Model Auditing, Error Checking, and Quality Control

Introduction

A financial model that contains an error is worse than no model at all, because it produces output that looks precise but is wrong. In investment banking, where model output directly influences deal pricing, board recommendations, and fairness opinions, an undetected error can have consequences measured in hundreds of millions of dollars. Model auditing is the systematic process of verifying that the model works correctly before its output is used for any decision.

In investment banking, models are never built and used by the same person without review. The standard process:

Analyst builds the model. The analyst constructs the three-statement model, DCF, comps, or transaction model from scratch (or adapts an existing template).

Associate reviews the model. The associate checks every formula, verifies data inputs against source documents (10-K filings, press releases), tests the logic of the assumptions, and verifies that the output is reasonable. This review typically takes 2-4 hours for a standard model.

Model Audit (Model Review)

The systematic process of verifying that a financial model is free of errors, internally consistent, and produces reasonable output. A thorough model audit checks formula accuracy (every cell references the correct inputs), data integrity (inputs match source documents), logical consistency (assumptions are coherent across tabs), balance sheet balance (assets = liabilities + equity in every period), and output reasonableness (implied multiples and growth rates are within expected ranges). In investment banking, the model audit is performed by the associate reviewing the analyst's work and is the single most important quality control step before the model output is used in any client-facing deliverable.

VP or MD spot-checks. The senior banker reviews the key assumptions and output for reasonableness but does not check every formula. They bring judgment about whether the output makes sense given their knowledge of the company and the market.

1. Balance Sheet Balance Check

The most fundamental check: Assets must equal Liabilities + Equity in every period. A dedicated row at the bottom of the balance sheet should calculate the difference (Assets - Liabilities - Equity), which must be zero. Use conditional formatting to turn this row red if the check fails. If the balance sheet does not balance, there is an error, most likely in the cash flow statement, that must be traced and fixed before anything else.

2. Formula Consistency Check

For each row in the projection, verify that the formula in the first period is the same formula used in all subsequent periods. Inconsistent formulas (where Year 3 uses a different calculation than Year 4) are a major source of errors. Excel's "Go To Special > Row Differences" tool can quickly identify cells where the formula differs from adjacent cells.

3. Sign Convention Check

Verify that costs, debt repayments, and other outflows are consistently treated (all positive with subtraction in formulas, or all negative with summation). Inconsistent sign conventions are one of the most common causes of model errors and are among the hardest to detect visually.

4. Reasonableness Checks

After verifying the mechanics, check whether the output makes sense:

5. Sensitivity Stress Testing

Run the sensitivity analysis at extreme values to verify the model behaves correctly:

If the model produces counter-intuitive results at extreme inputs, there is likely a formula error or a broken link.