Many employers today are looking at the social media accounts of potential employees to get an idea of the type of person they might be hiring. They’re not the only ones — lending companies are also getting in on the act. And new research shows that some of the more unusual things you post or the people you might be connected to could have an impact on your credit score.
The paper, “Credit Scoring with Social Network Data,” was authored by Yanhao Wei, a Ph.D. student in economics at the University of Pennsylvania, Wharton marketing professors Christophe Van den Bulte and Pinar Yildirim, and Boston University professor Chrysanthos Dellarocas. Yildirim recently discussed their findings on the Knowledge@Wharton show on Wharton Business Radio, which you can find on SiriusXM Channel 111.
An edited transcript of the conversation appears below.
On the challenges lenders face in the developing world:
For the lending companies, there are a few things they would like to verify before they give their money to you. They would like to know that you are who you say you are and that you are indeed a good person. And when I say a good person, [I mean] someone whom they can trust, whom they can lend money to and expect it to be paid back to them.
It is a challenging task. In countries like the United States and [other] developed nations, this is perhaps slightly easier to do. There are alternate ways of getting data about you. [Lenders] can go find you, talk to you or talk to the people around you. In fact, there are more institutionalized ways of doing this, such as looking at your credit history, which is a way of collecting your financial data over a period of time. They can look up your credit score, and then make a judgment about you.
But this is a very challenging task when we talk about other parts of the world, the nations that are developing. There are four reasons why, in my opinion, [social media] became an alternate method of deciding whether to give a loan to somebody or not.
Think of India, Mexico, the Philippines or Columbia. These are the countries where we see some of these alternate credit-scoring practices most often. The reasons why they find these countries suitable is, first of all, there is a lack of financial access, and yet a huge number of people are moving into the middle class. They used to be in a family that had been farming for decades. And now there is the young person who just for the first time in his family graduated from school and is looking for a loan and has to somehow go through the hurdles of the banking system to obtain credit….
The bigger part of the problem for the majority of these countries, and people in the remote areas of these countries, is they have a lack of resources to [gain] financial access. There is no bank branch where they can go and apply for loans. The closest banking branch might be three hours away. And if you think about all of the possibilities or, let’s say, the lack of possibilities of transportation, that means if you want to get even the smallest loan you need to essentially do a day trip going back and forth. So there’s the lack of physical access to financial services, and this is something that we don’t normally think about. Around the Wharton campus we have a bank branch probably on every corner, but in other parts of the world it’s a major, major problem.
“Who you know matters. Your social network matters.”
Third, of course, there is a high level of bureaucracy and a high level of paperwork involved in any application in any part of the world where you need credit. I have been talking to a lot of the companies that do this type of business. The chief technology officer of … one of the companies that does this type of work cites this example: If you’re in one of the Asian countries [and] you need a loan, if you’re a decent person — if you’re making decent money, if you have a great job — then you go and apply for a loan. And they have these somewhat irrelevant, almost unacceptable criteria for you to obtain a small loan, such as you should be employed by the same company for about a year. It just so happens that in many of these developing nations, if people are good [at what they do], especially if they’re in technology environments, they keep changing jobs every six months or so. They are looking for the best next opportunity for themselves.
So, without looking at anything else, you just immediately disqualified someone from obtaining access to credit. On top of that, they have these ridiculous rules, such as I need to call your employer, and I’ll make only one call to your employer. And if your employer happens to be away from his desk at that point, you’re again disqualified for a loan.
The fourth hurdle is that, even if I can call people, I don’t have well established ways of obtaining data about you. There is no credit history in many countries. There are no histories of financial payments whatsoever. This might be the first time somebody’s applying for a loan anyway, so I don’t have an ongoing relationship with this person. In such environments, I need to look for alternate ways of finding data about people’s character.
We talk a lot about the Internet revolution that has happened in many countries, such as we talk about how, in the United States and in developed nations, the Internet has changed [commerce]. But the Internet revolution hasn’t happened yet in many countries. In many developing countries where landlines never found access to the remote areas, the Internet didn’t find access either. In fact, if you look at even the current numbers, only about 26% of the world has broadband Internet access. It’s actually a relatively low number.
Let’s look at an alternate source — mobile phones. Mobile phone [penetration] in the developed world [is] actually more than 100%. People have more than one cellphone. In developed nations, it’s over 90%. So, indeed, people have much greater access to mobile phones. And because of that, they also have access to some of the systems that come with mobile phones, such as social media. In countries like Thailand and the countries surrounding Thailand, there are plans for cellphones where you may not necessarily have an Internet package, but your cellphone has an option that you might, for example, pay a small fee so that you have access to Facebook. It’s a Facebook-app-specific Internet package. And Facebook invests in this, the companies, the social media companies invest in these.
On the growing use of social media in lending:
In an environment of this sort, what you are going to end up with are very detailed data about people’s mobile phone use. And because you know mobile phone use, you end up having information about social network [use] as well. And second, because they are also on Facebook, Twitter and other social networks, you also have access to their information from these sources.
For a smart company, this creates an alternate way of getting data about people’s personalities. And that’s exactly what these companies that have disrupted the financial access systems have figured out, that they can find data about your social networks, which would provide some sort of verification for who you are — first of all, that you are indeed a real person. Second, if you’re a good person you must have some good connections with people around you, you must have a certain number of friends, and you must have had this account for a certain period of time. [These companies] can gather so much more information about you using social media than by looking at your financial data. So it became just natural for them to move toward this direction.
We know that this is happening in many countries. Lenddo is one great example. It operates in Mexico, the Philippines and Columbia. It operates with lenders. It provides and collects information with the permission of the users. It collects information from social media and then crunches it. It runs the numbers and figures out whether you’re a good person — and again, I’m saying this in terms of creditworthiness — whether you’re a real person and whether you should be deserving a loan. And it gives this information to the lender, which then decides to essentially qualify you for a loan.
On why it’s more of a throwback than a trend:
As new as it sounds, this type of practice goes back many decades, in fact, I think about a century or more. In the old days when you needed a loan, [the bank] looked at who you knew. It tried to qualify you by talking to the people you knew. There’s a story that when someone asked for a loan from a big-named banker back in the day, he said, “Well, I can’t give you a loan, but just walk with me to the park and somebody else will.”
It’s the same idea. [The] concept that these companies are banking on — the idea that who you are with, the people who are in your social network, will have very similar traits and behaviors as you — is a very strong, very powerful idea here that they are taking advantage of. They’re simply looking at the behavior of people around you, and by looking at that, they can guess [with accuracy] how you are likely to behave.
On how the practice is being used in the U.S.:
Of course, the reason why we don’t see this in the United States is partially because there are a lot of data [available] about people. But we could still improve somebody’s credit score by looking at his or her social network. The reason why we don’t see this is, of course, regulation. Financial data, or anything that may relate to financial data, are heavily protected in the United States. And we have rules about discrimination. The reason why … a company that is smart and operating in this business may prefer not to be in the United States is because, if I’m looking at your social network and if I’m denying you for a loan based on your social network, that may end up as a discrimination lawsuit.
In some countries, “there is a lack of financial access, and yet a huge number of people are moving into the middle class.”
For example, let’s say my friends are gay, and that might be one of the reasons why, for example, I may then feel that I’m denied. Many of the traits that [bring] people together in a social network could be the basis for a discrimination lawsuit. And I believe that most of the companies operating in this environment are simply afraid that, even though they’re not looking at any of these things as part of their credit-scoring system, they may feel that this puts them at risk for heavy regulation and perhaps some lawsuits in the United States….
On how the research formula was created:
The way that this started was first hearing about some of the companies that requested your password for Twitter or Facebook in order to approve you for a loan. Of course, it’s a very unusual, very untraditional business. And then we started learning more about this practice that had been happening in the United States and in different parts of the world. We [thought], “Can we study the system? And can we look at, first, whether using social network data is going to result in a more accurate assessment of someone’s credit-worthiness? Second, once [companies] use this network and once people realize that [companies] are using these types of data, could they try to game the system, and at some point would this result in less accurate information?”
We are not necessarily classifying info as A, B, C, D. We’re looking at general information collection and if there is a level of similarity between your information and my information. Let’s say we are in the same social network. That says we must have similar traits to a certain degree. It could be a [very significant] similarity, or it could be something very small. But there has to be a certain level of mutual liking between the two of us, which is usually a function of our similarity, of the things that we enjoy doing together. And if we can collect additional information from people who are in someone’s social network, then we simply get better opportunities to clarify our beliefs about the person. We are simply getting verification checks….
On whether people could game the system:
Could you start gaming the system? We looked at this possibility. What we are finding is that yes, indeed, individuals [could game the system], if they could know somehow that you are a financially responsible person and I am financially responsible, and we all need to show that we are good individuals to the company on social media so that we can be considered worthy of a loan. We find that individuals will have some incentives to drop their friends or at least make the information, the connection of having a friendship with [certain people] less visible….
What that could do over time is [cause] some sort of fragmentation in social networks. Good types, people who are more financially responsible, have incentives to drop the bad types. That’s also true for the bad types as well. They have an incentive to be connected to a higher number of [financially responsible people]. And they have an incentive to be connected to a smaller number of bad types. That’s going to result in, over time, a segmentation or fragmentation of the markets. The people who are credit-worthy [will] be more connected to other credit-worthy individuals, and people who are not credit-worthy being connected to, again, other non-credit-worthy individuals. As a result of this then, firms [that use social media as a tool for determining credit-worthiness] are facing two challenges: First, now they are getting information from a smaller number of people because everyone has an incentive to get rid of the bad types in their own social networks. So I’m ending up with a smaller number of data points for every single person….
But the second effect that companies are facing is actually something that’s positive. If they can get information, if they can look at you and know that you’re trying to game the system or when everyone is trying to game the system, if you’re a high type, then the people around you are more likely to be high types. When I see a group of good types around you, then I have more confidence and a higher belief that you are indeed [financially responsible]. So that is something that can increase the accuracy. If enough people are trying to game the system, and they do actively change … their networks, the social network environment overall can still provide more accurate information compared to the data that would come from only the individual’s history.
On how widespread the practice may become:
This business is certainly growing outside of the United States. And I think that it has the potential to grow in the United States as well. It just has to happen that the regulators need to be convinced that this is a way to reduce the amount of risk, the credit risk that is out there. Nobody wants people who are not going to be able to pay back their loans to obtain a loan, because [loans that are bad] are what led to the financial risk environment and all of the crises that we have been facing. We need to somehow take initiative and say we want individuals to be assessed in a better way. We want better models than this. And we should be, perhaps, relaxing some of these regulations or finding alternate exceptions, finding cases where an individual’s social media data, especially if he or she wants to share those data, could be used in order to assess them for a loan.
“If enough people are trying to game the system, and they do actively change their networks, the social network environment overall can still provide more accurate information compared to the data that would come from only the individual’s history.”
On getting consumer buy-in:
I get this question all the time: Why would someone want to share this information? Because in some parts of the United States, also in many parts of the world, what might end up happening is [the data show] that you’re a credit-worthy individual. You’re someone who can pay back a loan, but there are no good ways for you to communicate this information to the banking system, just like the examples that I was giving earlier. If I am a good worker, if I am someone who has obtained a good education, but the bureaucracy and the financial system, these old rules, are simply not adapting to the new environment and are denying me credit, then you’re closing the doors for me to obtain a better life. You’re preventing me from obtaining a mortgage, perhaps, or a car loan or a loan for furthering my education. Many doors are closed to these individuals….
On how the practice could stimulate economic growth:
We hear from politicians, executives and the financial world that we need to empower the middle class, we need to essentially take people out of poverty, give them an opportunity for a better life. And how can you do this if people don’t have any means of improving their lives by either creating a business, perhaps, by taking a loan to invest in their farms or to use it for their education? That’s exactly why we like these systems. It creates an opportunity for people to do better things.
On the key takeaways:
One of the things that is emphasized by the companies that operate in this business is that people whom you never met in your life, and whom possibly you will never meet in your life because they are your friends’ friends’ friends now have the ability to affect your credit score. We’re talking about a very connected world and a world that is connected at higher degrees [than before]. So that means some people in your network whom you’ve never met will influence your score. And on top of that, [these companies] can look at everything. Once you give them access to your social network, they have information on anything and everything that you do. They look at things like how many friends you have and how frequently you communicate with them. They don’t sit down and read the messages, but there is a certain level of text processing that’s done. I know that some companies are, for example, looking at the use of faulty words, certain words can indicate [certain life circumstances].
An interesting example that was given by one of the executives at one of these companies was it looks at the correlation essentially between the use of language and your ability to pay back. People who use the word “please” are apparently people who don’t pay back. So, even words like this that are relatively unexpected can have some correlation. It has something to do with the everyday communication that people are used to, perhaps. And in their lifestyles, of course that [can] affect their language, the words that they choose to use when they’re asking for things. And that in turn correlates with how likely they are to pay their loans.
On top of that, these companies also will, for example, make you write an essay or a small paragraph to explain why you need this loan. And then you would write it paying a lot of attention to the entire content. But what they would see the end is not necessarily what you put in there, but how many grammar mistakes you made. That’s the type of thing that, again, can indicate how careful you are, perhaps how much you pay attention to detail, and of course, your education level.
