Abstract
This study examines changes in some key indicators among 66 countries on six continents over a 56-year period, to compare the power of economic growth to improve human health and income distribution with its tendency to degrade the natural environment. The results indicate that growth depletes and pollutes nature far more than it benefits society. This suggests that public policy should shift toward enhancement of individual and social well-being in ways more direct and effective, and less ecologically damaging, than reliance on overall growth in gross domestic product. I illustrate this implication with a degrowth scenario for the United States to 2050 that draws on the empirical results for the period 1961 to 2016. And I consider certain reforms in the management and governance of organizations to implement such a scenario.
Keywords
Introduction
Since 1970, wild vertebrate populations have fallen by an average of 2% per year, leading to cumulative losses of 60% by 2014, and pointing toward further collapse to a mere one third of their initial abundances by the year 2020. The main causes of this global breakdown are humanity’s ongoing depletion of our fellow species’ habitats and other resources and addition of greenhouse gas and other pollutants. In turn, such harms result from expanding human population size and per-capita production and consumption—the two basic aspects of economic growth (World Wide Fund for Nature, 2018). Yet many view such growth as essential for enhancing human well-being. This prompts inquiry into the relative potency of growth for yielding social goods versus environmental “bads.”
The idea that economies should grow has deep historical and ideological roots. For example, Adam Smith argued that growth made “the great body of the people” “happiest and the most comfortable,” whereas a steady-state economy neither expanding nor contracting was “hard” on a country’s populace, and a shrinking economy “miserable” for them (Smith, 1776, Book 1, Chapter 8). Over the quarter millennium since Smith’s day, the degree to which human health has improved in different countries has indeed correlated with their respective rates of per capita economic growth (Bourguignon & Morrisson, 2002; Linden & Ray, 2017). However, more recent growth in some countries’ consumption and production per capita has actually diminished the health of their citizens (Ruhm, 2016; Tapia Granados & Ionides, 2008).
Smith also claimed that capitalism distributes necessary goods equitably: “The rich . . . are led by an invisible hand to make nearly the same distribution of the necessaries of life, which would have been made, had the earth been divided into equal portions among all its inhabitants” (Smith, 1759, Book 4, Chapter 1). Recent work on inequality and its detriments to health has refuted this proposition. For example, Piketty and Saez (2014) empirically disproved the “Kuznets curve” hypothesis that past a certain level of gross domestic product (GDP) per capita, further growth naturally reduces the gap between rich and poor. Meanwhile, health researchers have shown that inequality damages the well-being of individuals and societies (Pickett & Wilkinson, 2015). Piketty and Saez nevertheless theorized that among rich and poor societies alike, growth slows down the increase of inequality. Jackson and Victor (2016) challenged this hypothesis on theoretical grounds, and other studies have challenged it on empirical grounds (Mikkelson, 2017; Rubin & Segal, 2015).
Finally, Smith expressed no concern about the environmental degradation wrought by economic expansion and instead opined that putting every bit of suitable land under cultivation would yield “the greatest of all public advantages” (Smith, 1776, Book 1, Chapter 11). Below, I examine whether another Kuznets hypothesis—the environmental Kuznets curve or EKC—might help excuse Smith’s lack of concern. Whereas Kuznets himself formulated the hypothesis about inequality mentioned above, it was others, inspired by Kuznets, who devised the EKC. The EKC predicts that while growth increases pollution and other ecological harms up to a certain level of GDP per capita, further growth beyond that level reduces those harms. However, the best-cited review of work on this topic concluded that “EKC results have a very flimsy statistical foundation” (Stern, 2004).
Tracking Recent Economic Growth, Health Improvement, Rising Inequality, and Environmental Degradation
Previous studies, such as those cited above, have separately examined empirical links between economic growth and human health, economic inequality, and environmental damage. But methodological differences impede use of these studies to rigorously compare the relative strengths of these relationships. This article therefore applies a common statistical framework to all three, in order to facilitate such comparison. I analyze 66 countries around the world over the 56-year period from 1961 to 2016. 1 This sample ranges across six continents and a wide spectrum of income levels, from Bangladesh’s 2016 GDP per capita of $3,959 to Luxembourg’s $103,764.
To increase the resolution of the analysis, I break down GDP growth into four widely studied components: the rates of change in population size, employment (the percentage of the total population with paying jobs), average yearly hours worked per employee, and labor productivity (the dollar value of goods and services produced per “man”-hour of work). Multiplying population, employment, working hours, and productivity all together yields overall GDP. I use life expectancy to indicate individual human well-being, because it is perhaps the most comprehensive, widely reported measure of it. I employ the Gini index of net household income inequality as the most commonly tallied kind of economic inequality, and a social-level indicator of human “ill-being.” Finally, I use the “ecological footprint” as the most thorough metric of environmental damage (see below for an explanation).
In what follows, I first describe the data and statistical tools used for this comparison of social versus environmental impacts of recent economic growth. I then report the results of the analysis and discuss their scientific implications for the pro- and antigrowth ideas mentioned above in the Introduction section. Finally, I spell out some policy implications of the study, through a macroeconomic scenario for social and environmental improvement in the United States and a microeconomic application of the analysis and scenario to business organizations.
Data and Analysis
This study draws on four sources of data. The Conference Board (CB) compiles information about population size, employment, working hours, and labor productivity (CB, 2019). The CB productivity statistics correct for both inflation over time and differential purchasing power between countries (2018 US$, purchasing power parity). As for the dependent variables, data on life expectancy at birth come from the World Bank (2019), income inequality from the Standardized World Income Inequality Database (SWIID), and ecological footprints from the Global Footprint Network (GFN, 2018). The SWIID is perhaps the best data set of income inequality currently available, with a standardized Gini index of net household income inequality for many different countries in many different years (Solt, 2016). This index measures the disparity of income between households, after accounting for taxes paid to and benefits received from governments. It ranges from 0 (perfect equality of income) to 100 (maximal inequality). The SWIID corrects this index for variation in the way the original data were collected and processed.
The GFN’s “ecological footprint” quantifies the renewable resources depleted, and carbon emitted, to produce the goods consumed within any given country, and support the built infrastructure there. The ecological footprint makes different categories of depletion and pollution commensurate in terms of standardized “global hectares” (gha). This is the area of forests, fields, farms, and fisheries that would be required to renew the resources depleted, and absorb the carbon emitted, in order to sustain a country’s current level of consumption. The GFN compares this measure of a country’s demand for ecological sources and sinks with the corresponding measure of the country’s supply of them, termed biocapacity—the gha of the country’s biologically productive land and water. The ecological footprint is perhaps the most comprehensive measure of environmental damage that is widely available (Borucke et al., 2013). It combines in a principled way the information supplied by other such measures, such as greenhouse gas emissions, deforestation, overfishing, agricultural exploitation, and urban sprawl.
To facilitate comparison of links between economic growth and social versus environmental impacts, I focus on the countries for which estimates of all seven underlying variables are available: population size, the employment rate, working hours, labor productivity, life expectancy, income inequality, and ecological footprints. This criterion includes 67 countries. However, I exclude China because the CB singles it out as having unreliable official estimates, whereas the CB’s alternative estimates for that country do not include its working hours or labor productivity. These data form an unbalanced panel, with 66 countries, 56 years (1961, 1962, . . ., 2016), and a total of 2,406 observations.
For each dependent variable, I performed a panel regression of first differences of the natural logarithms of the raw variables, while controlling for initial values of the underlying dependent variable as well as of GDP per capita. These latter two controls allow for the possibilities that high values may limit further increases and that rich versus poor countries may respond differently to further economic growth. First-differencing transforms the panel into one with the 66 countries referenced above, 55 time intervals (1961-1962, 1962-1963, . . ., 2015-2016), and 2,340 observations (combinations of a country and a time interval).
First-differencing helps control for unmeasured peculiarities of different countries that might otherwise confound the results (Croissant & Millo, 2008). It also entails that the main variables in the regression are changes from year to year, rather than values in a given year. This is important, because recent economic growth may relate differently to changes over time in certain variables than correlations at a single point in time suggest. For example, I noted above that recent growth has resulted in health declines in certain countries. This contrasts with the legacy of past economic growth and health improvement: a positive correlation between GDP per capita and life expectancy across different countries in any one single year.
Using logged variables means that the slope estimates are elasticities (Bailey, 2015). For each underlying dependent variable and component of GDP, the elasticity is the percent acceleration or deceleration of the variable that is associated with a 1% acceleration in the GDP component. For example, ecological footprints rise 1.3% faster for each 1% acceleration in working hours (see below). To check whether longer term results concur with those for 1-year periods, I redid the above analysis with first differences averaged over the entire temporal range within each country. Finally, in the course of discussing the main results, I report some other analyses of the same general type as that described above (panel regressions of first differences of logged variables).
Comparing Growth’s Links With Longevity, Inequality, and Footprints
Table 1 displays the short-term results. The top three rows of numbers indicate the changes from 1 year to the next in life expectancy, income inequality, and ecological footprints that are independent of any growth component. The top row—of y-intercepts—shows that independent of growth, longevity and inequality both tend to rise from year to year (these intercepts are positive and statistically significant). But the second row tells us that life expectancy and income inequality increase more slowly in times and places with higher initial levels of these variables (the slopes of these initial values are significantly negative). The estimate in the middle of the third row means that inequality also rises more slowly where and when initial GDP per capita is higher. In contrast, initial GDP per capita has no discernible effect on subsequent changes in longevity. For footprints, neither the y-intercept nor the slopes of initial footprints or GDP per capita are statistically significant.
Short-Term Analysis.
Note. For each model corresponding to a different dependent variable, n = 66 countries, T = 5 to 55 years (1961-1962, 1962-1963, . . ., 2015-2016), and N = 2,340 observations. The term ln means “the natural logarithm of.” “Initial value” means the level at the beginning of a given year of life expectancy, income inequality, or ecological footprints for Model 1, 2, or 3, respectively. The “growth” or “change” in any given underlying variable is measured by the difference between the log of its value at the end versus the beginning of a given year. “GDP” stands for gross domestic product. Statistically significant parameter estimates are in bold (p < .05), and standard errors in parentheses.
The fourth through seventh rows of statistics show that economic growth relates more strongly to changes in ecological footprints than to changes in either life expectancy or income inequality. Once initial longevity is taken into account, no growth component has any significant relationship to changes in life expectancy. The employment rate and labor productivity both do relate negatively to income inequality. But the effects are small compared with the links between GDP components and footprints. Ecological footprints accelerate significantly with all four of those components, and by greater percentages than income inequality’s deceleration with any GDP component.
To check whether long-term relationships between economic growth and changes in longevity, inequality, and/or footprints differ markedly from the short-term relationships indicated above, I repeated the first-difference panel regressions for long-term (up to 55-year) averages within countries (Table 2). In the long term as in the short, all links between growth components and changes in ecological footprints have greater absolute values than any links between growth components and changes in life expectancy or income inequality. However, in the longer term, life expectancy accelerates, albeit weakly, with labor productivity. All significantly negative relationships between GDP components and inequality disappear, replaced by a stronger—and positive—relationship between working hours and income inequality.
Long-Term Analysis.
Note. For each model, n = 66 countries. The term ln means “the natural logarithm of.” “Initial value” means the earliest level for a given country in the data set of life expectancy, income inequality, or ecological footprints for Model 4, 5, or 6, respectively. The “growth” or “change” in any given underlying variable is the average, over all years with data for a given country, of the difference between the log of its value in 1 year versus the previous year. “GDP” stands for gross domestic product. Statistically significant parameter estimates are in bold (p < .05), and standard errors in parentheses.
Scientific Implications
The results above indicate that economic growth has much stronger connections to environmental degradation than to human well-being. Figure 1 illustrates the contrast. Life expectancy and income inequality vary little with growth in GDP, compared with the much greater expansion of ecological footprints along with such growth.
Long-term relationships between economic growth and changes over time in life expectancy, income inequality, and ecological footprints.
The statistics reported in Table 1 provide no support for Smith’s claims about the benefits of economic growth (Smith, 1776). Model 1 picks up on the general trends toward improving life expectancy regardless of growth, and slower improvement once high levels of longevity have been attained. But with these trends accounted for, changes in life expectancy from year to year have no significant connections, positive or negative, with how fast population size, the employment rate, working hours, or labor productivity increase or decrease. Longer term patterns of up to 55 years reveal a positive correlation between changes in productivity and longevity. But the effect size is more than 14 times weaker than even the weakest link between any growth component and changes in footprints. For each 1% acceleration in labor productivity, ecological footprints accelerate by 0.68%, whereas life expectancy accelerates by less than 0.05% (see Table 2).
The results relate ambivalently to the Piketty–Saez hypothesis that growth limits inequality. In the short term, employment and productivity have small, inverse links with income inequality (Table 1). But in the long term, neither of these negative correlations hold up, whereas working hours correlate positively with inequality, and with an elasticity more than four times greater in absolute value than the one governing either of the short-term links between growth and inequality (Table 2). Overall, the results qualify and refine previous challenges to the predicted negative link between economic growth and income inequality (Mikkelson, 2017; Jackson & Victor, 2016; Rubin & Segal, 2015).
Compared with the nonexistent, weak, and/or ambivalent connections between GDP and the human well-being indicated by life expectancy and income distribution, the analysis presented here confirms a much more robust link between economic growth and the environmental damage indicated by the ecological footprint. All four components of GDP relate positively and significantly to footprints, with elasticities whose absolute values surpass those characterizing any component’s relationship to longevity or inequality. Furthermore, this applies at both time scales studied herein: short intervals of 1 year and longer ones of up to 55 years. I now elaborate the implications of these growth–footprint links in terms of the EKC cited above.
The EKC hypothesizes that after societies reach a certain level of GDP per capita, they reduce their environmental impacts even while their economies continue to grow. The data analyzed herein provide some support for this idea but not in a way that justifies economic growth as a path to environmental protection. Among these 66 countries, 55 1-year intervals, and 2,340 observations, the rate at which ecological footprints expand correlates negatively with the natural logarithm of GDP per capita at the beginnings of the time intervals (R2 = .02, p of log(GDP per capita) < 10−12). However, this bivariate model predicts countries to stop expanding their footprints, and start lowering them, only at a GDP per capita of $54,541 (2018 US$, purchasing power parity). This exceeds GDP per capita in the vast majority of countries in our sample. Given this, and the fact that the world’s ecological footprint already greatly overshoots global biocapacity, it is not a viable option to simply wait for the EKC to start reducing environmental degradation at some hypothetical point lying far in the future (Kitzes et al., 2008).
Moreover, the negative link between GDP per capita at a given point in time and subsequent expansion of ecological footprints over time is completely explained by the fact that richer countries tend to have slower rates of economic growth. Models 3 and 6 reported above show that after controlling for the four growth components, GDP per capita has no significant positive or negative connection with changes in ecological footprints (see Tables 1 and 2). This suggests that if a country—rich or poor—is to reduce its ecological footprint, its economy cannot grow very fast. Indeed, in 69% of the countries and years in which footprints declined, GDP grew more slowly than the sample-wide median rate of 3.6% per year. Conversely, in 77% of the country–year combinations with faster growth than that median rate, footprints increased. Thus in most times and places, economic growth more than canceled out any efficiency gains that may have driven down ecological damage per dollar of GDP (cf. York, 2006).
Policy Implications
The above considerations imply that governments must abandon the conventional goal of growing their economies by several percentage points each year and seek other, more effective, less ecologically harmful ways to improve human well-being. This shift is necessary to relieve and reverse the devastation of wild nature noted in the Introduction above. It is also needed to avert the environmental and economic collapse to come if humanity’s ecological footprint continues to overshoot the planet’s biocapacity. Public programs to satisfy human needs and aspirations offer a proven way to enhance well-being without relying on economic growth (Sen, 2001). The analysis above suggests that policies targeting specific components of growth could also help optimize individual and social welfare while reducing environmental impacts.
To illustrate this implication, consider an optimized “degrowth” scenario for the United States. In 2016, that country’s total ecological footprint of 2.6 billion gha stood at more than twice its own biocapacity of 1.2 billion gha. Moreover, its footprint per capita of 8.1 gha amounted to almost five times the global “fair-share one-Earth” level of 1.7 gha per person (Vale & Vale, 2013). The following scenario would bring the U.S. ecological footprint back to within its own biocapacity by 2050 while enhancing both human health and income distribution. In this scenario, birth and/or immigration rates fall to the point where they together just balance the sum of death and emigration rates. This brings overall population growth to zero, as currently experienced in countries such as Spain and South Korea. The employment ratio grows slowly, at a rate of 0.15% per year, and labor productivity improves by 1.00% per year—a bit faster than it did from 2010 through 2016. Crucially, however, all of that increased productivity translates into increased leisure time rather than increased overall production and consumption, as working hours fall by 1.0% per year (cf. Russell, 1932).
Applying the long-term parameter estimates from Table 2 to these altered rates of change yields the following projection. By 2050, the ecological footprint of the United States falls back to 1.2 billion gha, thus matching its own biocapacity. Its footprint per capita falls to 3.9 gha in 2050—a dramatic reduction from its 2016 level, albeit to a level still more than twice the fair-share one-Earth per-capita footprint. Meanwhile, U.S. life expectancy improves from 79 years in 2016 to 83 years in 2050, and its Gini index of net household income inequality improves from 38 to 35, thus reversing the steady rise of inequality afflicting the country since the 1970s.
This scenario must, of course, be taken with a grain of salt, like any projection into the future. Still, it highlights several important features of reality. The ecological disasters looming in the 21st century require governments to abandon their 20th-century fixation on growth in GDP. People in rich societies like the United States—as well as most other countries in the sample of 66 countries used in this study—must pressure their governments to rein their economies back to within sustainable ecological limits. The above analyses indicate that this can occur while enhancing both human health and economic equality.
Implications for the Management and Governance of Organizations
The statistical analysis and future scenario reported above also shed light on how organizations, including business firms, could help shrink the economy back to within ecologically healthy limits. Businesses could lead society to a new era in which gains in labor productivity translate into more leisure time rather than exponentially increasing production and consumption. The regressions reported above indicate that this would relieve our unjust and unsustainable demands on the biosphere. After controlling for initial ecological footprints and GDP per capita, and subsequent growth in population size, the employment rate, and labor productivity, footprints decelerate by more than 1% for each 1% deceleration in working hours (see Models 3 and 6 in Tables 1 and 2).
Even if growth in population, employment, and productivity are dropped from the regressions, footprints still decelerate by more than 0.8% for each 1% deceleration in work hours (R2 = .04, p of growth in working hours < 10−11). This confirms the findings of Knight, Rosa, and Schor (2013). Their analysis, like the present one, suggests that people working less tend “to engage in more self-sufficient activities (e.g., gardening) or time-intensive, low-impact activities such as walking and biking to work instead of driving.” These tendencies compensate for whatever inclinations people with more leisure time might have to “take more vacations by auto or air . . . or have greater involvement in . . . shopping” or “other energy consuming activities” (p. 694).
A few historical examples demonstrate the possibilities. One of them is the Kellogg cereal company, which moved all its workers from 8- to 6-hour days in 1930. The owner and president of this company wished not only to relieve the unemployment setting in at the start of the Great Depression but also to show that the “free exchange of goods, services, and labor in the free market would not have to mean mindless consumerism or eternal exploitation of people and natural resources.” Instead “workers would be liberated by increasingly higher wages and shorter hours for the final freedom promised by the Declaration of Independence—the pursuit of happiness.” (Hunnicutt, 1996, as cited in Kaplan, 2008)
The company, its employees, and the surrounding community all flourished as a result of this vision and policy. However, few other businesses followed Kellogg’s lead, and for the most part, the vision lost out to that of “Full-Time, Full Employment and ‘Salvation by Work’” (Hunnicutt, 2013).
Could the owners, directors, and managers of business corporations be convinced to change this tune? One incentive might be that shorter working hours enhance labor productivity. In our sample of 66 countries and 55 time intervals, and after controlling for initial productivity and GDP per capita, and growth in population size and the employment rate, productivity accelerates by 0.53% for each 1% deceleration in work hours (R2 = .15, p of growth in working hours < 10−15). This suggests that, other things being equal, a company could produce more by hiring more workers with each putting in fewer hours. 2 However, other things are never equal, and many of them drive employers to increase rather than decrease hours worked per employee. Booth (2004) cited the fixed costs of health and other kinds of insurance, defined benefit retirement programs, and on-the-job training. Coote, Franklin, and Simms (2010) thus recommended a number of public policy changes to ensure that “employers’ costs . . . reward rather than penalise taking on extra staff” and to promote “flexible arrangements to suit employees, such as job sharing, extended care leave and sabbaticals” (p. 3).
But even if new laws altered employers’ costs to favor shorter working hours, their long-term political and financial interests might still lead them to resist a transition to a time-rich society. This is because long hours maximize workers’ cost of job loss, and therefore facilitate the firm’s control. Long hours also increase the size of the unemployment pool, which improves the terms on which firms can hire labor. (Schor, 1991, p. 78)
In other words, longer hours empower employers over employees. This implies that if the employees were the employer—as they are in worker cooperatives—workers would spend fewer hours on the job (Schweickart, 2011). Indeed, Germany—where mitbestimmung laws require half the members of any large firm’s board of directors to be elected democratically by their workers (cf. Dow, 2003)—has the lowest work time statistics in our sample (1,363 hours per employee per year in 2016). More generally, leisure time correlates with labor power. For example, the working hours reflected in the data analyzed above have a negative link with labor union membership among the 15 countries reported by McCarthy (2017; Spearman’s ρ = −.65, p < .01).
Thus, while the managers of some capitalist (i.e., shareholder-controlled) firms can be expected to take advantage of the productivity gains that go with reduced employee working hours, a major shift in that direction is only plausible if the balance of power tips mightily toward workers. As Booth (2004) put it, The only final answer . . . is a transformation of existing economic arrangements that would bring corporations . . . under the influence of democratic decision making. One possibility is to move in the direction of an economy where employees own and control the businesses for which they work. (p. 233)
Only given such a transformation will we finally “unhook modern society from high rates of growth and . . . bring the process of environmental degradation to a halt” (p. 233).
Footnotes
Author’s Note
Data used for the analyses reported herein are available on request from the author.
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author received no financial support for the research, authorship, and/or publication of this article.
