Endogeneity
Also known as: Endogenous regressor
Endogeneity means a regressor is correlated with the error: $\mathrm{Cov}(X,\varepsilon)\neq 0$. Then OLS $\hat\beta$ is biased and inconsistent for the structural slope.
Classic causes: omitted confounders, reverse causality / simultaneity, measurement error in $X$. Fixes lie outside plain OLS (instruments, experiments, richer models) — not more in-sample $R^2$.