The macro-comparative decoupling literature has often sought to test the arguments made by the treadmill of production (TP) and ecological modernization (EM) theories. However, due to data limitations, these studies have been limited to analyzing the years after 1960. Given that both theories discuss historical processes operating before 1960, analyzing pre-1960 data is warranted to more comprehensively test the propositions made by both theories. We assess the long-term relationship between economic growth and CO2 emissions from 1870 to 2014 using a sample of global North nations. We use Prais-Winsten regression models with time interactions to assess whether, when, and how much CO2 emissions have decoupled from economic growth over time. We find that significant relative decoupling has occurred twice since 1870: during the last 30 years of the nineteenth century, the timing of which is contrary to what both the EM and TP theories might expect, and after 1970. We also observe that the relationship remained relatively stable from the turn of the twentieth century to approximately 1970, which aligns with the arguments made by the classical TP work. We conclude that shifts in the global organization of production have shaped the magnitude of the economic growth–CO2 emissions relationship and its changes over time, which has implications for climate mitigation policy.

## INTRODUCTION

Since the first Industrial Revolution, societies have increasingly relied on the burning of fossil fuels to power a growing economy based on capital accumulation. This expansive economic growth has disrupted the Earth's carbon cycle and resulted in the highest atmospheric concentration of greenhouse gases in at least 800,000 years (IPCC 2014a).1 This process has caused the Earth's surface temperature to increase by between 1.06 and 1.21 degrees Celsius since the mid-nineteenth century (Blunden and Arndt 2017). Though climate change affects the entire planet, its causes are uneven across space and time. Historically, most of the greenhouse gas emissions can be attributed to the nations of the global North,2 who collectively are still responsible for approximately 75 percent of current annual global emissions (Harlan, Pellow, and Roberts 2015). Only recently has the global South begun to make a significant contribution to the global carbon footprint, almost entirely due to the development of China and India, the two most populous countries. However, global North nations still have far greater emissions per capita than the global South (Harlan, Pellow, and Roberts 2015). The distribution of emissions over time has also been unequal, with roughly half of all cumulative anthropogenic CO2 emissions occurring in the last 40 years (IPCC 2014b). This phenomenon is tied to the “great acceleration” that occurred after World War II, which witnessed unprecedented economic, demographic, and ecological change (Dietz 2017; Schnaiberg 1980; Steffen et al. 2015).

In sociology, the treadmill of production (TP) and ecological modernization (EM) theories have been advanced to explain the relationship between ecological degradation and economic development. TP argues that ecological degradation is a product of the capitalist system, which is predicated on capital accumulation, which demands ever-increasing quantities of natural resources to reproduce and expand itself (Gould, Pellow, and Schnaiberg 2004; Schnaiberg 1980). In contrast, EM posits that although economic growth is likely to harm the environment at early stages of development, as countries modernize they develop institutions, technologies, and norms that take environmental harms into account and ultimately mitigate them. In the decoupling literature,3 quantitative studies have sought to evaluate the propositions made by both theories through analyzing the relationship between economic growth and environmental harm over time to see whether and when decoupling has occurred (Jorgenson and Clark 2012; Thombs 2018a). Constrained by data availability, these empirical studies focused on the post-1960 period, leaving the beginning of the exponential rise of CO2 emissions in the mid-nineteenth century largely unexplored.

We compiled a longitudinal data set consisting of territorial CO2 emissions, GDP per capita, population, and trade openness (exports + imports as a percentage of GDP) for global North nations from 1870 to 2014 (Barbieri and Keshk 2016; Bolt et al. 2018; World Resources Institute and Center for Advanced Infrastructure and Transportation 2018). We employ Prais-Winsten regression models with two-way fixed effects that interact GDP per capita with time to investigate how the association between economic growth and territorial CO2 emissions has changed over the course of that 145-year period. We find relative decoupling during the last 30 years of the nineteenth century, which is contrary to what either theory would expect, and after 1970. We also observe that the relationship remained relatively stable from the turn of the twentieth century to approximately 1970, which supports the arguments of classical TP theory.

Our findings contribute to the literature in two ways. First, evaluating long historical changes in the economic growth–CO2 emissions relationship allows us to situate the current period in historical context and see whether it is different from or similar to previous periods. Second, using long historical data is important for theory building because it allows us to better evaluate the propositions made by both the TP and the EM theories, and particularly whether their core tenets are generalizable across time.

TP was originally developed by Schnaiberg in his book The Environment: From Surplus to Scarcity (1980). In this book, Schnaiberg set out to explain the immense ecological degradation in the United States after World War II, which he attributed to changes in economic and social conditions. The TP perspective argues that this great acceleration in ecological degradation was tied to the increase in capital accumulation in global North nations after the war, which led to, and was in turn partly fueled by, the implementation of advanced technologies that replaced human labor and made production more efficient (Gould, Pellow, and Schnaiberg 2004; Lewis 2018; Schnaiberg 1980). More efficient production reduces the energy and resources used per unit of production, but in a capitalist economy, these lower marginal costs often lead to an increase in the total units produced, which can increase ecological harm at scale. This apparent contradiction between efficiency and scale is also known as the Jevons paradox, after William Stanley Jevons (1866), who observed that total coal consumption increased as coal production became more efficient.

Even though Schnaiberg developed the theory to explain the changes occurring in the postwar period, he was explicit that many of these changes were set in motion at the beginning of the twentieth century, tied to the rise of monopoly capitalism and of educational institutions that were supported by private and public sources to research and develop advanced technologies. However, it was the cooperative relationship between capital, the state, and labor that fueled the unprecedented economic growth after the war. Therefore, TP argues that the effect of economic growth on environmental harm, including CO2 emissions, is likely to remain steady or intensify through time and would suggest that it is highly unlikely for decoupling to occur at any point.

## ECOLOGICAL MODERNIZATION: DECOUPLING THROUGH MODERNITY

In contrast, EM posits that societies can reduce ecological harm through modernity and technological progress. In fact, Mol (1995:42) asserts that “the only way out of the ecological crisis is by going further into the process of modernization.” Rather than viewing degradation as an inevitable consequence of the global capitalist system, EM argues that processes such as economic growth can decouple from environmental harm through state and institutional reforms along with technological innovation that increases efficiency in production processes (Mol 2002; Mol and Spaargaren 2000; Spaargaren and Mol 1992). EM also asserts that these “greening” processes are likely to first occur in global North countries, because they have more mature institutions and greater access to advanced technologies (Mol 2002; Spaargaren and Mol 1992).

Huber (1982, 1985) asserts that industrial societies go through three stages of development: industrial breakthrough; the formation of industrial society; and the shift to superindustrialization (i.e., eco-efficient production). EM argues that economic growth likely causes substantial environmental problems at the first two stages of development due to institutions and people embodying an “economic rationality” that neglects ecological issues (Mol 2002). But as these societies continue to modernize and reach the stage of superindustrialization, an “ecological rationality” is said to emerge, where environmental considerations begin to be taken into account (Mol 2002). In most of the EM literature, there is a consensus that development and ecological harms began to decouple in the 1970s and 1980s in Western European nations (Mol 2002). Therefore, EM would expect to find (more specifically) that economic growth's association with CO2 emissions increases at first and then decouples beginning in the 1970s or 1980s.

## PREVIOUS EMPIRICAL ANALYSES: EMERGENCE OF SUPERINDUSTRIALIZATION OR A PERSISTENT TREADMILL?

Prior macro-level comparative analyses on decoupling have yielded a mixture of evidence for both TP and EM theories. However, the general takeaway has been that the relationship between economic growth and anthropogenic CO2 emissions differs across time and space. For example, Jorgenson and Clark (2012) examined economic growth's association with three measures of CO2 emissions (total CO2, CO2 per capita, and CO2 per unit of GDP) over time. They found a small, relative decoupling for total CO2 emissions in their global and developed-countries samples, whereas the association remained constant in less developed countries since the 1960s. For per capita emissions, they found an intensification in the global and less-developed-country samples, and a slight decoupling between economic growth and CO2 per unit of GDP in developed nations. Based on these findings, they assert that both EM and TP have shortcomings, and they both should engage more with theories that discuss the ways in which the global organization of production influences and shapes the relationship between the economy and the environment.

In a replication and extension of Jorgenson and Clark's (2012) work, Thombs (2018a) found a significant relative decoupling between GDP per capita and all three measures of CO2 emissions in developed nations, whereas the effects remained constant in less developed nations over time. He indicated that these findings suggest that carbon-intensive growth had “tilted” to the global South since the 1970s. Knight and Schor (2014), though not directly testing TP and EM, investigated the relationship between economic growth and two measures of CO2 emissions—territorial and consumption emissions—in high-income countries. Their results differed depending on the measure of emissions. They found that the effect of economic growth was greater for consumption-based emissions and stayed constant over time, whereas economic growth relatively decoupled from territorial emissions. Thus, the arguments made by both theories may also depend on how CO2 emissions are operationalized. Consumption-based emissions account for the emissions embodied in imports, whereas production-based emissions do not. In other words, only examining the emissions produced internally within a nation may give a skewed view of what is really occurring, since countries are embedded in global trade networks (Givens, Huang, and Jorgenson 2019; Jorgenson and Clark 2012; Knight and Schor 2014; Peters and Hertwich 2008; Peters et al. 2011; Thombs 2018a).

Longhofer and Jorgenson (2017) examined whether integration into the world society influenced economic growth's effect on CO2 emissions, observing that countries more embedded in the environmental world society experienced a more substantial relative decoupling over time.4 Though world society theory and EM differ in many of their core tenets, these findings suggest that one way ecological modernization might occur is through the dissemination of pro-environmental values through the world society (Longhofer and Jorgenson 2017). However, world society theory does not attribute the decoupling to greater development or superindustrialization per se, but rather to transnational institutions, organizations, and norms (Longhofer and Jorgenson 2017).

These studies suggest that there may be a variety of processes at work that are not fully captured by either the TP or the EM theory. The main limitation of these studies is that they were unable to examine data from before 1960. Given that many of the countries in the global North began to industrialize well before 1960, the field has been unable to fully capture the rise of industrialization and how the relationship between economic growth and CO2 emissions has changed in the long run.5 We seek to remedy this gap in the literature by examining whether decoupling has occurred in a sample of global North nations from 1870 to 2014. Evaluating this relationship enables a more comprehensive assessment of the propositions made by EM and TP and provides a longer historical evaluation than any of these previous studies.

## DATA AND METHODS

The study relied on a balanced panel data set from 1870 to 2014 for 13 countries: Belgium, Canada, Denmark, France, Italy, Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, United Kingdom, and United States. These countries were included in the sample because they had complete data for CO2 emissions, GDP per capita, and total population over the entire period of interest. We excluded other countries from the sample because balanced panels were necessary to prevent bias due to differences in the sample over time (Jorgenson 2014; Thombs 2018a). However, three of the selected countries (Belgium, Denmark, and the Netherlands) were missing trade openness data for 1942 due to World War II. Therefore, the models that include trade openness should be interpreted with additional caution. The 13 nations included in the analysis constitute a majority of global North nations from 1870 to 2014 and were collectively responsible for 41% of global cumulative CO2 emissions over the period (World Resources Institute and Center for Advanced Infrastructure and Transportation 2018). Moreover, this sample is appropriate to evaluate EM propositions over the long run because EM argues that decoupling is likely to first occur in global North countries due to their head start in the modernization process. A balanced panel data set in four-year intervals was used. Studies using time interactions often use five-year intervals, but this is an arbitrary number. We used four years because the total time period is divisible by four and this gives us a larger number of observations to be analyzed.

## STIRPAT

Our study used the STIRPAT model, a stochastic version of the IPAT formulation (York, Rosa, and Dietz 2003a, 2003b). STIRPAT is commonly used to test hypotheses concerning the effect of population, affluence, and technology on CO2 emissions (Dietz 2017; Knight, Rosa, and Schor 2013; Rosa, York, and Dietz 2004; York, Rosa, and Dietz 2003a, 2003b). The basic STIRPAT model is

$ln(I)=a+b[ln(P)]+c[ln(A)]+e$

I refers to an environmental impact, a is the constant term that scales the model, b and c are the coefficients for population (P) and affluence (A), and e is the error term. Consistent with prior studies, the technology component of STIRPAT is captured in the error term, as it includes all the drivers that are not affluence or population (York, Rosa, and Dietz 2003a). Alternatively, a vector of variables can be included in STIRPAT models to operationalize technology. Past studies have included urbanization, renewable energy, trade openness, and various institutional components in their models (Dietz 2017; McGee, Clement, and Besek 2015; Rosa and Dietz 2012; Thombs 2018a, 2018b). Because STIRPAT is a multiplicative model, the dependent and independent variables are natural-logged, making them equivalent to elasticity models. In other words, a 1% change in the independent variable is associated with a constant percentage change in the dependent variable.

### Dependent Variable

The study used total territorial CO2 emissions (excluding land use change and forestry) as the dependent variable.6 The data were obtained from the World Resource Institute's Climate Analysis Indicator Tool (2018), which compiles CO2 emissions data from multiple sources, including the International Energy Agency.

### Independent Variables

The two main independent variables controlled for in the study were GDP per capita and total population. These data were obtained from Bolt et al. (2018), who are part of the Maddison Project.7 We used Bolt et al.'s RGDPNApc measure, which estimates real GDP based on the national accounts of countries, measured in 2011 U.S. dollars.

In additional analyses, we also controlled for trade openness (exports + imports as a percentage of GDP) using data over a shorter time period (1922 to 2014). Trade openness has been observed to be a driver of territorial CO2 emissions (Jorgenson and Clark 2012; Thombs 2018b), but EM also argues that it may be key for modernization purposes (Mol 2002).8 We created this measure by first collecting exports and imports data from the Correlates of War project (Barbieri and Keshk 2016; Barbieri, Keshk, and Pollins 2009) that estimate values of trade flows for individual nations dating back to 1870. Because several nations in our sample were missing a considerable amount of data for before the early twentieth century, we only use the trade data that were available from 1922 to 2014. These data were measured in current U.S. dollars. We used the inflation conversion factor tables from Sahr (2017) to convert them to constant 2011 US dollars. We then divided the total trade value (exports + imports) by total GDP (Bolt et al. 2018) to calculate the measure for trade openness. Although we were unable to confirm the accuracy of our trade openness estimates prior to 1960, they were strongly correlated with the World Bank's (2018a) trade openness data for 1960 to 2014 (Pearson's r = 0.88). This suggests that our estimates were fairly accurate, at least for 1960 to 2014. The descriptive statistics are presented in Table 1. As mentioned, there are typically additional predictors controlled for in STIRPAT models, but we refrained from including them due to the lack of available data over the whole period analyzed.9

TABLE 1.

Descriptive statistics

 Total CO2 emissions 3.896(1.988) GDP per capita 9.288(0.916) Total population 9.539(1.195) Trade openness 3.326(0.731) Total observations 481*
 Total CO2 emissions 3.896(1.988) GDP per capita 9.288(0.916) Total population 9.539(1.195) Trade openness 3.326(0.731) Total observations 481*

Note: All numbers are natural-logged. Standard deviations are in parentheses.

*

The trade openness data consist of 451 observations due to availability over a shorter time span and several missing data in 1942 due to World War II.

### Estimation Techniques and Models

We estimated the STIRPAT models using time-series cross-sectional Prais-Winsten regression models with an AR(1) correction and panel-corrected standard errors that allow for disturbances that are heteroskedastic and contemporaneously correlated across panels (Beck and Katz 1995).10 We also included country-specific and year-specific intercepts to control for unobserved time-invariant heterogeneity within countries and time periods, analogous to a two-way fixed effects model (Baum 2006). The use of two-way fixed effects explained much of the variation in the models, making this approach particularly robust when there is limited availability of independent variables (Jorgenson and Clark 2012).

We estimated four models with the Prais-Winsten estimator. Models 1 and 2 are baseline models. Model 1 estimates the linear coefficients for GDP per capita and population for the period from 1870 to 2014. Model 2 extends Model 1 by including trade openness as an independent variable but relies on panel data from 1922 to 2014 due to the dearth of available trade data. Models 3 and 4 test for decoupling by including a series of interactions between GDP per capita and time, operationalized as dummy variables in four-year intervals. Model 3 excludes trade openness, whereas Model 4 includes it. The coefficient for each interaction is the change in the association between GDP per capita and CO2 emissions in relation to the 1870 GDP per capita coefficient.11 Therefore, if the coefficient for the interaction term is statistically significant for a particular year, then the effect of GDP per capita in that year is the sum of the 1870 GDP per capita coefficient and the coefficient for that interaction term (Huang 2018; Jorgenson and Clark 2012; Thombs 2018a). Models 1–4 are specified below, where ui are the country-specific intercepts, and eit is the disturbance term for each country in each year:

Model 1: total CO2 emissionsit = β1 GDP per capitait + β2 populationit + β3 year 1874t + … + β38 year 2014t + ui + eit

Model 2: total CO2 emissionsit = β1 GDP per capitait + β2 populationit + β3 tradeit + β4 year 1926t + … + β26 year 2014t + ui + eit

Model 3: total CO2 emissionsit = β1 GDP per capitait + β2 populationit + β3 year 1874t + … + β38 year 2014t + β39 GDP per capitait × year 1874t + … + β74 GDP per capitait × year 2014t + ui + eit

Model 4: total CO2 emissionsit = β1 GDP per capitait + β2 populationit + β3 tradeit + β4 year 1926t + … + β26 year 2014t + β27 GDP per capitait × year 1926t + … + β49 GDP per capitait × year 2014t + ui + eit

## RESULTS

The results for the main effects are reported in Table 2 for Models 1–4. Model 1 represents a baseline model controlling for GDP per capita and total population. Due to rounding procedures, the GDP per capita and population coefficients were identical. On average, from 1870 to 2014, a 1% increase in either measure is associated with a 1.179% increase in total CO2 emissions. Model 2 estimates the same model but with the addition of the trade openness measure and using the period between 1922 and 2014. The coefficients for GDP per capita and population change slightly compared to Model 1, but both remain positive and statistically significant (1.253 and 0.945, respectively). The coefficient for trade openness is positive and statistically significant, indicating that from 1922 to 2014, a 1% increase in trade openness is associated with a 0.122% increase in total CO2 emissions.

TABLE 2.

Prais-Winsten regression model estimates with panel-corrected standard errors and an AR(1) correction, 1870 to 2014 and 1922 to 2014

Model 1 1870–2014Model 2 1922–2014Model 3 1870–2014Model 4 1922–2014
GDP per capita 1.179*
(0.113)
1.253*
(0.093)
2.775*
(0.170)
1.446*
(0.101)
Population 1.179*
(0.146)
0.945*
(0.192)
1.311*
(0.128)
0.968*
(0.213)
(0.036)
0.112*
(0.036)
R2 0.839 0.943 0.894 0.945
N 481 308 481 308
Estimated coefficients 51 39 87 62
Model 1 1870–2014Model 2 1922–2014Model 3 1870–2014Model 4 1922–2014
GDP per capita 1.179*
(0.113)
1.253*
(0.093)
2.775*
(0.170)
1.446*
(0.101)
Population 1.179*
(0.146)
0.945*
(0.192)
1.311*
(0.128)
0.968*
(0.213)
(0.036)
0.112*
(0.036)
R2 0.839 0.943 0.894 0.945
N 481 308 481 308
Estimated coefficients 51 39 87 62
*

p < .05.

Notes: Panel-corrected standard errors are in parentheses. Unit-specific and period-specific intercepts are unreported. Models 3 and 4 also include interactions between GDP per capita and time, which are separately reported in Tables 3 and 4.

Models 3 and 4 tested whether economic growth decoupled from CO2 emissions over time. These are the same models as Models 1 and 2, but adding the interaction between GDP per capita and time. The coefficients for GDP per capita, total population, and trade openness are positive and statistically significant, consistent across models. As a reminder, the coefficients for GDP per capita in Table 2 are the effects of economic growth on CO2 emissions in 1870 for Model 3 and 1922 for Model 4. Tables 3 and 4 report the coefficients for the interaction between GDP per capita and time in Models 3 and 4, respectively, and also include the main effects of GDP per capita in each model for reference purposes.

TABLE 3.

GDP per capita coefficients excluding trade, 1870 to 2014 (Model 3)

 1870 2.775*(0.170) 1874 −1.224*(0.081) 1910 −1.726*(0.140) 1946 −1.793*(0.152) 1982 −1.975*(0.171) 1878 −1.218*(0.108) 1914 −1.699*(0.147) 1950 −1.799*(0.154) 1986 −1.859*(0.168) 1882 −1.527*(0.135) 1918 −1.433*(0.147) 1954 −1.773*(0.152) 1990 −1.989*(0.184) 1886 −1.619*(0.122) 1922 −1.640*(0.155) 1958 −1.833*(0.155) 1994 −2.038*(0.177) 1890 −1.667*(0.128) 1926 −1.685*(0.163) 1962 −1.757*(0.150) 1998 −2.102*(0.187) 1894 −1.612*(0.137) 1930 −1.734*(0.158) 1966 −1.671*(0.157) 2002 −2.299*(0.208) 1898 −1.575*(0.135) 1934 −1.596*(0.151) 1970 −1.616*(0.156) 2006 −2.244*(0.213) 1902 −1.672*(0.137) 1938 −1.563*(0.151) 1974 −1.755*(0.166) 2010 −2.085*(0.206) 1906 −1.710*(0.140) 1942 −1.792*(0.153) 1978 −1.746*(0.177) 2014 −2.162*(0.195)
 1870 2.775*(0.170) 1874 −1.224*(0.081) 1910 −1.726*(0.140) 1946 −1.793*(0.152) 1982 −1.975*(0.171) 1878 −1.218*(0.108) 1914 −1.699*(0.147) 1950 −1.799*(0.154) 1986 −1.859*(0.168) 1882 −1.527*(0.135) 1918 −1.433*(0.147) 1954 −1.773*(0.152) 1990 −1.989*(0.184) 1886 −1.619*(0.122) 1922 −1.640*(0.155) 1958 −1.833*(0.155) 1994 −2.038*(0.177) 1890 −1.667*(0.128) 1926 −1.685*(0.163) 1962 −1.757*(0.150) 1998 −2.102*(0.187) 1894 −1.612*(0.137) 1930 −1.734*(0.158) 1966 −1.671*(0.157) 2002 −2.299*(0.208) 1898 −1.575*(0.135) 1934 −1.596*(0.151) 1970 −1.616*(0.156) 2006 −2.244*(0.213) 1902 −1.672*(0.137) 1938 −1.563*(0.151) 1974 −1.755*(0.166) 2010 −2.085*(0.206) 1906 −1.710*(0.140) 1942 −1.792*(0.153) 1978 −1.746*(0.177) 2014 −2.162*(0.195)
*

p < .05.

Note: Panel-corrected standard errors are in parentheses.

TABLE 4.

GDP per capita coefficients including trade, 1922 to 2014 (Model 4)

 1922 1.446*(0.101) 1926 −0.089*(0.028) 1950 −0.314*(0.065) 1974 −0.144(0.105) 1998 −0.414*(0.125) 1930 −0.173*(0.037) 1954 −0.273*(0.065) 1978 −0.185(0.102) 2002 −0.599*(0.132) 1934 −0.061(0.045) 1958 −0.321*(0.066) 1982 −0.393*(0.108) 2006 −0.535*(0.122) 1938 −0.049(0.052) 1962 −0.229*(0.069) 1986 −0.293*(0.098) 2010 −0.359*(0.121) 1942 −0.246*(0.060) 1966 −0.111(0.076) 1990 −0.351*(0.124) 2014 −0.414*(0.115) 1946 −0.301*(0.061) 1970 −0.043(0.089) 1994 −0.383*(0.122)
 1922 1.446*(0.101) 1926 −0.089*(0.028) 1950 −0.314*(0.065) 1974 −0.144(0.105) 1998 −0.414*(0.125) 1930 −0.173*(0.037) 1954 −0.273*(0.065) 1978 −0.185(0.102) 2002 −0.599*(0.132) 1934 −0.061(0.045) 1958 −0.321*(0.066) 1982 −0.393*(0.108) 2006 −0.535*(0.122) 1938 −0.049(0.052) 1962 −0.229*(0.069) 1986 −0.293*(0.098) 2010 −0.359*(0.121) 1942 −0.246*(0.060) 1966 −0.111(0.076) 1990 −0.351*(0.124) 2014 −0.414*(0.115) 1946 −0.301*(0.061) 1970 −0.043(0.089) 1994 −0.383*(0.122)
*

p < .05.

Note: Panel-corrected standard errors are in parentheses.

Overall, the results for Model 3 (Table 3) suggest that there was a relative decoupling in the 13 global North nations over the 145-year span. The coefficient for GDP per capita decreased in magnitude from 2.775 in 1870 to 0.613 (2.775 – 2.162) in 2014. The results for Model 4 (Table 4) reflect a similar relative decoupling trend from 1922 to 2014, during which the GDP per capita coefficient decreased from 1.466 to 1.052 (1.446 – 0.414). However, in neither model was the decoupling consistent over time. In Model 3, there was a significant relative decoupling over the latter part of the nineteenth century, but the coefficient stabilized around the year 1890. The coefficient remained relatively constant through the early 1970s, at which time another episode of relative decoupling occurred. The results from Model 4 follow a similar pattern.

Though Models 3 and 4 were similar in their results, there are noticeable differences. The GDP per capita coefficient for each year tends to be smaller in Model 3 than in Model 4, which controlled for trade openness (see Figure 1 for a plot of the yearly coefficients for Models 3 and 4). And the interaction terms for six years (1934, 1938, 1966, 1970, 1974, 1978) are not statistically significant in Model 4, indicating that the GDP per capita coefficients for these years were not significantly different from the year 1922.

FIGURE 1.

GDP per capita coefficients for Models 3 and 4, 1870/1922 to 2014

FIGURE 1.

GDP per capita coefficients for Models 3 and 4, 1870/1922 to 2014

## DISCUSSION: RETHINKING TP AND EM

The results reveal several noticeable long-term and short-term patterns that do not align neatly with either the TP or the EM theory. The models that interacted economic growth with time showed that an overall pattern of relative decoupling occurred over time in global North countries, which supports arguments made by EM. However, in the previously under-studied period of 1870 to 1960, we also observed an episode of significant relative decoupling in the late nineteenth century. This finding is contrary to what both the TP and EM literatures might expect, since decoupling does not appear to be a phenomenon unique to the post-1960 period. Rather, decoupling goes through mini-cycles, which is similar to the findings of Jorgenson and Clark (2012) and Thombs (2018a). The early decoupling occurred concurrently with the Long Depression from 1873 to 1896. During this time, industrial production stagnated (Mitchell 2013a, 2013b), which stalled the expansion of coal-based energy systems (Arrighi 2010; Engels 1887; Podobnik 2006). This likely led to a reconstruction of the political economy, where money flowed out of the productive economy and into the financial sector, which augmented the power of financiers, propping up financial speculation and diverting investment away from the industrial sector (Arrighi 2010). These processes may have curtailed economic growth's impact on CO2 emissions in the short term.

However, as the Long Depression came to an end, the late nineteenth century through the 1970s witnessed a positive and consistent association between economic growth and CO2 emissions. The stable association is likely attributable to the extensive industrialization and mass consumption that occurred in the global North, which was largely powered by the massive expansion of oil-based production and consumption (Podobnik 2006). These countries substantially increased industrial output, achieved nearly universal electrification, and opened up access to motor vehicle ownership on an unprecedented scale during this period. This finding is consistent with assertions made by TP, which argues that the post–World War II period witnessed unparalleled ecological degradation as a result of the increase in capital accumulation in global North nations (Gould, Pellow, and Schnaiberg 2004; Schnaiberg 1980).

Nevertheless, economic growth began to decouple from CO2 emissions with the rise of a new wave of globalization coinciding with the economic and energy crises in a number of global North countries in the 1970s (Chase-Dunn, Kawano, and Brewer 2000; Harvey 1990; Marglin and Schor 1991). This observation aligns with propositions made by EM, but the decoupling may be due to structural changes in the global economy rather than the rise of an “ecological rationality.” We posit that this change is likely tied to shifts during this period in the global organization of production that relocated energy-intensive industries from the global North to the global South, particularly to emerging economies in Asia. For example, by 2014, China was responsible for over half of the world's most carbon-intensive industrial production (e.g., steel, iron, and cement), followed by India and a host of other Asian countries (U.S. Geological Survey 2015a, 2015b). As energy-intensive industries shifted to these nations, growth in energy consumption per capita in the global North remained relatively stagnant from the 1970s onward, growing a mere 4.6% from 1970 to 2014 (World Bank 2018b) (Figure 2).

FIGURE 2.

Energy use per capita (kg of oil equivalent per capita) for 13 countries in the global North, 1970 to 2014

FIGURE 2.

Energy use per capita (kg of oil equivalent per capita) for 13 countries in the global North, 1970 to 2014

However, there were notable changes in the structural makeup of the energy sector during this time as well. Countries such as France, Belgium, and Sweden partly shifted to nuclear energy, a carbon-neutral energy source at the site of production, and renewables and natural gas also gained a larger share in the energy mix. Figure 3 shows how the organization of the economy and the energy systems of these 13 nations have changed since 1970, illustrated by average industrialization (industry percentage of GDP) and the percentage of energy use obtained from non-fossil-fuel sources, including nuclear.12 This may be interpreted as evidence for EM, since these changes likely avoided additional CO2 emissions, but if we frame these transitions in a broader socio-ecological context, many of these changes have given rise to new ecological risks. For example, the switch to natural gas occurring in countries like the United States substantially increases methane emissions (Howarth, Santoro, and Ingraffea 2011), and the rise of the nuclear industry in the latter half of the twentieth century created new risks tied to waste disposal, nuclear war, and unsafe labor conditions (Cable, Shriver, and Mix 2008; Macfarlane 2011).

FIGURE 3.

Average industrialization (1870 to 2008) and non-fossil-fuel energy consumption (1970 to 2008) for 13 global North countries

FIGURE 3.

Average industrialization (1870 to 2008) and non-fossil-fuel energy consumption (1970 to 2008) for 13 global North countries

Nevertheless, the amalgam of these structural changes likely resulted in the relative decoupling observed after 1970. However, as we have stated, this is likely tied to structural shifts in the global organization of production and the amplification of new risks and ecological harms, rather than a result of “modernization” processes per se. It should also be noted that even though a relative decoupling has been observed since circa 1970, the magnitude of the GDP per capita coefficient remained large (0.613 in Model 3 and 1.052 in Model 4) as of 2014. This continuance of carbon-intensive economic growth aligns with the arguments made by TP.

Before concluding, several limitations of the study should be acknowledged. First, the data we analyzed may be prone to error, particularly for the earlier years in the analysis. Therefore, we advise readers to interpret the early-period findings with caution. However, the sources of data we used are the best currently available. Second, limited by data availability and the need for a balanced sample, we only analyzed data on 13 global North countries. Therefore, the findings cannot be generalized to the countries excluded from the analysis. Nevertheless, the relative decoupling trend observed since 1970 is similar to what has been found in previous decoupling studies (e.g., Jorgenson and Clark 2012; Thombs 2018a), which lends support to the validity of our findings.

## CONCLUSION

By examining the economic growth–CO2 emissions relationship back to 1870, this study adds to the growing body of literature that has tested the propositions made by TP and EM. Over the long run, the results support the arguments made by TP, given that the effect of economic growth on CO2 emissions remains large throughout the analysis. However, the observation of various episodes of decoupling suggests that TP does not fully explain the long historical relationship between economic growth and CO2 emissions. The post-1970 decoupling is consistent with the arguments made by EM. But, contrary to expectations of EM about earlier stages of industrialization, a substantial decoupling also occurred in the late nineteenth century. This finding casts doubt on the assertion that decoupling is tied to the rise of an ecological rationality or superindustrialization. Rather, the late-nineteenth-century and post-1970 episodes of decoupling correspond to significant concurrent changes in the global political economy. Thus, while it is not necessarily wrong to expect that most global North nations will experience decoupling as they continue to develop, this decoupling is likely due to changes occurring at transnational and global levels rather than within a nation-state's borders. In other words, decoupling in one place may be tied to intensification in another. Yet, it is easy to lose sight of this connection if we fail to situate the changes within a nation in a broader global context.

These findings also have implications for climate mitigation policy. First and foremost, focusing on climate policy at the national level is necessary but insufficient given the global scale of the economic system. This study has demonstrated, alongside previous decoupling studies (Jorgenson and Clark 2012; Thombs 2018a), that decoupling may be misleading if not discussed in relation to global political economic processes. In particular, if carbon-intensive processes are regulated only in certain regions, they might simply shift to other spaces in the global economy that lack effective regulation of CO2 emissions.

Moreover, solely emphasizing the deployment of renewable energy is likely to be an insufficient mitigation strategy. Nations surely have to switch to renewable energy, but renewable deployment alone is unlikely to curtail the use of fossil fuels. Studies have indicated that they simply add to the energy mix to fulfill additional energy demand, rather than replacing existent demand for fossil fuels (Sinn 2012; Thombs 2018a; York 2012). There is also evidence that renewable-energy growth may intensify economic growth's association with CO2 emissions by spurring further energy demand that is not fully met by renewables (Thombs 2017; York 2016). Therefore, countries and international treaties should prioritize curtailing fossil fuel use and energy demand at both national and transnational levels, along with the deployment of renewables.

Furthermore, reevaluating the goals of policymaking, particularly the focus on maximizing economic growth, is a necessary endeavor. There is a long history of critiquing the growth paradigm over environmental concerns (Dale 2012; Georgescu-Roegen 1970, 1972; Kallis 2018; Meadows et al. 1972), and as most of the rewards of economic growth become increasingly concentrated in the hands of the few, the social benefits of growth are also questionable (D'Alisa, Demaria, and Kallis 2015; Hirsch 1976; Kallis 2018; Mazur and Rosa 1974). Although increases in efficiency could theoretically reduce the resources and energy needed for production, empirical evidence suggests that producers may instead leverage these efficiency gains to increase production (e.g., York and McGee 2016). Thus, sufficiently mitigating CO2 emissions is likely contingent on countries transcending the growth paradigm and renegotiating human–natural system relations, encouraging the fulfillment of basic needs and sufficiency while limiting the need for an ever-growing throughput of resources and energy (Jorgenson 2014; O'Neill et al. 2018; Steinberger et al. 2012; Thombs 2019).

## REFERENCES

REFERENCES
Arrighi, Giovanni.
2010
.
The Long Twentieth Century: Money, Power and the Origins of Our Times
.
London
:
Verso
.
Barbieri, Katherine, and Omar M. G. Keshk.
2016
.
“Correlates of War Project Trade Data Set Codebook, Version 4.0.”
Barbieri, Katherine, Omar M. G. Keshk, and Brian M. Pollins.
2009
.
“Trading Data: Evaluating Our Assumptions and Coding Rules.”
Conflict Management and Peace Science
26
(
5
):
471
91
.
Baum, Christopher.
2006
.
An Introduction to Modern Econometrics Using Stata
.
College Station, TX
:
Stata Press
.
Beck, Nathaniel, and Jonathan N. Katz.
1995
.
“What to Do (and Not to Do) with Time-Series Cross-Section Data.”
American Political Science Review
89
(
3
):
634
47
.
Blunden, Jessica, and Derek S. Arndt.
2017
.
“State of the Climate in 2016.”
Bulletin of the American Meteorological Society
98
(
8
):
Si
-
S280
.
Bolt, Jutta, Robert Inklaar, Herman de Jong, and Jan Luiten van Zanden.
2018
.
“Rebasing ‘Maddison’: New Income Comparisons and the Shape of Long-Run Economic Development.”
University of Groningen
.
Cable, Sherry, Thomas E. Shriver, and Tamara L. Mix.
2008
.
“Risk Society and Contested Illness: The Case of Nuclear Weapons Workers.”
American Sociological Review
73
(
3
):
380
401
.
Chase-Dunn, Christopher, Yukio Kawano, and Benjamin D. Brewer.
2000
.
“Trade Globalization since 1795: Waves of Integration in the World-System.”
American Sociological Review
65
(
1
):
77
95
.
Cohen, Gail, Joao Tover Jalles, Prakash Loungai, and Ricardo Marto.
2018
.
The Long-Run Decoupling of Emissions and Output: Evidence from the Largest Emitters
.
International Monetary Fund
.
Dale, Gareth.
2012
.
International Socialism
(
134
).
D'Alisa, Giacomo, Federico Demaria, and Giorgos Kallis, eds.
2015
.
Degrowth: A Vocabulary for a New Era
.
Abingdon, Oxon
:
Routledge
.
Dietz, Thomas.
2017
.
“Drivers of Human Stress on the Environment in the Twenty-First Century.”
Annual Review of Environment and Resources
42
(
1
):
189
213
.
Engels, Friedrich.
1887
.
The Condition of the Working Class in England in 1844
.
New York
:
J. W. Lovell
.
Georgescu-Roegen, Nicholas.
1970
.
“The Entropy Law and the Economic Problem.”
Pp.
49
57
in
From Bioeconomics to Degrowth: Georgescu-Roegen's “New Economics” in Eight Essays
, edited by M. Bonaiuti.
London
:
Routledge
.
Georgescu-Roegen, Nicholas.
1972
.
“Energy and Economic Myths.”
Pp.
58
92
in
From Bioeconomics to Degrowth: Georgescu-Roegen's “New Economics” in Eight Essays
, edited by M. Bonaiuti.
London
:
Routledge
.
Givens, Jennifer E., Xiaorui Huang, and Andrew K. Jorgenson.
2019
.
“Ecologically Unequal Exchange: A Theory of Global Environmental Injustice.”
Sociology Compass
13
(
5
):
e12693
.
Gould, Kenneth A., David N. Pellow, and Allan Schnaiberg.
2004
.
Organization & Environment
17
(
3
):
296
316
.
Haines, Michael R., and Richard Sutch.
2006
.
“Population: 1790–2000 [Annual Estimates], Table Aa6-8.”
In
Historical Statistics of the United States, Earliest Times to the Present: Millennial Edition
, edited by S. B. Carter, S. S. Gartner, M. R. Haines, A. L. Olmstead, R. Sutch, and G. Wright.
New York
:
Cambridge University Press
.
Harlan, Sharon L., David N. Pellow, and J. Timmons Roberts.
2015
.
“Climate Justice and Inequality.”
Pp.
127
63
in
Climate Change and Society: Sociological Perspectives
, edited by R. E. Dunlap and R. J. Brulle.
Oxford
:
Oxford University Press
.
Harvey, David.
1990
.
The Condition of Postmodernity: An Enquiry into the Origins of Culture Change
.
Malden, MA
:
Blackwell
.
Hirsch, Fred.
1976
.
Social Limits to Growth
.
Cambridge, MA
:
Harvard University Press
.
Howarth, Robert W., Renee Santoro, and Anthony Ingraffea.
2011
.
“Methane and the Greenhouse-Gas Footprint of Natural Gas from Shale Formations.”
Climatic Change
106
(
4
):
679
.
Huang, Xiaorui.
2018
.
“Ecologically Unequal Exchange, Recessions, and Climate Change: A Longitudinal Study.”
Social Science Research
73
:
1
12
.
Huber, Joseph.
1982
.
Die verlorene Unschuld der Ökologie. Neue Technologien und superindustrielle Entwicklung [The lost innocence of ecology: new technologies and superindustrial development]
.
Frankfurt
:
Fisher
.
Huber, Joseph.
1985
.
Die Regenbogengesellschaft. Ökologie und Sozialpolitik [The rainbow society: ecology and social policy]
.
Frankfurt
:
Fisher
.
IPCC
.
2013
.
“Summary for Policymakers.”
In
Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change
.
IPCC
.
2014a
.
Climate Change 2014 Synthesis Report Summary for Policymakers
.
IPCC
.
2014b
.
“Summary for Policymakers.”
In
Climate Change 2014: Mitigation of Climate Change. Contribution of Working Group III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change
.
Jevons, William S.
1866
.
The Coal Question: An Enquiry Concerning the Progress of the Nation, and the Probable Exhaustion of Our Coal-Mines
.
London
:
Macmillan
.
Jorgenson, Andrew K.
2014
.
“Economic Development and the Carbon Intensity of Human Well-Being.”
Nature Climate Change
4
(
3
):
186
89
.
Jorgenson, Andrew K., and Brett Clark.
2010
.
“Assessing the Temporal Stability of the Population/Environment Relationship in Comparative Perspective: A Cross-National Panel Study of Carbon Dioxide Emissions, 1960–2005.”
Population and Environment
32
(
1
):
27
41
.
Jorgenson, Andrew K., and Brett Clark.
2011
.
“Societies Consuming Nature: A Panel Study of the Ecological Footprints of Nations, 1960–2003.”
Social Science Research
40
(
1
):
226
44
.
Jorgenson, Andrew K., and Brett Clark.
2012
.
“Are the Economy and the Environment Decoupling? A Comparative International Study, 1960–2005.”
American Journal of Sociology
118
(
1
):
1
44
.
Kallis, Giorgos.
2018
.
Degrowth
.
UK
:
Agenda
.
Knight, Kyle W., Eugene A. Rosa, and Juliet B. Schor.
2013
.
“Could Working Less Reduce Pressures on the Environment? A Cross-National Panel Analysis of OECD Countries, 1970–2007.”
Global Environmental Change
23
(
4
):
691
700
.
Knight, Kyle W., and Juliet B. Schor.
2014
.
“Economic Growth and Climate Change: A Cross-National Analysis of Territorial and Consumption-Based Carbon Emissions in High-Income Countries.”
Sustainability
6
(
6
):
3722
31
.
Lewis, Tammy L.
2018
.
Environmental Sociology
1
13
.
Liddle, Brantley, and George Messinis.
2018
.
“Revisiting Carbon Kuznets Curves with Endogenous Breaks Modeling: Evidence of Decoupling and Saturation (but Few Inverted-Us) for Individual OECD Countries.”
Empirical Economics
54
(
2
):
783
98
.
Longhofer, Wesley, and Andrew K. Jorgenson.
2017
.
“Decoupling Reconsidered: Does World Society Integration Influence the Relationship between the Environment and Economic Development?”
Social Science Research
65
:
17
29
.
Macfarlane, Allison M.
2011
.
“The Overlooked Back End of the Nuclear Fuel Cycle.”
Science
333
(
6047
):
1225
26
.
Marglin, Stephen A., and Juliet B. Schor, eds.
1991
.
The Golden Age of Capitalism: Reinterpreting the Postwar Experience
.
New York
:
Oxford University Press
.
Mazur, Allan, and Eugene Rosa.
1974
.
“Energy and Life-Style.”
Science
186
(
4164
):
607
10
.
McGee, Julius Alexander, Matthew Thomas Clement, and Jordan Fox Besek.
2015
.
“The Impacts of Technology: A Re-evaluation of the STIRPAT Model.”
Environmental Sociology
1
(
2
):
81
91
.
Meadows, Donella H., Dennis L. Meadows, Jorgen Randers, and William W. Behrens.
1972
.
The Limits to Growth: A Report for the Club of Rome's Project on the Predicament of Mankind
.
New York
:
Universe Books
.
Mitchell, Brian R.
2013a
.
“D1 Europe: Indices of Industrial Production.”
In
International Historical Statistics, 1750-2010
.
Basingstoke
:
Palgrave Macmillan
.
Mitchell, Brian R.
2013b
.
“D1 North America: Indices of Industrial Production.”
In
International Historical Statistics, 1750–2010
.
Basingstoke
:
Palgrave Macmillan
.
Mol, Arthur P. J.
1995
.
The Refinement of Production: Ecological Modernization Theory and The Chemical Industry
.
Utrecht
:
Van Arkel
.
Mol, Arthur P. J.
2002
.
“Ecological Modernization and the Global Economy.”
Global Environmental Politics
2
(
2
):
92
115
.
Mol, Arthur P. J., and Gert Spaargaren.
2000
.
“Ecological Modernisation Theory in Debate: A Review.”
Environmental Politics
9
(
1
):
17
49
.
O'Neill, Daniel W., Andrew L. Fanning, William F. Lamb, and Julia K. Steinberger.
2018
.
“A Good Life for All within Planetary Boundaries.”
Nature Sustainability
1
(
2
):
88
95
.
Peters, Glen P., and Edgar G. Hertwich.
2008
.
“CO2 Embodied in International Trade with Implications for Global Climate Policy.”
Environmental Science & Technology
42
(
5
):
1401
07
.
Peters, Glen P., Jan C. Minx, Christopher L. Weber, and Ottmar Edenhofer.
2011
.
“Growth in Emission Transfers via International Trade from 1990 to 2008.”
Proceedings of the National Academy of Sciences
108
(
21
):
8903
08
.
Podobnik, Bruce.
2006
.
Global Energy Shifts: Fostering Sustainability in a Turbulent Age
.
:
Temple University Press
.
Rosa, Eugene A., and Thomas Dietz.
2012
.
“Human Drivers of National Greenhouse-Gas Emissions.”
Nature Climate Change
2
(
8
):
581
86
.
Rosa, Eugene A., Richard York, and Thomas Dietz.
2004
.
“Tracking the Anthropogenic Drivers of Ecological Impacts.”
AMBIO
33
(
8
):
509
12
.
Sahr, Robert.
2017
.
“Individual Year Conversion Factor Tables.”
School of Public Policy, Oregon State University
. Retrieved March 5, 2018 (https://liberalarts.oregonstate.edu/spp/polisci/faculty-staff/robert-sahr/inflation-conversion-factors-years-1774-estimated-2024-dollars-recent-years/individual-year-conversion-factor-table-0).
Schnaiberg, Allan.
1980
.
The Environment: From Surplus to Scarcity
.
New York
:
Oxford University Press
.
Sinn, Hans-Werner.
2012
.
.
Cambridge, MA
:
MIT Press
.
Spaargaren, Gert, and Arthur P. J. Mol.
1992
.
“Sociology, Environment, and Modernity: Ecological Modernization as a Theory of Social Change.”
Society & Natural Resources
5
(
4
):
323
44
.
Steffen, Will, Wendy Broadgate, Lisa Deutsch, Owen Gaffney, and Cornelia Ludwig.
2015
.
“The Trajectory of the Anthropocene: The Great Acceleration.”
Anthropocene Review
2
(
1
):
81
98
.
Steinberger, Julia K., J. Timmons Roberts, Glen P. Peters, and Giovanni Baiocchi.
2012
.
“Pathways of Human Development and Carbon Emissions Embodied in Trade.”
Nature Climate Change
2
(
2
):
81
85
.
Thombs, Ryan P.
2017
.
“The Paradoxical Relationship between Renewable Energy and Economic Growth: A Cross-National Panel Study, 1990–2013.”
Journal of World Systems Research
23
(
2
):
540
64
.
Thombs, Ryan P.
2018a
.
“The Transnational Tilt of the Treadmill and the Role of Trade Openness on Carbon Emissions: A Comparative International Study, 1965–2010.”
Sociological Forum
33
(
2
).
Thombs, Ryan P.
2018b
.
“Has the Relationship between Non-Fossil Fuel Energy Sources and CO2 Emissions Changed over Time? A Cross-National Study, 2000–2013.”
Climatic Change
148
(
4
):
481
90
.
Thombs, Ryan P.
2019
.
“When Democracy Meets Energy Transitions: A Typology of Social Power and Energy System Scale.”
Energy Research & Social Science
52
:
159
68
.
U.S. Census Bureau
.
2012
.
“Census of Population and Housing, Population and Housing Unit Counts, CPH-2-1, United States Summary: 2010.”
Retrieved December 21, 2017 (https://www.census.gov/prod/cen2010/cph-2-1.pdf).
U.S. Census Bureau
.
2016
.
“National Intercensal Datasets: 2000-2010.”
Retrieved December 21, 2017 (https://www.census.gov/data/datasets/time-series/demo/popest/intercensal-2000-2010-national.html).
U.S. Geological Survey
.
2015a
.
Cement
.
U.S. Geological Survey
.
2015b
.
Iron and Steel
.
World Bank
.
2018a
.
“Indicators.”
Retrieved May 24, 2018 (https://data.worldbank.org/indicator).
World Bank
.
2018b
.
“Energy Use (kg of Oil Equivalent per Capita)”
(https://data.worldbank.org/indicator/EG.USE.PCAP.KG.OE?locations=CN-XT).
World Resources Institute and Center for Advanced Infrastructure and Transportation
.
2018
.
“Climate Analysis Indicators Tool: WRI's Climate Data Explorer.”
Retrieved March 5, 2018 (http://cait2.wri.org).
York, Richard.
2012
.
“Do Alternative Energy Sources Displace Fossil Fuels?”
Nature Climate Change
2
(
6
):
441
43
.
York, Richard.
2016
.
“Decarbonizing the Energy Supply May Increase Energy Demand.”
Sociology of Development
2
(
3
):
265
72
.
York, Richard, and Julius Alexander McGee.
2016
.
Environmental Sociology
2
(
1
):
77
87
.
York, Richard, Eugene A. Rosa, and Thomas Dietz.
2003a
.
“Footprints on the Earth: The Environmental Consequences of Modernity.”
American Sociological Review
68
(
2
):
279
300
.
York, Richard, Eugene A Rosa, and Thomas Dietz.
2003b
.
“STIRPAT, IPAT and ImPACT: Analytic Tools for Unpacking the Driving Forces of Environmental Impacts.”
Ecological Economics
46
(
3
):
351
65
.

## NOTES

NOTES
1.
The carbon cycle consists of the biogeochemical processes by which carbon is cycled through the atmosphere, living organisms, soil, rocks, and oceans of the Earth.
2.
We refer to the industrialized countries in Western Europe and North America as the global North.
3.
The decoupling literature uses the terms decoupling and intensification to describe the economic growth–CO2 emissions relationship. Decoupling can be either relative or absolute. Relative decoupling is when the association between an ecological harm and an economic indicator lessens over time but remains positive. Absolute decoupling is when the association between an ecological harm and an economic indicator is zero or becomes negative over time. Intensification is when the association increases over time.
4.
Broadly, world society theory examines “the role of global institutional structures in influencing social change and environmental outcomes” (Longhofer and Jorgenson 2017:19).
5.
Outside of the sociological literature, recent papers by Cohen et al. (2018) and Liddle and Messinis (2018) have examined the long-run relationship between economic growth and CO2 emissions. Cohen et al. examined decoupling for a handful of countries since the nineteenth century, and found evidence for decoupling. Liddle and Messinis also examined this period, seeking to test for a carbon Kuznets curve and finding mixed results. Both studies used time-series analysis rather than panel data.
6.
Data for CO2 emissions from land use change and forestry were not available for most of the period of interest. Nevertheless, we acknowledge that land use change is an important source of greenhouse gas emissions (IPCC 2013).
7.
The Maddison Project builds on the work of Angus Maddison and provides estimates for income and other economic indicators for nations over time. The project is now housed at the Groningen Growth and Development Centre at the University of Groningen.
8.
EM argues that trade may lead to the emergence of more environmentally conscious production and consumption, though not definitively, as certain economic interests may impede it (Mol 2002).
9.
One such predictor we exclude is urbanization (Jorgenson and Clark 2012; Thombs 2018a). In addition to the lack of data, we find the typical urbanization measure (percentage of the population living in urban areas) to be problematic to use because the definition of “urban” has changed over the period analyzed and varies across countries. The measure is also highly correlated with population. For example, using census data from the United States (Haines and Sutch 2006; U.S. Census Bureau 2012, 2016) we calculated the correlation coefficient between population and urbanization to be 0.96 over the period 1870 to 2010. Thus, we believe that the three predictors included in this study, along with the two-way fixed effects, enable a relatively robust analysis.
10.
We found autocorrelation to be present using the xtserial command (Wooldridge test) in Stata 15.
11.
Jorgenson and Clark (2010, 2011, 2012) were the first to advance this approach to examine decoupling over time.
12.
Industrial data from Mitchell (2013a, 2013b), with the exception of Switzerland, whose data are from the World Bank (2018a). The non-fossil-fuel energy data include renewables and nuclear and are also from the World Bank (2018a).