The growth of Peer-to-Peer lending platforms: Institutionalization and cross-border model

(1)

The growth of Peer-to-Peer lending platforms:

Institutionalization and cross-border model

Student: Valentina Raco Student number: S3047628 MSc International Financial Management

Faculty of Economics and Business University of Groningen Supervisor: W. Westerman

13-01-2017 Abstract

The recent emergence of the FinTech sector has led to the development of new innovative economic players. This research focuses on the recent development of the online Peer-to-Peer (P2P) lending industry analysing the nature of the impact of Institutional investments within P2P lending platforms, and comparing a domestic-based platform with a cross-border platform. The comparative analysis shows that crossing the domestic boundaries benefits lenders, borrowers and the platform itself. In order to analyse the institutionalization effect, six Binary Logistic regressions have been performed on 589,635 loans originated between 2012 and 2015, and publicly disclosed in the historical databases of lendingclub.com. The study expected to observe a negative effect of institutional investments on loans’ quality. Instead, the regressions’ results show that the institutionalization does not decrease the loans’ quality, instead increase their volume. Therefore, to verify if the same effect occurs also in the cross-border P2P platform Bondora, several interviews have been conducted, suggesting that institutional capital would support also the growth of the cross-border platform.

(2)

1

1. Introduction

Economy and Technology have been the major global players of the last fifteen years. Technology has experienced a very rapid growth and development, whereas in 2008 the economy has been hit by the global financial crisis. The resulting lack of confidence within investors and the technology penetration of financial markets, has led to the creation of a new economic reality, driven by the “merger” of Finance and Technology. The FinTech sector uses innovative technology to implement and improve business models, such as E-commerce.

As Einav et al. (2015) have reviewed, the FinTech sector has been facing a high growth rate and profits in recent years, and nowadays continues to be an innovative sector studded with new companies and marketplaces, different in size and in geographical location. Online marketplaces, such as eBay, Airbnb and Uber, facilitated the access to different opportunities and products through a direct matching of supply and demand. This innovative environment allowed the flourishing of many alternatives to traditional finance, such as microfinance. This alternative was first used in developing countries in order to compensate for the population’s lack of capital, and then was implemented and adopted by small business sector in developed countries (Freedman, 2000), together with crowdfunding platforms. After the 2008 crisis, banks reduced credit access to Small Business Enterprises (SME), as well as for individuals, due to the stricter regulations in terms of capital requirement (Morgan Stanley, 2015). According to the World Economic Forum (Report October 2015), based on International Finance Corporation (IFC) data, in 2010, the global

percentage of SMEs which bank financing access have been denied, was about 50% of them. Thus, crowdfunding platforms were created to provide, as much as possible, the financial services which banks’ complex system refuse to give (World Economic Forum, 2015).

(3)

2

The first online Peer-to-Peer lending platforms was the English Zopa, in 2005, which represents the outset of the P2P lending global market (Hulme and Wright, 2006). According to Moenninghoff and Wieandt (2013) own calculations, Zopa’s loan volume has passed from €29 million in 2006 to €1.1 billion at the end of 2011, with a compound annual growth rate of about 100%. Over the years, Peer-to-Peer lending market carried on its rapid growth. As shown by Zhang et al. (2016:2), UK’s P2P consumer lending market reached £909 million in 2015, and an average growth rate of 78% for the period 2013-2015. These successful trends, have encouraged the spreads of P2P online lending model across the world (see e.g., Wardrop et al., 2015 and 2016). Moreover, Morgan Stanley’s (2015) calculations expected a global annual growth rate of 51% between 2014 and 2020. Even though the online Peer-to-Peer lending business has been growing fast and successfully since its launch, recently most of the platforms decided to open the market to institutional investors. Therefore, this study proposes to analyse the impact of such “new lenders” on the P2P lending market.

The study proceeds as follows: Section 2. Literature review, research question and hypotheses,

analyses the theoretical literature and relevant evidence, introducing the research questions and related hypotheses; Section 3. Platforms background information, describes the two platforms’ history and model; Section 4. Methodology, describes the research methods of the qualitative and quantitative analyses: the cross-border diversification analysis, the institutionalization analysis, the analysis of the effect of institutionalization on the cross-border model, and shows the empirical design, the samples characteristics and displays the results. In addition, in this section are presented the discussions on the analyses’ results; Section 5. Conclusion, combines each discussion of the three analyses and presents the conclusion of the research; Section 6. Recommendation, attempts to give advice for the practice and offers directions for future research.

2. Literature Review, research questions and hypotheses

Peer-to-Peer lending could also be defined as “the household credit implementation of

crowdfunding” (Morse, 2015), which earns profits from peers’ connection and loan’s distribution. These debt-based platforms have increased their business throughout the years, expanding

(4)

3

2000), the cutting-off of the financial intermediaries. According to Moenninghoff and Wieandt (2013), P2P lending gains competitive advantage over the traditional financial institution lending, by lowering operational cost (due to the reduced cost of online platform and disintermediation) and increasing both interest rate and small loans volumes (which are turned to the customers in the form of lower fees and zero interest spread). P2P lenders face the trade-off between cost and benefits given by high expected rate of return and the possible exposure to both credit and default risk. Often the investors do not have the expertise to evaluate such risks, even though P2P companies apply a screening process based on individuals’ credit history disclosure, such as “real time” online data and credit agency. These mechanisms help to assess the repaying probability of the unsecured loans, and to assign an individual default rate (see, e.g., Morgan Stanley, 2015; Lin et al., 2013; Einav et al., 2013). Along with the screening process, P2P’s borrowers and lenders might create groups aiming to improve members’ information. Freedman and Jin (2008), as well as Everett (2010) find evidences on the US-based P2P Lending Prosper, regarding decrease of default rates within these groups, due to exchange of information. In addition, Liu et al. (2015), identify this “friendship” as one of the main differences between these platforms and commercial banks.

Generally, Peer-to-Peer lending platforms do not have a deposit of guarantee for the investments. The extent of regulation applied is based on the different classification of the platforms, which depend on the country jurisdiction. When they are classified as intermediaries, registration is required together with licence need to provide credit. In Germany and France, P2P platforms are regulated as banks, so they are required to have a banking licence and full compliance to the

regulations (Kirby and Worner, 2014). US regulation treats them as public companies, so they must disclose platform activities and loan originations. Indeed, the regulation is more complex because platforms must register with the SEC (Securities and Exchange Commission) and meet the requirements both at the federal level and at the state level (Kirby and Worner, 2014). Therefore, the lender bears the full risk of investing in the P2P market, which makes the probability of credit risks more likely to occur. In the worst scenario, the default of such platforms could have a significant impact on the lending market and economy, as well as a mortgage crisis (Pokorná and Sponer, 2016).

(5)

4

These potential risks could lead to credit rationing, because they affect the lenders ability to choose the best investments related to his needs, due to the lack or the uncertainty of information provided by the borrowers. Furthermore, information uncertainty lead to the raising of herding behaviours within lenders. Therefore, Pokorná and Sponer (2016) suggest different approaches to mitigate and remove risks arising in the Peer-to-Peer lending market, such as increase individual investors’ risk consciousness and the use of econometric models to predict borrowers’ credit scoring precisely. They also underline the related increase in profits both for investors and for the platforms if the said approaches are fully applied. Furthermore, Berger and Gleisner (2009) argue that P2P groups create value because their activity, which is similar to the intermediaries’ one, helps in reducing

information asymmetry and in improving borrower’s credit rating. The same findings emerge in Freedman and Jin (2014) research on Prosper’s transaction and on the role of online social networks in mitigating risks related to borrowers’ creditworthiness.

In 2011, the US-based platform Prosper decided to cooperate with several financial institutions, such as investment banks and private equity funds. This opening to other than retail investors, leads to the P2P lending switching from Peer-to-Peer to “Institution-to-Peer” platforms. As Light (2012) illustrates, one year later the percentages of institutional shares increased up to 50%. In Europe, UK platforms faced a gradual growth rate of institutional involvement from 11% to 45% in the period 2013-2015 (Zhang et al., 2016). This switching offers more guarantees to lenders and helps in managing risks, but drastically transforms the early post-crisis idea of “disintermediation”. Furthermore, there is a growing awareness about “institutionalization”. Institutional involvement may mitigate loans’ default risk, but on the other hand, it arises concerns regarding other investors’ protection, due to similarities shared with the mortgage crisis. The main concerns are against the on-going creation of secondary markets within the platforms, where securitized loans can be sold before maturity, and the uncertainty regarding the quality of the loans sold in that market (Manbeck and Hu, 2014). In Berndt and Gupta (2009) study of bank credit models(based on originate-to-distribute model), the existence of a secondary market increases the loans funding volume, but may decrease their quality. According to the authors, one explanation could be related to lack of

observable information when selling unsecured loans on the secondary markets.

Even though P2P lending platforms are usually mainly focused on their domestic market, many platforms, such as the European Funding Circle recently has announced huge international

(6)

5

crossing the national frontier could increase the probability to have their loans’ funded; and the investors, as their portfolios could benefit from a cross-country diversification (Yoshimura, 2015). However, P2P platforms who decide to spread their business globally have to deal with additional risks, such as currency exchange rate fluctuation, and countries’ economic regulation (Milne and Parboteeah, 2016).

Since the launch of the Peer-to-Peer lending model, researchers have mainly focused on the behavioural explanations of P2P actors’ strategies and related market risks, testing the model also for discrimination aspects (see, e.g., Pope and Sydnor, 2011; Ravina 2012). Few researchers have analysed the economic effect and the rapid growth of this alternative model (see, e.g., Wardrop et al., 2016), and even less the “institutionalization” effect. Emekter et al. (2015) have found

significant empirical evidences. In this recent research, P2P loans’ credit risk as well as loans’ performances have been evaluated and measured using Lending Club data. The analysis reveals that the platform assigns certain credit risk to loans, which predicts with certainty the chance of default, except for the riskiest investments. Moreover, the results show that the longer the riskiest loans’ maturity is, the higher would be the mortality risk. However, the economic implications of this “disruptive” innovation have been not sufficiently examined by the P2P lending literature.

In order to have a complete up-to date view and to better explain P2P lending mechanism, this study first compares the US’s platform, Lending Club performances with the cross-border European platform, Bondora, to investigate the potential effect of internationalization on these platforms. Following, it investigates the impact of Institutional investment on P2P lending market, using public available data from one of the major US P2P platforms, Lending Club. Lastly, the study attempts to predict the nature of the potential effect of such an institutional involvement on the cross-border platform. Therefore, the following questions are addressed: what impact has “institutionalization” on Peer-to-Peer platforms? and does the cross-border expansion of P2P marketplaces lead to better performances?. Moreover, with respect to the cross-border model the study addressed the following question: what impact has the “institutionalization” on a cross-border platform?.

According to the literature, crossing the frontier might lead to increase of revenues and to spread the risk on different countries. Thus, the following hypothesis is generated:

(7)

6

In line with the first research questions and with the Berndt and Gupta (2009) evidence on the deterioration of loans’ quality due to the existence of secondary markets (where institutional investors have open access), the following hypothesis is generated:

Hypothesis 2: Institutional investments have a negative impact on loans quality

Furthermore, with respect to the European P2P lending market, Taylor (2015) analysis on the level of institutional involvement shows interesting results for the three major UK platform, Zopa, RateSetter and Funding Circle. The findings revealed that the institutional capital has helped Zopa and RateSetter to expand their business and to offer their services to new borrowers, such as those of high credit risk; whereas Funding Circle used the new capital to increase its liquidity. These findings indicate that the institutional capital help to increase the loans’ origination of the platforms also in the UK.

Despite the lack of certain institutional data available regarding Bondora, this study aims to observe the effect of the institutional involvement on the cross-border platform. According to the results obtained when testing hypothesis 2 on Lending Club data and together with the evidence on the UK market (Taylor, 2015) the following hypothesis is generated:

Hypothesis 3: Institutionalization has a positive effect on cross-border platforms

3. Platforms background information Bondora, the cross-border platform

Bondora is one of the major Pan-European cross-border P2P platform, and it was launched in 2009 in Estonia. It represents one of the European leaders in non-bank digital industry, which provides consumer loans. In 2012, it opened to European retail investors and in 2013, it extended its business in the Finnish and Spanish markets. Over the years, it has become the first marketplace providing loans and investments across multiple countries. Generally, Bondora financed all the loans it originates by selling the receivables to retail investors from 40 countries around the world.

Bondora has decided to base its business on unsecured consumer loans with a minimum amount of €500 to a maximum of €10,000, and repayments terms from 3 to 60 months. The platform

(8)

7

controlled defaults and strong performance in recoveries. Bondora has the license to do business in Estonia, because the Estonian Financial Supervision Authority has recognized the platform as a credit provider. To expand its activities over the domestic market it has obtained licences also in Finland, Spain, and recently in Slovakia. Bondora’s strategy has been to penetrate those European markets that lack of a competitive credit market, because of their size, as the Estonian and Finnish markets, or adverse macroeconomic conditions, as the Spanish one. Finland, Estonia and Spain are characterized by a high level of digital technologies, with Spain showing a great digital

convergence. The results of this strategy were the increase of the borrower’s probability to have their loans funded and investors’ opportunity to diversify their investments in national and non-national markets. The current percentages of loans originated by Bondora are 51.7 % in Estonia, 28.9 % in Finland, 18.3 % in Spain and 1.2 % in Slovakia (Bondora.com, public statistics, 2016). Bondora has an international business structure, which leverages Swedish banking infrastructure, German technology, Estonian development skills, worthwhile FinTech back-office operations and American marketing expertise.

Bondora earns its revenues from financing and servicing consumer loans. The revenues are

collected by several fees charged to the borrowers: contract fees, charged when the amount is paid out; monthly fees; debt collection fees. In addition, Bondora earns from interest on unsold loan receivables. The borrowers’ profiles are checked thorough their financial background, income and expense verification. Moreover, to assess a borrower creditworthiness and fraud risks, Bondora relies on transactional, credit bureau and big data.

Unlike the other type of platform, Bondora offers a single platform allowing both for lending and for borrowing, regardless for the residency, language and currency. This technology brings Bondora to gain extensive economies of scale, improvements and costs are shared across all markets in which it operates. Therefore, it gains competitive advantages through its business strategy

consisting in efficient ways of serving near-prime borrowers originating risk-adjusted net returns on loans portfolio, and maintaining costs under control.

(9)

8

Figure 1 – Percentages of Bondora’s shares per channel of investment

Source: Bondora.com, Blog Nov. 2016

Recently, Bondora released the October 2016 statistics on Loan funding per channel. Figure 1 shows that despite the introduction of the API system, the Portfolio Manager channel, which is an automated channel run by Bondora’s algorithm, remains the highly chosen way for investment (86% of the investors). Through Portfolio Manager, investors can have their investment sized and the algorithm would offer different investment options that meet their targets. Instead, the API interface gives more flexibility and access to detailed information, as well as data and services, which will help investors to make better decisions. Anyway, this new interface had not so much success among investors (about 6% of them use it). This statistic shows that Bondora’s investors have no time or have no specific financial knowledge to actually analyse and go through all the different loan’s offers.

Along with the geography diversification of Bondora, the online platform allows also risk seeking investors to sell before maturity their portfolio’s loans that are current and overdue as long as they are not more than 60 days in late, providing an internal secondary market. Bondora’s secondary market was launched in 2013, and it works as follows: loan parts are traded at principal value and remain listed until they are sold or cancelled by seller 30 days. Any unpaid interest, overdue interest, overdue principal and unpaid late charges are omitted from the sales prices and will

(10)

9 Credit scoring and pricing model

In January 2015, the platform has instituted a new rating system, called Bondora rating, aiming to obtain more predictable risk-adjusted returns, taking into account: historical data on the customer, such as employment records; external data, such as country of residence, credit history and public records data; behavioural patterns, based on customer observations Bondora’s website. In addition, because Bondora has activities in four different European countries, and to improve scoring and pricing accuracy, its pricing model is adjusted to reflect changes in other relevant components, which concur to determine the risk-based price. Therefore, the returns delivered to investors are adjusted to country-specific risks, updated for macro-financial data; risk-free rate of return, monitored to reflect the actual monetary conditions; market liquidity conditions. The platform’s pricing model is based on the Expected Loss Rate E (L) that can have a minimum and a maximum range (Table b, Appendix B).

The Expected Loss Rate E (L), as used by traditional lending institutions, is expressed as:

E (L) % = PD * LGD * EAD% (1)

According to the formula, the Expected Loss Rate E(L) is given by multiplying two risk metrics,

PD and LGD, and the exposure at default, EAD. PD and LGD are the risk metrics representing: the

probability of default (PD), the likelihood of loan’s default, including the borrower’s credit history and loan’s purpose; the loss given default (LGD) represent the percentage of outstanding exposure that an investor might lose if a loan defaults (Shuermann, 2004).

LGD = 1 – recovery rate (RR) where RR = 𝑉𝑎𝑙𝑢𝑒 𝑜𝑓 𝐶𝑜𝑙𝑙𝑎𝑡𝑒𝑟𝑎𝑙

𝑉𝑎𝑙𝑢𝑒 𝑜𝑓 𝑡ℎ𝑒 𝐿𝑜𝑎𝑛 (2)

This means the fraction of funds lost for the investor after all the recoveries, are estimated based on historical value. Therefore, precise estimation of LGD cannot be done in Bondora’s new markets (Finland, Spain and Slovakia), instead is assumed a LGD of 90%. The EAD represents the total exposure to credit risk percentage, the amount that the borrower owes to the lending institution at the time of default, including outstanding principal amount plus accrued interests. Applying the traditional formula to estimate E(L) in a cross-border platform as Bondora, means that others factors must be taken into account when calculating the explicative default factors LGD, EAD and

PD. Differences in country risks (such as recovery rate) must be added to the macroeconomic

(11)

10

is shown that since the introduction of the new rating system, performances or the returns have increased reasonably in the European Bondora market (Table a, Appendix B)

Net returns are calculated applying the XIRR (Extended Internal Rate of Return). This measure is used to return the internal rate of return for a specified cash flows that are neither periodic (not on-time payment) nor always positive (portfolio has both outgoing and incoming payments).

XIRR can be seen as the discount rate that generate a net present value (NPV) of all cash flows of a

portfolio equals zero. XIRR includes in its calculation: loan issue date, amount, repayment dates, amounts, and the sum of the scheduled future principal repayments, which is assumed by the platform to equal the present value of the portfolio. In the calculation are not considered the unpaid principal and interest payments.

Bondora’s risk-based pricing model estimates the borrower’s interest rate I as:

I = E(L)% + E(R) (3) where E(L) is the expected loan’s loss and E(R) is the expected loan’s return. The estimation of the expected return 𝐸(𝑅) on the same loan is calculated from the lender’s perspective:

𝐸(𝑅) = 𝐼 – 𝐸(𝐿)% (4) where I is the interest rate and E (L)% is the percentage of expected loan’s loss,

The expected return should be enough to cover the cost of capital at a given risk level. The minimum acceptable return, or the cost of capital, is generally determined by the CAPM, with its parameters adjusted to credit business. Every expected return is related to each calculated loan beta in the same way, as the expected return is a function of the asset’s beta in standard CAPM.

Therefore, loan’s beta is composed by the unexpected loss, the loan’s specific risk (based on Basel II IRB Risk Weight Function, used by banks to calculate capital requirements), and a country risk factor, because different European countries might have different reaction to systemic risk. Lending Club, the Home-based marketplace

Lending Club is a US-based marketplace, which have gained huge success and market shares, since its foundation in 2006. It is the world’s greatest online marketplace matching borrowers and

(12)

11

possibility within the marketplace and to rise up the amount of loans issued to $2 billion. According to at the time Lending Club’s CEO Renaud Laplanche, the platform had gained a long-lasting competitive advantage over banks due to the combination of advanced technology and costs

(lendingclub.com). Given the huge success faced, in 2014 Lending Club completed an IPO of $900 million. Later on, the platform has continued to strongly grow over the years, extending its business to Small Business Enterprises sector, partnering with banks and opening its market to Institutional Investors.

In line with Peer-to-Peer lending system, Lending Club leverages new technologies to operate in the marketplace at lower cost compared to the banks. Borrowers can access a loan through online or mobile interface at a low interest rate and lenders can provide capital in exchange for higher returns. Lending club loans’ target are unsecured personal loans, business loans and financing for elective medical procedures. Personal loans must have a minimum of $1,000 and a maximum of $40,000 and can have terms of 36 or 60 months. Small Business Enterprises can access loans up to $300,000, but the business has to be active at least from 2 years and applicants must own at least 20% of the business and a good credit history. Since 2014, when Lending Club acquired

Springstone Financial, a company specialized in medical financing, it is possible to apply to receive

loans for medical scopes. Lending Club relies on FICO score to assess potential borrower’s degree of creditworthiness. The FICO score, provided by Fair Isaac Corporation (an American analytics software company, which uses Big Data and mathematical algorithms to predict consumer behaviour), is calculated as a mathematical formula based on borrower’s credit report data. The score ranges from 350, indicating low level of repayment probability, to 850 high level of repayment. Generally, 720 or higher is considered a solid FICO credit score, even though

creditworthiness standards are different among investors. Each FICO score’s ranges is associated to alphabetic letter from “A” to “G”, being “A” the most prime loans. Consequently, depending on the type of FICO score, different interest rates are allocated. Investors rates of return can range from 5.32% to 30.99% and the Annual Percentage Rate (APR) for borrowers consequently may range from 7.46% to 34.34%, depending on the current market interest rate.

(13)

12

portfolio before the loans are charged off. Hence, this measure includes future losses estimation of any notes that are in "past due" status, not already charged off, based on historical charged off rate over nine-month period. Investors can model their Adjusted NAR calculation to their own

assumption about loans’ the future performances. Lending Club’s NAR formula:

(5)

Source: Lendingclub.com

NAR is calculated for any period from month 1 to month 𝑁, where 𝑖 is the monthly period.

Adjusted NAR formula:

(6)

Source: Lendingclub.com

calculated for any period from month 1 to month 𝑁, where 𝑖 represent the monthly period, which means that the estimation of portfolio returns is the cumulative annualized return on all dollars invested in loans over their life, with an adjustment for estimated future losses.

4. Methodology

In order to test the three hypotheses, this study refers both to qualitative and quantitative research methods. Hypothesis 1 is examined through a comparative analysis between the American Lending Club and the Pan-European Bondora. Following, to test hypothesis 2, to determine the effect of the independent variable (initial status of the loans) on the dependent variable (the loan current status), the Binary Logistic regression is used, as is done in the study by Emekter et al. (2015). With the aim of examining the predictive capability of the variables, six binary logistic regressions were

(14)

13

4.1 Cross-border diversification analysis

Comparative analysis

The comparative analysis on the two platforms is divided in three tables, located below the

discussion section. Table 1, displays the platforms’ main characteristics; Table 2, reports the loan’s

details; Table 3, shows the lenders returns and characteristics. 4.1.1 Discussion

According to the comparative analysis evidence, the first hypothesis is supported. In fact, the comparative analysis section shows the advantage of going over the national frontiers. Bondora offers the possibility to diversify within different grades of riskiness and different countries, increasing the probability of success for both borrowers and lenders.

Firstly, as stated before the platforms have business in two different countries: United States of America and four countries in Europe (Estonia, Finland, Spain and Slovakia). Lending Club can be defined as a home-based platform, because only borrowers who live in the US can apply to it. Nonetheless, foreign investors have recently been allowed to invest in its loans (LendingClub.com). Whereas, Bondora is a cross-border platform, which core business is based in Estonia, but it also operates in three other European countries. Historically, the US lending market is different from the European one. US lending market traditionally reported a high volume of requested loans, which can vary from mortgage’s loans to university loans. Instead, the European lending markets have always reported smaller requests of loans compared to the US. Lending Club reported more than $22 billion of loans issued by September 2016, reporting an increase of $1,972,033,973 between the second and the third quarter of 20161, whereas Bondora reported a little more than €70 million, reporting an increase of €2,445,710 between October and November 2016 (Figure c, Appendix B). Despite the cross-border activities of Bondora, the difference in the amount of loans issued is still high. However, this can be due to both the size of the market in which the platform operate, and the type of loans it issued (business or consumer loans).

From a lender perspective, Bondora might offers potential returns rate higher than Lending Club, but they might require to pay higher fees on the investments (Table 1 and Table 2). The higher returns can be due to the opportunity of cross-border investment diversification (Figure 2), but also the diversification based on different type of loans’ grade. However, Bondora reported a higher

(15)

14

default rate compared to Lending Club. On the borrowers’ side, the average interest rates and the APR are smaller in Lending Club than Bondora. Overall, both the platforms reported increases in expenses and revenues for 2015 (Table a, Appendix A). Lending Club reported losses, which are due to the expansionary strategy in marketing and engineering. On the other hand, Bondora reported small amount of profits for the same year, and decided to stop loans’ issuance in Slovakia, as its market did not grow as expected. Instead all others countries loan issuance continues to grow

(Bondora AS, Consolidated Annual Report 2015).

Table 1 – Platforms Overview

Notes: The APR includes all fees, all costs of loans’ procurement, and the nominal interest rate on loan. Sources: lendingclub.com; bondora.com.

Lending Club Bondora

Country United States of America Estonia HQ, Finland, Spain and Slovakia

Loan Range Personal loans = $ 1,000 - $ 35,000 Business loans = $ 1,000 - $ 300,000

Personal loan = € 500 - € 11,000

Reported Return Rate 8.5 % 16 %

Average Interest Rate 14.08 % 30.8 %

APR 7.46% to 34.34%, 18.89 % to 58.47 %

Platform features  Automated Portfolio Recommendations  Individual Note Search  Secondary Marketplace

 Automated Investment Selections

 Portfolio Managers  Portfolio Builders  API

 Secondary Marketplace

Fees and Payments  Origination Fees = 1.1% - 5%  Late fees = 5%

 Contract fee, % of loan amount = 5.95%

(16)

15

Table 2 – Loan details

Total Loan issued $ 22, 659, 522, 884 2 _{€ 70, 551,068}3

Loan Term 12 Months

24 Months 36 Months 48 Months 60 Months 12 Months 24 Months 36 Months 48 Months 60 Months Types of Loan offered Consumers loans 4 Business loans Consumer loans Requirement for Borrowers  660 FICO score

 Borrowers must have at least 36 months of credit history and a satisfactory debt-to-income ratio.

 Be in paid employment with sufficient income to cover all monthly expenses  Not have a history of bad

credit such as arrears, default, bankruptcy or an open enforcement

proceeding

 Not have a history of gambling problems

Notes: Consumer Loans include: Automotive, Medical Expenses, Property, Renewable Energy Miscellaneous, Debt Management. Sources: lendingclub.com; bondora.com.

Table 3 – Lenders returns

Total Interest paid to investors

$ 2, 429, 816, 661 € 15, 657, 493

Default Rate 5.13 % Estonia = 14.1 %

Finland = 30.7 % Spain = 46.2 % Slovakia = 76.1 % Total = 24.9 % Portfolio Returns by loans grades  A = 5.12 %  B = 6.98 %  C = 7.78 %  D = 7.77 %  E = 5.77 %  F + G = 7.06 %  AA = 11.71 %  A = 12.95 %  B = 13.06 %  C = 14.49 %  D = 15.94 %  E = 16.99 %  F = 19.2 %  HR = 21.13 %

Notes: The issue date of total interest paid starts 2007-Q1 ends 2016-Q1; Default rate per Rating based on defaulted principal amount. Regarding the default rate per countries, it has to be taken into account that Bondora has expanded its business recently in these countries (since 2013, and Slovakia since 2014). Sources: lendingclub.com; bondora.com;

(17)

16

Figure 2 – Percentages of loans issued by Bondora in the different countries.

Source: https://www.bondora.com/en/public-statistics

4.2 Institutional involvement analysis

Regarding the institutional involvement, the platforms do not disclose certain information regarding Institutional investment. A 2015 survey by Richards Kibbe & Orbe and Wharton FinTech has investigated how institutional investors related themselves to peer-to-peer lending and what are their main concerns regarding these growing online marketplaces. According to this survey, institutional investors prefer to invest mainly in the whole amount of the loans, rather than the fractional one. This finding has been confirmed in the same survey of 2016. Hence, considering the lack of certain information regarding the institutional investors in the platforms, this study assumes: the institutional investors’ data are those related to whether a loan is funded in its entirety, whole funded loan; or its amount is spread within several investors, fractional funded loan. Thus, every whole loan reported in the platform’s databases is considered as if an Institutional investor funded it.

Institutional involvement

The latest evidence of Institutional interest in Peer-to-Peer lending market, is dated 2011 when

Prosper, a US-based peer-to-peer platform decided to collaborate with several financial institutions. Lending Club’s whole loan type became available in 2012, as shown by the occurrence of whole

(18)

17

$100 mm of fractional loans, whereas the volume of whole loans originated has been $175 mm. Even though Lending Club maintains institutional investment around 25% of the total funding (Lending Club, 2015), in 2014 the percentage of whole loans originated passed by 50% the fractional loans’ percentage, indicating a future expected increase.

Figure 3 – Lending Club origination volume 2012-2014

Source: Pargeter (2014), Orchardplatform.com5

4.2.1 Data

The analysis of institutional involvement is carried on using public Lending Club loans’ data downloaded from www.lendingclub.com. Lending Club is one of the few platforms who allow public access to loans’ database, and disclose the initial status of the loans. On the platform’s website are available loans’ data from its foundation (2007) to the latest quarter. The relation between Institutional involvement and loans quality is observed in three different periods of time 2012-2013, 2014 and 2015. In order to better explain this relation, the interest rate paid on the loan (which is related to loans’ rating status) is included as independent variable. Among the data, this study considerate only subsamples of loans whose repayment term is set at 36 months, taking out 60 months’ loans, since most of them are still outstanding loans. The number of loans observed is 589,635, respectively: 143,892 observations for 2012-2013; 162,570 for 2014; 283,173 for 2015.

(19)

18 Statistical research design

Assuming 𝑦_𝑖 (dependent variable), to be an unobserved continuous number representing the likelihood of Charged Off loans, higher value of 𝑦_𝑖 indicates the higher probability of a loan to be

Charged Off. Given the nature of the Binary Logistic regression, the dependent variable is the

probability of a certain event to occur; in this case is the probability of a loan to be Charged Off. Furthermore, as Emekter et al. (2015), it is assumed that n independent variables in this type of regression are linearly related to the dependent 𝑦_𝑖 (or the binary outcome variable), thus the model is as follows:

𝑦 = 𝑏₀ + 𝑏₁𝑥_𝑖1 + 𝑏₂𝑥_𝑖2 +….+ 𝑏_𝑛𝑥_𝑖𝑛+ 𝑒_𝑖 (7) where 𝑥 𝑖 is the independent variable 𝑖 (the predictor variables) and n the number of covariates.

The logistic regression of y on 𝑥 _𝑖 calculates predictor variables for 𝑏_𝑖 using the maximum likelihood method:

𝑙𝑜𝑔 ( 𝑝

1−𝑝 ) =𝑏0+ 𝑏1𝑥𝑖1 + 𝑏2𝑥𝑖2 +….+ 𝑏𝑛𝑥𝑖𝑛+ 𝑒𝑖 (8)

y could assume value of 0 or 1, indicating failure or success, and p is the probability of y to be 1 p = prob (y =1) (9)

Specifically related to this study:

Loan status = 𝑏₀ + 𝑏₁*𝑖𝑛𝑖𝑡𝑖𝑎𝑙 𝑠𝑡𝑎𝑡𝑢𝑠 + 𝑒_𝑖 (10)

Where dependent variable is the Loans’ current status, and the independent variable observed is the loans’ listing status at the time when it was funded.

The second regression applied in this study includes the interest rate of the loans as independent variable:

Loan status = 𝑏₀ + 𝑏₁*𝑖nitial status + 𝑏₂*interest rate + 𝑒_𝑖 (11)

Variables descriptions

(20)

19

the event that a loan is past due, but still in its 15 days of grace period; Late (31-120 days ) loans are those that has not been current for 31 to 120 days; Charged Off loans represent those for which there is no longer a feasible expectation of sufficient payment to prevent the charge off. Charge Off status is reached no more than 30 days after the loan has been reported as Default; Default Loans described those loans that have not been repaid due to borrower’s failure to make periodically payments within the life of the loans, and the defaulted after 121+ days past due. Defaulted data are not provided by the platform to the public, but it appears only in the borrowers’ notes status. This study does not need default data to assess whether a loan is defaulted; the final status of a loan that is going to be repaid is when it is reported as Charged Off. Indeed, it occurs when a loan is 150 days past due (30 days after Default status). A Loan’s date of charged off depends on the borrowers’ filing for bankruptcy or not. In the case, the borrower has filed for bankruptcy it might be charged off earlier based on bankruptcy notification.

The independent variable of the first regression, the Loans’ initial listing status, assumes value equal to one, whenever the loan is reported as whole loan. Whereas if the loan’s initial listing status is reported as Fractional, the value assigned will be zero. Indeed, when an investor decides to purchase loan in its entirety, the investor will be funding the whole amount required by the borrower. Otherwise, the loan amount will be divided and spread in small investments within a group of different investors, as fractional loan. This way Lending Club allows investors to diversify their portfolios in several small investments, reducing the overall lending risk. By the borrower perspective, this process would provide “instant” funding. Lending Club introduced this whole loan funding possibility because of the Institutional investors’ request to fund the entire amount of a loan. Thus, along with the request, and to insure access to equal quality loans, Lending Club randomized a subset of loans by grade, available to be purchased as whole loan. These loans are available to be funded only for a short time period (12 hours), while others are instantly available as fractional loans. If the loans were not purchased as whole loans in that time-period, they would become available for purchase in the fractional manner.

(21)

20

4.2.2 Empirical results Subsample of years 2012-2013

Descriptive analysis

Mean Standard deviation

Variable Dep = 0 Dep = 1 Total Dep = 0 Dep = 1 Total

Initial loan status 0.216 0.199 0.214 0.41 0.391 0.410 Loan interest rate 0.129 0.148 0.131 0.04 0.037 0.039

Notes: The mean and the standard deviation of Initial loan status assumes the same values in both the regressions Loan status frequencies

Cumulative

Dep. Value N. of obs. Percentage N. of obs. Percentage

0 125,684 87% 125,684 87%

1 18,208 13% 143,892 100%

Table 4 reports the results of the binary logistic regressions. This section analysed data available for

years 2012-2013, as it was previously assumed 2012 as the starting year in which Institutional investments are reported in the Lending Club Historical Notes reports. In both the regressions the observations with dependent variable equal to 0, are those “Otherwise than Charged Off or Late (31-120 days)”. Whereas observations with dependent variable equal to 1 are “Charged Off” or “Late (31-120 days)”. Furthermore, the original dataset for year 2012-2013 included 188,181 number of loans, and it has been restricted to 143,892 number of loans issued, which represent 36 months’ loans. The results show that the initial status of the loan (the independent variable) has a weak negative impact on the outcome of the loan, even when including the percentages of interest rate. The estimated coefficients are significant at the 1% level for both the regressions. The models are statistically significant as the p-value for both is 0.000. Multicollinearity is not significant, since all the standard errors are smaller than 2.

The following formulas are the Logistic formulas related to the regressions’ outputs: One-predictor regression

𝑙𝑜𝑔 ( 𝑝

(22)

21

Inclusion of loans’ interest rate 𝑙𝑜𝑔 ( 𝑝

1−𝑝 ) = −3.617 − 0.072*𝑖𝑛𝑖𝑡𝑖𝑎𝑙 𝑠𝑡𝑎𝑡𝑢𝑠 + 12.281*𝑖𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑟𝑎𝑡𝑒 (13)

Table 4 - Binary Logistic Regressions results, Jan 2012 – Dec 2013.

One-predictor regression

Variable Coefficient z-Statistic

Constant - 1.912*** - 215,342

(0.009) Initial loan status

Log likelihood = – 54631.14

- 0.098*** - 49.701 (0.019)

Interest rate included

Constant - 3.617*** - 1.153.997

(0.031)

Initial loan status - 0.072*** - 35.879 (0.020)

Interest rate

Log likelihood = – 52806.08

12.281*** 595.189 (0.206)

Notes: *** indicates significance level at 1%. Numbers below the coefficients represent the coefficients’ standard errors. Convergence has been achieved after five itineration for both the regressions.

In both the regressions, the observations with dependent variable equal to 0, so “Otherwise than

Charged Off or Late (31-120 days)”, were 125,684 (87%), whereas there were 18,208 (13%)

observations of “Charged Off or Late (31-120 days)”.

(23)

22

estimation does not represent the probability of Charged Off loans, the odds ratio to whole loans to the odds of fractional loans needs to be estimated as 𝑒𝛽1_{, and then the probability is calculated as:}

𝑝 = 𝑒(𝛽0+𝛽1∗𝑥1+⋯+𝛽𝑘∗𝑥𝑘)

1+𝑒(𝛽0+𝛽1∗𝑥1+⋯+𝛽𝑘∗𝑥𝑘) (14)

The odds ratio for the first regression is equal to 0.906 (𝑒−0.098), which means that the odds for whole loan are 9.4% higher than the fractional one (1 − 0.906); whereas for the second regression are 0.931, 7% higher than fractional loan (for the initial status variable). The probability of whole loans to be Charged Off, or Late (31-120 days) is 0.118, approximately 12%, taking into

consideration only the first regression’s coefficients. Subsample 2014

Mean Standard Deviation

Variable Dep. = 0 Dep.= 1 Total Dep.= 0 Dep.= 1 Total

Initial loan status 0.479 0.456 0.476 0.499 0.498 0.499 Loan interest rate 0.122 0.145 0.124 0.038 0.038 0.038

Notes: The mean and the standard deviation of initial loan status assumes the same values in both the regressions. Loan status frequencies

Cumulative

Dep. Value N. of obs. Percentage N. of obs. Percentage

0 144,693 89% 144,963 89%

1 17,607 11% 162,570 100%

the year 2014. As in the previous subsample, both the regressions’ observations with dependent variable equal to 0 refers to “Otherwise than Charged Off or Late (31-120 days)”. Those

(24)

23

initial status coefficient is not statistically significant. The models are statistically significant as the p-value for both is 0.000. Multicollinearity is not significant, since all the standard errors are smaller than 2.

Table 5 - Binary Logistic Regressions results, year 2014.

Constant - 2.066***

(0.011)

- 190.399

Initial loan status - 0.089*** (0.016)

- 5.575

Log likelihood = - 55738.61

Interest rate included

Constant - 4.051***

(0.031)

- 130.429

Initial loan status 0.009 (0.016) 0.591 Interest rate Log likelihood = - 53078.50 14.542*** (0.200) 72.618

Notes: *** indicates significance level at 1%. Numbers below the coefficients represent the coefficients’ standard errors. Convergence has been achieved after 5 itineration for both the regressions.

observations of “Charged Off or Late (31-120 days)”. The related formulas are:

One-predictor logistic regression 𝑙𝑜𝑔 ( 𝑝

(25)

24

1−𝑝 ) = −4.051 + 0.009*𝑖𝑛𝑖𝑡𝑖𝑎𝑙 𝑠𝑡𝑎𝑡𝑢𝑠 + 14.541*𝑖𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑟𝑎𝑡𝑒 (16)

As stated before, in the regression with one predictor, for one-unit increase of initial status it is expected a -0.089 decrease in the log-odds of the dependent variable (Loan status). The odds ratio to whole loans to the odds of fractional loans and the probability are estimated following the aforementioned formulas in subsample years 2012-2013. Therefore, in the one-predictor regression the odds ratio is equal to 0.915, which means that the odds for whole loan are 8.5% higher than the fractional one; whereas for the second regression they are 1.009, 0.009% higher than fractional loan. The probability of whole loans to be Charged Off, or Late (31-120 days) is 0.104, so approximately 10%, taking into consideration only the first regression’s coefficients. Subsample year 2015

Mean Standard Deviation

Variable Dep. = 0 Dep. = 1 Total Dep.= 0 Dep. = 1 Total

Initial loan status 0.593 0.464 0.587 0.491 0.498 0.49 Loan interest rate 0,111 0.139 0.113 0.036 0.037 0.04

Notes: The mean and the standard deviation of Initial loan status assumes the same values in both the regressions. Loan status frequencies

Cumulative

Dep. Value N. of obs. Percentage N. of obs. Percentage

0 266,918 94% 266,918 94%

1 16,255 6% 283,173 100%

(26)

25

both the regressions. The models are statistically significant as the p-value for both is 0.000. Multicollinearity is not significant, since all the standard errors are smaller than 2.

Table 6 - Binary Logistic Regressions results, year 2015.

Constant - 2.519***

(0.011)

- 226.196

Initial loan Status

Log likelihood = - 61703,66

- 0.526*** (0.016)

- 32.434

Interest rates included

Constant - 5.027***

(0.032)

- 152.345

Initial loan Status - 0.188*** (0.017) - 11.131 Interest rate Log likelihood = - 57877.95 18.644*** (0.214) 87.207

Notes: *** indicates significance level at 1%. Numbers below the coefficients represent the coefficients’ standard errors. Convergence has been achieved after 6 itineration for both the regressions.

observations of “Charged Off or Late (31-120 days)”. The related formulas are:

One-predictor logistic regression 𝑙𝑜𝑔 ( 𝑝

(27)

26

1−𝑝 ) = −5.027 − 0.188*𝑖𝑛𝑖𝑡𝑖𝑎𝑙 𝑠𝑡𝑎𝑡𝑢𝑠 + 18.644*𝑖𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑟𝑎𝑡𝑒 (18)

As stated before, in the regression with one predictor, for one-unit increase of initial status it is expected a -0.526 decrease in the log-odds of the dependent variable (Loan status). The odds ratio to whole loans to the odds of fractional loans and the probability are estimated following the

aforementioned formulas. Therefore, in the one-predictor regression the odds ratio is equal to 0.591, which means that the odds for whole loan are 41% higher than the fractional one; whereas for the second regression are 0.828, 17% higher than fractional loan. The probability of whole loans to be Charged Off or Late (31-120 days) is 0.045, approximately 4.5%, taking into consideration only the first regression’s coefficients.

4.2.3 Discussion

The regression results show that the inclusion of interest rate as independent variable does not help in explaining the hypothesis. Therefore, this study will consider the results obtained in the one-predictor logistic regression to examine the impact of institutional investors on loans’ quality, with the US-based platform Lending Club.

(28)

27

Figure 4 – Probability of a loan (term 36 month) to be defaulted and “whole” funded

The regressions’ results do not support hypothesis 2, according to which the access and the permanent inclusion of institutional investors in P2P lending platforms should have a negative impact on loans’ quality.

In 2012-2013, the extent of negative impact is represented by the probability of 12%. Therefore, 12% out of the total of 36 months’ loans funded represent overall a small percentage of the

maximum negative impact on loans’ quality. In the same time-period, 15% of the loans originated have defaulted, and 21% of the total loans were funded in their entirety. The finding is confirmed also by the statistics regarding the number of loans issued counted for each loan rating (Table 3, Appendix C). The highest percentage of loans originated overall, has been 33% rating B loans, with a range of associated interest rates of 8.24% – 11.49%, which are generally ranked as low risk investments. Thus, in the institutional inception-years, loans’ quality does not seem to significantly decrease.

In 2014, the extent of negative impact slightly decreases to 10%, which represents a small

percentage of the total of loans. In the same year, the percentage of whole loan originated has faced a sharp increase to 52.40% of the total loans, and the percentage of defaulted loans has slightly decreased to 13%. Moreover, the highest percentages of loans originated overall have been 28% rating C and 26% rating B, with respectively a range of 12.74% – 15.99%, and 8.24% – 11.49% of interest rates (Table 11, Appendix C). The results obtained in this year should explain better the nature of the relationship because whole loans represent more than half of the total loans issued. Thus, also over 2014 loans’ quality does not seem to significantly decrease.

Regarding the last period of examination 2015, the probability of negative impact has drastically decreased to 4.5%, which itself demonstrated the weak negative impact of the institutional

(29)

28

investments. Furthermore, the percentage of whole loan originated for 2015 has been approximately 63% of the total, and the loans’ defaulted has been 6% of the total loans. In addition, also for 2015, the highest percentages of loans issued have been 29% rating C and 28% rating B, with the

aforementioned interest rates’ range (Table 18, Appendix C). Hence, over 2015 the loans’ quality does not decrease.

Overall, the findings are confirmed by the yearly platform’s trend, and reaffirm the Orchard platform’s study. The “whole” loans origination for Lending Club has been raising all over the time-period analysed (Figure 5), confirming also the Wharton survey’s result regarding the preference for “whole” loans investments.

Figure 5 - Funded loans’ percentages of the total issued per year

Source: Lending Club loans data

4.3 Analysis of the effect of institutionalization on the cross-border model

Generally, the interest of institutional capital on P2P loans has advantages in terms of restrain the interest rates range and their presence represent an indicator of platform’s stability for retail investors. This indicator is represented by the perception that institutional investors have more information regarding the platform, as their investment choice depends on the platform’s

characteristic and capacity to meet the diligence requirement asked by the institutional investors’ clients. Moreover, their presence ensure that net returns will stay at an attractive level; otherwise, they would exit the platform (Oxera, 2016). In the Lendit Europe 20156, the annual conference of peer-to-peer lending, emerged a large consensus among most of the investors and P2P marketplaces regarding the relevance of institutional capital for the platforms’ growth and for scaling the

business. This has happened in the US market, as also shown in this study, while the European market is taking longer to develop, due to the effects of the 2008 crisis, which were not the same in terms of size in continental Europe as they have been in US. Furthermore, the European consumer

(30)

29

base is historically more conservative than the American one. Nonetheless, Europe is trying to achieve scale through a more regulated market, which might encourage investors to invest in the European’s loans.

In order to verify the presence of institutional investors of Bondora and to estimate the potential effects of these capitals, the study proceeded with several interviews. The first will be the interview with two retail investors of the cross-border platform, then will be reported the interviews with two experts of P2P market, and the last one reported is the interview with the platform.

Investors

Investor 1

The interviewee chosen for this part of the study is a 48 years old German lender, who holds a university degree. He has opened many portfolios in European P2P platforms. In addition, he is the editor and main contributor of a Peer-to-Peer lending website, which keeps the readers up to date on Peer-to-Peer developments.

He started to invest in this alternative online sector in 2007. He firstly invested in a German

platform, as at the time he was not allowed to invest in other countries’ platforms different from his country of residence. He then diversified his investments in several platforms. At the time, he started to invest in Bondora because the yield was high 20%, which is no more achievable because of the current market, “but it is still providing high returns which are my reasons for choosing Bondora”. According to him, the platform is still one of the fewer platforms that gives back high returns. The lender’s Bondora portfolio is dated October 2012, when he deposited 14,000 Euro and withdrawn over the time 13,380 Euro. He still has a large loan book despite the huge withdrawal because as he said “at the moment I mostly reinvest the money that are in my portfolio” which includes interests and principal repayments from 2012 until 2016. In 2015, he calculates a

pessimistic scenario where summing withdrawals, current loans and cash, the initial amount of the investment has grown “in three years to a value of 19,500 Euro”. Currently, according to his own calculation of XIRR, he estimates a return of 17.0%, which is 7.6% less than what Bondora estimated as his net return.

Institutional capital

(31)

30

stated “the challenge is not getting investors, but find a solution with compliance on the borrower side, because the borrower’s markets are very fractioned, different within Europe”. There is not a unique borrower’s regulation within Europe, so the market is fractionated. Once these issues will be resolved, the institutional capital will come automatically in the platforms, as they would perceive the stability in the European market, providing good product and high returns. Otherwise, the problem could be that because of the poor borrowers demand the platforms have to turn down some institutional money, as happened at Lending Club.

Wil you be willing to invest despite huge amount the institutional capital?

As a lender, he cares about the platform’s default risk and the equal treatment among investors. The presence of the institutional capital would “increase the lender’s confidence in the platform”,

because the institutional investors have to check the financial stability before investing their capital.

Cross-border model

According to the lender, the platforms’ home-markets are still profitable, so at the moment they do not need to expand their business in other countries. Nonetheless, he sees Bondora cross-border model as the first step of the model, “but it is a long race until they have a more unify market”, to which within 5 to 10 years’ platforms have to think about.

Investor 2

The second investor interviewed is a 27 years old student at a Dutch University. In 2011, he decided to invest part of his salary, 1000 euro from his student job, in Bondora loans. He chose Bondora as he “stumbled somewhere over an advertisement or a newspaper article of Bondora and researched it well”. Moreover, the competition was low and the platform fitted his risk profile. Even though, he had a low default rate and returns of 5-8%, in 2012 he decided to close his portfolio as he “lost interest, the surplus of money and studied full-time”. Moreover, he claims “ I was worried a little about what happens legally when I run into problems with the platform or with foreign lenders”. The student was keeping the money in to speculate and gain some returns and he was surprised of his portfolios returns.

(32)

31

Wil you be willing to invest despite huge amount the institutional capital?

As a student, he answered he would invest in if he had enough funds to do so. In addition, his decision would depends also on the platforms’ expected risks and returns. Anyway, he thinks “the higher the involvement of institutional investors the less return will be possible for a private investor”. Thus, it would depend on his financial availabilities.

The investor claims to like the idea of a platform diversified across Europe, as he is “a big fan of the European idea”, and therefore, he likes the cross-border model as “Bondora was born into the model of the European Union, and used it accordingly”. Moreover, he likes “the idea that 400€ here can do much more for a person in East Europe than in North”.

Experts

Expert 1

The first expert interviewed is the founder and publisher of a website, which works as an

informative and educational source of peer-to-peer lending industry. Moreover, he is Co-Founder of the Lendit Conference.

According to him, at the moment in the USA the P2P lending market is recovering from a “difficult year” but it is trying to maintain a “sustainable growth” in 2017.

The expert states that the involvement of institutional investors will help the platform to “achieve scale and profitability more quickly”. Moreover, “if managed well there can be very few negatives for retail investors”. This is confirmed both in the US and in the European market. Regarding the US market, he has been an investor in Lending Club since 2009 and according to his experience the platform has managed to balance the level of retail and institutional investors, that is why uit

represents one of the two leading platform in the US. Regarding the European market, the platforms have recently opened to institutional investors, and bringing the example of the UK Zopa, he

(33)

32

Overall, the expert sees the coming of institutional capital in the industry as positive, because “the platforms will be able to achieve scale more quickly”. Anyway, he pointed out a potential negative effect: “the platforms may be tempted to loosen underwriting standards as they try to deploy these large amounts of capital more quickly”.

Regarding the cross-border model of Bondora, the expert believes that institutional investors will check the financial stability of the platform in each countries in which it operates and invest in what they like. However, “this could create a supply and demand imbalance but I think these are

surmountable obstacles”.

Expert 2

The second expert interviewed has more than 13 years of experience in investment banking. He is a researcher of the alternative finance space, such as crowdfunding and P2P lending. In addition, he is founder and CEO of a company that provides online updates, opinions and studies about the

alternative finance industry, which are useful for both the borrowers and the investors. Moreover, it provides quantitative indices for the industry through a sister company which works as a consulting business company.

According to him, P2P lending is a “great innovation”, as “technology has allowed [lenders and borrowers] to be matched much more efficiently within the internet”. Anyway, at the moment the industry is facing some problems, “investors are being asked to buy loans that the originator [the platform] is not investing in itself”. As a result, “the investors need to have confidence that the loans they are buying are of good quality and the industry is seeking to give that confidence using disclosure”. Nonetheless, he states that “at the moment the standards of disclosure are not yet good enough”.

(34)

33

represent the “regular fixed income investors” needed to scale the business. Regarding the US-based Lending Club opening to institutional investors, he said “it is a good news […] the previously lack of institutional participation was a very worrying sign that there was a problem in Lending Club”. In fact, institutional investors “reassure retail investors […] doing the due diligence on the platform’s origination of loans”. Hence, institutional capital need to be encouraged because they help in scaling the business and they reassure retail investors. Furthermore, the expert underlies that the institutional involvement in the European platforms would make them more “valuable”. At the moment, the percentage of institutional investments ranges between 20-35%, lower than the

previous year (2015), when was estimated to be more than 50%. According to him, the reason of the decrease could be partly related to the increase of retail’s shares and to the specialist institutions investing in the platform’s loans. Generally, the institutional investments would improve the standards of disclosure, as they are “not likely to invest in unless they are able to review the platforms on a like for like basis”.

Overall, the expert’s opinion is that it would make no sense to inhibit the institutional involvement in the platform, as well as it does not “saying that should be only retail investors in the stock market”. Therefore, their involvement should be encouraged, despite the initial idea of a

marketplace only for peers. Hence, he would rather rename it as “either market based lending” or “online lending”.

The expert states that the cross-border model “brings complexity”, because it exposes the platform to different currencies and jurisdictions issues. Therefore, he said “the diversification is only good if it decreases risks, but also understanding the assets you are buying reduces risks […] diversification across loans is certainly good, and I am less certain about diversification across countries” .

Nonetheless, according to him, institutional investors should be “able to handle the complexity of the cross-border model” and they would “encourage the origination of good quality loans” through due diligence process.

Additional remark

Regarding the active and passive institutional loan selection, the expert underlies that it is not necessary for them to do active loan selection because investing in microloans gives “insufficient reward from doing due diligence”. In addition, there is the “misconception […] that institutions are being given the option to do active loan selection”, instead, “they are offered a randomized

(35)

34

avoiding the “cherry picking” loans selection, or active selection, in order to have an equal allocation of loans within the investors.

Bondora

The Bondora’s interview was conducted with the investor relator’s employee. According to

Bondora, P2P market is facing a decrease in the growth rate compared to the previous years, both in the US and in Europe. Nonetheless, “the number of players has grown and different market niches are now covered with peer-to-peer opportunities. In some markets, there are too many companies already and some of them might not survive”.

Bondora’s opinion on the institutional investment is positive and “growing share of institutions in marketplace lending is natural, as it is now definitely a separate asset class that needs to be looked at in the composition of investment portfolio”. Specifically regarding Lending Club, because of their size they had to open to non-retail investors in order to grow. Furthermore, their developed secondary market has allowed institutions to have “direct investments to loans” which “were more natural than in other markets, as there is clear exit strategy and earning opportunity from yield compression”. With respect to Europe, “more [institutional] involvement would definitely support growth and there could be significantly more volume in Europe”. Nonetheless, their involvement has been less than in the US. The only market, which experienced much more institutional participation, is the UK market.

Bondora platform

The employee states that at the moment Bondora is “doing well, and revenues are growing at the rate of 70% per year and monthly originations are well over € 3 million”. They are open to

institutional investors, and they have dealt with them: “our cooperation with institutions so far has been positive and we expect to see more of them among our investors”. At the moment, “the share of institutions is less than 10%”, but the platform states that “the number of institutions we are having discussions on is significant”. Overall, the employee confirms that the future involvement of institutions would help them grow and explore other market segments.

Additional remark