Diversification by Risk Source: Assessing Business Portfolios, Channels and Products

Diversification begins where a loss can originate

Diversifying by risk source means identifying the causes of default, loss, delay or unavailable cash, then examining which exposures respond together to those causes. Credit risk, liquidity pressure, fraud, customer concentration and operational failure can coexist within one transaction. Grouping exposures by product helps organize the analysis, but does not establish independence. Receivables financing and working capital loans, for example, may ultimately depend on the same sales generated by one business group.

This educational article provides a method for assessing these dependencies when appropriately authorized institutions examine business origination, correspondent arrangements and financing products. It does not prescribe personal asset allocation. Diversification can reduce particular concentrations while adding costs, contracts and points of failure. Its usefulness must be weighed against the institution's ability to analyze, monitor and recover each exposure, including when several favorable assumptions fail together. Variety has economic value only when the resulting risks remain understandable and manageable.

Move from business records to a dependency map

A useful starting point is a map connecting each transaction to whoever generates the money needed for repayment. Alongside the formal borrower, include important buyers, controlling owners, related businesses, sectors, regions, critical inputs and guarantors. A carrier may depend on one shipper; a manufacturer on electricity and a distributor; a service business on renewing a few contracts. Record the evidence supporting each connection, distinguishing a suspected relationship from a documented economic dependency.

Consider two businesses with different Brazilian CNPJ registration numbers and different managers that earn almost all their revenue from the same retail chain. Their registration records suggest two separate risks, but a payment suspension by that chain could affect both. Different collateral assets can also depend on the same resale market. Machinery used in a shrinking industry may lose value precisely when many borrowers attempt to sell it, weakening expected recoveries across the portfolio.

The unit of analysis should follow the mechanism of loss. Customer concentration may require grouping by ultimate buyer; fraud by originator or related network; liquidity by funding maturities; business continuity by technology supplier. No single grouping replaces the others. A portfolio needs overlapping, traceable views showing information sources, the reporting point covered by the documents, and unresolved gaps. An unidentified relationship should remain an uncertainty, rather than being silently classified as evidence that exposures are independent.

Different products create different rights

A working capital loan depends on the business's overall repayment capacity. Financing backed by receivables requires examination of existing rights, eligibility, supporting documents, potential disputes and collection arrangements. Expected sales are not established receivables. If payment depends on a future delivery that remains uncertain, the analysis must show that performance risk. A commercial label associated with advancing receivables cannot turn hoped-for revenue into a documented claim or remove the conditions that must be met before payment becomes due.

Equity, loans and convertible instruments are not interchangeable. A loan creates repayment obligations under its contract. Equity exposes the investor to business performance and the applicable corporate rights and rules. A convertible instrument depends on conditions that may change the investor's future position. A donation to a nonprofit is neither equity ownership nor a promise of financial return. In Brazil, a business cannot admit equity partners while retaining MEI status, so its legal form matters before an equity proposal is considered.

Assess economic exposure separately from the legal right available when something goes wrong. Two instruments can fund the same project while offering different collection, information and exit rights. Public securities distribution and regulated financial activities require appropriate structures and providers. A technology interface or a registered CNAE economic activity code does not grant permission to conduct them. Product selection therefore starts with the funding purpose and a permissible structure; apparent variety cannot substitute for understanding enforceable rights and responsibilities.

More origination channels can reproduce the same risk

Different channels may broaden access to businesses and information while reaching essentially the same customer population. A trade association, a business services firm and a specialist portal might all refer suppliers of one buyer. Measuring only the volume generated by each channel conceals that overlap. The institution should connect proposals and completed transactions by customer, business group and material dependencies, while observing applicable rules on data protection and use. Channel diversity and economic diversity need separate measures.

In Brazilian correspondent arrangements, activities depend on a contract and a defined scope with the institution. Distinguish who introduces an opportunity, who gathers information, who performs contracted activities and who makes the financing decision. The funder retains the responsibilities assigned to it by applicable regulation. Paying an origination commission does not automatically transfer credit risk, control obligations or product responsibility, even where a contract provides for another party to bear a specified share of certain losses.

Assess a channel through the performance of its origination cohorts, requested exceptions, document quality and correction costs, alongside volume. A large commission may encourage unsuitable proposals if compensation rewards completed transactions alone. A small channel can also require disproportionate operational effort. Diversifying origination improves a portfolio only when the additional source provides information, access or economic independence that justifies its cost and can be monitored. Those benefits must be demonstrated through evidence rather than inferred from a new referral relationship.

Make credit policy, limits and approval authority workable

Credit policy should translate the risk map into decisions people can execute. It should specify eligibility, limits by dependency, minimum documentation, exception handling and review triggers. Limits based only on CNPJ are insufficient when the underlying concentration is a shared buyer. The denominator also matters: a limit may apply to disbursed balances, approved commitments or potential exposure. These measures can produce materially different conclusions while a portfolio is growing, even if the stated percentage remains unchanged.

Approval authority needs to account for aggregate exposure after the proposed transaction. A small application can breach a concentration limit that is already nearly reached. Decision makers therefore need visibility into existing balances and commitments awaiting disbursement. Policy should also identify who may approve exceptions, what justification must be recorded and when independent review is necessary. Commercial discretion without that consolidated view can allow many individually small concessions to accumulate into a substantial unrecognized exposure.

For investors, an important sign of governance is whether an approval can be reconstructed: information available, criteria applied, responsible people and conditions imposed. Internal disagreement need not disappear, but it should be documented. An institution may accept temporary concentration for a specific economic reason if the decision fits its governance framework. Calling that position diversified merely because it contains multiple contracts prevents a clear assessment of the risk accepted and of the conditions under which the decision should be reconsidered.

Use KYC and financial crime controls to identify dependencies

Knowing a customer involves understanding identity, business activity, ownership and control, and the economic rationale of a transaction. Institutions must conduct know your customer procedures and financial crime prevention according to their applicable obligations and policies. For diversification analysis, these controls help identify related businesses, common beneficial owners, inconsistent transactions and arrangements that place seemingly independent applications within one risk group. Their contribution is broader than checking whether every required field has been completed.

An inconsistency in customer records does not prove fraud, just as a complete file does not establish repayment capacity. Investigate proportionately, record what has been clarified and restrict transactions when necessary conditions remain unmet. Several proposals sharing documents, contacts or bank accounts without an adequate economic explanation deserve joint examination. Applications distributed across channels may conceal concentration or attempts to bypass internal limits. The relevant conclusion should follow the evidence, with unresolved issues kept visible to decision makers.

Personal data and business documents require a defined purpose, controlled access, appropriate retention and protection. Sending a file to more providers does not make it more reliable. The institution needs to know who can view, correct and share each item. An investor may appropriately receive a control report describing its scope and limitations instead of unrestricted access to private records. Transparency about the process must coexist with confidentiality and data subject rights, including when information is reviewed by outside specialists.

A numerical example: four blocks and a hidden concentration

Consider a hypothetical portfolio of R$12 million, divided into four blocks of R$3 million. Block A contains industrial suppliers' receivables; B contains loans to distributors; C provides working capital to service businesses; D finances equipment for rural businesses. Each represents 25% of outstanding balances. Every amount in this example is an arithmetic assumption used to compare risk mechanisms. None describes an actual institution's transactions, pricing or performance, and the equal block sizes do not establish equal risk.

Examining payment sources reveals that R$2.4 million in A depends on a buyer called Âncora, as does R$1.8 million in B. Combined exposure is R$4.2 million, or 35% of the portfolio: R$4.2 million divided by R$12 million. An apparently balanced allocation across four products therefore contains substantial customer concentration. Blocks C and D still require investigation for shared dependencies. The absence of an identified connection is not evidence that their cash flows are independent of Âncora or of each other.

In a simplified stress scenario, Âncora stops paying and the final loss equals 40% of the R$4.2 million exposed. The loss would be R$1.68 million, or 14% of the R$12 million portfolio. This assumes recovery of the remaining 60% of that exposure and no additional losses elsewhere. Those assumptions define the calculation's limits. It measures one specified shock rather than the full range of possible portfolio outcomes, and it does not establish how quickly the assumed recoveries would generate cash.

Check where exposure goes when concentration is reduced

Suppose the institution replaces some Âncora-linked transactions as contracts mature. Exposure to that buyer falls from R$2.4 million to R$1.2 million in A, and from R$1.8 million to R$600,000 in B. The R$2.4 million released is allocated to other borrowers, leaving each block at R$3 million and the total portfolio at R$12 million. Combined Âncora exposure becomes R$1.8 million, or 15% of outstanding balances. The product allocation is unchanged, while the identified dependency has decreased.

Repeating only the previous shock, a 40% loss on Âncora exposure now produces R$720,000 of losses, or 6% of the portfolio. The difference from the earlier R$1.68 million is R$960,000. This is not a realized gain and does not establish that the revised portfolio is superior. The comparison assumes replacement transactions are unaffected by the same shock. If they depend heavily on another buyer, the institution may simply have moved its vulnerability rather than reduced it.

Before making replacements, compare the new risks, document quality, potential recovery and monitoring costs. A gradual reduction may avoid forced sales, contractual breaches or rushed lending decisions. Record which dependencies decrease and which increase. Investors should seek that explicit comparison: more borrowers and a lower figure against one limit do not by themselves establish an economic improvement. The key question is whether the revised sources of repayment remain resilient under plausible common shocks, at a cost the institution can support.

Expected loss, concentration and financial results measure different things

Expected loss offers another perspective. Keep the four R$3 million blocks and assume hypothetical default probabilities of 2%, 3%, 4% and 3% for A, B, C and D. Assume losses given default of 40%, 50%, 60% and 40%, respectively. Multiplying balance by default probability by loss given default produces R$24,000 for A, R$45,000 for B, R$72,000 for C and R$36,000 for D. Total expected loss is R$177,000, equivalent to 1.475% of the R$12 million portfolio.

For simplicity, this calculation assumes stable exposure and parameters covering the same horizon. It does not replace accounting or prudential methodology. Adding expected losses does not require independence between blocks, but dependence strongly affects simultaneous losses and extreme outcomes. Two portfolios can therefore share the same R$177,000 expected loss while having different concentration risks. The Âncora shock is a separate scenario, not an amount to add automatically to expected loss. Combining them requires a coherent explanation of what each calculation already includes.

Now assume hypothetical contractual revenue of R$1.8 million over that horizon, funding costs of R$600,000, operating costs of R$180,000 and third-party costs of R$120,000. Deducting those R$900,000 of costs and R$177,000 of expected losses leaves R$723,000 before taxes, other expenses and compensation for equity capital. This is not a promised return. Delays, larger losses, recovery costs and uncollected revenue can change the result. Contractual income and cash actually received must remain distinguishable when assessing performance.

Cash can run short before a final loss is established

A portfolio can remain economically recoverable while experiencing a cash shortage. To illustrate, add a separate R$2 million cash reserve to the R$12 million portfolio, giving R$14 million of resources considered in this exercise. Across three successive windows, assessed at 30, 60 and 90 days, assume inflows of R$1.4 million, R$1.9 million and R$2 million, and outflows of R$1.5 million, R$1.8 million and R$1.7 million. There is no new origination in this initial calculation.

Cash ends the first window at R$1.9 million: the R$2 million reserve, plus R$1.4 million received, less R$1.5 million paid. It then rises to R$2 million and R$2.3 million. In a delay scenario, inflows instead become R$800,000, R$1.2 million and R$2.1 million, with outflows unchanged. Closing balances become R$1.3 million, R$700,000 and R$1.1 million. The lowest balance reveals pressure that an assessment based only on total collections could overlook.

If the institution adds R$900,000 of new origination at the end of the second adverse window, cash falls from R$700,000 to negative R$200,000, assuming no other available source. Commitments must therefore be reviewed before disbursement. This exercise concerns payment timing rather than permanent loss, so it should not be mechanically added to the Âncora scenario's loss. Positive closing balances also do not rule out shortages within a window: the sequence of receipts and payments can matter as much as their totals.

Settlement and reconciliation must account for the money's full journey

Diversification loses value if money cannot be identified and allocated correctly. Before contracting, define who pays, who receives, which obligation is settled, how the balance is updated and who resolves discrepancies. Naming an account in a contract does not establish that funds will follow the intended route. Verify the mechanisms actually available, the necessary authorizations, responsibilities and evidence connecting each transfer to the correct financial event. An expected payment process should not be treated as an implemented control.

Reconciliation should connect the contract, installment or receivable with the expected amount, actual receipt, transaction identifier and accounting status. Partial payments, fees, reversals, duplicates and arrears need different treatment. If a R$100,000 receipt settles only R$80,000 of principal and includes other components, recording the entire receipt as principal reduction distorts exposure and results. Unresolved differences need an owner, an internal resolution deadline and an escalation rule. The explanation must cover both the movement of cash and the resulting obligation balance.

Investors should examine whether the institution can reconstruct a complete sample from disbursement through settlement and balance reduction. Concentration in banks, payment processors and collection providers also matters, along with continuity procedures consistent with contracts and access rights. An outage may delay information without creating a credit loss; misdirected funds may cause loss and legal disputes. A technology platform alone establishes neither payment integrations nor functioning payment routines. Each claimed capability requires evidence of its actual scope and operation.

Build adverse scenarios around coherent causes

A useful stress test explains the causal chain. A major buyer postpones payments; suppliers spend cash to maintain payroll; businesses miss installments; collateral takes longer to sell. One source of stress affects default probability, recovery and liquidity. Applying separate percentages to every stage without checking what is already included can count a loss twice or produce a scenario with no clear economic interpretation. State which effects arise together, which assumptions connect them and which consequences remain outside the calculation.

An operational scenario might involve an originator submitting inconsistent documents, with the problem detected only after several disbursements. Sector diversification offers limited protection when the same documentation control fails across transactions. A regional drought could affect rural business revenue and collateral, with indirect effects on transport and local commerce. Scenarios should examine those connections without assuming that every business in a region behaves identically. Different contracts, reserves, customers and operating conditions can change the severity and timing of each exposure's response.

A stress test should lead to observable decisions: suspend particular applications, revise limits, preserve cash, investigate documents or strengthen collections. Define triggers before commercial pressure makes consistent action harder. Compare the scenario with the institution's capacity to absorb losses and continue meeting obligations, while explaining uncertainties. An exercise that always appears comfortable because assumptions are repeatedly adjusted to fit the budget provides little information. A useful result may identify a constraint that requires slower growth or a change in product design.

Evidence, origination cohorts and favorable samples

Organize performance data by origination cohort, channel, product and stage of maturity. A recently originated portfolio may look stronger simply because it has not reached the periods when arrears usually emerge. Comparing its delinquency with a mature portfolio without adjusting for that difference can mislead. The denominator also matters: contract count, original balance and current balance answer different questions about the frequency and size of problems. A falling current balance can alter a ratio even when no additional borrower has missed payment.

Distinguish arrears, renegotiation, write-offs, recoveries and normal settlement. Extending contracts can reduce reported delinquency without improving a borrower's economic repayment capacity. Selling troubled exposures may remove future losses from portfolio reports, but the discount accepted on sale remains relevant. For each indicator, investors need to understand classification rules, exception treatment and whether totals reconcile to financial records. Otherwise, a change in reporting or transaction structure may be mistaken for an improvement in the underlying credit performance.

Small samples call for clear limitations rather than artificial precision. No observed defaults in a small group does not establish low risk. External information may help if its comparability with the institution's customers, collateral and collection practices is examined. Documentation should distinguish observed facts, estimates and assumptions about expansion. Apply the same discipline to new channels: expected performance is not a verified track record. A plausible forecast can support discussion, but its uncertainty and dependence on untested assumptions must remain visible.

Disagreements and negotiation reveal incentives

Commercial teams, credit teams, operations and investors can disagree about concentration without anyone acting in bad faith. Commercial staff understand demand and the cost of losing a channel; credit identifies dependencies; operations assesses execution capacity; investors examine losses, compensation and information. Negotiation improves when each argument includes evidence and consequences. Calling a customer strategic does not explain how the portfolio would withstand that customer's default. The discussion needs to connect commercial value with the resources and controls required to accept the exposure.

Suppose an originator requests a higher limit after delivering complete documentation and substantial volume. The institution can recognize those strengths while requiring less dependence on one buyer, revised commissions or controls specified in the contract. If remuneration is withheld or the originator shares losses, examine the arrangement's validity, enforceability and the responsible party's financial capacity. A contractual claim against a party without sufficient resources offers limited economic protection. Its stated amount should not be treated as equivalent to readily available cash.

Investors should identify who is paid before the risk matures, who selects the proposals submitted and who bears the consequences of incorrect information. Compensation need not be identical for every party, but conflicts should be understood and addressed. Additional collateral, reporting access and substitution rules may be negotiated without assuming they eliminate losses. Commercial agreements must fit the applicable legal and regulatory structure, with responsibilities described clearly enough for the parties to understand both their obligations and the limits of any protection.

Correct execution errors without disguising performance

A frequent error is recording the same economic dependency under different names, splitting what should be one concentration limit. Correcting it requires standardized identifiers, reviewed connections and recalculated exposures, while preserving the earlier classification history. Another problem is using stale collateral values or ignoring enforcement costs. Review comparable transactions as well, since the defect may lie in the valuation method rather than one contract. A corrected figure should prompt examination of decisions that relied on the previous assessment.

A duplicate balance reduction can artificially lower exposure and release lending capacity that should remain unavailable. Appropriate treatment includes blocking reuse of that capacity, reconciling the payment, correcting the record and checking whether subsequent disbursements relied on the error. Incomplete documentation requires analysis of its effect on eligibility and recovery, rather than merely collecting missing files. Fixing the customer record without reviewing dependent decisions leaves part of the problem unresolved, particularly where the original information influenced approvals or concentration calculations.

Governance should assign responsibility for detection, containment and correction, with enough independence to challenge the originating team. Investors need information about material incidents, their effects and the results of corrective measures, within confidentiality limits. Methodology changes also need explanation. Recalculating historical series may be necessary, but the reason and a comparison should accompany the revision. That allows readers to distinguish an actual change in performance from a legitimate correction to how past performance was classified or measured.

Make decisions traceable and keep their limits clear

A decision can follow a short set of substantive questions: where does repayment come from, which exposures could fail together, what legal right supports action, who decides, and how is the money reconciled? Compare loss scenarios with cash availability and execution capacity. Record conditions for proceeding, limits, information gaps and review signals. When essential information is missing, lowering confidence in an estimate is more defensible than filling the gap with an arbitrary score. The record should explain what remains unknown and why it matters.

KEZS connects entrepreneurial projects and potential funders through technology and intermediation. In a project-oriented approach, transparent documentation can help connect the use of funds, economic activity, proposed rights and identified risks. Its usefulness depends on information quality and verifiability. Investors perform their own due diligence and make their own decisions; KEZS does not assign a credit score. A documented project does not guarantee approval, repayment, profitability, liquidity or preservation of capital. The method supports examination rather than replacing the investor's judgment.

Research entry points include Banco Central guidance on correspondents, CVM materials on investment structures and distribution, Portal do Empreendedor information on MEI, and ANPD guidance on data protection. These sources do not establish KEZS authorization or replace case-specific analysis. Blockchain integrity does not prove underlying truth, legal title, reserve solvency or payment, and private documents should not be placed on public blockchains. Useful diversification results in verifiable decisions, understood costs and clearly stated possibilities of loss, with uncertainty carried into the conclusion.

Frequently asked questions.

References and context

Institutional reference points for checking applicable rules and guidance. Calculations and scenarios are original examples, not quotes, historical results or individualized recommendations.

• Banco Central — correspondentes no País ↗
• CVM — Resolução 88, texto e alterações ↗
• CVM — Resolução 175, fundos de investimento ↗
• Portal do Empreendedor — perguntas sobre MEI ↗
• ANPD — orientações e perguntas frequentes ↗

AI-assisted educational content. Decisions require verification of the information, instruments and providers in each actual case.