Research Article: Peer-to-peer lending and bias in crowd decision-making

Date Published: March 28, 2018

Publisher: Public Library of Science

Author(s): Pramesh Singh, Jayaram Uparna, Panagiotis Karampourniotis, Emoke-Agnes Horvat, Boleslaw Szymanski, Gyorgy Korniss, Jonathan Z. Bakdash, Brian Uzzi, Lazaros K. Gallos.

http://doi.org/10.1371/journal.pone.0193007

Abstract

Peer-to-peer lending is hypothesized to help equalize economic opportunities for the world’s poor. We empirically investigate the “flat-world” hypothesis, the idea that globalization eventually leads to economic equality, using crowdfinancing data for over 660,000 loans in 220 nations and territories made between 2005 and 2013. Contrary to the flat-world hypothesis, we find that peer-to-peer lending networks are moving away from flatness. Furthermore, decreasing flatness is strongly associated with multiple variables: relatively stable patterns in the difference in the per capita GDP between borrowing and lending nations, ongoing migration flows from borrowing to lending nations worldwide, and the existence of a tie as a historic colonial. Our regression analysis also indicates a spatial preference in lending for geographically proximal borrowers. To estimate the robustness for these patterns for future changes, we construct a network of borrower and lending nations based on the observed data. Then, to perturb the network, we stochastically simulate policy and event shocks (e.g., erecting walls) or regulatory shocks (e.g., Brexit). The simulations project a drift towards rather than away from flatness. However, levels of flatness persist only for randomly distributed shocks. By contrast, loss of the top borrowing nations produces more flatness, not less, indicating how the welfare of the overall system is tied to a few distinctive and critical country–pair relationships.

Partial Text

The “flat-world hypothesis” is an idea of a new “level” playing field where global economic equality gradually improves, is seductive [1]. Models of financial markets suggest that international capital flows are reaching more countries [2] and dominating national institutional policies [3], thereby laying a groundwork for global equality in access to capital that can promote new possibilities for prosperity among the world’s poor [4–7]. However, others have countered that outside of a handful of cities/countries the vast majority of economic activities (e.g., institution and government investment, web traffic, and telecommunications) have remained domestic over time [8, 9]. As crowdfinancing grows is it a flat-world mechanism for creating opportunities for the world’s poor, or does it follow the biased-patterns exhibited by other established economic activities and mechanisms? The Lucas Paradox [10] indicates that, counterintuitively, the liberalization of international capital regimes has not produced an open club, but rather a rich club–that is, a group of countries with similarly well-developed monetary institutions, cultures, and wealth that display in-group preferences [11] in lending and borrowing, thus restricting capital to poor nations [12–13]. Whether the Lucas Paradox occurs with philanthropic crowdfinancing is an open question and a means for testing the flat-world hypothesis.

Crowdfinancing networks differ in orientation. Some crowdfinancing systems provide funds in exchange for equity in an investment (e.g., Equitynet.com, CrowdCube.com, Seedrs.com) or supply interest-bearing investments (e.g., Prosper.com). Others, promote interest-free loans, in which no monetary interest is gained by the lender, but contributions are made for the developmental aid of the borrower (e.g., Kiva.com). In addition, other forms of crowdfunding such as funding a project for non-financial returns (Kickstarter) and charity where no return is expected also exist [34]. Our dataset of lenders, borrowers, and loans includes all transactions made on Kiva.com, 2005–2013. Although the average loan size on Kiva is about $700, the vast majority of loan contributions are made in multiples of $25.00 and most contributions are for $25.00 and $50.00. These loans typically support purchases of machinery for petty entrepreneurs, livestock for farmers, or domestic items such as water purification systems that improve living conditions (see S1 File for case examples). For each loan we know the:

To analyze the structural property of the network, we used degree-preserving network randomization, a common technique for assessing the statistical significance of observed network properties, including biased links between nodes [35–38]. Using the randomization method for weighted (multiedge) networks, we generate many synthetic networks by randomly rewiring the loan transactions in the observed network [39] while preserving the total transactions made to and from, for each country (i.e., in- and out-degree of every node). Many synthetic networks provide a distribution of every bilateral exchange, giving an expected mean and standard deviation across all links in the network, which are used to determine how far observed relationships are from expected values (see S1 File). A comparison between the null model and the observed data enables us to identify country-level lending biases in this network–that is, which countries have a lending–borrowing relationship that is greater or smaller than expected by chance, where chance theoretically reflects a system without bias [22]. To measure the flatness of the lending network, we count the number of country–pairs (positive as well as negative) where the observed links are statistically different from what is expected using a z-score for each pair of countries. The z-score zij of any link ij is given by
zij=Oij-Εijσij(1)
where Oij is the observed number of transactions from a country i to country j. Eij and σij are the expected number of transactions and the associated standard deviation according to the null model. For a country–pair, the z-score provides a normalized and relative measure of how far away the observed number of transactions is from what is expected by chance. A pair is classified as biased if its observed number of transactions is 2 standard deviations above or below the null model (p<.05). Global interconnectedness has raised the possibility that the world is becoming flatter and offering more equality of opportunity worldwide. Online crowdfinancing platforms like Kiva provide alternative channels of capital flow to traditional institutions raising the question as to whether peer-to-peer financing is making the world flatter. To the contrary, we find continued and increasing bias in an inter-country, peer-to-peer crowdfinancing network. This drift towards a less flat world may arise from individual level preferences or global factors. Although crowdfinancing provides a lending platform that connects lenders with borrowers and eliminates conventional intermediaries such as banks, it is the individual lenders who decide whom they give loan to and can often be biased in their decisions. These biases are reinforced and made even stronger by the rapid growth of the crowdfinancing platform itself (“rich gets richer” effect). An example of this growing bias in the crowdfinancing network is seen in the form of self-loops (lenders lending to borrowers in the same country), which are consistently biased in the positive direction. Nonetheless, whether or not these biases will continue to persist in the long run, remains an open question. We explored the effects of hypothetical disruptive events on system-level flatness with simulations and found that the lending network is not vulnerable to random losses of countries or bilateral ties. However, the targeted removal of a few high-volume lenders or high-transaction links could cause the network’s flatness to increase significantly. This implies that the decreasing flatness is not centered on all lending, but on the lending bias of a few giant lenders that skew the overall system. In this way, the flatness of the system is directly linked with the dominance of a few big players. This lending bias by a small number of countries combined with simulation results targeting these lenders, suggest that increasing inequality may be attributed to preferential attachment (“rich get richer”) [47].   Source: http://doi.org/10.1371/journal.pone.0193007

 

0 0 vote
Article Rating
Subscribe
Notify of
guest
0 Comments
Inline Feedbacks
View all comments