Cash flow sensitivity analysis: how to test financial risks

Cash flow sensitivity analysis measures how much the cash balance changes when one assumption moves. It answers questions such as: what happens if customers pay ten days later, revenue falls 8%, exchange rates rise, or interest increases?

The analysis does not try to predict one future. It shows which variables put the most pressure on cash and how much time finance has to act.

While scenario analysis combines assumptions into coherent narratives, sensitivity changes one variable at a time or applies controlled shocks. This reveals the driver that truly moves liquidity.

What cash flow sensitivity analysis is

It is a structured test applied to a base forecast. Finance selects a driver, defines a range, and recalculates:

• minimum balance;
• date of the first gap;
• funding requirement;
• facility usage;
• covenant headroom;
• recovery time.

Example: if DSO increases by five days, what is the largest cash reduction over the next 13 weeks? If the answer is BRL 2 million, collection and credit deserve more attention than a small expense cut.

Formula, baseline, and metrics

The baseline is the reconciled and approved version of the forecast. For each period t:

Closing cash(t) = opening cash(t) + inflows(t) - outflows(t)

For a driver x, sensitivity can be expressed as:

Sensitivity = [cash under test - baseline cash] / change in x

If ten extra DSO days reduce the floor from BRL 4 million to BRL 1.6 million, the impact is BRL 2.4 million. Because the relationship may not be linear, recalculate rather than extrapolate. Also compare the floor date, closing cash, facility usage, and covenant headroom.

Sensitivity vs scenario vs stress test

Method	How it works	Purpose
One-way sensitivity	Changes one driver at a time	Identify critical variables
Two-way sensitivity	Crosses two drivers	Find interactions and risk zones
Scenario	Combines coherent assumptions	Compare possible futures
Stress test	Applies a severe but plausible shock	Measure resilience and contingency response

Sensitivity measures isolated effects. Scenarios combine consistent movements, such as lower volume, margin, and collections. A stress test takes those relationships to a severe but plausible level to test liquidity and response.

How to select drivers by materiality

Prioritize drivers by potential impact, volatility, concentration, speed, and ability to intervene. Use actuals to estimate ranges, contracts to identify limits, and business teams to anticipate events. Start with five to eight material drivers.

Collections

• average customer delay;
• delinquency;
• concentration in major accounts;
• lower sales;
• changes in mix and terms.

Payments

• supplier cost increases;
• shorter terms;
• due-date concentration;
• payroll and taxes;
• unavoidable CAPEX.

Funding

• interest rates;
• foreign exchange;
• facility renewal;
• repayments;
• covenant changes.

Operating events

• project delay;
• supplier loss;
• returns;
• chargebacks;
• tax changes.

One-way sensitivity and a two-way matrix

In one-way sensitivity, vary only one driver and test points on both sides of the baseline. For revenue, use -15%, -10%, -5%, base, +5%, and +10%.

A two-way matrix crosses two drivers. Rows can represent revenue and columns, collection delays. Each cell shows minimum cash:

• green: balance above minimum cash;
• yellow: positive balance but below the operating minimum;
• red: negative balance or covenant breach.

Two shocks that are manageable in isolation may bring a gap forward when combined.

Tornado chart explained in words

A tornado chart ranks horizontal bars by the difference between favorable and unfavorable tests. If DSO takes cash from BRL 4.8 million to BRL 1.6 million while rates take it from BRL 4.1 million to BRL 3.7 million, DSO ranks first. Without a chart, sort a table by spread.

Break-even and headroom

Break-even is the point at which a variable takes cash to its limit. If the minimum is BRL 2 million, DSO break-even may be +8.3 days.

Headroom = baseline result - limit

With a projected floor of BRL 4 million and a BRL 2 million minimum, headroom is BRL 2 million. Calculate it for facilities and covenants too, then translate it into supported delay days or revenue decline.

Correlated effects

"All else equal" helps diagnosis, but lower revenue also reduces taxes and, later, purchases. Run the isolated test first, then model relationships such as variable cost, delay, and elasticity. Document correlations and treat combinations as scenarios or stress tests.

How to run the analysis

1. Build a reliable base forecast

Use a cash budget with balances, open items, payroll, taxes, debt, and realistic dates. Sensitivity on stale data creates false precision.

2. Select a few drivers

Start with five variables that explain most of the movement. Use history, concentration, and materiality to select them.

3. Define plausible ranges

For each driver, create variations such as:

• DSO: +5, +10, and +15 days;
• revenue: -5%, -10%, and -20%;
• cost: +5%, +10%, and +15%;
• interest: +1, +2, and +3 percentage points.

4. Recalculate indicators

Record lowest balance, week of the gap, borrowing requirement, minimum cash, and covenant headroom.

5. Assign a trigger and action

Analysis without a decision becomes a presentation. Define:

• warning signal;
• owner;
• action;
• deadline;
• approval threshold.

Detailed numerical example

A distributor starts with BRL 6 million. Over 13 weeks, it forecasts BRL 30 million in collections, BRL 27 million in payments, and a BRL 4 million floor in week 6. Policy requires BRL 2 million.

With BRL 12 million in receivables and 40-day DSO, ten extra days suggest BRL 12 million × 10 / 40 = BRL 3 million deferred. The invoice-level model refines the impact to BRL 2.4 million in the horizon.

Shock	Lowest balance	Effect
DSO +5 days	BRL 2.8 million	Within threshold
DSO +10 days	BRL 1.6 million	Below minimum cash
Revenue -10%	BRL 2.2 million	Requires discretionary cuts
Interest +2 p.p.	BRL 3.7 million	Low short-term impact

Between +5 and +10 days, cash falls BRL 1.2 million. Approximate break-even is +8.3 days. In the matrix, revenue -10% with DSO +10 days produces BRL 0.4 million because taxes and commissions also fall. The response prioritizes collections, negotiations, and a contingency facility. Warning occurs at 45-day DSO; the critical trigger is 48 days or cash below BRL 2 million.

Examples by business type

In SaaS, test churn, renewals, and delinquency. In retail, test volume, margin, inventory, cards, and seasonality. In manufacturing, test capacity, raw materials, foreign exchange, and CAPEX. For services, test milestones and concentration; for marketplaces, chargebacks, payouts, and reserves.

How to automate sensitivity analysis

Automation recalculates tests when actuals change:

1. banks update balances;
2. ERP systems update open items;
3. CRM and billing update inflows;
4. rules apply shocks;
5. a dashboard shows gaps and triggers;
6. alerts route actions.

The model should preserve versions, dates, and assumptions. The committee then knows whether it is reviewing today's test or last month's spreadsheet.

Governance, cadence, and controls

Treasury owns the model; sales, collections, procurement, controllership, and FP&A own assumptions. Refresh a 13-week forecast weekly, or more frequently under stress.

Controls include bank and ERP reconciliation, protected formulas, independent review, logs, versioning, and segregation between editing and approval. Record each assumption's source, owner, date, and rationale.

Implementation plan

1. Define horizon, granularity, minimum cash, and decisions.
2. Reconcile banks, open items, taxes, payroll, and debt.
3. Approve and version the baseline.
4. Select material drivers, ranges, and owners.
5. Build one-way tests, a matrix, and a stress test.
6. Calculate break-even, headroom, gaps, and funding.
7. Link limits to alerts, actions, and approvals.
8. Validate formulas, automate inputs, and monitor accuracy.

Common mistakes

• testing dozens of variables without prioritization;
• using extreme percentages as the base case;
• changing several assumptions and calling it sensitivity;
• ignoring correlation among revenue, tax, and costs;
• looking only at closing balance rather than the daily path;
• failing to assign action to a trigger;
• updating assumptions without version history.

Control checklist

• Is the base forecast reconciled?
• Have the main drivers been prioritized?
• Are ranges plausible?
• Does the test show the date of the gap?
• Are minimum cash and covenants included?
• Does every threshold have an action and owner?
• Is the model refreshed with actuals?
• Are correlated effects documented?
• Have formulas, access rights, and versions been reviewed?

Sensitivity analysis FAQ

How many variables should I test?

Start with five material drivers. Expand only after the first tests already guide decisions.

What range should I use?

Combine historical percentiles, contractual limits, and forward-looking events. Use asymmetric intervals when risk is not balanced. A regulated price, for example, may have little downside potential and substantial upside risk.

What horizon should I use?

Thirteen weeks works for operating liquidity. Debt and CAPEX decisions may require a monthly view over 12 to 24 months.

Should sensitivity be daily, weekly, or monthly?

Use daily granularity when concentrated due dates can cause gaps within a week. Weekly is suitable for short-term treasury. Monthly is better for structural decisions, provided it does not hide liquidity peaks.

Does sensitivity replace the forecast?

No. The forecast is the expected base. Sensitivity shows how that base changes when an assumption moves.

How should nonlinear relationships be handled?

Calculate more points across the range and avoid interpolation where there are step changes, penalties, covenants, or credit limits. Recalculate break-even close to the threshold.

When should I use a two-way matrix?

Use one when two material drivers may occur together or interact, such as volume and margin, DSO and revenue, or foreign exchange and raw material prices.

What should happen when a test breaches minimum cash?

Quantify the amount and date of the gap, confirm the assumption, notify the owner, and compare actions by timing, cost, and reversibility. Do not wait for the actual balance to reach the limit.

Can AI choose the shocks?

AI can suggest historical patterns and anomalies, but thresholds and actions should follow financial policy, risk appetite, and business context.

How often should tests be reviewed?

Review them at least whenever the forecast is materially updated. Review immediately after a customer loss, exchange-rate move, debt renegotiation, or material variance in actuals.

Conclusion

Sensitivity analysis turns uncertainty into measurable thresholds. Instead of asking only whether cash closes, finance learns how much delay, revenue decline, or cost increase the operation can absorb.

Abstra connects bank, ERP, and forecast data to recalculate sensitivities, alert on gaps, and route actions with traceability.

To automate cash scenarios and stress tests, talk to an expert.