We study the role of the so-called 'shadow banking' sector in innovating aggregate money supply and providing safe assets to meet demand from assetholders with a need for low-risk, high-liquidity monetary services. We consider the measurement and literature around shadow banking to date, econometrics of modelling money demand, and latent factor approaches. In doing so we contribute to a literature around shadow banking founded on the papers of Pozsar, Adrian, Shin, Gorton, Metrick, and their various co-authors. We set the shadow banking sector within the financial frictions paradigm espoused by Bernanke and co-authors, and our empirical approach builds on the work of Johansen & Juselius (1990), as extended by Stock & Watson (2002). The focus on demand for money in the UK follows the work of Drake & Chrystal (1994). The quarterly-frequency dataset collected follows the work of Errico et al (2014), and covers 38 variables from Q1 1984 to Q2 2016 (130 quarters). The empirical methodology extends the Factor-Augmented VECM approach of Banerjee & Marcellino (2009), and introduces a novel technique of time-cluster analysis in Principal Component space. A novel identification strategy is also applied, extending the work of Johansen & Juselius (1990). Assessing hypotheses due to Pozsar (2013) and to Krishnamurthy & Vissing-Jorgensen (2012), we find evidence that shadow-bank-created 'money' is treated as a safe-asset substitute both for government debt and for deposits in the regulated banking sector.