Business Lending

Managing Risk in Small Business Lending

Two years ago, I left a promising career at PayPal, a major technology giant, for what some considered a risky move: I joined BlueVine, a young fintech startup. My title: vice president of risk.

This year, I took on an even bigger role when I was named chief risk officer of the Silicon Valley company, which offers working capital financing to small and medium-sized businesses.

My promotion comes at a time when risk is becoming a bigger concern in fintech, which is ushering in big changes in banking and financial services.

Fintech revolutionizes financial services

Data science technology has dramatically improved access to financing and the way we manage our money. The fintech wave that began with my former company, PayPal, and the world of payments, has spread to other aspects of personal finance, from mortgages to student and auto loans to investing.

This expansion was accompanied by growing concern that the fintech boom is fraught with risks that, if left unchecked, could lead to a major bust in the financial services industry that could in turn cause harm to the broader economy.

In a speech in January, Mark Carney, the governor of the Bank of England, cited the need to “ensure that fintech develops in a way that maximises the opportunities and minimises the risks for society.” “After all, the history of financial innovation is littered with examples that led to early booms, growing unintended consequences, and eventual busts,” he said.

Risk management as key to success

Risk management certainly has been a focus area for BlueVine from the beginning.

BlueVine joined the revolution in small business financing in 2014 when it rolled out an innovative online invoice factoring platform.

Factoring is a 4,000-year old financing system that allows small businesses to get advances on their unpaid invoices by providing easy, convenient access to working capital. BlueVine transformed what had been a slow, clunky, paper-based solution into a flexible and convenient online financing system that enables entrepreneurs to plug cash flow gaps that often hamper business growth.

Because the BlueVine platform is based on cutting-edge data science technology that can process and analyze information to make quick funding decisions, managing risk inevitably became a major challenge in building our business. As Eyal Lifshiftz, our founder and CEO, recalled in a recent column, in BlueVine’s first month of operation, almost every other borrower defaulted.

In fact, that was partly the reason Eyal invited me to join his team. BlueVine serves small and medium-sized businesses seeking substantial working capital financing of up to $2 million. To succeed, we needed to build a robust data and risk infrastructure.

Small startup with big data needs

Joining BlueVine also posed a personal challenge.

At PayPal, where I started as a fraud analyst and then moved into the company’s data science division where I helped develop behavior-based risk models, I had enormous amounts of data to work with to do my job. Now, I was joining a young startup with very limited data history, but with big data needs.

This meant putting together exceptional and experienced teams of data scientists and underwriters and developing a technology that becomes progressively more precise and accurate as it draw lessons from our steadily expanding data and underwriting decisions. It was important for us to have a group of super smart, highly-motivated and technologically-strong people working closely with a team of experienced and sharp underwriters.

Here’s how the process works: Our underwriters develop a robust methodology which is then translated into detailed logic decision trees.

Each decision tree includes dozens, even hundreds of branches, made up of question sets on different underwriting situations.

For example, a decision tree could focus on approving new clients coming from a specific industry, such as transportation or construction, or on increasing the credit line for a client with a specific financial profile.

A typical decision tree would drill down on further financial questions: What’s the expected cash-flow of the business in three to six months? What’s the pace at which it has accumulated debt over the past year? Are the business sales seasonal in a material way?

The questions could also focus on non-financial areas: Does the company’s website look professional? How does it compare with major companies in its industry? Does the business actively maintain its Facebook and Twitter accounts?

The goal is to build a risk infrastructure that steadily becomes more efficient in answering questions in an automated, large-scale and highly accurate manner. Our data scientists leverage multiple external data sources and use dynamic advanced machine learning models to answer these questions pretty much in real-time and with a high degree of accuracy.

So it’s a combination of technology and human input. There will always be gray areas, questions and situations that cannot immediately be addressed by our computer models.

But as the models get better and more robust, the gray areas will shrink. Our models are constantly and automatically enhanced, re-trained and expanded by the most recent data and input from our underwriters.

Think of it as the fintech version of Deep Blue and AlphaGo, the powerful computer programs that famously outplayed topnotch chess grandmasters. Both programs were based on similar principles: the more they played, the more knowledge they absorbed and the more formidable they became at chess.

Technology and teamwork

An even better example is the self-driving car powered by Google’s artificial intelligence technology. Human input is still required, but the more driving the car does, the smarter and more autonomous it becomes.

Building our risk infrastructure is an ongoing process for BlueVine. But it already has helped us steadily expand our reach, making us stronger, smarter and even faster in financing small and medium-sized businesses.

In just a couple of years, the strides we’ve made in managing risk more effectively enabled us to increase our credit lines to $2 million for invoice factoring and $100,000 for business lines of credit, which means we’ve been able to serve bigger businesses with bigger financing needs.

While we initially focused mainly on small businesses with annual revenue of under $250,000, today we have an increasing number of clients with annual sales of more than $1 million and increasingly, we’ve been able to serve clients with revenue of more than $10 million a year.

By the end of 2016, BlueVine had funded roughly $200 million. We’re on track to fund half a billion dollars by the end of this year.

We’ve accomplished this in a time of heightened skepticism about fintech in general and alternative business lending in particular. But rather than scoff at this skepticism, I’d point out two things.

First, fear often accompanies the rise of a new technology. Second, in the wake of the 2009 financial crisis, it’s prudent to raise hard questions about the rapid emergence of new financial technologies.

While building technologies and companies that can provide financial services faster and easier is a laudable goal, It’s wise to move cautiously and with humility.

The BlueVine experience underscores this.

Risk is still a challenge we take on every day. But we have found ways to take it on confidently and effectively with a vigorous combination of technology and teamwork.