When bankers or investment committees insist on a single WACC “because that’s how it’s done,” they are not defending economics. They are defending habit.

The request usually sounds reasonable. One number. Comparable across deals. Easy to explain. Easy to approve.

But that demand quietly assumes something that no longer holds: that project risk can be averaged, smoothed, and treated as stationary through time.

It can’t. A project is not a value. It’s a system, and systems don’t de-risk smoothly.

Projects resolve risk in steps, not gradients

Project finance risk doesn’t decline gradually as time passes. It resolves through events.

A permit is granted. A contract is signed. Financial close is reached. Construction completes. Performance is proven. Debt amortises.

Each step removes an entire class of uncertainty. Nothing incremental about it.

Yet most valuation models ignore these step-changes. They apply a single weighted average cost of capital to every dollar, regardless of whether that dollar is speculative, contingent, or contractually protected.

That simplification is convenient. It is also misleading.

A project is a sequence, not static.

A useful way to think about a project is as a sequence of economically distinct exposures:

  • Development behaves like an option: asymmetric outcomes, high abandonment risk
  • Construction is dominated by binary completion risk and counterparty performance
  • Early operations test assumptions rather than scale returns
  • Contracted operations often resemble long-dated credit instruments
  • Residual tails reintroduce market risk

Lumping these together into a single “project risk” does not make them comparable. It hides their differences.

When capital is allocated on the basis of that average, it tends to reward familiarity over understanding.

What the standard valuation actually assumes

The standard formulation is familiar:

(Equation: NPV as the sum, over every period t from 0 to T, of the period’s cashflow CFt discounted at (1 + WACC)t, shown in the original post.)

What’s less often examined is what this assumes to be true.

It assumes that:

  • Risk exposure is broadly stable through time
  • Uncertainty decays smoothly rather than discretely
  • Early-stage uncertainty can be treated as a scaled version of long-dated certainty

These assumptions are tolerable in diversified corporate settings. In project finance—where path dependence is the defining feature—they are doing real work.

Usually unnoticed or quietly ignored.

Why the single WACC survives

The single WACC persists for reasons that have little to do with economics.

It is:

  • Easy to explain
  • Easy to benchmark
  • Easy to defend in committee

It provides a common language, even if that language is blunt.

There is another reason it survives: it keeps judgement implicit.

Once the discount rate is agreed, the rest of the analysis feels mechanical. Disagreement shifts to operational assumptions, where it is easier to argue and easier to test.

Risk attribution stays buried, unnecessarily.

Better models won’t fix this

At this point, it is tempting to reach for a technical solution.

More detailed models. More tabs. Better templates. Cleaner spreadsheets.

None of that addresses the problem - we need better judgement.

The issue is not computational power. Modern spreadsheets are already more than capable of handling phase-specific discounting, scenario analysis, and structural complexity.

The constraint is not the tool. It is the judgement applied within it.

Valuation errors here do not arise because models are too simple. They arise because market participants are trained to optimise process and precedent rather than reason about risk mathematically through time.

Improving the models without improving the underlying judgement simply allows the same assumptions to be embedded more efficiently.

“But WACC barely moves value”

There is a widely held belief—often true—that valuation is more sensitive to operational assumptions than to discount rates.

Small changes in availability, capex, or operating cost can dwarf the impact of a 50–100 basis point change in WACC.

This observation is usually taken to mean that precision in discounting doesn’t matter much.

That conclusion is incomplete - so many operationally competitive and hones deals come down to WACC.

Sensitivity is not the same as importance

Operational assumptions shape how much cash is generated.

Discounting determines when that cash is treated as valuable.

Errors in operational parameters tend to be:

  • Local
  • Testable
  • Gradually revealed through performance

Errors in discounting are:

  • Global
  • Structural
  • Hard to detect once embedded

A mis-specified WACC doesn’t announce itself. It quietly reweights the entire project—shifting value between early and late cashflows without leaving a clear trace.

When a single WACC is applied across all phases, an error doesn’t just move value up or down. It reallocates value across time and that matters.

Systems thinking changes the valuation structure

If you think in systems rather than values, the question stops being:

What is the right WACC?

And becomes:

What risk exists at each stage, and what return is required for capital at that stage?

Mathematically, that change is simple but profound.

Instead of treating NPV as a single sum, it becomes a sum of sums—one for each phase of risk:

(Equation: NPV as a sum of sums, over each phase p from 1 to P and each period t in that phase, of CFt discounted at (1 + rp)t, shown in the original post.)

Where:

  • p indexes distinct project phases
  • t_p reflects the required return for risk in that phase

This is not about precision. It is about structure.

Isn’t this double-counting risk?

A common objection is that project cashflows are already conservative.

Engineers haircut outputs. Construction models include contingencies. Contracts embed credit judgement. Why add more risk via discounting?

The issue isn’t conservatism. It’s coherence.

Phase-aware discounting doesn’t add risk. It locates it.

Capital structure doesn’t solve this

Capital structure reallocates downside. It does not define economic value.

Treating structure as a substitute for valuation simply pushes judgement into places that are harder to see - which is a bigger problem I’ve witnessed time and time again.

What actually needs to change

This is not a tooling problem.

Project finance does not need better spreadsheets - well it would help solve a bigger problem but AI will solve that with the likes of www.formuly.ai

It needs better mathematical judgement and that requires people or systems better trained at identifying the financial risk dynamics beyond gearing / leverage and whether or not a contract or counterparty is strong vs medium risk.

Specifically:

  • Clearer reasoning about when risk is borne
  • Willingness to make risk attribution explicit
  • Comfort with judgement that cannot be reduced to precedent

Until that changes, no amount of modelling sophistication will fix the underlying mispricing.

Conclusion

Risk is not stationary. Scarce capital deserves models that reflect that reality.

Insisting on a single WACC because “that’s how it’s done” is no longer a serious position. It is a refusal to engage with how projects actually work.

Valuation is not about producing a number. It is about understanding a system.

If you made it this far you probably get it. DM me for a conversation or if you have a subject you’d value my thoughts on.

First published on Nick Crawley’s Substack.