# Bank Desk · What it means how to read every measure on the bank desk. Live page: https://getbookiq.com/bank-desk/#meaning --- HOW TO READ THIS DESK A bank's filing is a set of claims about itself. The useful questions are always the same four: can it fund itself if depositors leave, is the capital real once you mark the securities, is the business earning enough to absorb mistakes, and is the loan book telling the truth yet. Every number below is placed under one of those questions. The percentile chips compare a bank against banks of similar size in the same quarter; a chip turns red when the number sits in the worst fifth of the band for that measure, green when it sits in the best fifth. CAN IT FUND ITSELF · LIQUIDITY AND DEPOSITS Loans / deposits LNLSDEPR · net loans over total deposits MEANS How much of the deposit base is already lent out. Deposits are the cheapest, stickiest funding a bank has; everything above them must be borrowed. WHEN HIGH Above 100% the bank has lent more than depositors gave it and the difference is borrowed money, priced by markets and callable. Growth is outrunning the franchise. These banks need to sell assets, raise deposits, or slow. WHEN LOW Below 70% the bank is sitting on cash and securities it cannot deploy locally. It is liquid, safe, and yield-starved: the natural buyer of loans and participations, because its problem is finding assets, not funding them. VS PEERS Community banks cluster 75-90%. A bank far above its band is funding differently from its peers; ask how. OUR LINE > 100 = flag · < 70 with liquid > 25% and leverage > 9% = natural buyer screen LIMITS A wholesale-funded model can be stable if the funding is long and matched. The ratio says where the money came from, not whether it will stay. Liquid assets / assets (CHBAL + FREPO + SC) / ASSET · cash, fed funds, securities MEANS The buffer that pays depositors who leave before loans can be sold. This is what a run consumes first. WHEN HIGH Comfortable, provided the securities are worth their book value. Pair with the unrealized loss line below: a bank that is 60% 'liquid' in bonds trading at 80 cents is not 60% liquid. WHEN LOW Under 10% the bank has no cushion. Any funding shock becomes a forced sale of loans at whatever price is offered. VS PEERS Read together with uninsured deposits. High uninsured plus low liquid is the 2023 profile. OUR LINE < 10 = flag · < 20 with uninsured > 50% = run vulnerability (double weight in the score) LIMITS Book values, not market values. Securities count at par here and at the fair value of what they would actually fetch only in the securities-loss line. Uninsured share of deposits DEPUNINS / DEP · deposits above the $250K insurance limit MEANS How much of the deposit base has a reason to leave the moment confidence wobbles. Insured depositors rarely run; uninsured ones run in a day, by phone. WHEN HIGH Above 50% the bank's funding is a confidence contract. Silicon Valley Bank was 87% uninsured four quarters before it failed, Signature 92%, First Republic 74%. Their credit was fine. Their depositors were not obligated to stay. WHEN LOW Retail-heavy banks sit below 25% and their deposits behave like furniture. Boring is safe here. VS PEERS Business and wealth-focused banks run structurally high. That is a model choice, not a flaw; the question is whether liquidity and capital are sized for it. OUR LINE > 50 = flag · combined with liquid < 20 = run vulnerability · uninsured down 20%+ YoY = the large depositors are already leaving LIMITS Operating accounts of businesses are uninsured but sticky; hedge-fund and venture cash is uninsured and gone at the first headline. The filing does not distinguish them. Brokered deposits · wholesale funding BRO/BROR · (OTHBOR + FREPP + BRO) / ASSET · FHLB advances MEANS Money the bank rented rather than earned: deposits bought through brokers, Federal Home Loan Bank advances, fed funds. It is flexible, priced at market, and leaves when rates move. WHEN HIGH Brokered above 10% or wholesale above 15% of assets means the bank is leaning on markets to fund its book. Regulators restrict brokered deposits for banks that fall below well-capitalized, so a stressed bank loses this funding exactly when it needs it. WHEN LOW Low reliance is the mark of a franchise that funds itself. Core deposit ratio is the mirror image. VS PEERS FHLB reliance is common in mortgage-heavy banks and is not by itself a warning. Trend matters more than level. OUR LINE brokered > 10 · wholesale > 15 = flags LIMITS FHLB advances are collateralized and reliable until the bank's condition triggers collateral haircuts. The line is a funding dependence measure, not a default predictor. Deposit growth · uninsured deposit growth DEP and DEPUNINS vs the same quarter one year earlier MEANS Direction of the funding base. Balances shrinking while loans grow is the squeeze; uninsured balances shrinking faster than total is the early run. WHEN HIGH Fast deposit growth funded by brokered money or rate specials is bought growth and reprices. Fast growth in uninsured balances concentrates run risk. WHEN LOW Deposits down 5%+ year over year while loans are flat forces the bank into wholesale funding or asset sales. Uninsured down 20%+ means the depositors with options are exercising them. VS PEERS Compare against the band: in a year when the whole industry loses deposits, a bank at the median is not the story. OUR LINE deposits < -5% YoY = flag · uninsured < -20% YoY = flight flag LIMITS Mergers and acquisitions distort year-over-year growth; a bank that bought a competitor shows a jump that is not organic. IS THE CAPITAL REAL · CAPITAL Tier 1 leverage ratio RBC1AAJ · Tier 1 capital over average assets, the prompt-corrective-action measure MEANS The plainest capital number: how much loss the bank can absorb before depositors and the insurance fund take it. Regulatory categories attach to it directly. WHEN HIGH Community banks typically carry 9-12%. High leverage capital plus low ROA can mean a bank that cannot deploy its equity profitably, which is its own problem, but not a safety one. WHEN LOW Below 5% the bank is no longer well-capitalized and supervisory restrictions begin. Below 4% is undercapitalized, below 2% critically so, and closure follows quickly. The median modern failure was at 6.1% four quarters before the end and 1.65% at the final filing. VS PEERS A bank under 7% while its band sits at 10% is running with a third less cushion than everyone its size. OUR LINE < 7 = watch · < 5 = critical, double weight · falling two straight quarters = trend flag LIMITS Book capital ignores unrealized losses on held-to-maturity securities. A bank can report 9% leverage while its bonds are underwater by more than its equity. That is the next line. Unrealized securities loss / equity (SCAF − SCAA) + (SCHF − (SC − SCAF)), over EQTOT · both AFS and HTM books MEANS What the bond portfolio would lose if sold today, across both accounting books, measured against the equity on the books. Held-to-maturity losses never touch reported capital, so this is computed from filed fair values rather than read off the balance sheet. WHEN HIGH Above 30% of equity, a meaningful share of the capital exists only if the bank is never forced to sell. Republic Bank carried 337% of equity in unrealized losses a year before failing; Metropolitan Capital 505%; Citizens Bank Sac City 141%. Silicon Valley Bank's final filing showed 114%. WHEN LOW Near zero means the portfolio is short or the rate environment is kind. Low is not a virtue; it simply removes one way to die. VS PEERS The whole system carried losses equal to 13% of equity at mid-2026. A bank at 40% is three times more exposed than average to the one shock that turns liquidity into insolvency. OUR LINE < -15 = watch · < -30 = danger LIMITS Unrealized losses only become real when liquidity forces a sale. A bank with sticky insured deposits can hold to maturity and the loss evaporates. Read this line together with uninsured deposits and liquid assets, never alone. CET1 · equity / assets · AOCI RBCT1CER · EQTOT / ASSET · EQCCOMPI / EQTOT MEANS Three complements to leverage. CET1 is risk-weighted, so a bank full of mortgages looks stronger than one full of construction loans at the same leverage. Equity/assets is the raw book. AOCI is the part of the securities loss that already flows through equity (the available-for-sale book). WHEN HIGH A gap between high CET1 and modest leverage means low-risk-weighted assets, usually mortgages or Treasuries. Large negative AOCI means the AFS book is underwater and it is already visible in reported equity. WHEN LOW AOCI near zero with a large total securities loss means the losses are parked in held-to-maturity, where they are invisible to book capital. That gap is the SVB structure. VS PEERS Compare AOCI against the total securities-loss line: the difference is what the bank chose to hide in HTM. OUR LINE AOCI < -20% of equity = flag LIMITS Risk weights are regulatory conventions, not market judgments. CET1 rewarded holding long Treasuries right up until 2022. IS IT EARNING ENOUGH · EARNINGS Return on assets ROA · annualized net income over average assets, as published by the FDIC MEANS The single best summary of whether the business works. It is what builds capital when things go right and what drains it when they do not. WHEN HIGH Above 1% is the classic healthy community bank. Very high ROA at a small bank deserves a look at where the income comes from; fee-driven or one-time gains do not repeat. WHEN LOW Negative ROA eats capital every quarter it persists. The median modern failure was already losing money eight quarters before the end and never recovered. Chronic low-positive ROA (under 0.3%) means no cushion is being built for the next bad year. VS PEERS The chip compares against banks the same size in the same quarter, which strips out the rate environment. A bank at p10 in a good year has a business problem, not a cycle problem. OUR LINE < 0 = flag LIMITS Annualized from year-to-date, so first-quarter figures are the noisiest. One-time items (securities gains, tax events, sale of a business line) can swing a small bank's ROA by a full point. Net interest margin NIMY · interest income less interest expense, over earning assets MEANS The spread the bank earns for turning deposits into loans. It is the engine; everything else is cost and credit. WHEN HIGH Wide margins come from cheap deposits, higher-yielding loans, or both. Wide margins with rising delinquency mean the bank is being paid for risk it is taking. WHEN LOW Under 2.5% is thin. A margin compressing while cost of funds rises is the squeeze that turns adequate banks into marginal ones without a single bad loan. VS PEERS Band context matters: large banks run structurally thinner margins on fee income. A small bank at 2.3% is in trouble; a money-center bank at 2.3% is normal. OUR LINE < 2.5 = flag LIMITS NIM measures spread, not volume. A shrinking balance sheet can hold NIM while net income falls. Efficiency ratio EEFFR · noninterest expense over net revenue, as published MEANS How many cents it costs to earn a dollar of revenue. It is the cost-structure measure, and it is the widest single separator between banks that survive and banks that disappear. WHEN HIGH Above 85% the cost base is eating almost all the margin. Above 100% the year cost more than it earned. Small banks die of this more often than they die of bad loans. WHEN LOW Below 60% is efficient. Very low ratios at small banks can mean underinvestment in systems and compliance, which shows up later as findings. VS PEERS The chip is band-relative because scale drives cost. A $200M bank at 75% may be fine; a $5B bank at 75% has a problem. OUR LINE > band p90 = flag LIMITS Ratio of two flows; a bank shrinking revenue faster than expense looks worse without any new spending. Provision / assets ELNATR annualized over ASSET · what the bank set aside for expected losses this period MEANS Management's own forecast of credit losses, in dollars. It is the one number on the income statement that is a prediction rather than a record. WHEN HIGH A spike means management sees losses coming or an examiner made them see it. Provisions rising ahead of charge-offs is the honest sequence. WHEN LOW Provision persistently below net charge-offs is reserve bleed: the allowance is paying for today's losses instead of funding tomorrow's. It flatters earnings until it cannot. VS PEERS Compare provision against the bank's own charge-off line, not against peers. The question is internal consistency. OUR LINE no standalone flag · used in the reserve-coverage read below LIMITS CECL changed the accounting in 2020-2023; year-over-year comparisons across the transition are not clean. IS THE LOAN BOOK TELLING THE TRUTH · ASSET QUALITY Noncurrent loans / loans NCLNLSR · loans 90+ days past due plus nonaccrual, over gross loans MEANS Loans that have stopped performing and are on their way to loss. This is the lagging measure: by the time a loan is here the decision that made it bad happened a year ago. WHEN HIGH Above 2% is elevated for a community bank. The modern failure signature runs from 4.59% eight quarters before the end to 7.48% at the final filing. In the 2011 to 2014 crisis tail it ran 8.58% to 14.31%: those banks were already broken when the panel picks them up. WHEN LOW Under 0.5% is clean. Very low noncurrent alongside negative earnings is an anomaly worth asking about: a lender with no credit problems that cannot make money has a cost problem or a reporting problem. VS PEERS The chip is band-relative; the system sits near 0.9%. A bank at p90 of its band has a problem book regardless of the absolute level. OUR LINE > band p90 = flag LIMITS Noncurrent is a promise already broken. It confirms trouble; it does not predict it. Pair with the 30-89 day line for the pipeline. 30-89 days past due · net charge-offs · reserve coverage P3LNLS / LNLSGR · NTLNLSCOR · LNLSRES / NCLNLS MEANS Three views of the loss pipeline: what is entering it (early delinquency), what is leaving it as realized loss (charge-offs), and whether the allowance covers what is already identified (reserve coverage). WHEN HIGH Rising 30-89 with flat noncurrent means the wave is arriving. Charge-offs above the band p90 mean losses are being realized, not just brewing. Reserve coverage well above 100% means management is ahead of the problem. WHEN LOW Reserve coverage under 100% means the allowance does not even cover loans already identified as nonperforming. Every additional dollar of noncurrent comes straight from earnings or capital. VS PEERS Charge-off comparisons only mean something within a band and a loan mix; a credit-card bank and a mortgage bank are not comparable on this line. OUR LINE charge-offs > band p90 = flag · reserve coverage < 100% with noncurrent > 0 = flag LIMITS Charge-off timing is a management choice within limits; a bank can look clean by holding losses in nonaccrual, then recognize them all at once. WHAT IS IT CONCENTRATED IN · EXPOSURES Loans to nonbank financial institutions LNNDEPD · Schedule RC-C item 9.a · shown in dollars and as % of equity MEANS Lending to other lenders: warehouse lines to mortgage companies, auto finance companies, consumer lenders, private credit funds. This is the funding rail behind nonbank credit, and it lives here, not on the bank's own auto or mortgage lines. WHEN HIGH Above 50% of equity the bank has a second-order credit exposure larger than half its capital: it is exposed to the underwriting of institutions it does not control. Above 100%, a single borrower's failure can be existential. Origin Bank carried 84% when its auto-lender borrower failed. System-wide the line grew from $238B to $1.5 trillion in eleven years. WHEN LOW Zero means the bank does not play in this market. Most community banks do not. VS PEERS The chip compares exposure-to-equity within the band. Fast year-over-year growth in this line is worth as much attention as the level. OUR LINE > 50% of equity = flag · > 100% = high LIMITS The line is all nonbank financial institutions, not auto-specific. The FFIEC began sub-categorizing it in recent reporting changes; those splits are the natural next field. Exposure here is only as good as the bank's collateral monitoring, which the filing cannot show. Construction / equity · real estate · C&I · consumer mix LNRECONS / EQTOT · LNRE, LNCI, LNCON over gross loans MEANS What the book is made of. Concentration is not a flaw; it is a bet, and the filing shows which bet. WHEN HIGH Construction above 100% of equity crosses the supervisory concentration threshold that examiners have used since 2006; construction is the loan type that fails fastest in a downturn. Very high real estate share means the bank's fate is the local property market's. WHEN LOW Diversified books absorb shocks. But a small bank that is diversified into things it does not understand is not safer than one concentrated in what it knows. VS PEERS Compare against band and, ideally, against the bank's own history: a construction share that doubled in two years is a strategy change. OUR LINE construction > 100% of equity = flag LIMITS Loan categories are regulatory buckets. A 'C&I' loan to a car dealer secured by inventory is auto exposure that the auto line will never show. The auto book LNAUTO · P3AUTO + P9AUTO + NAAUTO · NTAUTOR MEANS Direct auto lending: size, share of loans, delinquency, and net charge-offs. Bank auto is mostly prime indirect paper bought from dealers. WHEN HIGH Auto above 20% of loans with charge-offs above the band p90 is a concentrated book that is souring. The system auto charge-off rate runs near 0.9%; a bank at 2% is losing money on the product. WHEN LOW Small or zero auto book is the norm at community banks. The indirect exposure, if any, is usually on the nonbank-lender line above. VS PEERS Auto charge-offs are only comparable across banks with similar credit tiers; the filing does not show tier. OUR LINE auto > 20% of loans and auto charge-offs > band p90 = flag LIMITS No repossessed-vehicle inventory line exists for banks the way it does for credit unions, so the leading indicator used on the credit union desk does not port here. PUTTING IT TOGETHER The composite score one point per flag, capital below 5% counts double · 25 flags across the dimensions above MEANS A count of how many things are wrong at once. Banks rarely fail from one number; they fail when liquidity, capital, earnings and credit go wrong together. WHEN HIGH Banks scoring 7 or more stopped filing within four quarters at 16.0%, against a 4.6% base rate, a 3.5x lift. The 42 modern failures scored a median 8 four quarters before the end and 10 at the final filing, against a survivor median of 1. WHEN LOW Zero to two is ordinary banking life. Most banks live here. VS PEERS The score is not band-adjusted itself, but most flags feeding it are, so it compares reasonably across sizes. OUR LINE 5-6 = elevated · 7+ = danger LIMITS Bank departures are overwhelmingly voluntary mergers, so the departure lift is flat compared with the failure signature. The 2023 run failures scored only 1-3 four quarters out: a low score with high uninsured deposits and large securities losses is not a clean bill. Read the run-risk flags separately. FOUR BANKS YOU WILL MEET THE LIQUIDITY-RICH BUYER Loans under 70% of deposits, a quarter or more of assets in cash and securities, leverage above 9%, ROA below its band. It cannot find enough loans locally and its excess sits in bonds earning less than it should. It buys participations and pools because that is the only way to deploy. Natural buyer screen. THE FUNDING-STRETCHED SELLER Loans above deposits, wholesale funding above 15%, deposits shrinking. Growth got ahead of the franchise. It sells loans, participations, or branches to get back inside its deposit base, and it does so on a timetable set by its funding, not its preference. Natural seller screen. THE SLOW BLEED Noncurrent loans rising quarter after quarter, provision chasing charge-offs, ROA turning negative, leverage eroding. This is almost every failure outside 2023. It is visible eight quarters out to anyone reading the filings in sequence, and the composite score sees it. THE RUN CANDIDATE Uninsured deposits above half, a bond book underwater by a large share of equity, liquid assets thin. Credit is fine, earnings are fine, capital is fine on paper, and it can be gone in seventy-two hours. This is 2023. The score alone misses it; the run-risk and securities-loss flags exist for it. WHAT THIS DESK CANNOT TELL YOU Whether uninsured deposits are sticky operating accounts or hot money. What the collateral behind a nonbank-lender line actually looks like. Whether management knows what it is doing. Anything about credit tier inside the auto book. Anything that happened after the last filing date. A filing is a quarterly photograph taken by the subject. Read it as evidence, argue with it, and never mistake a flag for a finding.