Response to “A critical look at ‘Energy savings, emissions reductions, and health co-benefits of the green building movement’”

A Correction to this article was published on 20 January 2020

This article has been updated

We appreciate the opportunity to respond to the commentary on our paper “Energy savings, emissions reductions, and health co-benefits of the green building movement” by John Scofield and Jakob Cornell [1, 2]. Our research has three main analytical components that we join to estimate the health and climate co-benefits of energy-efficiency measures in buildings: (1) analysis of energy utilization in buildings, (2) calculation of air pollutant emissions from that energy use in buildings, and (3) quantification of the resulting public health and economic impact of those emissions. The commentary by Scofield and Cornell focuses on the first component—energy utilization in buildings. (Scofield and Cornell acknowledge that our methodology in estimating emissions and health co-benefits is reasonable.) We will show here that the commentary by Scofield and Cornell related to the first component suffers from serious shortcomings and, as a result, does not impact our conclusions as originally written; our findings stand.

The crux of Scofield and Cornell’s argument is their assertion that our analysis did not account for source energy production losses, but rather only site energy. This assertion is incorrect. Our analysis relies on emission factors for electricity from the Environmental Protection Agency’s (EPA’s) Emissions & Generation Resource Integrated Database (eGRID) database. This database reports average emission rates for the main pollutants emitted from electricity generating power plants in each of the sub-regions in the United States, inclusive of losses during generation. Thus, the vast majority of the energy losses associated with electricity use are already included in our estimate of pollutant emissions.

While the generation losses are already accounted for, the transmission and distribution losses are not included in the eGRID emission factors are on the order of 5.82–8.21% depending on the region (Table 1).

Table 1 Grid gross loss factors reported by the EPA for each eGRID sub-region in 2009

Thus, the scalar to adjust our energy estimates is on the order of 1.05–1.08, which is significantly lower than that cited by Scofield and Cornell of 3.23 based on a source-site conversion efficiency of 31%, which is predominantly driven by energy losses during power generation at the plant. Table 2 shows that correcting for line losses may marginally increase the U.S. co-benefit estimates as it increases the source energy consumption of conventional buildings as well as LEED-certified buildings.

Table 2 Cumulative emissions reductions (kilotons) and climate/health co-benefits (million 2016 USD) from the U.S. LEED-certified projects from 2000 to 2016 under different scenarios

The commentary by Scofield and Cornell has additional important shortcomings. In an attempt to demonstrate how considering source energy influences the results, Scofield and Cornell selectively chose to present data from a very small and limited subset of data of buildings in one city, Chicago, to support their argument. However, of that limited sample, they then further narrowed the dataset by preferentially selecting just three of the worst performing green buildings to base their analysis on. Curiously, even among this sample, the data show that green buildings outperform conventional buildings: their energy use intensities (EUIs) are 791, 734, and 804 MJ/m3 compared to a conventional building EUI of 1006 MJ/m3. As they acknowledged, our estimated EUI reduction for these buildings was close but slightly more conservative than the performance data: 780 MJ/m3 versus an average of 776 MJ/m3 for these three buildings. As mentioned above, the rest of our methodology accounts for energy losses from combustion, both from the grid and on-site combustion, so our analysis is very consistent with the dataset that Scofield and Cornell attempted to use to discredit our analysis; the three buildings selected by Scofield and Cornell actually strengthen our findings.

One of the strengths of our approach is that we include all green buildings that have LEED certification, mitigating bias due to regional differences. The Chicago subset analyzed by Scofield and Cornell is not representative of the cohort of buildings we analyzed. Chicago is located at the RFC West sub-region which has a very high use of coal (49.8% vs. national average 30.4%) in power generation resource mix (eGRID Summary Tables 2016), and has the lowest reliance on electricity (percentage of electricity use is 53% compared to national average 60%) (CBECS 2012). If Scofield and Cornell had selected California (where only 4.3% of electricity production comes from coal power plant), which has approximately the same certified floor space as the RFC sub-region, buildings that use electricity would have fewer emissions than those that rely entirely on on-site natural gas or fuel oil combustion.

On a minor note, Scofield and Cornell do correctly point out that in one of our opening paragraphs we wrote that, ‘As of 2016, 90,500 commercial buildings in 165 countries (though mainly in the United States) have achieved LEED certification.’ The number we cite refers to the number of commercial projects, not fully certified buildings. This is a minor clarification and the sentence should read, ‘As of 2016, 90,500 commercial projects in 165 countries (though mainly in the United States) have participated in LEED, including 28,994 certified commercial projects published on GBIG.’ That said, since 2016, the global LEED building market has continued to grow, and the latest statistics on the USGBC website already increased to 92,200, as of October 2017. The number of certified commercial projects published on GBIG had also increased to 32,583 by the end of 2017.

Scofield and Cornell also point out that it was possible that a few LEED-certified projects with multiple certifications could get counted more than once in our co-benefit calculation, yet they made no attempt to determine if this would have any meaningful impact. We did. And, while possible, the impact would be trivial in our overall analysis. The total repeat count of duplicated LEED-certified projects in the six countries is 297, accounting for 1.2% of the analyzed building stock. Considering this, the total co-benefit estimates could be slightly inflated by the same scale. However, these 297 duplicate buildings are significantly fewer than the 3589 projects certified since 2016 that were not included in our analysis, again suggesting that we are in fact underestimating the real health co-benefits of LEED-certified buildings.

As discussed in the original paper, there is some debate about whether or not green buildings have lower EUIs than conventional buildings and whether these are consistent with design specifications. The most robust discourse has been around the 121 buildings certified before 2006 under the LEED standard New Construction V2. This dataset was originally analyzed by Turner & Frankel before being reanalyzed by Newsham et al. and Scofield. Based on differences in analytical approaches, the authors find site energy savings of 30–35%, 18–39%, and 10–17%, respectively. Interestingly, there is a wide variability in EUI for LEED buildings. Newsham et al. found that while mean EUI was 18–39% lower than paired conventional buildings, 28–35% of the LEED buildings actually performed worse than conventional buildings. In our analysis, we anticipate that for individual buildings, our estimated EUI may be significantly different than the actual EUI, but on average across the full set of buildings it appears that the range of EUIs is relatively consistent with the range of design EUIs in our analysis, which is from 20 to 40% depending on the level of certification and year. Again, differences in site and source energy are already accounted for in our analysis.

In addition to these three papers just discussed, Scofield and Cornell attempt to use a lengthy reference list to suggest there is much support for their claim about an energy performance gap (citations 2–28). On closer inspection, of these 27 sources, three are discussed above, and, of the remaining 24, only 12 are peer-reviewed, five are self-referential (i.e., written by Scofield et al.), and one is cited twice. Scofield and Cornell include a summary of these papers in Table 1 of their commentary. The papers that have been peer-reviewed consistently show that LEED buildings use less site energy than conventional buildings with the magnitude of performance varying based on the sample of buildings investigated. In general, analyses of high energy intensity, LEED-certified research and educational buildings tended to perform closer to conventional buildings, while medium energy intensity, LEED-certified buildings had consistent energy reductions compared to conventional buildings. This finding is best exemplified by the Turner & Frankel dataset which included both high and medium energy intensity buildings.

In conclusion, we stand by our research and the findings presented in the original paper. While our analysis relies on several assumptions, as all modeling approaches do, our choices of data and methodology are: (1) clearly described in our original manuscript, (2) based on sound scientific evidence and assumptions, and (3) follow a methodology that was intentionally conservative and erred on the side of underestimating benefits.

On that last point, the estimated benefits would most certainly be larger if we: (1) included green buildings certified by other standards and high-performance buildings that did not pursue any certification (e.g., we analyzed LEED data only, which accounts for ~30% of all green buildings) and (2) refined the damage estimates related to climate and air pollution to include: (a) new research showing consistent exposure-response functions of air pollution at levels below our current air pollution standards, (b) included air pollution and climate damages that were not only focused on combustion, but also extraction, transport, and the rest of the life cycle of energy production, and (c) included direct health effects of NOx and SO2.

Overall, our findings support a goal of increasing the adoption of energy-efficiency in buildings due to the immediate and long-term social benefits through averted emissions related to fossil-fuel combustion and the role that green buildings can have in meeting the Paris Agreement and the achieving progress toward the Sustainable Development Goals set by the United Nations. The commentary by Scofield and Cornell, if anything, strengthens the original assertions made in our paper.

Change history

  • 20 January 2020

    An amendment to this paper has been published and can be accessed via a link at the top of the paper.


  1. 1.

    MacNaughton P, Cao X, Buonocore J, Cedeno-Laurant J, Sprengle J, Bernstein A, et al. Energy savings, emission reductions, and health co-benefits of the green building movement. J Expo Sci Environ Epidemiol. 2018;28:307–18.

  2. 2.

    Scofield J, Cornell J. A critical look at “Energy savings, emissions reductions, and health co-benefits of the green building movement”. J Expo Sci Environ Epidemiol. 2018.

Download references

Author information



Corresponding author

Correspondence to Joseph G. Allen.

Ethics declarations

Conflict of interest

The authors declare that they have no conflict of interest.

Additional information

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

MacNaughton, P., Cao, X., Buonocore, J. et al. Response to “A critical look at ‘Energy savings, emissions reductions, and health co-benefits of the green building movement’”. J Expo Sci Environ Epidemiol 29, 594–596 (2019).

Download citation

Further reading


Quick links