Enlarged carbon footprint inequality considering household time use pattern

We apply the EEIO to estimate the household carbon emission intensity. Originating from the economic input-output (IO) model, EEIO is considered an extension of IO by connecting environmental impact to monetary flows. The basic equation of an economy is described as:

$$\begin{align}{X}= \left({I} - {A} \right)^{ - 1}{F}\end{align} \tag{ 1 }$$

where ${X}$ is the vector of output, ${I}$ is the identity matrix, ${A}$ is the input coefficient matrix, and ${F}$ is the vector of final demand, consisting of consumption, investment, and export. To note, due to the lack of sufficient data on imported products, the Leontief Matrix is transformed to ${\left( {{I} - \left( {{I} - {\overline{M}}} \right)} \right)^{ - 1}}$ type, which indicates only the domestic products is considered. Here, ${M}$ indicates the imports column matrix, which is calculated by equation (2) to represent the import portion of the direct requirement coefficient:

$$\begin{equation}{M_i} = \frac{{{m_i}}}{{{x_i} - {e_i} + {m_i}}}\end{equation} \tag{ 2 }$$

where ${e_i}$ and ${m_i}$ indicate the export and import of item $i$, therefore, $\left( {{x_i} - {e_i} + {m_i}} \right)$ is known as the total supply to the domestic market. Then, using the diagonal matrix $\overline {M}$ to express the import coefficient, the carbon footprint intensity of domestic producer goods is shown as

$$\begin{equation}{eA} - {e}\overline {M A} + {d} = {e}\end{equation} \tag{ 3 }$$

here, ${d}$ and ${e}$ indicate the direct and indirect carbon footprint intensity of per unit production, respectively. And solving equation (3), the embodied carbon footprint intensity is shown as:

$$\begin{equation}{e} = {d}{\left({I} - {A}\left( {{I} - \overline{M}} \right)\right)^{ - 1}}.\end{equation} \tag{ 4 }$$

Connected with household monetary consumption, the embodied carbon footprint generated by household consumption can be expressed as:

$$\begin{equation}{E^m} = {d^m}{\left( {I} - {A}\left( {{I} - \overline{M}} \right) \right)^{ - 1}}*Ex{p^m}\end{equation} \tag{ 5 }$$

where ${E^m}$ indicates the household embodied emission by consumption item $m$; $Ex{p^{im}}$ refers to monetary consumption on consumption item $m$;${\text{ }}{d^m}$ is the direct emission intensity of consumption item $m$. In this study, direct and indirect carbon emission intensity are driven from Embodied Energy and Emission Intensity Data for Japan Using Input-Output Tables by National Institute of Environmental Studies in 2015 (Nansai et al 2002, 2009, 2012, 2014). Household monetary inventory is driven by the Family Income and Expenditure Survey (FIES) of 2015, a dataset compiled by the Statistics Bureau of Japan's Ministry of Internal Affairs and Communications with the purpose of understanding the daily life of residents in Japan. Within this survey, a total of 8471 households are meticulously compiled, with income-group distribution proportions as follows: 11% (Low), 19% (Low-mid), 22% (Mid), 23% (Mid-high), and 24% (High).

In view of the complexity of household lifestyles, a residential consumption inventory provides a window for investigating how individual characteristics affect final environmental consequences. In this study, we apply the Survey on Time Use and Leisure Activities (STULA), entailing 18 types of residential daily activities by min per day to investigate their lifestyles. The original STULA data is classified into ten income groups and reclassified into five groups to match the FIES data of 2015. Items of FIES consist of 504 detailed categories covering from durable goods to semi-durable goods. However, household energy consumption on electricity, gas, kerosene, and sewage is difficult to attribute to certain activities. This issue also be mentioned as a limitation of previous studies (De Lauretis et al 2017, Wiedenhofer et al 2018). Therefore, the following allocation is conducted to distinguish the energy-related emissions considering Japanese residents' living habits.

All the household consumed items are classified into different activities by considering the using purpose. Semi-durable goods such as electricity, and natural gas are classified as follows. First, electricity consumption is multi-purpose while using. The percentage of household electricity consumption for a different purpose is derived from the Handbook of Japan's & World Energy & Economic Statistics (EDMC 2019) released by the Institute of Energy Economics of Japan. Shown as in table 1, lighting, appliance, hot water supply, and heating are found as major terminal purposes. The hot water supply here is identified as showing activity considering its major purpose. Electricity consumed by kitchen use is classified as cooking. Cooling, heating, lighting, and other appliance use are allocated to all the items occurring at home. Not just electricity, other accommodation-related items that cannot be classified by daily activities are allocated to in-home activities. Natural gas, which includes both city gas and LNG, is allocated to showing and cooking activities. The allocation ratio is 76% (Hot water supply) and 24% (Kitchen), which is referred from the Agency for Natural Resources and Energy and previous research (SUZUKI and Tanabe 2016, Agency for Natural Resources and Energy 2020). Besides, in 2015, 18% of household water usage was for cooking, and other purposes including toilet, showing, and washing ⁴ are all allocated to personal stuff (In-home activities). Second, durable goods such as furniture, interior decorations, curtains, floor covering, and rents (for dwelling and land) are considered necessary to conduct in-home activities. Therefore, although that consumption can be directly connected to certain behaviors, the correlated consumption is allocated to all the in-home activities ⁵ . The aforementioned activities are classified as activities in-home (relatively higher likelihood to occur at home) and out-of-home (relatively higher likelihood to occur outside the home) to allocate their carbon footprints reasonably (see figures 1 and S1, supplementary data).

Zoom In Zoom Out Reset image size

In this research endeavor, the disparity in carbon footprints across income groups is quantified through the utilization of the Gini index. The Gini index is a statistical metric that assesses the distribution of a certain parameter, often employed to gauge disparities within populations or income brackets. For this study, the Gini index is applied to the carbon footprints specific to each income group. The Gini index of income-group-specific carbon footprints is calculated based on:

$$\begin{equation}\alpha = \frac{1}{{{n^2}}}\mathop \sum \limits_{i = 1}^n \mathop \sum \limits_{j = 1}^n \left| {{x_i} - {x_j}} \right|{\text{ }}\end{equation} \tag{ 6 }$$

$$\begin{equation}\beta = \frac{\alpha }{\mu } = \frac{1}{{{n^2}\mu }}\mathop \sum \limits_{i = 1}^n \mathop \sum \limits_{j = 1}^n \left| {{x_i} - {x_j}} \right|\end{equation} \tag{ 7 }$$

$$\begin{equation}{\text{Gini}} = \frac{\beta }{2} = \frac{1}{{2{n^2}\mu }}\sum\limits_{i = 1}^n \sum\limits_{j = 1}^n \left| {{x_i} - {x_j}} \right|\end{equation} \tag{ 8 }$$

where ${x_n}$ indicates the carbon footprints among income groups ($n = 5$), $\mu = \left( {{x_1} + {x_2} + {x_3} + \ldots {x_n}} \right)/n$, and $\alpha$ and $\beta$ refer to absolute mean difference and relative mean difference, respectively. Households are divided into quantile groups based on the household yearly income, including Low (below 2440 000 JPY), Low-mid (2440 000–3590 000 JPY), Mid (3590 000–5000 000 JPY), Mid-high (5000 000–7310 000 JPY) and High (Above 7310 000 JPY).

To understand the extent to which lifestyle changes can contribute to global climate warming mitigation efforts, this study leveraged insights from prior research (Akenji et al 2019) on the aggregated reduction impacts of low-carbon options in Japan under the temperature goals of the Paris Agreement. To be more specific, a literature review on low-carbon lifestyle options and their associated emission reductions has been conducted, which incorporated nearly 50 lifestyle changes for this assessment. Table S9 summarizes the estimated reduction impacts of low-carbon lifestyle choices at various adoption rates, ranging from 15% to 100%. Using the data gathered, the potential reduction in carbon footprints could be estimated by altering the intensity and/or quantity of relevant components for each option. Notably, consistent adoption rates for the introduced low-carbon lifestyle changes, encompass 15%, 30%, 65%, 75%, and 100% adoption. Although the same adoption rates are assumed for lifestyle changes, the magnitude of carbon footprint reductions varied across different categories of daily activities. Moreover, emission reductions arising from a cleaner electricity grid, characterized by progressively lower grid emission factors, are also taken into consideration. The future grid emission factors are estimated based on the model by Sugiyama (Sugiyama et al 2021), which are estimated to be 0.1061, 0.09, 0.071, 0.046, 0.026, 0.007tCO₂/GJ in 2025, 2030, 2035, 2040, 2045, and 2050. Therefore, this study explores decarbonization pathways for households of varying income levels in future scenarios, based on assumptions related to the adoption rates of low-carbon lifestyle choices and evolving grid intensity.

Previous studies have demonstrated a positive correlation between household carbon footprints and income (Ravallion et al 2000, Hubacek et al 2017b, Liu et al 2019). However, in Japan, this relationship exhibits a U-shaped curve, indicating a nuanced pattern of emissions across income groups, as depicted in figure 2(a). The results indicate that the per capita carbon footprint varies among different income groups, with 'Dining' and 'Personal stuff' contributing most regarding the absolute volume. The specific carbon footprint information for each activity can be found in table S1. Importantly, we need to point out that the activities evaluated encapsulate overarching household demands, not mere individual consumables. For instance, 'Dining' comprises all constituents of eating, ranging from food items, and beverages, to related energy consumption and waste management.

Zoom In Zoom Out Reset image size

Figure 2(b) presents a waterfall chart of the activities contributing to carbon footprint gaps, showing that 'Personal stuff', 'Dining', and 'Rest and relaxation' significantly drive the disparity between low-income and middle-income groups, with table S2 provides the estimation results of specific contributions of various daily activities. Conversely, 'Transport and moving' is identified as the main activity that narrows the gap between these two groups. For middle-income versus high-income households, the increment in footprints is largely due to personal mobility and education expenditures.

Furthermore, incorporating the data from tables S3 and S4, we assess the carbon footprint variations associated with household consumption types, shedding light on the role of basic living demand and other auxiliary demands. Figure 2(c) shows the carbon footprint variation from low to high-income groups by aggregating household consumptions into two types of demands, namely the basic living demand (e.g. sleep, personal stuff) and other demands (e.g. Schoolwork, child care). Interestingly, the carbon footprints related to basic living demand are equally distributed among income groups. On the contrary, figure 2(d) shows that auxiliary demands, such as leisure and social activities, are where disparities emerge. This suggests that carbon footprint inequality stems more from discretionary spending than from essential needs.

Delving into specifics, figure 3(a) identifies major contributors within the 'Dining' category, which include Energy use (20%), Cooked food (13.5%), Grain (8.7%), Vegetables (8.6%), and Eating out (8.2%). The high-income group's predilection for meat and frequent dining out, accentuates its footprint. Meanwhile, lower income groups manifest heightened carbon footprints from energy use and staple foods, underscoring a dietary shift with income escalation. In terms of transport, figure 3(b) shows that the high-income and middle-high-income groups have the highest carbon footprint from gasoline use, automobile purchases, and the utilization of public transport, reflecting increased mobility with higher income. Conversely, the low-income group demonstrates the lowest carbon footprint in these categories due to reduced utilization of both personal and public transportation. Notably, 'Personal stuff' is the primary factor in enlarging the carbon footprint gap. In figure 3(c), the high-income group shows the highest carbon footprint in clothing, footwear, beauty products, and services than other groups. In contrast, the low-income group has the highest expenditure on hot water supply in terms of both electricity and gas use, suggesting higher utility costs or less energy-efficient appliances. As income increases, there's a noticeable decrease in these expenses, likely reflecting more energy-efficient choices or lifestyle preferences for out-of-home services.

Zoom In Zoom Out Reset image size

In this segment of the study, an emphasis is placed on examining the role of time use in exacerbating carbon footprint inequalities. As Smetschka et al (2019) (Smetschka et al 2019) and Druckman et al (2012) (Druckman et al 2012) have mentioned, how people arrange their time matters in their associated carbon footprints embodied in consumption, casting light on the genesis of inequality. In other words, the time dimension can provide clues for dealing with the unequal carbon footprint distribution since time is a limited resource akin to energy.

Consequently, we measured the quantile income groups' carbon footprint intensity from the time dimension and mapped it with 18 types of daily activities, the results of which are displayed in figure 4 (estimation results can be referred to in table S5). Different from previous energy intensities such as CO₂ per unit of production or monetary scale, we use the income-based per-minute consumption to quantify the lifestyle formation and its inner correlation and try to discover the concealed inequality. Moreover, before jumping into the result, we should note that Japan is one of the most developed countries as well as one the most equal countries economically. According to the Ministry of Health, Labor and Welfare of Japan, the Gini index is found to stay at 0.31 after income redistribution, tax, and performance in kind (Providing public assistance, services or any other means other than by providing performance in money.) in 2014 and 2017 (MHLF 2017). Japan shows a relatively low level (around 0.3) in residential income Gini Index after social income retribution (MHLF 2017) and slightly unbalanced regional or age-based carbon footprints (Shigetomi et al 2014, Long et al 2019).

Zoom In Zoom Out Reset image size

Activities with a Gini index greater than 0.35, as shown in figure 4(a), are mainly related to schoolwork, reflecting disparities in education-related consumption. Wealthier households tend to choose private schools, contributing to higher carbon footprints due to factors like extracurricular activities. Luxury goods in the 'Personal stuff' category also exhibit inequality among income groups. However, it is important to note that luxury consumption is not the primary driver of environmental inequality; education choices play a more significant role. Carbon footprint disparities related to schoolwork are amplified among income groups. Figures 4(b) and (c) delve deeper into schoolwork, classified by education stages and types. Apart from carbon footprints from preschool education and sanitation, the Gini index remains above 0.5 from elementary education through college and university. Specifically, elementary education, which includes both public and private school fees, sees the Gini index peak at 0.57. This signifies that the carbon footprint inequality begins as early as the elementary phase and continues to escalate through university. To note, 'Other education' represents Japanese Senshu (Specialization) University, and 'shadow education' refers to fee-based supplementary education outside of traditional school.

Figure 4(c) further dissects the carbon footprints from education by type. Interestingly, alongside private schooling, remedial school fees also contribute to the widening inequality, as wealthier households tend to spend more on education-related services such as entrance exam preparation for high schools and universities. This trend is common in various Asian countries and regions like mainland China, Korea, Japan, etc. where competition in education has spilled over from traditional schools to other educational institutions requiring additional payment (Yamamoto and Brinton 2010, Byun et al 2012, Cheng 2017). Consequently, correlated consumption has been found to enlarge social environmental inequality from the consumption perspective. However, it should be acknowledged that education, while contributing to disparities in carbon footprints within current circumstances, also plays a pivotal role in cultivating environmental awareness and sustainable consumption practices among the next generation (Zsóka et al 2013, Ardoin et al 2020).

In addition to using the Gini index to measure the degree of inequality, the carbon intensities in terms of time use are also presented, revealing trends in carbon footprints per unit of time, as presented in figure 4(d). Here, 12 of 18 household activities are observed with inequality enlargement by time unit intensity, which can generally demonstrate a wide social inequality is happening in Japan considering environmental impact by involving time dimension. Notably, with respect to 'Schoolwork' and 'Child care', the highest-income group substantially exceeds the carbon footprint of the lowest-income group, and the utilization of the gCO₂-per capita/min metric highlights the pivotal role of time in evaluating emission intensity. Consequently, this approach not only elucidates which activities yield higher carbon footprints but also identifies those with greater emission intensity, denoting higher carbon footprints per unit of time.

The per-time-unit carbon footprints for 18 activities differentiated by income are presented in figure S2, supplementary data. Per-time-unit carbon footprints vary across income groups (detailed data on per-time-unit and the deviation from the average carbon footprint across 18 daily activities are provided in tables S6 and S7). Low-income households have higher carbon footprints in leisure and personal errands due to limited access to energy-efficient options and economic constraints. Middle-income groups show a more even distribution, with a focus on formal education. Middle-high-income groups make eco-friendly choices but have higher carbon footprints in leisure and social activities. High-income households prioritize sustainability but contribute more carbon emissions in education, reflecting resource-intensive practices and wealth disparities. Despite relatively balanced overall carbon footprints across incomes in Japan, wealthier households invest more in health and child-rearing activities, impacting carbon emissions.

The Paris Agreement has underlined the ambitious 1.5 °C climate target. Based on prior research, to attain the goals of the Paris Agreement, annual per capita carbon footprint goals for the years 2030, 2040, and 2050 have been established at 3.2 tCO₂e, 2.2 tCO₂e, and 1.5 tCO₂e, respectively (Lettenmeier et al 2019). Given the decarbonization potential of household lifestyle change, an assessment of its impact on future climate targets has been conducted as figure 5, with Panels figures 5(a)–(e) delineating the projected outcomes from 2025 to 2050 for different income groups.

Zoom In Zoom Out Reset image size

Delving into the findings, different income groups showcase analogous decarbonization trends at varying rates of adoption of low-carbon lifestyles (see table S8). Upon dissecting the data, a consistent trend emerges across income groups: higher adoption rates of low-carbon lifestyles correlate with reduced individual carbon footprints. Moreover, we have demarcated the intersections corresponding to the adoption rates and the 1.5 °C global warming targets, signifying the levels of adoption required to align with these ambitious climate objectives. For example, both Panels figure 5(a) (Low-Income) and figure 5(e) (High-Income) suggest that to meet the per capita emission target of 3.2 t CO₂ by 2030, an adoption rate exceeding 30% is vital. In contrast, the Low-Middle, Middle, and Middle-High income brackets can achieve this with a mere 15% adoption rate. It is particularly noteworthy that for the Middle-High-Income households, this goal is most attainable. This trend holds through to 2040, with middle-income groups maintaining lower necessary adoption rates than their wealthier and poorer counterparts.

By 2050, to reach the stringent 1.5 tCO₂e per capita target, all groups must surpass a 30% adoption rate. The projections indicate that lower-income groups face a tougher challenge in achieving the 1.5 °C target, even at higher adoption rates. This is partly due to the higher resource-intensive activities, like private education, which are prevalent among wealthier households. It also reflects the inherent difficulty in expecting significant lifestyle reductions in high-income groups.

Additionally, these projections have not accounted for potential future income fluctuations among these groups, and the per capita carbon footprint under the 1.5 °C target integrated here includes the introduction of carbon capture and storage technologies. The sensitivity of these targets to negative emission technologies is significant. Absent these technologies, the targets become markedly stringent: 2.5, 1.4, and 0.7 tCO₂e for 2030, 2040, and 2050, respectively.