Key Stats

• The analysis covers 4,720 estimates from 462 studies, using electricity-demand data spanning 1934 to 2024.
• The preferred bias-corrected short-run elasticity is about −0.16, meaning a 10% electricity price increase corresponds to less than a 2% decline in consumption.
• Among the best-identified studies, the short-run estimate is about −0.09, equivalent to less than a 1% decline following a 10% price increase.
• Price responsiveness rises with adjustment time, from about −0.16 in the short run to −0.38 in the long run.
• The researchers find no statistically significant upward trend across nine decades in how strongly total electricity consumption responds to prices.

Note: The study combines results from hundreds of previous studies rather than following the same consumers for 90 years. Its strongest-design short-run benchmark is based on eight studies.

Continue reading ↓

Smart meters, time-based electricity pricing and decades of technological change have not been accompanied by a measurable increase in how strongly total electricity consumption responds to prices, according to a new research paper examining nearly 90 years of evidence.

The researchers analyzed 4,720 price-elasticity estimates from 462 studies, using underlying data spanning 1934 to 2024. Their preferred estimate suggests that a 10% increase in electricity prices corresponds to a decline of less than 2% in short-run consumption.

The finding challenges a long-running expectation that technologies such as smart meters, automation, storage and time-varying tariffs would make electricity demand progressively more flexible.

Instead, the researchers find that price responsiveness has remained broadly stable across generations of technological change.

Better-designed studies find even smaller responses

The estimated effect becomes smaller when the researchers focus on studies using stronger methods to isolate the relationship between electricity prices and consumption.

Naive studies produced a raw average short-run elasticity of about −0.37, meaning a 10% price increase would correspond to consumption falling by somewhat less than 4%.

Among the best-identified studies, the estimate was only about −0.09. That translates into a consumption decline of less than 1% following the same 10% increase in electricity prices.

After adjusting those stronger studies for other characteristics, the estimate was statistically indistinguishable from zero, although the uncertainty around it was relatively wide.

The researchers therefore do not conclude that electricity users are completely unresponsive to prices. Their results instead suggest that the response may be substantially smaller than less rigorous parts of the literature have indicated.

Consumers respond more when given time

Price responsiveness does increase when households and businesses have more time to adjust.

The paper estimates a bias-corrected elasticity of about −0.16 in the short run, −0.33 over an intermediate period and −0.38 in the long run.

That longer-term response can reflect changes that are difficult to make immediately, such as replacing appliances, improving buildings or investing in equipment that uses electricity differently.

Among 100 studies reporting both short- and long-run estimates, 86% found consumers were more responsive over the longer period.

The distinction is important: demand appears to become more responsive as people have more time to adapt, but the researchers find little evidence that successive generations of technology have made total consumption increasingly price-sensitive.

Time-based pricing did not produce the expected trend

Time-of-use electricity pricing is designed to charge different prices at different times, giving households and businesses an incentive to shift consumption away from periods when electricity is expensive.

Those settings might be expected to show particularly strong price responsiveness as smart meters and other technologies become more common.

Instead, the researchers report that estimates from time-of-use settings were generally closer to zero than estimates elsewhere in the literature.

That does not mean smart meters or time-based tariffs cannot help customers shift when they use electricity. The analysis focuses on total electricity consumption, rather than every form of short-term load shifting within a day.

The authors’ broader finding is that the historical record does not show total electricity demand becoming progressively more responsive to price as these technologies have spread.

Publication bias may have exaggerated earlier estimates

The analysis also finds evidence that selective publication affected much of the observational research on electricity demand.

Across the full short-run sample, the raw average elasticity was about −0.27. Methods designed to correct for publication bias generally moved the estimate closer to zero.

The corrected estimates ranged from approximately −0.04 to −0.23, depending on the method used, with −0.16 chosen as the researchers’ preferred short-run estimate.

The best-designed studies showed less evidence of the same publication-bias pattern, although that evidence comes from a much smaller group of studies.

Why it matters

The question has become more important as electricity systems add larger amounts of wind and solar power.

Because renewable generation can vary with weather and time of day, electricity systems benefit when consumers can shift demand toward periods of abundant supply and away from tighter periods.

If prices alone generate a strong response, that flexibility could reduce some of the need for storage, backup generation and other measures used to balance electricity supply and demand.

The new analysis suggests planners should be cautious about assuming that improved technology will automatically make total electricity demand much more sensitive to prices.

The authors argue that greater flexibility may instead require technologies, contracts and programs specifically designed to change when and how electricity is used.

The findings remain preliminary. The paper was released online in July 2026 and has not been verified here as having undergone peer review.

The authors also note limitations. No individual study covers the entire 90-year period, statistical adjustments could potentially absorb part of a genuine historical trend, and the strongest-design short-run benchmark is based on only eight studies.