 We have today our keynote speech and afterwards the panel discussion and It's my pleasure to announce to you the keynote speaker of today Professor Ming Feng Ling who came all the way from the Georgia Institute of Technology I think it's a 14-hour flight. So we really appreciate the effort of flying over to Berlin I thought we make it more appreciatable since tomorrow is the 30th anniversary of the German reunification So you can actually do two things come to the conference and also attempt this very historic event So Ming Feng is an associate professor at the Georgetown Institute of Technology He's at the business school and the Department of Management Information Systems He did his PhD in business and management from the University of Maryland At the University of Maryland. He also earned his MA in economics But he also earned another MA in economics from Peking University Where he also received his BA. He has a lot of fellowships at different universities I just named one of the most important. So he's also a fellow as Cambridge University varies at the Judge Business School and the Center for Alternative Finance and Ming Feng was one of the first scholars I have to say I found Publishing in the area of peer-to-peer lending and crowd investing and crowdfunding in general in particular in 2013 you had your first paper on management science already on peer-to-peer lending and so that was really the start of the literature I would say and Since then you didn't stop. So I just counting your iterating in your publications It's either management science you have five by now and four information system research And there are a couple of them already in the pipeline again I would say if I have to make a predictions probably the next one is information system research because now it's then it's five to five I would say and so Your honors and awards list is very very impressive. So in total I'll count it 33 awards and honors and So I cannot name all of them just I would suggest to you to see the website of Ming Feng and then you will be very impressed So among others he received the best paper award of management science also for management science he became the Distinguished Service or got the Distinguished Service Award and I must say there are a lot of productive people here in this room and I really appreciate that Losses among one of them I would say and but Ming Feng is really topping topping the edge I would say so he's really one of the most productive people I know Today he actually talks very much on the topic or the overall theme of our conference because his Presentation or his keynote speech today will be about something borrowed something new from peer-to-peer lending to real estate crowdfunding and beyond and we are very much looking forward to your presentation and the floor sewers Thank you very much for coming to Berlin. Thank you so much Lars for the very nice introduction And also for the honor to be here to share my work with you guys and thank you all for being here today So as Lars mentioned I've been working on peer-to-peer lending for quite a while Never counted actually. Oh, it's almost like 12 years now since when I was a student At that time it was quite uncertain how far it's gonna go and I'm gonna share my personal story in that period But if you guys know like today, it has become since the time I was a student It has become a global phenomenon from the US China and then UK Germany and many other places as well around the world So it has it has produced many many great success stories, but also produced some horror stories So like so that's why it is the Chinese government crack down on peer-to-peer lending a few years ago So but back then it was just started So this was the homepage of prosper when I was starting my PhD dissertation It looks very dated as it should be because that was back in 2006 when they just started so for me at the time because I came from an econ background so we are always looking for data and This one seems to be a very promising kind of arena So that's why and also it seems really exciting. I just got really excited by the phenomena itself So I collect the data study processing it and then try to formulate research questions based on that data set But at that time it was actually not all peachy So in fact, I was quite stressed out to be honest for two reasons one is prosper itself at that time It was just starting so there was a lot of Good news coming out from day to day But sometimes they're also bad news like November 24th 2008 SEC the security Exchange Commission in the US send a cease and desist letter to prosper to make sure they want them to shut down So that I saw the news and then my heart almost stopped So and then like two days later They got sued by a class action lawsuit by the investors probably following the cease and desist letter And then they also run into some troubles along the way, especially they are ups and downs in the fundraising process so At that time it was a startup and I tried to come for myself say that hey even if it goes out of business It's still real money invested by real people on the real marketplace So even if it goes out of business, this is still good place that generates great data that we can study as researchers So that's one thing that's kind of trying to come for myself I try to encourage me myself to keep working on my dissertation But there's another source of stress, which is the fact that it's public data So because everybody's looking at that I think there are many other people like in the finance in finance area also marketing area They're also looking at the same data set So as a student with not much resources to myself that was always a stress I my biggest fear is that because actually that was the time I was about to go on the job market and my biggest fear or nightmare would be that wake I wake up Sunday and some people published or Submitted something exactly the same idea as I was having so fortunately that didn't happen So but I always try to come convince myself that after all these papers that the first two papers I was having my dissertation. I should get out of this data set So but after all these years Prosper has changed a lot. Okay, so they have gone through many many ups and downs as well But for myself, I often somehow just keep coming back to prosper data set. This is so for several reasons I always find myself coming back to the same data set I've been working on So even today like a few weeks ago. I have met a friend of mine who was a student in Maryland We haven't met each other for 10 years I think and then he asked me what are you working on today as I'm still working on prosper data in addition to other data sets So but there are several reasons for this They are always coming back to prosper data for several reasons. One is the fact that the data is extremely rich There are two reasons for this one is that it is from a debt based or loan based Profunding context and as loans as you know, it's a heavily regulated industry and also is a very data rich industry So you have a lot of information about the borrower and so on so even from the traditional bank You have that information But here is actually not just that the nature of the context But also the fact that prosper made the data publicly available for a long time So from 2006 until 2013 they make the daily snapshot of all the entire platform publicly available So that was essentially the data source that I have been using So I was using the data set for many many papers. So this is just a yard diagram So it's not not just they didn't just provide the data for anybody to look through they actually provide great documentation which was a really fantastic way for us as a students at the time and So even though I have newer data, I think in many ways This is still very very unique and very helpful in many ways Okay, and there are other reasons the other one is that because it's a loan based crowdfunding It's a loan with a fixed maturity date So for us, we feel like there's a one of the biggest benefit is as a very objective measure for quality So for whether you are especially if you are a lender, you're trying to decide who to invest in For loans, you have a very specific date at which you can look back and say whether you made a good judgment or not So if you mess in somebody if you blend the money to that person three years down the road You know for sure whether that was a good investment or not once once the loan matures whether defaults or paid back Okay, so that was the objective for quality. So if you look at rewards crowdfunding that based crowdfunding I have some of those data sets as well But that's always been a struggle how to define the underlying the quality of that campaign So for loans, this is much less controversial So you can actually just look at that and then you can calculate not just whether it's repaid You can also look at for instance the return of investment for investors point of view Okay, so that's really the nature of that data that makes it much amenable to answer a lot of the research questions Especially from economists point of view when we're trying to answer information asymmetry issues The second the third one is the fixed time horizon. So for equity crowdfunding, which is great for businesses It goes on could go on theoretically forever, but for loans is a fixed date, right? So by that date you can actually make a determination and then the last one is actually not the least is the rich history Prosper has gone through so I've often found myself using this a lot because I've been using the data set for so long I think I know quite a few things that people probably wouldn't notice if you're not working on the data set So possible has gone through a lot of changes in their policy So many of the work that I do actually uses the natural experiments either policy changes from prosper or like the sec Shutdown prosper also use that as a natural experiment to help me tease out the causal influence in the data So that has always been a valuable thing in the publication process I think is as you guys will know, it's kind of difficult if you don't have a clear identification reviewer as pretty tough on that. So So the so these are the reasons that keep finding myself going back to prosper data. So these are the Some studies have been working on so far using this data set and the reason I wanted to bring them up today is that I Think even though the research are done on the prosper data set, which is on the peer-to-peer lending or debate that base crowdfunding I think at least to some extent some of the findings could be useful in informing The decisions in the other core funding market such as equity crowdfunding rewards Call funding or even real estate crowdfunding and even future innovations in this field So this will be some of the things I want to talk about today. I'm trying to summarize my research has been done so far So some of them are have been published some of them has still ongoing and so I think there are still some things that we can We can work on from the data set and I hope that the conclusion that I drew from those places can also be useful in in in the in other contexts, so Actually some of these questions like the last one here on the secondary market We actually when we started out we were trying to answer that question in the equity crowdfunding market and But when I talked to my co-authors actually we decided that actually it's better to answer that question Because of the nature natural experiment that we experienced. So that's why we actually come back to use the prosper data So I classify my work in this area into three broadly Broadly defined areas one is the primary market design meaning that the features of the primary market where the borrowers and lenders Interrupt each other. So essentially we're trying to understand the features that platforms roll out to try to solve the Isometric information problem between borrowers and lenders So this is different features that platform has and we're trying to understand and whether they have influence in mitigating that issue Trying to make the market work. Okay, so that's the the the one thing So for instance the first one which the things that we the underlying are the ones I want to spend a bit more time on The first one is the market mechanism. So that's that's how the market is actually organized So for any platform, that's probably the first question. How would you decide? How the borrowers and lenders or even in equity crowdfunding how the investors and entrepreneurs should be matched to each other So that's the market market Mechanism that was referring to the second part is the individual difference between investors. So for that one This is essentially the heterogeneity of my investors So we also look at institutional investors versus noise traders and then also some other information some other angles such as local investors and information about bits and then the last one is the secondary market and so so this is some of the Studies that we've done so far each one of the bullet points can probably expand to a whole hour of talk Which I won't do I promise so I just want to highlight for each one of them I just for the three things that underline I just want to give a one slide overview and then so we can talk more I can go into as much detail as you like me to Okay So the first one is the market mechanism this is also using prosper data and What we did at that time was to use a natural experiment on prosper when they Changed the whole platforms design from an auction format to a post-it price format So in the auction format is that you as investors when you come to the website you can look at the options and then you can decide which one to bid on and then There's an auction process meaning that as funds come in if it exceeds the amount that borrowers requesting the interest rate could potentially go down So from an economics point of view it seems like a very good mechanism Right you can actually reached probably the one with the least regret from both sides So it because you can essentially like a top-to-top momentum process auction Clearly you can find the market clearing price for each borrower But the downside of that offer obviously is that it takes time for auctions to complete So that's also one of the reason that prosper switched to post-it price in which in that kind of context in post-it prices Prosper would come up with the decision on how much each borrower should be borrowing at so they actually priced along For the lenders and then the lenders decision when they come to the website will be simplified in the sense that you only need to Decide whether you want to be part of that loan or not So you cannot really change the interest rate even if you are willing to provide more than they are asking for so that's really the idea So always so that's the original assignment the The announcement that they send out is that they're trying to convince the investors that this is a good thing Which I think this is true because if you are feeling of thinking from the platform's point of view The speed of funding is definitely important right because you make profit when the loans have been funded But what we found using that natural experiment because they shut down the the website and shut down the The auction based format and switch it over to post-it price We did it before enough to comparison using that as a natural experiment and what we found is that yes The funding speed did increase meaning that it is much faster than before to get the loans funded Okay, but on the other hand the default rate is actually getting higher and what we How how we reached that conclusion of course empirically is always that's what we found but theoretically In our model we found is the intuition is actually pretty straightforward meaning that in the past When the borrower Borrow money they start with their asking interest rate So they come up with the the original for instance I want to borrow money at ten thousand dollars starting at twenty five percent So I know that if I give that interest rate too low I may not get the money But if I give the interest rate too high then I'm going to be stuck with that rate for the whole life of the Long so I will be less likely to give you a very high rate Okay, so I will be trying to be conservative in deciding where the starting point should be for the auction But for the platform, they know that then the borrowers will have that kind of a reservation So they actually take advantage of that because if the price comes from prosper the platform is going to carry a much more Important or much more impactful is impact Effect on the investors because they know that it's kind of a platform's decision-making They have all the data about the borrowers So they should be able to be able to come out with an interest rate that is appropriate for the borrower so because of that lender should be able to react more strongly to the Price that prosper give to a borrower. Okay, so for that reason Prosper prosper as a platform could command that as a premium They can actually charge a higher interest rate than what the borrower will normally ask for at the beginning if they were to price them along themselves So this is also what we found in empirically So again the funding rate is higher But the default rate is also higher and because of the default rate is higher the overall return of to investment for investors It's actually decreasing at least under some circumstances and we also did some welfare analysis Looking at the whole social welfare from the platform to the borrowers and lenders and at least under some Circumstances the default rate will be the the overall social welfare will be decreasing as well So meaning that the overall message that we took away from that exercise is that even though from a platform point of view It may be better to use a post-depress mechanism. It actually comes at a cost that may not be obvious otherwise meaning that it could come at the cost of a higher Likely of a default. So that's something that we want to highlight and this is also highlighting the fact that there are so many stakeholders in the Craw funding market no matter what kind of crowdfunding market it is Whoever has the pricing power actually has a lot of impact on the market outcome So that's I think that's one thing that we want to take away from that study. Okay The second one is Going back to the three things I want to use this one as an illustration for the differences between Different kind of different investors As you guys know, they are both Retail investors who are just like playing around we call them the noise traders in the literature meaning that they don't have a whole lot of asset They are tend to be more Discoverers and then they tend to not react too well into the information that's in the market They are also institutional investors in the prosper So what we have been working on a few projects trying to understand the relationship between the two and also how they compare to each other so I had two studies here. One of them is the effect when prosper study labeling institutional investors I mean in that before some investors some institutional investors could participate But they are just not labeled as institutions and after a certain date They become labeled as institutions and the bidding history is visible to retail it to everybody on the platform And we found out that after they become the label was shown to retail investors The retail investors definitely had a significant reaction to the institutional investors In other words the fact that an institution shows up in the loan that actually carries a very significant impact on retail investors Even though that that actually ended up to be a not a very rational kind of decision for them to make So that was one and then the other one which we thought was quite Interesting it was based on a prosperous mispricing. So which we were just talking about the Prosper pricing the loans But there was actually a period of time between December 10th of 2010 and July of 2012 when prosper had a mispricing meaning that For people who are very different In terms of the underlying credit risk, they should have very different interest rates But because of the pricing scheme that they had at that time, they will actually charge the same interest rate So we use that as a this is one example. So The the interest rate was charged based on a number of factors. One of them is the credit score and so on So for instance this in this one example one borrow has a credit score between 640 and 660 this is the credit score in the US credit system. So We do not know the exact credit score, but we know the range the 10 point 20 point range where they are for into So we can know that borrow once credit score somewhere between 640 and 660 and then borrow to credit score is 760 and 780 so by any kind of indication the second borrower should be a much less credit risk So you want you lend money to them you should give them a better interest rate lower interest rate But during that period of time, they actually both of them were assigned to see at least this is actually what we observe in the data and because they are assigned to the C credit grade in the Prosperous pricing scheme. They were both charged the same interest rate, which is 21 point four eight percent. So that was a very unique period and the actually went on for quite a long time actually So we use that as an opportunity to identify the behavior of institutional investors and retail investors Meaning that when you face when you are faced with opportunities to in a sense to arbitrage Are you able to take advantage of the opportunity? So what we found is that for loans that were underpriced you do You do see or actually overpriced meaning that charging a high interest rate You do see a higher presence of institutional investors So meaning that they are actually able to take advantage of the information much more bad much better than retail investors would so retail investors They actually face with the same set of information if they had if they even look at the platform They should know that this is actually a bad idea meaning that for people in the borrow one case If you are thinking about these two you should go with borrow two for sure, right? But actually a lot of retail investors still go with borrow one So we use that to as opportunity to identify Trying to distinguish between the ability of retail investors and institutional investors. That's always being a question in the literature Whether smart money is really smart. Okay, so that's so that's essentially the question We're trying to Understand we pay them a lot of money to manage our money So whether they're actually able to deliver better performance so we what we found here is that indeed Institutional investors are able to Take advantage of the information they do invest better So if you look at their overall return is actually higher, but the difference is actually very very small So Michael other Rick's eyes as his abundance Professor in Arizona. He did a very interesting calculation. So it's just like a mental exercise So if both retail investors and institutional investors Start with equal share of the market and then we know that they have a difference in the performance How long would it take them for the retail investors to be completely irrelevant? Meaning that their their share of the market will go down to less than 10% and his calculation is that it takes about 400 years Meaning that there is a difference, but the meaning the difference is actually very very small So that was one thing that was kind of interesting and the uniqueness of the data is that this has been a question That's been going on, but we have a very unique opportunity in the data to understand the ability of the retail investors versus the Institutional investors. Okay, so so this is the second part which I think is It's also very important part for any kind of Coral funding especially in equity Coral funding. You have some people who are really experts They are not do a lot and some people who are they want to Get a share in the future Facebook and so on but they may not be able to process information That Correctly, so the question is to what extent we should protect them Okay, so as your retail investors, it's actually very interesting in the finance literature Even though they are retail investors There are noise traders meaning that they are almost like acting on noises. They're very sensitive noises It is actually very essential for the liquidity of the market So you don't want a market where they are only institutional investors participating So I think this is kind of where it's a give us a lot of things to think about to what extent we want to protect them So this is actually how? Institutional investors and retail investors can interact with each other. So this is actually has some implications for other Coral funding for mass as well. Okay The third one which we were looking at again, we look at the three parts one is the primary market design the second one is the Investor heterogeneity and then the third one is the secondary market So I think this is actually another amazing development in the fintech world in the past decade So in the past we're trying to figure out how the primary market works I'm not actually stepping into the secondary market because people want liquidity, right? So so this one was really kind of a lucky thing for me because I was just collecting data on prosper Back in 2016 I think for a while and then they suddenly send out an announcement because I'm an investor on prosper So they said that I saw I forget the exact date We're gonna they are just an email saying that on On the certain day, I think three months in the future. We're gonna shut down the secondary market So because it has not been used that much So I just kept on collecting the data and then we're trying to I try to understand how the shutdown of the secondary market affects The primary market, so it was kind of a lucky thing that was collecting the data before they actually shut down the market but at that time I also knew that this is not a question that Folks in my field I mean in the information systems field that they will be interested in because it's a very finance question But it's very lucky for me that actually Got into collaboration with quite a few finance co-authors and we actually got this project up and running very quickly So this one has been We almost have a working people ready So it's a but this is what we are trying to do is trying to understand when you shut down the secondary market in the sense Whether there is a secondary market or not how that affects the primary primary markets liquidity So the liquidity in terms of the funding time how much time it takes for you to fund the loan the cost of funding and then the total Total amount that can be funded. So that those are the typical definition in the finance literature That we use to define liquidity and what we found is that even though this is the interesting part to us I think at least even though prosper the reason that they shut down the secondary market is that not many people are using it But what we found is that even though this is a nominally used minimally used secondary market the fact that exists Still provides the comfort that investors need to trade in the primary market So when you shut down the secondary market, even though it was almost nobody was using that using that it still has a very significant impact On the liquidity of the primary market. So that was the the main finding and again I think it's pure luck that actually I came right into that natural experiment So we study how investors react to the announcement and then the actual shutdown of the market Okay, so so I think secondary market is always a very interesting thing I think especially for loans right it because many of the loans are three years in land or five years in land So what if something happens to the investor? They want that liquidity and the same thing for equity and also other kinds of Co-funding especially like real estate crowdfunding as well So I think understanding better how secondary market works and how investors react to the secondary markets existence It's a very important thing for many of us to look into. Okay, so so those are the three things that I want to highlight And there are other stuff that I don't want to go into but that's there. So okay, so Those are what I call something borrowed borrowed because it was in crowdfunding for loans but there's also More importantly for this conference I think it's something new in in the sense that how we can use something that we gather something that we found in Reward in the peer-to-pea lending market to inform what happens in equity crowdfunding real estate crowdfunding and so on So I think there's these are really new phenomena I think it's really interesting that people actually we can actually do a lot of things with equity and real estate and so on so So a lot of exciting developments I think particularly in Germany here and also in the US I think there has been some growth in the real estate crowdfunding market as well equity crowdfunding obviously after the jobs act has been picking up as well So but I Don't think the findings that we have found the peer-to-pea lending will completely apply to these new markets for several reasons One is that so I have been collecting data on rewards crowdfunding such as Kickstarter and Indiegogo and also equity crowdfunding Such as some of the platforms in the UK as well But the challenge that I always found because I'm as academics We have to write papers for the reviewers first, right? So for us, I think at least the most challenging thing is the fact that there's a whole lot of arts Element in rewards crowdfunding equity crowdfunding and even real estate crowdfunding I call them arts because every single project is different so for That crowdfunding is a little bit easier because loans are loans We can package them at the end of the day is just a bunch of numbers and it's almost like something you can package But for this is just a screenshot. I took from Crawl Cube yesterday You can see how varied they are they go into all different kind of industries and so on So if you really want to do a project level analysis To convince reviewers that one of them is a counterfactual of another that is very very difficult to do You can control for a lot of things about each project the funders and so on is I think is Objectively speaking, it's very very hard to find something that's truly a counterfactual of each other So that creates a lot of problem for for instance the the causal inference that we're trying to make for academic researchers So so that's at least one thing that I think is a challenge if we want to general generalize some of the findings from P2P lending to this new context Okay So this is one and for really say real estate I think it's probably even more challenging because every house is different So that this is even gonna be even more interesting There's a lot of numbers, but how we use the numbers to actually make causal inferences That's gonna be a very interesting and worthy of research obviously The second one is the use foot not just for loans, but also for other caps types of crowdfunding There's a lot of work on predictions as well So for instance, you can use prediction to detect fraud activities and so on and they are also Algorithm tree trading has been developed for peer-to-pea lending and I'm sure that as the scale develops There's gonna be similar developments for reward for equity crowdfunding and even real estate crowdfunding So the use of algorithms and the machine learning in this field I think it's gonna be opening up a whole lot of Opportunities on the other hand as we all know this could be potentials for biases, right? So for instance, I think just the what we saw in the news when you use a machine to train Data on how to hire people that will translate into the bias that you don't want to have in the hiring process So this is actually from a long time ago when landing club was I think it was I Think it's the second iteration of landing club when they're trying to create a match between borrowers and lenders Which was very interesting But they also have a what they call the secret sauce how to connect the borrowers and lenders to each other So so algorithm is actually not not something new by shipping going on for a long time But on the other hand I think there still needs to be a lot of research on how these algorithms or the predictions Effect the market participants and also the competition environment in this different context I think that will be a very important thing for us all to think about So obviously for instance in loans You could use a lot like in some of my work we look at text, right? So how people write when they are requesting funding It may seem very Inocuous that you are just trying to predict whether somebody's going to default or not But the by writing itself sometimes could reflect some flag reflect some of the protective information for instance whether The gender the race they all could have a bearing on how you write so if you develop an algorithm to To predict somebody's default likely who based on what they write How is that going to play into the fact that you need to make sure that your lending is done fairly and not Violating any of the red lines. So I think it's not a very obvious question. There's always a big trade-off between the Avoiding discrimination versus achieving the market efficiency that we have auto think about and I don't think that's unique to That crowdfunding but also applies to other kind of crowdfunding markets as well. Okay, so This is my last slide and with a lot of challenges. I think they are still some We are here because of the challenges, right? We are if there are no challenges there, there's no need for us to do research So but I think the more important thing is actually how we can proceed So I think these are three for my limited experience. I think these are some of the things that are very helpful One is the discussion and the collaboration between different stakeholders in the in the in the market So I think this conference is precisely a very good Representation of that kind of effort. The second one is the interdisciplinary research I'm very lucky to be in the field of information systems So this is a relatively small field But I get to work on whatever that I like to do and I can reach out to different experts in different fields So I have collaborators in I think almost all Disciplines in the business school. So it's very I think it's also very helpful like even statistics and so on you can use different Angles to look at the same phenomenon. So even for P2P lending data I think these days have been looked at by computer scientists statisticians and so on so I think it's we once we have the data is very easy to have a Interdisciplinary approach to tackle the problems together and then the last one of course is probably More difficult just like what I mentioned before We always have to balance between the need to protect and also the need to experiment So this is not very easy to do but I think it's ultimately How much do we want to Experiment and how much we want to protect the person people that would need to be protected such as smaller investors So again, that's what just my Something I wanted to share with you guys based on what I've what I've done in the peer-to-peer lending market And I hope that some of them will be useful for Newer iterations of fintech innovations such as real estate crowdfunding. Okay. Thank you all very much