Revised practical article; source details and figures are dated to their cited filing or event, with current evidence reviewed through 24 September 2026.
A bank’s debt is not like a manufacturer’s debt. Customer deposits are part of the product that lets the bank lend and earn a spread. An insurer’s premium is not an immediate profit cheque: the insurer has agreed to cover future claims. That is why an industrial-company owner-earnings formula cannot be dropped onto a bank or insurer without changing the analysis.
For an industrial company, investors may estimate operating cash flow after tax, reinvestment and working capital, then allocate claims between lenders and shareholders. For a bank, loans, deposits, interest expense and credit losses are central operating items. There is no clean boundary between “debt financing” and “operating input” in the usual enterprise-value framework. For an insurer, premiums, claims, reserves, reinsurance, investment return, and service obligations shape the value of the business.
Banks: profit is not the same as distributable capital
Start with return on equity, but ask what generated it. Net interest income depends on loan yields and funding costs. Fee income depends on customer activity and fee rules. Credit-loss charges depend on loan quality and future recoveries. The balance sheet should be followed from opening to closing: new lending, repayments, defaults, write-offs, collateral, provisions, securities, deposits, and wholesale funding.
Then examine capital. The bank’s ability to lend and distribute capital depends on the rules for its regulated entity, the quality and risk weights of assets, potential losses, and buffers. The Central Bank of Kenya’s Bank Supervision Annual Report for 2024 reports sector-level capital and liquidity statistics, including a 19.6% total capital adequacy ratio at December 2024 against a minimum ratio of 14.5% for that reported system measure [1]. Those are aggregate, dated figures, not the capital position of every bank or a universal payout limit. For an individual issuer, use the latest company/regulator data and confirm definitions, entity scope, and any newer rule.
A bank dividend is sustainable only if earnings and capital generation can support it after expected losses, growth and regulatory constraints. The analyst should distinguish accounting equity, regulatory capital, retained earnings, capital buffers, and freely distributable cash. A high dividend yield can reflect a distribution that cannot repeat or a risk the market has not priced correctly.
Peer-reviewed accounting research finds that fair-value disclosures and financial-instrument information can matter for bank share valuation [2]. This supports reading the asset and liability notes closely. It does not eliminate credit-risk judgment. For a bank, one must ask which loans may not be repaid and whether allowances and capital adequately reflect that exposure.
Insurers: premiums create future claims
An insurer receives premiums and accepts obligations whose amount and timing are uncertain. The cash can be invested, but the owner does not own the full premium balance as free cash. Claims, reinsurance, expenses, policyholder rights, and required solvency capital remain.
IFRS 17 requires insurance contract profit to be recognized as insurance services are provided and as the entity is released from risk; it also separates insurance service results from insurance finance income or expense [3]. The standard’s contractual service margin represents unearned profit to be recognized over the coverage period. It is not a cash reserve available for distribution today. A reported earnings line needs to be read beside claims development, reserves, reinsurance, investment balances, and solvency information.
An analyst can assess the insurer’s operating result by product line and underwriting period, then evaluate the investment portfolio and capital backing those obligations. Mortality, medical utilization, catastrophe losses, claims inflation, lapse rates, and investment returns can have different implications across the portfolio. Group-level earnings can also conceal differences among legal insurers and markets.
Research on capital allocation in insurance discusses how regulatory risk-based capital and the economics of separate lines of business affect capital decisions [4]. Use that lens to ask where the company must retain capital and whether a reported return is high because the business is profitable or because it operates with greater risk.
Estimate value from the equity claim
Ohlson’s peer-reviewed valuation framework links value to book value, earnings and dividends under its stated assumptions [5]. For banks and insurers, an equity-oriented residual-income or dividend model is often more natural than subtracting “net debt” from enterprise value. But the model is only as good as its path for equity earnings, capital, loss reserves, payout and return on equity.
A transparent model might forecast:
- core income and operating costs;
- credit-loss or claims assumptions;
- ending book equity and regulatory capital proxy;
- retention required to support a chosen balance-sheet growth rate;
- maximum distribution consistent with the selected capital assumptions;
- return on equity and cost of equity under base and downside cases.
The regulatory-capital proxy must be labeled as a proxy unless the model uses the actual regulatory calculation. Do not assume one shilling of accounting equity equals one shilling of eligible regulatory capital. Test a credit/claims shock and a lower capital-generation case.
A bank analysis that begins with the loan book
A bank model should start with the legal bank and its consolidated group boundary. A listed parent may own banks in multiple countries, a payments business, insurance subsidiaries or asset managers. Dividends received by the parent may be constrained by subsidiary capital, local rules, minority interests, currency controls or upstream approvals. Keep each regulated entity and currency visible before combining them.
For the core bank, build a multi-year schedule with average earning assets, loan yield, securities yield, deposit cost, wholesale funding cost, net interest margin, fee income, operating expenses and credit-loss charges. Use average balances when estimating yields and costs; ending balances can distort a year in which lending or deposits grew quickly. Split business and consumer loans, secured and unsecured exposures, sectors and currencies where disclosures permit. Track restructured and re-aged loans separately from loans that are current under original terms.
A single non-performing-loan ratio is not enough. Follow the amount and direction of Stage 2 and Stage 3 balances, overdue days, collateral valuation, loan-loss provisions, write-offs, recoveries and new lending. Compare the allowance with gross exposures and test collateral realization cost and timing. Loan loss allowances depend on expected future conditions and model judgments; a benign current default rate can coexist with weak underwriting if loans are newly originated or restructured. The question is not just “how many loans are bad today?” but “what losses could emerge from the book as it seasons?”
The Central Bank of Kenya's 2025 supervision report, issued in September 2026, reports sector aggregates including a 20.7% total capital adequacy ratio, 59.3% liquidity ratio and 16% gross NPL ratio at the reported year-end [6]. The August 2026 MPC background data provides a later system snapshot [7]. These figures help frame the operating environment, but they are not an estimate for a named bank and should not be plugged into its model as if they were. Verify each issuer's latest ratios, definitions, legal entity, reporting date and capital rules.
Turn capital into an explicit distribution test
A bank's dividend capacity is a capital-allocation question. Start with reported eligible capital by tier, risk-weighted assets, leverage measure, deductions, buffers, minimum requirements and any supervisory or board limits. Then prepare a bridge from opening capital to closing capital: retained profit, dividends, share issues, credit losses, other comprehensive income, valuation movements, acquisitions, foreign exchange, regulatory changes and changes in deductions. A change in the capital ratio can come from either numerator or denominator; do not attribute it to earnings alone.
For a shareholder estimate, use a simple capital roll-forward beside the actual regulatory calculation. The simple version is a planning proxy, not a compliance model. Suppose, purely for illustration, a bank starts with KES 100 billion of eligible capital and KES 500 billion of risk-weighted assets, a 20% ratio. It earns KES 20 billion after tax, plans a KES 10 billion dividend, and risk-weighted assets grow by 12% to KES 560 billion. Ignoring all other changes, closing capital is KES 110 billion and the ratio falls to 19.6%. If management's chosen operating floor were 18%, the simplified headroom would be KES 9.2 billion: KES 110 billion less 18% of KES 560 billion. This is not the bank's statutory distributable amount; real capital deductions, buffers, legal restrictions, supervisory review, risk weights and other adjustments can materially change it.
The calculation shows why profit growth does not guarantee dividend growth. If assets grow quickly, capital may need to be retained. If credit quality worsens, both earnings and capital may fall. If asset risk weights rise or a new regulatory buffer applies, the amount available to distribute can shrink even with unchanged book profit. If a dividend is funded by issuing new shares or selling assets, report that financing source clearly.
Normalize bank earnings through the credit cycle
Use several years of net interest income, fees, expenses, impairment charges and recoveries. Mark extraordinary credit events, acquisitions, tax benefits, restructuring and changes in accounting or regulation. A mechanical average is not enough if years include very different interest-rate and credit environments. Build a central case from loan growth, pricing and funding costs that can be explained from the disclosures; build a downside where credit loss and funding costs change together.
A useful stress has four connected parts:
- Borrower stress: slower turnover, weaker collateral values, higher unemployment or sector-specific pressure increases arrears and expected loss.
- Funding stress: deposit competition raises rates or moves deposits to more expensive channels; wholesale funding rolls at a higher cost or shorter tenor.
- Capital stress: higher provisions reduce earnings and retained capital, while risk-weighted assets may grow or shift toward higher-risk exposures.
- Distribution stress: management reduces or suspends distributions to preserve the bank's ability to lend and meet requirements.
Do not assume the earnings impact ends when a loan is classified. Recovery costs, collateral realization time, legal expenses, restructuring, foregone interest and customer relationships also matter. A bank may recover some collateral, but the time and net proceeds should be evidenced. Track write-offs and recoveries separately from the provision expense so that a year with large recoveries does not look like permanently improved underwriting.
Barth's research on fair-value disclosure of investment securities and the market valuation of banks supports close attention to securities and valuation notes [2]. In a rising-rate or volatile-market period, distinguish accounting classification, cash-flow timing, realized sales and economic sensitivity. A security held to collect contractual cash flows is still exposed to credit and funding risks; a sale to meet withdrawals can crystallize a loss that was previously unrealized.
What a bank workbook should calculate
A bank companion schedule should contain: statement history; average balance sheet; net interest income bridge; loan quality and provisions; funding mix and cost; sovereign and sector concentration; capital by regulated entity; risk-weighted asset growth; dividends and share count; regulatory minimums and buffers; base and stress cases; and a source/date ledger. Link each ratio to its inputs. If the exact regulatory ratio is not available in public disclosures, show the missing source instead of approximating it from total assets.
The model should calculate return on average equity, but place it beside growth, credit cost, capital retention and cost of equity assumptions. A high return on equity can come from good economics, a low equity base or high leverage. Ask whether book value per share grows after distributions and whether the bank can maintain risk controls while expanding. Compare an earnings or residual-income value with dividend capacity; a dividend discount model should not simply project the last payout forward.
An insurer analysis that follows the promises
For an insurer, separate insurance service result, insurance finance income or expense, investment income, expenses, claims, reinsurance and capital. Under IFRS 17, the liability for remaining coverage relates to service not yet provided, while the liability for incurred claims relates to claims for events that have already occurred but may not yet be settled. Contractual service margin represents profit associated with future service under the standard; it is recognized over coverage rather than paid out immediately [3]. The analyst should understand the issuer's disclosure labels and avoid treating the margin or gross premium cash as distributable cash.
Build the insurance analysis by product and legal entity where data exist. For each major line, follow gross and net premiums, acquisition costs, claims incurred, claims paid, expenses, reinsurance recoveries, reserve development and retention. Compare prior estimates with eventual claims and identify whether favorable experience is a repeatable pricing or underwriting improvement or the release of an old reserve. A reserve strengthening can reduce earnings even when current cash collections are strong; a reserve release can increase earnings without a matching new cash inflow.
Use a claims development view. Group claims by accident or underwriting year and compare paid and incurred amounts over time. Examine large losses, medical cost inflation, catastrophe exposure, lapse and renewal rates, concentration in corporate schemes and the terms of reinsurance. Then test counterparty risk: a reinsurance receivable only helps if the contract responds and the reinsurer can pay. Gross exposure, net exposure, collateral and dispute status should be shown separately.
Capital, asset allocation and dividend capacity at an insurer
Insurance capital supports policyholder promises and bears the risk of adverse claims or asset values. Map eligible own funds, solvency requirements, capital quality, stress results and any restrictions on distributions from regulated subsidiaries. Cummins' work on capital allocation in insurance shows why capital needs vary across lines and risks rather than following one group-wide payout ratio [4]. The analyst should ask which entity holds capital and which entity declares the dividend.
Inspect the investment portfolio by duration, credit quality, currency, liquidity and accounting treatment. Compare asset duration with the expected timing of claims. A bond portfolio can produce recurring income, but selling assets to meet claims or dividends during market stress may crystallize losses. Real estate or private investments may be less liquid than the claims they support. Keep investment gains, recurring coupon income, realized disposals and fair-value movements distinct.
Then compare proposed dividends with after-tax earnings, solvency headroom, cash at the operating entity, debt maturity and the timing of claims. A group can report profit while a regulated subsidiary needs to retain capital. A special dividend may be supported by a one-time disposal or reserve release. Each distribution should be traced to the legal entity and source of cash.
Jubilee Holdings: read the dates and measures separately
Jubilee's FY2025 results announcement describes an 18% increase in net profit and a record dividend [8]. Its H1 2026 announcement reports profit before tax of KES 4.4 billion, up 30% [9]. These are useful, dated issuer claims about reported performance. They do not establish that the H1 result will recur for a full year or that the announced dividend is a sustainable owner-earnings rate.
The analysis should reconcile the annual report and half-year statement by segment, insurance service result, investment outcome, tax, non-controlling interests and cash distribution. Check the dividend amount and record date against the parent and regulated subsidiaries' available capital and cash. Compare the H1 profit before tax measure with after-tax profit attributable to ordinary shareholders; the two should never be substituted without a bridge. If the company reports growth due to investment performance or one-off items, isolate that contribution before normalizing the underwriting result.
A peer review and governance routine
Use a second reviewer to challenge the loan-loss or claims assumptions, not just check formulas. Ask which credit segment or policy line drives the downside, whether the collateral or reinsurance assumption has independent support, and whether the capital calculation uses the same entity as the dividend. A reviewer should be able to trace a result from source page to input, formula, scenario and recommendation.
At each reporting date, update the model only after the filing and official regulatory data are available. Preserve the prior version and write a change log: new fact, affected assumption, result before, result after, and reason. Set triggers such as a material increase in arrears, loss ratio, reserve strengthening, a fall in capital buffer, a change in dividend policy or a large shift in funding cost. The triggers guide investigation; they are not a universal buy or sell rule.
For a bank or insurer, the practical owner-earnings answer is not “cash from operations minus capital expenditure.” It is the sustainable equity earnings that remain after expected credit or claims costs, capital retention, regulatory constraints, liquidity needs and dilution. If a key solvency or credit measure is unavailable, state what cannot be concluded and identify the exact disclosure needed. That is an actionable analysis, not a refusal to analyze the business.
A dividend decision has two separate tests
A dividend announcement answers what the board proposes or declares under company law and the issuer's policy. An owner-earnings analysis asks whether the distribution is repeatable after the business meets its expected obligations and preserves the capital base. Put both in the same table but do not treat one as proof of the other.
For a bank, compare: proposed dividend; profit attributable to ordinary shareholders; retained profit after distribution; change in risk-weighted assets; capital ratio before and after the payout; headroom to regulatory and internal buffers; and the downside effect of higher credit costs. The ratio of dividend to profit is a useful descriptive figure, not a regulatory capacity test. For an insurer, compare the dividend with parent-company cash, each regulated subsidiary's solvency headroom, claims and reinsurance exposures, reserve development, and debt service. A group can pay from accumulated cash while a subsidiary retains most current earnings.
Use a dated case sheet. Record the declaration date, approval date, record date, payment date, amount per share, total cash cost, source of cash and whether the figures are actual or proposed. If a full-year dividend uses half-year results, keep the periods separate. For an investor computing a yield, match the share price to the date and distinguish declared from proposed amounts.
What changes the bank's estimate most?
A first-pass sensitivity should show the effect of several credit-loss cases and funding-cost cases on profit, capital generation and distributable capacity. For example, compare current credit cost, a moderate rise to a level observed in a weak year, and a severe sector-specific stress. Build the scenario from the institution's own loan mix and history where disclosed. Do not apply one system-wide NPL number to every bank. Then vary deposit repricing and loan yield separately: a bank may not pass a policy-rate move through to every borrower and depositor at the same speed.
At minimum, the analyst should track four patterns across filings: credit growth faster than deposits; rising restructured or overdue loans; provisions falling while risk exposure expands; and dividends growing faster than both earnings and eligible capital. None proves a problem in isolation. Together they identify a question for the next report and a scenario to quantify. Use the CBK's system report as context, and use institution-specific statutory and audited data for the actual model [6, 7].
What changes the insurer's estimate most?
Claims inflation, product pricing, renewal retention, catastrophic loss, reserve development, reinsurance recoverability, investment returns and solvency capital can move in different directions. A scenario table should not merely reduce investment income while leaving claims and lapses unchanged. Ask what macroeconomic, health, weather, legal or competitive mechanism drives the shock, which contracts are affected and how quickly new pricing can respond.
The H1 result or annual profit can be a starting observation, not a run-rate. Jubilee reported FY2025 net profit growth and a record dividend, then reported H1 2026 profit before tax growth [8, 9]. To decide whether that is repeatable, bridge segment earnings, tax, investment results, claims experience, non-controlling interests and available subsidiary capital. A sharp investment gain may be volatile; an underwriting improvement may persist if pricing and retention support it; a release of prior-year reserves does not necessarily repeat.
A practical source and review checklist
Before sharing a bank or insurance valuation, make sure the reviewer can answer:
- What is the legal entity whose capital supports the dividend?
- Which earnings measure is attributable to ordinary owners, and which is before tax or includes minority interests?
- What balance-sheet risks were newly originated, restructured, revalued or transferred?
- Does the stress change both earnings and capital, or only the valuation multiple?
- How are liquidity, maturities, asset duration and claims timing reflected?
- Which official report or regulatory return would change the model, and when is it due?
This discipline is useful even when a public report omits a crucial detail. Label the missing item, describe its possible direction and size, and avoid replacing it with a convenient average. The actionable output may be “no reliable payout capacity can be estimated until the regulated bank's core capital and asset-quality schedules are available.” It can also be a range that is wide but tied to specific credit-loss assumptions. The investor now knows exactly what evidence to request or monitor.
Workbook map
The Project 2 workbook keeps CBK system capital, liquidity and NPL ratios distinct from Jubilee group profit and premium/contribution figures. Evidence Vectors distinguishes the CBK FY2025 observation window from its September 2026 report publication and Jubilee's 31 December 2025 year-end from its 9 April 2026 results release. Its simplified bank capital illustration shows how earnings, dividends, risk-weighted-asset growth and an internal floor affect a proxy ratio and headroom. The Profile Stress sheet adds separate bank credit-loss/retention and insurance claims/reserve/solvency shocks on normalized indices; Appendix 04 explains why those outputs cannot be compared with operating-company residuals. The workbook does not calculate Jubilee statutory solvency or claims sufficiency because entity-level inputs are unavailable; use issuer and regulator records for that separate test.
Technical appendix
Technical appendices
The worked calculations and source boundaries below extend this article's examples. They keep assumptions and unresolved inputs visible so readers can reproduce the steps without mistaking an illustration for an issuer or sector estimate.
References
[1] Central Bank of Kenya, Bank Supervision Annual Report 2024, https://www.centralbank.go.ke/uploads/banking_sector_annual_reports/1328318827_2024%20Annual%20Report.pdf.
[2] M. E. Barth, “Fair Value Accounting: Evidence from Investment Securities and the Market Valuation of Banks,” The Accounting Review, vol. 69, no. 1, pp. 1–25, 1994.
[3] International Accounting Standards Board, “IFRS 17 Insurance Contracts,” https://www.ifrs.org/issued-standards/list-of-standards/ifrs-17-insurance-contracts/.
[4] J. D. Cummins, “Allocation of Capital in the Insurance Industry,” Risk Management and Insurance Review, vol. 3, no. 1, pp. 7–27, 2000, https://doi.org/10.1111/j.1540-6296.2000.tb00013.x.
[5] J. A. Ohlson, “Earnings, Book Values, and Dividends in Equity Valuation,” Contemporary Accounting Research, vol. 11, no. 2, pp. 661–687, 1995, https://doi.org/10.1111/j.1911-3846.1995.tb00461.x.
[6] Central Bank of Kenya, Bank Supervision Annual Report 2025, issued 23 September 2026, https://www.centralbank.go.ke/uploads/banking_sector_annual_reports/1241268828_ANNUAL%20REPORT%202025.pdf.
[7] Central Bank of Kenya, Monetary Policy Committee Meeting Background Data, August 2026, https://www.centralbank.go.ke/uploads/presentations/2005474214_Monetary%20Policy%20Committee%20Meeting%20Background%20Data%20August%202026.pdf.
[8] Jubilee Holdings Limited, “Jubilee Holdings Declares a Record Dividend Following 18% Increase in Net Profit,” 9 April 2026, https://jubileeinsurance.com/group/newsroom/jubilee-holdings-declares-a-record-dividend-following-18-increase-in-net-profit/.
[9] Jubilee Holdings Limited, “Jubilee Holdings Limited Delivers 30% Growth in H1 Profit Before Tax to KES 4.4 Billion,” 28 August 2026, https://jubileeinsurance.com/group/newsroom/jubilee-holdings-limited-delivers-30-growth-in-h1-profit-before-tax-to-kes-4-4-billion/.
Companion resources
The workbook and technical appendices are available with research membership. The video production guide maps this article to a planned episode; a published video is not yet available.
No public comments yet. The first thoughtful question can start the discussion.