Assessing Energy and Climate Effects of Digitalization: Methodological Challenges and Key Recommendations

Digital technologies hold enormous potential to help or hinder climate change mitigation. They have been claimed to be a major threat—responsible for a large and growing emissions footprint while accelerating consumption and fossil fuel extraction—as well as a critical enabler of emission reductions across sectors. These claims and estimates are wide-ranging and inconsistent, frustrating public understanding and thoughtful policymaking. For example, published estimates of the lifecycle greenhouse (GHG) emissions of the global information and communication technology (ICT) sector for 2015 range by a factor of 2, while projections for 2025 diverge by up to 25-times. These considerable differences and uncertainties reflect the use of inconsistent – and in some cases, problematic—methodologies. The lack of consensus is also prominent in attempts to quantify the indirect effects of digital technologies, which are even more poorly understood.

Given the urgency of the climate crisis and vast but uncertain impacts of digitalization, there is a critical need to develop and employ robust and consistent methodologies to assess, review, and evaluate the energy and climate effects of digitalization. To identify key challenges and potential solutions, over 70 leading researchers, practitioners and policy makers participated in a two-day expert workshop in May 2021. This paper provides essential background on the need for the workshop and highlights key outcomes of the workshop, including actionable insights that were put forward by the expert participants.

Day 1 of the workshop focused on the energy and emissions footprint of digital technologies (direct effects), including from raw materials acquisition, production, use, and end-of-life treatment. Day 2 focused on indirect effects on emissions resulting from the use of digital technologies across sectors and services, such as efficiency improvements, rebound effects, and long-term socio-economic changes. Each day included two breakout sessions, where participants were organized into parallel breakout groups focused on specific methodological challenges, including:

• A lack of consistency in system boundaries and methodologies;

• Challenges  in data availability, accessibility, transparency, and quality;

• A need to further develop methodologies, including for specific applications/services, validating and improving models, dealing with uncertainties, environmental impacts beyond GHGs, hypothetical baselines, rebound effects, limitations of case studies, and emissions accounting for avoided emissions.

Each of the breakout groups developed a short list of key actionable insights for policy makers, industry, researchers, and other key stakeholders related to their topics (see the Appendix for the full list). The need for access to data and transparency of methods was a central theme across all breakout groups. The breakout group insights have been synthesized into several key recommendations:

• All stakeholders must contribute to foundational work to improve consistency in terminology, methodology, and data, such as developing common terminology and boundaries for central concepts (e.g., digital sector, machine learning, the internet) and taking stock of existing data gaps and quality issues. They also need to contribute to validating models, conducting reality-checks on data, and transparently addressing uncertainties in results.

• Intergovernmental organizations and standardization bodies should increase engagement with academia and industry to raise awareness and promote the use of key existing standards (e.g., ITU L.1410 and L.1450). Standards and protocols can play a key role in promoting consistency through clear definitions, boundaries, and methodologies, but awareness and use to date have been limited.

• Companies along digital technology value chains should improve the systematic collection and public reporting of timely and high-quality data. Empirical data on energy, emissions, and other environmental indicators can provide the foundation for modeling and developing and evaluating baselines.      Consistent and systematic disclosure of data needs new systems and channels to encompass all segments of the digital value chain, with business incentives potentially playing an important role (e.g., pressure from customers).

• Policy makers and regulators can also play an important role in requiring the stent and comprehensive collection of data across the ICT sector. Solid data provides an essential foundation for crafting and implementing sound policies and regulations. Policy makers should be aware of the complexities of life cycle assessments (LCA) and understand that the rch is still maturing. From a regulatory perspective, topics such as data accessibility and handling are currently high up on the political agenda, presenting a window of opportunity to advance policy in these areas.

• Researchers and organizations involved in standards and disclosure should increase engagement and collaboration with companies in the ICT sector and beyond.     Engaging with the ICT sector could help improve the quality and consistency of reporting, increase the availability of disaggregated and context-specific data, and improve benchmark studies for life cycle assessments. When setting up such studies and establishing basic average data sets, opportunities to learn from other sectors should be explored.

• Researchers must work to increase the rigor of peer reviews and improve methodologies, including evaluation of rebound effects from digitalization and potential measures to counteract undesirable effects. Further work is needed on the complexity of baselines for assessing indirect effects as well as in dealing with uncertainties in data, models, and results.

• Researchers should engage in development of methods for prospective modeling that integrate a deeper understanding of future technology usage by changing industries and societies. Such methods should balance dealing with the uncertainties of predictions over a period of 10-15 years when product roadmaps only cover a fraction of this time, while allowing sufficient guidance to shape the development of systems in a sustainable direction.

• Diverse perspectives from across disciplines are needed to develop robust criteria for case studies and set their system boundaries. This implies a need to develop specific but consistent methodologies for different types of studies which can provide complementary benefits—comprehensive studies can provide robust conclusions while smaller-scale studies can provide initial insights and advance methodological approaches.

• All actors should consider the requirements on studies in relation to their purpose and handle methodology choices and trade-offs accordingly. Accuracy depends on the research question—qualitative results and magnitudes may be more important than the precise level of impact and there is a crucial balance between accuracy and complexity. There is also a conflict between timely results and accurate review processes, and allocation of emission reductions to specific actors is often less relevant than describing the overall impact of applications. In any case validation of models and reality checks of data is important.

Advancing these recommendations can increase consistency and robustness of estimates, while also helping to identify key levers to maximize the potential benefits of digitalization while minimizing its risks. Progress will require coordinated and complementary efforts from policy makers, industry, researchers, and other organizations.