A new study in the journal Nature Communications warns that most American cities are likely underreporting their greenhouse gas emissions by nearly 20%, an alarming statistic given that the world’s cities already produce the lion’s share of CO2 emissions—and many are expected to triple in size by 2030.

Conducted by a research team from Northern Arizona University, the study looked at 48 U.S. cities, comparing their “self-reporting inventories” (SRIs) to independent estimates from the Vulcan Project, a NASA/Department of Energy initiative that “aims to estimate ‘bottom-up’ CO2 emissions,” writes The Independent.

While common, SRIs have attracted concern as they are conducted entirely by the cities themselves, and are not subject to any systematic or peer-reviewed assessment.

In their investigation, the researchers found that emissions were being underreported by an average of 18.3%. Extrapolating that difference across all U.S. cities, they wrote that the consequent emissions overshoot “would be 23.5% larger than the entire 2015 California state emissions.”

As to how SRI-generated emissions estimates come to their lowballed figures, the study identified errors like the exclusion of certain fuels, as well as non-standard reporting of transportation emissions. It also cited budget concerns as a likely driver for the weakness in SRIs, as the creation of these inventories is “a costly endeavour” that burdens both staff and resources.

While there were exceptions, the majority of the cities studied were found to be underreporting, though there was a wide variation in degree. Indianapolis underreported its emissions by 26.9%, while Flagstaff, Arizona and Madison, Wisconsin were among the few that overestimated their emissions.

The true danger in such faulty reporting lies in how it will affect the development of climate plans. “If cities are working with skewed data, it makes strategies to reduce emission levels less effective,” explains The Independent.

Indianapolis, for example, “has indicated that they aim to make a 20% reduction in building GHG emissions between by 2025 relative to 2016 values.” But, the study authors warn, “with the 26.9% underestimate found here, it will be difficult to know when and if this target is truly achieved or track progress towards it.”