Energy labels, feebates and standards: Lessons from the EU’s 2021 reform

The European Commission is simplifying the rules around the EU energy label. This column uses product-level sales data on refrigerators in Belgium, France, Germany, and Poland to evaluate the impact on energy use of the 2021 switch from the inflated A+++/A++/A+ scale to a more informative A–G scale. Although consumers prefer better-rated products, they value one euro of future electricity costs at only 42–79 cents. Relabeling alone reduces average electricity use per refrigerator sold by 0.7–2.2%, but other policies would be more effective: a revenue-neutral feebate would reduce electricity use by 2–6% and a stricter minimum standard by 10–16%, at the cost of higher prices.

Energy labels are back on the EU agenda. In June 2026, the European Commission proposed to simplify the rules and move more of the information online. Manufacturers would no longer have to pack a printed label with every appliance they ship, although shops could still ask for printed labels, and buyers would still have to be able to see the label wherever the product is sold. Product information sheets would become fully digital (European Commission 2026a). The accompanying implementation report estimates that labels and ecodesign standards together cut the average household’s annual energy bill by €317 in 2024, about a quarter of it attributed to the label (European Commission 2026b). The label’s reach also keeps growing: smartphones and tablets have carried one since June 2025, rescaled labels for heaters and air conditioners are in preparation, and computers and solar panels may follow. What can clearer labels achieve, and when are standards or financial incentives needed?

The ‘energy-efficiency gap’ refers to energy-saving investments that appear to pay for themselves but that consumers do not make (Gerarden et al. 2017). Earlier VoxEU columns show why the design of the policy response matters. European car buyers valued one euro of future fuel savings at 87 cents, yet fuel taxes still outperformed upfront vehicle taxes in reducing fuel use (Grigolon et al. 2015), and averages conceal large differences across drivers (Levinson and Sager 2021). For appliances, consumers value label classes beyond the energy savings they imply (Andor et al. 2020; Houde 2018). Adding operating-cost information can shift purchases towards lower-rated, cheaper refrigerators with similar overall energy use (d’Adda et al. 2022). A choice experiment found that rescaling increased willingness to pay for top-rated refrigerators (Faure et al. 2021). We complement this evidence with an analysis of actual purchases before and after the 2021 label reform in four countries.

The reform

The March 2021 reform replaced the A+++/A++/A+ scale with a new A–G scale whose top class was left empty at the outset to leave room for future improvement. A model rated A+++ before the reform could find itself in class C, D or E without becoming less efficient. Ecodesign rules that took effect in the same month tightened minimum efficiency requirements. The two measures were designed as complements, but their simultaneous introduction complicates evaluation, as does the 2022 energy crisis that followed Russia’s invasion of Ukraine.

What we did

In Doganoglu et al. (2026), we analyse monthly GfK (now NielsenIQ) data on refrigerator prices and sales in Belgium, France, Germany, and Poland during 2019–2022. Refrigerators run continuously, so their energy consumption is largely decided at the moment of purchase. Between March 2021 and December 2022, average electricity use per refrigerator sold fell by 3.1–7.3% across these countries. Adjusting for product characteristics gives declines of 3.2–3.9%. These changes do not imply causal effects of the label reform: minimum standards, electricity prices, and the product mix changed at the same time, and so did the protocol for measuring energy consumption. 

To obtain further insights into the effects of the label reform, we therefore estimate a model of refrigerator demand and manufacturer pricing, country by country. We investigate two complementary specifications. In the first one, consumers value the label classes directly. In the second specification, labels enter the decision problem of consumers via their effect on the expected 15-year electricity cost.³

Consumers value better labels but still undervalue operating costs

In the specification with direct label effects, we find that the willingness to pay rises with the efficiency class under both the old and the new scale. In Belgium, France, and Germany, a class at a given position in the distribution of models actually sold commands a higher premium under the new scale than under the old one, so the rescaling made a product’s position on the scale matter more for choice. The comparison is approximate, because the reference classes and the allocation of products to classes changed with the reform.

In the electricity-cost specification, buyers value one euro of future operating costs at 47 cents in Belgium, 42 cents in France, 60 cents in Germany, and 79 cents in Poland (Figure 1); before the rescaling the figures were 34, 38, 55, and 82 cents, respectively. Undervaluation thus narrowed where it was deepest and barely changed in Poland. These estimates assume a 15-year lifetime, zero discounting, and constant expected electricity prices; higher discount rates or shorter lifetimes would reduce the measured undervaluation.

Figure 1 Valuation of future electricity costs by refrigerator buyers, before and after the 2021 reform

Notes: Baseline electricity-cost specification, Doganoglu et al. (2026), Table 3. Purchase-price valuation per euro of expected electricity costs, assuming 15 years and zero discounting. A value of 1 means full valuation.

Relabeling, minimum standards and feebates

We use the estimated models to perform policy counterfactuals that predict the equilibrium effects of reforms on sales, prices, energy use, and welfare. More specifically, we compare the information reform associated with the new labels with two alternative instruments: feebates and minimum standards (Figure 2).

The first counterfactual applies the new labels to the February 2021 product range, holding energy consumption, production costs, and electricity prices fixed and letting refrigerator prices adjust. This isolates the information channel. Demand shifts mainly from class F to class E, and average electricity use per refrigerator sold falls by 0.7% in France, 1.8% in Belgium, 1.9% in Poland, and 2.2% in Germany. 

The second counterfactual investigates the impact of a stricter minimum efficiency standard by mandating that every class F and G product meet class E efficiency requirements. We adjust the production costs using our model to account for the change in the label,  while keeping all other characteristics fixed.  Electricity use per refrigerator falls by 9.7% in Germany and 12.8–15.6% in the other countries. The sales-weighted average price rises by between 4.1% in Germany and 8.8% in France, where F and G models dominate sales.

The third policy counterfactual evaluates the impact of a ‘feebate’ scheme, which taxes products in high-consumption classes and subsidizes those in low-consumption classes. France’s bonus-malus for cars, introduced in 2008, is the best-known example. In the basic feebate scheme, we set the tax and subsidy rate equal to the estimated undervaluation and the pivot point such that fees exactly cover rebates in the simulated equilibrium. In the extended feebate scheme (Feebate +), the rate also corrects for the carbon externality. Our counterfactual analysis indicates that electricity use per refrigerator decreases by 1.8–4.7% under the basic feebate scheme and by 3.2–6.0% under the extended feebate scheme. Manufacturers’ prices barely move; what changes is the price consumers pay once the fee or rebate is added. 

Figure 2 Change in average electricity use per refrigerator sold under alternative policies

Notes: Baseline counterfactual simulations, Doganoglu et al. (2026), Table 5. Relabeling uses February 2021; other policies use March 2021. The minimum standard upgrades F/G models to E efficiency. Both feebates are revenue-neutral; ‘Feebate +’ adds carbon damages using €100 per tonne of CO2 and a common gas-based generation benchmark.

We not only compare the effects of the policies on average electricity use but also evaluate the impact on total welfare for the two feebate policies and the minimum standard. Total welfare includes producer profits, the social costs of emissions, and consumer surplus. The latter accounts for the fact that consumers experience energy costs that they do not fully take into account (as implied by our estimates).  Our counterfactual analysis reveals that all three policy interventions raise welfare. The gains amount to roughly €50–100 per refrigerator sold for the minimum standard and €2–11 for the basic feebate, in purchasing-power-adjusted euros. These are total welfare gains, not cash savings for every buyer. Most of the gains come from correcting undervaluation rather than from avoided carbon damages. Distributional effects remain important, as Levinson’s (2018) Vox column explains.

What this means for the policy debate

First, clearer labels do influence purchases, but the information effect alone is modest. Our estimate is the reform’s immediate effect at a fixed product range; it does not include any longer-run adjustment of manufacturers’ portfolios.

Second, the two instruments differ in how they raise the price of the least efficient models. A revenue-neutral feebate leaves F- and G-labelled models on the shelf and adds a fee to their price. The minimum efficiency standard policy replaces them with upgraded versions whose prices are 8–9% higher because they cost more to produce. The larger energy savings due to our minimum standard policy follow from that upgrade, and its estimated benefits depend on what the upgrade is assumed to cost.

Third, a common label has different implications in different national markets, because product portfolios, electricity prices, and consumer valuations differ. In our baseline estimates, the welfare case for a price correction, per euro of electricity cost saved, is strongest in Belgium and France, where undervaluation is deepest, and weakest in Poland, where buyers already take most of the electricity bill into account at the time of purchase. The simulated gains from the basic feebate are €8–11 per refrigerator sold in Belgium, France, and Germany, and €2 in Poland. EU rules already require national incentives to go to the two highest significantly populated classes, which constrains the rebate side of a feebate: subsidies could reward only the top classes, not intermediate ones.²

Two limitations are central. First, our counterfactual simulations hold the product range fixed and so omit possible consequences of redesign, entry, and withdrawal. Firms may already bunch products just below class thresholds, as our data suggests. Second, market-level sales data cannot fully distinguish mistakes from financing constraints, shorter expected lifetimes, or other reasons for giving operating costs little weight. This may affect the welfare effects to some extent, and additional research with household-level purchasing data may provide further insights.

Our reading of the evidence from four European appliance markets is that the labeling reform is useful as a ranking device, but on its own it delivers relatively modest effects. Other policies, in particular minimum efficiency standards and tax incentives that provide price signals, can provide more substantial benefits in reducing electricity use and raising total welfare.

Source : VOXeu