Small-State Aviation Carbon Dynamics: Data Quality, Code Collisions, and Panel Construction in EUROCONTROL Emission Statistics — Reproduction Package

EUROCONTROL state-level CO₂ emission statistics have become a widely-used data point in European aviation climate science, but the steps between the raw files and an estimate panel are sparse — and they can be consequential. This note documents five of the hazards collected from auditing the monthly files for 2010–2026 and brings into relief what each does to inference. For COLLECTED using codes, two STATE_CODE collisions (LY shared by Serbia and Montenegro, LS shared by Switzerland and Liechtenstein) inflate code-keyed Montenegrin emissions on average 679% and five orders of magnitude for those of Liechtenstein. Three countries joins the panel later (Latvia in 2011, Georgia in 2014 and Estonia in 2018) contributing one additional cumulative growth point to naive aggregate trends. Only one microstate matters: Monaco, with its 12–40 flights per month, litter the across state population dispersion of log emissions increase by a factor of 33% and alters the panel central finding of convergence from significant (β = −0.0100, p = 0.009) to null (β = −0.0028, p = 0.38). You can see it all the timeHigh capacity estimation flags get perpetually stuck in four states, and non-ASCII state titles lead to whispering fizzle unfalteringly. Europe’s post-pandemic emission growth is concentrated in small states. Consequently, these data defects directly overlap with the most scientifically interesting variation. The note distils a five-rule construction protocol, releases a reproducible pipeline, and demonstrates that four defensible-looking panel constructions yield materially different headline statistics from identical raw data.