Why Cities Underestimate Water

In 1928, the Los Angeles Department of Water and Power published a report predicting that the city’s existing water supply would be inadequate for the projected population within thirty years. The report was accurate. The crisis it described arrived roughly on schedule, in the late 1950s, triggering the political battles over the State Water Project that eventually produced the California Aqueduct. The gap between the published prediction and the political response was not caused by ignorance. The 1928 engineers were right and their successors knew they were right. The gap was caused by the ordinary political economy of urban infrastructure: the people who would benefit from action were not yet born, and the people paying for it were very much alive and not enthusiastic.

This pattern repeats so consistently across urban water history that it qualifies as a law rather than a tendency. Cities learn what they need to know about their water supply, fail to act on that knowledge for fifteen to forty years, and then act expensively and under crisis conditions. The engineering question of when the shortage will arrive is usually answered correctly and early. The political question of who will pay, and who will be inconvenienced, and whose land rights will be affected — these questions are deferred until deferral is no longer possible, which is always more expensive than acting when the problem was still theoretical.

The ancient Romans, who built a water infrastructure of genuinely extraordinary sophistication, also provide the earliest clear example of urban water politics. The aqueducts that supplied Rome by the first century CE delivered roughly one million cubic meters of water per day to a city of perhaps one million people — a per-capita supply that would be considered generous by modern standards, and that required constant maintenance across hundreds of miles of channels, tunnels, and elevated arcades. The administrators who maintained this infrastructure (the curator aquarum, under the Republic and early Empire) were well aware that the lead pipes distributing water through the city were a health problem. Vitruvius, writing around 25 BCE, explicitly recommended earthenware over lead, noting that water from lead conduits could not be good for the body given what lead workers looked like.

The recommendation was ignored. Lead was cheaper to work, easier to shape, and already embedded in the capital stock of the system. The political economy of replacement favored deferral. Rome used lead pipes for four more centuries. Historians debate how much this contributed to the various neurological symptoms noted among Roman elites, which is a debate the Romans were structurally unable to have because they were, to varying degrees, experiencing those symptoms.

The Owens Valley story is the ur-text of American water politics, partly because it produced Chinatown and partly because its mechanics are unusually transparent. William Mulholland, Los Angeles’s water superintendent, understood by 1904 that the city’s growth trajectory would exhaust the Los Angeles River within a generation. His solution — a 233-mile aqueduct from the Owens Valley in the eastern Sierra Nevada — was technically brilliant and politically ruthless. Los Angeles agents, with the connivance of the federal Bureau of Reclamation, quietly purchased land and water rights in the Owens Valley before the aqueduct project was publicly announced, stripping the valley’s farmers of the water their land depended on before they understood what was happening.

What gets lost in the Chinatown narrative is that Mulholland was not simply corrupt. He was, in his own estimation, solving a genuine problem. The Los Angeles River really was inadequate. The city really did need additional supply. The methods he used — the land purchasing strategy, the political manipulation of federal water policy, the suppression of alternative options — reflected a genuine conviction that the city’s need overrode the valley’s. He was wrong about the ethics and right about the engineering, which is roughly the profile of most successful urban water projects in American history.

Mulholland’s failure was the Saint Francis Dam, which he built in 1926 and which collapsed in 1928, killing at least 431 people in a flood that destroyed communities from Santa Clarita to Ventura. The failure was engineering — inadequate site assessment on unstable rock — and it ended Mulholland’s career. But the larger project, the aqueduct and the water supply system, worked. Los Angeles grew. The Owens Valley dried. The moral accounting of this bargain remains contested.

The political structure of urban water crises has not changed much since Mulholland’s era. What has changed is the time horizon of the problem. Early-twentieth-century water crises were mostly about supply — getting enough water to where people were. Twenty-first-century water crises are increasingly about climate disruption to supply patterns that cities built their infrastructure around over decades or centuries.

The Colorado River compact, signed in 1922, allocated water between seven western states based on flow measurements from 1905 to 1922, which hydrologists now know was an anomalously wet period in the river’s long-term history. The compact allocated roughly 17.5 million acre-feet per year; the river’s actual long-term average flow is probably closer to 13 million. States built cities and agricultural systems around the higher number. The gap between the number states believe they are entitled to and the number the river can deliver has been known and documented since at least the 1960s. Political resolution has been deferred, repeatedly, through a combination of voluntary conservation agreements, groundwater pumping, and the optimistic assumption that something would work itself out.

Something is, in 2029, working itself out, but not in the way the optimists hoped. Lake Mead is at levels that would have been declared a crisis in any previous decade and is now just the baseline against which policy is measured. The seven-state compact renegotiation that has been described as urgent for thirty years is happening, slowly, under conditions that preclude any outcome that is not painful for someone.

The specific political mechanism that enables water underinvestment is the mismatch between infrastructure timescales and political timescales. A water system is planned over decades, built over years, and operated over generations. A political career is measured in election cycles. The politician who funds the early-stage studies and the land acquisition and the initial design work for a project that will be completed after they have left office — and whose costs are visible now and whose benefits will be felt later — is doing something that the political system does not reward. The politician who defers the investment, avoids the rate increases, and points to the current system’s apparent adequacy is doing something the political system does reward, at least in the short term.

This is not a character flaw. It is a structural feature. The economists who study it call it hyperbolic discounting — the tendency to weight present costs and future benefits in ways that consistently favor deferral. Applied to infrastructure, hyperbolic discounting produces systems that are always somewhat behind where they should be, always catching up, always reacting to crises that were foreseeable and whose political resolution was available years earlier at lower cost.

Cape Town’s near-miss with Day Zero in 2018 is the most recent canonical example of the pattern. The city had been warned about the trajectory of its reservoir levels for years. The population had grown substantially. The infrastructure had not grown commensurately. The rainfall patterns had shifted. By early 2018, the city was ninety days from running out of water and had to implement the kind of emergency restrictions — fifty liters per person per day, enforced with monitoring and significant fines — that any water manager who had read the supply projections from 2010 could have told you would eventually be necessary. The emergency measures worked, barely. Day Zero was averted. The cost of the avoidance — economic, political, social — was vastly higher than the cost of building the resilience that would have made the emergency unnecessary.

This is what makes water crises so frustrating to study. They are not surprises. They are scheduled events that arrive on roughly the schedule the engineers predicted, after the political system has done everything in its power to avoid acknowledging the schedule. The knowledge is always there. The action is almost never proportionate to the knowledge, because the political economy of water infrastructure consistently rewards inaction and punishes foresight.

The cities that have managed water well over time share a characteristic that is not primarily technical. Amsterdam has sustained a sophisticated water management system since the sixteenth century not because Dutch engineering is uniquely excellent (though it is very good) but because the water boards that govern it have, by historical and structural circumstance, interests that are aligned with long-term system health rather than short-term political performance. Singapore made water security a national priority in 1965 not because Lee Kuan Yew was especially far-sighted about hydrology but because the political circumstances of the moment made water dependence on Malaysia an existential concern that concentrated attention.

Good water policy tends to require either catastrophe or an unusual concentration of political will — a moment where the normal political economy is disrupted enough that foresight becomes viable. The catastrophe route is expensive and traumatic. The political will route is rare. Both are available. Neither is as common as the third option, which is knowing what is coming, writing accurate reports about it, filing those reports, and waiting to see what happens.

The reports are still being written. The water is still running. For now.