What Drives the Expansion of the Peer-to-Peer Lending?

Transcription

1 What Drives the Expansion of the Peer-to-Peer Lending? Olena Havrylchyk 1, Carlotta Mariotto 2, Talal Rahim 3, Marianne Verdier 4 1 LEM, univerisity of Lille; CEPII and LabexReFi 2 ESCP-Europe, LabeX ReFi and Mines ParisTech - Centre for Industrial Economics (CERNA), Paris 3 Boston University 4 CRED (TEPP), Université Paris II Panthéon-Assas and Mines Paristech (CERNA), Paris 27 March th Financial Risks International Forum Institut Louis Bachelier, Paris 1 / 19

2 Introduction Prosper and Lending Club have been launched in in the US. They match savers with borrowers who need personal and business loans. P2P lending amounted to 12.5 per cent of the retail lending in the US at the end of / 19

3 Purpose of LendingClub and Prosper s borrowers Consumer loan amounts vary between a minimum loan of $1,000 for Prosper and $500 for Lending Club and a maximum loan of $35,000 for both platforms. When applying for a loan, borrowers declare a purpose: The platform pre-sets the interest rates based on the FICO score, the debt-to-income ratio and on some offline verifications (e.g. employment status) 3 / 19

4 Pricing structure A simplified scheme for a P2P lending platform Prosper s interest rates Prosper s Platform fees composition in 2013 LendingClub s interest rates 4 / 19

5 Purpose Purpose First explanation of what are the main drivers of the expansion of P2P lending in the US Hypothesis 1 Competition-based hypothesis 2 Crisis-based hypothesis 3 Technology-based hypothesis Identification strategy Our identification strategy relies on the exploration of the geographic heterogeneity of the P2P lending expansion at the county level. 5 / 19

6 Our Dataset We are the first to match LendingClub and Prosper s datasets and to merge it with other datasets Lending Club: observation points, a total volume of funded loans equal to $3.2 billion,from January 2007 to December 2013(99.25% of the Lending club portfolio.). Prosper: observation points, a total volume of originated loans equal to $662 million, from January 2006 to 30 October 2013 (100% of the total Prosper portfolio.). Variables from the loan book data of Prosper and LendingClub LendingClub Vol(in mln $) N. of counties N. of loans Prosper Vol(in mln $) N. of counties N. of loans / 19

7 Our Dataset 7 / 19

8 Spatial relations Various channels of interdependence : Theory of human interactions (Comin et al., 2012) Regional business cycles and economic shocks Technology diffusion Policy coordination Regional disparities for which we do not control with our right-hand variables Boundary mismatch problems when the economic notion of a market does not correspond well with the county boundaries (Rey and Montouri, 1999). 8 / 19

9 Methodology and empirical results Model specification: a spatial autoregressive model (SARAR model) Our objective is to test: The three hypothesis on the adoption of P2P lending (β, γ, δ); Whether adopting P2P lending in a county has a positive impact on the adoption of P2P lending in neighboring counties (λ). A SARAR model: y i = λ n w ij y j + β i competition i + γ i crisis i + δ i technology i + α i X i + u i j=1 i, j = 1,..., n u i = ρ n w ij u j + ε i with ε i N(0, σ 2 I ) j=1 Model estimation: Maximum Likelihood Estimation: cross-sectional spatial regression 9 / 19

10 Market structure variables Branches per capita: Number of branches in a county divided per population C3: The share of deposits of the three largest deposit taking institutions in a county HHI: Herfindahl-Hirschmann index, computed in terms of deposits Pay Day loans: Number establishment divided by population. Non-depository consumer lending Competition variables Obs Mean Std.Dev. Min Max Branches C HHI Payday / 19

11 Credit market during crisis period Crisis Leverage: The average leverage ratio of deposit taking institutions present via branches in a county, during crisis years of Crisis Tier 1 capital: The average Tier A capital ratio of deposit taking institutions, during crisis years of Failed banks: % of deposits affected by bank failures in a county during the whole period Crisis variables Obs Mean Std.Dev. Min Max Failed Crisis Tier Crisis leverage / 19

12 Socio-economic and demographic variables Population density: Population number divided by area in sq.m. in a county Bachelor: % of county population with at least bachelor education Income: Income per capita per county in logs Poverty: % of county population below poverty line Black: % of Afro-Americans in the county population Hispanic: % of Hispanic population in the county population Age 20 to 34: The share of the population between years State level dummies Other variables Obs Mean Std.Dev. Min Max Density Bachelor Income Poverty Hispanic Black Age 20 to / 19

13 Crisis and competition variables Online lenders are more present in counties with a poor branch network and with overleveraged banks. Counties with a more concentrated banking structure have witnessed slower growth of online lenders. Leverage ratio is significant and negative 13 / 19

14 Innovation and Internet variables Patents: Number of patents per population Broadband: % of county population with access to any broadband technology (excluding satellite) Mobile: % of county population with access to Mobile Wireless (Licensed) technology Speed: % of county population with access to upload speed 10 mbps or higher Speed: % of county population with access to upload speed 50 mbps or higher Innovation variables Obs Mean Std.Dev. Min Max Patents Broadband Mobile Speed 10000k Speed 50000k / 19

15 Innovation variables Volume of P2P loans per capita Number of P2P loans per capita 15 / 19

16 Differences between Prosper and Lending Club 16 / 19

17 Marginal effects We compute: ATDI: average total direct impact ATII: average total indirect impact ATI: average total impact ATDI ATII ATI Branches -0,001 0,000-0,002 HHI -0,196-0,063-0,259 Payday 0,008 0,003 0,011 Crisis Leverage -1,057-0,338-1,394 Density 0,046 0,015 0,061 Broadband -0,432-0,138-0,570 Poverty -0,641-0,205-0,847 Hispanic 0,647 0,207 0,854 Income log -0,188-0,060-0,248 Bachelor 0,324 0,103 0,427 Black 0,005 0,002 0,006 Age 20 to 34 0,219 0,070 0, / 19

18 Conclusion First attempt to explore the drivers of the expansion of online lenders. We have proposed three hypotheses and we account for spatial effects and socio-economic and demographic characteristics. Online lenders have made inroads into counties that have a poor branch network and with overleveraged banks. Counties with a more concentrated banking structure have witnessed slower growth of online lenders. Internet played a positive role only for the Prosper Marketplace. Higher education and higher propensity to innovate play a significant and positive role. Spatial effects play a crucial role. 18 / 19

19 Thank you for your attention 19 / 19