Commercial Real Estate and Bank Systemic Risk

The link between banking system performance and the outlook for commercial real estate (CRE) is complex. Bank CRE loans include more than dubious quality loans made to finance office buildings and shopping centers. They include acquisition, development and construction (ADC) loans for 1- to 4-family residential properties, ADC loans for other types of properties, loans secured by multifamily properties, owner occupied and non-owner occupied commercial properties and commercial real estate loans not secured by real estate.

While not every one of these bank CRE loan categories is expected to suffer from a “COVID hangover,” the outlook for the performance of all CRE loans has been negatively impacted by the increase in interest rates. As existing CRE loans come due, borrowers will face higher loan servicing costs which could make some projects and properties uneconomic to refinance.

Regardless of your view about the future performance of CRE and its implications for the banking system, it is important to understand that, using the favored supervisory measure of bank CRE loan concentration – bank CRE loans outstanding to bank Tier 1 capital plus the bank’s allowance for loan and lease losses (ALLL) – the banking system’s current CRE exposure and performance statistics look remarkably similar to the banking system’s CRE exposure and performance statistics circa December 2019. That was the last regulatory data observation before the onset of the COVID crisis.

Supervisory CRE loan concentration measures are little changed between December 2019 and December 2023. To the best of my recollection, bank CRE exposure was not identified as problematic in 2019, so presumably, without the COVID hangover created by lockdowns and government-stimulus-induced inflation, bank CRE exposures would not be viewed as problematic today. Bankers continued to make commercial real estate loans as they had done prior to COVID. Their mistake, if they have made one, was their failure to recognize that the government policies implemented to battle COVID would eventually undermine the demand for some types of real estate and sow the seeds for inflation and higher interest rates.

Not “Seismic” – Yet

Bank CRE loan performance is not yet a significant stability issue. The December 2023 banking system CRE loan performance profile looks similar, maybe even slightly better, when compared to the system’s CRE loan performance profile in December 2019. There is some evidence of bank “extend and pretend” behavior in some CRE loan classes, primarily non-owner occupied, nonfarm, nonresidential loans made by some large banks and “other ADC” loans made by many smaller banks. But the widely anticipated CRE seismic loan default event has yet to materialize, at least according to December 2023 regulatory data.

AEI’s Paul Kupiec: Interest rates and unrecognized losses.

What has changed since COVID is the prevailing level of interest rates. The post-March 2022 increase in interest rates has inflicted large unrecognized mark-to-market losses on bank securities, loan and lease portfolios.

For the most part, these losses are not reflected in bank Tier 1 capital levels – a key component of the supervisory measure of CRE concentration risk. When banks’ Tier 1 capital balances are adjusted to reflect estimates of the unrecognized interest rate losses banks have suffered since early 2022, banks’ market-value-adjusted CRE concentration ratios are much higher than December 2023 supervisory CRE concentration ratios suggest.

Higher interest rates impact both the demand and the ability to refinance bank CRE loans. Current interest rates are much higher than when many existing bank CRE loans were struck. Defaults will occur when returns from commercial property cash flows do not justify refinancing properties at a new higher interest rate.

Moreover, refinancing properties at higher rates may not be an option in some cases since banks typically require minimum interest coverage ratios to qualify a loan – ratios some borrowers may not satisfy at today’s higher interest rates. In some cases, banks and borrowers may find a troubled debt restructuring preferable to a CRE loan default, but there is widespread anticipation that, in many cases, default will be the preferable option.

Concentration Ratios

Regulatory guidance suggests that bank CRE concentration ratios above 3 may be indicative of excessive CRE loan exposure. (The preferred supervisory CRE concentration ratio is a bank’s total CRE loans divided by the sum of its Tier 1 capital plus ALLL balances.)

At a minimum, regulatory guidance suggests that supervisory CRE concentration ratios greater than 3 warrant additional examiner scrutiny. Using the supervisory CRE loan concentrations measure, there are 1,763 banks with CRE concentration ratios greater than 3, including 274 with concentrations greater than 5 and 77 over 6. Most of these CRE concentrations are at smaller banks, but there are 2 banks with assets between $50 billion and $250 billion with supervisory CRE concentration ratios greater than 5.

Banking system CRE concentration risk is more pronounced once a bank’s unrecognized mark-to-market interest rate losses are taken into account.

Source: Paul H. Kupiec, Commercial Real Estate and Bank Systemic Risk, page 8.

My adjustments to bank-reported Tier 1 capital levels account for actual unrecognized interest rate losses on a bank’s securities, and estimated losses on its loans and leases. After incorporating unrecognized interest rate losses into the CRE loan concentration measure, the number of banks with market-value-adjusted CRE concentration ratios above 3 swells to 2,916, including 19 banks with assets between $50 and $250 billion and 2 with assets greater than $250 billion. Of the aforementioned 19 banks, 8 have market-value-adjusted CRE concentration measures greater than 5, 7 of which are greater than 6.

Marking to Market

I construct a profile of banking system systemic risk using my CRE loan concentration measure that uses the mark-to-market estimate of a bank’s Tier 1 capital in place of the bank’s reported Tier 1 capital level. I consider CRE loan concentrations for each specific CRE loan type as well as for bank aggregate CRE loan exposures. This disaggregated approach provides insight into the impact of alternative CRE loss events on the market-value solvency condition of banks.

Market-value insolvency has been the proximate cause of recent bank failures like First Republic and SVB in 2023. The April 2024 failure of Republic First Bank demonstrates the merit in such an approach. Republic First had a high concentration of CRE loans and a large unrecognized interest rate driven market-value loss on its assets.

As of December 2023, the most recent regulatory data available, Republic First reported $5.88 billion in assets including $1.6 billion in CRE loans. With $288 million in regulatory capital, Republic First had a Tier 1 leverage ratio of 4.4% and a supervisory CRE Tier 1 + ALLL concentration ratio of 5.19.

According to my calculations, on a mark-to-market basis, Republic First had over $600 million in unrecognized mark-to-market losses, making the real market value of its Tier 1 capital negative $400 million. The FDIC cost of resolving of Republic First largely agrees with this unrecognized-loss estimate considering that the FDIC loss also likely reflects an additional discount for the bank’s CRE loan concentration. The Republic First resolution cost the deposit insurance fund $667 million.

Test for Deposit Insurance

I estimate the number of banks and share of total banking system assets that would become market-value insolvent from CRE loan losses that range between 10% and 25%. Such loss rates are lower than several publicized loss rates including the 30% rate recently incurred when the FDIC sold CRE loan assets held by Signature’s bridge bank.

The systemic risk created by a wave of CRE loan defaults that render a large number of banks market-value-insolvent with a high risk of failure, and the large share of banking system assets held in these banks, will in part depend on how the public reacts to this realization. Public confidence in the adequacy of the FDIC deposit insurance fund will be especially important. The deposit insurance fund’s current balance is equivalent to but 0.51% of the $23.69 trillion in banking system total assets (as of December 2023).

After adjusting bank CRE concentration ratios for banks’ unrecognized mark-to-market interest rate losses, the data suggests that even modest CRE loan stress losses could test the solvency of hundreds of banks. While many of these banks are small, in aggregate the number of potential failures would likely pose a challenge to the deposit insurance fund’s resolution resources.

Bank resolutions would likely be strung out over time and increase insurance fund losses, as happened in the 1980s Savings and Loan Crisis and 2008 Great Financial Crisis. Larger CRE losses, or even modest simultaneous losses across several CRE loan categories, would almost certainly test the public’s confidence in the deposit insurance safety net.

Paul H. Kupiec is a senior fellow at the American Enterprise Institute and previously was an associate director of the Division of Insurance and Research within the Federal Deposit Insurance Corp. Center for Financial Research. This article is a lightly edited excerpt from Mr. Kupiec’s paper Commercial Real Estate and Bank Systemic Risk.