Banking regulators recently revised their Uniform Bank Performance Report (UBPR) to improve the report’s ability to identify and analyze interest rate risk (IRR). This publicly available report is designed to provide a comprehensive view of a bank’s performance and risk profile. Using some real-life examples, I’ll show where the UBPR has improved, where it falls short, and discuss potential improvements to the source data.
What is the UBPR?
The UBPR is an analytical report developed by the FFIEC. Reports for individual banks are available on the FFIEC website. The report compiles Call Report data to show trends in a bank’s performance and risk profile. It also compares that performance and risk to the bank’s peers. The UBPR displays five quarters of data. The default approach will show data for the current quarter, the same quarter for the prior year, and for the three previous year-ends. You can, however, customize the trend. I usually found using the past five consecutive quarters better for monitoring purposes, but the default approach can be better at spotting longer term trends.
The UBPR uses both dollar amounts and ratios. The report shows the bank’s ratios, the median ratios for its peer group, and the bank’s percentile within its peer group. Although the UBPR lists 120 different possible peer groups, all commercial banks with more than $100 billion in assets fit into the same peer group. This means that JP Morgan Chase is in the same peer group as Regions Bank even though it’s more than 20 times the size.
The UBPR has also tended to be less useful when it comes to analyzing IRR. This is partly attributable to the scant amount of IRR data found on the Call Report. However, the UBPR had also not optimized the IRR information that was available. Recently, however, the FFIEC added 58 new ratios to the IRR section of the UBPR. The new report includes trends in net interest income. (Some of this information was already reported on other schedules.) The revised schedule also provides additional details on unrealized gains and losses on held-to-maturity (HTM) and available-for-sale (AFS) securities and on the composition of loan and investment portfolios. The IRR schedule now also includes further details on deposits and other liabilities.
These revisions took effect not only prospectively but also on historical reports going back to 2001, subject to data availability. This allows analysts to see how well the revised schedule would have identified banks with excessive IRR. Legacy UBPRs using the old ratios are not readily available, however.
Some Examples
Net Interest Margin
The revised UBPR added several ratios related to net interest margin (NIM). The IRR schedule includes not just net interest income (already available on a separate UBPR schedule) but also changes to NIM over time. This can show how vulnerable a bank’s earnings were to interest rate risk. Here is a screenshot for First Republic as of March 31, 2023:
First Republic showed a NIM of only 1.75%, well below the peer group median (PG 1) of 2.80% and in the 19th percentile (PCT) for its peer group. The trend is even worse, with the NIM down 94 bps YOY, worse than 97% of its peer group. This clearly shows a bank in a lot of trouble. But look just six months earlier (shown below). In September 2022, First Republic’s income ratios were worse than peers, but its NIM was just 9 basis points below the peer group median and was slightly up YOY. You didn’t see real deterioration until December 2022.
NIM trends provided even less of an early warning indicator for SVB, where NIM was still up slightly YOY in the quarter before the bank failed. NIM just isn’t that much of a leading indicator, especially if the NIM decline is due to interest rate risk.
Unrealized Losses
Unrealized losses on securities are a better early warning indicator. Some of this information was already on the UBPR, but it has been expanded with more granular data. Levels and trends in unrealized losses can tell us a lot about a bank’s condition and its IRR profile. As of December 2022, unrealized losses on securities at SVB represented more than 100% of capital. Even more dramatic was the change between 2020 and 2021, where the bank went from an unrealized gain of 19.77% (90th percentile) to an unrealized loss of 7.67% (3rd percentile). (The real turning point came between March and June 2021, nearly two years before the bank failed.)
Focusing on unrealized losses has its limitations. Banks like First Republic, which had a concentration in long term mortgages rather than MBS, won’t show up as outliers. However, both banks stick out in the section “Long Assets with Options.” First Republic’s concentration of Mortgage Loans & Pass-throughs placed it in the 96th percentile as of December 2022. SVB placed in the 90th percentile. Paradoxically, this is a legacy line item that the FFIEC plans to eventually phase out. There is some overlap between this section and a newly added section related to long-term assets. However, mortgages and MBS differ from, say, long-dated U.S. Treasuries in that the duration of mortgage-related assets tends to increase as market interest rates increase.
Deposits and Funding
Looking only at the asset side won’t tell the whole story. As of December 2022, Bank of America’s unrealized loss represented 62% of capital. But while 93.82% of SVB’s deposits were uninsured, only 34.07% of BofA’s were. Moreover, depositors, even large ones, are less likely to run for the exits at a too big to fail bank. But also consider Charles Schwab Bank SSB. By December 2022, unrealized losses were 88.20% of capital. However, only 20.67% of deposits were uninsured. Banks with less concentrated funding face less threat to their viability. That doesn’t mean we should be dismissive of IRR at BofA or Schwab. Though bank runs are less likely at these banks, they certainly are possible. And the large depreciation of their investment portfolios increases the potential cost to the FDIC. Nor was IRR exposure at these banks temporary. Unrealized losses still represented 45.45% of capital at Charles Schwab as of March 2025 (6th percentile). They represent 51.07% of capital (3rd percentile) at Bank of America.
Potential Improvements to the UBPR
The UBPR reports levels and trends in nonmaturity deposits (NMDs), but it’s unclear how we should interpret the data. Trends can be useful, especially if we see banks replacing NMDs with borrowings or other wholesale funding. But treating NMDs simply as short-term liabilities can be misleading since, for most banks, rates on these deposits significantly lag market rates.
More useful would be to consider uninsured deposits. The worst-case scenario for a bank with high structural IRR is a combination of rising rates and deposit outflows. Banks with high concentrations of uninsured deposits are most vulnerable to this scenario. Banks over $1 billion already provide an estimate of uninsured deposits on the Call Report, so this information can easily be incorporated into the UBPR. Amazingly, total uninsured deposits are not reported on the Liquidity & Funding schedule of the UBPR either. Uninsured demand deposits, which are the most vulnerable to a run, are still considered “core deposits” on the UTPR.
Improving the Call Report
Other improvements to the UBPR would also require changes to the Call Report. IRR is most useful as a forward-looking measure, but the income measures in the Call Report are backward looking. One improvement would be to provide information on current yields on assets and costs of liabilities. This information can, at best, only be inferred from the current Call Report. Yield and cost information would be especially useful for deposits since the current nomenclature tells us so little. A “savings account” at Bank of America or Wells Fargo (each paying 0.01%) is much different from one at Synchrony (4.00%) or Goldman Sachs (3.65%).
More information on asset yields would also be useful. Low yielding mortgages and MBS tend to have longer weighted average lives since they provide borrowers with less incentive to refinance and prepay. This information would also allow analysts to at least estimate unrealized losses on mortgage loans. In addition, information on current yields and costs provides a more forward-looking indicator of net interest income sensitivity. Let’s look again at First Republic. While the current UBPR already showed a sharp decline in NIM, it didn’t tell the whole story. A look at yield information as of March 31, 2023 shows that First Republic had an asset yield of only 3.66%. Meanwhile, the bank was becoming increasingly reliant on borrowings, with a weighted average cost of 4.33%. While the bank’s overall spread remained positive, its spread on new (wholesale) funds was negative.
Naturally, we would expect that attempts to expand the Call Report to meet with industry resistance. While there can be real concerns about regulatory burden, keep in mind that this is information that most banks already have. Banks regularly report yield and cost information in their securities filings and need this information to run internal earnings-at-risk and economic value of equity estimates. Data mapping and aggregation processes may differ across banks, but regulators can take steps to avoid extensive data re-mapping. If a bank cannot produce weighted average coupons for its deposits, investments, and mortgages, that alone probably tells you something about the quality of its IRR management.
Leave a Reply