Correlated Errors Challenge Vulnerable Growth
研究发现，美国GDP增长波动主要由金融条件预测的下尾分位数驱动，但相关误差导致内生性问题；修正模型后，金融条件单独不再显著预测下尾增长，挑战了先前结论。
ABSTRACT We examine the impact of correlated errors on findings regarding fluctuations in US GDP growth, which are primarily driven by the lower quantiles of the distribution as predicted by financial conditions. A proposed zero‐correlation test reveals the presence of correlated residuals in the quantile regression, introducing endogeneity and raising concerns about the validity of the previous analysis. In response, we extend the model by including two additional lags of predictors. The revised specification shows that financial conditions alone no longer significantly predict lower‐tail GDP growth, although their joint effect with lagged values remains significant. These results challenge earlier conclusions about the sensitivity of GDP growth to financial conditions.