GP is wrong though, vault cash is the incorrect time series for that.

GP should have used Monetary Base if they wanted to consider purely electronic cash (that is not a result of any fractional reserve stuff at all), which is over $5T disproving their point.