The evaporation of trust in the banking system following the financial crisis fostered the growth of digital platforms offering peer-to-peer investment opportunities in the US, Europe, and China.1 In Europe, as the debate about the capital market union progresses (European Commission 2017), policymakers see the possibilities for the digital investment and lending industry to help foster a unified capital market, which has been missing for so long. While the entrenched differences in public listing and bankruptcy regulations render difficult the immediate creation and implementation of a unified equity market, the Commission notes that digital investment services might allow borrowers from various countries to obtain funding elsewhere. Such platforms would also serve the purposes of small and medium-sized enterprises, which represent a vast majority in many European countries, which find it hard to get funding from the traditional banking sector due to a lack of collateral or reputation. Against these merits, some warn about the risks posed by these new forms of financial investment. The absence of regulation, unsecured and uncollateralised lending, and a lack of delegated monitoring might indeed undermine investor protection.
The balance between these merits and risks has yet to be thoroughly investigated, and the main economic channels affecting prices in these markets are poorly understood. In recent work we address this gap in the literature, conducting an empirical analysis with data from Prosper and Lending Club, the two biggest platforms in the US (Faia and Paiella 2017).
In the absence of collateral and delegated monitoring, digital peer-to-peer lending should be plagued by information asymmetry. Since investors cannot distinguish the quality of the project, prices and returns should embed an information premium. Moreover, adverse selection of borrowers would arise, so that we should expect higher default rates than those associated with credit through traditional banking. Data show instead that over the years, default rates in the platforms have decreased steadily and are much lower than those in the traditional banking system. Lending rates have gone down (though still remaining attractive for investors), and trade volumes have steadily increased.
The reason for this success stems from the innovation in screening technology rather than new forms of financial products. As borrowers enter the platform, machine learning algorithms collect massive amounts of information and allow the platform to compute and transparently publish credit scores (hard information), effectively reproducing a publicly available credit registry. On top of this, on platforms like Prosper, investors offer other investors the possibility of observing recommendations and the actual investment decisions of ‘friends’, namely groups of investors linked to each other by similar characteristics or by social networks. This constitutes additional soft information signals. These publicly available signals facilitate screening and also act as discipline devices for borrowers.
Our empirical analysis shows that, altogether, these hard and soft signals facilitate sufficient screening of projects, thereby improving borrowers' selection, so that default rates decline over time. Overall, the information premium declines. Indeed, a one standard deviation increase in the credit score lowers the lending rate by over four percentage points (25% of the mean Prosper borrowing rate over the period 2006-2014). This confirms the hypothesis examined in previous studies of the superiority of (transparent) markets over financial intermediaries due to the value added by a diversity of opinions (Allen and Gale 1999). Banerjee (1992) and Bikhchandani et al. (1992) similarly stress the value of information obtained by observing other investors' actions. The role of friends' recommendations in reducing adverse selection is also akin to the role of social ties in informal lending (La Ferrara 2003).
We rationalise all this through a model of asymmetric information with heterogenous borrowers (along the lines of Stiglitz and Weiss 1981). We show that the availability of public signals (freely available credit scores on the platform, or groups' recommendations visible to all) reduces adverse selection and lending rates. Our model also shows that in the presence of public signals, the value of information, measured through a Theil index (Theil 1967) rises. One important finding is that, in the context of digital investment, private signals are not equally as informative as public signals. Private signals would be informative if they were costly, but since it is rather cheap for borrowers to enter the platforms and provide reports about their investment, bad borrowers can easily mimic good ones.
Our analysis also uncovers the channel that explains the exponential growth of these platforms. Our empirical results indeed show that lending premia tend to decline when the risk of bank runs or liquidity dry out – proxied in our analysis by an indicator similar to that in Gorton 1988 – in the banking sector rises. When investors and borrowers expect fragility in the banking sector, fearing the early liquidation of projects and the possibility of haircuts on deposits, they shift to other forms of investment, possibly these digital ones. Indeed, the growth of this sector became faster after the 2007-2009 financial crisis and the resultant fragility of the banking sector. This evidence speaks in favour of a substitution channel between the digital sector and the traditional banking sector. We do not rule out that complementarity might emerge in the future. As the banking industry adapts to the new digital technology, and possibly develops multi-faceted platforms where customers can choose between many different investment opportunities, competition between the two sectors might foster strategic complementarities. So far, however, this has not happened.
Allen, F and D Gale (1999), "Diversity of opinion and financing new technologies", Journal of Financial Intermediation 8: 68--89.
Banerjee, A V (1992), "A simple model of herd behavior", Quarterly Journal of Economics 107(3): 797–817.
Bikhchandani, S, D Hirshleifer and I Welch (1992), "A theory of fads, fashion, custom, and cultural change as informational cascades", Journal of Political Economy 100(5): 992–1026.
European Commission (2017), Midterm review of the capital markets union action plan.
Faia, E and M Paiella (2017), “P2P lending: Information externalities, social networks and loans’ substitution”, CEPR Discussion Paper 12235.
Gorton, G (1988), "Banking panics and business cycles", Oxford Economic Papers, New Series, 40(4): 751–781.
La Ferrara, E (2003), "Kin groups and reciprocity: A model of credit transactions in Ghana", American Economic Review 93(5): 1730-1751.
Liberum Alternative Finance (2015), Direct lending report.
Stiglitz, J and A Weiss (1981), "Credit rationing in markets with incomplete information", American Economic Review 71(3): 393-410.
Theil, H (1967) Economics and Information Theory, Rand McNally and Company: Chicago.
 According to research firm Liberum Alternative Finance (2015), in 2015 the online lending industry surpassed $28 billion in the US and Europe, and reached $157 billion in China. Venture Capital firm, Foundation Capital, predicts that by 2025, $1 trillion in loans will be originated in this manner globally.