our identification assumption is that any change in the composition of borrowers within a risk category, for example, due to changes in the economic environment, changes in the borrowing options outside of LC, or changes in how LC assigns borrowers to risk categories, does not affect differentially borrowers opting to take loans between $ ${\$}$ $12,000 and $ ${\$}$ $16,000 in March and borrowers opting to take loans between $ ${\$}$ $10,000 and $ ${\$}$ $12,000 in July relative to loans issued at control amounts. Under this assumption, comparing the change in performance of eventually selected amounts before and after the menu expansion at those amounts with the change in performance of the control amounts in the same risk category isolates the effect of maturity selection induced by the menu expansion. We further include a comprehensive set of granular borrower controls, which ensures that the estimations come from comparing borrowers who took loans at selected amounts to observationally equivalent borrowers taking loans at nonselected amounts. Figure 3 View largeDownload slide Stylized depiction of identification strategy This figure shows a stylized depiction of our difference-in-differences strategy using the expansion of the menu of borrowing options. Figure 3 View largeDownload slide Stylized depiction of identification strategy This figure shows a stylized depiction of our difference-in-differences strategy using the expansion of the menu of borrowing options. Before providing evidence to support the identification assumption (see section 2.2.5 below), we discuss here its plausibility. First, even though it is unlikely that changes in economic conditions may have affected the demand for loans between $ ${\$}$ $10,000 and $ ${\$}$ $16,000 exactly at the same month of the menu expansion, to check whether there were any aggregate changes in the demand for LC loans, we plot in Figure 4 the total dollar amount of LC loans issued by month. There is no indication that the growth rate of LC lending changed around the dates of the two 60-month loan expansions. Second, in web searches we found no evidence of a change in the outside borrowing options that exclusively targeted the eventually selected loan amounts ($ ${\$}$ $10,000 to $ ${\$}$ $16,000) in a manner that corresponds with the staggered expansion of the menu. Third, we found no evidence that LC released advertising targeted at 60-month loans between $ ${\$}$ $10,000 to $ ${\$}$ $16,000 during the analysis sample. On the contrary, according to the information reported in the website Internet Archive, LC continued to advertise that 60-month loans were available only for amounts above $ ${\$}$ $16,000 until November 2013, after our analysis period ends.19 Fourth, any change in LC’s loan qualification criteria or assignment to risk categories cannot, by construction, affect borrower selection across different amounts within a risk category. The reason is that both eligibility for an LC loan and the assignment to risk categories are determined using borrowers’ observable information before the borrower selects a loan amount from the menu. Nevertheless, we verify that the criteria used to determine eligibility for an LC loan (the minimum FICO score of 660, minimum credit history length of 36 months, and maximum nonmortgage debt to income threshold of 35%) remain constant over the sample period. Figure 4 View largeDownload slide Total dollar amount issued by LC by month of listing This figure shows the time series of the total dollar amount of LC loans (of both maturities) by listing month since 2012. The vertical dashed lines show the 2 months in which the 60-month loan minimum amount was reduced. Figure 4 View largeDownload slide Total dollar amount issued by LC by month of listing This figure shows the time series of the total dollar amount of LC loans (of both maturities) by listing month since 2012. The vertical dashed lines show the 2 months in which the 60-month loan minimum amount was reduced. It is important to emphasize that there is no appropriate counterfactual for borrower selection on the 60-month loans. This is why our empirical strategy relies exclusively on a comparison of 36-month loans taken before and after the expansion, and ignores any changes in the composition of borrowers who take 60-month loans. The mix of borrowers taking a 60-month loan could have changed, for example, because some borrowers who take the 60-month loan would have not borrowed at all before this option became available. Since we are unable to account for such selection on the extensive margin for 60-month loans, we are limited in how much we can infer about the determinants of the performance of the 60-month loans. The focus on 36-month loans also implies that our approach for measuring the effect of selection is based on a revealed-preference argument, which relies on the axiom of independence of irrelevant alternatives. Specifically, we assume that a borrower who prefers not to borrow from LC over taking a 36-month loan when there is no 60-month option will not prefer to take the 36-month loan once the 60-month loan becomes available. Finally, we note that the empirical approach is aimed at estimating the effect of selection on maturity of LC loans. If LC borrowers have access to 60-month loans between $ ${\$}$ $10,000 and $ ${\$}$ $16,000 at a similar price elsewhere during the analysis period, we should fail to reject the null hypothesis and conclude that there is no screening on maturity in LC (since borrowers who wish to select long-term loans would already be taking them elsewhere). In effect, any impact of the menu expansion at LC can also be interpreted as indirect evidence that consumer credit markets are imperfectly competitive. This might be true because some intermediaries have a technology advantage over others that generates some market power or because there are search frictions in the market.20 2.2 Evidence of selection We start by measuring the amount of selection induced by the menu expansion: how does the number of borrowers who take the short-term loan at any given amount change after the long-term option becomes available at that amount? To do so we collapse the data and count the number of loans $ $N_{jkt}$ $ at the month of origination ($ $t$ $)$ $\times$ $risk category ($ $j$ $)$ $\times$ $$ ${\$}$ $1,000 loan amount bin ($ $k$ $) level for all 36-month loans issued during our sample period (amount bins measured starting from $ ${\$}$ $10,000, e.g., $ ${\$}$ $10,000 to $ ${\$}$ $11,000, $ ${\$}$ $11,000 to $ ${\$}$ $12,000, and so on). We define a “selected” dummy variable $ $D_{kt}$ $ equal to one for those loan amount bin-month pairs where a 60-month option was available, and zero otherwise. That is, \[D_{kt}=\begin{cases}1 & \mathit{if}\,\, \$16,000>\mathit{Loan}\ \mathit{Amounts}\geq\,\, \$12,000\;\&\;t\geq \mathit{March}\ 2013\\1 & \mathit{if}\,\, \$12,000>\mathit{Loan}\ \mathit{Amounts}\geq\,\, \$10,000\;\&\;t\geq \mathit{July}\ 2013\\0 & \,\mathit{otherwise}\end{cases}\] Then we estimate the following difference-in-differences regression: \begin{equation}\mathit{log}(N_{jkt})=\beta'_{k}+\delta'_{jt}+\gamma'\times D_{kt}+\epsilon_{jkt}.\label{eq:log_number_regression}\end{equation} (1) The coefficient of interest is $ $\gamma'$ $, the average percent change in the number of short-maturity loans originated for eventually selected amounts (i.e., amounts in which a long-maturity loan was not available at the beginning of the sample and became available due to the menu expansion) relative to control amounts. We include amount bin fixed effects $ $\beta'_{k}$ $, which control for level differences in the number of loans in each $ ${\$}$ $1,000 bin. In turn, risk category$ $\times$ $month fixed effects $ $\delta'_{jt}$ $ control for any changes over time in the number of borrowers who are approved at each of the 25 different risk categories. Table 2, Column 1, shows the results of regression (1), estimated on the full sample of borrowers who took a 36-month loan between $ ${\$}$ $5,000 and $ ${\$}$ $20,000 during the sample period (December 2012 to October 2013). The point estimate of $ $\gamma'$ $ is negative and significant, and implies that the number of borrowers who took a short-term loan is 14.5% lower once the new long-term loan option for the same amount becomes available. This estimate provides us with a magnitude for the number of borrowers who would have taken a short-term loan if the long term option had not been available.21 Table 2 Regression results: Selection into long-maturity loans (1) (2) (3) (4) (5) $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $D_{amount1000,t}$ $ –0.1451*** –0.0825 0.0586 –0.0355 –0.0441 (0.033) (0.065) (0.064) (0.048) (0.028) Main 60m 36m 36m Sample 16k - 24k 16k - 24k 1k - 10k Placebo Observations 3,663 1,738 1,637 2,374 3,861 $ $R^{2}$ $ 0.817 0.724 0.802 0.761 0.862 (1) (2) (3) (4) (5) $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $D_{amount1000,t}$ $ –0.1451*** –0.0825 0.0586 –0.0355 –0.0441 (0.033) (0.065) (0.064) (0.048) (0.028) Main 60m 36m 36m Sample 16k - 24k 16k - 24k 1k - 10k Placebo Observations 3,663 1,738 1,637 2,374 3,861 $ $R^{2}$ $ 0.817 0.724 0.802 0.761 0.862 This table shows that selection into the new 60-month options was higher among borrowers who would have selected a 36-month loan in the same range of amounts as the new 60-month options. The sample corresponds to loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $20,000 with a list date between December 2012 and October 2013. Column 1 shows the coefficient of the regression of log(N), the logarithm of the number of loans at each month, the credit-risk category, and the $ ${\$}$ $1,000 loan amount interval level, on a dummy that equals one for loan amounts at which the 60-month loan was first not available and then made available, and zero otherwise. Columns 2, 3, and 4 show the regression results on different samples where we redefine $ $D_{amount1000,t}$ $ in an ad hoc manner for each column. Column 2 restricts the sample to 60-month loans issued in the main sample period for amounts between $ ${\$}$ $16,000 and $ ${\$}$ $24,000; $ $D_{amount1000,t}$ $ is defined as one for loan amounts between $ ${\$}$ $16,000 and $ ${\$}$ $20,000 on and after March 2013, and zero in other cases. Column 3 restricts the sample to 36-month loans issued in the main sample period for amounts between $ ${\$}$ $16,000 and $ ${\$}$ $24,000; $ $D_{amount1000,t}$ $ is defined as one for loan amounts between $ ${\$}$ $16,000 and $ ${\$}$ $20,000 on and after March 2013, and zero in other cases. Column 4 restricts the sample to 36-month loans issued in the main sample period for amounts between $ ${\$}$ $1,000 and $ ${\$}$ $10,000; $ $D_{amount1000,t}$ $ is defined as one for loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $10,000 in and after July 2013 and zero in other cases. Column 5 reports the tests of a placebo sample that includes loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $20,000 issued between July 2013 and May 2014. Standard errors are robust to heteroscedasticity. *, **, and *** represent significance at the 10%, 5%, and 1% level, respectively. Table 2 Regression results: Selection into long-maturity loans (1) (2) (3) (4) (5) $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $D_{amount1000,t}$ $ –0.1451*** –0.0825 0.0586 –0.0355 –0.0441 (0.033) (0.065) (0.064) (0.048) (0.028) Main 60m 36m 36m Sample 16k - 24k 16k - 24k 1k - 10k Placebo Observations 3,663 1,738 1,637 2,374 3,861 $ $R^{2}$ $ 0.817 0.724 0.802 0.761 0.862 (1) (2) (3) (4) (5) $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $log\left(N\right)$ $ $ $D_{amount1000,t}$ $ –0.1451*** –0.0825 0.0586 –0.0355 –0.0441 (0.033) (0.065) (0.064) (0.048) (0.028) Main 60m 36m 36m Sample 16k - 24k 16k - 24k 1k - 10k Placebo Observations 3,663 1,738 1,637 2,374 3,861 $ $R^{2}$ $ 0.817 0.724 0.802 0.761 0.862 This table shows that selection into the new 60-month options was higher among borrowers who would have selected a 36-month loan in the same range of amounts as the new 60-month options. The sample corresponds to loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $20,000 with a list date between December 2012 and October 2013. Column 1 shows the coefficient of the regression of log(N), the logarithm of the number of loans at each month, the credit-risk category, and the $ ${\$}$ $1,000 loan amount interval level, on a dummy that equals one for loan amounts at which the 60-month loan was first not available and then made available, and zero otherwise. Columns 2, 3, and 4 show the regression results on different samples where we redefine $ $D_{amount1000,t}$ $ in an ad hoc manner for each column. Column 2 restricts the sample to 60-month loans issued in the main sample period for amounts between $ ${\$}$ $16,000 and $ ${\$}$ $24,000; $ $D_{amount1000,t}$ $ is defined as one for loan amounts between $ ${\$}$ $16,000 and $ ${\$}$ $20,000 on and after March 2013, and zero in other cases. Column 3 restricts the sample to 36-month loans issued in the main sample period for amounts between $ ${\$}$ $16,000 and $ ${\$}$ $24,000; $ $D_{amount1000,t}$ $ is defined as one for loan amounts between $ ${\$}$ $16,000 and $ ${\$}$ $20,000 on and after March 2013, and zero in other cases. Column 4 restricts the sample to 36-month loans issued in the main sample period for amounts between $ ${\$}$ $1,000 and $ ${\$}$ $10,000; $ $D_{amount1000,t}$ $ is defined as one for loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $10,000 in and after July 2013 and zero in other cases. Column 5 reports the tests of a placebo sample that includes loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $20,000 issued between July 2013 and May 2014. Standard errors are robust to heteroscedasticity. *, **, and *** represent significance at the 10%, 5%, and 1% level, respectively. In the Internet Appendix Table A1 we conduct robustness tests where we vary the dimensions along which we collapse the loan-level data to count the number of loans. There we show that the selection result is slightly smaller in magnitude, ranging from 6.3% to 10%, but statistically significant across all specifications when we collapse the data in month of origination$ $\times$ $risk category$ $\times$ $$ ${\$}$ $1,000 loan amount$ $\times$ $ four-point FICO score bins (Column 1 in panel A), month of origination$ $\times$ $risk category$ $\times$ $$ ${\$}$ $100 loan amount $ $\times$ $ four-point FICO score bins (Column 1 in panel B), and month of origination$ $\times$ $ four-point FICO score$ $\times$ $five-point debt-to-income bins (Column 1 in panel C). We emphazise the estimated coefficient of 14.5% shown in Column 1 of Table 2 as our baseline result because it implies the smallest difference in default rates for individuals who choose the short term loan. We note that, qualitatively or quantitatively, none of our results, except the magnitude of the average difference in default rates, depend on this choice. 2.3 Selection and repayment Having shown that the expansion of the menu of borrowing options induced a significant amount of self-selection from short-term to long-term loans, we run our main test to uncover the unobserved quality of the borrowers who selected into the new long-term contract. We estimate the following difference-in-differences specification on the sample of 36-month loans: \begin{equation}Default_{i}=\beta_{i}^{1000bin}+\delta_{i}^{jt}+\gamma\times D_{i}+X_{i}+\epsilon_{i},\label{eq:selection_regression}\end{equation} (2) where data is at the loan level $ $i$ $. The outcome variable, $ $Default_{i}$ $, is defined as a dummy that equals one if the loan is late by more than 120 days measured as of April 2015. Standard errors are clustered at the state level (45 clusters). The main explanatory variable of interest, $ $D_{i}$ $, is a dummy equal to one if the 36-month loan $ $i$ $ is issued at a time when a 60-month loan of the same amount is also available, and zero otherwise: \[D_{i}=\begin{cases}1 & \mathit{if}\ \$16,000>\mathit{Loan}\ \mathit{Amount}_{i}\geq \$12,000\;\&\;t\geq \mathit{March}\ 2013\\1 & \mathit{if}\ \$12,000>\mathit{Loan}\ \mathit{Amount}_{i}\geq \$10,000\;\&\;t\geq \mathit{July}\ 2013\\0 & \,\mathit{otherwise}\end{cases}\] The coefficient of interest, $ $\gamma,$ $ measures the change in the default rate of 36-month loans for eventually selected amounts before and after the expansion of the menu options, relative to the change of the default rate for never-selected and always-selected amounts, which were not affected by the menu expansion. We include granular month of origination $ $t \times {\rm risk}$ $ category $ $j$ $ fixed effects, $ $\delta_{i}^{jt}$ $, which ensure we compare borrowers who took a loan on the same month with the same contract terms and with similar observed measures of credit risk (same risk category). We also include a vector of control variables observable at origination, $ $X_{i}$ $. In our baseline specification, $ $X_{i}$ $ includes four-point FICO score at-origination bin and state fixed effects. The rich set of fixed effects ensures that we perform the difference-in-differences estimation by comparing borrowers who are observationally equivalent. We also report results including as controls the full set of variables that LC reports and that investors observe at origination. These variables (61 in total) include annual income, a dummy for home ownership, stated purpose of the loan, length of employment, length of credit history, total debt balance excluding mortgage, revolving balance, and monthly debt payments to income, among others. Table 3, Columns 1 and 2, reports results of regression (2). The negative point estimate for $ $\gamma$ $ indicates that borrowers who take a 36-month loan once a 60-month option is available are significantly less likely to default than borrowers who take the same 36-month loan when the long-term option is not available. The point estimate of $ $-$ $0.0081 means that the default rate of the borrowers that are selected on maturity is 0.8 percentage points lower than the default rate of the nonselected borrowers (Column 1), and the magnitude is unchanged when we include as additional controls every single variable observable at origination in LC’s data set (Column 2). That our estimate is virtually unaffected by including this full suite of additional controls demonstrates that the granular fixed effects structure in our baseline regression is sufficiently comprehensive to absorb any changes in the composition of observed borrower characteristics. Table 3 Regression results: Screening with maturity (1) (2) $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $D_{i,t}$ $ –0.0081** –0.0080** (0.004) (0.004) Sample MAIN MAIN Observations 60,511 57,263 $ $R^{2}$ $ 0.035 0.047 # clusters 45 45 (1) (2) $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $D_{i,t}$ $ –0.0081** –0.0080** (0.004) (0.004) Sample MAIN MAIN Observations 60,511 57,263 $ $R^{2}$ $ 0.035 0.047 # clusters 45 45 This table shows that the default rate of borrowers who selected into a short-term loan when they have taken take a long-term loan is higher than for borrowers who could not take a long-term loan. The table shows the output of the regression of each outcome on a dummy for the staggered reduction of the minimum amount threshold for long-maturity loans in March 2013 (to $ ${\$}$ $12,000) and July 2013 (to $ ${\$}$ $10,000). The outcome is $ $Default$ $, a dummy that equals one if a borrower is late by more than 120 days, measured as of April 2015. The sample corresponds to loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $20,000 with a list date between December 2012 and October 2013. All regressions include risk category $ $\times$ $month and four-point FICO score bin, state, and $ ${\$}$ $1,000 loan amount bin fixed effects. Column 2 includes all borrower-level variables observed by investors at the time of origination as controls. Standard errors are clustered at the state level. *, **, and *** represent significance at the 10%, 5%, and 1% level, respectively. Table 3 Regression results: Screening with maturity (1) (2) $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $D_{i,t}$ $ –0.0081** –0.0080** (0.004) (0.004) Sample MAIN MAIN Observations 60,511 57,263 $ $R^{2}$ $ 0.035 0.047 # clusters 45 45 (1) (2) $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $D_{i,t}$ $ –0.0081** –0.0080** (0.004) (0.004) Sample MAIN MAIN Observations 60,511 57,263 $ $R^{2}$ $ 0.035 0.047 # clusters 45 45 This table shows that the default rate of borrowers who selected into a short-term loan when they have taken take a long-term loan is higher than for borrowers who could not take a long-term loan. The table shows the output of the regression of each outcome on a dummy for the staggered reduction of the minimum amount threshold for long-maturity loans in March 2013 (to $ ${\$}$ $12,000) and July 2013 (to $ ${\$}$ $10,000). The outcome is $ $Default$ $, a dummy that equals one if a borrower is late by more than 120 days, measured as of April 2015. The sample corresponds to loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $20,000 with a list date between December 2012 and October 2013. All regressions include risk category $ $\times$ $month and four-point FICO score bin, state, and $ ${\$}$ $1,000 loan amount bin fixed effects. Column 2 includes all borrower-level variables observed by investors at the time of origination as controls. Standard errors are clustered at the state level. *, **, and *** represent significance at the 10%, 5%, and 1% level, respectively. This decline in the default probability is due to the borrowers who self-select into the long-term loan, which we estimated to be 14.5% of the borrowers in the nonselected sample (Table 2, Column 1). Combining the two results allows us to obtain an estimate of the default probability of the borrowers that self-selected into the 60-month loan: it is $ $^{0.8\%}/{14.5\%}=5.5\%$ $ higher than for those who self-select into the 36-month loan when the long-term loan is available (significant at a 10% level, based on bootstrapping with 1,000 repetitions). This is an estimate of the counterfactual probability we are after: it is the default rate that borrowers who self-selected into the 60-month loan would have had if they had taken the 36-month loan. The economic magnitude of this difference is large compared with the average default rate of 36-month loans issued before the menu expansion, 9.2% (Table 1). The comparison implies that among observationally equivalent borrowers, those who self-select into a long-maturity contract are 59% more likely to default than those borrowers who self-select into the short-term contract, ceteris paribus (e.g., holding constant the contract characteristics). The results suggest that maturity choice reveals unobserved heterogeneity among borrowers. The lower default rate of borrowers who self-select into a short-maturity loan cannot be predicted by variables available to investors at the time of origination, as attested by the comparison between the estimates with and without controls for observables. Although we do not control for the exact FICO score but for scores within each four-point FICO bin, the predictive power of FICO on default in our sample is too small for selection within four-point FICO bins to account for our results. Indeed, a regression of $ $Default_{i}$ $ on the high end of the FICO four-point range at origination, including risk category by $ ${\$}$ $1,000 loan amount bin by month fixed effects, gives a coefficient of $ $-$ $0.0000362. That is, a one-point increase in FICO score at origination is correlated with a 0.004% decline in default rate, not statistically significant. Thus, variation in default rates within FICO score bins can at most account for a 0.012% difference in default rates (0.004% $ $\times$ $3), quantitatively irrelevant next to our estimated effect of 0.8% reduction in default. 2.4 The APR premium for 60-month loans For maturity to operate as a screening device, it must be that unobservably high-risk borrowers are self-selecting into loans with a higher APR. We estimate the difference in the long- and short-term loan by running a regression of $ $APR$ $ on $ $Long$ $, a dummy that equals one for long-term loans, controlling by credit-risk grade by month by $ ${\$}$ $1,000 loan amount by four-point FICO range fixed effects. As Internet Appendix Table C1 shows, LC charges a 3.3% higher APR for 60-month loans, holding all borrower and loan characteristics constant.22 Importantly, this APR differential cannot be explained by the upward sloping yield curve during our sample period. In Internet Appendix Section B we show that the upward sloping Treasury yield curve on March 1, 2013, would imply an APR premium between the 36- and 60-month loan of 0.2 percentage points. This implies that 3.1% of the 3.3% differential in our sample (94%) cannot be explained by the yield curve. Thus, borrowers who selected into the 60-month loan option were required to pay a premium for the insurance provided by this contract, consistent with any screening mechanism. Further, the APR difference between short- and long-maturity loans can be more than fully accounted for by the 5.5% higher expected default rate of those borrowers who self-select into the long-maturity option.23 2.5 Identification tests 2.5.1 Evidence to support the identifying assumption Our empirical strategy rests on the identifying assumption that there were no changes in unobserved borrower creditworthiness that differentially affected borrowers taking loans between $ ${\$}$ $12,000 and $ ${\$}$ $16,000 in March and borrowers taking loans between $ ${\$}$ $10,000 and $ ${\$}$ $12,000 in July. One potential threat to this assumption is the possibility that there is a gradual shift in the composition of borrowers over 2013 that approximately matched the pattern of the staggered expansion. We test for this possibility by running an amended version of (1) using a series of dummies that become active $ $\tau$ $ months after a 60-month loan is offered at each amount. Formally, we define: \[D\left(\tau\right)_{kt}=\begin{cases}1 & \mathit{if}\ \$16,000>\mathit{Loan}\ \mathit{Amount}\geq \$12,000\;\&\;t= \mathit{March}\ 2013+\tau\\1 & \mathit{if}\ \$12,000>\mathit{Loan}\ \mathit{Amount}\geq \$10,000\;\&\;t= \mathit{July}\ 2013+\tau\\0 & \,\mathit{otherwise}\end{cases},\] and we run the following regression:24 \begin{equation}log(N_{jkt})=\beta_{k}+\delta_{jt}+\sum_{\tau=-3}^{3}\gamma_{\tau}\times D\left(\tau\right)_{kt}+\epsilon_{jkt}.\label{eq:pretrends_regression-1}\end{equation} (3) Figure 6 shows the results of regression 3. The results show no differential pre-trends in the 3 months leading up to the expansion and then show a discontinuous fall in the number of loans made in these amounts exactly at the time of the expansion. This rules out that our results are coming from pre-existing trends in borrower demand or composition unrelated to the menu expansion. Figure 6 View largeDownload slide Pre-trends on number of loans originated This figure shows the regression coefficients ($ $\gamma_{\tau}$ $) and 90% confidence interval for the following regression: \[log(N_{j,t,amount1000})=\beta_{amount1000}+\delta_{j,t}+\sum_{\tau=-3}^{3}\gamma_{\tau}\times D\left(\tau\right)_{amount1000,t}+\epsilon_{i,t},\] which measures the difference in the number of loans issued between eventually selected and control amounts $ $\tau$ $ months after the threshold expansion. Standard errors are robust to heteroscedasticity. Figure 6 View largeDownload slide Pre-trends on number of loans originated This figure shows the regression coefficients ($ $\gamma_{\tau}$ $) and 90% confidence interval for the following regression: \[log(N_{j,t,amount1000})=\beta_{amount1000}+\delta_{j,t}+\sum_{\tau=-3}^{3}\gamma_{\tau}\times D\left(\tau\right)_{amount1000,t}+\epsilon_{i,t},\] which measures the difference in the number of loans issued between eventually selected and control amounts $ $\tau$ $ months after the threshold expansion. Standard errors are robust to heteroscedasticity. To further ensure that our results are not driven by differential trends in the demand for loans of varying amounts, we run regression (1) on a sample shifted forward to start when the 60-month loan option is available for any amount above $ ${\$}$ $10,000 (after the expansion in menus is complete). That is, we shift the definition of $ $D_{kt}$ $ forward by 7 months and run the regression on the sample of loans originated between July 2013 and May 2014. Column 5 of Table 2 shows the results. The coefficient on $ $D_{kt}$ $ equals $ $-$ $4.4% and is insignificant, and given the confidence interval we can reject the null that this coefficient equals our main estimate. 2.5.2 Simultaneous choice of maturity and loan amount A second identifying assumption behind our empirical approach is that the choice of loan amount is sufficiently inelastic to loan maturity. If this is not the case, the difference-in-differences estimate will be biased toward zero. This could be the case either because borrowers in what we classify as eventually selected amounts may be already selected on maturity before the menu expansion (e.g., if some borrowers who wanted to take a long-term loan at a treated amount before the expansion took a long-maturity loan at larger amount instead) or because borrowers in what we classify as never-selected amounts may be a selected group after the menu expansion (e.g., because some borrowers who wanted to take a long-term loan at a control amount after the menu expansion took a long-maturity loan at a treated amount instead). We first consider the possibility that eventually selected amounts are selected before the menu expansion. As an example, consider borrowers who would like to take a $ ${\$}$ $10,000 60-month loan, which is not available before the menu expansion. The closest feasible alternatives are (1) a $ ${\$}$ $10,000 36-month loan and (2) a $ ${\$}$ $16,000 60-month loan.25 Our empirical strategy will estimate the effect of maturity on selection if borrowers choose the first option; for example, they take a loan for the amount they prefer at a shorter maturity—36-months—when the 60-month option is not available. The reason is that these borrowers select out of the 36-month loan when the 60-month option is available, after the menu expansion.26 If, on the contrary, borrowers choose the second option, for example, they take a 60-month maturity loan but for a larger amount, then our difference-in-differences estimate will be zero. Indeed, these borrowers will not be in the eventually selected group of loans before or after the expansion because our estimation is exclusively based on the outcomes of 36-month loans. Thus, selection from one long-term loan to another will not affect our estimates.27 Figure 5 View largeDownload slide Pre-period loan amount histogram This top panel shows the number of 36-month loans issued by LC by loan amount in $ ${\$}$ $25 increments between $ ${\$}$ $5,000 and $ ${\$}$ $25,000 between December 2012 and February 2013. The bottom panel shows the same histogram for the same period of time for 60-month loans. Figure 5 View largeDownload slide Pre-period loan amount histogram This top panel shows the number of 36-month loans issued by LC by loan amount in $ ${\$}$ $25 increments between $ ${\$}$ $5,000 and $ ${\$}$ $25,000 between December 2012 and February 2013. The bottom panel shows the same histogram for the same period of time for 60-month loans. Now consider the second case, where the never-selected amounts are treated after the menu expansion. Take for example borrowers who would like to take a $ ${\$}$ $5,000 60-month loan, but since this option is not available before the menu expansion, they take a $ ${\$}$ $5,000 36-month loan instead. Although these borrowers are in the control group in our estimation, it is possible that they choose a $ ${\$}$ $10,000 60-month loan when this option becomes available in the menu. If this is the case, then the menu expansion will also cause self-selection into long maturity among the control group of loans, and the comparison between eventually selected and control loans will be biased toward zero. We investigate formally whether eventually selected loans amounts were affected prior to the expansion or if control loan amounts were impacted after the expansion. To do this we exploit the same setup as regression 1, which measures the change in the number of short-term loans issued at eventually selected and control amounts, before and after the menu expansion, and compare the evolution of the number of 60-month (36-month) loans in the $ ${\$}$ $16,000 to $ ${\$}$ $20,000 ($ ${\$}$ $5,000 and $ ${\$}$ $10,000) range relative to the evolution of the number of loans in the $ ${\$}$ $20,000 to $ ${\$}$ $24,000 ($ ${\$}$ $1,000 and $ ${\$}$ $5,000) range around the menu expansion. We estimate the same difference-in-differences regressions with a modified definition of the “selected” dummy $ $D_{kt}$ $ to equal 1 one after March 2013 or July 2013, for different loan amounts according to the timing of the menu expansion, as explained below. First, using the subsample of 60-month loans for amounts between $ ${\$}$ $16,000 and $ ${\$}$ $24,000, we define $ $D_{kt}$ $ to be equal to one after March 2013 for all amounts between $ ${\$}$ $16,000 and $ ${\$}$ $20,000. The coefficient on this dummy tells us whether the number of loans of amounts close to the $ ${\$}$ $16,000 expansion threshold declined relative to those further from the threshold. If so, it would be an indication that eventually selected loan amounts experienced selection to 60-month loans prior to the expansion. The coefficient on the interaction term is $ $-$ $8.25% and is not significantly different from zero (Table 2, Column 2). This suggests weak evidence that our estimates may understate the degree of selection because some borrowers in eventually selected amounts may have opted for 60-month loans above $ ${\$}$ $16,000 prior to the expansion.28 We repeat the exercise at the $ ${\$}$ $10,000 amount threshold using 36-month loans. We restrict the analysis to the sample of loan amounts between $ ${\$}$ $1,000 and $ ${\$}$ $10,000, and define $ $D_{kt}$ $ equal to one after July 2013 for amounts between $ ${\$}$ $5,000 and $ ${\$}$ $10,000 and zero otherwise. The coefficient on the interaction term is $ $-$ $3.6% and, again, not significantly different from zero (Column 4). Thus, there is no evidence that borrowers who in the pre-period selected a short-maturity loan below $ ${\$}$ $10,000 would have taken a larger long-maturity loan above the $ ${\$}$ $10,000 threshold when they became available in July. In other words, we find no evidence that the control group of loans in our main empirical design were affected by the menu expansion. Taken together the results in Table 2 confirm our conjecture that the bulk of any selection to longer-maturity loans induced by the expansion of the menu was in the eventually selected amounts.29 2.6 Robustness We present in Table 4 several tests that demonstrate the robustness of our results. First, Column 1 of Table 4 presents a counterpart to our main result in Column 1 of Table 2 but limiting the sample to loan amounts between $ ${\$}$ $6,000 and $ ${\$}$ $19,000 (a $ ${\$}$ $1,000 narrower window than our main sample, which uses loans from $ ${\$}$ $5,000 to $ ${\$}$ $20,000). The results are qualitatively similar, although the estimate is noisier and significant only at a 10% level. Table 4 Robustness (1) (2) (3) $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $D_{i,t}$ $ –0.0066* 0.0018 –0.0106 (0.004) (0.010) (0.007) Sample 6k–19k 36m, 16k–24k 36m, 1k–10k Observations 54,689 14,652 33,493 $ $R^{2}$ $ 0.037 0.061 0.035 # clusters 45 45 46 (1) (2) (3) $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $D_{i,t}$ $ –0.0066* 0.0018 –0.0106 (0.004) (0.010) (0.007) Sample 6k–19k 36m, 16k–24k 36m, 1k–10k Observations 54,689 14,652 33,493 $ $R^{2}$ $ 0.037 0.061 0.035 # clusters 45 45 46 The table shows the output of several robustness tests. Column 1 replicates Column 1 in Table 3 on a sample of loans listed between December 2012 and October 2013 and issued for amounts between $ ${\$}$ $6,000 and $ ${\$}$ $19,000 (an interval $ ${\$}$ $1,000 narrower than that in main sample). Columns 2 and 3 report the output for regressions run on a sample of loans listed between December 2012 and October 2013 for different loan amounts, where the independent variable is defined in an ad hoc manner using $ $\mathit{default}$ $ as outcome. Column 2 restricts the sample to 36-month loans issued in the main sample period for amounts between $ ${\$}$ $16,000 and $ ${\$}$ $24,000, and $ $D_{i,t}$ $ is equal to one for loan amounts between $ ${\$}$ $16,000 and $ ${\$}$ $20,000 listed on or after March 2013, and zero otherwise. Column 3 restricts the sample to 36-month loans issued in the main sample period for amounts between $ ${\$}$ $1,000 and $ ${\$}$ $10,000; $ $D_{i,t}$ $ is equal to one for loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $10,000 listed in or after July 2013, and zero otherwise. All regressions include risk category$ $\times$ $month and four-point FICO score bin, state, and $ ${\$}$ $1,000 loan amount bin fixed effects. Standard errors are clustered at the state level. *, **, and *** represent significance at the 10%, 5%, and 1% level, respectively. Table 4 Robustness (1) (2) (3) $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $D_{i,t}$ $ –0.0066* 0.0018 –0.0106 (0.004) (0.010) (0.007) Sample 6k–19k 36m, 16k–24k 36m, 1k–10k Observations 54,689 14,652 33,493 $ $R^{2}$ $ 0.037 0.061 0.035 # clusters 45 45 46 (1) (2) (3) $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $\mathit{Default}$ $ $ $D_{i,t}$ $ –0.0066* 0.0018 –0.0106 (0.004) (0.010) (0.007) Sample 6k–19k 36m, 16k–24k 36m, 1k–10k Observations 54,689 14,652 33,493 $ $R^{2}$ $ 0.037 0.061 0.035 # clusters 45 45 46 The table shows the output of several robustness tests. Column 1 replicates Column 1 in Table 3 on a sample of loans listed between December 2012 and October 2013 and issued for amounts between $ ${\$}$ $6,000 and $ ${\$}$ $19,000 (an interval $ ${\$}$ $1,000 narrower than that in main sample). Columns 2 and 3 report the output for regressions run on a sample of loans listed between December 2012 and October 2013 for different loan amounts, where the independent variable is defined in an ad hoc manner using $ $\mathit{default}$ $ as outcome. Column 2 restricts the sample to 36-month loans issued in the main sample period for amounts between $ ${\$}$ $16,000 and $ ${\$}$ $24,000, and $ $D_{i,t}$ $ is equal to one for loan amounts between $ ${\$}$ $16,000 and $ ${\$}$ $20,000 listed on or after March 2013, and zero otherwise. Column 3 restricts the sample to 36-month loans issued in the main sample period for amounts between $ ${\$}$ $1,000 and $ ${\$}$ $10,000; $ $D_{i,t}$ $ is equal to one for loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $10,000 listed in or after July 2013, and zero otherwise. All regressions include risk category$ $\times$ $month and four-point FICO score bin, state, and $ ${\$}$ $1,000 loan amount bin fixed effects. Standard errors are clustered at the state level. *, **, and *** represent significance at the 10%, 5%, and 1% level, respectively. As we mention above when describing our empirical strategy, the expansion in the menu of borrowing options may have induced selection in the unaffected or control group of amounts, above and below the $ ${\$}$ $10,000 to $ ${\$}$ $16,000 interval. In Table 2 above we show that the number of loans issued at the control amounts did not change, which suggests that no such selection occurred. However, it is important to independently verify that there is no change in the credit quality of loans issued at control amounts induced by the menu expansion. Here, we test for this possibility. Column 3 of Table 4 restricts the sample to loans issued between December 2012 and October 2013, between $ ${\$}$ $16,000 and $ ${\$}$ $24,000. The independent variable of interest equals one for loans between $ ${\$}$ $16,000 and $ ${\$}$ $20,000 after March 2013. The coefficient is positive and insignificant. Column 4 of Table 4 repeats the exercise for loans between $ ${\$}$ $1,000 and $ ${\$}$ $10,000 issued between December 2012 and October 2013. Here, the independent variable of interest equals one for loans between $ ${\$}$ $5,000 and $ ${\$}$ $10,000 issued after July 2013, and the coefficient is negative and insignificant. Thus, we find no significant differences in the default rate of loans issued at amounts bordering the interval of eventually selected amounts in both cases. The results in Columns 2 and 3 of Table 4 also serve as placebo tests and confirm that our results are not spuriously driven by shifting creditworthiness at different loan amounts. Overall, these tests point to a robust conclusion: borrowers who self-select into long-maturity loans are unobservably more likely to default, holding the loan contract characteristics constant. 3. Interpretation: Private Information About What? 3.1 Time structure of private information So far, our empirical results show that borrowers who select into a longer-maturity loan with a higher rate are privately informed about their increased propensity to default on a short-term loan with a lower rate, and that, therefore, maturity is an effective screening device. We turn to understanding what is the specific private information that borrowers are selecting on. Since maturity provides insurance against future changes in the price of credit, then the private information must relate to how borrowers value this protection. It is theoretically possible that borrowers who are privately informed about their own high degree of risk aversion select into longer maturity loans (De Meza and Webb 2001). If more risk averse individuals also default less (e.g., because they endogenously select less risky income streams), selection on risk aversion is inconsistent with the higher default rate that these borrowers exhibit. It follows that borrowers who select a long-maturity loan privately place a higher value on the insurance it provides either because (1) they are less likely to have sufficient funds to make loan repayments in the future (e.g., they face a higher the probability of job loss or illness or have less discretionary income), or (2) they have a higher exposure to rollover risk due to privately observed differences in the timing of their income (the cash flow timing hypothesis). These two explanations differ in the horizon after origination at which borrowers become risky. Borrowers who self-select into long maturity because they are more exposed to shocks to their observable ability to repay will tend to default more the longer the horizon after origination, as the negative shocks are realized. In contrast, borrowers who self-select into long maturity because their income arrives later will tend to default less with time after origination, as their income is realized. Therefore, the two selection mechanisms can be distinguished by their predictions of the time structure of default implied by the private information. We exploit the fact that we observe when a borrower in our sample enters default to differentiate between these two accounts. To do this, we redefine our baseline measure of default and create two variables for default at different horizons: borrowers who missed their first payment within the first 12 and 24 months of loan origination (for loans that are 120 days past due in April 2015). We label these variables $ $Default12m$ $ and $ $Default24m$ $, respectively, and use them as dependent variables in regressions that are otherwise identical to the one we estimated in Column 1 of Table 3. The results are presented in Columns 1 and 2 of Table 5. Column 1 shows that borrowers who self-select into long-term loans have no differential propensity to default within the first year of the loan. Since the hazard rate of default in our sample peaks at 13 months (Figure 2), this result is not mechanically driven by lack of statistical power due to a low frequency of default early in the life of the average loan (unconditionally, loans are as likely to default in the first 12 months after origination than later). Column 2 shows that the differential propensity to default is present at the 24-month horizon from origination. Table 5 Interpretation of results (1) (2) (3) (4) (5) (6) $ $\mathit{Default}12m$ $ $ $\mathit{Default}24m$ $ $ $\mathit{FICO}$ $ $ $\mathit{FICO}$ $ $ $\mathit{SD(FICO)}$ $ $ $\mathit{Prepayment}$ $ $ $D_{i,t}$ $ –0.0039 –0.0082* 2.7464** 2.6705** –0.5764** –0.0063 (0.003) (0.004) (1.032) (0.999) (0.266) (0.008) Sample Main Main Main Main Main Main Obs 60,511 60,511 60,511 57,263 60,511 60,511 $ $R^{2}$ $ 0.024 0.032 0.192 0.215 0.027 0.023 Clusters 45 45 45 45 45 45 (1) (2) (3) (4) (5) (6) $ $\mathit{Default}12m$ $ $ $\mathit{Default}24m$ $ $ $\mathit{FICO}$ $ $ $\mathit{FICO}$ $ $ $\mathit{SD(FICO)}$ $ $ $\mathit{Prepayment}$ $ $ $D_{i,t}$ $ –0.0039 –0.0082* 2.7464** 2.6705** –0.5764** –0.0063 (0.003) (0.004) (1.032) (0.999) (0.266) (0.008) Sample Main Main Main Main Main Main Obs 60,511 60,511 60,511 57,263 60,511 60,511 $ $R^{2}$ $ 0.024 0.032 0.192 0.215 0.027 0.023 Clusters 45 45 45 45 45 45 This table shows the output of the regression of each outcome on a dummy for the staggered reduction of the minimum amount threshold for long-maturity loans in March 2013 (to $ ${\$}$ $12,000) and July 2013 (to $ ${\$}$ $10,000). Outcomes include $ $\mathit{Default}12m$ $ and $ $\mathit{Default}24m$ $, dummies that equal one if a borrower is late by more than 120 days as of April 2015 and whose last payment occurred within 12 and 24 months after origination, respectively. FICO is the (the high end of the four-point bin) FICO score measured as of April 2015; SD(FICO) is the time-series standard deviation of (the high end of the four-point bin) FICO scores within an individual, using four observations per individual at origination, as of August 2014, December 2014, and April 2015; and $ $\mathit{Prepayment}$ $, a dummy that equals one if the loan is fully paid as of April 2015. In Column 4 we include all variables that are observable by investors at origination as controls, like in Column 2 of Table 3. The sample corresponds to loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $20,000 whose listing date is between December 2012 and October 2013. All regressions include risk category$ $\times$ $month and four-point FICO score bin, state, and $ ${\$}$ $1,000 loan amount bin fixed effects. Standard errors are clustered at the state level. *, **, and *** represent significance at the 10%, 5%, and 1% level, respectively. Table 5 Interpretation of results (1) (2) (3) (4) (5) (6) $ $\mathit{Default}12m$ $ $ $\mathit{Default}24m$ $ $ $\mathit{FICO}$ $ $ $\mathit{FICO}$ $ $ $\mathit{SD(FICO)}$ $ $ $\mathit{Prepayment}$ $ $ $D_{i,t}$ $ –0.0039 –0.0082* 2.7464** 2.6705** –0.5764** –0.0063 (0.003) (0.004) (1.032) (0.999) (0.266) (0.008) Sample Main Main Main Main Main Main Obs 60,511 60,511 60,511 57,263 60,511 60,511 $ $R^{2}$ $ 0.024 0.032 0.192 0.215 0.027 0.023 Clusters 45 45 45 45 45 45 (1) (2) (3) (4) (5) (6) $ $\mathit{Default}12m$ $ $ $\mathit{Default}24m$ $ $ $\mathit{FICO}$ $ $ $\mathit{FICO}$ $ $ $\mathit{SD(FICO)}$ $ $ $\mathit{Prepayment}$ $ $ $D_{i,t}$ $ –0.0039 –0.0082* 2.7464** 2.6705** –0.5764** –0.0063 (0.003) (0.004) (1.032) (0.999) (0.266) (0.008) Sample Main Main Main Main Main Main Obs 60,511 60,511 60,511 57,263 60,511 60,511 $ $R^{2}$ $ 0.024 0.032 0.192 0.215 0.027 0.023 Clusters 45 45 45 45 45 45 This table shows the output of the regression of each outcome on a dummy for the staggered reduction of the minimum amount threshold for long-maturity loans in March 2013 (to $ ${\$}$ $12,000) and July 2013 (to $ ${\$}$ $10,000). Outcomes include $ $\mathit{Default}12m$ $ and $ $\mathit{Default}24m$ $, dummies that equal one if a borrower is late by more than 120 days as of April 2015 and whose last payment occurred within 12 and 24 months after origination, respectively. FICO is the (the high end of the four-point bin) FICO score measured as of April 2015; SD(FICO) is the time-series standard deviation of (the high end of the four-point bin) FICO scores within an individual, using four observations per individual at origination, as of August 2014, December 2014, and April 2015; and $ $\mathit{Prepayment}$ $, a dummy that equals one if the loan is fully paid as of April 2015. In Column 4 we include all variables that are observable by investors at origination as controls, like in Column 2 of Table 3. The sample corresponds to loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $20,000 whose listing date is between December 2012 and October 2013. All regressions include risk category$ $\times$ $month and four-point FICO score bin, state, and $ ${\$}$ $1,000 loan amount bin fixed effects. Standard errors are clustered at the state level. *, **, and *** represent significance at the 10%, 5%, and 1% level, respectively. We present in Figure 7 the coefficients from estimating our main specification using as the dependent variable an indicator for whether the first missed payment occurred before 1, 2, and so on, up to 24 months after origination.30 The figure indicates that the cumulative default probability differential between the two groups of borrowers increases linearly with the months after origination. That is, borrowers who select into the 60-month loan have a propensity to default on the 36-month loan that is increasing in the time since origination of their loan.31 This evidence indicates that the source of private information that is driving maturity selection is borrowers’ exposure to shocks to their own future observable creditworthiness. Note that Figure 2 demonstrates that the hazard rate of default for 36-month loans peaks at 16 months.32 This indicates that the bulk of default at either maturity occurs well before the 24-month horizon possible in our analysis, thereby ruling out the concern that our results are too near to origination to account for default behavior for either type of loan. Figure 7 View largeDownload slide Default rate coefficient by number of months since origination This figure shows the estimated coefficient and 90% confidence interval for the following regression: \[default\left(\Delta t\right)=\beta_{amount1000}+\delta_{j,\overline{FICO},t}+\gamma\times D_{amount100,t}+X_{i,t}+\epsilon_{i},\] where the outcome is $ $default\left(\Delta t\right)$ $, a dummy that equals one if a loan is late by more than 120 days as of April 2015 and if the last payment on these loan occurred $ $\Delta t$ $ months after origination, on $ $D_{amount1000,t}$ $, a dummy that captures the staggered expansion of the 60-month loans for amounts above $ ${\$}$ $12,000 and $ ${\$}$ $10,000 in March and July 2013, respectively. Standard errors are clustered at the state level. Sample includes loans issued between December 2012 and October 2013 for loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $20,000. Figure 7 View largeDownload slide Default rate coefficient by number of months since origination This figure shows the estimated coefficient and 90% confidence interval for the following regression: \[default\left(\Delta t\right)=\beta_{amount1000}+\delta_{j,\overline{FICO},t}+\gamma\times D_{amount100,t}+X_{i,t}+\epsilon_{i},\] where the outcome is $ $default\left(\Delta t\right)$ $, a dummy that equals one if a loan is late by more than 120 days as of April 2015 and if the last payment on these loan occurred $ $\Delta t$ $ months after origination, on $ $D_{amount1000,t}$ $, a dummy that captures the staggered expansion of the 60-month loans for amounts above $ ${\$}$ $12,000 and $ ${\$}$ $10,000 in March and July 2013, respectively. Standard errors are clustered at the state level. Sample includes loans issued between December 2012 and October 2013 for loan amounts between $ ${\$}$ $5,000 and $ ${\$}$ $20,000. 3.2 Private information about future observable creditworthiness We provide additional evidence in support of our preferred interpretation. Specifically, we observe the realized observable creditworthiness measured by each borrower’s credit score (FICO score) as of April 2015, roughly 2 years after origination. In Table 5, Columns 3 and 4, we run our main regression model but replace the main outcome $ $Default_{i}$ $ with $ $FICO_{i}$ $, the borrower’s FICO score as reported in the latest LC data pull. In Column 4 we include as controls all variables that are observable by investors at origination, like in Column 2 of Table 3. The results imply a statistically significant increase of future FICO scores of approximately 2.7 points among selected short-term borrowers relative to unselected ones. In economic terms this means that the average future FICO score of the 14.5% of borrowers who self-select into the long-maturity loans is $ $^{2.7}/{14.5\%}=18.6$ $ points lower than for the average borrower who selects the 36-month loan. To further demonstrate that borrowers are selecting maturity based on private information about their exposure to shocks to their future observable creditworthiness, we use the volatility of a borrower’s future credit rating as a measure of the reclassification risk she is exposed to. If borrowers have private information about this reclassification risk, we expect borrowers who self-select into the 36-month loan to have less volatile future FICO scores. To test this hypothesis we present in Column 5 of Table 5 the results of estimating our main specification using the within-individual standard deviation of the FICO score as the outcome variable, using FICO scores obtained from 4 different pulls of the LC loan performance data: at origination, as of August 2014, as of December 2014, and as of April 2015, which is the same outcome variable used in Table 2.33 The cross-sectional average and standard deviation of this measure for loans in our sample that were issued in the three pre-period months are 24.5 and 19.1, respectively. The point estimate in Column 5 of Table 5 is $ $-$ $0.57 and statistically significant at the 5% level. This implies that borrowers who select the 36-month loan have a future FICO score that is 2.3% (equal to 0.57/24.5) less volatile when the 60-month loan is available than when it is not. This pattern is strongly consistent with the insurance rationale for the screening mechanism: borrowers who select long-maturity loans are (unobservably) more exposed to reclassification risk. Note that $ $FICO_{i}$ $ measures borrowers’ repayment status on all of their debts. In particular, it considers a borrower’s performance not only on the 36-month loan with LC, but on loans of different maturities as well. Thus it is unlikely that this result is driven by the incompatibility between the short-term LC loan and the time profile of borrower’s future income. Instead, this directly shows that borrowers who select long-maturity loans have private information that directly relates to shocks to their observed creditworthiness and the impact that this will have on the price at which future lending will occur. Finally, we study how maturity choice relates to a borrower’s unobserved propensity to prepay her loan prior to maturity. The LC data record loans that have been fully prepaid as of April 2015, which we code in $ $Prepayment$ $, a dummy variable. If borrowers select maturity based on private information about the timing of their income, we would expect that those borrowers who select into a short-term loan would prepay at a higher rate than borrowers in an unselected group. If this were the case, the main coefficient in regression model (2) where we replace the outcome variable $ $Default$ $ with $ $Prepayment$ $ should be positive. We document the output of this regression in Column 6 of Table 5. The point estimate is negative but insignificant (p-value is .44). Although this result is not conclusive, it does suggest that maturity choice does not seem to be driven by private information about the timing of borrowers’ income shocks. It is difficult to believe that selection based on private information about the timing of income would simultaneously generate a statistically significant reduction in default but would produce a change in loan prepayment that is statistically undetectable, when the prediction about the timing of payment is most directly tied to the hypothesis itself. On the other hand, this finding is fully consistent with the interpretation that borrowers who are privately informed of their increased exposure to shocks to their ability to repay select into long-maturity loans: positive realized shocks lead to early prepayment, while negative shocks lead to default. 3.3 60-month loan performance Further evidence about the underlying private information that is driving maturity choice can be provided by looking at the default rate of borrowers who took 60-month loans. If, as we hypothesize, these borrowers are more exposed to shocks to their ability to repay then, after controlling for observables, the default rate should be higher at the longer-maturity loans. In contrast, if borrowers are selecting to match the privately observed horizon of their income, then the default rate should be no higher. Before presenting this evidence, an important caveat that stems from our core empirical challenge is required. Our analysis has so far focused on the propensity to default holding the terms of the contract constant, that is, focusing exclusively on a sample of 36-month loans. Thus, our analysis tells us what the default probability of borrowers who self-select into 60-month loans would have been had they selected a 36-month loan. We cannot empirically identify what their default probability is for a a 60-month loan. This is because the default rate of 60-month loans is also driven by selection in the extensive margin: there are some borrowers who would have chosen not to take a loan at all in the absence of a 60-month option, but do so when it becomes available, and we cannot independently isolate the repayment propensity of these extensive margin borrowers. Notwithstanding this problem, we can provide suggestive evidence by comparing the average default rate of 36-month and 60-month loans that have the same measured expected default risk (initial risk category and four-point FICO score bin), are issued the same month, and are the same size ($ ${\$}$ $1,000 loan amount bin). The propensity to enter default by April 2015, which holds the repayment horizon equal across the two loan contracts, is 3% higher for the 60-month than for the 36-month loans. This is commensurate with the 3.1% APR risk premium for the 60-month loan that we documented in Section 2.4. This provides further evidence that selection is based on private information about exposure to shocks to creditworthiness. If, alternatively, borrowers were selecting maturity based on the time horizon of their income rather than their future creditworthiness, then we should not expect to see higher default or interest rates for the longer-maturity loan. A different but related question is whether increased maturity impacts a borrower’s propensity to repay a loan. The answer also hinges on the average ability to repay of borrowers who select 60-month loans on the extensive margin, which we cannot measure in our setting. If we make the stark assumption that their ability to repay is the same as for borrowers who are selected away from the 36-month loan, then our results suggest that 2 more years of maturity reduces the propensity to default by 2.5% over the horizon for which we observe these loans.34 If borrowers who take the 60-month loan on the extensive margin have a lower (higher) ability to repay, then this will under- (over-) state the effect. This unmeasured margin could reconcile our results with Dobbie and Song (2017), who use a randomized experiment on U.S. household credit card borrowers to show that increased maturity does not causally change a borrower’s propensity to default, or with Field et al. (2013), who find that increased maturity induces entrepreneurs to undertake risky projects and leads to higher default. 3.4 Price reaction to screening Our empirical analysis benefits from the natural experiment created by LC’s decision to expand the availability of long-term loan contracts without changing any of the characteristics or terms of the short-maturity contract. This implies that, within the window of the natural experiment, the default probability of 36-month loans between $ ${\$}$ $10,000 and $ ${\$}$ $16,000 dropped while the interest rate did not change. If LC was earning a competitive return on these loans before the menu expansion, then it must have been earning rents after the expansion. In theory, competitive pressures should eventually drive the interest rate on the short-maturity loan down to reflect the lower risk of the borrowers that self-select into short maturity. Indeed, after our analysis sample period (during which all lending terms were held constant), LC adjusted the APR of the 36-month loan in a way that is consistent with this conjecture. We show this in Figure 8 which plots the average APR charged to borrowers on 36-month loans in each month controlling for loan amount and borrower characteristics.35 Consistent with our conjecture, we see that the APR fell by roughly 0.8% for short-term loans after long-term loans were added to the menu. This number is of the same order of magnitude as our estimate in Column 1 of Table 2 that showed that the expected default rate of the 36-month loans fell by 0.8% as a result of the selection into long-maturity loans. Figure 8 View largeDownload slide Reduction in APR This figure shows the time series of the residual of a regression of loan APR on $ ${\$}$ $1,000 loan amount dummies, FICO score bin dummies, annual income, and address state dummies by month of origination for 36-month loans issued between $ ${\$}$ $10,000 and $ ${\$}$ $16,000. Figure 8 View largeDownload slide Reduction in APR This figure shows the time series of the residual of a regression of loan APR on $ ${\$}$ $1,000 loan amount dummies, FICO score bin dummies, annual income, and address state dummies by month of origination for 36-month loans issued between $ ${\$}$ $10,000 and $ ${\$}$ $16,000. 4. Conclusion We document that loan terms, in particular maturity, can be used to screen borrowers based on unobserved creditworthiness in U.S. consumer credit markets. Borrowers who are unobservably more exposed to shocks to their ability to repay self-select into longer maturity loans with higher APRs. Extrapolating from the results in our paper may help understand the broader unsecured consumer credit market that platforms like LC and Prosper operate in. Relative to their main competition – credit card debt– these platforms offer loans at significantly shorter maturities, allowing them, through the mechanism we document in this paper, to skim off low-risk borrowers from the credit card market. This may explain how these platforms offer investors competitive returns while offering APRs to borrowers below that offered on credit-card debt. As these options grow in size, it is possible that they will eventually impact the credit card market, which will be left with an increasingly screened pool of high-risk borrowers. It also remains an open question, from both an empirical and a theoretical perspective, whether screening maturity is also a first-order determinant of equilibrium loan prices in consumer credit markets in which lenders may screen using other dimensions of the contract, such as collateral in mortgage markets. Providing concrete evidence of these broader implications is left for future work. We thank Sumit Agarwal, Asaf Bernstein, Emily Breza, Tony Cookson, Anthony DeFusco, Theresa Kuchler, Adair Morse, Holger Mueller, Christopher Palmer, Mitchell Petersen, Philipp Schnabl, Antoinette Schoar, Amit Seru, Felipe Severino, and Johannes Stroebel and numerous seminar and conference participants for helpful comments. We thank Siddharth Vij for outstanding research assistance. A previous version of this paper was circulated under the title “Adverse Selection on Maturity: Evidence from Online Consumer Credit.” Supplementary data can be found on The Review of Financial Studies web site. Footnotes 1 Examples of other contractual terms that have been shown in theory to have a screening role are inside ownership (Leland and Pyle 1977), managerial incentives and capital structure (Ross 1977), mortgage points (Stanton and Wallace (1998)), and prepayment penalties (Bian and Yavas 2013). 2 For the relationship between observable creditworthiness and maturity, see Barclay and Smith (1995), Guedes and Opler (1996), and Johnson (2003), and, for that between observable creditworthiness and collateral choice, see Leeth and Scott (1989), Berger and Udell (1990), Booth (1992), Degryse and Van Cayseele (2000), and Jimenez et al. (2006). For the relationship between observable proxies for the degree of private information and maturity, see Berger et al. (2005), and, for that between observable proxies for the degree of private information and collateral choice, see Berger and Udell (1995) and Berger et al. (2011). 3 For examples of this approach, see Goyal and Wang (2013) and Gopalan et al. (2014). Since contract terms are an endogenous choice (either by the borrower or the lender), controlling for contract characteristics in a regression estimation heavily relies on functional form assumptions and is likely to yield biased estimates due to reverse causality (for an example of this approach, see Kawai et al. (2014)). 4Karlan and Zinman (2009) make a parallel argument for the identification conditions to isolate empirically adverse selection on loan prices. 5 Following the logic in Rothschild and Stiglitz (1976), when borrowers have private information about the value they place on this insurance, the market for loan maturity may not be characterized by a single price at which borrowers can buy all the insurance—maturity—they require. 6 That the higher default rate and APR at the long-maturity option are lower than the 5.5 percentage points suggests that the causal effect of longer maturity is to reduce the probability of default. 7 For comparison, the monthly installments on a $ ${\$}$ $10,000 5-year 10% APR LC loan would be $ ${\$}$ $210, whereas the minimum repayment per month in a credit card with the same balance and APR would be $ ${\$}$ $93. If the credit card APR were 20%, the minimum monthly payments would be $ ${\$}$ $157, which is still lower than the monthly installments on the LC loan. 8 Figures reported in the firms’ 2014 10K reports. 9 See http://kb.lendingclub.com/borrower/articles/Borrower/What-if-my-loan-isn-t-fully-funded-when-my-listing-ends/?l=en_US &fs=RelatedArticle. 10 Figures reported in the firms’ 2014 10K reports. 11 See http://kb.lendingclub.com/borrower/articles/Borrower/What-if-my-loan-isn-t-fully-funded-when-my-listing-ends/?l=en_US &fs=RelatedArticle. 12 This allows us to estimate a measure of time-series volatility in FICO scores for each individual. 13 The exact dates correspond to loans listed as of December 4, 2012, and October 25, 2013. Even though we refer to months as the borders of the interval, all our analyses consider these two dates as the starting and end points of the sample period, respectively. We verify empirically that the interest rates of all risk category-loan quantities pairs are unchanged over this period. For example, Figure E2 in the Internet Appendix shows supply schedules (rate versus amount) before and after the expansion of the menu of borrowing options for borrowers assigned to risk categories B1 through B5: the graphs are identical. We establish the same point in general in a tractable way in Internet Appendix E by regressing the interest rate of all 36-month loans in our sample on fixed effects for loan amount by risk category. The regression yields an $ $R^{2}$ $ of 99.7%, which confirms that the pricing of each menu was constant throughout the sample period for all 25 risk categories. 14 We exclude loans with the “policy code” variable equals 2, because they have no publicly available information, and, according to the LC Data Dictionary, they are “new products not publicly available.” In robustness tests, we limit the sample to loan amounts between $ ${\$}$ $6,000 and $ ${\$}$ $19,000, a $ ${\$}$ $1,000 narrower interval. Also, in some placebo tests we shift our sample to loans issued between July 2013 and May 2014. 15 See Internet Appendix D for details on this reverse-engineering procedure. The error in matching loans to their subgrade does not vary systematically over the same period or by loan amount. 16 We also define borrowers as being in default if they are reported to be on a “payment plan.” Our results are robust to not considering these borrowers as in default. 17 The date of default is determined by the last payment date, a variable available in the LC data. 18 Note that because of the upfront origination fee, borrowers who took short-maturity loans prior to the expansion could not costlessly swap them for long maturity ones after the expansion. This ensures that the pool of borrowers who select the short-maturity loan prior to the expansion is not changed ex post by the expansion itself. 19 Indeed, we found no evidence of any change in outside borrowing options or advertising campaigns at all. 20 For evidence of search frictions in consumer credit markets, see Stango and Zinman (2015). 21 Standard errors for estimates of equation (1) are robust to heteroscedasticity, but other alternatives, for example, clustering in any dimension, are irrelevant in terms of statistical significance. For example, when clustering at the risk category level (25 clusters), the standard error of the coefficient $ $\gamma'$ $ in Column 1 of Table 2 is 0.028. 22 The distribution of the long-short spread is shown in the Internet Appendix Figure C1, where we plot the median long- and short-term APRs for loans issued between $ ${\$}$ $15,000 and $ ${\$}$ $20,000 in the post-period by subgrade (top panel) and initial four-point FICO range (bottom panel). 23 As we discuss below, the fact that the APR gap is less than 5.5% suggests that longer maturity may causally lower the expected default rate on a loan. 24 The final 60-month threshold reduction takes place in July 2013. This leaves 3 more months in our sample period up to October 2013. Similarly, the first 60-month threshold reduction occurs in March 2013. This leaves 3 months in the pre-period (from December 2012). 25 These borrowers may also choose not to borrow at all when their preferred option is not available in the menu, and take the 60-month loan when it becomes available. This extensive margin will not affect our estimates, since our results are exclusively based on the behavior of 36-month loans before and after the menu expansion. 26 One way to test for whether borrowers at control amounts are selected before the menu expansion is to look for evidence of bunching at the borders of the treated interval. The top panel in Figure 5 presents the pre-period loan amount histogram at the short maturity. The histogram suggests that borrowers choose “round” numbers like $ ${\$}$ $10,000 and $ ${\$}$ $12,000 much more frequently than other intermediate amounts. In turn, this makes it very difficult to find evidence of bunching at specific amounts. 27 The bottom panel in Figure 5 presents the pre-period loan amount histogram at the long maturity. The histogram has the same pattern as the top panel. Evidence of bunching is, again, very difficult to establish because of borrowers’ preference for round numbers. 28 In an analogous test we also check whether borrowers of 36-month loans in control amounts above the $ ${\$}$ $16,000 threshold were affected by the expansion. The coefficient on the interaction term is 5.9% and is not significantly different from zero (Table 2, Column 3), indicating that the expansion of the menu did not induce selection away from short-term loans above $ ${\$}$ $16,000. Given that long-maturity loans were always available for these amounts, this is not a surprising result. 29 In the Internet Appendix Table A1 we conduct robustness tests that mimic the results in Table 2 where we vary the dimensions along which we collapse the loan-level data and count the number of loans. These robustness tests consistently show that selection along the margins of the interval of treated amounts is not significantly different from zero, aside from one case in which it is significant at the 10% level. 30 At horizons of 19 months and longer, the sample used to run the regression is right censored because loans issued late in our sample do not have sufficient time to enter default at these horizons. This affects loans in the eventually selected and control amounts in the same way and does not affect the identification strategy. 31 The finding that information asymmetries grow with the horizon from origination is itself new and potentially important in its own right. For example, this supports the assumed time structure of information asymmetry in Milbradt and Oehmke (2014). 32 The hazard rate of default on 60-month loans issued at the same time is similarly shaped and peaks at 17 months. This indicates that repayment over the first 24 months of a loan is the crucial determinant of default at either maturity. 33 The standard deviation is calculated as $ $SD(FICO_{i})=\sqrt{\frac{1}{4}\times\sum_{t=1}^{4}\left(FICO_{i,t}-\overline{FICO_{i}}\right)^{2}}.$ $ 34 We obtain this number as 5.5%-3%=2.5%, where 5.5% is the default probability on the short-term loan for the 14.5% of borrowers who chose the long-term loan when it became available in the menu expansion, measured in Section 2, and 3% is the average excess default rate of long-term loans relative to short-term loans. 35 These characteristics are FICO score bin, annual income, and state of residence. Note that variation in APR before November 2013 in this graph is entirely accounted for by the fact that we do not control for the borrower’s initial risk category, which we cannot estimate after October 2013. This also implies that we are unable to simply compare the APR for the 36-month loan at each menu. References Adams, W., Einav, L. and Levin. J. 2009 . Liquidity constraints and imperfect information in subprime lending. American Economic Review 99 : 49 – 84 . Google Scholar CrossRef Search ADS Agarwal, S., Chomsisengphet, S. Mahoney, N. and Strobel. J. 2015 . Regulating consumer financial products: Evidence from credit cards. Quarterly Journal of Economics 130 : 111 – 64 . Google Scholar CrossRef Search ADS Baker, M., Greenwood, R. and Wurgler. J. 2003 . The maturity of debt issues and predictable variation in bond returns. Journal of Financial Economics 70 : 261 – 91 . Google Scholar CrossRef Search ADS Barclay, M. J., and Smith. C. W. 1995 . The maturity structure of corporate debt. Journal of Finance 50 : 609 – 31 . Google Scholar CrossRef Search ADS Berger, A. N., Espinosa-Vega, M. A. Frame, W. S. and Miller. N. H. 2005 . Debt maturity, risk, and asymmetric information. Journal of Finance 60 : 2895 – 923 . Google Scholar CrossRef Search ADS Berger, A. N., Espinosa-Vega, M. A. Frame, W. S. and Miller. N. H. 2011 . Why do borrowers pledge collateral? New empirical evidence on the role of asymmetric information. Journal of Financial Intermediation 20 : 55 – 70 . Google Scholar CrossRef Search ADS Berger, A. N., and Udell. G. F. 1990 . Collateral, loan quality and bank risk. Journal of Monetary Economics 25 : 21 – 42 . Google Scholar CrossRef Search ADS Berger, A. N., and Udell. G. F. 1995 . Relationship lending and lines of credit in small firm finance. Journal of Business 68 : 351 – 81 . Google Scholar CrossRef Search ADS Bester, H. 1985 . Screening vs. rationing in credit markets with imperfect information. American Economic Review 75 : 850 – 55 . Bian, X., and Yavas. A. 2013 . Prepayment penalty as a screening mechanism for default and prepayment risks. Real Estate Economics 41 : 193 – 224 . Google Scholar CrossRef Search ADS Booth, J. R. 1992 . Contract costs, bank loans, and the cross-monitoring hypothesis. Journal of Financial Economics 31 : 25 – 41 . Google Scholar CrossRef Search ADS Bosworth, B. 1971 . Patterns of corporate external financing. Brookings Papers on Economic Activity 1971 : 253 – 84 . Google Scholar CrossRef Search ADS Cochrane, J. H. 1995 . Time-consistent health insurance. Journal of Political Economy 103 : 445 – 73 . Google Scholar CrossRef Search ADS De Meza, D., and Webb. D. 2001 . Advantageous selection in insurance markets. RAND Journal of Economics 32 : 249 – 62 . Google Scholar CrossRef Search ADS Degryse, H., and Van Cayseele. P. 2000 . Relationship lending within a bank-based system: Evidence from european small business data. Journal of Financial Intermediation 9 : 90 – 109 . Google Scholar CrossRef Search ADS Diamond, D. W. 1991 . Debt maturity structure and liquidity risk. Quarterly Journal of Economics 106 : 709 – 37 . Google Scholar CrossRef Search ADS Dobbie, W., and Skiba. P. M. 2013 . Information asymmetries in consumer credit markets: Evidence from payday lending. American Economic Journal: Applied Economics 5 : 256 – 82 . Google Scholar CrossRef Search ADS Dobbie, W., and Song. J. 2017 . Targeted debt relief and the origins of financial distress: Experimental evidence from distressed credit card borrowers. Working Paper , NBER . Field, E., Pande, R. Papp, J. and Rigol. N. 2013 . Does the classic microfinance model discourage entrepreneurship among the poor? Experimental evidence from India. American Economic Review 103 : 2196 – 226 . Google Scholar CrossRef Search ADS Finkelstein, A., McGarry, K. and Sufi. A. 2005 . Dynamic inefficiencies in insurance markets: Evidence from long-term care insurance. American Economic Review 95 : 224 – 8 . Google Scholar CrossRef Search ADS PubMed Flannery, M. J. 1986 . Asymmetric information and risky debt maturity choice. Journal of Finance 41 : 19 – 37 . Google Scholar CrossRef Search ADS Gopalan, R., Song, F. and Yerramilli. V. 2014 . Debt maturity structure and credit quality. Journal of Financial and Quantitative Analysis 49 : 817 – 42 . Google Scholar CrossRef Search ADS Goyal, V. K., and Wang. W. 2013 . Debt maturity and asymmetric information: Evidence from default risk changes. Journal of Financial and Quantitative Analysis 48 : 789 – 817 . Google Scholar CrossRef Search ADS Graham, J. R., and Harvey. C. R. 2001 . The theory and practice of corporate finance: Evidence from the field. Journal of Financial Economics 60 : 187 – 243 . Google Scholar CrossRef Search ADS Greenwood, R., Hanson, S. and Stein. J. C. 2010 . A gap-filling theory of corporate debt maturity choice. Journal of Finance 65 : 993 – 1028 . Google Scholar CrossRef Search ADS Guedes, J., and Opler. T. 1996 . The determinants of the maturity of corporate debt issues. Journal of Finance 51 : 1809 – 33 . Google Scholar CrossRef Search ADS Hart, O., and Moore. J. 1994 . A theory of debt based on the inalienability of human capital. Quarterly Journal of Economics 109 : 841 – 79 . Google Scholar CrossRef Search ADS Hart, O., and Moore. J. 1995 . Debt and seniority: An analysis of the role of hard claims in constraining management. American Economic Review 85 : 567 – 85 . Holmstrom, B. 1983 . Equilibrium long-term labor contracts. Quarterly Journal of Economics 98 : 23 – 54 . Google Scholar CrossRef Search ADS Jaffee, D. M., and Russell. T. 1976 . Imperfect information, uncertainty, and credit rationing. Quarterly Journal of Economics 90 : 651 – 66 . Google Scholar CrossRef Search ADS Jimenez, G., Salas, V. and Saurina. J. 2006 . Determinants of collateral. Journal of Financial Economics 81 : 255 – 81 . Google Scholar CrossRef Search ADS Johnson, S. A. 2003 . Debt maturity and the effects of growth opportunities and liquidity risk on leverage. Review of Financial Studies 16 : 209 – 36 . Google Scholar CrossRef Search ADS Karlan, D., and Zinman. J. 2009 . Observing unobservables: Identifying information asymmetries with a consumer credit field experiment. Econometrica 77 : 1993 – 2008 . Google Scholar CrossRef Search ADS Kawai, K., Onishi, K. and Uetake. K. 2014 . Signaling in online credit markets. Working Paper . Leeth, J. D., and Scott. J. A. 1989 . The incidence of secured debt: evidence from the small business community. Journal of Financial and Quantitative Analysis 24 : 379 – 94 . Google Scholar CrossRef Search ADS Leland, H. E., and Pyle. D. H. 1977 . Informational asymmetries, financial structure, and financial intermediation. Journal of Finance 32 : 371 – 87 . Google Scholar CrossRef Search ADS Levine, C. B., and Hughes. J. S. 2005 . Management compensation and earnings-based covenants as signaling devices in credit markets. Journal of Corporate Finance 11 : 832 – 50 . Google Scholar CrossRef Search ADS Milbradt, K., and Oehmke. M. 2014 . Maturity rationing and collective short-termism. Journal of Financial Economics 118 : 553 – 70 . Google Scholar CrossRef Search ADS Morse, A. 2015 . Peer-to-peer crowdfunding: Information and the potential for disruption in consumer lending. Annual Review of Financial Economics 7 : 463 – 82 . Google Scholar CrossRef Search ADS Myers, S. C. 1977 . Determinants of corporate borrowing. Journal of Financial Economics 5 : 147 – 75 . Google Scholar CrossRef Search ADS Ross, S. A. 1977 . The determination of financial structure: The incentive-signalling approach. Bell Journal of Economics 8 (1) : 23 – 40 . Google Scholar CrossRef Search ADS Rothschild, M., and Stiglitz. J. E. 1976 . Equilibrium in competitive insurance markets: An essay on the economics of imperfect information. The Quarterly Journal of Economics 90 : 630 – 49 . Google Scholar CrossRef Search ADS Stango, V., and Zinman. J. 2015 . Borrowing high versus borrowing higher: Price dispersion and shopping behavior in the U.S. credit card market. Review of Financial Studies 29 : 979 – 1006 . Google Scholar CrossRef Search ADS Stanton, R., and Wallace. N. 1998 . Mortgage choice: What’s the point? Real Estate Economics 26 : 173 – 205 . Google Scholar CrossRef Search ADS Stiglitz, J., and Weiss. A. 1981 . Credit rationing in markets with imperfect information. American Economic Review 71 : 393 – 410 . Taggart, R. A. 1977 . A model of corporate financing decisions. Journal of Finance 32 : 1467 – 84 . Google Scholar CrossRef Search ADS © The Author(s) 2018. Published by Oxford University Press on behalf of The Society for Financial Studies. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

The Review of Financial StudiesOxford University Press

Published: Mar 8, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off