Natural Field Experiment · Great Britain & Spain
As wind and solar grow, power grids increasingly face a new problem: too much clean energy at once. The usual response is curtailment — paying generators to switch off, and letting renewable electricity go to waste. We tested a radical alternative: expose consumers to free and even negative prices, and let them absorb the surplus instead.
residential customers enrolled
countries — GB and Spain
turn-up events triggered
demand increase when electricity was free (GB)
The Problem
The energy transition is creating a paradox. Wind turbines and solar panels produce the most electricity when conditions are right — but that often doesn't align with when people want to use it. The result is a grid with too much electricity and nowhere to put it.
When that happens, system operators have traditionally paid generators — especially renewables — to switch off. This is called curtailment. It means clean energy that could be powering homes is simply discarded.
Negative wholesale prices — where generators effectively pay the grid to accept their electricity — are already common in high-renewable countries, and they're growing fast. Germany, Spain, the UK, California, Australia: all have seen hundreds to over a thousand hours per year with negative prices.
Behind this headline number is a more subtle problem: even when national prices are positive, local transmission constraints can mean renewable energy is trapped in one area with no way to reach demand elsewhere. In Scotland, wind energy is often locally abundant, but transmission lines to England are full. The solution has been curtailment — wasting the wind. Demand turn-up is the alternative.
Interactive · Experimental Evidence
We randomized ~120,000 Octopus Energy customers into different price treatments during surplus events. Select a country to see the demand response.
Bars show increase in hourly consumption (kW) relative to control. Percentage figures show increase relative to baseline consumption.
Key Finding · Countries Differ
Both countries responded to free electricity, but the nature of the response differed — with important implications for carbon emissions and grid management.
Interactive · Technology Heterogeneity
Who responded most? Households with electric vehicles and rooftop solar were substantially more elastic. Select a household type to see the difference.
Why the difference? Electric vehicles can rapidly increase their charging rate during a one-hour event window. Rooftop solar customers may export less during the event, effectively consuming more of their own generation. As households electrify — heating, vehicles, cooking — the potential for demand turn-up grows.
This suggests demand turn-up becomes more valuable as the energy transition advances, not less — a complement to, not a substitute for, household electrification.
Welfare Analysis
We developed a welfare framework combining our experimental estimates with institutional features of electricity markets. The logic: a single national price masks locationally different marginal values. DTU realizes surplus that curtailment leaves on the table.
Implications
Free electricity works better than paying people. Demand responded strongly when prices hit zero, but offering negative prices (paying people to consume) added almost no additional response. The policy implication: free electricity events are likely the most cost-effective trigger.
The NPG opt-in design was more cost-effective. When customers sign up in advance, payments go only to those who actually change behavior — cutting cost per kWh of induced demand from £1.22 to £0.77. Targeted programs beat broad non-targeted ones on cost-effectiveness.
Demand turn-up and electrification are complements, not substitutes. EV owners and solar households respond far more strongly. As households electrify, the demand-side resource available to grid operators grows. This suggests the value of DTU programs rises over time.
Retail price architecture matters. Non-energy levies attached to volumetric charges dampen the effective price signal consumers receive. Reforming how network and policy costs are recovered would substantially increase the demand response achievable through dynamic pricing.