As you've observed, averages can be misleading because they're not robust to outliers. However, this seems most relevant to something like income, where we're interested in "most people" and don't care so much about the one ultra-rich person in the sample.
This appears to be less true for something like emissions, where the overall total is most relevant -- unless you're arguing that it's masking an "unfair" situation where improvements are coming from one group and should be coming from the other?