Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
The dynamic relationship between advertising and sales in the annual Lydia Pinkham data is re-evaluated. Past studies have found a feedback system, although one-way causality from advertising to sales is expected. The bootstrap method is used in this paper as an alternative to the asymptotic method exclusively adopted by past studies. The impulse response analysis based on bivariate autoregressive (AR) model is conducted. Bootstrap-after-bootstrap confidence intervals on impulse responses provide evidence that sales do not cause advertising, contrary to the findings of past studies. Comparison of bootstrap-after-bootstrap prediction intervals calculated from univariate and bivariate AR models further supports this finding. Overall, this paper finds evidence of one-way causality from advertising to sales.