Author: Pankaj Kulshreshtha
In the run upto 2008, I had an interesting experience as CRO for a loans portfolio. While the stress and uncertainty of COVID19 times are going to make 2008 look benign, I thought of sharing the story. There are likely a few lessons here that may be useful for lenders now.
This was a small liquidating portfolio of personal loans. Loss rates were running at about 7% when this portfolio came in my fold. When I saw the draft monthly PQR (portfolio quality report) that included this portfolio for the first time, I knew we had a problem at hand. My team's triangulations projected that we will have about 40% LTS (Loss to sales) over the lifetime of this portfolio. This seemed as toxic a portfolio as it gets for a risk manager. I called this out in the global risk review. As expected, folks thought this was just a new risk guy padding up the projections, but cursorily asked me to keep an eye out. A couple of months go by and we see all our assumptions to 40% LTS tracking as predicted. Come the next global risk review, people now take notice. Global CRO then asked the question that led to multiple insights. He asked, "Did you guys make a mistake in your campaigns ?"
Let me explain. These loans originated through a direct mail cross sell campaign to our existing credit card customers. We dutifully asked our marketing analytics colleagues to check the campaign rules for each mailing. To everyone's relief, we hadn't mailed the loan offers to delinquent customers ! We also confirmed that all the exclusions and cut-off criteria recommended by risk were applied correctly. So the plot thickened.
Several teams dug deeper into the data, we (risk team) also sat with collectors to gain insights why this portfolio was performing so badly. One big structural problem was the product itself. Loan sizes were too large, in fact, they were orders of magnitude larger than the credit limits these customers had on the credit card. Typical duration was 5 years, a lot longer than was good for this segment. And, the final straw on the back of the camel was the price. Price was high and drove a significant adverse selection. A large proportion of those who actually applied couldn't afford loans of this size, and most likely no one had offered them much credit. Basically response models messed up with risk models big time. It was all downhill from there though I am sparing you gory details given the surreal and scary climate we are in.
Now that brings me to the current time. Enough has been said about the mess COVID19 has put us humans and our economy in. There is no doubt that we are in a recession, U or V shape recovery notwithstanding. We can debate the extent of shrinkage and the time it will take to recover any semblance of normalcy to consumption, but it is imperative that financial services respond to rapidly evolving situation. In my informal discussions with colleagues in risk management across banks, the key word is BRACE. Which effectively translates to tighten underwriting, cut down accept rates. In many cases, that will be the right choice.
We have to remember though that the proverbial Black Swan has just hit us between the eyes and is still sticking there. Severity of current fear and the pervasiveness combined is likely once a 100 year event. Such deep fear and widespread panic triggers an antediluvian risk aversion response. Our genes haven't evolved as fast as our technology has ! So let us try to visualize January 2022. We would have a vaccine by then. The immunity levels in the community would have gone up. People will be desperate to bring back the quality of life of 2019. And humanity will realize that our fate is collective ... ok only a tiny bit !
How would our lending decisions look in 2022. Would we feel we used a sledgehammer where a pickaxe would have been more suitable ? Did we use all the customer information we had in the most meaningful manner ? Did we create new products in line with dramatically different customer needs ? Did we miss the opportunity to build deep customer loyalty by supporting them in the time of their need ? My biggest frustration in 2007 was that we weren't originating new loans on that platform. We could have taken the opportunity to reset the product, bring in sharper underwriting and pricing. That would have kept the program alive and largely reversed the loss problem within 18 months. It is hard to shrink out of a loss problem.
Reality rarely is as simple as blogs and linkedin posts like this make it out to be. We did want to be helpful in these unbelievable times in a more tangible manner. The Scienaptic team has over 150 years of experience with consumer credit and risk management. We have distilled our experience in a benchmarking and diagnostic toolkit that can help risk / lending leaders come up with prioritized list of 5-6 key initiatives to run immediately. Toolkit just takes a couple of interviews to understand your portfolio. We will happily do this free if that helps. Please contact us if we can help, stay safe and let us work towards better times.
PS: The pro-forma for this loan program assumed loss rates of 4.5%. This was duly stressed to 5.5% and the program still was profitable. So much for stress testing, but that is a story for another time.