Rationing by Intermittency

Over one billion people receive piped water on an intermittent schedule: the pipe is pressurized only a few hours or days per week. Economists have long studied non-price rationing via quantity caps and queues, but both assume the good remains continuously available. Intermittency rations a different dimension (access itself) and forces households to invest in private storage to smooth consumption across supply interruptions. This paper asks: what is the welfare cost of this form of rationing, and what prevents governments from replacing it with pricing plus redistribution?

Method.

• Setting: Amman, Jordan: one of the most water-scarce cities in the world, with households receiving water roughly 2 days per week at a heavily subsidized price.
• Large-Scale Data: The universe of meter-level billing data (≈910,000 households × 20 quarters) linked to a household panel survey with storage-capacity measurements.
• Causal Inference: A spatial regression discontinuity at zone boundaries, where adjacent neighborhoods receive different supply schedules for reasons unrelated to household demand.
• Structural Modeling: A two-stage model in which heterogeneous households first invest in durable storage, then allocate consumption across on- and off-days. The model is estimated by maximum likelihood on the survey and validated against the RDD.

Findings.

• Storage is a deadweight capital stock. Private tanks substitute for public delivery reliability. This is a novel stock-channel form of rent dissipation and is distinct from the flow-channel dissipation (time spent waiting) in classical queuing models.
• Intermittency is a regressive implicit tax. Storage capital costs consume roughly 7.5% of weekly expenditure for the poorest quintile but only 2.3% for the wealthiest, collected outside the fiscal system through forced private investment.
• Market-clearing prices dominate intermittency in aggregate welfare. Replacing the status quo with efficient pricing raises social welfare by 3.1% on the existing degraded network and 6.4% when combined with pipe rehabilitation. Mean storage falls by 38–85%.
• The gain flows through government revenue, not consumer surplus. Without redistribution of the fiscal surplus, every household is worse off under pricing. With a flat lump-sum rebate, 49% of households gain, but no feasible transfer rule that conditions only on income produces a Pareto improvement, because the flat volumetric Engel curve makes income a poor proxy for who loses under reform.

Takeaway. Intermittent supply is a costly accommodation of limited state capacity. The pipe’s physical constraints substitute for the transfer infrastructure that efficient pricing requires.