Long Chen, Director of Luohan Academy, he is also executive provost of Hupan University of Entrepreneurship, presented at Luohan Academy's Frontier Dialogue. Robert Townsend, Professor of Economics at MIT discussed Long Chen's presentation. The following texts features transcripted excerpts from. It has been lightly edited for clarity and length.
Speaker presentation by Long Chen
Good morning, good afternoon, and good evening, depending on where you are. I'm so happy to share my thoughts after hearing several other scholars talking about what data information can do to finance. What I'm talking about here is some joint work by my fellow colleagues at Luohan Academy, including some of our fellows we work together.
We all know the value of finance. Finance in nature is a service of individual information. So the two keywords, information and individual, corresponds to two key obstacles of finance, information cost and the lack of scalability. Now we look at current financial system, and see how it deals with these problems.
First, financial accounts. Globally, even very recent, there are still about 1.7 billion people who do not have bank accounts. Now, quoting the seminal work by [Narayana] Kocherlakota back in 1998, "money is memory"1, I could argue that a lot of people do not have financial accounts because they do not have trustworthy information being measurable and recorded in account. So cash and collateral are still are the hard currencies.
Next, business lending. even in the United States, about 60% of the commercial and industrial loans are secured. A large chunk of the SMEs are financially constrained because they do not have enough collaterals. The progress of the micro loans is not advancing as far as we have hoped, presumably because the cost of collecting and verifying information is not very scalable in a traditional way.
Then consumer lending. Similarly even in the United States, about 85% of the consumer loans are secured. The part of the non-secured loans of the consumer credit is very much related to the payment that requires a high cost of collecting and processing information.
Finally, wealth management. Piketty's point is well taken that wealthy people get wealthier because they have money to invest. The average people, they are under invest. One key reason is because of a lack of trustworthy information.
We can see that information has been a major obstacle in finance, because finance is a service based on individual information. Now, the question here is what the technology can do in this regard? I'm going to make four claims. I think they are major trends that can describe what's going on. Then I'm going to provide some evidences to verify those claims.
Technology and New Finance in the Digital Era | May 25th, 2021
My first claim is that there is an information revolution. There's a breakpoint around the arrival of the mobile internet. We've been talking about the information revolution for a long, long time, of course. But I think what's taking off is the arrival of the mobile internet, because right after that, both the consumption and production behaviors becomes much more measurable and tractable across space and time and that goes with the rise of the digital finance.
The second major trend is the capacity transformation. Starting from the mobile period, the digital revolution deeply transforms the accessibility, the risk assessment and the risk management of finance, so that both the supply and demand of finance become much more real time. So that's the capacity transformation.
The third is the integration of finance with real economy. So the rise of the digital finance goes with the pace of the digitalization of the real economy. But of course, the interesting thing is that it doesn't exactly take off as Darrell mentioned earlier, because the pace of the digital penetration is actually an art. It depends not only on technology. It also depends on unsatisfied demand and habitual resistance from that country. So it could be very complicated.
Finally, I think a major trend is the open finance. The rise of both the digitalization finance and the real economy requires that many financial institutions and the FinTechs have to work together, because they have different strategies to serve the same customers. So open finance is becoming more and more inevitable, even though there's a lot of competitions between them, but there's really a lot of needs for them to work with each other.
Now, let me give you some evidences we're seeing in China. Let me first talk about mobile payment. Now, if we look at the first graph on the left, you can see that in the past six and seven years, actually it's Asia’s non-cash payment, which is the mobile payment really grows the fastest. So it's not every continent that grow exactly the same speed, but I think that's led by China, India, and some other Asian countries.
Now if we look at China, then you can see that the right-hand graph is the breaking point I'm talking about. Back in around year 2010, 2011, China's e-commerce as the portion of the retail is only about 1% of the retail sales in China, which is actually below United States and many other countries. At that time, China's mobile payment is very similar to United States, around 15 billion, and US is about 8.3 billion US dollars.
But then China in the past 10 year, also is just really took off right surrounding the mobile internet period. So I agree with some of the previous speakers. This is particularly meaningful for the developing regimes. They can pick up the technology and grow faster. This is something called inclusive growth. That's the mobile payment.
Another really interesting story is the battle between the QR code and the traditional point of sales (POS) machine. Now as for your traditional payment method, you really have to have point of sales machine that tries to collect information from different parties and verify the information. It's costly and requires every merchant to have the deposit above several thousand Renminbi to have that post machine in China back in 2011. For every transaction, it charges 1% to 7% of fees. Now, if you use the QR code, you have zero collateral, and the payment fee becomes much lower and it's real time settlement. So that's why even beggars now use the QR code because that's the modern POS machine.
But it's not only just that it's much cheaper for the merchants and the customers. Actually, it's much safer too. Using alternative information Manju and other speakers just talk about, you have the real time risk management. If you compare it to the traditional payment methods, of which the fraud loss rate was about 0.7%, but if using the mobile payment, the alternative data, real time data and risk management,then your default rate becomes much lower, lower in several scale. So that's why I haven’t met a single person in my life that has a loss happened in a mobile payment.
What I was trying to talk about is that information and digital technology really have changed how payment is being done and become much more inclusive. But not only that, because in the beginning I talked about the mobile financial account that really means that it has a remote identity that is transparency, so it can spur a whole list of innovations include the sharing economy and unattended retailing, even buying things. For example, an interesting paper by Ouyang, who is a PhD candidate from Princeton and has been working at the Luohan Academy for about two years now. In this graph, you can see the interesting question here is that, if you want to develop the mobile payment, should you start with the supply side or demand side? So here that is looking at the horizontal axis is the number of bicyles in a city. If the bicycle amount is very small, then it's the activity of the users doesn't take off, but really it was the supply side of the bicycle, when it reaches to some points, the whole thing just takes off.
When the customer becomes more active, the effect of the data spills over to the ability for them to get credit. So that's the wonderful part, the footprint that brings more additional financial availability in this particular case. So that are some examples on the payment partners.
Now let me say something about SME lending. In the traditional way, we use the hard information, structured hard information like the credit score Manju just mentioned earlier, but what we're really seeing in practice is that a lot of the efforts have be used to change and find a hard information in soft information using the alternative data, and also take a lot of the alternative unstructured information into the structured data information. So in that way, it becomes much more useful. You can call this the deeper alternative score. Because that method is actually very scalable, it becomes a new way of providing the credit service that is scalable and sustainable.
In the past 10 years or so, MYbank has served more than the 35 million SMEs and half of them are women entrepreneurs. The percentage of credit divided by the scale of GDP is higher in the less developed regions. It is really inclusive in this way. Also, just a couple of the speakers worry about that if you have kind of AI to screen, the customers would lose and lead to discrimination, Thomas Philippon2 had a similar concern. If we look at here, the age group, education levels, gender group. We do not see that happens.
Now, this is other paper by Professor Huang Yi and he shows is that those SMEs, because of the alternative data, they are able to get the lending without collateral as Bengt Holmstrom had claimed, "Information as the new collateral". So you can see that it is very beneficial to those firms that are able to get those non-collateral lending because of the data. Finally, what I'm showing here is that this is not just providing credit to those SMEs, the ones who are lacking the collateral, but actually it's safe. You can see that a 90-day delinquency rate is clearly below the comparable banks, like rural commercial banks, or the city commercial banks. It went up a bit during the COVID period, but goes down soon because of a real time risk management. So you can see that this is a sustainable thing to do things. Now, let me spend a little bit more time to talk about consumer lending.
The consumer lending is very similar. We can see that the left graph here is the shock to China's consumption during COVID and then kind of recovered after that, earlier last year. If you compare to the righthand picture, you can see that it really benefits the people who are using it because compare the people who use and who does not use credit pay can clearly see that the ones using it get hurt much less during the COVID period. So that's on the demand side. On the supply side, it's actually the credit pay. It connects the supply with the demand. On the supply side, if you compare the merchants who accept the credit pay with the one that do not accept, then clearly the ones that do not accept get really hurt.
If you compare who has higher penetration with those with the lower penetration of credit pay, you can see that the ones having a lower penetration got hurt again more during the COVID period because of the lockdown. This graph shows that if you compare the regions, the regions with higher GDP on the right-hand side tends to accept and use the credit card even more, whereas the regions with a lower GDP on the left-hand side accept the credit pay using alternative data even more. So it's really complimentary to the parental financial system to make it more inclusive.
Finally, I want to say something about open finance. I think that's also very exciting. By the end of last year, there are about two hundred institutions on Alipay's the wealth management platform, 170 million investors, and 51 million users have watched live stream shows. There are 6,000 professional investors advisors on the platform in which thundred millions of contents and videos produced, thus connecting the investors and financial advisors with the financial paradox. As far as the first-time witness, I've been lived in US for a long time. I think it's first time to see that people have money like $10 and they can have a real time financial advisor to help them go through up and down. So investors can get help from education contents, discussion forums, news analysis, selective products, and also robot advisors which together are the platform we see nowadays.
Now you can see that these are the contents put out by the financial institutions. They spent a lot of time to make their contents interesting to attract their own investors and build their own communities on the platform. Next, every Chinese know that livestream is extremely popular in China so all those professional analysts help the customers who have very little or a lot of money. It tells them exactly what's going on to help them to boost up and down. Also, they have their own online forums for the investor to talk about what they should do, what they should buy and so on. Also they have the robot advisors, they stay doing very cool things to help them start right and help them improve their financial literacy for decision-making.
I'm trying to answer the question of what information revolution is doing to finance. So I make four claims. The first claim is that it's really taking off around the rival of the mobile internet, because at that time post consumption and production behaviors become much more measurable and tractable, and it was the rise of the large scale of the digital finance.
The second part is consistent with previous speakers, a lot of alternative data converts soft information to the hard information, unstructured information to structured information, deeply transforming accessibility risk assessment and management of the finance. Thus, both supply and demand become much more real time and the digitization of the finance is being increasingly integrated with the real economy. Of course, it depends on the country.
I agree with previous speakers that this is particularly wonderful for the emerging markets where there's a lot of unsatisfied demands and more attitude to embrace digital technology. Finally, because of this integration between the real economy and finance, only finance is not enough and we need a lot of technology and connectivity, which means that the financial institutions and Fintechs have lots of rooms for them to work together. An open financial system has become inevitable, even though there's competition. This is the future happening right now.
Discussant presentation by Robert Townsend
Thank you very much for asking me to participate. It's really a great event and a panel, and particularly to discuss Long Chen's paper. I wanted to pick up on something related to what he had said and his talk today, which is the big deal about payments. So, one premise - it's really all about the data. If you think about digital payments, it's a log of transactions like a cash flow statement. Indeed, that's exactly how we create ledgers in financial accounts. By having the underlying transactions data, you're going run checks, the balance sheet becomes integrated with the income statement and so on. So you can actually get, in principle, the financial accounts out from digital payments.
This has various desirable features. First of all, changes in the stocks of assets would be consistent in the flows whether or not you own it and if you own it, did you use it and so on. Shockingly, many famous and deservedly well-known US micro surveys, absolutely none of them meet this standard in terms of the consistency of stocks and flows. Another good feature of the payments data is that you're reconciling transactions across parties and doing it instantaneously. Now to harken back to something Durrell said, this actually gets harder when you have multiple payment providers, because you will have multiple databases, so one can think the big picture, which is how do you create better information infrastructure in the US and other places. I would propose that there are many insights available from data science if we're thinking about better designs that deal with synchronous versus asynchronous or something we've got used to from Zoom and lecturing. But it has a different meaning here.
Centralized versus decentralized systems and the difficulty of scaling as in completely centralized. Computer scientists refer to these as CRUD systems where only delegated people have the authority to create, read, update, and delete with potential problems they're in the de-centralized systems where there are theorems about the impossibility of having both consistent, available, and partition tolerance systems all at the same time.
This slide is at the heart of Long Chen's comments - payments. It doesn't stop with payments. From payments, you get to financial access and this is not a new idea. The G-20 adopted this as a formal part of the agenda years ago. Of course, from financial access built on payment systems, you get two impacts on the real economy and vice versa they all become intermingled together. So an example of this in Africa well-known is M-Pesa, running on a private system, Safaricom, but let me point out that it's actually tricky in many contexts to make the conversion from pure paper currency to these E-money systems. An example would be CBDC and the Central Bank of Bahamas they have it and they're doing it. But if you listen carefully to the comments of the governor, they've actually had problems in more rural and remote islands, getting people off of the currency.
There are of course success stories and Alipay is one of them. And you've heard already this morning in Long Chen's talk, credit reaching tiny businesses, operating in small markets instantaneously, including beginning of day borrowing and end of day payment. It's quite remarkable. Alipay also has run insurance portals with millions of users and has mutual aid through its system, as with Covid injections, the wealth management platform you've heard about and the commerce that takes place, e-commerce and so on that actually links back to the real economy because beneath all the e-commerce lies the real shipping load, just logistics and so on, where Alipay has also made a considerable investment. Another success story to balance this thing a bit is Pix in Brazil. They're making great strides. They require their commercial banks. Darryl again was saying how do we get banks to do this in the US is one thing in Brazil. They just mandated it.
I don't know the back-channel story of how they overcame the political economy forces. But all banks are required to offer electronic banking. It's not just payments as in Pix. They're building up credit registries. I think something akin to Alipay, but still caveat at the end. It seems to me there's the dangerous allure in these new technologies. The question is always what are we trying to build? As in the applications, insurance is potentially different from credit and wealth management, but there's a tendency to think that one can separate engineering as encoding up new CBDC for the US or other places as separate from the economics of what we're trying to accomplish. So we must bring the economics and the technology into it jointly and focus on the applications rather than thinking we're going to design some universal apiary, all-purpose system.
Another way to emphasize what I think is a danger here are the buzzwords, which we hear in the US and other countries. Fast payments, great. What's wrong with that? Real time gross settlement, atomic swaps of assets. As if these were the desirable features and they can be, but they do come with problems that we're all aware of already, namely liquidity problems, managing. No one holds enough liquidity to be able to honor all the payments that are at the beginning of the day in a given central bank. It doesn't happen. People systems economize on liquidity. Another issue that was raised earlier is the information revelation problem that can come with atomic swaps. So they're actually trade-offs with new versus legacy systems that need to be thought through carefully.
So, my last slide, I deliberately have only three pints I want to touch on. What I think is a very misleading debate about information, as you will hear, it said that we have to preserve privacy. That's paramount, in Europe but also US, other places. You'll also hear that a la Hershlifer3, once information acquired, it should become a public good. So data as public poses a dichotomy as if we had to choose one or the other of these systems and it's not the case. Encryption allows data to be kept private, but nevertheless allows analysis of the data in these encrypted homomorphic spaces. So the computer scientists have excelled in multiparty computation and fully homomorphic encryption. You can run the smart contracts as part of applications on these encrypted spaces. So now, again, we need to make choices depending on the applications, others that haven't come up so much yet you can have options to allocate financial products and so on.
You can run those options without auctioneers. You do not need trusted third parties, with everything encrypted actually. You can get insurance and embed that within the credit liquidity systems, and you don't need the planner of mechanism design. You use the guidelines from mechanism design and implement through encryption without the planner. You can get aggregation of data on the ledgers through encryption. Even though all the accounts are private, you nevertheless have some aggregate numbers can be used by private sector and government and so on. You can securitize small and medium enterprise debt with these new technologies.
Let me just conclude with a word about the nature of competition, especially on this slide having to do with information. When we talk about open competition, we should make a distinction. My experience is that what works well is ex ante competition among private, maybe in competition with public, for the rights to provide services. But not ex post competition, not siphoning off customers who are under a given contract, because that can undercut desirable incentives.
Similarly, you can have platforms mediating trade and so on, but the information collected on the platform should not be under the auspices of the party running the platform. For example, if they sell statistics for gain that can actually create an adverse incentive effect. This is the paper with Antoine Martin and Michael Lee. Likewise, you can program information and actually deliberately keep some of it private when it relates to Akerlof's lemon problem. If there's too much information out there, then basically the markets can train to zero. But here you don't let a private party do it. You preprogram it in code so it's the decision about how much information is being released to investors. Finally, this is not meant to be a contest between the public and private sector as proposed the comments about DM and so on. Distributed ledgers actually are helpful to regulators in allowing solutions to problems like coordination and runs on markets and so on.
For more information, please visit Luohan Academy's youtube channel: Luohan Academy