Card Issuers Look to Great Recession for Loss Reserve Guidance

Introduction

Journalists, industry research groups, government agencies, and others have spent several months pontificating about what the economic fallout from the COVID-crisis might look like in the lending industry. With the release of 1Q2020 earnings and regulatory reporting, we got our first look at actual impacts – sort of.

The timing of, and response to, this crisis will push observable consumer performance impacts out into 2Q2020 reporting and beyond. Widespread preventative measures like travel bans and business closures did not go into effect in the US until the second or third week of March. This means consumers were only impacted by economic worsening for a small portion of the quarter, and some consumers made all their 1Q2020 payments before they began to feel economic pressure. Moreover, widespread customer assistance programs further obscure true impacts as many institutions offered deferrals rather than marking accounts as delinquent. These factors, amongst others, minimized any observable risk impacts in 1Q2020 and early 2Q2020 reporting.

While the factors listed above limit performance impacts, banks gave a strong indication of their expectations with their reported modifications to 1Q2020 loss provisions. The FDIC requires banks to categorize consumer loss allowances for three lending categories: Credit Cards, Residential Real Estate, and Other Consumer. In this article, we will focus on credit cards as this is a key consumer lending category that touches consumers across the economic spectrum (and, it just so happens, is one in which our team has particularly strong expertise – specifically in card assets).

Before diving deeper into credit card loss provisions, it is important to note that financial reporting in 2020 includes a new expected credit loss accounting standard – Current Expected Credit Losses (CECL). Many of the largest card issuers split out their allowance builds for CECL adoption and general macroeconomic worsening. Looking only at issuers who attributed allowance shifts between these two impacts leaves us with a relatively small sample, but the apparent trends are interesting nonetheless.

Brief Note on CECL

We won’t go into too much detail on CECL – our friends at the big accounting firms have provided plenty of interesting coverage of the topic over the past several years – but comparing implementation across the largest issuers yields a few insights worth highlighting.

When CECL was first announced, the banks warned that the new accounting standard would increase their cost to lend to riskier subprime customers, which would in turn increase the cost of credit, and decrease the access to credit, for these consumers. However, initial data measuring CECL impacts does not necessarily support this. 

Capital One and Synchrony hold the riskiest portfolios, but their increases in allowances due to CECL adoption were similar to the increases at Citi and Discover (as can be seen in the table above). Conversely, Bank of America and Chase hold card portfolios with considerably less risk. However, both, respectively, nearly doubled allowances when they adopted CECL accounting standards.

Drawing firm conclusions about CECL adoption impacts from this limited sample is impossible, but the relative comparisons are strikingly different from what one might have expected based upon the comments made to regulators by these institutions. The greater driver of these relative differences may be the average life of loans and risk distribution across that period. While risk in subprime populations tends to be heavily concentrated early in the customer journey, losses in super-prime portfolios tend to be more evenly distributed over a longer period. Because CECL requires issuers to hold reserves against lifetime losses, rather than just those expected in the next 12 months, subprime issuers may have a “shallower curve” in adapting to CECL given a greater share of their losses were already accounted for under the previous accounting standard.

COVID/Macro Driven Increases to Allowances

Despite the lack of definitive consensus around the shape and timing of the economic impacts of COVID-19, banks were required at the end of 1Q2020 to develop internal loss forecasts to adjust their loss allowances. “Business and Economic Conditions” is explicitly listed as a qualitative adjustment factor in CECL guidance, so each bank made its own determination on COVID’s economic impact and the ensuing recovery (which were, in turn, incorporated into their allowance adjustments). The comparison of the firms’ allowance increases, and the potential drivers of those increases, highlights an interesting trend.

We investigated several hypotheses to explain what might be driving the variance in COVID-driven allowance ratio increases, with the strongest correlations suggesting an inverse relationship in current loss levels and expected increases. While this might surprise a casual observer, well-documented analysis of previous recessions shows that low loss portfolios experience greater relative worsening in response to macroeconomic worsening. In simpler terms, risk triples in a 2% loss portfolio much more quickly than it doubles in a 6% portfolio.

While the inverse correlation between recent losses and 1Q2020 loss allowance increases was apparent, interestingly, the positive correlation between the worsening multiple each bank experienced during the Great Recession and the 1Q2020 loss allowance increases was stronger. Excluding Bank of America, which bucks both trends, the latter linear correlation yields an R-squared value of 0.85, shockingly high for this small sample.

Based on this information, it may be reasonable to suspect that these firms leveraged various macroeconomic forecasts, and benchmarked losses to a period when economic indicators mirrored these forecasts. While this may have been a reasonable response early in the crisis, it fails to account for the unique dynamics of our current recession.

In the Great Recession, unemployment peaked at around 10%. This recession has seen unemployment jump from 3.5% in February, to 14.7% in April. Our detailed analysis of BLS data (which we will share in a subsequent post) quantified the variance in unemployment across industries, highlighting many relatively unaffected sectors and a few that experienced cataclysmic job losses. This analysis also highlighted a notable geographic disparity in unemployment rates. Given these variances, it is likely that unemployment in many issuers’ portfolios will differ from the national unemployment rate.

While unemployment trends may be the most notable difference between our current crisis and the Great Recession, issuers should also consider distinctions in the government response, shifting spend behavior (which we covered here), and a host of additional discrepancies. 

Putting it All Together

At the end of 1Q2020, these established issuers appeared to be expecting a gradual entry into a Great Recession-like loss environment. Armed with more information, it will be interesting to see how these issuers adjust their reserve ratios at the end of 2Q2020. Of course, the most important data points will be the actual loss and delinquency data we observe in the coming quarters. While no one has the clairvoyance to precisely predict future losses, we do have a few recommendations to better prepare for an uncertain future:

• The timing, magnitude, and shape of the loss curve will undoubtedly vary from what issuers experienced during the Great Recession. Issuers can increase the precision of loss forecasts by leveraging industry and geography specific unemployment data, and analysis of government response impacts. The variance in unemployment rates across various customer segments challenges he assumption that the jobless rate within any given portfolio will match that of the aggregate US economy.

• Ensure loss mitigation strategies and teams are operational and ready to scale. While there is no consensus on the size and timing of the brewing loss bubble, an increase of some magnitude has become increasingly apparent. AQN’s Pradeep Kalla outlined key collections strategies in his recent post here.

• Develop capabilities to improve competitiveness in any risk environment. Even if you believe the largest players overreacted with their 1Q2020 allowance builds, improved monitoring, better modeling of value drivers, and novel testing agendas can enhance profitability in any lending business. AQN’s Gaby Garcia recently discussed these initiatives, and several others, here. 

The information we shared here represents a fraction of the research AQN is currently doing on behalf of our clients. We will be sure to check back in once more data becomes available, but please do not hesitate to get in touch if you would like to discuss what we are seeing in the market.