This study suggests that some empirical findings against money-output causality can be the consequence of ignoring autoregressive conditional heteroskedastic (ARCH) errors. Monte Carlo results confirm that ARCH effects drastically reduce the power of the standard causality test. The maximum...