Working Time and the Environmental Intensity of Well-Being: A Cross-National Analysis of High-Income OECD Countries, 1970–2019

A key question in the sociology of development literature is how economic development can affect environmental conditions (Givens, Jorgenson, and Clark 2016). Prior to the implementation of the SDGs, international development was guided primarily by the “growth consensus,” which is the notion that the solution to social and economic problems is more economic growth (Hickel 2020; Jackson 2009a). This line of thinking led to the adoption of policies geared explicitly toward economic growth, with a disregard for the externalities associated with that growth, such as environmental degradation and increased economic inequality, on the assumption that beyond a certain level of economic growth these problems fix themselves (Goldman 2005). The SDGs are an intentional pushback on the idea that economic growth and development are interchangeable (Meadowcroft 2017; Sachs 2012). While the SDGs represent progress in considering the complexities of the concept of development, there are still substantial disagreements on the best approach to sustainable development.

Perhaps the most popular vision of sustainable development can be referred to as “eco-efficiency” or “green growth.” It continues to rely on the logic of the growth consensus: the idea that economic growth is the engine of development and that sustainable development is about minimizing the environmental impacts of growth through the development of more efficient methods of production and better social awareness of environmental problems. A green-growth approach to sustainable development is one that relies on decoupling economic growth from its environmental impacts. That economic growth remains a core feature of mainstream sustainable-development approaches is highlighted by the fact that SDG 8 calls for further economic growth (United Nations n.d.a). In environmental sociology, this perspective is most closely associated with ecological modernization theory, which builds on modernization theory, situated within the sociology of development (for foundational texts, see Chase-Dunn 1975; Rostow 1990). Ecological modernization theory asserts that, for several reasons, economic growth has a curvilinear relationship with environmental harms. Per this theoretical perspective, economic growth often increases environmental pressures in the early stages of development but will eventually lead to technological advancements and the development of an ecological consciousness among individuals, businesses, and the state (Mol, Sonnenfeld, and Spaargaren 2009; Mol, Spaargaren, and Sonnenfeld 2014).

Critics of the green-growth approach argue that economic growth beyond a certain point is antithetical to sustainability. For example, when considering issues of human well-being and development, economic growth does appear to improve well-being in poorer countries (Brady, Kaya, and Beckfield 2007; Clark 2011; Cole 2019; Firebaugh and Beck 1994), but the picture is far more complicated in wealthier countries, where further economic growth has, at best, diminishing returns, and, at worst, can reverse improvements in well-being (Diener, Kahneman, and Helliwell 2010; Easterlin 1974, 2015; Fanning and O’Neill 2019; Jackson 2009b; Tapia Granados and Ionides 2008). There is a similar skepticism about the ability of economic growth to be “green” and decouple from its environmental costs. For example, the evidence of decoupling of economic growth from greenhouse-gas emissions is relatively weak. Most empirical research finds that increases in GDP are associated with increases in emissions (Jorgenson and Clark 2012; Knight and Schor 2014; Longhofer and Jorgenson 2017). While there is evidence of relative decoupling, in that the rate of increase of emissions has slowed, it is unlikely that this will be enough to meet rapidly approaching climate-goal deadlines to avoid the worst impacts of global climate change (Hickel et al. 2021). Furthermore, the evidence of relative decoupling in wealthy countries is likely due in large part to the organization of the global economy, where emissions are typically attributed to where production occurs, not where the final consumption of a product occurs. When evaluating consumption-based emissions measures, the evidence of decoupling is far weaker (Knight and Schor 2014; Roberts et al. 2020; Thombs 2018). Research on the relationship between economic growth and ecological footprint, which is a measure intended to capture total environmental pressures, also finds that growth intensifies environmental impacts (Jorgenson and Clark 2011).

Other research suggests that the techno-optimism of the green-growth perspective is misguided. The contradictory relationship between technological advancement and resource consumption is often referred to as the Jevons paradox, after the classical political economist William Stanley Jevons, which highlights how the primacy of economic growth in capitalism is inherently unsustainable (Alcott 2005; Foster, Clark, and York 2010; Jevons 1865). In particular, while technological advancements can lead to improvements in efficiency, those improvements often lead to greater overall production, rather than the same amount with fewer inputs. This is because the primary goals are greater profits and overall growth rather than environmental sustainability.

The Jevons paradox is specifically about efficiency advancements, but the mechanisms underlying it can also apply to other technological advancements, like the development of alternative energy sources. York, Adua, and Clark (2022) refer to the “displacement paradox,” which is the idea that green technology alternatives, like renewable energy, do not displace but rather add to existing technologies. While some recent studies suggest that the development of alternative energy sources has helped to decouple economic growth from emissions (Wang, Assenova, and Hertwich 2021), other studies present less optimistic results for the development and deployment of renewable energy. For example, Thombs (2017) finds that while renewable energy consumption reduces emissions in poorer countries, it does not in wealthier countries. Similarly, York (2012) shows that renewables do not simply displace fossil fuels but instead supplement them, leading to greater overall energy production and consumption. This is again due to the fundamental imperative of growth rather than sufficiency.

Based on these findings, York and Bell (2019) suggest it is inappropriate to refer to the development of alternative energies as an “energy transition,” preferring the term “energy addition.” Needless to say, the issue of technological advancement is a complex one. Technological advancements aimed at improving the resource efficiency of production or displacing fossil fuel energy sources are necessary. However, the issue comes down to the question, technology for what? If the growth consensus remains the dominant approach, then the potential environmental benefits of efficiency improvements and alternative energy development are reduced, as the primary emphasis is on economic growth first and environmental impacts second. Indeed, Wang, Assenova, and Hertwich (2021) find that while alternative energy has contributed to decoupling, decarbonization of energy systems is not occurring quickly enough to meet the goals of the Paris Climate Accords.

Instead of relying on economic growth and technological advancements, in the form of efficiency gains or alternative technologies such as renewables, some scholars argue that achieving sustainability requires a focus on structural change and social and economic policies that move away from the growth consensus. One approach to this is referred to as “degrowth” (Kallis 2017). Proponents of degrowth, and related perspectives, argue that wealthy countries around the world need to engage in a planned reduction of energy and resource consumption with the goal of reducing the size of their economies to within planetary limits (Demaria et al. 2013; Hickel 2021). Degrowth goes against the dominant political economic paradigm, so it faces unique challenges in terms of implementation. However, degrowth scholars offer numerous suggestions for policies aimed at moving society in a more sustainable direction. One such policy is the reduction of working hours, which despite its challenge to growth imperatives has gained some support in recent years.

Proponents of working-time reduction (WTR) suggest that it has the potential to be a multi-dividend sustainable-development policy that improves social, economic, and environmental outcomes. In this section, we first describe how working time can affect environmental outcomes and provide a brief overview of the empirical research connecting the two. Then, we discuss how working time can affect human well-being and provide a brief overview of the empirical research on that relationship as well.

It is theorized that longer working hours lead to greater environmental pressures by increasing production and consumption while also promoting more ecologically intensive consumption through both competitive consumption practices and time-stress (Fitzgerald 2022; Knight, Rosa, and Schor 2013a; Schor 2010). There are two theorized mechanisms through which longer working hours can increase environmental pressure. The scale effect is the contribution of working hours to overall economic growth, or its contribution to GDP. Put simply, more work increases the scale of the economy through greater production and consumption. As noted above, greater economic growth tends to be associated with greater environmental impacts, so if longer working hours promote economic growth, they also contribute to greater environmental impacts. On the consumption side, working longer hours tends to be associated with more consumption of goods and services, which also contributes to greater overall economic growth. This is part and parcel of what Schor (1992) refers to as the “work and spend” cycle, where consumption dominates many aspects of modern life. Engagement in consumer culture is facilitated through longer working hours.

The second mechanism is the compositional effect: working hours affect the composition of consumption, net of its contribution to GDP. One way to think about this is as an issue of time scarcity (Becker 1965). Those who work long hours have less free time and are more likely to choose less time-intensive but more environmentally harmful products and services (Jalas 2002, 2005; Jalas and Juntunen 2015; Kasser and Brown 2003). For instance, food consumption and preparation have clear connections to time scarcity and environmental impacts. Those with more free time may prepare their meals at home more often, while those with less free time may choose more fast-food meals, which are generally more ecologically intensive. The compositional effect can go beyond time effects, however, and more research is needed in this area. For instance, Fitzgerald (2022) shows that inequality shapes the compositional effect, suggesting that competitive consumption pressures also lead people to buy more environmentally intensive goods and services to “keep up with the Joneses.”

The vast majority of empirical research on this relationship associates longer working hours with higher levels of a variety of environmental harms (Antal et al. 2020; Fitzgerald 2022; Fitzgerald, Jorgenson, and Clark 2015; Fitzgerald, Schor, and Jorgenson 2018; Hayden and Shandra 2009; King and van den Bergh 2017; Knight, Rosa, and Schor 2013a; Shao and Rodríguez-Labajos 2016). For instance, Knight, Rosa and Schor (2013a, 2013b) analyzed high-income OECD countries from 1970 to 2007 and found working hours to be positively associated with total carbon emissions, carbon footprint, and ecological footprint. Fitzgerald, Jorgenson, and Clark (2015) examined both developed and developing countries from 1990 to 2008. They find that working hours are positively associated with energy consumption and that the relationship is increasing in intensity over time. While most of the studies in this area are cross-national, some use survey data on individuals or households (Jalas and Juntunen 2015; Kasser and Brown 2003; Nässén and Larsson 2015). For instance, Nässén and Larsson (2015) examine data on Swedish households and analyze the relationship between working hours and a measure of household carbon emissions and energy consumption. Other recent research examines the relationship at the US state level. Fitzgerald, Schor, and Jorgenson (2018) examined US states from 2007 to 2013 and find that working hours and CO₂ emissions are positively associated with one another, with both scale and compositional effects. Mallinson and Cheng (2021) replicated that study but included newer data from 2014 to 2017 and also find that the effect increased in magnitude from 2014 to 2017. Most recently, Fitzgerald (2022) elaborated on the relationship between emissions and working hours at the US state level by considering the role of income inequality. Findings from that study suggest that the effect of working hours on emissions is larger in places with wider income inequality.

On the other hand, it is possible that reducing working time would have rebound effects, where more free time results in greater environmental pressures (Antal et al. 2020). For instance, more free time might encourage households to take more vacations, which would increase the ecological intensity of consumption. Reduced working time may also increase labor productivity. Indeed, one of the arguments of the four-day-workweek social movement is that shorter hours do not necessarily mean reduced production, and the trials mentioned earlier did find that worker productivity improved with shorter working hours (Autonomy 2023). So even if less working time would slow overall economic growth and thus reduce environmental impacts, this greater labor productivity, meaning more production per hour of work, could dampen the scale effect of shorter working hours.

Research on the relationship between working hours and health suggests multiple pathways through which working time is associated with human well-being outcomes. The first is referred to as the job-strain model (Bakker and Demerouti 2007; Karasek 1979; Karasek et al. 1981; Kleiner, Schunck, and Schömann 2015). This model emphasizes the structure of work environments and how job demands can have negative health consequences (Karasek 1979; Karasek and Theorell 1990; Pfeffer 2018). A job demand highlighted in this literature is working-time mismatch, where people work longer hours than they would like (Bakker and Demerouti 2007; Kleiner, Schunck, and Schömann 2015; Virtanen and Kivimäki 2018). The effects of the structure of work have also been identified by others, emphasizing how work has become more precarious over time, which increases work-related stress and anxiety (Kalleberg and Vallas 2017; Standing 2011). Beyond the precarity of work, much work can be alienating and unfulfilling, particularly when there is a lack of control over working conditions, which can also lead to worse mental health as people spend more time at work (Braverman 1998; Graeber 2018). In all, from the job-strain perspective, we should expect that more time at work is more damaging to health because work increases overall stress, which is associated with higher odds of unhealthy coping behaviors (Kleiner and Pavalko 2010; Ng and Feldman 2008). On the other hand, working fewer hours could result in improvements to mental health and life satisfaction, especially as it relates to leisure time and control over one’s time (Golden and Wiens-Tuers 2008; Schor 2010). And these effects on mental health are likely to be shared by entire families, not just the individuals who work long hours (Kleiner and Pavalko 2014).

A second pathway, similar to the compositional effect explained above, emphasizes that longer working hours structure time availability (Becker 1965; Fan et al. 2015; Virtanen and Kivimäki 2018). Those who work longer hours experience greater time scarcity, which reduces their mental and physical well-being. This is particularly important in the context of physical health: those with less free time are less likely to exercise (Fan et al. 2015; Kleiner and Pavalko 2010, 2014; Nomaguchi and Bianchi 2004). Time availability also shapes consumption choices and can be particularly problematic when it comes to food consumption. For instance, Fan et al. (2015) find that longer working hours are associated with more frequent fast-food consumption. Lack of sleep is another consequence of long work, and is associated with higher risks for a number of physical and mental health problems (Afonso, Fonseca, and Pires 2017; Virtanen and Kivimäki 2018). Time availability can also shape mental health outcomes. For instance, Kleiner, Schunck, and Schömann (2015) find that longer working hours affect work–life balance, especially with respect to time spent with family, which can lead to depression (Roxburgh 2012) and other mental health problems.

A third pathway is that work often involves exposure to health hazards (Dembe et al. 2005; Jorgenson et al. 2020; Virtanen and Kivimäki 2018). A collection of articles edited by Tim Driscoll and Paul Whaley (2022) explores this topic in depth. Numerous hazards may be present in the workplace, across different sectors of the economy, including hazardous chemicals and toxins, air pollution, noise pollution, and many others (Pega et al. 2021; Schlünssen et al. 2023). In addition to these hazards, work itself can be physically demanding or require unhealthy practices, such as sitting for long periods, which itself can result in various health problems, all of which are exacerbated with longer working hours (Hulshof et al. 2021).

There is a substantial empirical literature exploring the relationship between working hours and human well-being, though it is mostly at the micro level, using surveys. For instance, Kivimäki and Kawachi (2015) find that working longer hours, regardless of race or gender, is associated with increases in coronary heart disease and stroke. Similarly, Thomas et al. (2018) find that longer hours are associated with worse physical health among women. A recent study using data from the World Health Organization and the International Labor Organization found that longer working hours are associated with large increases in the incidence of heart disease and stroke (Pega et al. 2021). Hulshof et al. (2021) find that long working hours are associated with increased ergonomic risk factors, which are situations that cause bodily wear and tear, notably knee, back, and hip pain. Most recently, studies of people who took part in the four-day-work-week trials in the United Kingdom find that their stress levels declined significantly and levels of burnout dropped dramatically compared to their pre-trial surveys (Autonomy 2023).

While most research in this area is conducted at the individual level, some researchers have begun to consider how working time may affect health in a more macro context. For example, a recent study by Jorgenson et al. (2020) explores the relationship between working hours, air pollution, and life expectancy at the US state level. Their findings suggest that longer working hours are associated with shorter average life expectancy, and this relationship is stronger when air pollution is worse. In addition to the focus on objective measures of well-being, like life expectancy or specific health outcomes, other studies have examined subjective measures of well-being and found that people who work fewer hours have greater life satisfaction (Kasser and Sheldon 2009; Pouwels, Siegers, and Vlasblom 2008). In all, the empirical evidence suggests that reducing working hours could improve human well-being.

Research shows a strong positive association between working hours and environmental pressures and a strong negative association between working hours and human well-being. However, advocates of WTR suggest it can improve both environmental outcomes and human well-being simultaneously. To assess this claim, we turn to the concept of the environmental intensity of well-being, which is an alternative measure of sustainable development that puts human well-being, instead of economic growth, at the forefront. In this section, we first provide background on the environmental intensity of well-being and then discuss the empirical research on this unique measure.

Research on this topic draws on a rich literature that looks at connections between economic growth, use of the environment, and human well-being (Daly 2014; Dietz and Jorgenson 2014; Dietz, Rosa, and York 2012; Easterlin 1974; Jackson 2009a, 2009b; Mazur and Rosa 1974). As mentioned above, the dominant paradigm has long assumed that economic growth will lead to improvements in human well-being, but this assumption ignores environmental costs and their impacts on well-being. Indicators such as the environmental intensity of well-being allow researchers to gain a better understanding of the benefits and detrimental effects of human activities by combining environmental impacts and social outcomes into one indicator that is comparable across scales and locations. This indicator offers a valuable way to link two components of development and sustainability: “balancing human well-being with impacts on the biophysical environment” (Dietz, Rosa, and York 2009:114).

Early work on such indicators began with a concept of environmentally efficient well-being (Dietz, Rosa, and York 2009), followed by the environmental intensity measure (Dietz, Rosa, and York 2012). Research explored the effect of economic growth on EIWB, where “ecological footprint” represents the environmental impact captured in the measure (Jorgenson and Dietz 2015). Subsequent work expanded the indicator to look at the energy intensity of well-being (Jorgenson, Alekseyko, and Giedraitis 2014) and CIWB (Jorgenson 2014). Across these different measures, well-being is often operationalized as life expectancy, which is considered an objective measure of well-being at a macro level of analysis, although some studies explore subjective measures like average life satisfaction (Knight and Rosa 2011). Research in this area shows that social factors affect the environmental intensity of well-being differently depending on the group in both national (Briscoe, Givens, and Alder 2021; Jorgenson, Dietz, and Kelly 2018) and international (Greiner and McGee 2020) contexts. Similarly, Thombs (2022) finds that increasing fossil fuel dependency is associated with higher CIWB but that reducing dependence does not have the same effect in lowering CIWB.

Research using CIWB and EIWB examines the relationship with various types of drivers and shows how different types of inequalities matter to the relationship between emissions and population well-being, at various scales from the subnational to the cross-national. The various inequalities and other factors examined in relation to CIWB include national level of economic development (Jorgenson and Givens 2015; J. Wang et al. 2022), domestic income inequality (Jorgenson 2015), urbanization (Givens 2015; McGee et al. 2017; S. Wang et al. 2022), technology (Feng and Yuan 2016), world-society integration (Givens 2017), global trade (Givens 2018), gender inequality (Ergas et al. 2021; Jorgenson et al. 2018), and race (Briscoe, Givens, and Alder 2021).

Looking at the effects of working hours on the environmental intensity of well-being is in line with the broader interests of scholars in this area of research and contributes to a better understanding of ways to reduce CIWB and EIWB (Roberts et al. 2020). For example, research finds that resource use and pollution emissions are associated with affluence and that the growth imperative and consumption expansion obstruct needed societal change (Wiedmann et al. 2020). A decent living for all humans with minimum energy consumption is possible, but one component of the change needed is decreased demand, in addition to advances in technology (Millward-Hopkins et al. 2020). Similarly, O’Neill et al. (2018) find that a good life within planetary bounds is possible, but high-affluence lifestyles exceed planetary limits. A good life does not require high levels of material consumption and can be achieved through increasing time for leisure and other social activities, which is beneficial for both social and environmental outcomes. This shift can be facilitated through reduced working hours (Hickel et al. 2022; Schor 2010).

In this study, we analyze data for 34 countries (Table 1) from 1970 to 2019. We do not include years before 1970 because of a lack of data on key indicators. All of the countries examined are high-income countries, defined by the World Bank as the top quartile of GDP per capita, and members of the OECD.

We only examine high-income countries for two reasons. First, because the arguments for WTR are generally aimed at these wealthy countries, where further economic growth can be categorized as unnecessary, or as Herman Daly (2014) defined it, “uneconomic” growth. While discussions of WTR should not exclude developing countries, there are considerations and arguments for that context which are beyond the scope of this study. Second, data are more consistently available and comparable for these countries, making the analyses more robust. That said, there are still issues of data availability, and the data set we analyze is not perfectly balanced. For instance, only after 1990 do some former Republics of the Soviet Union (Czech Republic, Latvia, Slovakia, Slovenia, Lithuania, and Estonia) have data available on key measures. To maximize the use of total data, we allow the number of observations to vary across models; depending on the model, there are between 1,217 and 1,418 total observations.

We examine two dependent variables: CIWB and EIWB. CIWB is the ratio of territorial CO₂ emissions per capita to average life expectancy. The data on per capita CO₂ emissions come from the Global Carbon Atlas (2021) and include CO₂ emissions from fossil fuel sources and the manufacture of cement. In alternative models we use consumption-based CO₂ emissions, which more appropriately allocate emissions to where consumption takes place (Givens 2015, 2018). However, those data are not available until after 1990, so we use territory-based emissions in the reported models. The data on life expectancy come from the World Bank’s World Development Indicators database. Life expectancy is an objective indicator of population health (World Health Organization 2015) and is commonly used in cross-national analyses (Dietz, Rosa, and York 2009; Jorgenson 2014; Kelly 2020).

EIWB is the ratio of total ecological footprint of consumption per capita to average life expectancy in a given country. Ecological footprint of consumption is a composite measure intended to represent the amount of ecological material it takes to produce the natural resources consumed by a given population (Global Footprint Network n.d.). Data on ecological footprint are from the 2019 release of the National Footprint Accounts from the Global Footprint Network (2019). Due to data availability, EIWB models use data from 1970 to 2016 and do not include Iceland or South Korea because data are not reported for these countries.

As both CIWB and EIWB are ratios, we must ensure that neither the numerator nor the denominator has disproportionate influence. We follow the same approach as previous researchers using these measures (Briscoe, Givens, and Alder 2021; Dietz, Rosa, and York 2009; Huang and Jorgenson 2018; Thombs 2022) and constrain the coefficients of variation to be equal by adding a constant to each numerator. This shifts the mean without affecting the variance. These variables are created using the eiwb command in Stata, which was developed by Ryan Thombs (2021). The formulas are then:

The key independent variable is average annual hours of work per worker. These data come from the Conference Board’s Total Economy Database. They reflect the actual number of hours worked, including overtime and excluding paid time off. The Conference Board compiles data from national labor surveys and the OECD. This is the same source of data as used in previous cross-national studies of working hours (Fitzgerald, Jorgenson, and Clark 2015; Hayden and Shandra 2009; Knight, Rosa, and Schor 2013a; Shao and Shen 2017).¹ Following previous studies, we estimate the scale and compositional effects of working hours in separate models. In models that estimate the scale effect (the contribution of working hours to total GDP), we disaggregate GDP into three components: hours worked, labor productivity, and percentage of the population that is employed (Fitzgerald, Schor, and Jorgenson 2018; Hayden and Shandra 2009; Knight, Rosa, and Schor 2013a; van Ark and McGuckin 1999). When estimating the compositional effect, we control for GDP per capita to find the effect of working hours net of its contribution to economic growth. Labor productivity (GDP per hour worked), employed population percentage, and GDP per capita are also gathered from the Total Economy Database.² We also control for urbanization (the percentage of the population living in urban areas) and trade (the sum of exports and imports of goods and services, as a share of GDP). Both of these variables are from the World Development Indicators.

Descriptive statistics and pairwise correlations for all of the variables included in the analyses are presented in Tables 2 and 3, respectively.

We estimate first-difference models with a lagged dependent variable, two-way fixed effects, and panel-corrected standard errors. These models are robust to time-invariant unobserved heterogeneity as the first-difference estimator implicitly controls for time-invariant unit effects while the unit-specific fixed effects control for unit-specific time trends (Allison 2019). Year-specific fixed effects control for changes that impact all countries in a given year, such as a global recession (Allison 2009; Thombs 2020). Post-estimation tests revealed heteroskedasticity and cross-sectional dependence, which suggest the need for panel-corrected standard errors (Beck and Katz 1995, 2011). The Wooldridge test revealed autocorrelation in all models, which we control for by lagging the dependent variable (Pickup 2015).

The coefficient of the lagged dependent variable in models where the variables are measured in levels instead of differences is near 1.³ This suggests that the dependent variable is nonstationary (Dickey 2015). This is tested with the Fisher-type unit root test, which suggests that CIWB is nonstationary while EIWB is stationary. While stationarity is typically a more serious concern for time-series analyses, it can be an issue in panels where T > N and is corrected by using first-difference models (Wooldridge 2015). Thus, we report first-difference models with two-way fixed effects for the most conservative results.⁴ However, models that only include two-way fixed effects and not first differences have substantively similar results and are reported in the  appendix. We also tested for the presence of multicollinearity by examining pairwise correlations and calculating variance inflation factor (VIF) values. None of the variables that are included in the same model have high correlation coefficients; the strongest correlation is that between employed population percentage and GDP per hour, at r = 0.532 (Table 3). The highest VIF across each of the models is 2.99, which is for urban population percentage in the model estimating the scale effect of working hours on EIWB.⁵ This suggests that multicollinearity is not a significant problem in the models. The estimated models are

Consistent with previous research on the human drivers of environmental change, all of the variables in the model are log-transformed (Dietz, Rosa, and York 2007; Rosa, York, and Dietz 2004). This corrects for skewness and allows the estimation of elasticity models, where the interpretation of the coefficients is the percentage change in the dependent variable given a percentage change in the independent variable over time, while holding all else constant.

Table 4 presents the results of the models testing the scale and compositional effects of working time on CIWB and EIWB. Model 1 estimates the scale effect of working hours on CIWB, while Model 2 estimates the compositional effect of working hours on CIWB. Models 3 and 4 estimate the scale and compositional effects of working hours on EIWB, respectively.

Turning first to Model 1, the scale effect of working hours on CIWB is positive and significant (p < .000). The coefficient is 0.085, with a 95% confidence interval between 0.045 and 0.125. The point estimate suggests that a 1% increase in working hours is associated with a 0.085% increase in CIWB, holding all else constant. The control variables in this model show the expected relationships. Labor productivity and employed population percentage are positive and significant (p < .000). Interestingly, the effect of urban population percentage is negative and significant (p = .015), suggesting that increases in urbanization reduce CIWB. Previous studies of this relationship have been inconsistent; some find that urban population percentage is positively associated with CIWB (Givens 2015; McGee et al. 2017), while others find a negative association (S. Wang et al. 2022). The discrepancies are likely due to variation between studies and whether all countries are examined, versus subsets of countries based on economic development. The effect of trade is nonsignificant (p = .647).

Turning to Model 2, the compositional effect of working hours on CIWB is nonsignificant (p = .263). While this result is inconsistent with studies that have examined the relationship between working hours and CO₂ emissions in the United States (Fitzgerald 2022; Fitzgerald, Schor, and Jorgenson 2018; Fremstad, Paul, and Underwood 2019; Mallinson and Cheng 2021), it is consistent with other cross-national studies which did not find a significant compositional effect (Knight, Rosa, and Schor 2013a). The control variables remain the same, with trade nonsignificant (p = .647) and urban population percentage negative and significant (p = .016). GDP per capita is positive and significant (p < .000), with a coefficient of 0.075. This suggests that a 1% increase in GDP per capita is associated with a 0.075% increase in CIWB, holding all else constant. Considering the arguments on green growth discussed earlier, this would appear to contradict that idea that economic growth is sustainable.

Models 3 and 4 examine the effects of working hours on EIWB. Turning first to Model 3, the compositional effect of working hours on EIWB is positive and significant (p < .000). The coefficient is 0.144, with a 95% confidence interval between 0.083 and 0.206. The point estimate of the coefficient suggests that a 1% increase in working hours is associated with a 0.144% increase in CIWB, holding all else constant. GDP per hour (p < .000) and employed population percentage (p < .000) are both positive and significant as well. Urban population percentage (p = .747) and trade (p = .344) are both nonsignificant. The fact that urbanization has different effects on CIWB versus EIWB in OECD countries is interesting and is worthy of further analysis in future research.

Model 4 shows that the compositional effect of working hours on EIWB is positive, with a coefficient of 0.047 and a 95% confidence interval from –0.007 to 0.101. It is only significant at the .10 level (p = .087). This is somewhat consistent with previous research, where Knight, Rosa and Schor (2013a) find a significant effect for both scale and compositional effects when examining ecological footprint. However, the evidence here is not as strong, as p is greater than .05 and the 95% confidence interval ranges from negative to positive, suggesting a null effect. The control variables here match Model 3, with urban population percentage (p = .613) and trade (p = .232) both nonsignificant. GDP per capita is positive and significant (p < .000), with a coefficient of 0.135. The 95% confidence interval is between 0.106 and 0.163. This again suggests that economic growth increases EIWB (holding all else constant), which contradicts current assumptions regarding economic growth and sustainable development.

To ensure the robustness of the findings, we conducted several sensitivity analyses. First, we tested for the presence of overly influential cases. To do this, we systematically re-estimate each model while excluding one country at a time. This is a common and effective technique to ensure that no single panel is overly influencing the results (Beck and Katz 1995). Across these analyses, the effects of working hours on CIWB and EIWB remained consistent, with only slight variations in coefficient sizes. Our findings are also robust to the inclusion of additional control variables and different dependent variables.

For the additional control variables, we estimated models which include the contribution of industry to overall GDP and education spending as a percentage of GDP. The former captures the degree to which a given country’s economy is industrialized, and the latter captures some modernization arguments that more educated countries are more likely to be engaged with sustainability issues. The inclusion of these two variables as controls does not affect the findings reported above, as the estimated effects of working hours on CIWB and EIWB remain substantively the same. We do not report these models above because the data availability for these two variables reduces the overall sample size to 819 for models examining CIWB, and 675 for EIWB.

We also estimated models examining an alternative conceptualization of CIWB, where instead of using territory-based CO₂ emissions in the numerator we use consumption-based CO₂ emissions. Data on consumption-based CO₂ emissions are not available until 1990, resulting in a sample size of 918 instead of 1,418 for the territorial-based emissions, which go back to 1970 for most countries in the sample. It should be noted that the correlation coefficient for these two variables is very high, at 0.988. With that said, the results of these models are similar to those reported above. The main difference is that the estimated compositional effect for working hours is positive (β = 0.011) and significant at the .05 level (p = .042).

Proponents of WTR suggest that it has the potential to be an effective sustainable-development policy that improves both environmental and social outcomes. We empirically test this proposition by examining the relationship between working hours and both CIWB and EIWB. Longer working hours are positively associated with both measures, suggesting that shorter working hours could reduce both of them, a desirable outcome in terms of sustainability. Scholars have noted that the understanding of the collective impacts of WTR is limited (Hickel et al. 2022). Our study contributes to this understanding in a group of high-income OECD countries where it is suggested that further economic growth is “uneconomic” (Daly 2014) and, thus, would seem to be prime candidates for WTR policies. The issue of collective impacts highlighted by Hickel et al. (2022) emphasizes open questions regarding whether reduced work time will increase people’s consumption of and engagement in environmentally intense leisure activities such as shopping and travel. While our study does not directly examine this question (which is not possible at this unit of analysis), the research presented here provides some support for the idea that, on average, at a macro level, sustainability gains are still realized, as indicated by association between working time and CIWB and EIWB.

The findings reported here have several implications for theory and policy. First, regarding the broader question of approaches to sustainable development, they provide support for the idea that sustainable development does not require further economic growth. Rather, innovative social and economic policies have the potential to help countries achieve their sustainable-development goals. In this case, we show that shorter working hours is a pathway that can improve sustainability at the national level by reducing emissions and overall environmental impacts, while also improving human well-being.

Second, this research is the first to explicitly examine the claim that WTR is a multi-dividend sustainability policy. Previous research on the effects of working time has examined environmental costs and impacts on human well-being, but this research connects the two by examining CIWB and EIWB. Through these measures of sustainability, we can test the potentially broad benefits of WTR. While we find significant scale effects—the contribution of working hours to overall economic growth—the compositional effect, which captures how working hours affect the composition of consumption, is not significant at the .05 level for either CIWB or EIWB in our reported models (though it is significant at the .05 level in the models reported in the  appendix that only estimate two-way fixed effects without first-differencing). To a degree, this is in line with the results of previous studies on the environmental costs of working time.⁶ Nearly all research in this area finds significant scale effects of working hours on a variety of environmental outcomes. The compositional effect is more inconsistent in cross-national analyses. For example, Knight, Rosa, and Schor (2013a) find a significant compositional effect for the relationship between working hours and ecological footprint but not for the carbon footprint or total CO₂ emissions. On the other hand, Fitzgerald, at al. (2015) finds a significant compositional effect for energy consumption in both developed and developing countries. Similarly, research has found significant compositional effects on CO₂ emissions at the US state level (Fitzgerald 2022; Fitzgerald, Schor, and Jorgenson 2018; Mallinson and Cheng 2021). In this case, while our results are mostly consistent with previous studies, the differences likely come down to differences in sample (cross-national versus US state level), time period (this study uses much more recent data than previous cross-national studies), and the fact that our outcome measure is quite different in its inclusion of human well-being. In all, the results presented here provide further evidence that shorter working hours could have broad sustainability benefits for countries.

Future work in this area should continue exploring this issue in different contexts, specifically how it may vary by group or by economic circumstances. For example, research has shown that economic inequality shapes the relationship between working time and sustainability (Fitzgerald 2022). Evidence from France’s reduction in working time showed that lower-paid, less skilled workers did not get the same benefits from the shorter working hours, and in fact may have been harmed (Hickel et al. 2022). These findings raise questions about how WTR policies should be implemented or affect work in countries with wide inequality, and how these policies would affect part-time, temporary, and contract workers. Beyond issues of domestic inequality, how WTR fits into broader global environmental inequalities is another important issue. As it stands now, most research and policy efforts on this issue have been done in wealthy countries. On the one hand, this makes sense, as the focus of degrowth scholarship is on countries which already have a high standard of living. On the other hand, it means that the social and environmental benefits of WTR will mostly go to those in already privileged countries. Further consideration of this issue is needed, and empirical research ought to focus on how WTR may affect global environmental inequalities and also what WTR might look like in less developed countries.

In addition to issues of economic inequality, there are gendered components to work, both private and domestic, that should be considered and studied. Issues such as the gender pay gap and gendered disparities in household work suggest that WTR may not be equally feasible or beneficial for men and women. It is possible that WTR could lead to a more equal distribution of household labor, though research on the “second shift” suggests that gender norms may prevent this from happening automatically (Hochschild and Machung 2012). There are obvious social well-being implications with these issues, but there are also environmental implications, because pro-environmental behaviors, especially those within the home, have been feminized and performed at higher rates by women (Briscoe et al. 2019; Dzialo 2017; Kennedy and Kmec 2018). Recent research on CIWB shows that factors like inequality and economic growth affect groups differently based on race and sex (Briscoe, Givens, and Alder 2021; Jorgenson et al. 2018), and an analysis of how the relationship between working time and sustainability may differ between men and women would be insightful.

We must note some important limitations to our research. This study takes a macro-level approach and finds a significant positive association between working hours and both CIWB and EIWB. While critically important, this macro-scale approach also has some limitations. One is that the mechanism through which working time improves sustainability outcomes is not immediately clear. While that question is not addressed here, the pilot programs being conducted by 4 Day Week Global (n.d.) provide many insights on this question on a more micro level. We view our research as complementary to these exciting initiatives. A second limitation is that the macro approach can miss variation at smaller scales, such as the firm, industry, city, or region. This variation could be important in informing well-crafted policy on WTR, and future empirical research should focus on these scales as well.

Despite these limitations, a primary contribution of this work to the CIWB and EIWB literature is its clear and feasible policy implications. Prior research finds that economic growth is positively associated with both measures in higher-income countries (Jorgenson 2014; Jorgenson and Dietz 2015), which we also confirm in our analyses. This is an important theoretical finding, which represents a critique of the current political economic system, but in terms of policy implications, it would require drastic change that governments and companies are unlikely to pursue without a lot of pressure. WTR could address these issues in a more palatable way. While there is some resistance to the idea (Wade 2023), reducing working hours is a proposal that is gaining popularity among the public, governments, and companies (Veal 2023). It is an achievable change that our research shows would have clear sustainability benefits.