Through-Cycle Normalization: Setting the Right Earnings Baseline

Introduction

Through-cycle normalization is the most important valuation technique in industrials banking. Every other concept in this guide, from operating leverage to cycle positioning to LBO modeling, ultimately connects back to the fundamental question: what does this company actually earn in a "normal" economic environment? The answer to that question determines the valuation anchor for virtually every industrials transaction.

The core problem is straightforward. A capital goods company reporting $300 million in EBITDA at the peak of a cycle might earn only $150 million at the trough and $220 million at mid-cycle. Applying a 10x multiple to each produces three very different enterprise values: $3 billion, $1.5 billion, and $2.2 billion. Which one is "right" depends entirely on which EBITDA figure you use, and the answer is almost always the normalized mid-cycle figure, not the trailing twelve months. This article explains the three primary normalization methods and how bankers apply them in practice.

The most straightforward normalization method is to average a company's EBITDA (or EBITDA margins) across a full economic cycle, typically 5-7 years. The logic is simple: if you average across both peaks and troughs, the extreme observations cancel out, and the mean approximates what the company earns at a "normal" point in the cycle.

Mid-Cycle EBITDA (Historical Average Method)

The arithmetic mean of a company's EBITDA over a period covering at least one full economic cycle (typically 5-7 years, or longer if the cycle is extended). This figure represents the company's estimated earning power at a "normal" point in the economic cycle, stripping out the effect of unusually strong or weak demand. Applying a valuation multiple to mid-cycle EBITDA rather than trailing EBITDA is the standard approach for valuing cyclical industrial companies.

How to apply it. Pull 5-7 years of historical EBITDA, calculate the arithmetic mean, and use that figure as the earnings base for your valuation. For a company with EBITDA of $180M, $220M, $280M, $310M, $250M, $190M, and $160M over seven years, the average is approximately $227M. This is your mid-cycle EBITDA estimate.

Strengths: Simple, transparent, and easy to defend in a negotiation. Both buyers and sellers can verify the inputs. It does not require subjective judgment about future cycle trajectories.

Weaknesses: It assumes the past cycle is representative of the future, which may not hold if the company has structurally changed (acquired businesses, divested segments, implemented cost improvements). It also gives equal weight to every year, even though some years may have been affected by one-time events unrelated to the cycle.

The margin regression method normalizes by applying a mid-cycle margin percentage to the company's current or projected revenue base. This is particularly useful when a company's revenue has grown structurally (through acquisitions or organic growth) such that historical absolute EBITDA figures understate current mid-cycle earning power.

How to apply it. Calculate the company's average EBITDA margin over a full cycle period. Then apply that margin to the company's current or last-twelve-month revenue to estimate what EBITDA would be at a normal margin profile.

For example, if a company's EBITDA margins over the past six years were 18%, 21%, 24%, 22%, 16%, and 14%, the average margin is approximately 19%. If the company's current revenue is $600 million (reflecting recent acquisitions that were not present in earlier years), the mid-cycle EBITDA estimate is $600M x 19% = $114 million, even if the company is currently earning $85 million at a cyclical trough or $144 million at a cyclical peak.

Strengths: Accounts for structural changes in the business (revenue growth from acquisitions, geographic expansion, new product lines) that make historical absolute EBITDA levels less relevant. More applicable to growing companies where the revenue base today is meaningfully different from five years ago.

Weaknesses: Assumes the company's historical margin range is still achievable, which may not hold if the cost structure has changed, the competitive environment has shifted, or the revenue mix has evolved.

The capacity utilization method links a company's earnings to industry-wide capacity utilization data published by the Federal Reserve. The logic is that margins correlate with utilization rates: when utilization is high, manufacturers have pricing power and fixed cost absorption is favorable; when utilization is low, the opposite holds.

How to apply it. Regress the company's EBITDA margin against its relevant industry capacity utilization rate over 7-10 years. Use the regression equation to estimate what the margin would be at a "normal" utilization rate (typically the long-run average of ~79-80% for total manufacturing). Apply that normalized margin to current revenue.

Strengths: Provides an economically grounded link between the company's earnings and the macro environment. Particularly useful for companies in sub-sectors where capacity utilization data closely tracks company performance (metals, chemicals-adjacent, heavy manufacturing).

Weaknesses: Requires the company's performance to correlate meaningfully with the available utilization data, which may not hold for diversified companies with exposure to multiple end markets. The regression may also be unstable if the company has undergone significant operational changes.

In a typical industrials transaction, bankers do not rely on a single normalization method. They triangulate across all three (and sometimes additional approaches like management-guided normalization) to establish a range of mid-cycle EBITDA estimates.

On the sell-side, the banker's objective is to demonstrate that the company's current earnings are sustainable or that any above-mid-cycle component is offset by secular growth (using the secular vs. cyclical framework). On the buy-side, the objective is to ensure the acquisition price reflects normalized, not peak, earning power.