Signaling and the Separating Equilibrium

What this topic is

Signaling is what the party who knows the truth does when nobody will take their word for it. Assertions are free, so they carry no information: anyone can say the machine is reliable, the revenue is recurring, the candidate is capable. A signal is an action that is expensive — and, critically, more expensive for the party who would be lying. When that condition holds, doing it proves the claim, because the person with the weaker hand cannot afford to copy it. The market then splits into a separating equilibrium: each type takes a visibly different action, and the buyer can read type from action.

Why it matters

This is the repair for the market failure in the previous lesson, and it is the reason half the terms in a commercial contract exist. A meaningful escrow, an uncapped indemnity on a specific representation, a founder who keeps most of their equity, a warranty priced below what the risk would cost a weaker rival — each is a claim made in a currency that cannot be faked. Learning to design one, and to recognise when yours is not actually working, is worth several points of price on any deal where the other side is discounting you for uncertainty.

Who should learn it

Sellers being discounted for risk they know they do not carry, founders raising capital, anyone whose quality is genuinely above the market average and cannot prove it, and buyers who need to tell a real commitment from a decorative one.

What you will understand

• The single-crossing condition — the inequality that decides whether a signal separates or is copied
• How to compute the band of signal strengths that actually work, and why one exists at both ends
• Why the prize for mimicry rises as your signal succeeds, and what that does to the required strength
• Which signals are cheap for the economy and which are pure waste, and why you should prefer transfers

Prerequisites

• Strategy-proofness and Marketplace Rules People Cannot Game
• Mechanism Design and the Revelation Principle

Common misconception

"A strong guarantee shows we believe in the product." Belief is free and everyone claims it, which is exactly why nobody is persuaded by it. A guarantee only carries information if a weaker competitor would lose money by matching it — that is a statement about their cost structure, not about your conviction. The second misconception is that more is always better. Overshoot and the signal costs you more than the price premium it earns, which is the failure mode behind uncapped indemnities, unlimited return policies, and every founder who kept so much equity they could not fund the business.