Do Firms Follow through on Environmental Commitments? An Empirical Examination
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsIt is hoped in the study that firms with higher environmental performance and stronger environmental commitment obtain better financial performance in the long-term. Additionally, the study suggests that following through on environmental commitments is critical but more environmentally proactive countries might put additional pressure on firms to make environmental commitments. The paper concludes with suggestions for future research opportunities including examining how other firm attributes, such as senior leadership gender, may influence the propensity to follow through on environmental commitments and whether these results hold true for other types of negative firm events such as product recalls.
There are several concerns that might be addressed in further efforts. How do you validate using data from multiple sources? It is rarely said that the identification strategy is expressed by equations – the research strategy is first of all confirmed, which leads to the formulations. Therein, you have discovered the challenge of model selection. My question for the authors would be: did you go through a selection process to identify the best alternative among a collection of candidate models? This is based on examining the analysis's goals. Another comment is on Figure 1. I have carefully read its two parts. What are their differences?
Author Response
It is hoped in the study that firms with higher environmental performance and stronger environmental commitment obtain better financial performance in the long-term. Additionally, the study suggests that following through on environmental commitments is critical but more environmentally proactive countries might put additional pressure on firms to make environmental commitments. The paper concludes with suggestions for future research opportunities including examining how other firm attributes, such as senior leadership gender, may influence the propensity to follow through on environmental commitments and whether these results hold true for other types of negative firm events such as product recalls.
COMMENT 1 - There are several concerns that might be addressed in further efforts. How do you validate using data from multiple sources?
RESPONSE – We thank the reviewer for sharing their concern. The reviewer’s concern isn’t clear, since using multiple data sources is common in empirical studies. If we misunderstood the concerns, we will be happy to address any misunderstanding in a subsequent revision.
COMMENT 2 - It is rarely said that the identification strategy is expressed by equations – the research strategy is first of all confirmed, which leads to the formulations. Therein, you have discovered the challenge of model selection. My question for the authors would be: did you go through a selection process to identify the best alternative among a collection of candidate models? This is based on examining the analysis's goals.
RESPONSE – We understand the reviewer to ask two related, but separate questions. The first is about how we developed our empirical model and the second is our identification strategy, i.e., the assumptions we made that develop a case for causality. We discuss both topics in Section 3.3 of the manuscript. As to our empirical model, it is common in empirical studies to assume a linear relationship between the covariates and the dependent variables (unless contextual knowledge suggests otherwise). This is the approach we take. Once the model is established, the challenge is then to (1) use the proper empirical method (determined by the data) to evaluate the research questions, (2) ensure the proper set of control variables are in place to controls for alternate explanations for the results, and (3) conduct a proper set of robustness checks to address possible endogeneity and other explanations for the results. To point #1, as we justify in the second paragraph of Section 3.3, we use GEE (General Estimating Equations) as the primary empirical method. However, we also evaluate Hierarchical Linear Modeling (HLM) and fixed-effects analysis as robustness checks (Section 4.2). The paragraph goes into significant detail explaining why we choose GEE over other possible methods. As to #2, this is discussed in Section 3.2.4. We include as controls all variables that might influence our dependent variable. Further, our fixed effect model (robustness check) addresses the concern that variables not included in the model (but that don’t change over time) might affect the results. Finally, as to #3, we conduct the appropriate robustness checks to address all potential sources of endogeneity (see Section 4.2), including selection bias, reverse causality, and omitted variable bias.
Regarding our identification strategy, we address this in the first paragraph of Section 3.3. As stated, “Our identification strategy assumes that an environmental commitment and EMP adoption are tightly linked temporally (cause and effect occur within close time proximity). We also assume adoption of EMPs may begin slightly before or in anticipation of a commitment. Since we do not know the commitment date, EMP adoption is thus evaluated in the year of a commitment and in the following year.” Should the reviewer have specific concerns about this identification strategy, we would be happy to address them or better explain our logic.
COMMENT 3 - Another comment is on Figure 1. I have carefully read its two parts. What are their differences?
RESPONSE – We thank the reviewer for pointing out the confusion with Figure 1. The figure has been revised in the updated manuscript. The figure is attempting to show that ‘firm environmental performance’ (H3a) and ‘firm size’ (H3b) moderate whether firms follow through on an environmental commitment after an environmental accident.
Reviewer 2 Report
Comments and Suggestions for AuthorsDear authors
The article follows an interesting topic. However, there are some concerns about this study. You may need to consider all the comments below:
You used greenwashing in the keywords, and it's a hot topic, but neither in the title nor in the abstract. The abstract mentions examining 442 U.S. manufacturing firms, but it doesn't provide details on the methods or criteria used to evaluate whether firms followed through on their commitments. More specifics on the methodology would enhance understanding. The abstract does not highlight what makes this study unique or how it contributes to the existing literature on environmental commitments.
Some variable definitions are lengthy and could be more concise. Additionally, the section could benefit from a table summarizing all variables and their sources for quick reference.
While we have various data sources (in this study), it does not clearly explain why each source was chosen and how they complement each other. More justification for the selection of these data sources would be helpful.
The sample is limited to 442 U.S. manufacturing firms, which could introduce selection bias. The section does not discuss how this limitation might affect the generalizability of the findings.
In the conclusion and the discussion sections, the fact that the data for the study came from U.S. manufacturing firms undercuts any possible generalization of the findings to firms from a different region or industry with dissimilar regulatory, cultural, or economic contexts within which to operate. Relating to this, while the issue of 'greenwashing' is mentioned, there is little further exploration as to how firms might execute strategies for greenwashing or how stakeholders might be, in fact, misled by more nuanced tactics. Annualized data has fewer granularities, which may lose information on some of the more detailed, immediate actions firms take when promises are made, or accidents occur. It may, therefore, miss out on changes across a shorter timescale or those that are more nuanced.
Comments on the Quality of English LanguageSome minor corrections are needed.
Author Response
Dear authors
The article follows an interesting topic. However, there are some concerns about this study. You may need to consider all the comments below:
COMMENT 1 - You used greenwashing in the keywords, and it's a hot topic, but neither in the title nor in the abstract.
RESPONSE – We included ‘greenwashing’ as a keyword because firm motives when deciding NOT to follow through on a public commitment might be nefarious. However, our study does not investigate motive. Instead, we evaluate whether firms follow through (or not) on public commitments to reduce resource usage or reduce emissions. While we introduce greenwashing as a possible motive, there can be many reasons a firm may not follow through that are not ‘greenwashing’. For example, a firm may experience resource constraints or there could be poor communication between the parties making the commitment and the parties implementing the commitment. In any case, while we introduce greenwashing as a possible motive, our data does not allow us to go further in establishing a firm’s motive for NOT following through on an environmental commitment. As such, greenwashing does not play a role in our study.
Given the prior discussion, we have removed ‘greenwashing’ as a keyword.
COMMENT 2 - The abstract mentions examining 442 U.S. manufacturing firms, but it doesn't provide details on the methods or criteria used to evaluate whether firms followed through on their commitments. More specifics on the methodology would enhance understanding.
RESPONSE – We discuss how we measure ‘follow-through’ in the fourth paragraph of the Introduction. We write, “Follow-through in the study is measured as increases in the number of environmental management practices (EMP) a firm adopts. EMPs, such as revising processes, conducting audits, and changing supplier selection criteria, are the operational changes firms make to improve environmental performance. Adopting EMPs is thus how a firm would follow through on an environmental commitment. Further, since EMP adoption is directly associated with environmental performance (ex. Anton et al., 2004; King & Lenox, 2002; Klassen & Whybark, 1999b; Potoski & Prakash, 2005; Toffel, 2005), it is a proxy for, and leading indicator of, environmental performance.” We subsequently discuss how we capture information about EMP Adoption in the measurement section of the manuscript (Section 3.2.1).
Should the reviewer prefer additional details, we will be happy to add them.
COMMENT 3 - The abstract does not highlight what makes this study unique or how it contributes to the existing literature on environmental commitments.
RESPONSE – We thank the reviewer for the suggestion. We have now added the following sentences to the end of the abstract, “The results provide important insights for environmentally conscious stakeholders who use the commitments to determine whether to buy from, invest in, work for, or supply to a firm. The study also highlights the benefits to firm leaders of following through and provides input to ideas that can increase follow-through. Finally, the study contributes to several literature streams, including literature evaluating environmental management, environment commitments, and environmental accidents.” While we would have liked to add additional information, we were constrained by the 200-word limit.
COMMENT 4 - Some variable definitions are lengthy and could be more concise. Additionally, the section could benefit from a table summarizing all variables and their sources for quick reference.
RESPONSE – We thank the reviewer for the excellent suggestions. We reviewed the variable definitions and made them more concise. We also created a new ‘Table 1’ that summarizes the variables. This table will be very valuable to readers.
COMMENT 5 - While we have various data sources (in this study), it does not clearly explain why each source was chosen and how they complement each other. More justification for the selection of these data sources would be helpful.
RESPONSE – We have ensured that adequate references are provided for each variable in the study. The new Table 1 also makes the data sources much clearer for the reader. For the control variables sourced from Hardcopf et al. (2019), Table 1 has again clarified the connection to that paper. Should additional clarity be required, we will be happy to add it.
COMMENT 6 - The sample is limited to 442 U.S. manufacturing firms, which could introduce selection bias. The section does not discuss how this limitation might affect the generalizability of the findings.
RESPONSE – We address selection bias as a robustness check (see Section 4.2). Using Propensity Score Matching (PSM), we find that the results are robust to selection bias. Thus, we do not discuss it as a limitation.
COMMENT 7 - In the conclusion and the discussion sections, the fact that the data for the study came from U.S. manufacturing firms undercuts any possible generalization of the findings to firms from a different region or industry with dissimilar regulatory, cultural, or economic contexts within which to operate.
RESPONSE – We agree with the reviewer that focusing the study on firms in the U.S. impacts the generalizability of the study to other countries and regions around the world. As a comment, we evaluate only U.S. firms “… to eliminate country-based heterogeneity in environmental rules, regulations, and cultures that may influence what firms say and do about their environmental performance” (extracted from Section 3.1). To address the generalizability concern, we add a statement in Section 5.3 (Limitations and Venues for Future Research) which reads, “Another potential limitation is that our sample (public U.S. manufacturers) may limit generalizability of the results. For example, European countries have been noted to be more environmentally proactive. This may put additional pressure on European firms to make environmental commitments, whether they have the resources to follow through or not. Replicating our study in different countries might be a productive avenue for future study since understanding when firm follow through on environmental commitments is important to environmental stakeholders around the world.”
COMMENT 8 - Relating to this, while the issue of 'greenwashing' is mentioned, there is little further exploration as to how firms might execute strategies for greenwashing or how stakeholders might be, in fact, misled by more nuanced tactics.
RESPONSE – We refer the reviewer to our response to Comment #1. As mentioned, our study does not investigate motive. Instead, we evaluate whether firms follow through (or not) on public commitments to reduce resource usage or reduce emissions. While we introduce greenwashing as a possible motive, there can be many reasons a firm may not follow through that are not ‘greenwashing’. For example, a firm may experience resource constraints or there could be poor communication between the parties making the commitment and the parties implementing the commitment. In any case, while we introduce greenwashing in the study as a possible motive, our data does not allow us to go further in establishing a firm’s motive for NOT following through on an environmental commitment. As such, greenwashing does not play a role in our study. Note that the keyword ‘greenwashing’ has been removed from the study.
COMMENT 9 - Annualized data has fewer granularities, which may lose information on some of the more detailed, immediate actions firms take when promises are made, or accidents occur. It may, therefore, miss out on changes across a shorter timescale or those that are more nuanced.
RESPONSE – We agree with the reviewer that annualized data has limitations. However, annualized data might also be the best way to evaluate our research questions. We discuss this tradeoff in the manuscript in Section 5.3 (Limitations and Venues for Future Research). The discussion reads, “One potential limitation is that annualized data limits the ability to evaluate the ‘micro’ actions firms take following an environmental commitment or environmental accident. A detailed investigation might provide a better understanding of whether a firm intended to follow through on an environmental commitment or why they were unable to follow through. However, a firm’s response to an environmental commitment or environmental accident may only play out over a longer time horizon. If true, examining macro changes in firm behavior, as we do in this study, might provide a better understanding of management intent.”
Reviewer 3 Report
Comments and Suggestions for AuthorsThe article discusses the problem of enterprise environmental commitments.
Environmental commitments are one of fields which influence the enterprise activity and should be analyzed in conjunction with other factors.
Do environmental commitments influence corporate decision-making? Is it possible to develop a predictive model? What regulations cause environmentally friendly enterprise activity?
Cause-and-effect analysis will provide additional benefits.
Is it possible to divide resources and substances between those that circulate in a closed loop (e.g. water) and those that are disposed of irretrievably, such as coal.
Table 1 is unreadable.
The analyses are conducted at a high level of generality and the results are difficult to apply.
What about information on the companies surveyed, their products, type of resources, etc.?
Figure 2 is difficult to interpret. The values on the horizontal axis are discrete and therefore the linear relationship shown in the figure is not interpretable.
Author Response
The article discusses the problem of enterprise environmental commitments.
Environmental commitments are one of fields which influence the enterprise activity and should be analyzed in conjunction with other factors.
COMMENT 1 - Do environmental commitments influence corporate decision-making?
RESPONSE – The reviewer asks an important question. The short answer is ‘yes’. What we know is that firms feel pressure to be environmentally responsible. There is ample evidence of this in the news every day. Additional evidence is found in the United Nations summits organized to get world leaders aligned around environmental preservation (United Nations, 2024). We also know that firms respond by making public ‘commitments’ to be environmentally responsible. For example, in the United States (the focus of this study), 76% of Fortune 100 companies and 60% of Fortune 500 companies had made a commitment of one or both types by 2021 (Cervantes et al., 2021). What we (firm stakeholders) don’t know is whether or when the ‘commitments’ are valid proxies for action, i.e., when can they be believed. This is the focus of our study.
What we learn through the study is that the commitments DO typically influence corporate decision making, i.e., firms implement environmental management practices (EMPs), although whether and how many EMPs they adopt depends on context. As we discuss in the Introduction (Section 1), “… we find that firms generally follow through on an environmental commitment, whether the commitment is to reduce resource consumption or reduce hazardous emissions. … Results show that firms tend not to follow through on a commitment should they have recently experienced an EA. This suggests that commitment quality depends on contextual conditions at the time of the commitment. … The results (also show that) better environmental performers and larger firms are more likely to follow through on an environmental commitment, even after an EA.”
COMMENT 2 - Is it possible to develop a predictive model? What regulations cause environmentally friendly enterprise activity?
RESPONSE – The reviewer asks an interesting question. A predictive model could be developed or our model could be used to predict future ‘firm environmental activity’. However, a truly ‘predictive’ study would have different research objectives, different hypotheses, different analyses, different discussion, and different contributions. As such It would be best developed as a separate study.
In contrast, our ‘empirical’ study aims to find dependencies and relationships within a set of data. We build a case for causality through our identification strategy, the execution of our empirical model (with sufficient controls for alternate explanations), and the use of robustness checks to address possible endogeneity (selection bias, reverse causality, omitted variable bias).
COMMENT 3 - Cause-and-effect analysis will provide additional benefits.
RESPONSE – We are unclear of the reviewer’s suggestion or concern. Might this comment have been intended to be part of the prior feedback?
COMMENT 4 - Is it possible to divide resources and substances between those that circulate in a closed loop (e.g. water) and those that are disposed of irretrievably, such as coal.
RESPONSE – As the reviewer points out, resource inputs to production processes can be ‘renewable’ (ex. sunlight, wind, tides, plants, and trees) or non-renewable (ex. coal, oil, and natural gas). A unique component of our study is that in addition to evaluating firm commitments to reduce hazardous emissions (‘process outputs’), we evaluate firm commitments to reduce the use of process inputs, i.e., water, energy, and raw materials. In fact, our study is the first to evaluate such commitments (even studies that evaluate reduction ‘targets’ only investigate emission reduction targets, not targets to reduce the use of process inputs). However, our data does not capture whether a firm made a commitment to reduce non-renewable vs. renewable resources. It only captures whether a commitment was made or not. We are not aware of other data that would capture this nuance. As such, we are not able to investigate the reviewer’s suggestion.
COMMENT 5 - Table 1 is unreadable.
RESPONSE – We agree and thank the reviewer for their concern. This concern was also shared by Reviewer 1. The figure has been revised in the updated manuscript. P.S. The figure is attempting to show that ‘firm environmental performance’ (H3a) and ‘firm size’ (H3b) moderate whether firms follow through on an environmental commitment (EC) following an environmental accident (EA).
COMMENT 6 - The analyses are conducted at a high level of generality and the results are difficult to apply.
RESPONSE – We understand the reviewer’s concerns to be two-fold, about (a) the unit of analysis and (b) the generalizability of the results. As to the unit of analysis, the study is conducted at the firm level as that is the level at which ‘commitments’ are made. As discussed in the Introduction, firm stakeholders interpret the commitments firms make when they choose whether to buy from, invest in, work for, or supply to a firm.
As to generalizability (see Section 3.1 for details), we evaluate only U.S. firms “… to eliminate country-based heterogeneity in environmental rules, regulations, and cultures that may influence what firms say and do about their environmental performance”. We also evaluate only manufacturing industries “… because they are a primary source of pollution (Banerjee et al., 2003) and their environmental commitments are thus of greater concern”. However, while our choices were purposeful, it does reduce generalizability to countries outside the U.S. and non-manufacturing firms. As such, we include a statement in Section 5.3 (Limitations and Venues for Future Research) that discusses this generalizability limitation. The statement reads, “Another potential limitation is that our sample (public U.S. manufacturers) may limit generalizability of the results. Non-manufacturing industries harm the natural environment. Also, European countries, in particular, have been noted to be more environmentally proactive. This may put additional pressure on European firms to make environmental commitments, whether they have the resources to follow through or not. Replicating our study in different countries might be a productive avenue for future study since understanding when firm follow through on environmental commitments is important to environmental stakeholders around the world.”
COMMENT 7 - What about information on the companies surveyed, their products, type of resources, etc.?
RESPONSE – Our study includes all manufacturing industries, i.e., the industry sectors that have the greatest negative impact on the natural environment. The sectors included in the study were identified using the Industry Classification Benchmark (launched by Dow Jones and FTSE Russell in 2005). The specific sectors included in the study include Sector 530 - Oil & gas producers, Sector 580 - Alternative energy, Sector 1350 – chemicals, Subsector 1737 – Paper, Sector 1750 - Industrial metals & mining, Sector 1770 – Mining, Sector 2350 - Construction & materials, Sector 2710 - Aerospace & defense, Sector 2720 - General industrials, Sector 2730 - Electronic & electrical equipment, Sector 2750 - Industrial engineering, Sector 3350 - Automobiles & parts, Sector 3530 – Beverages, Sector 3570 - Food producers, Sector 3720 - Household goods & home construction, Sector 3740 - Leisure goods, Sector 3760 - Personal goods, Sector 3780 – Tobacco, Subsector 4535 - Medical equipment, Subsector 4537 - Medical supplies, Sector 4570 - Pharmaceuticals & biotechnology, and Sector 9570 - Technology hardware & equipment. See FTSE Russell (2024) for details about each industry.
We did not include sector details in the original manuscript for brevity. Should the reviewer prefer that we add more detail, we will be happy to do it.
COMMENT 8 - Figure 2 is difficult to interpret. The values on the horizontal axis are discrete and therefore the linear relationship shown in the figure is not interpretable.
RESPONSE – We appreciate the challenge to which the reviewer refers. Interaction plots are notoriously difficult to interpret. However, the approach we take to display the interactions is standard in empirical studies (a google search will show this to be true). If the reviewer has a specific suggestion for how to make the graphs more interpretable, we will be happy to make the change.
Round 2
Reviewer 2 Report
Comments and Suggestions for AuthorsDear authors
The article looks much better in its revised form.
Comments on the Quality of English LanguageOnly some minor corrections.
Author Response
We did not notice any specific recommendations from this reviewer, so did not make any corrections.
