September 15th, 2008. Do you remember what happened that day?
You may not recall the specific date, but if you were a risk professional or a financial services executive back then, you’ll surely recognize the time-period. This was the day that Lehman Brothers collapsed, perhaps the defining moment of the financial crisis that led directly to the Great Recession.
This was also a defining moment in my career, being in charge of credit risk for Capital One Financial. For those who haven’t experienced anything like it, I can tell you it is a surreal experience to wake up every morning with fear and uncertainty about not only the survival of your employer, but of an entire sector of the economy. The securitization markets had ground to a halt, our stock price had dropped by 90%, and a 158-year-old company that everyone presumed was too big to fail had just vanished overnight. We needed to come up with a plan to navigate through this perfect credit storm, and we needed it yesterday. To say time was of the essence is a massively gross understatement.
But survive it we did. I learned so much during that time, and I still carry the battle scars with me. I try to pass on as much of my learnings from that time as I can to my clients, so they can be prepared when the next downturn comes.
“Those who fail to learn from history are doomed to repeat it.”
· Sir Winston Churchill
The case for “Why Now”
Fast forward to today. We are in the midst of the longest running bull market in history; over 10 years and counting. Since World War II, there’s been a recession in the US every 6.1 years on average. In times like this, it’s easy to slip into complacency and assume the good times are here to stay. But you can predict with 100% accuracy that another recession is coming; it’s not a question of if, but when. And recently, there have been increasing signs that the next recession may be just around the corner.
Of course, I don’t have a crystal ball and even the best economists find it extremely difficult to predict the turn. But even if you believe the recession is a few years away, now is still the time to start preparing for it. Why? Because while we are still in the expansion, or growth, phase of the Economic Cycle, we are definitely on the wrong side of the Lending Cycle.
The Lending Cycle
In underwriting, although you are assessing the creditworthiness of an individual, macroeconomic factors also play a role in the performance of that loan. Credit conditions tend to follow a similar cyclical pattern to the economy, but with a lead-lag effect.
This is driven by supply economics:
· The initial recovery period coming off the trough is the best time for lending. This is because the supply of credit in the market is at its lowest as lenders are slow to expand credit policy, and their rear-view mirror approach to modeling will overestimate the risk of an individual loan during this period.
· Near the end of the boom period is actually the worst conditions for lending. The supply of credit in the market is at its highest, as business heads are pressured to hit growth targets on already high volumes, and the rear-view mirror has the opposite effect on risk assessment. This is where we are now!
Hopefully by now, you are convinced that the right time to prepare for the downturn is today. The next question, then, is what is your playbook?
Getting your ducks in a row
When I look back to 2008 and how we navigated through the storm at Capital One, my biggest learning is that we should have acted faster. We made the right calls that pulled us through in the end, but we could have been quicker to detect, and there was too much lag time between detection and action.
For the detection piece, I would recommend designing an early warning system that monitors both internal performance and macroeconomic indicators. What is most imperative is that you obsess with leading metrics – if you’re waiting for an increase in charge-offs, you’re going to be at least 6 months late to the party. For credit cards, look to metrics even earlier than the first missed payment, such as a ramp in balances that customers are carrying.
You should also kick the tires on your delivery pipeline for promoting changes into production. You want to be able to move fast at the first sign that something might be wrong with the canary’s breathing. Back then, we were on a shared platform with the US business, and we sat on critical changes to credit policy that were ready to go for 8 months before we could get them released. Now is the time to invest in technology and agile delivery before it is too late.
Rebuilding your models
The most important decision you will ever make in the customer lifecycle is the initial adjudication. The lifetime value (LTV) of an account is highly dependent on the initial exposure granted; it is really hard to correct for this on the backend if you get it wrong initially. For instance, a lot of credit card lenders used credit line decreases (CLD) as a tool during the Great Recession. I analyzed the results of a CLD program from a portfolio we purchased in 2011 and found it was highly unprofitable; while it did reduce losses, that was overwhelmed by the foregone revenue.
Thus, your new account acquisition models are your most important asset, and you should start rebuilding, or at least refitting, those now. Even a 2-3% increase in the Gini coefficient of a risk model can translate to millions of dollars for a large portfolio. I am often asked by clients how important is it to leverage machine learning (ML) or other AI techniques when building new credit risk models. My advice is to horserace as many techniques as you can, provided they still meet your bar for transparency and explainability, and can run error-free in your live production environment. The models we implemented in 2008 were built with SAS using logistic regression, and the vintages booked on these new models vastly outperformed the prior ones on the old models. I have found that 80% of the benefit comes from refitting your models so they are trained on newer, more relevant samples, and only 20% comes from new techniques and new data sources. That doesn’t mean you shouldn’t strive to capture that 20%, and the newest ML libraries can certainly help. But what it does mean is that what is even more important is to plan for rapid refit, where you can continually adjust your model over time on the most relevant samples.
Preparing for hardship
As credit conditions begin to turn, you will inevitably see an increase in the number of customers falling behind on their payments and flowing into later stages of delinquency. In 2008, we were overwhelmed with how quickly this deterioration occurred, and were unable to staff up quickly enough in collections to deal with the higher volumes. Now is the time to build contingency plans for scaling your collections network, as it is a high turnover industry and proper staffing is key to managing through a recession. If you are collecting in-house, you should consider onboarding a third-party vendor who can handle overflow. If you are currently outsourced, you should be increasing staffing targets in your forecasts and managing seat count closely with your collection vendors.
Digital messaging (email, SMS) is a great tool for communicating with customers falling upon hardship that the industry has been slow to adopt. Reaching out to customers who are over-utilizing their credit line or have missed their first payment can reduce the volumes flowing into later stage delinquencies. You can also provide tools through these messages, such as a temporary reduction in interest rate or fee waivers, that will help your customers manage their finances better and get back on track. During the Great Recession, one of the best levers we pulled was proactively putting our struggling customers on a hardship plan, where we reduced their interest rate to 0% and converted the line to an installment loan, with more affordable monthly payments. This proved to be a successful strategy, as the money we saved from the defaults we were able to prevent more than outweighed the foregone revenue – and it led to many more positive outcomes for our customers.
A final thought
For many, the biggest reason why we don’t all start planning for the downturn today is organizational psychology. I’ve seen this time and time again – during the good times, discretionary investment of budget and human capital flows to growth initiatives, and requests for new risk model builds or adding a new collection vendor fall upon deaf ears. Then, when things start to go south, Risk can get anything they want prioritized – but of course, by then the ship has already left the harbour. Hopefully, I’ve presented a case here that can help you to break this wheel; if you can get started today, you will certainly gain an edge over your competition as the recession nears.