Appendix: Example answer — Lesson 3 forecasting exercise

Don’t read this until you’ve done the exercise yourself. If your answer looks different to this one, that’s likely fine — the point is to practise the moves, not to hit a target number.

(a) Fermi. A large European fast-food chain might process ~50 million broilers per year. If the policy covers ~80% of their supply and represents a modest welfare improvement (slower growth, less crowding, better slaughter practices), you could land anywhere from 20–40 million chickens meaningfully affected per year. A good answer shows the components, not the final number. What “affected” means matters a lot — you might want to multiply by a welfare improvement factor (e.g., 0.1 on a 0–1 scale of suffering reduction), which gives you chicken-welfare-units comparable across interventions.

(b) Base rate. Reasonable final estimates sit between 15% and 40%. Factors up: strong track record, a political window, a supportive public. Factors down: hostile incumbent, similar campaigns already tried, weak enforcement mechanisms. The important move isn’t the exact number — it’s that you started with the 20–30% reference class rather than with your gut.

(c) Pre-mortem. Honest answers here tend to look like: “the company made a public commitment but quietly rolled it back once media attention moved on,” or “the campaign generated headlines but the actual implementation decisions happened at a lower level than we had access to.” The point isn’t to be right — it’s to notice the failure mode before it happens.