<![CDATA[Deforestation and rising greenhouse gas (GHG) emissions are widely assumed to impose measurable welfare costs, yet the empirical basis for that assumption in Indonesia remains thin. This study asks whether tree cover loss (TCL) and GHG emissions are associated with GDP per capita, the Human Development Index (HDI), the poverty rate, and regional investment, with population density as a control variable, across 34 Indonesian provinces over 2015-2024 (340 observations). Panel ARDL models are estimated by Pooled Mean Group (PMG) and Dynamic Fixed Effects (DFE) and are then subjected to a full diagnostic sequence: CIPS unit-root tests, Pesaran-Yamagata slope-homogeneity and Hausman tests, Pesaran cross-sectional dependence tests, collinearity diagnostics, Common Correlated Effects Mean Group (CCEMG) re-estimation, and subsample exclusion tests. The initial estimates indicate a strong association between TCL and poverty (-3.1989, p<0.01) and a weaker one with investment (-0.7036, p<0.10). Neither survives. The two degradation proxies are almost perfectly collinear (r = 0.994, VIF > 80), cross-sectional independence is rejected in every specification, and the poverty coefficient collapses to -0.0025 once a single anomalous year is excluded. A parsimonious specification carrying one degradation proxy, population density, and Driscoll-Kraay standard errors does return significant negative coefficients for GDP per capita, HDI and investment; all become insignificant, and two reverse sign, once the common factors shared across provinces are absorbed. The welfare losses that a conventional panel specification produces are therefore a cross-sectional-dependence artefact rather than an identified effect, so claims of a significant deforestation-welfare link in Indonesia should be treated with caution.]]>
Copyrights © 2026
