Why CapEx Is Easier to Approve Than to Measure

Most companies approve big CapEx more rigorously than they ever review it afterward.

Large investments usually clear the same internal hurdles: IRR above the hurdle rate, acceptable payback period, plausible strategic rationale. The investment gets approved. And then, quietly, value fails to appear.

Markets do not value projects. They value companies.

A project can show an attractive IRR and still dilute capital efficiency at the company level — if the incremental return on capital deployed is not strong enough to sustain or improve the firm's economic spread.

The relevant measure is not project IRR alone. It is:

Incremental ROIC = ΔNOPAT / ΔIC

And the value creation test is:

Value creation ≈ (ROIC − WACC) × ΔIC

Capital creates value only when incremental deployment preserves a positive spread. A project that clears IRR but compresses company-level ROIC is not growth. It is capital deployed without improving firm-wide economics.

Companies routinely label defensive spending as growth investment. But much of what is called growth CapEx is simply the cost of staying in the game — defending market share, retaining key customers, responding to competitors, avoiding strategic retreat.

From a management perspective, such investment may feel unavoidable. From a valuation perspective, capital deployed to preserve position rather than improve future economics is not value-creative, regardless of what it is called.

This is why management approves investments the market greets with skepticism. Management sees strategic necessity. The market sees weak incremental returns.

There is a compounding problem: incremental ROIC rarely stays flat as more capital is deployed.

The best projects are executed first. What follows is weaker project quality, higher execution complexity, more overhead, slower ramp-ups, and greater cannibalization. The same dynamic appears in M&A — even a sound target delivers deteriorating returns if the purchase price is too high.

Capital allocation does not get easier with scale. It gets harder. Returns compress either because operating economics weaken, or because the price paid becomes too expensive.

In theory, the solution is straightforward: measure ΔNOPAT properly after approval and hold management accountable.

In practice, this almost never happens cleanly — and that is the real problem.

Once a project goes live, the clean logic of the approval memo dissolves into operating reality. Was the revenue increase caused by the investment, or by market conditions and existing relationships? Did margins improve because of the project, or because of mix shift and cost allocation? Was the growth truly incremental, or partly cannibalized from elsewhere?

A plant expansion may be approved with a clean IRR case. Two years later, sales are higher — but no one can say with confidence whether that uplift came from the project, from a cyclical recovery, from price increases, or from customers who would likely have bought anyway. By then, working capital has risen, additional CapEx and maintenance spending has appeared, common costs have been reallocated, and the original investment case is no longer easy to isolate.

Add to this the follow-on CapEx and overhead that were not fully visible at approval, and the investment case becomes nearly impossible to measure after the fact.

Depreciation is an accounting outcome. Value creation is an economic outcome. They are not the same thing.

For routine expenditures, that gap may be tolerable. For M&A, large-scale industrial expansion, or AI infrastructure, it is not. Once capital is deployed at scale, the question is no longer whether the asset was booked correctly. It is whether anyone actually owns the investment case afterward — and whether it is being measured against a credible baseline.

In most companies, post-approval measurement is partial at best.

This is not just a measurement problem. It is an ownership problem.

The business unit that sponsored the project has little incentive to declare that returns fell short. Finance or FP&A may not have the time or authority to re-underwrite every major investment after approval. Accounting records the asset, starts depreciation, and moves on. The board often sees the investment most rigorously before approval — not two years later, when the economics have become harder to defend.

So the organization ends up with a familiar gap: everyone participated in the approval, but no one fully owns the post-approval proof of value creation.

That is why large investments are often approved with more discipline than they are later reviewed.

CapEx calls it growth. M&A calls it synergy. The structural failure is identical.

Expected value creation is hard to isolate, hard to measure, and easy to rationalize after the fact. Discipline breaks down not at the modeling stage, but after capital has already been committed. In M&A, this is simply more visible: if the purchase price is too high, incremental return compresses from day one. That is how a control/acquisition premium becomes an impairment.

There is no clean solution to this problem. Anyone who claims otherwise has not tried to measure ΔNOPAT two years after a plant expansion.

The structural gap starts at the top. Articles of incorporation define approval authority. They were not designed to track whether capital actually earned its return. No standard corporate charter contains a post-investment monitoring obligation. That clause does not exist — and its absence is not an oversight. The legal architecture of capital approval was built around decision legitimacy, not economic outcome.

What happens after approval is largely invisible to the same mechanism that authorized it.

Some of this can be addressed at the investment committee level. Two things are necessary, and they only work together: lock in a post-audit baseline at approval — expected ΔNOPAT, assumed ΔIC, measurement timeline — and require a formal review session for investments above a defined threshold at year two or three. Without the baseline, the review becomes a narrative. Without the review, the baseline is never consulted again.

Even then, the core problem remains. ΔNOPAT cannot be cleanly isolated from market movements, mix shifts, cannibalization, and cost reallocation. No governance mechanism fully resolves that.

What these measures do is narrow the accountability gap — and make it structurally harder for everyone to simply move on after approval.

That may be the most honest thing a CFO or board can actually offer.

The next time a CapEx proposal looks attractive, the key question is not only whether the project clears IRR.

It is whether the investment will improve company-level capital efficiency — and whether there is a credible mechanism to measure that after approval.

Because approved CapEx does not automatically become value-creative CapEx.

A project may pass the model. It may pass internal approval. It may even look strategically necessary. But if incremental ROIC weakens, or if ΔNOPAT cannot be observed cleanly after capital is deployed, valuation will not follow.

The real problem is not that spreadsheets are careless.

It is that approval is almost always easier than accountability.

Thank you for reading.
Dai Kadomae, CFA, CPA