“You can’t manage what you don’t measure,” the adage goes. But despite valiant efforts by companies to measure their supply chain emissions, the majority are missing a big part of the picture.
Widely used models for estimating supply chain emissions simplify the process by assuming that companies source all of their goods from a single country or region. This is obviously not how the world works, and manufacturing in the United States is often cleaner than in countries with coal-heavy grids, like China, where many of the world’s manufactured goods actually come from. A study published in the journal Nature Communications this week found that companies using a U.S.-centric model may be undercounting their emissions by as much as 10%.
“We find very large differences in not only the magnitude of the upstream carbon footprint for a given business, but the hot spots, like where there are more or less emissions happening, and thus where a company would want to gather better data and focus on reducing,” said Steven Davis, a professor of Earth system science in the Stanford Doerr School of Sustainability and lead author of the paper.
Several of the authors of the paper, including Davis, are affiliated with the software startup Watershed, which helps companies measure and reduce their emissions. Watershed already encourages its clients to use its own proprietary multi-region model, but the company is now working with Stanford and the consulting firm ERG to build a new and improved tool called Cornerstone that will be freely available for anyone to use.
“Our hope is that with the release of scientific papers like this one and with the launch of Cornerstone, we can help the ecosystem transition to higher quality open access datasets,” Yohanna Maldonado, Watershed’s Head of Climate Data told me in an email.
The study arrives as the Greenhouse Gas Protocol, a nonprofit that publishes carbon accounting standards that most companies voluntarily abide by, is in the process of revising its guidance for calculating “scope 3” emissions. Scope 3 encompasses the carbon that a company is indirectly responsible for, such as from its supply chain and from the use of its products by customers. Watershed is advocating that the new standard recommend companies use a multi-region modeling approach, whether Watershed’s or someone else’s.
Davis walked me through a hypothetical example to illustrate how these models work in practice. Imagine a company that manufactures exercise bikes — it assembles the final product in a factory in the U.S., but sources screws and other components from China. The typical way this company would estimate the carbon footprint of its supply chain would be to use a dataset published by the U.S. Environmental Protection Agency that estimates the average emissions per dollar of output for about 400 sectors of the U.S. economy. The EPA data doesn’t get down to the level of detail of a specific screw, but it does provide an estimate of emissions per dollar of output for, say, hardware manufacturing. The company would then multiply the amount of money it spent on screws by that emissions factor.
Companies take this approach because real measurements of supply chain emissions are rare. It’s not yet common practice for suppliers to provide this information, and supply chains are so complex that a product might pass through several different hands before reaching the company trying to do the calculation. There are emerging efforts to use remote sensing and other digital data collection and monitoring systems to create more accurate, granular datasets, Alexia Kelly, a veteran corporate sustainability executive and current director at the High Tide Foundation, told me. In the meantime, even though sector-level emissions estimates are rough approximations, they can at least give a company an indication of which parts of their supply chain are most problematic.
When those estimates don’t take into account country of origin, however, they don’t give companies an accurate picture of which parts of their supply chains need the most attention.
The new study used Watershed’s multi-region model to look at how different types of companies’ emissions would change if they used supply chain data that better reflected the global nature of supply chains. Davis is the first to admit that the study’s findings of higher emissions are not surprising. The carbon accounting field has long been aware of the shortcomings of single-region models. There hasn’t been a big push to change that, however, because the exercise is already voluntary and taking into account global supply chains is significantly more difficult. Many countries don’t publish emissions and economic data, and those that do use a variety of methods to report it. Reconciling those differences adds to the challenge.
While the overall conclusion isn’t surprising, the study may be the first to show the magnitude of the problem and illustrate how more accurate modeling could redirect corporate sustainability efforts. “As far as I know, there is no similar analysis like this focused on corporate value chain emissions,” Derik Broekhoff, a senior scientist at the Stockholm Environment Institute, told me in an email. “The research is an important reminder for companies (and standard setters like the Greenhouse Gas Protocol), who in practice appear to be overlooking foreign supply chain emissions in large numbers.”
Broekhoff said Watershed’s upcoming open-source model “could provide a really useful solution.” At the same time, he said, it’s worth noting that this whole approach of calculating emissions based on dollars spent is subject to significant uncertainty. “Using spending data to estimate supply chain emissions provides only a first-order approximation at best!”
