Blog | On the use (and misuse) of Gini Coefficients in Credit Scoring: Gini and Acceptance Rate

This is part 3 in a series of blog posts about Ginis in Credit Scoring. Read part 1 and part 2.

The relationship between Gini Coefficients and Acceptance Rate

One of the most frequent uses of Credit Scores is to decide whether to admit or reject an applicant applying for loan. This is usually called an “Admission score” or “Origination score”. A key decision around this use case is the selection of a score cut-off that will determine a threshold for admission. This cut-off value determines the acceptance rate of the population.

If the score is working well and predictive power is good, the relationship between acceptance rate and default rate will be positive. The higher the acceptance rate, the higher the default rate of the accepted population and vice versa. The direction of this relationship also has two implications: when acceptance rate is higher, the absolute number of bad loans (i.e. non-performing loans) or “bads” will also be higher, and the proportion of these “bads” in respect to the total loans in the accepted population will be higher too.

What does this mean in practical terms?

It means that the predictive power as measured by a Gini coefficient for the exact same score at different levels of acceptance rate for the exact same population will be different. The higher the acceptance rate, the higher the Gini coefficient and vice versa.

This is something that can be easily tested. If you have a portfolio and a score with good predictive power, you can calculate the Gini coefficient for different score cutoffs or acceptance thresholds and the results should look something similar to this example of a typical credit portfolio:

So for example, if there is a change in credit policy and the acceptance rate is lowered from 60% to 40%, the Gini coefficient for the same score over the new sample may also be lower. Does that mean the model is not working anymore? Absolutely not. All the contrary, it’s probably just a good signal that the score is doing a good job. Once a change in acceptance rate is implemented, results should be assessed by the change in default rate, not in predictive power.

Bottom-line:  To judge the predictive power of a Credit Score by the means of Gini, you also need to take into account the Acceptance Rate at which the Gini coefficient is measured. Lower Acceptance Rates will tend to have lower Gini coefficients by construction, even if it is the same exact score over the same population.

The fundamental reason behind this phenomenon was discussed in the part 2, where we explained why Gini coefficients should only be directly compared over the exact same data samples, even if the two samples correspond to the same population.

---

By: Carlos Del Carpio, Director of Risk and Analytics, LenddoEFL

Blog | Lessons from the field: How we created new group psychometrics to increase financial inclusion in Mexico

While Jonathan takes notes, Gerardo helps an applicant navigate our psychometric assessment on a mobile device. An essential component of our field work was to get direct usability feedback from applicants as they completed new psychometric content.

By Jonathan Winkle, Behavioral Sciences R&D Manager, LenddoEFL

An experimental psychologist by training, I am relatively new to the world of financial technology. Since joining LenddoEFL, I have embraced terms like information asymmetry, alternative data credit scoring, and financial inclusion. Yet it was only during a recent trip to the field that I was able to meet the people behind the FinTech jargon we use in our day-to-day, the small business owners whose lives we help improve in our mission to #include1billion.

In April of this year, I traveled with colleagues to Veracruz, Mexico to test new psychometric content for one of the top 3 microfinance institutions (MFI) in the country. Their group loan product extends a line of credit to a collection of business owners, but liability for payments is joint: if one person misses a payment, the group must still make that payment in full. Since many of those applying for these loans lack traditional credit histories, this MFI asked LenddoEFL to develop psychometric exercises that could quickly and reliably assess group traits that predict creditworthiness.  

There are traits that define a strong social group which are nonexistent for individual borrowers. A successful group has strong internal relationships that ensure they will help each other in times of need. A tenacious group can generate creative ideas to solve problems that arise when life presents hardships, as it is wont to do. And a cohesive group exhibits decision making abilities that allow it to act deliberately and with confidence. We designed new psychometric exercises to measure these core traits, and tested them in the field with groups of small business owners applying for loans.

Hiding from the Veracruz heat underneath a family’s palapa, Gerardo leads a collection of applicants through our group psychometric exercises while Jonathan makes observations about their behavior.

Measuring interpersonal relationships through social pressure
To measure the strength of a group’s interpersonal relationships, we examined the social pressure that exists among group members. Do individuals feel that they can answer sensitive questions honestly? Or do they feel pressure to conform to the opinions of the group majority? While the group was sitting together in one room, we asked them to raise their hands if they agreed with statements about the trustworthiness, fairness, and helpfulness of their local communities. We then asked individuals to answer these questions privately. The discrepancy between how the questions were answered in each setting could reveal how much social pressure exists, and thus how comfortable group members are being honest with each other. We expect that less social conformity means the group’s interpersonal relationships are stronger, an important factor for predicting whether the group will cover individuals who may miss payments throughout the loan cycle.

Measuring creativity through brainstorming
To measure a group’s creativity, we created a set of generative exercises. For both an easy and a hard problem, we had groups brainstorm as many solutions as they could in 60 seconds. The number of solutions generated was recorded as a creativity metric, and, as predicted, groups generated many fewer ideas for the harder exercise. We were also interested in the group’s dynamic as they performed these tasks. Were they apathetic or engaged? Was there a dominant member of the group? Ultimately, when a loan payment is due and some individuals are short on money, can the group come up with ideas for how to get the extra money? We hope that these generative exercises will shed light on this critical group trait.

Gerardo snags a picture with one of the applicants we met and her business, a stand selling eggs, candy, and other sundries. The small scale of some businesses we encountered, such as the one pictured above, reinforces their need for access to financial products. This woman’s entrepreneurial endeavors are only limited by the capital she can acquire.

Measuring decision making abilities through consensus
To measure a group’s decision making abilities, we created a time-to-consensus task. This exercise asks the group to solve a problem where all members must agree on the answer they provide. While we asked the groups to estimate the population of the state they live in, we actually don’t care how accurate their answer is! What’s more important in this exercise is how the group reaches consensus. Are they indifferent and accept the first estimate suggested? Or do they take their time and argue intensely while deliberating over possible solutions? What kind of strategies did they use to reach their estimate? Importantly, this task provides loan officers with a window into the group dynamic that might not otherwise be seen if the assessment merely collected static information such as sociodemographics and business revenues.

Financial inclusion is the mission of LenddoEFL, but working directly with the people we want to include allowed me to better understand how our assessments must be tailored to their cultures and experiences. The better we can measure group dynamics that predict creditworthiness, the more successfully we can extend financial services to those in need. As we continue to expand our credit scoring offerings across the world, looking past the business jargon we use and maintaining empathy for the humans we touch is essential on our path to #include1billion.

Blog | On the use (and misuse) of Gini Coefficients in Credit Scoring: Comparing Ginis

By: Carlos Del Carpio, Director of Risk and Analytics, LenddoEFL

This is part 2 of a series of blog posts about Ginis in Credit Scoring. To see the part 1, follow this link.

What is an AUC?

AUC stands for “Area Under the (ROC) Curve”. From a statistical perspective, it measures the probability that a good client chosen randomly has a score higher than a bad client chosen randomly. In that sense, AUC is a statistical measure widely used in many industries and fields across academia to compare the predictive power of two or more different statistical classification models over the exact same data sample [1].

How is AUC used in Credit Scoring?

In the particular case of Credit Scoring, AUCs are useful for example in the model development process, when there are several candidate models built over the same training data and they need to be compared. Another typical use is at the time of introducing a new credit score, to compare a challenger against an incumbent score over the same sample of data under a champion challenger framework.

How does AUC relate to Gini Coefficient?

The Gini Coefficient is a direct conversion from AUC through a simple formula: Gini = (AUC x 2) -1. They measure exactly the same. And it is possible to go directly from one measure to the other, back and forth. The only reason to use Gini over AUC is the improvement in the scale’s interpretability: while the scale of a good predicting model AUC goes from 0.5 to 1, the scale in the case of Gini goes from 0 to 1. However, all the properties and restrictions of AUC still translate into Gini Coefficient, and this includes the need to compare two different AUC values over the exact same data sample to make any conclusion about their relative predictive power.

What does this mean in practical terms?

Any direct comparison of the Gini Coefficients (or AUCs) of two different models over two different data samples will be misleading. For example: If a Bank A has a Credit Score with a Gini Coefficient of 30%, and Bank B has a Credit Score with a Gini Coefficient of 28%, it is not possible to make any conclusion about which is better or which is more predictive because they have been calculated over different data samples without accounting for the difference in absolute number of observations and the difference in proportion of good cases against bad cases. The only direct comparison possible is the one made about two scores side by side, over the exact same data sample.

Bottom-line: To affirm that a certain absolute level of AUC or Gini Coefficient is “good” or “bad” is meaningless. Such affirmation is only possible in relative terms, when comparing two or more different scores over the exact same data sample. Unfortunately this is often not well understood, which leads to the most frequent misuse of AUC and Gini Coefficients, such as direct, un-weighted comparisons of Gini values across different samples, different time periods, different products, different segments and even different financial institutions.

[1] Hanley JA, McNeil BJ. The meaning and use of the area under a Receiver Operating Characteristic (ROC) curve. Radiology, 1982, 143, 29-36.

Blog | On the use (and misuse) of Gini Coefficients in Credit Scoring

Over years of blogging, one of our most popular ever blog posts was about the Gini coefficient. In this series of posts, we revisit the Gini and dig further into its uses and the ways we see it misused in credit scoring.

What is a GINI?

For lenders around the world, the “Gini Coefficient” is an often heard, sometimes feared, and frequently misunderstood statistical measure. Commonly used to assess things like wealth inequality, Gini Coefficients are also used to evaluate the predictive power of credit scoring models. In other words, a Gini Coefficient can help measure how good a credit score is at predicting who will repay and who will default on a loan: the better a credit score, the better it should be at giving lower scores to riskier applicants, and higher scores to safer applicants.

Though calculating a Gini Coefficient is complex, understanding it is fairly simple:

A Gini Coefficient is merely a scale of predictive power from 0 to 1, and a higher Gini means more predictive power.

However, there are a few key aspects of Gini Coefficients that are not always well understood and can lead to their misuse and wrong interpretation. Over this series of blog posts we’ll discuss four of them:

1. People often compare Ginis when they should not. The only useful comparison across Ginis (or AUCs) is when looking at different scores over the exact same data. 

2. People forget that Gini will vary by acceptance rate. When presented with a Gini coefficient, always keep an eye on the effect of the acceptance rate.

3. People focus on Ginis, but are not always aware of its impact on the costs, benefits and overall economics of Credit Scoring.

4. People do not fully understand and often overestimate the role of Gini in the business of lending.

About the Author:

Carlos del Carpio is Director of Risk & Analytics at LenddoEFL. He has 10+ years of experience developing credit scoring models and implementing end-to-end credit risk solutions for Banks, Retailers, and Microfinance Institutions across 27+ countries in Latin America, Asia and Africa.

About LenddoEFL

LenddoEFL’s mission is to provide one billion people access to powerful financial products at a lower cost, faster and more conveniently. We use AI and advanced analytics to bring together the best sources of digital and behavioural data to help lenders in emerging markets make data-driven decisions and confidently serve underbanked people and small businesses. To date, LenddoEFL has provided credit scoring, verification and insights products to over 50 financial institutions, serving seven million people and lending two billion USD. For inquiries about our products or services please contact us here.

Blog | Digital Identities: Learnings from GSMA’s User Research in Sri Lanka

Smallholder farmers in Sri Lanka interviewed as part of the digital identity research

Recently, our friends at GSMA’s Digital Identity Programme and Copasetic Research set off to research digital identities and how they could support smallholder farmers in Sri Lanka. Their key hypothesis was:

“If MNOs, financial institutions, government and other service providers had access to a smallholder farmer’s ‘economic identity’ (income, transactional histories, credit worthiness, rights to/ownership of land, geolocation, farm size, and other vital credentials), they could provide access to more and better tailored services that enhance their productivity.”

In speaking with 40 smallholder farmers in Sri Lanka, as well as 7 stakeholders and 5 agri experts, GSMA learned a lot about the need for digital identities. GSMA invited LenddoEFL’s input in advance of the field research so we were keen to review the learnings. Below are some of the key findings of the report and how it relates to our work at LenddoEFL.  If this is interesting, we recommend reviewing the full report.

Source: Digital Identity for Smallholder Farmers: Insights from Sri Lanka

Identity is valued, but farmers are unclear how it relates to additional benefits
In Sri Lanka, the government is rolling out a new smart ID card giving increasing access to official identity. But farmers do not immediately understand how new forms of identity can be used to help them get access to more services (e.g. more tailored information services). Once they make the connection, they see the value clearly.

Identity is valued as it relates to accessing credit
Farmers and banks do not connect directly in many cases and farmers tend to have informal manners of connecting to credit through their buyers and agribusinesses. Banks don’t always have the information they need to cater to farmers. And the microcredit model can be more of a burden on the farmers than it’s worth. The research found that smallholder farmers are happy for their trusted service providers to work together and share information to enable access to credit. But since many farmers receive their income informally, the thought of sharing this information too widely (particularly with the government) caused some concerns.

Source: Digital Identity for Smallholder Farmers: Insights from Sri Lanka

Digital ID must build on face to face relationships
In Sri Lanka, farmers rely on and trust institutions with whom they have built local, personal, face to face relationships and these will be the best channels to roll out new systems and technology.

Farming is changing
Climate change and globalization mean that the work of a farmer is changing. Traditional farming skills are no longer enough. Farmers need to be constantly re-considering which crops they will grow now and in the future due to changing weather conditions and fluctuations in profitability. Younger farmers in particular are looking outside of their communities to the internet for new information. This new information needs to be combined with better access to financial services, allowing farmers to finance the transition to new crops, and hedge some of the risks in experimenting with new approaches.

ID needs vary across farmer types
The research found that a farmer’s financial stability and the extent to which they are embracing change (i.e. changes to farming practices, or the use of new technologies) have the most significant influence on their digital identity needs and priorities. GSMA mapped farmers across a 2 by 2 with the axes of poorer → wealthier and embracing change → stuck/fearful of change. In each quadrant is a unique farmer with unique needs. See report for more.

All of this means there is opportunity to better serve farmers (and other small business owners).

Farmers need better access to formal financial services:
Digital financial profiles could allow farmers to access savings, credit or insurance more conveniently and cheaply. Note that farmers were concerned about sharing their income information with a lender for fear it would get to the government and increase taxation or reduce welfare support. Credit scoring using psychometric data could be a good fit for farmers as it relies on personality profile data created at the time of assessment rather than existing financial data.

Read the full report

Contact us for more info on LenddoEFL’s credit assessment

Blog | Raising the Stakes on Psychometric Credit Scoring

An updated and expanded 2nd edition (first edition)

Why read this post?

Learn why high-stakes data is essential for building accurate credit-scoring models.

Introduction

Billions of people lack traditional credit histories, but every single person on the planet has attitudes, beliefs, and behaviors that can be used to predict creditworthiness. Quantifying these human traits is the focus of psychometrics, and the alternative data provided by this technique allows LenddoEFL to greatly expand financial inclusion in its mission to #include1billion.

But there is a catch: in order to build models that accurately predict default, applicants need to complete psychometric assessments in pursuit of actual financial products, a so-called “high-stakes” environment. This is because people answer psychometric questions differently when they have a chance to receive a loan (the high stakes) than they would in a hypothetical situation with no incentive (the low stakes).

Despite this fact, psychometric tools are frequently built using low-stakes data. For example, many companies develop psychometric credit scoring tools using volunteers. And many lenders want to validate psychometric credit scoring tools on their clients through back-testing: giving the application to existing clients and comparing scores to their repayment history, again a low-stakes setting.

These approaches are only valid if low-stakes data can be applied to the real world of high-stakes implementation, where access to finance is on the line for applicants. But it turns out that this is not the case. A recent study published by our co-founder Bailey Klinger and academic researchers proved that low-stakes testing has no predictive validity for building and validating psychometric credit scoring models in a real-world, high-stakes situation. The data below shows exactly how applicant responses shift as they move from one environment to another.

 

The Experiment

To test for differences between low- and high-stakes situations, LenddoEFL gathered psychometric data from two sets of micro-enterprise owners in the same east-African country. One group already had their loans (low-stakes) and another group completed a psychometric assessment as a part of the loan application process (high-stakes).

First, the low-stakes data. The figure below shows the frequency distribution for two of the most important ‘Big 5’ personality dimensions for entrepreneurs, Extraversion and Conscientiousness, as well as a leading integrity assessment[i].
 

You can see that when the stakes are high, people are answering the same questions very differently. The distribution of scores on these three personality measures shifts significantly to the right. When something important is at stake, like being accepted or rejected for a loan, people answer differently.

How do these differences in low- vs. high-stakes data matter for credit scoring?

To see how these differences impact the predictive value of psychometric credit scoring, we can make two models[ii] to predict default: one uses responses from applicants that took the application in low stakes settings, and the other uses responses from applicants that were in high stakes settings. Then we can use a Gini Coefficient—which measures the ability of a model to successfully rank-order applicants’ riskiness and for which a higher coefficient is a metric of success in this—to compare each model’s ability to predict default for the opposing population as well as its own.[iii]

These results clearly show that there is a significant change in the rank ordering when models built on low-stakes data are applied in high-stakes settings and vice versa.[iv] Importantly, we can see that a psychometric credit-scoring model can indeed achieve reasonable predictive power in a real-world, high-stakes setting. But, that is only when the model was built with high-stakes data.

Think about it like this: when the stakes are high, both less and more risky applicants change their answers. But, less risky applicants change their answers in a different way than riskier applicants. This difference is what is used to predict risk in psychometric credit scoring models: the difference between how low- and high-risk people answer in a high-stakes setting.

This also illustrates why we see that a model built on low-stakes data is ineffective in a real-world high-stakes implementation. In the low-stakes setting, the low- and high-risk people aren’t trying to change their answers, because they aren’t concerned with the outcome of the test. Once the stakes are high, however, this pattern changes.

 

Conclusions

Testing existing loan clients or volunteers has an obvious attraction: speed. That way you don’t have to bother new loan applicants with additional questions, and then wait for them to either repay or default on their loans before you have the data to make or validate a score, an approach that takes years.

Unfortunately, these results clearly show that this shortcut does not work. People change their answers when the stakes are high, so a model built on low-stakes data falls apart when used in the real-world. People answer optional surveys with less attention and less strategy than they do a high-stakes application, and therefore the only strong foundation to a predictive credit-scoring model is real high-stakes application data and subsequent loan repayment.

Consider an analogy: you can’t predict who is a good driver based on how they play a driving video game, where the outcome is not important. Conversely, someone who does well on a real-world driving test may not perform that well on a video game.  Whether it is driving skills or creditworthiness, you must predict the high-stakes context with high-stakes data.

TAKEAWAYS:

- Psychometric model accuracy is only guaranteed when you collect data in a high-stakes situation (i.e., a real loan application).

- Despite its speed, back-testing a model on existing clients in a low-stakes setting is risky because it might not tell you anything about how the model will work in a real implementation.

- If you want to buy a model from a provider, the first thing you should verify is what kind of data they used to make their model. Was it from a real-world high-stakes implementation similar to your own?

[i] These are indices from widely available commercial psychometrics providers. It is important to note that LenddoEFL no longer uses any of these assessments or dimensions in our assessment, nor any index measures of personality.

[ii] Stepwise logistic regression built on a random 80% of data, and tested on the remaining 20% hold-out sample. An equivalently-sized random sample was used from the other set (high-stakes data for the low-stakes model, and low-stake data for the high-stakes model) to remove any effects of sample size on gini.

[iii] Note that this exercise was restricted to those questions that were present in both the low- and high-stakes testing. It does not represent LenddoEFL’s full set of content and level of predictive power, it is only for purposes of comparing relative predictive power.

[iv] The results also show that using standard personality items, the absolute predictive power is lower in a high-stakes setting compared to a low-stakes setting. This is likely because of the ability to manipulate some items in a high-stakes setting makes them not useful within a high-stakes setting. This lesson has lead LenddoEFL to develop a large set of application content that is more resistant to manipulation and which has much higher predictive power in high-stakes models. This content forms the backbone of the current LenddoEFL psychometric assessment, all of which is built and tested exclusively with high-stakes data and subsequent loan repayment-default rather than back-testing.

Financial Services Veteran, Ray Ferguson, Joins LenddoEFL’s Board

Ferguson’s wealth of experience across banking, insurance and venture capital in Asia makes the appointment a strategic addition for the growing fintech company

LenddoEFL, whose credit score, verification and consumer insights help leading banks across emerging markets make data-driven decisions, is proud to welcome financial services industry veteran, Ray Ferguson, as a new member of its board. Ferguson is co-founder and Managing Partner of Caber Partners, former Regional CEO, SE Asia at Standard Chartered Bank, and currently chairs the boards of digital life insurance company, Singapore Life, and e-money software platform, Youtap. 

Read complete press release.

Blog | Can behavioral traits help financial institutions assess creditworthiness?

Despite progress in financial inclusion, 1.7 billion people still lack access to a bank account . And globally, only 11% of people over 15 borrow from a financial institution. This lack of access to credit is problematic for livelihoods, economic development, and much more.

Credit scoring is quickly evolving in favor of greater financial inclusion. At LenddoEFL, we spend 110% of our time thinking about ways to understand someone regardless of credit history or collateral. We pull together the best new sources of opt-in alternative data that applicants can choose to share so that we can assess creditworthiness and verify identities.

One of our sources of data is behavioral and psychometric data. Read this article published in Social Innovations Journal by Javier Frassetto, our Chief Risk and Analytics Officer, to learn how it works: what traits predict creditworthiness, how do we measure them, and how do we prevent gaming.
 

Find out how behavioral traits predict creditworthiness

Blog | What exactly do we mean when we say financial inclusion?

LenddoEFL Partner Success Manager, Gerardo Rivero, doing field research for our financial access tools

We started LenddoEFL to solve the problem of access to credit in emerging markets, where people find themselves unable to get a loan, and unable to build their credit. This excludes good people from financial services, limiting opportunity for individual livelihoods and economic growth. 

We realized that even though people may have limited financial data in a credit bureau, they have plenty of unique data that can be accessed to better understand who they are. For example, we found that analyzing the digital footprint of an individual (with full consent) helps us to get to know them and understand certain traits that relate to creditworthiness and credit risk.

Now, we are working with banks and lenders across 20+ countries to use non-traditional forms data - digital footprint, mobile behavior and psychometric to predict risk, and unlock access.

When we think about financial inclusion, there are really 3 levels, each necessary to get to the next one. 

1. Access comes first: Can you get a credit card or open a savings account? 1.7 billion adults around the world lack an account at a financial institution according to the 2017 Global Findex. Enabling these people to take that first step towards opportunity is foundational. 

2. Price: Often where access is scarce, the first loan can come from a payday lender or other institution at an unbearably high price/interest rate. So the next step to financial inclusion is bringing the price of a loan down to reasonable rates even without historical credit data. 

3. Convenience: Once you have access to credit at a fair price, the third step to financial inclusion is making it convenient to get. Historically, inclusive lending such as microfinance could involve arduous, time consuming processes with multiple in-person visits and copious document collection. We want to make borrowing easier and faster for people while maintaining safety. The beauty of moving from analog loan officer-based processes to machine learning and big data-driven processes like ours is that it becomes faster and easier. 

We believe that financial inclusion isn't simply about access to financial products, but about access to fast, affordable, and convenient financial products. Join us on our mission to #Include1Billion people around the world. We are hiring! 

CardRates.com | How LenddoEFL Uses Data and Personality Analyses to Increase Access to Financial Services in Emerging Economies

Credit is hugely important to people around the globe. You need it to obtain housing and higher education. You need it to start a business. You need it in case of emergencies and other unexpected expenses.

But in emerging economies, credit may not be accessible to many people. According to the World Bank’s 2017 Global Findex, 31% of the world’s population doesn’t have an account with a financial institution or a mobile money provider.

“We still have 1.7 billion people on the planet who don’t even have a basic bank account,” said Amie Vaccaro, Director of Marketing at LenddoEFL. “Only 11% of people around the world borrowed from a formal financial institution in the last year.”

Read full article

Blog | Score Confidence: Boosting Predictive Power

Note: This is a new and improved version of a popular post from last year.

Our unique platform has a big reason to live: we provide fast, affordable and convenient financial products for more than 1 billion people worldwide. And there is only one way to accomplish that: by facilitating more actionable, predictive, robust and transparent information to our clients to enable them to make the best possible lending decisions. However, data quality pose the most challenging problem we have faced along this journey as it threatens the predictive power we are delivering to our clients. Therefore, through the years we have developed and perfected a one-of-its-kind way to assess the quality of the data applicants are supplying: Score Confidence.

What exactly is Score Confidence?

Score Confidence is a tailored algorithm that scans and analyzes psychometric information gathered through LenddoEFL's Credit Assessment to generate a Green or Red flag which reflects how confident we are on our score’s ability to represent an applicant’s risk profile:

• The result will be Green if LenddoEFL is confident in the data quality such that we will generate and share a score based on it.
• Conversely, the outcome will be Red when LenddoEFL’s confidence in the gathered information has been undermined.

What does Score Confidence measure?

Once the applicant has taken our psychometric assessment, we put the data through our Score Confidence algorithm to find out whether we can be confident in a score generated using this data or not. We will return a Green Score Confidence flag if we believe the score accurately predicts risk, and also be transparent about the reasons behind a Red Score Confidence flag to empower our partners with increased visibility and actionable information.

LenddoEFL's Score Confidence system is comprised of five Confidence Indicators of key behaviors, each generated from a combination of different data sources. If we identify evidence of any of the following behaviours, the assessment will be rated as Red and no risk score will be returned in order to protect our partners:

• Independence – the assessment has not been completed independently, and LenddoEFL detects attempts to improve one’s responses with either the help of a third party or other supporting resources.
• Effort – the applicant has not put forth adequate effort and attention in completing the assessment.
• Completion – the applicant has not responded to a sufficient portion of the timed elements of the assessment.
• Scoring error – a connection issue or system error occurred and LenddoEFL is unable to generate a score.

What information feeds Score Confidence?

Our data quality indicators are constantly reviewed and updated and, over the years, we have added new and different data sources to our Score Confidence algorithms:

• Browser and device metadata surrounding the completion of the application
• User interaction information with LenddoEFL’s behavioural modules
• Self-reported demographic data

Our Score Confidence system flexibly combines all the available data in order to return a Red or Green status for each application.

How does Score Confidence help our partners make the best possible lending decisions?

To boost the predictive power we can deliver for our clients, LenddoEFL does not share a LenddoEFL score for applicants with a Red Score Confidence flag as we have learned that Red applications tend to have very limited predictive power whereas data coming from Green flagged assessments can effectively sort risk amongst applicants. Therefore, not lending against a score for Red flagged applications boosts the predictive benefit for our clients.

Blog | iDE Ghana increases access to sanitation with help of innovative credit assessment from LenddoEFL

Partnership allows Ghanaians to purchase their first toilets

Globally, 32% of people lack access to a toilet in their homes (Source: WHO UNICEF JMP). In Ghana an astonishing 87% of people do not own a toilet. And in rural Northern Ghana, it is worse still. Two out of every five children in northern Ghana are stunted, compared to approximately 20% of children stunned nationally (Source: UNICEF).

iDE Ghana, a nonprofit that creates income and livelihood opportunities for poor rural households, wanted to improve sanitation in the region. They began by applying design thinking to understand the low rate of toilet use. It turned out that people didn’t know where to buy a toilet, and if they did, it was prohibitively expensive to buy.  People could not afford the full cost all at once, and there were no options to pay for a toilet over time, as there were for other large purchases.

"What we found was the criteria for borrowing towards non-income generating loans were ridiculous. So we set up a one stop shop for toilets and sanitation products, selling them door to door,” explained Valerie Labi, WASH Director at iDE Ghana. “And the beauty of the model is that we give our customers 6 to 18 months to pay the toilet off over time.”

This seemed like the perfect solution given the challenges to toilet purchasing uncovered, but it was still challenging. “We allowed people to pay over the course of 6 to 18 months but we required for the customer or a guarantor to prove their income with bank statements or payslips. And this was a big deterrent. No one wanted to give their bank statements to a toilet company. And it would take an average of 40 days to get through the process” Labi shared. “We realized these requirements were scaring away customers as they’d never had formal credit before. So we asked ourselves, how else could we assess creditworthiness in a more inclusive way?”

That’s when they came across LenddoEFL universal credit assessment. By collecting behavioral and psychometric data at the time of application, iDE’s commercial agents will be able to assess risk and make a decision in a day or less, cutting down the time to sale greatly. Previously, the commercial agent made multiple calls and visits to collect the required documents. By using the LenddoEFL score, iDE removes the need for a guarantor or proof of income for the best scoring customers. Low scorers will need to pay 50% of the cost of the toilet in monthly installments before receiving the toilets.

iDE’s goal is to provide 20,000 to 25,000 toilets to households in Northern Ghana. At an average of 11 people per household, this will provide life-saving sanitation for 275,000 people. And the plan is to sell toilets as part of a fast, convenient customer-driven process and at affordable rates. With the LenddoEFL assessment in place since February 2018, iDE is already receiving positive feedback for customers who enjoy the process. Stay tuned for updates on this exciting partnership.

Blog | Our Commitment: Privacy, Responsibility, Choice and Control

By: Richard Eldridge, LenddoEFL CEO

Data privacy and security is a top priority at LenddoEFL and with the General Data Protection Regulation (GDPR) deadline coming up, we wanted to share our thoughts on this topic.

Our work toward a more financially inclusive future for one billion people brings with it important responsibilities, none more important than keeping customer data private and secure.

Our Responsibilities

Privacy is one component of a broader set of responsibilities we have as a global financial technology company.

1. Customer Protection and Privacy

We follow these five principles across our operations:

• Customer Data Ownership: Data we collect will always remain the property of the customer who shared their information with us and we will always safeguard the data as if it were our own. LenddoEFL uses world-class security standards in the transfer, storage, and processing of information to ensure that customer data is kept secure at all times. We never store data for longer than is necessary or authorized. Any information we permanently store is anonymized and encrypted. Where third party services are required, we only enlist the assistance of industry recognized players that adhere to the same or stricter standards than we do. In addition, security checks and penetration testing are conducted on a regular basis to ensure the security of our platform. See our full Security Policy here. 

• Consent-Driven Access: LenddoEFL only accesses data that customers share with us and all information gathered requires their explicit consent.

• Inclusive Use: Data shared with LenddoEFL is used with the sole purpose of enabling greater financial inclusion for each customer.

• Transparent Handling: Data shared with us is not--and will never be--shared without the consent of the person to whom it belongs. We will never share a customer’s data or sell it to another third party except their financial institution that is our client. Furthermore, we will only use the data for purposes the customer has agreed to.

• Unbiased Application: When building a credit model, no discriminatory variables—such as gender, race, and political or religious preferences— are taken into consideration.

For more details, read our full Privacy Policy.

2. Responsible Lending

When used properly, credit is a powerful tool for alleviating poverty, stabilizing income inequality, and empowering people to thrive. When used irresponsibly, credit can result in over-indebtedness, default, and economic instability. At LenddoEFL we are dedicated to building robust, proven models for our financial institution clients that enable safe, responsible data-driven decisions across the customer lifecycle with the goal of building a stable economy. 

3. Customer Choice and Control

Lastly, we believe in giving people options for financial inclusion, where they did not exist before. This involves using their own data to unlock access to savings, insurance products, and credit. With Europe’s second Payments Services Directive (PSD2) paving the way for open banking, people have increasing control over their data, and we know from experience that data can open doors to better, more affordable financial services. It makes sense to let each individual decide if and when to share their data. LenddoEFL’s credit scoring and verification tools are designed with this choice and control in mind. We allow customers to choose which data they want to share, if any, to get access to financial services from our clients. The more data someone grants us access to, the better we can understand them, and the better financial institutions can match them with appropriate offerings (pricing, terms, amount, etc).

Blog | Credit Scoring using Digital Footprints

Evaluation of Research Findings

Author: Carlos Del Carpio

On April 30th, an exciting new piece of research was published on SSRN - “On the Rise of Fintechs – Credit Scoring Using Digital Footprints.” It was as if we had commissioned it, as this piece helps to validate what our clients around the world already know: how someone behaves online is a reflection of their character, and can be used to measure risk and boost lending. We are grateful to the authors for their work as this paper provides a great insight into the predictive power that credit scoring models can achieve by using alternative data.

However, the methodology used to build and assess the models is as important as the data used as input. In this post we share a couple of observations about the methodology used in this paper based on over 10 years of experience building and deploying alternative data credit scoring models in production environments around the world.

#1 Using cross-validation alone to assess credit scoring models can inflate predictive power. We recommend to use out-of-time out-of-sample hold-outs to set more realistic performance expectations.

Credit scoring’s main objective is to predict credit repayment behavior in order to make credit risk decisions. In that sense, it is a forward looking predictive modeling exercise.

As in many applications of predictive modeling, in a credit scoring setting both the context and the behavior of the population being studied tends to evolve over time. The macroeconomic environment changes, the credit policies changes, the origination and collection processes change, and therefore the population itself is changing. All these changes combined often  introduce bias and systematic differences between the training and validation sets used to build credit scoring models. Other economic and financial applications of predictive modeling face similar challenges. For example, if a model for predicting stock values is trained on data for a certain five-year period, it is unrealistic to treat the subsequent five-year period as a draw from the same population. In the same way, using cross-validation alone, where testing samples are drawn from the same time period as the training sample to validate the results of predictive models could lead to predictive power expectations that will differ greatly from the actual predictive performance that can be achieved.

To address this, using an out-of-time out-of-sample set of data that is representative of the most recent time period is preferred and can enable more realistic results. Since the model in this paper only uses out-of-sample cross validation, the results may be too optimistic compared with the actual results it may return when implemented.

#2 The paper omits feature selection, an important part of the model building process. This decision could lead to completely different results.

Dimensionality reduction methods, such as feature extraction (FE) and feature selection (FS) are important components of the building process of credit scoring models. However, depending on the classification technique used to estimate the functional form of the final model, FS can be done independently of model estimation, or it can be embedded in the process (i.e. built into the classifier construction occuring naturally as a part model estimation).

Logistic regression, for example, does not perform FS as part of its estimation. Therefore it has to be performed independently each time before the actual estimation. If FS is required to be done independently by the modeling procedure, this must be repeated each time for every training set, which in turn creates different optimal models with different sets of variables for each iteration of the cross-validation, which adds an additional step in order to choose the final set of variables to go into the final model.

This paper omits this problem completely by forcing all variables everytime without using any criteria for feature selection. In other words, the authors forcefully and arbitrarily “select” to include all the features, something that is rare and unrealistic in most credit scoring settings where there are hundreds or thousands of variables to choose from and are impossible to fit within a logistic regression model due to the curse of dimensionality[1]. This is an intrinsic problem for credit scoring models that include big data sources such as digital footprint and social data, and the reason why feature extraction and feature selection methods can play a key role, sometimes as important as the techniques used to estimate the final functional forms. Had the authors included a feature selection process as part of each iteration in the cross validation process, it could have yield very different results.  

In conclusion, putting these modeling observations aside, we consider this paper important because it offers a clear example of the signal that can be found in digital footprint data, and the possibilities to improve current methods and data sources. We just provide a few examples on how methodological choices can affect how results are estimated and assessed, and why are they important to assess the full potential of the predictive power in this particular type of data.

[1] Donoho DL. (2000). High-dimensional data analysis: The curses and blessings of dimensionality. AMS Math Challenges Lecture, 1, 32 pp.

NewsWav | CTOS & LenddoEFL partner to boost financial inclusion in Malaysia

KUALA LUMPUR: CTOS Data Systems Sdn Bhd (CTOS), Malaysia’s largest credit reporting agency, has entered into a partnership with LenddoEFL to enable access to financing for Malaysian consumers with little to no credit history.

Both CTOS and LenddoEFL have aided banks, lending instit…

Read more in NewsWav

Yahoo Japan | Can Japanese banks use big data with "AI loan"? (日本の銀行は「AI融資」でビッグデータを活用できるか)

Attempts to calculate the creditworthiness of individuals by AI (artificial intelligence) and to finance using it are expanding. This is called "AI score lending". 

　The meaning of AI doing loan screening, which is one of the most important tasks of banks, is quite large. 

　However, the question is whether Japanese financial institutions can handle big data. If it can not do it, it will repeat the failure of the past score lending. 

Singapore's Lenddo is a service in emerging countries such as India, Vietnam, Indonesia, which have never had a history of credit. 

Read full article

Sina News Taiwan | How to break the credit assessment problem? (如何破解信貸評估難題?)

Bangladeshi banker and Nobel laureate Muhammad Yunus (Muhammad Yunus) the promotion of microfinance , is the poor through microcredit loans , so there is money to do a small business to support themselves, and thus get rid of poverty. However, due to the time-consuming and laborious credit evaluation of lenders, the large-scale application of microfinance is difficult to achieve once.

Nowadays, mobile banking comes. It can collect data to help people who have little formal financial records in the traditional sense to broaden their services. Labor costs are also greatly reduced. For example, Kenyan mobile telecommunications operator Safaricom and African Commercial Bank jointly launched the M-Shwari business in 2012, which can determine customers’ credit scores based on Safaricom’s user information and the trading history of its M-PESA mobile money business. Loan amount.

In addition to payment data, mobile phones (especially smart phones) can also provide more types of information for credit evaluation by borrowers . For example, a person's geographic location data can reflect whether he has a stable job and fixed residence; shopping records can even reveal whether the borrower is pregnant ; and the richness of information obtained by social media is not Yu.

The fintech start-up company Lenddo EFL also uses the Internet to conduct psychological tests on potential borrowers. The question concerns the concept of money (for example, choosing to pay $10,000 at a time, or $20,000 for six months), where your money is spent. , Evaluation of living communities, etc., to determine the reliability of testers loan repayment. To date, the company has completed more than 7 million credit assessments, helping consumers with a lack of traditional credit records to borrow 2 billion U.S. dollars from 50 financial institutions of varying sizes.

詳全文 如何破解信貸評估難題？-財經新聞-新浪新聞中心 http://news.sina.com.tw/article/20180514/26854022.html

Benzinga | Here Are The Benzinga Global Fintech Award Finalists For The Best Under-Banked Or Emerging Market Solution

The finalists for the Best Under-banked or Emerging Market Solution category are:

LenddoEFL
CEO: Richard Eldridge
Description: LenddoEFL's mission is to provide 1 billion people access to powerful financial products at a lower cost, faster and more conveniently.

See full list of finalists

Welcoming our New Behavioral Science Manager

In this photo, Jonathan demonstrates cultural differences in height during a field visit with loan applicants in Veracruz, Mexico.

Since our merger, we have welcomed a number of incredible new colleagues onto the LenddoEFL team. Jonathan Winkle joins us in our Boston office as our new Behavioral Science Manager. We cornered him to learn more.

Tell us about your background?

In undergrad I majored in psychology, where I developed a passion for researching the brain and behavior. To gain more experience after college, I worked in a systems neuroscience lab at MIT studying visual attention. Eventually I found my way to Duke where I earned my PhD in cognitive neuroscience. My dissertation focused on the behavioral economics of dietary choice, investigating how the mind is affected by “nudges” that can bias people towards healthy (or unhealthy) eating habits.

What brought you to LenddoEFL?

Studying behavior has always excited me because it is the ultimate endgame of our brains’ hard work, yet academic research on the topic can often be too disconnected from real-world problems. I found myself wanting to make more of an impact on society, and in this role I can leverage my experience to quickly and directly improve people’s lives around the world. As the Behavioral Science Manager for LenddoEFL, I can test a new hypothesis and apply that knowledge globally in a matter of weeks. And the better I do my job, the more people I can help get access to life-changing financial services.

What are your plans as Behavioral Science Manager?

My primary goal is to drive feature engineering. Features are the observations we collect about individuals to predict credit risk, and feature engineering is the process of discovering and creating new features to make our algorithms work better. For example, how honest a person is might be predictive of loan default, but we first need to quantify honesty as a feature to use it in a predictive model. As new features make our models more predictive and more powerful, our financial institution clients all over the world will gain a better understanding of their under-banked loan applicants.

If I am successful, we will be better at predicting if someone will repay their loans, thereby allowing our clients to make the best, most informed decisions possible. No pressure.

Across data sources, we look for ways to profile a person’s character, trying to understand how traits like honesty or conscientiousness relate to credit risk. This is a hard, but extremely important challenge.

LenddoEFL deals with both psychometric/behavioral and digital data sources. How do those differ and how do you think about each?

On the psychometric side, we engineer the form our data will take from the outset, then extract it by inserting new content (e.g., survey questions or psychometric games) into our simple, interactive assessment. We can be more hypothesis-driven when it comes to designing features in this realm.

On the digital side, we work with large, unstructured data sources where we necessarily have to be more exploratory and let the data do the talking.

Will you be working with our research advisors?

Absolutely! I am looking forward to working with leading researchers like Peter Belmi to push the envelope of our own research while also sharing the insights gained from our unique dataset with those in the field of behavioral economics. We will also be inviting more researchers to collaborate on our work.

Enough about work, what do you do for fun?

I like to rock climb, play Go, hang out with my dog Clementine (pic below), and try out new recipes in the kitchen.

What’s a fun fact about you?

I have a tattoo of Phineas Gage, a famous figure in the history of psychology and neuroscience. Gage was a railroad worker in 1848 that lost the left pre-frontal cortex of his brain when an accidental explosion sent a 3 foot iron rod rocketing through his head. Miraculously, he survived and was even able to walk himself to a doctor despite the 11⁄4 inch hole running behind his left cheek and out the top of his skull. He lived for 11 years after this event, but experienced marked changes in his personality that have been studied ever since. The story in itself is fascinating, and of particular interest to me is how Gage’s misfortune shaped theories of the mind for more than a century after the accident.

Look out for a future post from Jonathan about his field work in Mexico and learnings about group dynamics.

The Economist | Mobile financial services are cornering the market

Mobile money means more nimble financial services

KAUSAR PARVEEN, of Chakwal district in the north of Pakistan’s Punjab province, is a star beneficiary of the work of Karandaaz, a Pakistani financial-inclusion charity. The owner of just one buffalo, she borrowed 75,000 rupees (about $650) to buy another one and started selling milk. The business has done so well she now has four buffaloes and an assistant, and has taken out another loan to install a biogas plant, saving on firewood and sparing her family the woodsmoke.

This was how microcredit, as promoted by Muhammad Yunus, a Nobel-prizewinning entrepreneur from Bangladesh who launched his Grameen bank in 1983, was supposed to work: credit would allow the poor to establish microbusinesses and improve their lives. The idea has spread across the developing world. Sadly, in many places it has not worked out that way. A big expansion of microcredit in India’s Andhra Pradesh province caused a crisis in 2010 when the lenders were blamed for an increase in suicides by farmers. A World Bank paper last November, written by Robert Cull of the bank and Jonathan Morduch of New York University, considered evidence showing that microcredit has had “only modest average impacts on customers”. It has often been used to cover the normal ups and downs of household spending, which is helpful but not transformative. Read full article.