Stress testing restaurant costs: common mistakes vs. the right method for multilateral impact

Why doesn't my cost simulation work when demand drops?

Because it's built on a straight line. When you project that if food cost is 30% with 100 covers it will stay 30% at 60 covers, you're ignoring that fixed costs—rent, utilities, base payroll—don't drop with volume.

At 60 covers, your variables (COGS) decline, but those fixed costs spread across fewer dishes; your actual food cost rises to 34-36% even though you're buying less inventory. A proper simulation maps that elasticity: it breaks down fixed and variable costs, calculates food cost as a function (daily_fixed ÷ expected_revenue + %_variable_on_sales) and shows where margin becomes unviable. Masterestaurant applies this method in franchise audits where the stress test reveals that at 70% occupancy the operation can't cover core payroll. It's the maximum loss you can absorb without destroying core employment or discontinuing critical services. The typical error is to keep calculating 'margin' even when you're below that threshold, where cost cuts and renegotiation no longer work; only total closure or radical pivot remain.

What is the break-even point where my restaurant stops being viable?

According to the SATE Institute, 67% of credit failures in Latin America stem from simulations that ignored this break-even point: they granted credit on theoretical margins that couldn't withstand cost elasticity.

The break-even isn't a percentage margin—it's an absolute revenue level below which closing a section (pastry, bar, dining room) or selling equipment becomes unavoidable. In my experience auditing MSME portfolios, identifying that point before stress allows you to design a cost structure that tolerates 20-30% demand swings without breaking. Three scenarios require operational logic, not statistics. Base case: normal month of operation, fixed costs covered, positive margin. Moderate stress: demand drops 25-30%, some suppliers adjust volumes, discounts vanish, two payroll positions freeze—but kitchen and admin remain. Collapse: demand drops 40-50%, mandatory supplier minimums spike (because you buy little), equipment breaks with no cash to fix it, sections close. Each one maps which costs compress, which stay fixed, and where viability ends.

How do I build three stress scenarios without them being just numbers on a spreadsheet?

Masterestaurant has applied this framework across 43 countries, and the difference between a linear simulation and this three-scenario structure is the difference between granting credit blind and granting it with measured risk.

Verifiable figures with sourced data, not invented margins. According to the National Restaurant Association, the U.S. restaurant industry generated over $1.1 trillion in 2024 with an average operating margin of 3-5%—not the 10-15% you see on YouTube. That means a 100-cover restaurant with average check $20 ($2,000/day) operates on net daily margin of $60-100; any 15-20% stress on ingredient costs puts it in the red. In your scenarios, include: real food cost broken down (direct COGS, shrinkage, waste), base payroll plus benefits (CEPAL reports 52 of 100 tourism workers in Latin America are informal; model what happens if that changes), utilities (water, energy, connectivity) with seasonal swings.

What figures should I include in my simulation to make it credible to a bank?

Without those figures your simulation is fiction. Because your simulation doesn't account for operational discontinuities—points where viability breaks without transition.

An 8% margin in base case looks fine, but if that margin comes from ignoring input volatility or assuming fixed costs scale down proportionally, any real shock fractures the business. A seasoned banker runs the model backward: at what volume does your business stop paying payroll, utilities, and interest? If that happens at 70% of expected occupancy and your base model assumes 85%, credit risk exceeds your margin. Diego F. Parra has seen this hundreds of times in portfolio audits: credit granted on projections assuming linear costs, and within four months operational reality (minimum purchases, full-price utilities, fixed payroll) hit and triggered default. A robust simulation identifies that break-even first. Price elasticity in restaurants is low but real: if you cut prices 10%, you gain volume, but not linearly.

How do we handle price elasticity when simulating demand decline?

A pizzeria that drops from $8 to $7.20 doesn't double covers; it gains maybe 15-20% more. Your simulation must capture that:

base_price × (1 − discount_%) × (1 + volume_elasticity) = new_revenue. But here's the error: confusing elasticity with margin sacrifice. Cutting prices compresses your already thin 3-5% margin (per NRA 2024) toward unsustainable levels where even fixed-cost coverage disappears. The best stress strategy, tested on MSME portfolios across Latin America, isn't cutting prices but shrinking SKU (remove low-margin dishes), reducing expensive imported inputs, and boosting labor efficiency. When you model this, you see a 25% demand drop is absorbed without price cuts if you reconfigure menu and payroll. Your ingredient cost rises, compressing margin further. A supplier who negotiated with you at $800/day purchasing power might charge 15-20% premium if you drop to $500/day because their admin, logistics, and billing carry fixed costs too.

What happens when suppliers don't apply volume discounts when your orders shrink?

If your simulation assumes food cost stays 30% at reduced volume but the supplier raises $0.50 per kg on protein, your actual food cost scales to 33-35% automatically.

This is the 'mandatory supplier minimums' effect that regulators mention in stress-test analysis: the restaurant's only option is smaller portions, different protein, or discontinuing dishes. Based on Masterestaurant audits of multilateral credit portfolios, this effect only surfaces if your simulation maps critical suppliers, their minimum volumes, and their price scales. Most failing models ignore this. Currency swings amplify stress: if 60% of your COGS is imported (proteins, wines, equipment) and your currency depreciates 15-20%, your food cost jumps that same amount automatically. A Mexico City restaurant paying in pesos but buying imported U.S. beef in dollars saw costs spike in 2023 when the peso weakened; no model assuming fixed exchange rate survives that reality. Your stress scenario must include: (a) base case: today's rate, (b) moderate stress: 10% depreciation, (c) collapse: 20% depreciation.

How do I factor in exchange rate risk in my simulation if I have imported inputs?

In each you recalculate the COGS for imports, see what margin remains, and where viability ends. If you discover that at 15% depreciation payroll coverage disappears, then your restaurant carries high credit risk without local input diversification.

This analysis is what a multilateral bank expects to see; most applicants skip it. **Elasticity vs. linear assumption:** The error assumes food cost stays 30% if volume drops from 100 to 60 covers. Reality: variable COGS falls, but fixed minimums (services, base suppliers) don't compress; food cost rises to 34-36%. The correct simulation captures this with a function: food_cost = daily_fixed / expected_revenue + %_variable_on_sales. **Enterprise viability vs. theoretical margin:** The error keeps calculating "margin" even below the viability threshold, where structural closures or renegotiation become mandatory. The correct method flags the breakeven: maximum loss volume sustainable without destroying core-employment (kitchen + admin staff), signaling where credit becomes unsustainable.

Key operational differences

**Verified sources vs. opinion:** The error relies on owner projections without sector reference. The correct approach extracts verified benchmarks (CEPAL: ingredient costs by format; ILO: payroll by role; IDB-Lab: viable margins by territory) and cross-checks against real operations from Masterestaurant (8.400+ accounts, 43 countries). **Parameterized scenarios vs. vague intent:** The error labels scenarios "pessimistic" without defining it (-10% or -40% revenue?). The method defines: Moderate = 15% revenue drop + 10% input inflation; Collapse = 30% revenue drop + 20% input inflation, each enriched with territorial risks (devaluation, zone closure, supply disruption). **Granular payroll vs. payroll block:** The error treats all payroll as fixed; stress adjustments appear uncontrolled (arbitrary layoffs). The method disaggregates: kitchen (85-90% fixed: supervisor + base), floor (40% fixed: maître, coordinator; 60% variable: servers, helpers by covers), admin (100% fixed: accountant, manager). Stress scenarios show where areas close without collapsing the core operation.