Contact Menu
Back
machine learning in banking

It’s Official: Online Lenders Can No Longer Afford to Ignore Alternative Data

January 20, 2020 Explorium Data Science Team Data Enrichment

You don’t know what you don’t know, as the old saying goes.

But in the age of Big Data, you simply can’t afford to shrug off what Donald Rumsfeld famously called “unknown unknowns”. In a high-risk, high-reward business like online lending, incomplete information quickly puts you at a disadvantage – and your business in real trouble.

What’s more, if information that could make all the difference is out there, somewhere, you can bet your life someone else is extracting insight and value from it. If that person is your direct competitor, you could be shedding customers and market share.

That’s why the Federal Reserve’s statement on using alternative data for online lending, released earlier this month, has caused a stir. 

For companies that have already expanded their data horizons, it was a reminder that with power comes responsibility. For those for whom alternative data is the great unknown, it was a clear call: don’t get left behind.

Guide to Data Acquisition

What happened, exactly?

On December 4, 2019, the Federal Reserve Board released a statement on why consumer protections are important when using alternative data, in conjunction with the Consumer Financial Protection Bureau (CFPB), Federal Deposit Insurance Corporation (FDIC), the National Credit Union Administration (NCUA), and the Comptroller of the Currency (OCC).

A lot of acronyms there, I know. But the most striking thing is that it shows just how much alternative data for banking has gone mainstream. Whether for fraud detection, account management, credit writing, or a host of other banking operations, online lenders, banks, and fintech companies are turning to alternative data in droves.

Let’s take a closer look at what the fuss is all about.

What is alternative data?

Alternative data is information you get from sources that are non-traditional or hard to access. That could mean the IoT, sensors, PoS transactions, satellites, or even social media sources. From here, you use machine learning or deep learning algorithms to analyze the data and help you make predictions about patterns, performance, or individual behaviors.

This approach is incredibly effective for fintech providers and online investment managers – and those in the know are throwing serious resources behind it. In fact, JP Morgan reported in 2017 that asset managers were already spending up to $3 billion per year on alternative data collection and were employing four times as many data analysts than they had just five years earlier.

Now banks and online lenders are getting in on the action, using alternative data to better inform their credit decisions.

By collecting and analyzing data on borrowers’ income and outgoings, for example, they get a clear picture of the person’s cash flow over time. This means creditors can predict the borrower’s ability to repay a loan with far more accuracy than they would by relying on more traditional data points.

Not only does this improve confidence in lending decisions, it’s good news for would-be borrowers whose earning can be a little erratic, such as freelancers and contractors. As the world moves ever more in the direction of the gig economy, alternative data will be vital in helping consumers to tap into credit services that may have been closed to them in the past.

Why does this matter?

You can’t make nuanced, effective decisions without quality data. We all know this, but in some parts of the financial sector, awareness has been slow to translate into action. PwC estimates that most businesses use just 0.5% of the data available to them.

The problem often boils down to speed and responsiveness. Executives in the past often relied on long lead-time documents like quarterly reports or whitepapers to help them assess what’s happening in the marketplace, rather than seeking out real-time data that would allow them to seize opportunities and respond to challenges as they emerge. Data analysis and business intelligence simply weren’t embedded into their data-to-day.

This is changing – and for the better. 

More and more companies understand that it’s crucial to have total visibility and insight into how all business applications, processes, and decisions happen.

These companies are starting to realize that people’s financial realities are set up differently to how they were ten or even five years ago. Models that worked fine for assessing a borrower’s reliability or solvency in the past no longer cut it. At the same time, we all produce swathes of data every second of every day and getting a grip on this wealth of information provides far more accurate ways to predict behavior.

Alternative data and predictive models

Alternative data unlocks compelling ways to reduce risk while giving fintech companies and lenders a vital competitive edge. As the Federal Reserve’s report pointed out, when used right, this can improve the credit decision process for underserved consumers while helping businesses to match better benefits, repayment plans, and pricing options to their existing customers. Everyone’s a winner.  

Let’s say you’re concerned about improving the way you detect fraudulent applications and loan stacking. Machine learning models are an excellent way to do this, but the chances are you don’t have all the data you need in-house to make this truly effective.

With the right platform, you can connect to multiple alternative data sources, including external datasets like government filings, business registration, social media, domain information, search engine results, and foot traffic. This allows you to bypass lengthy data acquisition and data-matching processes, bringing all this information into a single, coherent depository.

From here, machine learning algorithms identify troubling patterns and red-flag anomalies with ever-improving accuracy, helping you to pinpoint suspicious behavior and detect fraud before it costs your business money. At the same time, you reduce the risk of false positives, ensuring that you aren’t turning the wrong customers away.

Final thoughts

Extracting actionable insights from alternative data can be tricky, however. 

You’re talking about streaming, merging, and harmonizing potentially conflicting data from multiple sources into one place. You’re talking about subjecting that data to AI-driven analysis engines or even incorporating machine learning into the production process.  

This data may need to be cleaned or reformatted to make it consistent. You may need to work hard to bring real-time streams in line with one another so that you’re always using up-to-date, accurate data in the right order and at the right time.

That means you need a powerful platform, too. One that’s flexible and highly scalable. One that streamlines the task of integrating models into production environments, making it easy to reproduce results, stress-test applications, and monitor data drift. Without this, implementing an alternative data strategy can quickly descend into chaos.

And lastly, as the Federal Reserve made clear, consumer protection is paramount. 

This isn’t the Wild West: there are strict rules governing how you collect and use people’s data, especially when it comes to financial decisions. Be responsible and make sure you stay in line with fair lending and credit reporting regulations.

With the right technology and approach, expanding your pool of data brings extensive benefits to you and to your customers, helping you to stay ahead of the curve in a fast-changing industry. For that, there is simply no alternative.

Guide to Data Acquisition

Subscribe Today! Get the latest updates with our newsletter.
We promise you'll love it.

Follow us