ECO.10 - Trace Efficiency Savings through Demand Responses (Rebound Effects)
Type: Method pattern Status: Stable
ECO.10:1 - Problem frame
Use this Method when a process or technology needs less resource per useful result and you want to understand total use. Identify the useful result, resource, boundary, period and comparison without the improvement. Trace any change in activity induced by the improvement.
Return a conditional or supported comparison of total use, or the output threshold relevant to it. For a fixed-output question, an engineering comparison can suffice under the stated output assumption. To apply that result to actual use, establish whether output and other material conditions were fixed.
ECO.10:2 - Problem
Lower unit consumption can make a service cheaper, encourage more use, release spending for other activity or make a new use feasible. A response can offset part or all of an initial saving. If it more than offsets a positive fixed-output saving, the result is backfire, also called Jevons’s paradox.
Demand might have grown without the improvement. To attribute a later rise in use to the efficiency change, distinguish the induced response from other causes. Judge the consequence against the actual purpose, such as growth, access or keeping use below a cap.
ECO.10:3 - Forces
Physical efficiency, price, demand and total use can refer to different activities and time scales. A threshold may answer a practical question while future output remains uncertain. Assess the cost and likely value of a broader demand investigation before commissioning it.
ECO.10:4 - Solution
ECO.10:4.1 - State the comparison
Identify the useful result and choose a unit that preserves its performance. Specify the resource, activities included, period and comparison without the proposed improvement. Calculate cost, energy and emissions as distinct quantities, using a supported conversion if the question requires it.
Calculate three totals on the same resource account. Find use without the improvement, use with the improvement at the counterfactual output before any induced response, and use with the response. Subtract the second from the first for fixed-output saving. Subtract the second from the third for the response-induced change. Include relevant equipment, infrastructure and transition costs in the appropriate totals. A burden that arises at fixed output belongs in the second total as well as in the third if it continues. Use applicable PHY, MMP, OPS, MA or FIN results.
ECO.10:4.2 - Follow the responses that can change the answer
Establish how the improvement affects feasible action. Users may buy more, producers may increase output, released funds may support other work, or a new activity may become practical. Identify constraints such as price, capacity, time, access rules or a resource cap. Use a demand estimate that fits the mechanism claimed.
Distinguish resource use in the focal activity from consequential use elsewhere. Add effects within the declared boundary, count each once and state which remain uncertain. Consider new modeling, causal inquiry or measurement when it could materially improve the finding, choice or intended use at an acceptable cost.
For one unchanged service with resource use e per unit and output q, total use is e × q. Compare the improved total with total use in the counterfactual. Let q0 be counterfactual output and e0 and e1 the old and new unit resource requirements. If e1 > 0 and no other terms enter the account, equal total use occurs at q = q0 × e0/e1. This gives a break-even output under the assumptions. Demand at that output is a separate question. When other terms matter, compare the full totals.
ECO.10:4.3 - Return the consequence to the actual purpose
State whether use with the response is below, equal to or above use without the improvement. When the fixed-output saving is positive and well defined, divide the response-induced change by that saving if the requested result needs the fraction offset. A fraction above one indicates backfire under this account. When there is no positive, well-defined fixed-output saving, report the totals and induced change directly. Identify the accounting boundary, since including new equipment or transition can change both the saving and the percentage.
For a resource cap, assess whether aggregate use remains within the limit and whether released capacity needs allocation. If the goal is useful growth, compare the additional output’s value and cost. For a proposed policy, assess feasibility and effects on the parties concerned. A rebound estimate can inform those choices.
Return the supported total-use finding, its assumptions and any material unknown. If the task asks for a choice or operating limit, compare the feasible responses and state which conditions matter to that result.
ECO.10:5 - Archetypal Grounding
ECO.10:5.1 - Half the energy per unit
Suppose production without the improvement would be 100 equivalent units at 2 kWh each, using 200 kWh. The improved process uses 1 kWh per unit, so at the same output it would use 100 kWh and save 100 kWh.
At an induced output of 150 units, total use would be 150 kWh, a saving of 50 kWh against the 200 kWh counterfactual. The added demand would offset 50% of the 100 kWh fixed-output saving. At 220 units, total use would be 220 kWh, a net saving of −20 kWh and a 120% offset. That is backfire. At 200 units the old and new totals are equal.
Now suppose that without the improvement output would already have risen to 180 units. The comparison totals for 220 improved units are then 360 kWh without the improvement, 180 kWh with it at fixed output and 220 kWh with the response. The extra 40 kWh offsets about 22.2% of the 180 kWh fixed-output saving. An observed rise from last year’s 100 units to 220 would not by itself identify either the counterfactual or backfire.
Return to the case with 100 units of output without the improvement. Include 150 kWh for new equipment and transition in the same period. These costs arise at any output with the improvement and are absent without it. If output with the improvement reaches 150 units, the three totals are 200, 250 and 300 kWh. Fixed-output saving is 200 − 250 = −50 kWh. The response adds 300 − 250 = 50 kWh and total use rises by 100 kWh. The process still saves operating energy per unit. On the broader account there is no positive fixed-output saving from which to calculate a fraction offset.
ECO.10:5.2 - An AI task becomes cheaper
Suppose a firm lowers the cost per accepted analysis and considers additional analyses. Establish which output is comparable, including acceptance and any required human revision or quality. Compare costs and resource use per accepted result, feasible volumes and the value to the work that receives them. Token count alone need not represent an accepted analysis.
If unit cost falls, the same expenditure may buy more useful analyses. Spending the full budget, using more electricity and using more hardware are separate consequences. Establish each from the corresponding resource account.
ECO.10:6 - Bias-Annotation
A dramatic paradox can replace a causal account. Engineering optimism can assume fixed demand without saying so; pessimism can assume unlimited expansion. Preserve both the technical gain and the conditions of its economic response.
ECO.10:7 - Conformance Checklist
Identify the useful result, resource, boundary, period and counterfactual. State the mechanism assumed for each induced response and what the evidence supports. Check the totals, any saving-offset percentage and the result for the stated purpose. Distinguish the consequences of stipulated scenarios from an observed causal effect.
ECO.10:8 - Common Anti-Patterns and How to Avoid Them
- A unit saving is reported as a total saving. Carry it through the relevant volume and other induced uses.
- Every subsequent increase is called rebound. Compare with what would have happened without the change.
- Backfire is assumed for every efficiency gain. Find the attainable response and its constraints.
- A carbon, energy or cost figure changes meaning mid-calculation. Keep each total and its conversion grounds separate.
ECO.10:9 - Consequences
A comparison can preserve the engineering gain while showing the demand response and total resource consequence. A break-even threshold may settle a question about a bounded trial or operating limit. Report broader effects conditionally when the evidence does not support a precise rate.
ECO.10:10 - Architectural Rationale
Physical input per useful result supplies one term in the economic account. The improvement can change feasible uses and actual demand, while other causes can change demand independently. Distinguish those effects to interpret total use for the receiving purpose.
ECO.10:11 - SoTA-Echoing
Brockway and colleagues (2021) compare evidence and models for economy-wide energy rebound. Their estimates depend on the scope and assumptions of the studies. The FAccT 2025 analysis of AI rebound identifies possible AI demand responses and limits on attributing changes to efficiency.
Use the mechanism and boundary analysis at the level of detail the question requires. A fixed-output engineering calculation can suffice for a fixed-output question. Use a broader economic model when additional responses matter to the resulting finding or decision and the assumptions fit. Reassess the affected result when performance, demand, access, prices or the resource boundary changes.
ECO.10:12 - Relations
Use ECO.2 for price responses, ECO.3 for newly attainable exchanges and ECO.7 or ECO.9 for a selected shared-resource or rule question. Use PHY or MMP for models, OPS for flow and capacity, and MA or FIN for financial and cost comparisons. Use FPF choice or portfolio Methods if the receiving task requires a choice among alternatives. ECO.10 supplies the demand-response contribution to that comparison.