The other day, a reporter asked me the question posed in the title. And I looked at several bank 10-Ks and call reports accessible from the Federal Reserve’s website. I could not tell or even begin to guess an answer to this question. Why? Because publicly listed banks are not required to explicitly report fair values of their loans collateralized by commercial real estate. Neither are they explicitly required to report the fair value of the commercial real estate underlying these loans with sufficient granularity.
Valley National Bank example
Consider an example to make this problem more concrete. Valley National Bank, based in New Jersey, is rumored to have a large exposure to commercial real estate. I pulled out their 10-K for the year ended December 31, 2023, and found the following disclosures.
As you can see below from the loan portfolio disclosures that appear on page 67 of their 2023 10-K, $28 billion of their $50 billion loan portfolio (64%), is concentrated in commercial real estate. But these numbers reflect the historical cost of these loans, not their fair value. Meaning, we do not know how much of these loans are likely to be recoverable once the potential deterioration in the value of the collateral is considered.
Another disclosure on page 67 of their 2023 10-K states that their commercial real estate is well diversified across Florida, Alabama, New Jersey and Manhattan and the combined loan to value ratio is only 57%. That is, the value of the real estate could fall by 43% before loan values equal the real estate. This sounds reasonably safe, but the ratio represents a combined number across regions. The commercial real estate problem is arguably far worse in Manhattan than in Alabama.
Under the level 1/2/3 hierarchy disclosures reproduced below, Valley reflects the combined fair value of all loans. As you can see, the carrying amount of loans is $49.7 billion whereas the fair value of these loans is $47.9 billion. Is this $1.8 billion of unrealized losses the smoking gun we are looking for w.r.t commercial real estate? May be. It is hard to know for sure as real estate loans are explicitly broken out. Interesting though, is that the fair value of HTM (held to maturity) residential MBS (mortgage backed securities) are explicitly reported and, as you can see, embed $330 million of unrealized losses ($2.88 billion - $2.252 billion).
Somewhat disconcerting is that the sum of unrealized losses of roughly $2.2 billion ($1.8 billion in loans, $442 million in HTM securities offset by a saving in liabilities of 72 million) is almost a third of the bank’s book value of equity of $6.7 billion.
So, where does all this leave us? Perhaps there is no need to worry about Valley’s solvency. May be. We don’t know for sure.
Macro evidence on the problem
I then took the question to two of my brilliant Columbia Business School colleagues who have written clever papers trying to estimate the scale of the crisis: Stijn Van Nieuwerburgh and Tomasz Piskorski. They have calculated the scale of the brewing crisis at the macro level for the whole economy but were somewhat unsure of which specific banks might be affected.
Stijn pointed me to his paper titled “Work From Home and the Office Real Estate Apocalypse.” Stijn and his co-authors try to infer the extent of value destruction in the office market by looking at lease-level data for 105 office markets throughout the United States over the period January 2000 to December 2022. They document a 17.38% decrease in lease revenue in real terms between December 2019 and December 2022. As of 2023.Q3, they find that 22.1% of office space was available for rent in New York and 30.4% in San Francisco. After a lot of clever modeling, they find a $664.1 billion decline in office values over the three-year period 2019-2022.
This begs the question of whose books are these losses sitting on. At an overall level, we know that commercial banks have about $2.4 trillion in commercial real estate loans on their balance sheets as of June 2022 according to Call Report data. Banks account for 61% of CRE (commercial real estate) debt holdings. The question I am interested in, of course, is which specific banks are sitting on these losses. Well, that is hard to know.
The 10-K of a bank lists the “amortized cost” of the loans given to the commercial real estate sector. If the system worked well, the allowance of loan losses provided against these loans should ideally reflect the losses that Stijn and his co-authors document. But does the system work as intended? I am not sure and there is no easy way to find out.
Tomasz pointed me to his paper titled, “Monetary Tightening, Commercial Real Estate Distress, and US Bank Fragility.” In that paper, Tomasz and his co-authors, have access to a sample of 35,253 loans totaling $825 billion in aggregate principal balance from the CMBS (commercial mortgage backed securities) market. These loans, drawn from the outstanding CMBS loans as of December 2023, were obtained from the DBRS Morningstar database. Tomasz and co-authors estimate bank losses in the order of $100 to $200 billion given 10 to 20% default rate. Based on the CMBS data they analyzed: the following proportion of loans appear to be underwater: (i) 14% of all CRE loans; (ii) 44% of all office loans; and (iii) 12% of multifamily loans. They go on to estimate that more than 1/3 of all loans and majority of office loans would not be able to refinance in regular refinance market.
Then they access publicly available data on the amortized costs of loans on bank balance sheets. They make a bunch of assumptions about credit default and interest rate scenarios and estimate potential losses that banks will have to swallow. They conclude that the 2022 interest rate hike, commercial real-estate distress would add up to an additional $160 billion of losses to more than $2 trillion decline in the value of bank assets due to higher interest rates.
This is brilliant forensic work but its pretty hard for an investor to know whether these estimates are accurate given that they are necessarily based on coarse historical cost data of loans. In particular, how does the investor know whether CEOs have adequately provided for these credit losses in their financial statements? Or whether the auditors have forced them to do so and/or issued going concern opinions on these banks?
Opacity hobbles governance and oversight
Lest you think that hedge funds can easily guess which banks are most affected and hence short the relevant bank stocks, the case of Arbor Realty Trust, a real estate financing company might prove instructive. Arbor is a publicly traded mortgage REIT that is potentially managing earnings. On paper, it should be easy to short the stock. But so far, the shorts have struggled because the fair value of Arbor’s multi-family loans portfolio are not disclosed, and such fair value can be difficult to estimate as the REIT changes contract terms on potentially delinquent loans.
To put numbers in perspective, I pulled up the 10-K of Arbor for the year ended December 31, 2023. Arbor recorded a large increase in allowance for credit losses for its multifamily portfolio from around $19.2 million in 2022 to $72.8 million in 2023. On top of that, Arbor reports $272 million of multi family loans as newly impaired in 2023. To place these numbers in context, Arbor reports book equity of $3.2 billion and loans and investments of $12.377 billion, as of December 31, 2023.
In 2023, Arbor has restructured several loans in the multifamily category of around $520 million. It is not obvious what these restructuring initiatives made the reported balances for loans, allowance for losses and non-performing assets look better. Most likely they did.
The fair value disclosures, for all loans combined, do not seem to suggest a full blown crisis. If anything, the fair value of loans is higher than carrying amount, presumably because the impact of lower interest rates on fair value has offset any distress related impairments of loans. However, it would be nice to be able to see the fair value of each loan type separately (multi-family, land, retail etc.).
So, where does this leave us. On the surface, things don’t look as alarming for Arbor, if we take these somewhat aggregate and opaque disclosures at face value. Are the shorts wrong? And, how do we know for sure?
Policy fixes
So, what to do? Here are suggestions for policy makers:
- The Fed likely has loan by loan data for loans given out by banks. Is there a case to make that data public or even allow limited release to investors and short sellers or so that they can assess which bank has the highest exposure to commercial real estate?
- Make the loan level data on mortgages more easily available, in general. All mortgages are recorded in the deed records and hence, in theory, already publicly available. One can purchase some data from companies like CoreLogic and Black Knight, who aggregate and clean them. The process is needlessly complicated and expensive which ends up thwarting research on CRE debt. Maybe the Fed should make this data easily and freely available for academic purposes.
- Can the FASB or the SEC ask banks to disclose fair value of loan assets for each loan category? Or declare the loan to value ratios of the commercial real estate underlying these loans, separately by region and by type of loan (commercial real estate, apartment loans etc.)? The regional breakdown is important. Presumably, the work-from-home movement has had a bigger impact on Manhattan office buildings than the ones in say Charlotte, North Carolina.
- Does the PCAOB want to investigate audits of banks likely to have the highest exposure to commercial estate to ensure that auditors have asked management to provide for these loan losses in a timely manner? And/or, why have going concern opinions not been issued by the auditors?
The PCAOB, in particular, needs to think hard about what the auditor can and cannot factor in when auditors consider going concern opinions. When my co-authors and I documented that auditors had not issued enough going concern opinions leading up to the 2008 financial crisis, we were told that auditors had factored in the possibility of a Fed sponsored bailout of the weak banks.
Some will no doubt argue that strategic ambiguity about the value of bank collateral is necessary to avoid a depositor led run on the affected banks and hence stop the fallout from any contagion that such a run may cause in the financial system. But that sounds like a strange argument to me. Why are we keeping investors of banks in the dark and bailing out CEOs who made aggressive bets that may have gone bad after interest rates rose?
This article was originally published on Forbes.com.