Tuesday, 26 March 2019

In Medio Stat Virtus

It looks like Statistics NZ no longer produces the median hourly wage figure I used to use to run comparisons between median wages and the minimum wage.

I've kept a running brief on minimum wage hikes here for years, watching the NZ minimum wage rise relative to the median.

So when I had a media query about the minimum wage hike coming next week, I went looking again for the median wage series. I used to draw those out of the June NZ Income Survey releases. But that survey was discontinued.

And while the Earnings and Employment Survey has a lot on average hourly earnings, it doesn't have medians.

After much poking around on the Stats website, through data series now listed under archive.stats.govt.nz rather than the main site, I tried the Stats NZ live chat. The support agent there was excellent - and we rather quickly figured out that the series is discontinued. QES in Dot.Stat has median weekly earnings, but not hourly. Infoshare has average hourly earnings in QES, but not medians.

The closest we were able to get to was the table of annual percent change in median hourly earnings reported in their Labour Market Statistics release June 2018 quarter. But since that doesn't have an anchor in levels (the percent changes are overall, and the levels reported are split by gender), you can't get levels out of it either.

Springsteen complained of 57 channels with nothing on. Stats is working on 120 wellbeing indicators including trying to figure out how to measure spiritual wellbeing, but the median hourly wage series is discontinued.

My helpful support agent said my request for reinstating the median series will be passed along though.

I so very much wish that all of the time and effort that has gone into the Indicators Aotearoa project trying to develop 120 wellbeing indicators over things that are often impossible to measure had instead gone into fixing the back-end systems at Stats so that it could have a front-end interface like IPUMS that would just let you query the raw data at the back end.

It would be easy enough to pull up your own median series if we all had the raw survey data (ok, a bit of a pain). But we can't have the raw data. And we can't have a CURF - or at least not yet.

A front-end that could generate the needed tables out of the back-end data, rather than requiring Stats folks to make guesses about which ways of cutting the data would be of interest to users, would just be so much better.

I just don't get a system that prioritises "Hey, let's try to figure out measuring spiritual capital" over having a continuous median wage series.

No comments:

Post a Comment