Reproducibility and transparency in academia, and implications for statistical agencies