基于短时多源回归算法的P2P平台风险观测方法

P2P网络借贷作为当代互联网金融领域中流行的借贷方式,具有借款金额小、还款周期长短不一的特点,导致传统的年度风险评估方法因时间粒度过粗而容易给 平台投资者造成损失。基于此,提出一种基于短时多源回归算法的网络借贷平台运营风险的动态评估方法。通过动态时间窗对借贷记录进行切分,并以线性回归来量化平台的动态风险指数。实验结果表明,该方法能够及时反映P2P平台的风险宏观运营情况,并向投资者提供平台的动态风险评估和预测指标。     

Herding behavior in online P2P lending: An empirical investigation

We study lender behavior in the peer-to-peer (P2P) lending market, where individuals bid on unsecured microloans requested by other individual borrowers. Online P2P exchanges are growing, but lenders in this market are not professional investors. In addition, lenders have to take big risks because loans in P2P lending are granted without collateral. While the P2P lending market shares some characteristics of online markets with respect to herding behavior, it also has characteristics that may discourage it. This study empirically investigates herding behavior in the P2P lending market where seemingly conflicting conditions and features of herding are present. Using a large sample of daily data from one of the largest P2P lending platforms in Korea, we find strong evidence of herding and its diminishing marginal effect as bidding advances. We employ a multinomial logit market-share model in which relevant variables from prior studies on P2P lending are assessed.     

基于风险和剩余价值的在线P2P借贷投资推荐方法

在线P2P(peer-to-pear)借贷是一种新兴的在线个人财富分配和管理系统,它允许投资人直接对借款人创建的借款标的进行竞标和投资.在P2P借贷平台中,存在一个重要的问题即如何合理分配投资人的投资金额给合适的借款人.针对该问题,提出了一种基于风险和剩余价值最大化的投资推荐框架RTSM(risk total surplus maximize).RTSM首先对借款标的进行风险评估,然后基于经济学中的剩余价值理论,使用投资人和借款人在有风险情况下的剩余价值假设,将风险评估与投资推荐结合在一起,为投资人推荐高收益低风险的投资决策.实验在风险评估和投资推荐2个阶段对美国和中国知名的P2P借贷平台(Prosper、拍拍贷)的真实数据进行分析和验证.从实验结果可以看出:RTSM可以更好地降低风险和提高投资人与借款人的整体利益.     

考虑投资者朋友关系和预期效用的P2P借贷个性化投资推荐方法

随着互联网技术的快速发展,在线P2P借贷市场投资推荐已经成为网络金融领域的重要研究方向.对于P2P借贷市场潜在投资者来说,需解决的关键问题包括2个方面：1)如何选择真正符合自己投资需求和偏好的投资项目;2)如何将自己的投资金额在这些投资项目中进行合理分配.以往关于这两者的研究主要是侧重在借贷项目的违约风险预测、投资项目全局推荐及投资组合优化等方面.而随着研究的深入可以发现,仍在投资者效用无差异假设及投资者历史交易数据的基础上设计推荐模型,将难以满足不同风险偏好投资者的投资决策需求,保证推荐的有效性.鉴于此,1)基于Prosper平台历史数据建立P2P关联网络模型,并分别计算借贷项目和投资者的概念特征,得出相应的概念模型;2)进一步考察P2P关联网络模型中的投资者朋友关系,以捕获投资者之间投资行为的相互影响,发掘投资者投资行为的影响因子,并将其应用于借贷项目兴趣度的预测,以提高投资项目推荐的有效性;3)在此基础上,从预期效用理论出发,进一步考虑投资者风险偏好对投资需求的影响,建立个性化投资组合推荐框架,以提高其投资的满意度和经济性能;4)将其推荐结果与其他基准模型的推荐结果进行对比分析,以综合评价其推荐效果.在Prosper平台真实数据的基础上进行了详细的实验测试,结果表明:该方法相较于传统的投资推荐方法具有更好的推荐效果.     

考虑社会网络关系的P2P借贷项目违约风险预测

针对P2P（Peer to Peer）借贷项目违约风险预测中财务信息不完全或质量较低、预测准确率不高等问题,提出了一种考虑平台社会网络关系的P2P借贷项目违约风险预测的方法。通过对P2P借贷平台社会网络相关信息进行分析,从社会资本的结构维度、关系维度和认知维度发掘其中具有风险预测价值的关键特征,即社会网络风险特征,并将这些特征作为预测指标用于违约风险预测,依据多种非线性预测方法分别构建基于传统财务指标预测模型和引入社会网络风险特征后的混合指标预测模型,并对模型的预测结果进行了对比分析。实验结果表明,P2P借贷社会网络关系中蕴含着与借贷项目违约风险显著相关的特征,通过对这些特征进行有效挖掘并将其合理引入P2P借贷项目违约风险预测模型,有助于提高借贷项目违约风险预测效果,为投资者的投资风险规避及P2P借贷市场风险管理提供支持。     

P2P系统中面向异构平台的信任模型

针对P2P系统中可信计算平台与传统的非可信计算平台所组成的异构系统间跨平台信任建立的问题,基于可信计算技术,提出一种全新的信任模型,对其架构和认证流程进行研究。仿真结果表明,系统中节点具有较高的匿名度,同时该模型具有良好的抵抗恶意节点行为的能力。     

A trust model for online peer-to-peer lending: a lender’s perspective

Online peer-to-peer (P2P) lending is a new but essential financing method for small and micro enterprises that is conducted on the Internet and excludes the involvement of collateral and financial institutions. To tackle the inherent risk of this new financing method, trust must be cultivated. Based on trust theories, the present study develops an integrated trust model specifically for the online P2P lending context, to better understand the critical factors that drive lenders’ trust. The model is empirically tested using surveyed data from 785 online lenders of PPDai, the first and largest online P2P platform in China. The results show that both trust in borrowers and trust in intermediaries are significant factors influencing lenders’ lending intention. However, trust in borrowers is more critical, and not only directly nurtures lenders’ lending intention more efficiently than trust in intermediaries, but also carries the impact of trust in intermediaries on lenders’ lending intention. To develop lenders’ trust, borrowers should provide high-quality information for their loan requests and intermediaries should provide high-quality services and sufficient security protection. The findings provide valuable insights for both borrowers and intermediaries.     

IDEA:一种基于P2P借贷网络的投资决策分析算法

互联网金融P2P借贷平台上存在着较大的贷款投资风险,为协助投资人获得更佳的贷款收益,本文综合考虑贷款坏账风险、流标风险、利率和投资人风险偏好等要素,提出投资决策算法IDEA（Investment DEcision Analysis）：构建投资人-贷款网络,充分利用网络中的贷款投资行为信息来度量贷款坏账风险,利用图半监督学习方法度量贷款流标风险,为投资决策提供依据. 在真实数据集上的实验结果表明,相对于现有算法,我们的算法不仅可以取得更佳的投资收益,而且能够协助具有不同风险偏好的投资人进行投资决策.     

Visual analysis of risks in peer-to-peer lending market

Online peer-to-peer (P2P) lending financial products have been developing rapidly in recent years. This investment method is designed for people free of high-rate debts. However, the lending and borrowing affairs between anonymities may potentially produce risks, including wash sale and money laundering. Apart from the well-documented research on the causal factors and economic influence of the P2P lending market, limited attention has been paid to the risk management of individual P2P lending platforms. This study presents a visual analysis method that detects and analyzes risks in P2P lending transactions. Moreover, we evaluate our approach on real-world P2P data sets and report our findings.     

Who can get money? Evidence from the Chinese peer-to-peer lending platform

This paper explores how borrowers’ financial and personal information, loan characteristics and lending models affect peer-to-peer (P2P) loan funding outcomes. Using a large sample of listings from one of the largest Chinese online P2P lending platforms, we find that those borrowers earning a higher income or who own a car are more likely to receive a loan, pay lower interest rates, and are less likely to default. The credit grade assigned by the lending platform may not represent the creditworthiness of potential borrowers. We also find that the unique offline process in the Chinese P2P online lending platform exerts significant influence on the lending decision. We discuss the implications of our results for the design of big data-based lending markets.     

基于神经网络模型的网络借贷高危企业信用风险的识别研究

网络借贷的飞速发展在一定程度上缓解了小微型企业融资难的问题,但也暴露出网络借贷平台信用风险的识别问题。为充分识别高危网贷企业的特征,以中小型网贷企业为样本,通过指标筛选,挑选出与风险识别相关度较高的指标作为指标变量。并利用BP神经网络算法模型得出高危网贷企业在不同条件下的信用风险识别率和信用风险分类正确率。实验结果表明,高危网贷企业的信用风险具有高度识别性,高召回率、高正确率的特点。     

P2P环境下的扩展主观逻辑信任模型

为了缓解P2P环境下信任缺失、抑制恶意节点攻击,降低不可靠服务风险,通过扩展J?sang主观逻辑思想,提出了一种新的P2P信任模型—ESLTrust。为更精确地描述信任关系的复杂性,采用肯定信任值和否定信任值取代J?sang主观逻辑中肯定事件数和否定事件数,并引入时间衰减因子和风险值,计算得到节点信任度。仿真实验与分析表明,该模型使节点的信任度受恶意行为的影响更灵敏,可有效增强对恶意节点的抵御能力,提高系统交互成功比例。     

Information asymmetries and identification bias in P2P social microlending

The Internet has created new opportunities for peer-to-peer (P2P) social lending platforms, which have the potential to transform the way microfinance institutions raise and allocate funds used for poverty reduction. Depending upon where decision-making rights are allocated, there is the potential for identification bias whereby lenders may be motivated to give to specific projects with which they have an affinity without regard to whether it represents a sound financial investment. Using data collected from Kiva, we present empirical evidence that distant upstream lenders do not have adequate information about local business and loan conditions to make sound microfinance funding decisions, but instead make decisions based on identification biases. Furthermore, more information provided on the P2P lending site about the prospective loan does not improve the lender’s information about the loan conditions, but rather exacerbates the identification bias effect.     

A decision tree model for herd behavior and empirical evidence from the online P2P lending market

Herd behavior has been studied in a wide range of areas, such as fashion, online purchasing, and stock trading. However, to date, little attention has been paid to the herd behavior in online Peer-to-Peer lending market. With a decision tree, we model the formation of herding when decision makers with heterogeneous preferences are facing costly information acquiring and analyzing. Data from Prosper.com provides us with a good opportunity to explore empirical evidences for herd behavior. When herd behavior arises, individuals follow the behaviors of other people and generally ignore their own information which might cost them too much to obtain or analyze. Following this idea, we propose to detect herd behavior by focusing on investors’ decision-making time variation. We observed that friend bids and bid counts impose significant effects on the decision-making time of investors, which is considered as the evidence of herding. We also conduct empirical analyses to address the impact of herd behavior on an individual’s benefit. We reveal that lenders are more likely to herd on listings with more bids and friend bid, but their benefit will be reduced as the consequence of the behavior.     

P2P系统中基于声誉的信任评估机制

P2P系统能够使陌生节点之间进行在线交易,但安全问题越来越受到人们的关注。因此,如何在P2P系统中构建一个有效的信任机制来帮助在节点之间建立信任是目前P2P技术研究的一个热点。提出了一种新的P2P系统中基于声誉的信任评估机制,该机制较全面地考虑了影响信任度的因素,改进了局部声誉与全局声誉的计算方法,降低了信任度的计算负载,并且引入了黑名单机制。实验结果表明,该机制能有效地评估节点的信任度,识别和隔离恶意节点,提高系统交易成功率,并能有效地应用于P2P系统中。     

From the wisdom of crowds to my own judgment in microfinance through online peer-to-peer lending platforms

Information asymmetry is one of the fundamental problems that online peer-to-peer (P2P) lending platforms face. This problem becomes more acute when platforms are used for microfinance, where the targeted customers are mostly economically under-privileged people. Most of the prior empirical studies have been based on data from Prosper.com or similar sites that compete in traditional consumer loan markets. Our study examines P2P lending in microfinance for which borrowers are unbankable so that signals on creditworthiness of new borrowers are very limited. In addition, microfinance customers have more incentive to repeatedly seek loans from the market. Under this microfinance setting, we examine how lenders change their decisions as creditworthiness inference becomes increasingly possible through the accumulation of transaction history. Our findings confirm that lenders seek the wisdom of crowds when information on creditworthiness is extremely limited but switch to their own judgment when more signals are transmitted through the market. Different information sets are utilized according to the structures of decisions. Due to the possibility of a repeated game, it is also shown that borrowers try to maintain a good reputation, and direct communication with lenders may adjust incorrect inference from hard data when their creditworthiness is questioned.     

对等网络信任评价模型研究

对等网络已经成为用户共享服务的平台,如何结合用户评价进行服务选择已经成为其中的难点和关键问题。文中针对在开发、动态的网络环境中对服务评价可信度的问题,本文提出了一种基于评价的信任度更新算法,并给出了相关策略。在服务选择的过程中同时考虑服务的信任度和社交度,从而提高查找的准确率,并结合搜索行为聚类算法,降低群体欺骗的可能性。     

Are investors rational or perceptual in P2P lending?

This paper addresses the research on investor decision-making behaviors in peer-to-peer (P2P) lending from the perspective of rationality and sensibility not only to more thoroughly examine the factors affecting P2P lending but also to contribute to P2P platform builders’ and investees’ knowledge. We test investors’ rational choice behaviors using indices such as interest rates and investees’ monthly income and test perceptual choice using the identifiable victim effect. These tests attempt to determine whether investors prefer identifiable investees and whether this identification, as measured by social distance, affects the amount of investment. The panel data collected through the experiments are used to construct Probit and Tobit models, which address a combination of rationality and sensibility. The empirical results show that investors prefer large, short-term, high-interest loans and that investors are more likely to bid for such loans. In addition, investees find it easier to obtain funding when they share similar characteristics—in particular, a birthplace, location or ethnicity—with investors. Moreover, investees find it easier to obtain more funding when they share a similar birthplace, location or occupation with investors, whereas investees with an “identifiable” educational background find securing more funding to be more difficult. Furthermore, for a specific bidding amount, there are substitution effects between occupation and location, occupation and ethnicity, birthplace and education, and birthplace and age, which make it disadvantageous to increase the similarities across those dimensions. Finally, there are complementary effects between education and occupation and between education and age. However, there is an inverted-U relationship between social distance and bidding amount that determines whether rationality or sensibility dominate investors’ decisions.     

DisCoP2P: an efficient P2P computing overlay

CodiP2P and DisCoP are two peer-to-peer (P2P) computing overlays aimed at sharing computing resources (CPU, Memory, etc.) to execute parallel applications. Their component nodes are basically PC’s and a wide range of computer servers, desktops or laptops. This paper joins these two platforms into a new one, DisCoP2P, to combine the features from both overlays. CodiP2P is highly scalable, and DisCoP has an efficient searching mechanism and the ability to classify computing resources. The new platform takes advantage of these features and uses them to offer new facilities to schedule and execute parallel applications efficiently. This is accomplished at null cost because the platform is made up of nodes that share resources for free. This research field can also be classified in desktop computing. The success of this platform depends greatly on the added overhead. This overhead is produced mainly in searching for resources and system administration. The obtained results in a preliminary prototype, although not sufficiently conclusive, demonstrate the applicability of DisCoP2P in the real world, i.e. Internet.     

Analyzing default risk among P2P platforms based on the LAS-STACK method by considering multidimensional signals under specific economic contexts

Electronic Commerce Research - P2P platform default risk seriously affects the returns of investors, which may cause systemic financial risks. The existing literature mostly focuses on borrower...