Environmental regulations and labor markets Updated

Conceptual framework

The effect of environmental regulation on labor markets is conceptualized using the neoclassical theory of labor demand [1], [2]. Environmental regulations generally require firms to install pollution abatement equipment that does not necessarily increase their productivity. So environmental regulations can be introduced in the standard labor demand model as an increase in the rental rate of productive capital. An increase in the cost of capital leads to lower output (output effect) and to a shift away from capital (substitution effect). As a result, the net effect on labor demand is indeterminate and depends on whether the output effect is larger than the substitution effect.

Changes in labor demand caused by regulations can also lead to reductions in workers’ wages. The incidence of wage changes will depend on macro- and microeconomic attributes. If regulations lead to increases in labor demand, short-term wage gains are possible. If regulations reduce labor demand, workers exiting the regulated industry and moving to new industries may face transitional costs, depending on multiple factors. Frictional unemployment, arising from transitions between jobs, can open a large time gap between jobs. Displaced workers may lose industry-specific skills or industry rents and face a large wage penalty as they move across jobs. Studies of displaced workers typically show that less educated, longer tenure, and older workers face larger wage losses [3]. So the incidence of the wage cost of environmental regulations is likely to vary across workers, reflecting differences in observable measures of productivity. There may also be wage losses for workers who remain in the regulated industries.

Research designs and data

Several factors make it difficult to identify credibly the effect of environmental regulations on labor market outcomes. In the ideal case for empirical evaluations, regulations would be randomly assigned across workers, firms, industries, and geographic areas. This would ensure that comparable workers and firms are observed across regulatory regimes in similar local labor markets. But this is not always the case. In the US, for example, more polluted areas are more likely to be regulated. They tend to be more densely populated, urbanized areas where polluting firms are older and larger [4]. In addition, there is considerable heterogeneity in wages across workers, firms, and locations. Thus, simple comparisons of wages or employment rates across areas or industries that face different environmental regulations are unlikely to reveal the true effect of regulations on labor market outcomes.

Credible studies (such as internally valid studies) must therefore use quasi-experimental research designs to identify and exploit exogenous sources of variation in regulatory pressure. A common approach is to leverage changes in local regulatory status that result from changes in national environmental standards. In the US, the design of the Clean Air Act has led to such variation in regulatory intensity across years, counties, and sectors.

Specifically, the 1977 Amendments to the Clean Air Act stipulate that, starting in 1978, every county in the US is designated annually as in-attainment or out-of-attainment (non-attainment) of the National Ambient Air Quality Standards. Polluting plants in non-attainment counties are subject to regulations requiring the installation and operation of specified pollution abatement equipment. But polluting plants in attainment areas face weaker regulatory standards and thus face substantially lower capital costs for pollution control. Those differences in capital cost can have differential effects on labor demand.

One approach to exploiting this variation is to compare the outcomes for workers in polluting plants of newly regulated counties, before and after the introduction of the regulations, with the outcomes for workers in similar plants in counties that remain unregulated. The most prominent studies of environmental regulation effects on employment and wages in the US are based on such comparisons [1], [2], [5], [6].

An important matter of interpretation is that such difference-based estimators may overstate the national employment loss due to the regulation. This “double-counting” will occur when the workers displaced in the regulated sectors find new employment in the unregulated sectors [2]. Since there are frictions in labor and capital mobility, this overstatement may be limited in practice, although theoretical analyses suggest such reallocation effects can be important [7]. But other measures of labor market sensitivity to environmental regulations, such as job destruction rates, should also be considered, since they are immune to double-counting [5].

An equally important challenge is to gather the data to exploit these research designs. The ideal data for studying the effect of regulation on labor market outcomes would be a panel of establishment-level microdata, enabling individual-level wages and hours worked to be compared across establishments and over time (before and after changes in regulatory intensity). Moreover, the transitional costs of regulations can be identified only if individual workers (or groups of workers) can be tracked over time as some change their employer (and some remain with the same one). Finally, information on establishment-level regulatory status is needed to assign establishments to “treatment” and “control” groups. To date, studies based on such rich data collection have been primarily implemented in the US due to better data availability, but recent studies in Canada and Europe have also made use of highly granular worker-level data to study the effect of environmental policies.

A final challenge relates to the generalizability or external validity of the results derived from an internally valid empirical study. For example, is the evidence identified from regulatory changes in the 1970s and 1980s relevant for a correct evaluation of the welfare effects of a prospective environmental policy in 2018? Similarly, can a study identified from an environmental reform in one specific labor market (e.g. a state or province in one country) be used to accurately predict the effect of a prospective reform in a different labor market? In the presence of any significant change to the structure of labor markets and to the policy environment over time or differences across countries, provinces, or states, this may not be the case. Thus, studies of the effect of environmental regulations on labor markets need to be carefully designed to strike the right balance between internal and external validity.

Characteristics of polluting industries

The incidence of environmental regulation depends on the industrial composition of a regulated sector and on the characteristics of workers in polluting plants. In the US, attainment (non-regulated) counties tend to be more rural, with lower population density, lower urban population shares, lower median household income, and lower median home values. Research on US manufacturing plants also indicates that polluting plants tend to be larger and older than non-polluting plants [2], [6]. In addition, workers in polluting firms are older, have higher than average education, and earn up to 25% more than workers in comparable, less polluting plants. These unadjusted differences between the polluting and non-polluting sectors show that job displacement caused by environmental regulation can lead to substantial earnings losses for the affected workers, since the cost of job displacement varies across workers of different ages and education levels.

Employment

California introduced air quality regulations in the late 1970s that were more stringent than the federal standards under the Clean Air Act, providing variation in regulatory intensity between parts of California and the rest of the US. A 2001 plant-level analysis of the impact of increased local nitrogen oxides regulation in California's South Coast Air Basin (Los Angeles) area measured the effect of the added regulation on manufacturing plant outcomes-specifically on plant-level pollution-control capital investments, employment, and value added using data from the Annual Survey of Manufactures [1]. The study concluded that the added regulation resulted in sizable investments in abatement capital (especially in oil refineries and other highly polluting industries), without any significant effect on employment. The regulations did impose real costs on manufacturing firms, but had no detectable employment effects.

Another detailed study looked at the effect of the increased stringency of the emission standards under the 1970 and 1977 Amendments to the Clean Air Act in the US [2]. These Amendments represented the first air quality standards introduced in the country and the first attempt by the federal Environmental Protection Agency (EPA) to enforce them. The empirical analysis is based on detailed plant-level input and output data for more than 1.75 million plants drawn from the 1967?1987 US Census of Manufactures. The preferred empirical estimates suggest that carbon monoxide and ozone regulations have the strongest depressing effects on labor demand. A carbon monoxide non-attainment designation leads to a 3.3% reduction in annual employment in carbon monoxide-emitting plants, while an ozone non-attainment designation leads to a 1% reduction in annual employment in ozone-emitting plants. Regulations for excessive sulfur dioxide and suspended particulate emissions are not associated with significant changes in employment.

The study also examines the heterogeneity of the measured effect of the regulations on employment across industrial sectors. While the evidence suggests that regulatory effects on employment do not differ statistically across industries, the total impact of the regulations is particularly severe for industries that emit multiple pollutants in counties that are designated as non-attainment for those pollutants, particularly for the pulp and paper and the iron and steel industries.

Overall, the evidence suggests that in the first 15 years of implementation of the Clean Air Act (1972–1987), regulated non-attainment counties lost close to 600,000 jobs (relative to the unregulated counties) [2]. Ozone and carbon monoxide regulations were the prime source of the employment loss. Although the decline in manufacturing employment was substantial in non-attainment counties, it was modest in relation to the size of the entire manufacturing sector: the 600,000 jobs lost correspond to about 3.4% of total employment in the sector over the study period.

Other studies of the strengthened emission standards under the Clean Air Act amendments of the early 1990s in the US have used rich data on establishment-specific employment and payrolls to create a panel of plant-level observations by county, year, and sector over 1985–2005 [5], [6].

The analysis of employment rates in these studies is the most complete to date, since it precisely measures job dynamics for the affected sectors. In particular, the employer–employee sample allows individual workers to be tracked over time as some are displaced from their jobs following increases in air quality regulatory pressure. The study shows a prolonged decline in employment associated with the new regulations. Employment in polluting sectors fell by 15% in the ten years following the change in regulation. A decomposition of the overall employment effect indicates that employment losses were driven mostly by higher job destruction rates in regulated sectors (as opposed to lower job creation rates). So, workers displaced by the regulations may suffer significant costs associated with involuntary job loss. The results also indicate that sectors regulated because of violations of the ozone, particulate, and sulfur dioxide standards faced the largest reductions in employment over the long term.

The difference in the pollutant-specific employment effects reported in these studies highlights the change in pressure imposed on labor markets by the regulation of specific pollutants from the late 1970s and 1980s to the 1990s [2], [5].

Environmental regulations are sometimes applied to specific sectors or fuels as opposed to mandating a baseline air quality standard. Regulations imposed on specific sectors that produce an important input used in many other sectors can cause a ripple effect on labor demand throughout the economy. In particular, power plants in some states of the US are subject to regulations regarding emissions of nitrogen oxides (NOx) and sulfur dioxide. The employment effects associated with the US NOx Budget Trading Program, a cap-and-trade market regulating emissions of nitrogen oxides in the Midwest and Eastern US, are the focus of a 2018 study [8]. Using a triple-difference estimator comparing employment in manufacturing industries pre and post implementation, in regulated and non-regulated states, and across various levels of industry-specific energy intensity, the study finds that the regulation reduced manufacturing employment by 1.3%. Since the NOx Budget Trading Program mostly affected the electricity generation sector, these employment effects should be driven by higher electricity prices charged to manufacturing industries, although this mechanism is not directly considered in the study. Examination of labor market flow data suggests that the employment effect is primarily driven by a reduction in hiring as opposed to an increase in job separations (including layoffs and firings) [8].

A growing literature now reports empirical evidence of employment effects of environmental policies outside the US. An earlier example examined the effects of the climate change levy on manufacturing plant activity using data from the UK's production census [9]. The study compares outcomes between plants that have to pay the full tax rate under the levy and plants that were granted an 80% discount on the tax after voluntarily joining a climate change agreement (voluntary agreements containing targets to increase energy efficiency or reduce carbon dioxide emissions). Fixed-effect and instrumental variable methods are used to control for the selectivity of joining a climate change agreement. The study finds that the climate change levy leads to large declines in plant-level electricity use but has little effect on overall economic performance, employment, and productivity.

In 2008, the province of British Columbia in Canada introduced a revenue-neutral carbon tax on the purchase of all fossil fuels for all businesses and industries, and the residential sector. Two recent studies examine the impact of this carbon tax using difference-in-differences methods that compare outcomes in British Columbia with outcomes in the other provinces in the rest of Canada before and after the introduction of the carbon tax.

The first study argues that revenue-neutral carbon taxes can affect employment through two channels: an output effect (a reduction in labor demand due to a reduction in output) and a redistribution effect (an increase in labor demand due to the redistribution of the tax proceeds which can increase product demand) [10]. Thus, conceptually the employment effect of a revenue-neutral carbon tax is a priori ambiguous, depending on which of the two channels dominate.

Using industry-level employment rates and average wages for 2007–2013, the author finds that the carbon tax had a small positive effect on aggregate employment. Another key finding is the varying effects across industrial sectors: employment fell in carbon-intensive and trade-exposed sectors in response to the tax while it grew in less carbon-intensive service sectors.

A second, similar study alternatively uses individual-level data to examine the possible heterogeneous response to the carbon tax across worker skill level [11]. The author reports that the British Columbia carbon tax increased the fraction of workers reporting being unemployed while leaving the average hours of work per week and the labor force participation rate unchanged. Moreover, the increase in the proportion unemployed was especially large among lower-educated workers.

A striking feature of both studies is the divergence of the main results, which underscores the challenge associated with credibly measuring the impact of environmental regulations on labor market outcomes. To this end, the author of the second study argues that some of the differences between the results may be explained by a failure of the common trend assumption necessary for the implementation of the difference-in-differences method.

Air and water quality regulations are also being established in China and India to counteract the large impacts pollution has on health in those countries. However, in general, much less is known about the labor market effects of environmental regulations in low- and middle-income countries. One of the first comprehensive studies of the impact of such regulations on employment in China estimates the effect of more stringent wastewater discharge regulations on textile and dyeing firms in the Jiangsu region [12]. Using a difference-in-differences method, the authors find that the stricter standard reduced employment by 7%. Additionally, the reduction in employment was found to be concentrated in domestically-owned private firms as opposed to foreign-owned or state-owned firms, perhaps indicative of differential enforcement or inspections across ownership type.

Productivity

Two US studies measure the effects of environmental regulation on productivity in the manufacturing sector, revisiting earlier studies that examined the role of environmental regulation in explaining the productivity slowdown of the 1970s.

A large-scale analysis from 2012 of the effect of US air quality regulations on manufacturing plant productivity, measured by plant-specific total factor productivity, used detailed plant-level production data for 1.2 million plants drawn from the 1972–1993 Annual Survey of Manufactures [13]. The main finding is that, for surviving polluting plants, stricter air quality regulations are associated with total factor productivity declines of about 2.6%. In other words, output at regulated polluting plants declined by 2.6%, holding constant the labor, capital, and material inputs. Once estimates are corrected for the confounding effects of price increases and output declines in the manufacturing sector over 1972–1993 and for selection based on plant survival, the measured effect of air quality regulations is a 4.8% decline in total factor productivity for polluting plants in regulated areas. Of individual air pollutants, regulation of ozone has the largest negative effect on productivity. By contrast, carbon monoxide regulations increase measured total factor productivity, especially among refineries. Together, the results indicate an annual economic cost of air quality regulations on manufacturing plants of $21 billion (2010), or roughly 8.8% of average manufacturing sector profits over the study period.

In contrast to these results is an analysis of the relationship between measured environmental compliance costs and plant-level productivity, defined by the real value of shipments per worker [14]. In the 1980s and early 1990s, productivity and environmental compliance costs were only weakly correlated in the US.

Earnings

Only one study measures the effect of environmental regulations on long-term earnings [6]. The key starting point of the prior studies examining employment effects is that workers displaced by regulations generally find new employment, perhaps in new locations or industries. In the absence of frictional unemployment, workers move across jobs quickly, so measures of regulatory effects on job losses would not be informative about the costs of environmental policies. In reality, however, some displaced workers may experience long periods of unemployment following layoff and may lose industry- or job-specific skills. This study provides an important summary measure of these kinds of costs by studying the long-term wage effect of job displacement due to environmental regulations [6].

The analysis, based on longitudinal employer?employee data, tracks US workers across jobs over time. This permits the measurement of long-term wage costs for workers who remain in regulated sectors and for workers displaced by the regulations. The results indicate that the earnings costs generated by the 1990 Clean Air Act Amendments are significant: workers in sectors affected by the new regulations lose more than 5% of their pre-regulation earnings in the three years after the regulations’ implementation, and the declines are persistent, since earnings begin to recover only five to six years after the regulations are introduced.

In other words, air quality regulations in the US appear to impose long-term costs on the affected workers. On average, affected workers in the regulated sectors experience a total earnings loss equivalent to 20% of their pre-regulatory earnings. These losses are almost entirely driven by workers displaced from the regulated sectors, rather than by workers who remain employed in the regulated sectors. Further, the evidence suggests that the effects are concentrated among older workers displaced from large plants in areas with weaker local labor markets. While the estimated aggregate wage displacement costs of the 1990 Clean Air Act Amendments are large ($5.4 billion in foregone earnings), they remain small compared with the estimated benefits associated with increased air quality. The EPA estimates that these benefits range from $160 billion to $1.6 trillion.