Coase asked why firms exist at all if markets are efficient, and answered: a firm internalizes transactions until the marginal cost of organizing one more inside equals the cost of doing it through the market. That equality is an optimum, not a maximum — past it, the firm is worse than what it replaced. The same balance of an inward (grow) and an outward (shrink) pressure sets the right size of every collective structure — firm, nation, empire, bureaucracy, church. There is no scale that is good as such; the optimum is wherever the two curves cross, and it moves as conditions change. This is foundational enough to sit beneath the political question of whether the nation or the global collective is the “right” unit — both are collectives subject to the same law.
Links: Economics, Symmetry Breaking (the mechanism — optimal size is the doctrine of the mean), Did Christian Societies Actually Outperform? (the historical validation — decentralization won), The Gödel Governance Problem (the same law applied to the scale of power), Externalities Fact-Check (Coase’s other half), Market Efficiency and Human Limits, Jevons for Software (cheaper internal production of software moves the Coasean build/buy boundary inward — the firm-level face of the elasticity argument)
Ronald Coase, The Nature of the Firm (1937): if the price system allocates resources efficiently, why is so much economic activity organized inside firms — by command, not by price? Because using the market has costs: discovering prices, negotiating, writing and enforcing contracts. When those transaction costs exceed the cost of just telling someone what to do inside an organization, the activity gets internalized. The firm therefore grows until:
the marginal cost of organizing one more transaction inside the firm = the marginal cost of carrying it out through the market.
That is a boundary, and it cuts both ways. Below it, the firm is too small — it’s paying market transaction costs it could have internalized. Above it, the firm is too big — it’s paying internal coordination costs the market would have borne more cheaply. The efficient firm is at the crossing, not at the frontier of maximum size. Oliver Williamson (Nobel 2009) sharpened the cost side — asset specificity, bounded rationality, opportunism — but the shape is Coase’s: an interior optimum bracketed by two failure modes.
| Inward (grow) | Outward (shrink) |
|---|---|
| Economies of scale | Hayek’s knowledge problem — local information can’t be centralized without loss (see econ README) |
| Internalizing externalities & transaction costs | Principal–agent drift; monitoring & incentive costs |
| Risk pooling | Coordination overhead — communication channels grow ~n(n−1)/2 (Brooks’s Law) |
| Defense / survival (big enough to persist) | Bureaucratic rent-seeking, X-inefficiency, sclerosis |
| Market power | Loss of accountability and local responsiveness |
Optimal size is where these net to zero at the margin. Note the symmetry with Hayek’s price-system argument: the knowledge problem is the outward pressure — it’s the reason a single coordinating mind (a planner, a head office, a capital) degrades as it scales, and why the market beats the firm past a point.
None of this is about firms specifically; it’s about coordination, so it ladders up. The formal generalization to polities already exists: Alesina & Spolaore, The Size of Nations (2003) model optimal country size as the tradeoff between economies of scale in public-goods provision (inward) and the heterogeneity cost of governing a larger, more-diverse population (outward) — Coase’s firm with “transaction cost” replaced by “preference heterogeneity.” Their corollary is the tell: free trade and security guarantees lower the inward pressure (a small state can buy scale on the open market), so the optimal number of states rises as the world integrates economically — which is exactly what decolonization and the proliferation of small states after 1945 looks like.
Run the same logic on empires, agencies, and churches and you get one law: every collective has an interior optimal scale set by the balance of a growth pressure and a coordination-cost pressure.
This is not a separate principle from the vault’s dynamics; it’s an instance of one already mechanized in symmetry-breaking. That page argues a virtue is “a metastable basin between two opposing vices,” and is explicit that “a virtue can’t have only one opposed vice, because any single attractor would just pull the agent in — stability requires opposed attractors.” Optimal size has exactly that shape: too-small (deficiency — can’t capture scale, defenseless, high external transaction costs) and too-big (excess — knowledge problem, agency rot, sclerosis) are the two vices, and the optimum is the basin held in place by their opposition. It’s not a midpoint you average to; it’s a path-dependent equilibrium maintained by two cost curves pulling opposite ways.
And the same page’s second move transfers: the mean shifts with substrate conditions (courage in wartime ≠ courage in peacetime). Optimal scale shifts with transaction-cost conditions — and technology moves the curves. The printing press, the railroad, the corporation, the container ship, the internet, and now AI each re-price coordination, so the efficient boundary of firms and states is a moving target, not a constant. (The platform-giant + gig-economy split is one symptom: digital coordination lowered some internal costs enough to enable continent-spanning firms and lowered market transaction costs enough to push other activity back out to contractors — the curves moved in both directions at once.)
A common Austrian-adjacent claim, stated by TIK and by others, is that firms become inefficient at scale because there are no prices inside a firm — no internal exchange, therefore no calculation, therefore no way to know if you are efficient. Chris’s correction, and it repairs the argument rather than rejecting it:
“This is inaccurate. Most companies do very well tracking prices even internally. The issue is more that they don’t compete. It is fine to know your widget cost $3.75 to make, but the question isn’t asked if it could be made or bought cheaper. So the company knows if it is making a profit or not; what is lacking is the bidding process.”
The distinction matters because the naive version is empirically false and the repaired one is not. Firms have cost accounting, transfer pricing, activity-based costing and internal chargebacks; a large firm usually knows its unit costs to the penny. What it does not have is the counterfactual: the price a rival would have charged. A cost is a number about your process; a bid is information about everyone else’s. Only the second tells you whether the process should exist.
⭐ And this lands exactly on Coase’s margin. “Could it be made or bought cheaper” is the make-or-buy question — the boundary condition above. So the two literatures converge: the reason the firm’s boundary stops expanding is that beyond some point internal provision cannot be evaluated, and it cannot be evaluated because nobody bid against it. Profitability is a whole-firm signal; it does not decompose. A firm can be comfortably profitable while carrying a division that a market would have replaced, and nothing in its accounts will say so.
Two consequences worth keeping:
The vault has already reached the empirical verdict this page supplies the reason for. Christian-structural-advantage credits Europe’s divergence not to scale or unified sovereignty but to fragmentation:
That is the optimal-scale law as economic history: the civilization that refused to collectivize into one empire and stayed as competing mid-sized units out-innovated the unified empires that had passed their efficient governance scale. The verdict was already in the vault; the firm theory is why.
Here is what makes this foundational rather than partisan. The optimal-scale law refutes both poles of every “what is the right unit?” argument at once:
The correct frame is subsidiarity: each function has its own optimal scale (defense may optimize continental, schooling local, trade global), set by that function’s transaction costs. There is no end-goal collective — only a function-relative, condition-relative optimum. This is why the principle sits beneath left and right: both sides routinely commit the same category error of treating a particular scale as a terminal value, when the theory says sacralizing any fixed scale is the error — collectives are instruments with optimal sizes, not ends, and the optimum moves.
Chris’s large HOA (400+ homes, his estimate) is a Coasean collective. Its real value is bulk bargaining: driveways and roofing contracts “probably in the $millions”, weekly lawn care in the $10ks. The trouble is that it bundles bulk purchasing with control over timing and scope:
“Did I need a roof? Did I need a driveway?… the theory is the ability to take advantage of bulk pricing without losing control over what is done.”
Chris floated an insurance-like structure: owners file claims for work. He named the risks himself: free-riding and risk-sharing problems.
Claude’s notes:
Context: Magpie review — HOAs as competing governments, Protection Against the Gradient (friction favours the current holder).
This page is the foundation several vault conclusions silently stand on: christian-advantage’s decentralization verdict, the Gödel-governance worry about the scale of power (a monopoly is a collective past its optimal size whose outward pressure has been suppressed), and the diagnosis (on the unreplaced-event page) that a foundation myth which sacralizes one scale fails when conditions move the optimum.