Can peer-to- peer insurance replace
insurance companies?
Do you know that you pay a higher car insurance price because insurance companies cannot tell that you are a responsible driver, but your twenty-something neighbor uses streets for practicing his drifting skills? The money insurance companies collect from policyholders must at least match all insurance claims. The more diligent drivers a company insures, the lower insurance price it can charge. Peer-to-peer insurance enables individuals to profit by telling who is naughty and who is nice.
Lessons from peer-to-peer lending
With peer-to-peer insurance, the role of insurance companies is changing. Something similar has started happening with banks a few years ago. Traditionally, banks allocated savings of individuals to the most suitable borrowers, as they have been considered to have the most expertise in selecting such borrowers and being the most efficient in doing so. However, in the last few years, banks have been partially replaced (especially in the allocation of credit to smaller borrowers) by peer-to-peer lending platforms, which connect borrowers and investors in a faster and, as it seems, also a more efficient way.
The biggest of them all is Prosper.com, which has funded over $10 billion loans to 800,000 U.S. borrowers. According to data on their website, the average FICO score[1] of Prosper borrowers was 710 (in the period between October 2016 and March 2017), compared to 700, which was the U.S. average in April 2017. Is this an indication of a positive selection of Prosper borrowers? Can individuals truly efficiently select better credit applicants?
Friends as financial experts
Prosper (to my knowledge) does not publish average default rates of the funded loans. A non-independent website,[2] however, reported default rates to be between 2.6% for the highest ranking and 15.9% for the lowest ranking loans. Research (Lin et al., 2013) which determines a default as payment being late by at least two months[3] shows that borrowers with friends on peer-to-peer lending platforms have a higher probability of successful funding and lower interest rates. Furthermore, they find that borrowers with a lender-friend are about 9% less likely to default. The likelihood of default is even lower (14%) if a borrower has a friend who bid on his/her listing. If such friend won the bid, the probability of default decreases by 21% compared to borrowers without friends, but otherwise identical characteristics.
Another research (Iyer et al., 2009) suggests that lenders on Prosper can justly differentiate between borrowers within the same credit rating category — although they do not observe the actual credit score of borrowers, they are (fairly) able to distinguish between high-scoring and low-scoring individuals in the same category. Lenders pay greater attention to the more credible signals and infer the borrower’s creditworthiness mainly from other standard-banking variables like the debt-to-income ratio, the number of current delinquencies or the number of credit inquiries. Lenders also learn from non-standard variables, with the borrower’s maximum rate he or she is willing to pay as the most crucial variable.
Besides friendship networks that are established outside the online market, Prosper allows its members also to join group networks that are formed within the peer-to-peer lending market. However, while friendship networks of borrowers seem to influence performance in peer-to-peer lending platforms positively, the opposite appears to hold for group networks (Chen et al., 2016). Although borrowers’ social capital generated from group networks leads to improved funding performance, there is no improvement in the repayment performance. In fact, group trust (such as group leader endorsement) worsens the repayment performance.
All different — all differently funded
The peer-to-peer lending platforms seem to be able to select the creditworthy borrowers. The result: only 1.8% of listings successfully collects funds in the lowest credit category, as opposed to 30.9% of listings funded in the highest credit category (Iyer et al., 2009). However, empirical studies have provided evidence that some lenders use decision methods that many national legislations would consider illegal if done by institutions.
Borrower’s appearance deducted from the enclosed picture can account for as much as 25 to 35% difference in the likelihood of receiving a loan and 60 to 80 basis points[4] difference in the interest rate. Moreover, borrowers who avoid providing a picture suffer adverse consequences from lenders. Investors also discriminate against the elderly and the overweight but in favor of women and those that signal military involvement (Pope and Sydnor, 2011). Similarly, trustworthiness based on appearance judgments allows borrowers to promise a lower interest rate compared to the average borrower and keep the same probability of obtaining a loan. In fact, such individuals also pay lower interest rates if they get funded (Duarte et al., 2012).
Is such bias economically irrational? In other words, are investors seeking for the highest risk-adjusted return on their investment making an efficient decision? Apparently, yes. Duarte et al. (2012) and Pope and Sydnor (2011) provide evidence that appearance attributes can explain a part of the variation in default probabilities. The explanation for such observations might be that lenders understand correlations between appearance features and “important characteristics for predicting default that they cannot perfectly observe, such as education and social-support networks” (Pope and Sydnor, 2011).
Recap of the lesson
Lenders in peer-to-peer lending can distinguish trustworthy borrowers from the crowd by hard financial and soft information, as well as heuristic rules. They perform even better if they are friends with the borrower (but not if they only join the same group), as they probably know more about them and perhaps also influence their behavior. So, why couldn’t it work in insurance as well? Prepare your naughty or nice list!
Written by Dr. Tjaša Bartolj
[1] FICO score is a credit risk score that provides information on credit worthiness of a borrower. It ranges between 300 and 850, with scores above 650 indicating a very good credit history.
[2] www.investmentzen.com/peer-to-peer-lending-for-investors/prosper
[3] If a loan is late for two months or more, it is sent to a collection agency.
[4] 1 basis point = 0.01%
References:
Lin, M., Prabhala, N. R., and Viswanathan, S. (2013). Judging borrowers by the company they keep: Friendship networks and information asymmetry in online peer-to-peer lending. Management Science, 59(1):17–35.
Iyer, R., Khwaja, A. I., Luttmer, E. F., and Shue, K. (2009). Screening in new credit markets: Can individual lenders infer borrower creditworthiness in peer-to-peer lending? NBER working paper.
Chen, X., Zhou, L., and Wan, D. (2016). Group social capital and lending outcomes in the financial credit market: An empirical study of online peer-to-peer lending. Electronic Commerce Research and Applications, 15:1–13.
Pope, D. G. and Sydnor, J. R. (2011). What’s in a picture? evidence of discrimination from Prosper. com. Journal of Human Resources, 46(1):53–92.
Duarte, J., Siegel, S., and Young, L. (2012). Trust and credit: The role of appearance in peer-to-peer lending. The Review of Financial Studies, 25(8):2455–2484.
