Behavioral Biases in Peer-to-Peer (P2P) Lending

Courtesy of Shahar Ayal, Daphna Bar-Haim and Moran Ofir

Peer-to-Peer (P2P) lending refers to online marketplaces where lenders lend to individuals or small businesses. This rapidly expanding new source of credit eliminates the traditional financial intermediary and reduces financial exclusion by allowing more people to borrow and lend. At the same time, the replacement of financial intermediaries by individuals poses new challenges, since both the borrowers and the lenders are human beings who are prone to a variety of behavioral biases.

In our paper, Behavioral Biases in Peer-to-Peer Lending, we survey the growing literature on P2P lending and identify typical cognitive biases that may affect borrowers and lenders’ financial decisions. Specifically, we review two core biases studied in the traditional behavioral literature and discuss their implementation in the context of P2P lending; these are Familiarity Bias as well as Stereotypes and Representativeness.  Then, as a case in point, we focus on Debt Account Aversion, which describes an individual’s tendency to consistently pay off small debts first to reduce the nominal number of debts, even though they also have larger debts with higher interest rates. We conduct an online experiment to measure DAA in the context of P2P lending.

The Peer-to-Peer Lending Market

The 2008 financial crisis shattered the public’s confidence in the regulated banking system. This was due in part to the realization that the financial system cannot survive the failure of several large banks – so-called Too Big to Fail banks. In addition, the crisis, and new post-crisis regulations, incentivized banks to become highly selective when granting loans, which thus excluded many borrowers. Unlike large companies, small and medium-sized enterprises (SMEs) as well as individual consumers cannot turn to capital markets to fund their borrowing needs. P2P lending aims to solve this problem by connecting lenders directly with borrowers – no intermediary needed. These platforms have provided new sources of credit for individuals and SMEs who have lost confidence in traditional banks or were denied credit due to stringent underwriting criteria. Since there is no third party involved in the loan contract, the interest rate can be negotiated by the borrower and lender, resulting in more attractive rates for both parties.

Online P2P lending platforms emerged during the early 2000s, before the 2008 financial crisis. The first online P2P lending platform, Zopa, was set up in the UK in 2005. Zopa’s founders have indicated that the idea came about when they realized that banks essentially serve as intermediaries that connect two financial markets: a borrower’s market (loans), and a lender’s market (deposits). Since Zopa, the P2P lending market has experienced massive growth and has developed into a large global industry. As of December 2017, the total value of transactions conducted through online P2P lending platforms amounted to $86 billion dollars[1]; this number is expected to reach nearly $292[2] billion dollars by 2022. The largest markets are in China ($56b), the US ($25.7b), and the UK ($4.2b)[3]. Surveys conducted in the U.S indicate that 10%[4] of the population uses P2P lending andin China, there were 5,500 P2P[5] lending platforms in 2016. However, the Chinese market has suffered because of an $8[6] billion pyramid scheme by the P2P lending platform Ezubao. In the U.S., the most well-known P2P lending platforms include Lending Club and Prosper. Lending Club is the largest platform in the US, which had a 45% market share in 2017 and 2.8 million users per month[7], more than triple its closest competitor, Prosper[8].

Behavioral Biases in Peer-to-Peer Lending

Our paper examines the role of three behavioral biases in the financial decisions made by P2P lending platforms. We deal in particular with three well known biases known to affect financial decisions in other contexts. The first, Familiarity bias, refers to investors’ tendency to concentrate their portfolio holdings in familiar assets. The second, Stereotypes, are generalizations about types of individuals that people use to make probability judgments when influenced by the representativeness heuristic. The third is Debt Account Aversion (DAA) which describes individuals’ tendency to consistently pay off small debts over larger debts in order to reduce the nominal number of debts they have. DAA is especially important in P2P lending where small borrowers probably have multiple loans outstanding.

Current empirical evidence indicates that Familiarity bias exists in P2P lending. Lenders are more likely to fund borrowers who are similar to them in ethnicity, gender, occupation, or place of residence[9]. If familiarity bias is driven by psychological and not economic reasons, then investors should invest in home state loans despite low returns and equal information, or when location information is emphasized. Accordingly, it was found that loans with a greater proportion of home state funds have worse returns, default sooner, and have higher default rates. These data are consistent with a psychological explanation of familiarity bias in P2P lending rather than an economic one.

As for Stereotypes, while in some platforms, borrowers are not permitted to disclose information that reveals race, religion, gender, and other personal attributes, many borrowers still do so. Studies have found that lenders are biased against different borrower characteristics. For example, P2P lenders are biased against young borrowers who are considered riskier and less likely to repay. In addition, it was found that single women paid 0.4 percent[10] less interest than men, indicating a bias in favor of female borrowers. In contrast, on PPdai.com, a Chinese P2P lending platform, female borrowers pay higher interest rates, but are funded more often and default less[11].

To test whether the presence of P2P loans affected the tendency of multiple debt account holders to exhibit Debt Account Aversion, we ran an experiment in our laboratory in which participants were saddled with multiple debts and forced to decide how to allocate this money for effective repayments. Participants in the experiment were randomly assigned to two conditions: ‘Bank’ and ‘Mixed’. Participants in both conditions were given $8,000 to repay four loans of different amounts and interest rates: 1) $10,000 at 3.8% interest, 2) $4,000 at 3.2% interest, 3) $4,000 at 3.2% interest, and 4) $5,000 at 3.5% interest. In the ‘Bank’ condition, all loans were from an institutional bank. In the ‘Mixed’ condition, the smallest loans were P2P loans, “from a private individual through a P2P lending app”, and the largest loans were from an institutional bank. In this setting, optimal financial behavior was defined as the amount of money repaid to close out the highest interest loan. Debt Account Aversion behavior was defined as specifically closing out the two smallest loans.

Our findings show that participants in the ‘Mixed’ bank and P2P debt condition behaved less optimally than those in the ‘Bank’ condition. Specifically, participants in the ‘Mixed’ condition repaid on average $3,407 to the highest interest loan compared to $5,128 that was paid by the participants in the ‘Bank’ condition. In addition, more participants in the Mixed condition exhibited the behavioral pattern expected by the Debt Account Aversion (i.e., closing the two smallest debts first) than in the Bank condition. These results suggest that the contrast between bank and P2P loans intensifies Debt Account Aversion. When given the option to repay four bank loans, participants behaved more rationally than when given the option to repay either high interest bank loans or low interest P2P loans.

Conclusions

The rapidly growing P2P lending market is a particularly interesting context for the study of behavioral finance because of its unique characteristics. P2P borrowers and lenders are exposed to different information and must make a variety of interpersonal decisions. For instance, lenders are often exposed to personal information and pictures of borrowers. In addition, P2P lenders may be confronted with more and different decisions, such as bidding on multiple loans. As a result, any behavioral bias that impacts these decisions is bound to have a greater effect on financial outcomes since the bias affects more financial decisions.

Moreover, regulators should take into account the behavioral limitations of these actors when considering ways to regulate P2P platforms. Revealing the behavioral patterns of both lenders and borrowers in P2P lending will enable financial institutions and policy makers to devise tools and procedures that help make this new credit market better serve the needs and limitations of different individuals.

 

 

References

Ayal, Shahar and Bar-Haim, Daphna and Ofir, Moran, Behavioral Biases in Peer-to-Peer (P2P) Lending (July 3, 2018). Forthcoming in Behavioral Finance: The Coming of Age (Venezia I. ed., World Scientific Publishers), 2018. Available at SSRN: https://ssrn.com/abstract=3207163

[1] Statista (December, 2017). Fintech Report 2017 – Alternative Lending. Available at https://www.statista.com/study/50625/fintech-report-alternative-lending/

[2] Id

[3] Statista Survey. (n.d.). Consumer awareness of crowdlending in the United States in 2016. In Statista – The Statistics Portal. Retrieved January 22, 2018, from: https://www.statista.com/statistics/638575/familiarity-with-crowdlending-usa/

[4] Id

[5] Gough, N. (2016, February 1). Online lender Ezubao took $7.6 billion in ponzi scheme, China says. The New York Times. Available at http://www.nytimes.com; China’s $8.6 billion P2P fraud trial starts: Xinhua (2016). Available at https://www.reuters.com/article/us-china-fraud-ezubao/chinas-8-6-billion-p2p-fraud-trial-starts-xinhua-idUSKBN1450I2; Tao, L. (2017, March 28). Just a few big Chinese P2P lenders seen surviving in sector tarnished by scandal. South China Morning Post. Available at http://www.scmp.com/frontpage/international.

[6] Id; Id; Id.

[7] SimilarWeb. (n.d.). Leading peer-to-peer lending websites in the United States from August 2016 to July 2017, by number of monthly visits (in 1,000’s). In Statista – The Statistics Portal, January 22, 2018, Available at https://www.statista.com/statistics/763623/p2p-lending-websites-us-by-visits/

[8] Galland, D. (2017). The 4 Best P2P Lending Platforms For Investors In 2017 – Detailed Analysis. Available at https://www.forbes.com/sites/oliviergarret/2017/01/29/the-4-best-p2p-lending-platforms-for-investors-in-2017-detailed-analysis/#63328b6f52ab

[9] Galak, J., Small, D., & Stephen, A. T. (2011). Microfinance decision making: A field study of prosocial lending. Journal of Marketing Research, 48(SPL), 130-137.; Ravina, E. (2008). Beauty, personal characteristics, and trust in credit markets. Working Paper, Columbia Business School, New York.

[10] Pope, D. G., & Sydnor, J. R. (2011). What’s in a Picture? Evidence of Discrimination from Prosper.com. Journal of Human Resources, 46(1), 53-92.‏

[11] Chen, D., Li, X., & Lai, F. (2017). Gender discrimination in online peer-to-peer credit lending: evidence from a lending platform in China. Electronic Commerce Research, 17(4), 553-583.‏