From Coase to the Agent
A reading of ninety years of organizational economics — from Ronald Coase to Oliver Williamson to the agent-mediated firm now forming — and what the unit economics of autonomous knowledge work mean for the institutions that absorb them.
Coase asked a question economists had ignored for sixty years: if markets allocate resources so efficiently, why do firms exist at all? Why not contract every job, every decision, every coordination problem on the open market? His answer became foundational economics. Markets carry hidden costs — finding the right counterparty, negotiating the terms, monitoring whether the work got done, suing when it did not. The firm absorbs these costs by replacing market transactions with managerial authority. Inside the firm, you tell people what to do and pay them a salary. The firm exists as the institutional workaround for the friction of human exchange.1
Oliver Williamson sharpened the argument across the four decades that followed. Three forces, he showed, determine where firm boundaries fall: bounded rationality (you cannot foresee every contingency), opportunism (counterparties may cheat), and asset specificity (some investments lose their value outside the relationship that produced them). The Nobel committee recognized Coase in 1991 and Williamson in 2009. Their combined work explains, with more durability than any rival theory, why companies grow, why they shrink, why they integrate vertically, and why they spin work back out to vendors.2 One variable runs underneath all of it: the cost of getting humans to coordinate.
That cost is collapsing. AI agents — software that perceives, reasons, decides, and acts on behalf of a principal — perform the cognitive work of coordination at the marginal cost of compute. A November 2025 NBER paper from Shahidi, Rusak, Manning, Fradkin, and Horton calls the resulting condition the Coasean Singularity: a state where the activities that once made coordination expensive — eliciting preferences, negotiating terms, monitoring compliance, verifying identity — drop toward zero variable cost.3 Coase’s variable, the one underneath everything, has begun to move for the first time in ninety years.
When the input variable that built the modern firm shifts by an order of magnitude, the firm reconstitutes around the new economics. Companies that read this clearly will redesign their org charts, their workflows, and their unit economics around what coordination now costs them. Companies that read it as automation will drop agents inside organizational structures built for the old economics, capture a small fraction of the available value, and watch their margins compress as the redesigners overtake them. The choice plays out over the next thirty-six months.