The Value of “New” and “Old” Intermediation in Online Debt Crowdfunding
Fabio Braggion - Tilburg University, CEPR and ECGI;
Alberto Manconi - Bocconi University and CEPR;
Nicola Pavanini - Tilburg University and CEPR;
Haikun Zhu - Erasmus School of Economics
-- Online debt crowdfunding is an increasingly popular investment (Rau 2019). The segment has moved from an older “peer-to-peer” model, where lenders pick the individual loans they fund, to a “marketplace” model, where the crowdfunding platform sells loan portfolio products to lenders (Balyuk and Davydenko 2019, Vallée and Zeng 2019). That has brought platforms closer to traditional banks, in that portfolio products are shorter-term liabilities invested in longer-term loans. Unlike bank depositors, however, marketplace lenders bear liquidity risk: they can only cash out their investment once the underlying loans are sold on the platform’s secondary market.
In our recent paper, "Fintech and the Value of Financial Disintermediation," we study the impact of the new business model on credit provision, lenders, and platform profits. A welfare comparison between marketplace, peer-to-peer, and traditional bank credit is not obvious. Marketplace lenders are exposed to liquidity risk; but compared to peer-to-peer lenders, they face lower search, diversification, and adverse selection costs; and compared to bank depositors they earn higher returns. In turn, increasing returns for lenders, and shielding the platform from liquidity risk, may incentivize credit provision, benefiting borrowers. Quantifying these tradeoffs is crucial to inform regulation and to address growing concerns about liquidity risk on online credit platforms (BIS 2017). Thus, we must assess the costs and benefits of alternative platform designs on the data.
To address this challenge, we build a model of online credit that nests the marketplace, peer-to-peer, and bank-like platform designs, and we estimate it on a database covering the universe of loans and loan applications on Renrendai (人人贷), a leading Chinese debt crowdfunding platform. We can observe the investment choices of each lender and the composition of the portfolio products offered on Renrendai, and thus quantify liquidity risk exposure. Moreover, our model allows us to simulate counterfactuals to compare the welfare effects of alternative platform designs.
Consistent with practitioner accounts, we observe a transition to marketplace credit: in 2010, when Renrendai was launched, 100% of lending was peer-to-peer; by the end of our sample in early 2017, over 95% was marketplace. This trend produced non-trivial liquidity risk: whereas most of Renrendai’s portfolio products have maturities of 3, 6, or 12 months, the underlying loans typically mature in 36 months. Our estimates also indicate that lenders’ preferences are heterogeneous: the more active lenders have a stronger preference for yield and a weaker disutility from liquidity risk, whereas the opposite is true for less frequent investors and first-time users. We interpret this as evidence that lenders with more appetite for yield might benefit from the marketplace model, while others, more concerned about liquidity, might be better off under the bank-like model.
We use our model to compare marketplace credit with two counterfactual scenarios: peer-to-peer credit, where only direct lending is allowed, and bank-like credit, where the platform sells portfolio products but bears liquidity risk. The marketplace model appears welfare-improving relative to the peer-to-peer model, raising credit provision and lender surplus by over 60%. Moreover, with a baseline level of liquidity (time to loan resale on the platform’s secondary market around half a day), bank-like credit provides identical loan volumes and lender surplus as marketplace credit, and only a small drop in platform profits (0.20%).
That comparison is very different when liquidity is low, as illustrated in Figure 1. As a “stress test,” we raise the time to loan resale to one month (long, but well within the range observed on Renrendai). Under that scenario, the marketplace model exhibits a larger decline relative to the bank-like model in credit provision (8% vs 1%) and lender surplus (34% vs 0.5%), but a smaller drop in platform profits (9% vs 12%). In other words: when liquidity is low the marketplace model is preferable from the platform’s point of view, but worse for lenders and borrowers. The potential conflict between the interests of the platform, lenders, and borrowers might reflect the current reach of online debt crowdfunding and the features of the lender population. When, in a final counterfactual, we alter the lenders’ composition to have weaker utility from yield and stronger disutility from liquidity risk on average, we find that the bank-like model is a Pareto improvement, raising platform profits too.
Figure 1 Marketplace and bank-like credit in a “stress test” scenario with low liquidity, and more/less risk-averse lenders
The figure compares the outcomes of a “stress test” scenario with low liquidity on the platform’s secondary market, under alternative assumptions about the platform’s design (marketplace vs a “bank-like” design in which the platform, rather than the lenders, bears liquidity risk) and the population of lenders (more vs less averse to liquidity risk). For each counterfactual scenario, we plot the percentage changes in credit provision, lender surplus, and platform profits relative to a baseline scenario with high liquidity and marketplace credit (i.e. corresponding to the typical conditions observed in the segment).