*2.3. Limitations*

Our analysis was affected by limitations regarding the representativeness, harmonization and measurement errors of the HBSs. A detailed account of these limitations [26] and their implications for the carbon and energy footprint calculations can be found elsewhere [3,31]. There may be higher sampling error and inflated variation associated with infrequent purchases [26], for instance second homes [40], personal vehicles, flights or furniture, and their associated environmental impacts.

There are some limitations regarding the environmental impact assessment. EXIOBASE offers details of 200 products and services across 44 countries and five rest-of-the-world regions, and can thus only distinguish the country-level carbon and energy intensities of largely heterogeneous product groups. Particularly in the context of household dynamics, the product detail was insufficient to distinguish between consumption of items that are more likely to be shared within and between households (e.g., use of shared appliances vs. individual equipment). Difficulties in allocating land use change emissions to specific economic activities have been previously recognized [41,42].

Some products and services may also be purchased directly by households in some countries but are provided through governmental spending in others. Focusing solely on household expenditure may thus result in substantial variation in terms of spending on health, social work, education and transport services, disregarding impacts associated with public provision, which affects comparative analysis [43]. As a result, our analysis may not capture well country differences in the between-household sharing opportunities through the provision of public infrastructure.

Furthermore, as household carbon and energy footprints are based on monetary expenditure, there are limitations due to potential price differences within products. Therefore, we likely overestimated the environmental impact of expensive products (wealthier individuals) and underestimated the impact of cheap products (and less wealthy individuals) [44]. In addition, we could not examine the effect of "green consumerism" [16] on carbon and energy intensities, e.g., buying a fuel-efficient car, opting for a green energy provider or a more expensive but energy efficient dwelling. Larger households may also be more likely to purchase items in bulk and, thus, pay lower prices per item. Prior work discusses the limitations associated with the monetary-based approach [2,31,44].

The HBS uses household size or type in the stratification criteria for most countries in order to make the survey sampling more accurate [26,32]. Yet, there may be an under-representation of less common household types such as intentional communities (e.g., eco-villages, co-housing). All collective households such as elderly homes, boarding schools and others, where individual spending cannot be distinguished from collective spending, have been excluded from the HBSs [26].

Furthermore, the population density variable and interaction effect are based on the LAU2 classification and as such it can only capture potential consumption and footprint differences between cities, towns and suburbs and rural areas. We cannot capture differences in the between-household sharing potential and opportunities on dwelling-, close community- or neighborhood levels.

#### **3. Results**

#### *3.1. Descriptive Statistics and Bi-Variate Regressions*

## 3.1.1. Household Size, Carbon and Energy Footprints

In per capita terms, one-person households have the highest average carbon and energy footprints in the EU at 9.2tCO2eq/cap and 0.14 TJ/cap per year (Figure 1). They contribute 17-18% of the EU's total carbon and energy footprints, but constitute less than 13% of the EU population. Two-person households are most numerous with 27% of the EU population. They also contribute the largest share of the EU's total carbon and energy footprints with 31-32%. The EU per capita average of carbon and energy footprints for two-person households amounts to 8.4 tCO2eq/cap and 0.12 TJ/cap, respectively. The largest households (>4 persons) contribute about 9-10% to total EU emissions and energy use and represent 14% of the population. They have the lowest average carbon and energy footprints of 4.6 tCO2eq/cap and 0.07 TJ/cap, respectively (Figure 1).

(**b**)

Figure 2 depicts the relationship between average per capita carbon and energy footprints and average household sizes across EU countries. The figure shows a negative trend across countries, suggesting a substantial overlap between countries with high average carbon and energy footprints and relatively low household sizes. The average household size in EU amounts to 2.4, varying between

2.2 and 2.9 across countries. The supplementary material (SM2) provides more detail about the distribution of carbon and energy footprints, and household sizes across EU countries.

**Figure 2.** Association between average household size and average per capita carbon (**a**) and energy (**b**) footprints in the EU. The carbon footprints are measured in tCO2eq/cap and energy footprints in TJ/cap. Household weights provided by the HBS have been applied.

The countries with the highest per capita carbon and energy footprints in the EU include Luxembourg, Greece (previously found to have one of the highest carbon footprints in the EU [4,43], with a large vessel fleet in relation to its size, requiring a high use of fuel from bunkers [45]), Ireland, Finland, United Kingdom, Belgium, Germany and Denmark, with carbon footprints between 14.1 and 9.1 tCO2eq/cap, and energy footprints between 0.2 and 0.13 TJ/cap (Figure 2, SM2). These are also the

countries with some of the lowest household sizes: Germany (2.0), Denmark and Finland (2.1), Belgium and the United Kingdom (2.3). Finland and Denmark have the highest share of one-person households from the total number of households at 40%, followed by Germany at 39%. These observations broadly agree with the Eurostat statistics on household sizes (SM3).

The countries with the lowest per capita carbon and energy footprints include Romania, Croatia, Hungary, Latvia, Poland, Bulgaria, Spain, Portugal and Slovakia, with carbon footprints between 3.6 and 6.2 tCO2eq/cap, and energy footprints between 0.05 and 0.09 TJ/cap. The countries with the highest household sizes include Romania and Cyprus (2.9), Slovakia, Malta, Poland and Croatia (2.8), and Spain (2.7). Romania, Malta and Spain have the lowest share of one-person households (19%) from the total number of households.

Figure 3 shows average per capita carbon and energy footprints per household size across EU countries. It confirms a drop in the environmental per capita impact with rising household size within EU countries. While the slopes vary in steepness, we consistently confirm this trend for all EU countries. For example, the average carbon footprint of Luxembourg ranges from 18.8 to 7.4 tCO2eq/cap for one-person and six-or-more persons households, respectively. Similarly, the per capita energy footprint of the average one-person household in Luxembourg is 0.27 TJ/cap, while that of an average six-or-more persons household amounts to 0.11 TJ/cap. According to Figure 3, the spread of the average carbon and energy footprints across EU countries is much larger for smaller household sizes compared to larger household sizes. Additionally, the absolute change in environmental impacts with the addition of one more household member is decreasing in magnitude with the rising household size.
