Big Is Fragile: Why Large Systems Break

Lehman Brothers failed in September 2008. Its collapse froze credit markets globally. Required hundreds of billions in emergency interventions from central banks. Triggered a cascade of failures across the financial system.

A corner store fails every day in cities across the world. Its employees find new jobs. The owner absorbs the loss. Someone else opens a store in the same space within a year. The failure is contained.

The difference is size. Large systems have failure modes that small systems don't.

The Concentration of Failure Modes

When a system grows large, the number of failure modes doesn't scale linearly. It scales faster. More connections. More dependencies. More ways that a local failure can cascade into a global one.

Lehman's business touched every major financial market. Its borrowing involved hundreds of counterparties. Its trading positions were enormous. Its failure didn't merely hurt Lehman. It hurt everyone connected to Lehman. Which was almost everyone.

A corner store's failure affects the store, its employees, and local customers. The rest of the economy doesn't know it failed. The failure is absorbed locally.

This is the core principle: the cost of failure for a large system is often borne by the entire system, not just by the failed entity.

Lehman vs. the Corner Store

Lehman was systemically important. Its failure was everyone's problem. This sounds like Lehman mattered more. It mattered in the sense that Lehman was bigger, but not in the sense that Lehman was more robust.

The label "too big to fail" is usually read as reassurance. It should be read as a warning. An institution whose failure has system-wide consequences is one whose size is a systemic liability, not an asset.

The correct policy is not to make such institutions "unfailable"—to guarantee they will never fail. The correct policy is to make them small enough that their failure is survivable. Size should be capped by the cost that the system can absorb, not by the profits the institution can generate.

We have not done this. Instead, we have treated Lehman-sized institutions as too important to let fail. We have bailed them out, propped them up, and allowed them to grow even larger.

The cost of the next Lehman failure will be even higher.

Switzerland Outlasts Empires

Switzerland is smaller than most U.S. states by population. Smaller than most U.S. cities by area. It has survived every major European upheaval of the last five hundred years—the Thirty Years' War, the Napoleonic Wars, two world wars, the Cold War, the EU's expansion—with its governance structure largely intact.

The Soviet Union, the Austro-Hungarian Empire, and the Ottoman Empire all dwarfed Switzerland. All collapsed.

Switzerland's secret is decentralization: 26 cantons with substantial autonomy. Decisions pushed to the lowest competent level. No single point of failure at the national level.

When a canton faces a crisis, it affects that canton. The system persists. When an empire faces a crisis at the center, the entire empire is vulnerable.

Small and distributed beats large and centralized in terms of time-tested survival.

The Concentration of Power

Large systems concentrate power. A large corporation has decisions made by a small group at the center. A large government has power concentrated in a capital. A large army has command flowing from a single general.

When the center is right, everything flows. When the center is wrong or breaks, nothing works.

Smaller systems distribute power. Many small businesses compete, innovate, adapt. Many local governments experiment, learn, adjust. Many independent teams self-organize, adjust, adapt.

The large system looks more efficient at coordinating. Until it isn't. Then its size becomes a liability.

The Chip Supply Disaster

A major automotive manufacturer, optimizing its supply chain, consolidated its chip supply to a single specialized vendor. For a decade, costs were reduced and efficiency was measured in millions of dollars saved.

Then the vendor's fab caught fire.

The manufacturer could not obtain chips. Production halted for months. Losses from the single event exceeded all the savings from consolidation.

This pattern—cost optimization by concentration producing catastrophic exposure—is so common that it now has a formal literature. Every decade produces another round of manufacturers rediscovering that optimization and fragility are two names for the same choice when the underlying risk is Extremistan.

The solution is not to optimize to the single-best supplier. The solution is to fragment supply so that no single failure can halt production.

That fragmentation looks wasteful. It is insurance.

The Roman Empire's Slowness

The Roman Empire at its height was the largest and most sophisticated state structure ever assembled. The Empire's bureaucracy, at its peak, could take months to transmit orders from Rome to the frontier.

Information about crises arrived slowly. Responses arrived slower.

Smaller polities—barbarian confederations, Byzantine successor states, local warlords—could decide and act in days.

Over a few centuries, the slow, centralized Empire lost to the fast, decentralized alternatives. Scale produced inertia that scale couldn't compensate for.

The pattern recurs in modern corporations facing smaller competitors. The large firm has more resources but loses speed. By the time the large firm's decision process produces a response, the smaller competitor has already adapted.

Network Effects and Too-Big-to-Fail

Some industries develop network effects. The value of your membership depends on how many other members there are. Financial networks, telecommunications, payment systems, social networks.

These industries naturally concentrate. The largest network has the most value. Scale becomes an advantage.

But scale in networks creates systemic risk. If the network fails, the entire system collapses. Everyone who depends on the network loses access.

The 2008 financial crisis demonstrated this. The financial system had become a single large integrated network. A failure in one node cascaded to all nodes. The system broke.

The correct response is to keep financial networks smaller and more decentralized. Instead, we have continued to concentrate. Banks are larger. Networks more integrated. The next crisis will be even more expensive.

How Size Produces Fragility

Size produces fragility through several mechanisms:

Couplings increase: more connections, more ways for failures to cascade.

Speed decreases: larger organizations respond more slowly to disruptions.

Failure modes change: a single node failing doesn't break a large system; a central component failing does.

Regulatory capture increases: large institutions can influence rules in their favor, reducing real competition and resilience.

Agency problems increase: the incentives of decision-makers diverge from the interests of those bearing risk.

None of these problems exist in small systems.

The Practical Question

When evaluating any large structure—a bank, a company, a nation, a supply chain, a technology platform—ask: "What happens when this breaks?"

If the answer is "most things downstream also break," the structure's size is a liability, not an asset.

The correct response is to keep things small enough that their failure is survivable.

We know how to do this. We choose not to. The profits from size overwhelm the costs that size imposes on others.

Until those costs are imposed on the decision-makers themselves, the incentive structure won't change.