4  Discussion

While spillover effects were detected, the overall global impact from vessels that comply with the targets set by the BWBS program remains minimal, representing only a 0.6% increase in emissions compared to the global counterfactual (Figure 3.1). This spillover appears to be driven by a subset of vessels that, while reducing their speed inside the VSR zones, “catch up” to their regular trip times by speeding up outside the zones (Figure 3.5). These are 136 vessels (56% of the compliant fleet) whose net global CO2e emissions are 5.3% higher compared to their baseline (Figure 3.6). Conversely, isolating the 107 vessels (44%) that do not exhibit this catch-up behavior reveals a net global emissions reduction of 5.9% relative to their counterfactual (Figure 3.6).

When spatially exploring these two groups, CO2e emissions and speeds decrease within the VSR areas for both groups, but catch-up vessels show a distinguishable increase in speeds around the VSR zones (Figure 3.11), particularly significant inside the U.S. West Coast EEZ compared to outside it. Catch-up vessels also show a smaller speed reduction inside the VSR areas than that of non-catch-up vessels. It is the increase in speed, and the corresponding rise in CO2e and all other pollutants (Figure 3.7), of these vessels outside the VSR zones that drives the increase in global emissions, which is ultimately capped at just 0.6% by the compensating net global reductions from non-catch-up vessels. These results highlight the potential benefits of the program if vessel operators adjust their behavior within VSR zones while maintaining baseline behavior outside them. Furthermore, these changes should be interpreted in the context of background activity from non-compliant vessels, whose emissions profiles show larger net increases outside the VSR zones than those observed among compliant vessels (Section 3.4).

Inside the VSR zones, all compliant vessels, whether considered as a whole or split by catch-up and non-catch-up groups, show a substantial reduction in CO2 and most other pollutants, with the exception of CH4, CO, and VOCs. The increases in these three particular pollutants are related to nonlinear engine combustion dynamics at reduced speeds. Although lower speeds reduce overall fuel consumption and CO2 emissions, combustion characteristics, reflected in emission factors, engine loads, and operational phases, affect other pollutants differently. Reduced speeds can decrease combustion efficiency for certain pollutants, leading to higher per distance emissions. In short, slowing down saves fuel and reduces CO2 but can worsen combustion efficiency for some pollutants. The emissions model captures these nonlinearities, which explain the observed increases within the VSR zones even among vessels that clearly reduced speed. The same dynamics explain why some local pollutants are reduced less than others. See Appendix A for more details.

When examining program emissions impacts by company, we see how some companies’ vessels engage more deeply with the voluntary speed reductions, driving global reductions in emissions, while others do not (Figure 3.13). Companies also exhibit different amounts of ‘catch-up’ behavior outside the VSR zones, with some exhibiting little or none. This heterogeneity offers a potential avenue to investigate and understand what drives differential engagement across companies.

4.1 Future work

We believe there are three exciting areas for potential future work:

  1. Building off this analysis to develop a robust causal inference framework: While this project has provided an initial exploration and assessment of the BWBS program’s emissions impacts, the rapid nature of the project prevented us from performing a full causal inference assessment of BWBS program. This type of work would require us to build out more robust counterfactuals. These counterfactuals could help us incorporate information from vessels that participated in the program but that did not meet our data inclusion criteria, as well as information from vessels that do not participate in the program at all. Such a framework would also allow us to account for confounding factors that may influence spatiotemporal dynamics but that are not currently included in the model such as fuel prices, shifts in global trade, and seasonal weather and shipping patterns. Thus, a robust causal inference assessment that accounts for such confounding factors and allows us to consider all vessels in the program would enable both attribution of the BWBS program’s impact and quantification of its true emissions reduction potential.
  2. Building an analysis pipeline and platform for performing this type of analysis each year: This current analysis was a one-off attempt at a rapid program assessment. If it is anticipated that this sort of analysis could be desired annually moving forward, we could build an analysis pipeline and platform for repeating this analysis each year and visualizing the results. This could either be an internal tool for the BWBS program, or an external public-facing tool.
  3. Expanding the analysis to other VSRs: As the BWBS program expands state-wide, and as this type of VSR program expands to other areas around the globe such as Chile, this presents an important opportunity to continue assessing the new global impact these programs have. We would be excited to explore further analyses in this space.