Banking Financial Analysis and Modelling

Introduction

Banks differ fundamentally from non-financial firms. Deposits and wholesale borrowings are both funding sources and financial products, while loans and securities generate income but also expose the institution to credit, liquidity and market risks. High leverage means that relatively small changes in asset quality, funding costs or interest rates can materially affect capital.

Bank analysis therefore requires an integrated view of:

• balance-sheet structure;
• net interest income and margins;
• liquidity and asset–liability management (ALM);
• credit quality, non-performing assets and provisions;
• regulatory capital and risk-weighted assets (RWA);
• profitability and efficiency; and
• stress transmission across earnings, capital and liquidity.

The precise definitions and minimum requirements used in practice depend on the applicable accounting framework, Basel implementation and national regulation.

Learning Objectives

After completing this chapter, students should be able to:

1. interpret a bank’s balance sheet and income statement;
2. calculate and explain net interest margin, spreads and profitability ratios;
3. evaluate liquidity using contractual gaps, LCR and NSFR;
4. measure asset quality, non-performing loans and provision coverage;
5. model expected credit losses using PD, LGD and EAD;
6. analyse CET1, total capital, leverage and RWA;
7. model earnings and economic-value exposure to interest-rate changes; and
8. conduct an integrated regulatory stress scenario linking losses, RWA, capital and liquidity.

1. The Bank Balance Sheet

The accounting identity is:

All quantities in this identity must be measured at the same reporting date and in the same currency unit.

1.1 Typical structure

Loans commonly represent the largest earning-asset category. Deposits are often the principal funding source, although their stability varies by customer type, concentration, insurance status and digital withdrawal capability.

1.2 Gross and net loan presentation

The allowance is a contra-asset representing estimated credit losses already recognised. A provision expense is a period flow through profit or loss; the allowance is the accumulated balance-sheet stock. They are related but not interchangeable.

1.3 Average balances

Income and expense accumulate over a period, while balance-sheet values are measured at points in time. Ratios combining them should normally use average balances:

This two-point average is suitable when balances change approximately evenly. Monthly or daily averages are preferable for volatile or rapidly growing portfolios.

2. Interest Income, Margin and Spread

2.1 Net interest income

Net interest income (NII) is measured in currency per reporting period. It is influenced by asset and liability volumes, contractual rates, repricing dates, defaults, deposit behaviour and hedging.

2.2 Net interest margin

Interest-earning assets include loans, interest-bearing placements and eligible securities. NIM is usually annualised and expressed as a percentage. Analysts must use a consistent definition when comparing banks.

2.3 Interest spread

A commonly used balance-sheet spread is:

where:

Spread is not generally equal to NIM because earning assets and interest-bearing liabilities may have different volumes. Non-interest-bearing deposits can cause NIM to exceed the simple spread.

Worked Example 1: Margin, spread and profitability

A bank reports average earning assets of $900 million, average interest-bearing liabilities of $850 million and average total assets of $1,000 million. The asset yield is 6.2%, and the liability cost is 3.4%.

Interest income is:

Interest expense is:

Therefore:

The difference arises because the denominators used for asset yield and funding cost are not equal.

2.4 Deposit beta and funding sensitivity

Deposit beta measures how strongly deposit rates respond to a reference-rate change:

For example, if a policy rate rises by 2 percentage points and the average deposit rate rises by 0.8 percentage points, the beta is 40%. Beta is an empirical behavioural measure, not a fixed contractual parameter.

3. Profitability and Efficiency

3.1 Return on assets and equity

Both ratios normally use annual net income or annualised interim income. ROE is amplified by leverage and should not be interpreted independently of risk and capital adequacy.

3.2 A denominator-consistent ROA decomposition

Assuming net income equals NII plus non-interest income less non-interest expense, credit-loss provision and tax, ROA can be decomposed as:

Here ATA is average total assets, and every component uses the same denominator. Other material income or expense items must be added when applicable.

Conventional NIM cannot simply replace NII/ATA because NIM uses average interest-earning assets. The valid bridge is:

In Worked Example 1, suppose non-interest income is $12 million, non-interest expense $25 million, provision expense $7 million and tax $1.5 million. Net income is:

Thus:

The denominator-consistent decomposition is:

The first component also equals 2.99% × (900/1,000) ≈ 2.69%, subject to rounding.

3.3 Cost-to-income ratio

Operating income commonly includes NII and non-interest operating income. Definitions may exclude exceptional items. A lower ratio normally indicates greater operating efficiency, but aggressive cost reduction can weaken controls, cybersecurity or customer service.

4. Asset Quality and Non-Performing Assets

A loan is commonly classified as non-performing when payment is materially overdue—often more than 90 days—or when full repayment is otherwise unlikely. Exact regulatory definitions vary.

4.1 Core asset-quality ratios

The definition of net NPLs may differ across jurisdictions, so disclosure methodology must be checked.

A 100% coverage ratio does not necessarily eliminate loss risk: collateral values, cure rates, collection costs and timing remain uncertain. Conversely, secured portfolios may justifiably have coverage below 100% if collateral is enforceable and conservatively valued.

4.2 Migration and vintage analysis

Portfolio averages can conceal deterioration. Analysts should examine:

• movement between performing, underperforming and default categories;
• delinquency buckets such as 30, 60 and 90 days past due;
• default rates by origination vintage;
• sector, region and borrower concentration;
• restructurings and forbearance;
• write-offs, recoveries and cures; and
• collateral type and loan-to-value ratios.

Rapid loan growth can temporarily reduce the reported NPL ratio because new loans enlarge the denominator before defaults emerge.

5. Credit-Loss Modelling

5.1 PD, LGD and EAD

A basic one-period expected-loss model is:

Here:

• PD is the probability of default over the specified horizon;
• LGD is the proportion of exposure lost if default occurs, net of recoveries and costs;
• EAD is exposure at default in currency units.

PD and LGD are dimensionless proportions, so expected loss has the same currency unit as EAD. The model assumes that parameters are defined consistently for the same scenario and horizon.

5.2 Multi-period discounted expected loss

A general discrete formulation is:

MPD_t is the unconditional marginal probability of default during period t, DF_t is the discount factor to the reporting date, and T is the model horizon.

If one constant effective annual discount rate r applies:

If r_t is the annual-compounded spot rate specifically applicable to maturity t, then:

Thus, r denotes a constant rate, whereas r_t denotes a maturity-specific spot rate. Accounting standards may instead require discounting using the instrument’s original effective interest rate.

Worked Example 2: Two-year ECL

A $100 million portfolio has an LGD of 40%. Unconditional marginal PDs are 2% in year 1 and 3% in year 2. EAD remains $100 million. Annual-compounded spot rates are 5% for year 1 and 5.5% for year 2.

The result depends on the assumed exposure path, marginal PDs, LGD and discount rates. Correlation and scenario effects are embedded only if the parameters have been conditioned accordingly.

5.3 Accounting and regulatory perspectives

Under IFRS 9, performing assets generally begin with 12-month ECL, while assets experiencing a significant increase in credit risk generally move to lifetime ECL. Credit-impaired assets are subject to lifetime ECL. US CECL generally recognises lifetime expected credit losses for assets within its scope from origination. Regulatory capital rules may treat accounting allowances differently from accounting equity.

Robust models use probability-weighted macroeconomic scenarios. A simplified scenario combination is:

where scenario weights w_s ≥ 0 sum to 1. Scenarios should be internally consistent; for example, unemployment, property prices and default rates should not be combined arbitrarily.

6. Liquidity and Funding Analysis

Liquidity is the capacity to meet obligations when due without unacceptable loss. Solvency and liquidity are related but distinct: a solvent bank can fail if it cannot obtain cash quickly enough.

6.1 Liquidity gap

For maturity bucket t:

Contractual gaps should be supplemented by behavioural assumptions concerning deposit withdrawals, loan prepayments, credit-line drawdowns, collateral calls and market access.

6.2 Liquidity Coverage Ratio

Under the Basel framework:

High-quality liquid assets (HQLA) are measured after applicable eligibility restrictions and haircuts. The 75% cap prevents a bank from relying excessively on expected inflows. The Basel minimum is generally 100%, subject to national implementation and permitted use during stress.

6.3 Net Stable Funding Ratio

Available stable funding (ASF) and required stable funding (RSF) apply regulatory weights based on funding stability and asset liquidity over a one-year horizon. The Basel minimum is generally 100%, subject to national rules.

6.4 Funding concentration

A satisfactory aggregate LCR can coexist with vulnerabilities such as:

• reliance on a few uninsured depositors;
• excessive short-term wholesale funding;
• foreign-currency mismatches;
• encumbered securities;
• dependence on central-bank facilities; or
• operational inability to monetise collateral promptly.

Contingency funding plans should specify early-warning indicators, governance responsibilities, collateral availability and tested funding actions.

7. Regulatory Capital and RWA

7.1 Capital categories

Common Equity Tier 1 (CET1) includes high-quality, loss-absorbing common equity after regulatory adjustments. Additional Tier 1 and Tier 2 contain other eligible instruments subject to specified conditions.

RWA includes credit-risk, market-risk and operational-risk requirements calculated under applicable standardised or approved model-based approaches. Capital ratios should be compared with regulatory minima plus applicable conservation, countercyclical and systemic buffers.

7.2 Leverage ratio

The exposure measure includes specified on-balance-sheet and off-balance-sheet exposures. Unlike risk-based ratios, the leverage ratio is intended as a relatively risk-insensitive backstop.

7.3 Why RWA can rise during stress

Even without new lending, RWA can increase because borrowers migrate to weaker risk grades, collateral values fall, default exposures receive higher weights, or market and operational risk requirements rise. A bank can therefore experience both a falling capital numerator and a rising RWA denominator.

8. Interest-Rate Risk and Asset–Liability Management

Interest-rate risk in the banking book (IRRBB) affects both short-term earnings and the economic value of equity.

8.1 Repricing-gap model

For a selected horizon:

Under the restrictive assumptions of an immediate parallel rate change, unchanged balances and equal repricing sensitivity:

For example, a one-year positive gap of $200 million and a 1 percentage-point increase imply an approximate $2 million rise in annual NII. The approximation may fail when deposit betas, optionality, yield-curve movements or basis spreads differ.

8.2 Economic value of equity

Using a constant annual discount rate r, the present value of cash flows is:

Alternatively, maturity-specific spot rates r_t may be used as CF_t/(1+r_t)^t. Cash-flow assumptions must include prepayments, early withdrawals and embedded options.

For a small parallel yield change, a duration approximation is:

Here D_A and D_L are modified durations in years, A and L are present values in currency, and Δy is the yield change in decimal form. The approximation assumes small, parallel shifts and stable cash flows; convexity and option effects matter for larger shocks.

8.3 Funds transfer pricing

Funds transfer pricing (FTP) assigns an internal funding or investment rate to business units. It separates customer pricing performance from centrally managed interest-rate and liquidity risks. Effective FTP incorporates maturity, liquidity premiums, optionality and currency, reducing incentives for business units to maximise volume while ignoring funding risk.

9. Integrated Regulatory Stress Testing

Stress testing projects the effect of severe but plausible conditions on earnings, provisions, capital, RWA and liquidity. It complements, rather than replaces, historical ratios.

Illustrative Case: Meridian Bank

This is a fictional application designed to demonstrate integrated modelling.

#### Background

Meridian Bank has the following baseline position, in millions of dollars:

Baseline capital adequacy is:

Because inflows of $60 million are below 75% of $180 million, net outflows are $120 million. Therefore:

#### Problem and stress assumptions

Management applies a recession scenario involving:

• $24 million of incremental credit-loss expense;
• a 25% tax benefit that is recognised and considered recoverable;
• credit migration increasing RWA from $650 million to $700 million;
• stressed HQLA of $108 million after valuation changes and regulatory haircuts;
• 30-day outflows rising to $210 million; and
• 30-day inflows rising to $70 million.

The exercise assumes no dividend, capital issuance or other profit during the initial shock.

#### Analysis: loss and capital transmission

The after-tax reduction in retained earnings is:

Assuming the loss is fully reflected in CET1:

The stressed CET1 ratio is:

The ratio falls for two reasons: the loss reduces CET1, while deteriorating risk grades increase RWA. Whether 7.43% breaches requirements depends on Meridian’s jurisdiction-specific minimum and buffers.

#### Analysis: liquidity transmission

The regulatory inflow cap is:

Actual inflows of $70 million are below the cap, so net outflows are:

Thus:

The bank experiences simultaneous capital and liquidity pressure. Credit concerns can increase withdrawals, while asset sales intended to generate cash may crystallise valuation losses.

#### Approach and management response

Meridian models three actions:

1. issue $15 million of common equity, with the cash retained as eligible HQLA;
2. replace $15 million of funding due within 30 days with funding maturing beyond the stress horizon, reducing outflows to $195 million; and
3. convert $5 million of unencumbered non-HQLA securities into eligible HQLA without material loss.

Post-action CET1 becomes $67 million, while RWA remains $700 million:

HQLA becomes $108m + $15m + $5m = $128m. Net outflows become:

because the $70 million inflow remains below the 75% cap. Therefore:

#### Outcome and learning points

The actions restore LCR above 100% and improve the CET1 ratio, although management must still compare capital with all applicable buffers and assess whether the actions are feasible under market stress. The case demonstrates that:

• credit losses affect earnings and CET1;
• borrower migration can increase RWA independently of losses;
• deposit and wholesale-funding stress affects LCR;
• equity issuance can support both capital and liquidity when proceeds remain liquid; and
• stress testing must avoid evaluating capital and liquidity in isolation.

10. Model Construction and Governance

A practical banking model commonly contains:

1. historical financial statements and ratio diagnostics;
2. loan, deposit and securities schedules;
3. interest-rate, volume and deposit-beta assumptions;
4. credit migration, PD, LGD, EAD and provision modules;
5. RWA and regulatory capital schedules;
6. contractual and behavioural liquidity cash flows;
7. baseline, adverse and severe scenarios; and
8. validation checks and management dashboards.

10.1 Essential controls

Every model should enforce:

• balance-sheet equality;
• reconciliation of opening and closing allowances;
• consistency between loan growth, funding and cash;
• traceability from assumptions to outputs;
• version control and independent review;
• unit, sign and time-period consistency;
• sensitivity and back-testing; and
• transparent treatment of overlays and expert judgement.

10.2 Technology and AI

Machine learning can improve transaction-based early-warning systems, default prediction and deposit-flow forecasting. However, complex models can produce unstable relationships, proxy discrimination and weak explanations. Governance should include representative data, fairness testing, human oversight, explainability appropriate to model risk, cybersecurity and privacy controls.

Generative AI may assist documentation or scenario development, but unverified output must not be inserted into regulatory models. Material assumptions, calculations and disclosures remain the responsibility of accountable professionals.

11. Limitations, Ethics and Contemporary Developments

Bank models depend on historical data that may not represent structural breaks. Default correlation, customer behaviour and market liquidity often worsen simultaneously during crises. Stress assumptions can also become outdated as digital banking accelerates deposit withdrawals.

Ethical and governance concerns include concealing problem loans through repeated restructuring, delaying loss recognition, manipulating model assumptions, discriminatory credit decisions and overstating collateral recoverability. Boards should challenge optimistic forecasts and ensure that remuneration does not reward growth without regard to risk-adjusted performance.

Contemporary analysis increasingly addresses climate-related credit risk, cyber incidents, instant-payment liquidity, commercial real-estate concentration and the interaction between social media and deposit runs. These risks should enter models through defensible channels—such as cash flows, PD, LGD, collateral values or operational losses—rather than unsupported adjustments.

Chapter Summary

Bank financial analysis integrates profitability, credit quality, liquidity, interest-rate risk and capital adequacy. NIM uses average interest-earning assets, whereas ROA uses average total assets; valid decompositions must preserve denominator consistency. Asset quality is evaluated through NPLs, coverage, migration and cost of risk, while credit-loss models combine PD, LGD, EAD and appropriate discount factors.

LCR evaluates short-term liquidity, NSFR assesses structural funding, and ALM models repricing and economic-value exposure. Regulatory capital analysis must consider both changes in eligible capital and changes in RWA. Integrated stress testing shows how credit losses, risk migration and funding withdrawals can jointly weaken CET1 and liquidity. Reliable models require transparent assumptions, reconciliations, scenario analysis, independent validation and strong ethical governance.

Key Terms

Asset–liability management: Coordinated management of balance-sheet interest-rate and liquidity risks.

CET1: Highest-quality regulatory capital, principally common equity after regulatory adjustments.

Cost of risk: Annual credit-loss provision expense divided by average gross loans.

Deposit beta: Sensitivity of deposit rates to changes in a reference interest rate.

EAD: Estimated exposure outstanding when a borrower defaults.

Economic value of equity: Present value of asset cash flows less the present value of liability cash flows.

Expected credit loss: Probability-weighted estimate of discounted credit losses.

Funds transfer pricing: Internal pricing system assigning funding and liquidity costs to business units.

HQLA: Eligible high-quality liquid assets available to meet stressed cash outflows.

Interest-rate risk in the banking book: Risk that interest-rate changes affect banking-book earnings or economic value.

LCR: Ratio of HQLA to net cash outflows over 30 calendar days.

LGD: Proportion of exposure expected to be lost if default occurs.

Net interest margin: Net interest income divided by average interest-earning assets.

Non-performing loan: Loan materially overdue or otherwise unlikely to be repaid in full.

NSFR: Ratio of available stable funding to required stable funding.

PD: Probability that a borrower defaults over a specified horizon.

Provision coverage ratio: Allowances allocated to NPLs divided by gross NPLs.

Risk-weighted assets: Exposures adjusted under regulatory rules for credit, market and operational risk.

Stress testing: Forward-looking assessment under severe but plausible adverse conditions.

Review Questions

Short Answer Questions

1. Why should average balances normally be used when calculating bank profitability ratios?
2. Distinguish net interest margin from interest spread.
3. Explain the difference between provision expense and the credit-loss allowance.
4. Why can RWA increase even when a bank does not expand lending?
5. What is the purpose of the 75% inflow cap in the LCR?

Long Answer Questions

1. Explain the major components of a commercial bank balance sheet and their associated risks.
2. Discuss how NIM, non-interest income, operating costs and provisions influence ROA.
3. Compare contractual liquidity-gap analysis, LCR and NSFR.
4. Explain how PD, LGD, EAD, discounting and macroeconomic scenarios are used in ECL modelling.
5. Critically evaluate the strengths and limitations of repricing-gap and EVE models.

Analytical / Application Questions

1. A bank has NII of $45 million, average earning assets of $1.5 billion and average total assets of $1.8 billion. Calculate NIM and the NII contribution to ROA, and explain why they differ.
2. Construct a two-year ECL estimate for a $200 million portfolio using independently justified PD, LGD, EAD and discount-rate assumptions. Explain each assumption.
3. Design a stress scenario that simultaneously affects credit losses, RWA, CET1 and LCR. Recommend feasible management actions and identify possible second-round effects.

Multiple Choice Questions with Answers

Question 1

A bank reports annual net interest income of $30 million and average interest-earning assets of $1,000 million. What is its NIM?

A. 2.0%

B. 3.0%

C. 3.3%

D. 30.0%

Correct Answer: B. 3.0%

Question 2

Which expression provides a denominator-consistent component in an ROA decomposition?

A. NIM plus net income divided by average equity

B. NII divided by average interest-earning assets

C. NII divided by average total assets

D. Interest income divided by average gross loans

Correct Answer: C. NII divided by average total assets

Question 3

A bank has gross NPLs of $40 million and gross loans of $800 million. What is the gross NPL ratio?

A. 4%

B. 5%

C. 8%

D. 20%

Correct Answer: B. 5%

Question 4

Allowances allocated to NPLs are $36 million, while gross NPLs are $45 million. What is the NPL coverage ratio?

A. 20%

B. 45%

C. 80%

D. 125%

Correct Answer: C. 80%

Question 5

For a one-period exposure of $50 million, PD = 4% and LGD = 45%. What is expected loss?

A. $0.09 million

B. $0.90 million

C. $2.00 million

D. $9.00 million

Correct Answer: B. $0.90 million

Question 6

A bank has CET1 capital of $60 million and total RWA of $750 million. What is its CET1 ratio?

A. 6%

B. 8%

C. 10%

D. 12.5%

Correct Answer: B. 8%

Question 7

A bank has HQLA of $105 million, 30-day outflows of $160 million and inflows of $50 million. The inflows are below the regulatory cap. What is its LCR?

A. 65.6%

B. 87.5%

C. 95.5%

D. 105.0%

Correct Answer: C. 95.5%

Question 8

In the Meridian Bank stress case, what was the stressed CET1 ratio before management actions?

A. 7.43%

B. 8.00%

C. 9.57%

D. 10.77%

Correct Answer: A. 7.43%

Question 9

Under a simple repricing-gap model, a bank has a positive gap of $200 million and rates rise by 1 percentage point. What is the approximate annual change in NII?

A. A decrease of $20 million

B. A decrease of $2 million

C. An increase of $2 million

D. An increase of $200 million

Correct Answer: C. An increase of $2 million

Question 10

Which ratio is designed principally to assess the adequacy of stable funding over approximately a one-year horizon?

A. Cost-to-income ratio

B. CET1 ratio

C. Liquidity Coverage Ratio

D. Net Stable Funding Ratio

Correct Answer: D. Net Stable Funding Ratio