No. 092Strategy & Enterprise

The Nature of the Firm (Transaction Costs)

Firms exist because using the market has costs; an activity belongs inside when organising it there is cheaper than buying it, so when those costs change, the boundaries should too.

A firm grows until organising one more activity inside costs more than buying it; AI changes those costs, so it moves the boundary.

To an executive

We do work in-house when coordinating it is cheaper than buying it. AI changes those costs, so it should change what we own, outsource and automate.

To an engineer

Every boundary between teams or suppliers carries search, contract and monitoring costs. When AI cuts coordination costs, revisit which work sits in which team, system or vendor.

The idea

In "The Nature of the Firm", published in Economica in 1937, Ronald Coase asked a question economists had skipped: if markets coordinate work so well through prices, why do firms exist at all? Inside a firm, people are directed by managers, not prices.

His answer was that using the market has costs. You have to find out who can supply what and at what price, negotiate and write a contract for each exchange, and check that it is honoured. A firm saves some of these costs by replacing many separate contracts with one longer relationship in which someone directs the work. But organising inside has costs too, and they rise as the firm grows and management becomes stretched. So a firm expands until the cost of organising one more transaction inside equals the cost of carrying it out in the market, or in another firm.

These came to be called transaction costs. In 1991 Coase received the Nobel Memorial Prize in Economic Sciences for clarifying the significance of transaction costs and property rights for how the economy is organised.

Why it works

Boundaries follow costs, not tradition. What a firm makes, buys or partners for reflects how expensive each option is to coordinate. When those costs move, the right boundary moves with them.

Technology changes coordination costs. The telephone, the internet and shared data changed what was cheap to arrange inside and outside firms. AI changes them again: it can search, draft contracts, translate between systems and monitor work at low cost, but it also adds new costs of checking its output.

The same logic applies inside the enterprise. Teams, shared services and platforms are small firms within the firm. Every hand-off between them carries its own search, agreement and monitoring costs.

Where it applies

Sourcing and build-buy-partner decisions, where AI changes the cost of building in-house, managing suppliers or using an AI service instead.
Organisation and team design, where AI may let smaller teams do work that once needed coordination across several.
Integration and platform strategy, where the cost of connecting many applications decides what to centralise and what to federate.
Shared services and outsourcing contracts that were set up when coordination was expensive.

How to apply

List the coordination costs. For an activity, write down what it costs to arrange it inside (management, hand-offs, alignment) and outside (finding suppliers, contracting, monitoring quality).
See which ones AI changes. Be specific: AI may cut the cost of drafting and checking contracts, but add the cost of verifying AI output and managing model risk.
Redraw the boundary. Decide whether the activity should move inside, to a supplier, or to an automated service, and what team size it needs.
Revisit regularly. Coordination costs keep falling unevenly, so review sourcing and team structures as the technology matures.

Example scenario

A retailer outsourced product-content work to an agency: writing descriptions, translating them and checking them against supplier data. The contract existed because doing it in-house needed a large team and the agency had the scale. Coordinating the agency still took briefs, review rounds and disputes over quality.

With AI able to draft and translate from structured product data, the enterprise architects recosted both sides. Organising the work inside now needed a small editorial team and a reliable product data feed; the remaining agency work was mostly specialist copy. The retailer brought routine content in-house, kept the agency for campaigns, and invested in the product data the AI depended on.

Questions to ask

For this activity, what does it cost us to coordinate it inside compared with finding, contracting and monitoring a supplier?
Which of those costs does AI actually lower here, and does that move the boundary towards doing it ourselves, buying it, or handing it to a machine?
Are our team and unit boundaries still set by coordination costs that no longer apply?

When not to use it

Strategic or capability choices, where an activity is core to your advantage whatever it costs to coordinate.
Decisions dominated by regulation or risk, such as data residency, where transaction costs are not the deciding factor.

How it misleads you

It is easy to count only the costs AI lowers and forget the ones it adds: checking output, managing models, and losing the supplier's expertise. Coase's argument also concerns the margin, one more transaction, not wholesale change, so beware of using "AI cuts transaction costs" to justify sweeping insourcing or headcount cuts without costing each activity.