Current risk models don’t have sufficient data to account for today’s volatile markets. What’s the true impact of factors like unemployment and inflation on the credit losses incurred by banks, and could the looming 2023 recession actually provide banks with the ammunition they need to build better, more accurate models?
Friday, October 28, 2022
By Tony Hughes
The global economy is not yet in a recession, but things sure feel miserable. In the context of strong demand and low unemployment, inflation is currently elevated in all major economies.
After 30 years of price stability, our collective consciousness is not used to dealing with this. When asked for advice recently, I suggested coaxing a 1985 CFO out of retirement!
Risk models built using data exclusively from the Great Moderation/Great Recession era cannot adequately explain the impact that high inflation and rising interest rates will have on bank financial performance. The data we collect this year and next will help to fill this gap. A high-inflation recession in 2023, which many say is probable, will be wretched, but at least it will provide us with a lot of interesting data to improve future model performance.
Let’s now discuss some of the theoretical relationships between inflation and bank financial data, and then tie these back to what we've seen in recent financial data and models.
In terms of credit performance, the real burden of fixed-interest exposures always falls due to inflation. This is easy to see – if my mortgage repayment is $1000 per month and my nominal income doubles overnight, I will allocate a smaller proportion of my salary to meet debt obligations. If property prices are also inflating, the loan-to-value ratio on my mortgage will be in decline, reducing the loss-given default (LGD) for my lender if I ever happen to default.
For variable rate loans, meanwhile, the interest component of the repayment will be impacted by rising rates. If this increase is faster than the growth rate of my nominal disposable income, the interest burden of my loan will rise and I will be at greater risk of default. Offsetting this is the fact that rapid nominal income growth will erode the real burden of the principal component of the repayment.
We’ve focused thus far on consumer loans, but the forces described should be even more apparent for corporate exposures. It’s generally accepted that businesses, especially large ones, have more pricing power than consumers/employees – and can consequently better ensure that revenues rise faster than expenses in the context of high inflation. Households generally lack this bargaining power, and therefore tend to take a harder hit as inflation bites.
Where banks lose out during a period of high inflation is in noninterest income, which is mainly driven by fees earned during loan origination, with mortgages being perhaps the most lucrative sector. Unsurprisingly, higher long-term rates strangle the market for new home loans, and U.S. banks have thus reported a rapid recent decline in fee revenue. In the last couple of weeks, for example, Wells Fargo reported that its Q3 noninterest income declined by 25%, relative to 2021.
The liabilities side of the balance sheet is also impacted by inflation. Those holding noninterest-bearing deposits, normally available on demand, will be incurring a rapidly increasing opportunity cost for the convenience of instant access.
Deposit duration tends to increase markedly during periods of rising rates, which has powerful implications for liquidity management. The rates currently available on longer-term deposits, for instance, are higher than they have been for years, enticing many householders and corporate treasurers to move unneeded cash into such products.
Of course, it’s also true that the current data on deposits is affected by a considerable amount of noise stemming from unusual activity during the pandemic. In 2020, U.S. total deposits rose at a record pace, because corporations used existing credit lines to boost cash on hand and because of a sharp rise in the household savings rate bolstered by generous government-support programs. In more recent quarters, however, we have seen sustained declines in total deposits, as many of these pandemic-era actions have shifted into reverse.
Inflationary Impact on Models
A “deposit recession” like this is very unusual; normally, we see robust growth in aggregate balances – even during recessions – whether inflation is contained or otherwise.
So, how should inflation impact our risk models?
On the credit loss side, the key point is that a combination of high unemployment and high inflation should have a more muted effect on losses than a recession marked by grinding deflation. While the economy is expanding, inflation is not exactly positive for credit performance, but it’s not especially terrible either. This suggests we should consider a regime-switching model where credit losses attributed to unemployment can be dialed down if inflation is expected to be high.
Bear in mind that what we’ve covered thus far is based on distant memories of past inflationary episodes, with a considerable pinch of chin-scratching and speculation. Indeed, right now, inflation doesn’t even show up as a significant driver in most quantitative credit models.
My contention – that inflation will prevent excessive losses in a recession – can’t be tested until we actually experience such a thing. If prognosticators are correct and a recession happens in 2023, the evidence gathered will guide the way risk models should be subsequently updated.
I have written in the past of how recessions, pandemics and outbreaks of inflation are weird and rare. Our job as risk managers is to understand, to the fullest possible extent, the range of human experiences and their potential impact on the portfolios we shepherd. We can’t ever afford to let a “good recession” go to waste.
The future will always surprise us. The point of models is to summarize our past learnings and point us to where the evidence leads.
The evidence-based approach is impossible to better.
Tony Hughes is an expert risk modeler. He has more than 20 years of experience as a senior risk professional in North America, Europe and Australia, specializing in model risk management, model build/validation and quantitative climate risk solutions.