Evaluating the impact of lifestyle changes: a scenario-based analysis for Europe’s residential buildings sector

PRIMES-BuiMo (Capros 2018) is a hybrid economy-engineering model used for the development of scenarios for the residential and services buildings sector of EU member states under alternative policy futures (Fotiou et al 2019). Decision-making of agents concerning energy efficiency interventions, including choices about the upgrade of heating equipment, and the timing and depth of renovation, is modelled endogenously based on microeconomic theory, while respecting engineering constraints. The modelling framework represents market (i.e., true costs relating to hidden up-front investments and limited access to capital resources) and non-market (i.e., perceived costs relating to the lack of information, presence of uncertainty and economic factors related to individual consumers) barriers which influence energy efficiency investments. A range of economic (e.g., taxes and subsidies) and regulatory (e.g., minimum energy performance standards) measures to accelerate the energy transition are also modelled (Fotiou et al 2022).

PRIMES-BuiMo treats consumer heterogeneity by splitting households into several categories, according to the location, age of construction, income class, and type of household (single or multi-story building). Deviating from the single representative consumer hypothesis, the model portrays agent groups with idiosyncratic behaviours and energy preferences. Subjective discount rates are differentiated by income class to reflect differences in the availability of funding opportunities, access to information and uncertainty. Moreover, PRIMES-BuiMo employs discrete choice theory to capture consumer heterogeneity within household groups. In this way, the model represents heterogeneous consumer decisions, respecting the purchase of energy-related equipment and technologies or improvement of building thermal insulation.

EDGE-Buildings is a simulation model that projects global energy demand for buildings, following a top-down approach driven by population, GDP and climate data projections (Levesque et al 2018, Levesque et al 2019). Per capita useful energy demand for end uses is projected into the future based on the observed correlation with macro-economic drivers and, in the case of space heating and cooling, influenced by changing climatic conditions. The observed elasticity between the useful energy demand and macro-economic drivers can be altered to represent behavioural or legislative changes. The building shell efficiency is represented by an aggregated thermal transmittance indicator that improves over time. The effect of increased renovation activity is calibrated to pathways with a known renovation rate. As EDGE-Buildings can only simulate scenarios that follow current trends, for transformative scenarios under stringent climate policy, assessments are performed using REMIND-Buildings. REMIND is a global multi-regional IAM which represents the full economy, with a detailed representation of the energy system (Baumstark et al 2021). REMIND-Buildings is calibrated to baseline scenarios of final energy demand devised with EDGE-Buildings.

Both the PRIMES and EDGE/REMIND modelling suite treat lifestyle transformations in the residential sector, not relating to energy efficiency measures, via ad-hoc exogenous assumptions. These assumptions are applied to model parameters controling the ativity level for different residential end-uses. More specifc details on model characteristics can be found in the Supplementary Information (SI) in section S1.

For this study, both models have undergone enhancements first regarding the breakdown of energy demand to individual residential end-uses, especially for thermal services (i.e. including space and water heating, and space cooling). This is achieved through bottom-up engineering-based functions with an improved distinction between the behavioural and non-behavioural factors driving useful energy demand, that is the amount of energy that should be supplied to final consumers:

• For space heating and cooling, useful energy demand is derived through functions incorporating building-level technical values (e.g., thermal conductivity of building shell/U-value, air infiltration rates), climatic conditions (e.g., heating and cooling degree days—HDDs/CDDs- or the external temperature and length of heating/cooling season), adjusted to include potential behavioural shifts (e.g. changes in internal heating/cooling temperature).
• For water heating, the refined bottom-up functions decompose demand for hot water for personal hygiene purposes to the effect of technical (e.g., showerhead flow rate) and behavioural factors, including the option to reduce the duration of showers.
• In PRIMES-BuiMo, an additional step was taken for calibrating these functions to official energy data from Eurostat (energy balances and split of energy to end uses), thus enabling a more accurate quantification of the effects of behavioural change on useful energy demand.

Second, in order to assess alternative trajectories of floor space growth and their effect on space heating and cooling demand, regional upper limits of floor space growth were set in both models. These caps were defined based on sustainability-focused climate mitigation pathways (Fishman et al 2021, Mastrucci et al 2021) to reflect policies aiming to limit the expansion of building sizes. Third, the representation of electrical demand for appliances and lighting was improved to better incorporate the effect of behavioural changes. In PRIMES-BuiMo this was achieved by integrating in the scenarios energy saving estimates resulting from the choice of eco-mode over standard washing programmes for dishwashers and laundry machines, and from eliminating standby power from appliances. The impact of these measures was reflected implicitly in the scenarios created by EDGE-Buildings.

Scenarios developed for the residential buildings sector of EU-27 countries and the UK extend from a historical baseline year common between the models (2015) to 2050, in 5-year intervals. To enhance comparability across scenarios, the sectoral models use harmonised socio-economic assumptions regarding GDP and population growth. Population estimates are based on Eurostat’s EUROPOP2019 projections, updated in the short-term to reflect the influx of refugees from the Ukraine⁴. GDP projections are based on forecasts developed by the Directorate General for Economic and Financial Affairs (DG ECFIN 2021, 2022). The scenario analysis for the residential sector considers two distinct climate targets based on the current EU climate policy landscape. On the one hand, the ‘Baseline’ scenario (existing policy framework) considers only currently implemented climate and energy policies in EU Member States, such as those presented in their National Energy and Climate Plans (NECPs) and long-term renovation strategies. It follows the general energy, economic and emission trends presented in the latest EU Reference scenario of the European Commission (Capros et al 2021).

On the other hand, the ‘Decarbonisation’ scenario assumes much more ambitious EU climate policies in 2030, including the 55% GHG emission reduction target relative to 1990, and stringent energy efficiency and renewable energy targets (as those presented in the Fit for 55 package (EC 2021)). Furthermore, this scenario leads to climate neutrality by 2050 in line with the goals of the EU’s Green Deal (EC 2019). The decarbonisation scenarios assume that various regulatory (e.g. building energy codes, carbon tax) and non-regulatory policies (e.g. subsidies, information and education policies) facilitate the green transition supporting the up-take of clean energy technologies and corresponding investment. However, both the basic Baseline and Decarbonisation policy context overlook the potential energy and emission impacts from more ambitious lifestyle-led demand transitions.

On top of the two climate policy scenarios, three distinct assumptions have been imposed concerning future lifestyle changes: (a) a scenario that assumes no shift in consumer behaviours and a continuation of current trends in energy consumption habits, (b) a Low Ambition lifestyle change scenario variant assuming that a small share of consumers adopts lifestyle changes (gradually until 2050) to reduce their energy consumption and carbon footprint, and (3) a High Ambition lifestyle change scenario variant assuming that a larger share of consumers adopts lifestyle changes of similar or increased intensity compared to the low-ambition case. Combining the two levels of climate policy ambition (‘Baseline’ and ‘Decarbonisation’) with the three levers of lifestyle change (‘No’, ‘Low’ and ‘High’), results in six possible scenario configurations simulated with the two models (labelled as ‘Base’, ‘Base_LC’, ‘Base_LC_High’, ‘Decarb’, ‘Decarb_LC’, and ‘Decarb_LC_High’).

2.2.1. Selection of lifestyle changes

2.2.2. Quantification of lifestyle changes

Useful energy demand for residential thermal uses in the EU-27 & UK region increases slightly over time in the Baseline case, as shown in figure 1. This is driven mainly by the growing floor space (see SI/section S2 for energy service level projections), while improvements in the thermal properties of building envelopes have a more limited effect. Integration of lifestyle changes leads to significant reductions in useful energy demand relative to the Baseline case, with the effect reaching a maximum level in 2050 when more households adopt energy-saving lifestyles. Relative to the Baseline case, useful energy demand in 2050 is projected to decline by 17% in the Low Ambition (Base_LC) and 29–34% in the High Ambition (Base_LC_High) scenario, with the two models showing similar trends. Due to its magnitude, the largest contributor to the projected decrease is space heating (which is reduced via thermostat set-point adjustrements, declining floor space and increased renovation), with a 80–90% contribution share (see SI/section S2 for detailed figures)¹¹ .

Zoom In Zoom Out Reset image size

In the decarbonisation scenarios family generated with PRIMES-BuiMo, useful energy demand in 2050 is reduced by 13–42% relative to the Baseline case, similarly driven by the declining space heating needs. The scenario achieving the strongest reduction in useful energy demand is as expected the Decarb_LC_High, which assumes high annual renovation rates and depth, and the most intense lifestyle changes combined with ambitious policies towards net-zero emissions. Decarbonisation scenarios from REMIND are not shown as the model does not report end-uses at sufficient granularity.

Based on models’ projections, final energy use in the residential sector of the EU-27&UK region in the Baseline case changes from 12 EJ yr⁻¹ in 2015 to 10–12 EJ yr⁻¹ in 2030 and decreases to 9 EJ yr⁻¹ in 2050, representing a ∼20% reduction from current levels (see SI/section S2 for a breakdown to fuel and end-use shares). PRIMES-BuiMo shows that stronger energy savings are realised earlier compared to REMIND-Buildings. This is the result of ambitious energy efficiency and renewable energy policy targets in 2030 explicitly represented in the modelling framework of PRIMES. REMIND-Buildings on the other hand projects higher mid-term baseline demand. There is however good agreement of the relative reduction from LC between the models, as shown in figure 2.

Zoom In Zoom Out Reset image size

The impact of lifestyle changes on energy consumption in the baseline scenarios is significant, especially in the long-term: in the low and high-ambition LC cases, residential energy use is respectively shown to decrease by 15–16% and 27–30% in 2050 relative to the Base scenario with no lifestyle change. The two models also agree about the relative decrease in final energy by 2050 under the decarbonisation scenario, projecting a 25–26% reduction with respect to baseline levels. Furthermore, the energy savings from behavioural shifts in the decarbonisation context are slightly smaller compared to baseline scenarios, as the reduction in final energy in 2050 relative to the basic Decarb case is only 14% and 26–27%, respectively for the low and high-ambition scenario variant.

Moreover, the two models agree about the additional benefits from raising the ambition level of lifestyle changes in the residential sector from ‘low’ to ‘high’. In the current policy context, more widespread adoption of low-carbon behaviours increases the energy savings by 12–14 p.p. in 2050; while the corresponding savings in the decarbonisation context are relatively smaller at 9–10 p.p., showing the diminishing returns from an increased shift to low-carbon behaviours in a progressively decarbonised system.

Baseline CO₂ emissions in the EU-27&UK residential sector (figure 3), are projected to decrease in PRIMES-BuiMo by 25% and 51% in 2030 and 2050 respectively, relative to 2015 levels. Conversely, REMIND-Buildings projects a somewhat lower decrease in Base scenario emissions , which amounts to 18% in 2030 and 44% in 2050 below 2015 levels. The less pronounced reduction in baseline emissions in 2030 can be explained by the smaller change in total energy demand from current levels; while in 2050 this can be traced back to the higher share of remaining coal and liquids compared to the final energy demand of PRIMES.

Zoom In Zoom Out Reset image size

Similarly in the decarbonisation scenario, PRIMES-BuiMo projects CO₂ emissions from households to reduce in 2030 by 52%, while the corresponding reduction in REMIND-Buildings is lower at 36%. Nevertheless, emissions in both models virtually reach zero by 2050, as strong decarbonisation policies drive the significant electrification of the sector and massive displacement of fossil fuels combined with the uptake of zero-carbon options, such as heat pumps and distric heating.

Emission reductions increase in the lifestyle change variants with emissions in the low(high)-ambition LC baseline scenario dropping further by 4 p.p. (10 p.p.) in 2030 and 7–8 p.p. (16–17 p.p.) in 2050 compared to the core baseline case. In agreement with final energy demand results, the impact of lifestyle change on emissions is less significant in a decarbonisation setting, as the maximum additional emission reduction in the two models is 6 p.p. in 2030 and almost negligible by 2050 as net zero emissions are already achieved in the region. CO₂ emissions in both models display greater variation between the core decarbonisation and baseline case, compared to the variation observed between the core scenario and its lifestyle change variants.

This section compares the costs incurred by consumers in the residential sector under the various scenarios, based on PRIMES-BuiMo results (figure 4). The cost indicators selected are annual fuel expenses and capital costs, which comprises payments for direct energy efficiency measures (renovation and new building constructions) and for new energy equipment (e.g. heating equipment, electric and electronic appliances etc).

Zoom In Zoom Out Reset image size

Fuel purchasing expenses reduce in the decarbonisation scenario relative to the baseline, around 1% when cumulated over the period 2021–2050. Factors like the introduction of carbon price in non-ETS sectors from 2030 onwards which increases fossil fuel prices, mitigate the reduction in fuel expenses due to lower energy demand. Lifestyle changes reduce even more the cumulative fuel expenses over the period 2021–2050 by 6–7% and 13–14% in the low- and high- ambition LC scenarios respectively, compared to the Base and Decarb scenarios without lifestyle changes. In this way, lifestyle changes are effective in reducing energy expenditure¹².

Compared to the baseline case, capital costs in the decarbonisation scenario increase by about 20% over the 2021–2050 period. This is influenced by policies enabling additional capital investments in energy efficiency, based on a higher rate and depth of buildings’ renovation and the accelerated uptake of capital-intensive low-carbon technologies, including heat pumps, as well as energy efficient equipment and appliances. As a result, capital costs in 2050 for the Base and Decarb case are respectively 80% and 120% higher than the corresponding 2020 level.

In the lifestyle change variants, changes in capital costs are driven by two opposite effects, namely the increased spending on renovation measures assumed in the LC variants counterbalanced by the reduced investments in energy-using equipment as thermal needs are lower (lower household space, lower temperature set-point). As the latter is a larger cost element, cumulative capital costs under the LC variants are lower than the cases without lifestyle changes both in the baseline and decarbonisation context. Over the 2021–2050 period, capital costs reduce in the LC (LC_High) case by 1% (3%) in the baseline, and 1% (4%) in the decarbonisation scenario, relative to the no lifestyle change cases. Cumulative total energy-related costs (comprising capital costs and energy purchases) are therefore reduced by 4% and 8% under the low and high-ambition lifestyle change case respectively, for both the baseline and decarbonisation scenario. The share of cumulative operating to total costs drops from 49% in the baseline scenario to 41% in the Decarb_LC_High case while the share of capital costs increases due to the investment in buildings’ renovation and the purchase of more capital-intensive (and energy efficient) technologies and appliances.

This section elaborates on the results of the decomposition analysis of energy savings achieved in the decarbonisation scenarios incorporating individual high-ambition behavioural changes compared to the Decarb case without lifestyle changes. Amongst the assessed measures/lifestyle changes, the one having the strongest impact on final energy use for both PRIMES and REMIND is the reduction of average household space, with cumulative savings of 12–14 EJ in the 2021–2050 period relative to the Decarb case (figure 5). The second most effective measure is thermostat set-point adjustments according to PRIMES, with cumulative energy savings of around 7 EJ. Conversely in REMIND, increased renovation has the second largest effect (11 EJ savings), followed by thermostat adjustements (8 EJ savings).

Zoom In Zoom Out Reset image size

In PRIMES, intensification of renovation measures has a much less pronounced contribution to overall enegy savings, as in the basic Decarb scenario the modelled energy efficiency policies already lead to an increased rate and depth of renovation leaving a smaller room for further energy demand reductions in the LC cases. The decreasing effect on useful energy demand via increased building renovation under the basic Decarb case is represented more coarsely in REMIND such that the effect of an additional increase of the renovation rate has to be calibrated to results from models including explicit renovation activity. Appliance use has a moderate impact on energy savings in both models (∼6EJ savings), while hot water conservation measures has the lowest impact (∼3EJ savings).

Adding together the energy savings resulting from individual behavioural measures leads to a level of total savings that is higher than the ones achieved under the Decarb_LC_High scenario (plotted in figure 5). This could be explained by the potential (but moderate) trade-offs between lifestyle changes targeting demand for space heating and cooling, namely the interplay between building renovation choices and exogenous assumptions about changing thermostat temperature set-points and dwelling size. In the scenarios where renovation rates are unconstrained, the introduction of individual exogenous measures on top of the Decarb scenario leads to energy savings according to the mitigation potential (in % terms) of each action. Enforcing higher renovation rates and depth in the Decarb_LC_High case results in a lower useful energy demand for space heating relative to the Decarb case, thereby limiting the room for additional savings from other behavioural measures.

The critical role of behavioural shifts in shaping future energy consumption patterns in the residential sector is emphasised in both models’ results, especially in the long-term when the uptake of lifestyle changes is maximised. This is particularly evident under the baseline scenario’s specifications where lifestyle shifts result in substantial reductions in energy demand and CO₂ emissions from the residential sector. This showcases the significant contribution of radical demand-side transitions and lifestyle changes in achieving energy efficiency and CO₂ reduction goals.

Savings from lifestyle changes in the baseline context for the low-ambition scenario variant are broadly consistent but on the high end of the range of estimates from previous similar studies. EU-level residential emissions in the Low Ambition scenario decrease by 18–20% from their baseline level in 2050, which compares well with the 16% and 12% emissions reduction estimates in the global and European lifestyle case of van Sluisveld et al (2016) and Costa et al (2021)¹³ respectively. The impact on CO₂ emissions in 2050 is somewhat larger compared to the 9% reduction esimate derived from the ‘Pocket Lifestyles’ scenario in van den Berg et al (2024). The smaller impact can be explained by the lower saturation rate of various lifestyle changes in 2050 compared to the 100% rate adopted in our study and the non inclusion of building renovation in the list of behavioural measures. Furthermore, comparison with previous studies demonstrates the optimistic nature of our High Ambition lifestyle change case, which requires a radical up-scaling of behaviours and social norms in the residential sector. In the baseline high-ambition scenario variant, our models project CO₂ emissions to reduce by 61–68% between 2015 and 2050. This is still within the 40–70% range of the technical potential for emissions reduction in 2050 through demand-side actions in all end-use sectors, as estimated by the Intergovernmental Panel on Climate Change (IPCC) (Creutzig et al 2022b).

Lifestyle change impacts on energy efficiency and decarbonisation goals are smaller under a climate-neutrality policy framework (in agreement with van Sluisveld et al (2016)). This suggests that the effectiveness of lifestyle changes depends also on broader climate action. While most of the emission reduction effort is attributed to the overall decarbonisation measures, lifestyle changes could still have a complementary role and support medium-term goals in case of weaker implementation of decarbonisation policies. Results suggest that lifestyle changes, such as reducing average dwelling size and thermostat set-point adjustments, can act as a substantial complementary driver to climate policies in a decarbonisation setting to accelerate the EU’s transition.

Moving to economic considerations, scenario results highlight that lifestyle changes can lower energy expenditure for residential consumers as they lead to reduced fuel expenses, irrespective of the policy context. Moreover, under the decarbonisation context where ambitious policies drive additional investments in energy efficiency, lifestyle changes can be effective in mitigating the increases in capital expenditure incurred by consumers. Overall, this provides evidence that the role of lifestyle changes in the policy agenda could be two-fold, one for the contribution to decarbonisation goals and another for reducing energy costs for consumers. Lower expenditures due to lifestyle changes could potentially aid in the redution of the risk of energy poverty for low-income Eurpean households, and support broader SDGs with clear co-benefits in terms of reduced air pollution, improved health and enhanced sustainability.

The overall findings of this study underline the need for designing mitigation strategies tailored to the residential sector’s challenges and potentials, with a strong representation of radical demand-side transitions and lifestyle changes. However, caution is needed in the promotion of lifestyle changes to the extent to which they do not compromise well-being and hinger progress towards achieving decent living standards (Millward-Hopkins et al 2020). Currently, these elements lack significant representation in most of the current EU-wide policies and national climate strategies (Salem et al 2021). Some exceptions exist for EU countries, such as in the long-term strategies of Austria and Portugal (Federal Ministry Republic of Austria 2019, Republica Portuguesa 2019), which incorporate some behavioural changes for buildings, however mostly focusing on building renovation measures.

The robust integration of diverse lifestyle changes and behavioural shifts into leading IAMs and ESMs, an aspect which is often overlooked in traditional modelling approaches, allows for a better understanding of their potential contribution to a low-carbon transition. This study has focused on modelling lifestyle changes and their impacts on the EU-27&UK residential sector, through improving two leading sectoral energy models. The performed analysis informs policy makers about the potential contribution and added value of environmentally sustainable lifestyle changes of increased intensity and adoptability for the net-zero transition goals.

Despite the rich representation of lifestyle change impacts in the modelled scenarios, our approach has some limitations. First, with the exception of building renovation in PRIMES-BuiMo, lifestyle changes are represented in our models in an exogenous manner; a common feature of ESM/IAM-based approaches (Andreou et al 2022, Krumm et al 2022). As a result, a comprehensive analysis of the implementation barriers, enabling factors and policy costs incurred by lifestyle transitions was not performed. Second, due to the unavailability of information on heterogeneous energy preferences, the scenarios have applied a uniform change in lifestyles across household groups. Not comprehensively representing actor heterogeneity does not allow studying the varying responses of consumer types to policies targeting lifestyle changes (Keppo et al 2021). Third, despite our effort to validate our scenario assumptions about lifestyle changes through reviews and expert insights, the real extent to which such demand-side transitions can happen has remained outside the scope of this study.

Future work could focus in endogenising lifestyles in large-scale IAM/ESM models and representing the heterogeneity of energy preferences between diverse consumer groups. For buildings, research could focus on the drivers and more importantly the psychological and infrastructural barriers of behavioural shifts with a high mitigation potential, such as reducing average living space size (Huebner and Shipworth 2017) or changing the thermostat set-point. A promising approach is the linking of IAMs/ESMs with tools grounded on rich empirical data on energy use behaviours, such as statistical lifestyle modules (Pettifor et al 2024) and agent-based models (Niamir et al 2020, Niamir et al 2024). A linkage with an agent-based model can in particular aid in capturing complex social dynamics, such as lifestyle changes, as they have rich representation of heterogeneous actor characteristics and can simulate social interaction and learning effectively.

Moreover, the scope could be broadened to include other domains with a large carbon footprint and high sensitivity to behavioural change such as mobility, diet and consumer goods (Grubler et al 2018, Ivanova et al 2020). Systemic analysis -using full-system IAMs- can also provide a more holistic representation of lifestyle change impacts on the energy system, including energy supply, but also in land-uses and in the broader economy and society. Testing the generalisation of findings could involve expanding the geographical scope to encompass other regions, including other developed or less-developed economies (van den Berg et al 2021).