The British Railway Mania
The oldest case in this site's catalog, and still the clearest proof that a technology can change the world and ruin its investors at the same time.
Updated
What happened
In the early 1840s, railways were Britain's proven miracle: faster than anything in human history, and the early lines paid handsome dividends. Then belief outran arithmetic. Between 1843 and 1846, the railway capital authorized by Parliament exploded from roughly £60 million to about £240 million, on the order of 40% of a year's GDP by one accounting, and by the broadest counts the proposed capital approached a full year's national output (CEPR (opens in a new window), Odlyzko (opens in a new window)). Railway shares roughly doubled in a short span, and promoters floated lines to everywhere, justified by traffic that existed mostly in prospectuses.
What inflated it
The fuel was a financing structure: partly-paid shares. Investors put down a fraction of a share's price and committed to pay the rest later, whenever the company "called" for it. That gave small investors option-like leverage: control of large stakes for little cash up front. Many funded new subscriptions by selling or borrowing against other partly-paid shares, pyramiding commitments they could never all meet, an early form of circular financing (MPRA (opens in a new window)). Researchers estimate this structure let the mania run about two years longer than cash-only investing would have allowed (Odlyzko (opens in a new window)).
What popped it
Reality arrived as a cash call. As authorized lines moved to construction, boards demanded the unpaid portion of subscribed capital, just as the Bank of England tightened creditWhen a central bank raises interest rates or drains money from the system to slow the economy. Most historical bubbles peaked during or within 18 months of tightening.Full definition in the glossary into the crunch of 1847. Investors who had committed money they didn't have defaulted; the defaults revealed that much of the "subscribed" capital had never really existed (Queen's University Belfast (opens in a new window)). Notably, pamphlets and market commentary had been warning about over-promising since 1845, roughly 6 to 18 months before the bust: early, loud, and ignored.
How far it fell, and how long recovery took
Representative railway shares fell roughly 50-70% from their 1845-46 peaks to the lows of 1848 (CEPR (opens in a new window)). Surviving major lines often took more than a decade to regain their peaks; many minor lines never did. About half of the authorized schemes were never built at all.
Who captured the value
Britain got its railway network, and the economy captured enormous value from it over the following decades: cheaper freight, connected cities, industrial growth. But the first-wave shareholders largely paid for a gift to everyone else. Consolidators bought distressed lines at deep discounts to construction cost, and peripheral branch lines took 20 to 40 years of traffic growth to justify their capacity, where they ever did (Odlyzko (opens in a new window)). The pattern, first-wave investors wiped, second-wave owners and the public enriched, repeats in every infrastructure bubble since.
Echoes in today's AI boom
- Spending scaled to the economy. Railway authorization approached a year's GDP; AI-related investment was estimated near 5% of US GDP by 2026, high by any historical standard though below the railway extreme. See Building Ahead of Demand.
- Commitments outrunning cash. Partly-paid shares let 1840s investors promise capital they didn't have; today, AI companies have signed compute contracts an order of magnitude larger than their revenue, promises that likewise depend on future money arriving on schedule.
- Warnings early and ignored. The 1845 pamphleteers were right, 18 months early. The Case Against page shows why being early has usually looked the same as being wrong.
- The technology was real. Railways mattered for a century. The question was never whether trains worked; it was who would still own the shares when the traffic finally showed up.