--- slug: firm-theory type: concept summary: "Ronald Coase's transaction-cost answer to why firms exist and where their boundaries sit, and why AI lowering coordination costs is pushing those boundaries toward smaller, leaner companies." created: 2026-05-26 updated: 2026-06-07 related: lean-team-economics: relation: upstream-of note: "Transaction-cost theory explains why falling coordination and execution costs shrink the cost-minimizing team; lean-team economics is that boundary moving in the 2025–2026 hiring data." one-person-company: relation: upstream-of note: "The theory predicts firm boundaries contract as AI lowers the cost of doing and coordinating work; the one-person company is that boundary pushed to its smallest possible size." solo-founder-viability: relation: upstream-of note: "Whether one founder can carry a company is a question about where the make-vs-buy boundary now sits for a firm of one; transaction-cost theory is the frame that answers it." fractional-contract-talent: relation: upstream-of note: "The decision to buy a function through a fractional executive or contractor rather than employ it is a transaction-cost calculation about which work to internalize and which to source through the market." capital-efficiency: relation: related note: "A firm sized at its cost-minimizing boundary spends less to produce the same output, so the make-vs-buy discipline the theory describes is the structural source of capital efficiency." hiring-sequence: relation: related note: "Each first-hire decision is a choice to move a function inside the firm rather than buy it through the market, which is the boundary question transaction-cost theory poses directly." --- # Theory of the Firm *Ronald Coase's transaction-cost answer to why firms exist and where their boundaries sit, and why AI lowering coordination costs is pushing those boundaries toward smaller, leaner companies.* > **Concept** > > Vocabulary that names a phenomenon. Most founders never stop to ask why a company employs anyone at all. If markets are efficient, a founder could buy every function through contracts: engineering from a shop, design from a freelancer, sales from a commission rep. Yet companies hire, and they hire in particular shapes. The theory of the firm explains why that happens and where the line between "do it inside" and "buy it outside" gets drawn. For most of a century it was graduate-seminar economics. In 2025 it became the cleanest frame for why startups are getting smaller. ## What It Is The theory of the firm is Ronald Coase's answer, set out in his 1937 paper "The Nature of the Firm," to why organizations exist instead of every exchange happening through the open market. His insight was simple: using the market isn't free. Finding the right counterparty, negotiating a price, writing and enforcing a contract, and repeating that work for every transaction all carry a cost. Coase called it the cost of using the price mechanism; economists now call these **transaction costs**. A firm forms when coordinating an activity inside an organization, through authority and employment, costs less than coordinating it through market contracts. The firm's boundary sits where those two costs balance. Internalize too much and the company carries overhead for work it could have bought cheaply. Internalize too little and it keeps contracting for work it does often. Oliver Williamson, who shared the 2009 Nobel in part for this work, sharpened Coase's frame into **transaction-cost economics**. Three conditions matter most. **Asset specificity** means an investment is valuable only inside one relationship, such as a custom integration or a process tuned to one partner. The market becomes risky because the counterparty can hold you up once you're committed, so the firm tends to internalize the work. **Uncertainty** makes fixed contracts weaker: the harder it is to cover every future contingency, the more useful employment's open-ended authority becomes. **Frequency** points the other way. Work done constantly can justify the fixed cost of building it in-house; work done rarely is usually cheaper to buy. The boundary is not arbitrary. It is a calculation, run consciously or not, every time a company decides whether to hire. ## Why It Matters The 2025–2026 relevance is direct. AI is lowering transaction costs on both sides of Coase's ledger at once. It lowers the cost of doing work directly, since a small team with AI tooling can produce what a larger one used to produce. It also lowers the cost of buying and coordinating outside work, because search, specification, and management overhead are partly automatable. When both costs fall, the cost-minimizing boundary moves. The firm gets smaller, more work shifts to tools or markets, and the headcount needed to reach a milestone drops. This is the economic engine beneath [lean team economics](lean-team-economics.md) and [the one-person company](one-person-company.md): those entries describe the trend in the hiring data; this one names why it is happening. The frame changes how each reader reads a company. The **founder** facing a hire is making a make-vs-buy decision whether they name it that way or not. Is this function specific, frequent, and hard enough to contract that it belongs inside the firm, or can the market or a model now carry it? The **investor** evaluating a lean team can ask a sharper question than "why so few people." The real question is whether the company drew its boundary where the costs balance, or merely deferred hiring while pushing work onto vendors and inference budgets. Those costs will show up as margin pressure later. The **senior operator** weighing a [fractional or contract engagement](fractional-contract-talent.md) is on the other side of the same boundary. The fractional model exists because, for many functions, buying expertise through the market now costs less than employing it full-time. This is a pattern in flux as of 2025–2026. Coase's and Williamson's theory is stable and decades-proven; what's moving is the input. How far AI lowers transaction costs is still being learned. So is how far the firm boundary contracts. Any confident claim about the durable size of an AI-era company is a forecast rather than a measured result. ## How to Recognize It The theory shows up wherever a company decides what to own versus what to source. A few questions surface it. - **Is this a make-or-buy decision in disguise?** Every hire, vendor contract, and "should we build this ourselves" debate is a boundary question. The honest version names what's being internalized and why the inside cost is lower than the market cost for that specific work. - **How specific is the asset?** Work tied to one relationship or one custom configuration resists the market and pulls toward in-house, because a contract can't protect against holdup as well as ownership can. Generic, substitutable work pulls the other way. - **How frequent and how contractible is it?** Constant, hard-to-specify work justifies an employee; occasional, easily-specified work is cheaper bought. A function that's drifted to the wrong side of that line is a recurring source of either overhead or friction. - **What did AI just change?** When a tool lowers the cost of doing or coordinating a function, the boundary that was right last year may be wrong this year. Recognizing the theory means re-running the calculation as the inputs move, not freezing a 2020 org chart into a 2026 company. ## How It Plays Out The clearest illustration is the one the whole [lean-team](lean-team-economics.md) shift rests on. A 2020 SaaS company reaching a few million in revenue carried dozens of employees because building, supporting, selling, and marketing were frequent, specific, and expensive to coordinate through contracts. The costs favored internalizing nearly all of it. By 2025, AI had lowered the cost of doing several of those functions and coordinating the rest. A company hitting the same milestone could hold a tighter boundary: a small core team, with more work bought through tools and services that were too costly to coordinate that way before. The org chart shrank not because founders got more disciplined, but because transaction costs fell and the cost-minimizing boundary moved with them. The instructive failure is the founder who treats the boundary as fixed. One carries a 2020 headcount plan into a 2026 company and over-hires for functions AI can now handle, burning runway on internalized work the market would supply more cheaply. Another over-corrects, buying everything through vendors and contractors to look lean, then discovers that the asset-specific, holdup-prone work (the custom enterprise integration, the core product judgment) should have stayed inside. Now the company is hostage to suppliers it can't replace. Both drew the boundary in the wrong place because they read it as a preference rather than a calculation. ## Consequences **What the frame gives you.** Naming a hire, a vendor choice, or a build-vs-buy debate as a transaction-cost decision turns instinct into a question with an answer: which costs less to coordinate, inside or out, for this specific work right now. It explains why lean teams are a structural shift rather than a trend and gives investors a principled way to read a small org chart. It also tells a founder where the AI-era boundary has moved, and where the old boundary still holds because the work is too specific or uncertain to buy. **What it costs to use badly.** The theory is a frame, not a formula: it names the forces but doesn't compute the answer. A founder still has to judge asset specificity and contractibility for each function, and judging wrong is expensive either way. The inputs are moving fast enough that a boundary correct today can be wrong within a year, which makes the calculation a standing discipline rather than a one-time decision. Companies that use the frame well re-ask the make-vs-buy question as costs shift; the ones that use it badly run the calculation once, at founding, and let the answer ossify. ## Sources - Ronald Coase, "The Nature of the Firm" (1937) — the founding paper of the theory, which asked why firms exist at all and answered with transaction costs: organizations form where coordinating activity internally costs less than coordinating it through market contracts. - Oliver Williamson, *Markets and Hierarchies* (1975) and *The Economic Institutions of Capitalism* (1985) — the development of Coase's insight into transaction-cost economics, naming asset specificity, uncertainty, and frequency as the conditions that decide where a firm's boundary falls; recognized with the 2009 Nobel Memorial Prize in Economic Sciences. - The 2024–2025 strand of management research applying Coasian transaction-cost theory to AI agents and the firm boundary — including work in the *California Management Review* on how AI shifts the make-vs-buy calculus — read as the current, still-forming application of a settled theory rather than a measured result. - The empirical counterpart to the theory lives in this book's [Lean Team Economics](lean-team-economics.md) and [The One-Person Company Frontier](one-person-company.md) entries and their sources, which document the falling-headcount trend the theory explains. --- - [Next: Early Traction](early-traction.md) - [Previous: Revenue Model Selection](revenue-model-selection.md)