Insight, Industrial Efficiency
Industrial efficiency projects with three-year paybacks routinely do not get done. A new US$100 million platform in India treats that as a financing problem. It is mostly a risk and attention problem, and the product being sold is a counterfactual.
Singapore's August Energy and India's Energeia have launched a US$100 million platform to finance energy-efficiency upgrades for commercial and industrial customers (as reported), covering cooling, heating, compressed air, motors, fuel switching and renewable systems, with customers avoiding upfront capital expenditure.
The problem it addresses is one of the oldest and most stubborn in energy. Industrial efficiency projects routinely show paybacks of two to four years. They routinely do not happen. Engineers produce the calculation, management agrees it looks good, and the chiller is still there five years later.
If the returns were as good as the spreadsheet claims, firms would take them. Something in the spreadsheet is wrong, or something outside it is missing.
Adam Jaffe and Robert Stavins gave the phenomenon its name in 1994, asking what the energy efficiency gap actually means. The modern review by Gerarden, Newell and Stavins sorts the explanations into three groups: market failures, behavioural explanations, and flaws in the models that generate the estimates.
The third group is the one efficiency advocates dislike, and it is frequently correct. The engineering payback omits real costs.
Production risk. Replacing a working utility system in a running plant means downtime, commissioning, and a period where the process behaves differently. In a factory operating near capacity with penalty clauses on delivery, that risk is worth more than the energy saving.
Maintenance capability you do not have. A high-efficiency chiller with variable-speed drives and a control system requires skills the existing maintenance team may not possess. A system that is not maintained correctly reverts to poor performance quietly, and the savings disappear without anybody noticing.
Management attention, which is the scarcest input in the building. A plant manager has a finite number of projects they can push through procurement, installation and commissioning in a year. Efficiency competes against everything else for that slot, and it is not what they are measured on.
Capital rationing that is entirely rational. A firm with limited capital is not comparing the chiller against its own hurdle rate. It is comparing the chiller against a new production line. The line often wins at a worse return, because it grows revenue and the chiller only reduces a cost.
Read those together and the diagnosis behind an as-a-service model is right, and it is not principally about the absence of capital. The customer is buying the removal of performance risk, maintenance obligation and decision burden. Finance is the wrapper, not the product.
Here is the difficulty that has broken energy-service companies repeatedly for thirty years, and it is worth naming precisely because the model is otherwise sound.
The thing being sold is energy that was not consumed.
Payment depends on savings measured against a baseline, and the baseline describes a world that did not happen. Every party to the contract is settling against a counterfactual, and counterfactuals are contestable in a way that delivered goods are not.
Four failure modes follow, and all four are predictable.
Baseline gaming, in both directions. A customer coming off a bad year has an incentive to set a baseline that flatters future savings. A provider negotiating hard has the opposite incentive. Neither party is behaving badly. The instrument invites it.
Production change. Output rises thirty per cent, energy consumption rises with it, and measured savings turn negative against an unadjusted baseline. The adjustment formula is where these contracts are actually decided, and it is usually drafted last and read least.
Rebound. A cold room that is cheaper to run gets run colder, or longer, or with the door open. That is a genuine gain in service and a loss in the savings calculation. The customer is better off and the contract says otherwise.
Attribution across bundled measures. Install new motors, a chiller and a control system together and the saving is real but unallocatable between them, which matters the moment one component underperforms.
The invoice depends on the gap between two lines, and only one of them is data.
Only the first line is measured. The second is an estimate. Production changes and rebound both move it, and neither is directly observable. The payment lives in the gap between them.
What makes the model work in practice is unglamorous: measurement and verification to a recognised protocol, sub-metering at the equipment rather than at the site boundary, adjustment formulas agreed and worked through with real production scenarios before installation, and a dispute mechanism that does not require anybody to go to court over a chiller.
It is worth noticing what that implies. The provider is measuring the quantity that determines its own revenue. That is a structural conflict, not an allegation, and it is resolved in mature markets by independent verification. In a new platform in a new market, it is usually resolved by whoever wrote the contract.
For an operator or finance director being offered this, the useful questions are narrower than the sales conversation.
Who sets the baseline, and can it be reopened?
Ask what happens if the baseline year turns out to be unrepresentative. If the answer is that it cannot be revisited, understand which direction that cuts before agreeing to it.
What happens to the payment when production changes?
Take the adjustment formula and run your own worst case through it. A thirty per cent output increase, a product mix change, a shift pattern change, a year of low utilisation. If the formula has not been tested against those, it has not been read.
Who owns the equipment at the end, and in what condition?
Transfer terms, residual value, and the maintenance standard the asset must meet on handover. This is where the economics of the last three years of the contract sit.
What happens if you close the line?
This is the question nobody asks and the one most likely to matter. A long service contract attached to a chiller in a plant you might mothball, relocate or sell is a liability that survives the decision to stop production. Establish the exit terms while you are still an attractive customer.
The efficiency gap is real and this class of platform is a sensible response to it. It fails when the counterfactual at its centre is treated as an accounting detail rather than as the product.
Get the measurement architecture right before the equipment arrives, because afterwards it is a negotiation rather than a design.
The Lab works on industrial and energy questions across energy access and local manufacturing. The problem of settling against a baseline that describes a world that did not happen is the same one we set out in The Distance Between the Work and the Reward, arriving through a commercial contract instead of a carbon credit, and the same reason what gets measured is decided by the financing structure.
If you are signing or underwriting one of these contracts and want the measurement terms reviewed by somebody with no stake in the savings number, tell us what you need to know.
This is an independent insight piece by Transitions Lab. For the Lab's applied work, see Energy Access & Off-Grid Systems and Local Manufacturing & Supply Chains. See also Who Does It Fail For? on the same structural conflict of a supplier reporting its own accuracy figure. To discuss a study, see Contact.