Real Options in Capital Budgeting

DCF’s blind spot

Static NPV assumes management commits to a project today and runs it to completion regardless of what happens. Real-world projects have choices built in: management can delay, expand, or shut down in response to information that arrives over time. These choices have value. DCF ignores them.

That gap is called real options value.

Three common real options

Option to defer (timing option). A firm has the right to invest later. Like a financial call: strike = investment cost, underlying = project NPV. Value comes from waiting to see how uncertainty resolves.

Example. An oil company holds drilling rights to a lease. Today’s NPV at $70/bbl oil is barely positive. Drilling next year, when oil prices may have moved, could be much more attractive. The firm keeps the rights and waits.

Option to expand (growth option). Building a small initial project preserves the option to scale up if it works. The value of the option to expand is often much larger than the standalone NPV of the small initial investment.

Example. Amazon Web Services in 2006 was a small experimental project. The growth option — to scale into the largest cloud platform on Earth — was the value, not the initial unit economics.

Option to abandon. Management can shut down a money-losing project. Like a put: strike = abandonment value, underlying = remaining project NPV. Value comes from limiting downside.

Example. A movie studio cancels a film mid-production when test screenings come back terrible. The cancellation cost is the strike; abandoning prevents larger losses on continuing.

Why DCF underestimates

Conventional DCF takes a single point estimate of cash flows. It treats uncertainty as risk to be discounted, never as opportunity.

But a real option’s value comes precisely from asymmetry:

• If things go well, exercise (expand, continue, drill).
• If things go badly, don’t exercise (delay, abandon, hold off).

The asymmetry — “I get the upside but cap the downside” — has positive value that DCF can’t capture. The technical name is convex payoff in uncertainty: more volatility makes the option more valuable, the opposite of standard risk-aversion.

A worked deferral example

A pharma firm can invest $100M today to launch a new drug. Two states next year: blockbuster ($300M PV, 50% probability) or flop ($50M PV, 50%). Discount rate 10%.

Static NPV today:

$$NPV = -100 + \frac{0.5(300) + 0.5(50)}{1.10} = -100 + \frac{175}{1.10} = -100 + 159 = +59M.$$

Project looks good. Launch today.

But what if they could wait until next year, see whether it’s a blockbuster or flop, and only launch in the blockbuster case?

NPV with deferral option:

• Blockbuster (50% prob): wait, then NPV = -100 + 300 = +\200M in year 1, discounted: \200/1.10 = $181.8M$.
• Flop (50% prob): don’t invest. NPV = \0$.
• Expected NPV with optionality: $0.5 \times 181.8 + 0.5 \times 0 = $90.9M$.

The option to wait is worth $90.9M vs the immediate $59M static NPV — an extra $31.9M of value from doing nothing for a year. The exact magnitude is sensitive to discount rate and probability estimates, but the qualitative pattern holds: deferral is valuable when uncertainty is large.

When is real-options value largest?

Three drivers:

1. High volatility of project value. More uncertainty → more asymmetric payoffs → more option value.
2. Long time to decision. More time = more chance to learn.
3. Big upside-downside asymmetry. Truncating losses is most valuable when losses would have been large.

Three contexts where real options matter most:

• R&D-intensive industries (pharma, biotech, semiconductors).
• Resource extraction (oil, gas, mining — commodity price volatility creates big optionality).
• Tech platform investments with staged scaling potential.

How practitioners value real options

Three approaches, in increasing sophistication:

1. Decision-tree analysis with explicit branches and probabilities. Easy to set up, sometimes hard to defend probabilities.
2. Binomial models (lattice approach). Discretize stock-price movements like in Unit 5’s binomial option-pricing example, but with the “stock” being the underlying project value.
3. Black-Scholes-style closed-form for simple cases (e.g., a single deferral option on a single asset).

For undergraduate work, the decision-tree and binomial approaches are standard. The honest answer for many real-world decisions is that the qualitative recognition of optionality — and the bias- correction it brings to DCF — matters more than the exact number.

Trap to avoid

The most common abuse of real-options analysis is inflating NPVs to justify projects that should be killed. “It has a real option value of $50M!” is easy to assert and hard to verify. A discipline: the option only has value if management actually has the ability and willingness to exercise it. If the firm never abandons a running project for organizational reasons, the abandonment option is worth $0 regardless of theoretical value.

Real options are real. They’re also easy to overestimate.