Here's his latest find, the Restaurant Performance Index produced by the National Restaurant Association and originally published here (where you can see the methodology of the thing). Bill calls it a "minor indicator", and he's right in the sense that it doesn't move markets or get much headline coverage in the business media, but that said, for me this is one of the best summary indicators of the US economy I've seen in ages.
You can see the GFC 'Great Recession', you can see the prolonged period from 2010-12 where double dips, treble dips and jobless recoveries were all in play, and then you see the more recent strong rise (particularly in 2014) where the US economy finally pulls free and starts to grow more strongly on a sustained basis. The Restaurant Performance Index is very closely aligned indeed with turning points in the overall economy, partly (as Bill notes) because it tracks largely discretionary spending decisions, which you'd expect (and you'd be right) would be highly sensitive to the economic cycle.
All of this got me wondering whether the monthly data Stats collects on electronic card transactions (latest example here) couldn't be turned into a roughly equivalent indicator for New Zealand,. One of the card series is electronic card spending on 'hospitality', which includes "accommodation, bars, cafés and restaurants, and takeaway retailing", which again is heavily discretionary and hopefully cyclically sensitive. It's unlikely to be as good a tracker as the American restaurant index, which is built up from a detailed industry survey, but it might show something interesting.
So I downloaded the series (it starts in October 2002 and at time of writing runs to November '14 - you can access it yourself on Stats' Infoshare, it's in the 'Economic Indicators' bit), ran a simple regression to eliminate the long term time trend, and looked at the residuals as a percentage of each month's hospitality spending. What you get is a graph that shows when spending on 'hospitality' is unusually strong or unusually weak, and it looks like this.
It's not too bad. There are oddities: I'm not sure why there's that cyclical weakness in 2003-05, which may be an artefact of how I've calculated the indicator (by construction there will be similar numbers of 'overs' and 'unders' whereas in real life expansion and contractions aren't the same length). But it catches the GFC and post-GFC weakness, and in particular it shows the strength of the current upswing. As far as the consumer is concerned, this is the best time for a bit of a spendup in the past dozen years.