So-called "spurious regression" relationships are generally accompanied by clear signs of residual autocorrelation. A conscientious researcher would likely re-estimate with an autocorrelation correction. Simulations indicate that resulting test statistics are close to true values, so do not yield spurious results.