Skip to content

Direct and Indirect Rebound Audit

Diagnostic audit — instantiates Rebound-Aware Efficiency Governance

Traces where an efficiency gain's freed capacity and freed money actually went — same-service demand, cross-category spending, and induced supply — to see how much of the intended saving rebounded.

After an efficiency gain lands, two things are set loose: freed capacity (the service is now cheaper to consume) and freed money (the bill went down). Direct and Indirect Rebound Audit follows both, channel by channel, to measure how much of the engineered saving came back. Its defining discipline is the split between the direct channel — the same service consumed more because it got cheaper — and the indirect channel — the money saved re-spent on other resource-using goods, plus the higher-order responses of suppliers and induced capacity. Where a forecast assumes the saving simply banks, the audit is empirical and ex-post: it goes and finds the leaks that already happened.

Example

A commercial landlord retrofits an office tower to LED lighting and cuts lighting wattage by more than half. The Direct and Indirect Rebound Audit follows the escapes. On the direct channel, tenants — now that light is nearly free — leave floors lit longer and light space they used to leave dark; measured lighting hours climb, clawing back an illustrative ~30% of the engineering saving. On the indirect channel, the money each tenant no longer spends on the lighting bill does not vanish: some funds longer operating hours, some funds extra IT and plug load, each carrying its own energy. The audit maps every path, sizes it, and reports a blunt figure — the net saving after rebound is materially smaller than the wattage cut implied. That corrected number, not the nameplate reduction, is what the rest of the governance scheme gets to act on.

How it works

The audit's distinguishing move is to refuse a single lump "saving" and instead decompose it by channel. It separates direct rebound (more of the same service) from indirect rebound (savings spent elsewhere) and from second-order effects (supplier responses, induced capacity). It follows both the freed capacity and the freed money, because tracing only one misses half the leak. And it is realized rather than projected — it waits for the response to occur and measures it, which is what separates it from an ex-ante scenario exercise.

Tuning parameters

  • Channel breadth — how many hops of indirect rebound to chase. Deeper tracing catches more but attribution weakens with each hop.
  • Attribution method — simple before/after versus a constructed counterfactual. More rigor, more cost and delay.
  • Money vs. capacity tracing — follow the freed money, the freed capacity, or both. Both is fuller; one alone systematically under-reads.
  • Observation window — how long after the gain to keep tracing, since rebound is often delayed and a short window reads as "no rebound."

When it helps, and when it misleads

Its strength is exposing the re-spending and same-service growth that engineering forecasts simply omit, using the standard direct/indirect rebound taxonomy[1] so the leaks are named rather than lumped. Its failure modes are attribution: indirect spending is genuinely hard to trace to one cause, and an over-eager audit can double-count a resource on both the capacity and the money side. The classic misuse is to stop tracing at the first channel that looks small and declare "no rebound found." The discipline that guards against it is to predeclare the channels before looking, follow both money and capacity, and hand any rebound that crosses the accounting boundary to a dedicated boundary review rather than pretending it disappeared.

How it implements the components

Direct and Indirect Rebound Audit fills the archetype's channel-tracing components — the map of how rebound flows, not the estimate of its size before the fact:

  • rebound_channel_map — its primary output: the enumerated map of direct, indirect, and higher-order channels through which the saving leaked, each one sized.
  • substitution_path_monitor — the strand that follows freed money into cross-category spending, tracking where the saved resource re-appears as other consumption.

It does not quantify the underlying behavioral elasticity (that is the Elasticity Experiment) or range future rebound before commitment (that is Rebound Scenario Stress Test); nor does it set the tolerance its findings are judged against — that is the Absolute Resource-Budget Protocol.

Notes

The audit measures realized rebound, so it needs enough time to elapse before it reads true — run it a month after the change and it will report a saving that later erodes. That is the mirror image of the stress test, which runs before commitment on scenarios; the two are complementary, one ex-ante and one ex-post, and neither substitutes for the other.

References

[1] The rebound effect: an efficiency improvement lowers the effective cost of a service, and part of the expected saving is taken back as increased use. The standard split is direct rebound (more of the same, now-cheaper service) and indirect rebound (money or resource saved is spent on other resource-using goods). ↩