VoxEU Column Environment

More evidence for technology’s role in the clean-up of manufacturing

Pollution emitted by manufacturers has been falling in Europe and the US. A concern with this clean-up is that developed countries have been offshoring the production of pollution-intensive parts and products. This column presents evidence refuting this concern. Using a new approach, the author calculates that almost all of the clean-up in US manufacturing can be explained by technological changes.

Pollution emitted by manufacturers has been falling for decades in Europe and the US, while the real value of manufacturing output has been growing (Brunel 2014). What accounts for this clean-up? A worrisome explanation is that rich countries have been offshoring the pollution-intensive, or ‘dirty’, parts of their manufacturing sectors, producing the clean goods and doing final assembly at home while importing the dirty goods and resource-intensive intermediate inputs. That’s worrisome for two reasons.

  • First, that pass-the-buck clean-up cannot be replicated by all countries. The last of the world’s countries to clean up will have no place left to offshore their own pollution, so somebody is going to be stuck inhaling those pollutants. 
  • Second, pass-the-buck clean-up doesn’t work at all for global pollutants like greenhouse gases that cause climate change. If pollution reductions in Europe and the US simply reappear elsewhere, then that clean-up has been entirely illusory. Europe, the US, and the global climate are not better-off.

The promising explanation for the European and US clean-up is technology. Perhaps Europe and the US are producing the same mix of goods as they did decades ago, but producing those goods in a less pollution-intensive manner -- using less energy and cleaner fuels, recycling, adding pollution control equipment, etc. If technology explains the clean-up, then other countries can replicate that success, perhaps even more easily if they can learn from early adopters’ experiences. And if technology explains the clean-up, then it represents real improvement even for global pollutants.

What accounts for the pollution reduction?

To assess the degree to which composition changes (worrisome) or technology improvements (promising) account for developed countries’ pollution reductions, a number of researchers have taken a simple and straightforward approach.1 They examine the size and composition of countries’ manufacturing sectors over time, and predict what would have happened to pollution if the pollution per dollar of output in each of those industries had remained constant. I wrote about this approach in an earlier VoxEU column (Levinson 2008).  

The approach can be understood by examining Figure 1. The top line depicts the real value of manufacturing output in the US from 1990 to 2008, scaled so that 1990 equals 100. The bottom line depicts total sulphur dioxide (SO2) emissions from manufacturers, similarly scaled. The gap between lines (1) and (2) represents the clean-up to be explained by either composition or technology (see Levinson 2014 for details).

Figure 1. US manufacturing output and sulphur dioxide

Source: NBER-CES Manufacturing Productivity Database and EPA National Emissions Inventory.

As you can see, the pollution growth predicted by the scale and composition of industries in the US (line 3) is close to the growth in the dollar value of manufacturing overall (line 1). 

In other words, if pollution had increased one-for-one with overall manufacturing output, pollution would have increased like line (1). But if pollution increased one-for-one with the output of each industry separately, holding emissions intensities (technology) fixed, pollution would have increased like line (3), nearly as much. 

  • Most of the clean-up – the difference between lines (1) and (2) – must be explained by technology, represented in Figure 1 by the difference between lines (2) and (3).

The studies in footnote 1 all use versions of this same approach, estimating the composition effect using emissions intensities from some initial year and explaining the technology effect as a residual unexplained clean-up. But this approach has several problems. Most importantly, it assumes there are no interactions between scale, composition, and technology – that changing the scale of an industry does not affect its pollution intensity. Any such interactions would all be erroneously included in the remainder term and attributed to technology changes in this approach.

What’s more, all of the prior approaches use emissions intensities from a single year – most often the 1987 Industrial Pollution Projection System (IPPS) developed by the World Bank (Hettige et al. 1992). If over time the faster growing industries cleaned up more in percentage terms, then using base-year pollution intensities misses that change and overstates the role of technology. If the slower-growing industries cleaned up more, then using base-year pollution intensities understates the role of technology.

New approach to estimating the technology effect

In Levinson (2014), I address both shortcomings by estimating the technology effect directly using time-varying measures of pollution intensity.2 Rather than holding technology fixed and attributing the residual clean-up to technology, I hold the composition of industries fixed, measure technology directly, and attribute the residual clean-up to composition changes.

In particular, I calculate an index of technological change as follows. The numerator is the pollution intensity of each industry, multiplied by its valued added in 1990, summed across all industries. So, the numerator is a prediction of total manufacturing pollution if emissions intensities had changed over time, but manufacturing scale and composition had remained fixed.3

Line (4) in Figure 1 plots this index. It drops nearly as fast as emissions, suggesting that nearly all of the clean-up of US manufacturing can be explained by technological changes within each of the 400-plus industries within the US manufacturing sector.  This is completely consistent with the indirect calculation of technology’s role from equation (1). To be precise, the indirect calculation that generates line (3) suggests that 88% of the clean-up of US manufacturing SO2 emissions come from technology changes within industries; the direct calculation that generates line (4) suggests that 92% of the clean-up come from technology. Calculations using alternative indexes and other pollutants yield similar results.

Concluding remarks

This conclusion, and the earlier indirect calculations it confirms, contradicts widespread perceptions about the effects of environmental clean-up on manufacturing. While I don’t assess the cause of that clean-up here, one natural speculation would be that it has been the result of environmental regulations. If so, those regulations have not worked by reducing the share of polluting industries in the US manufacturing sector – driving those industries overseas or reducing consumption of those industries’ products. Instead, the regulations have worked by reducing the emissions intensities on an industry-by-industry basis. That finding should be welcomed by anybody concerned that environmental regulations might only be succeeding by reducing the menu of products available to consumers or by shifting pollution from developed to developing countries. The results here refute that concern. 


Brunel, C (2014), “Pollution Offshoring and Emission Reductions in EU and US Manufacturing” .

Cole, M  A and R J R Elliott (2004), “Determining the Trade-Environment Composition Effect: The Role of Capital, Labor, and Environmental Regulations”, Journal of Environmental Economics and Management 46:363-383.

Ederington, J, A Levinson and J Minier (2004), “Trade Liberalization and Pollution Havens”, Advances in Economic Policy and Analysis 4(2). Berkeley Electronic Press.

Hettige, H, R E B Lucas and D Wheeler (1992), “The Toxic Intensity of Industrial Production: Global Patterns, Trends, and Trade Policy”, The American Economic Review Papers and Proceedings, 82(2): 478-481.

Kahn, M E (2003), “The Geography of U.S. Pollution Intensive Trade: Evidence from 1959 to 1994”, Regional Science and Urban Economics 33(4): 383-400.

Levinson, A (2008), “What accounts for the clean-up of US manufacturing: Technology or international trade”, VoxEU.org, 2 January

Levinson, A (2009), “Technology, International Trade, and Pollution from US Manufacturing”, American Economic Review 99(5) pp. 2177-92.

Levinson, A (2014.), “A Direct Estimate of the Technique Effect: Changes in the Pollution Intensity of US Manufacturing 1990–2008.” NBER Working Paper No. 20399.

Shapiro, J and R Walker (2014), “Why is U.S. Air Quality Improving? The Roles of Trade, Regulation, Productivity, and Preferences” working paper.


1 See Brunel 2014, Reed and Shapiro 2014, Levinson 2009, Cole and Elliott 2014, Kahn 2003

2 See the US National Emissions Inventories at www.epa.gov/ttn/chief/eiinformation.html, accessed August 2014.

3 The index is analogous to a Laspeyres price index, with emissions intensities in place of prices. In Levinson (2014) I also describe the results with a Paasche version of the index – comparing concurrent emissions with predicted 1990 emissions – with similar results.

630 Reads