Friday 23 November 2012

Always extend the data series

The New Zealand Drug Foundation's submission on alcohol reform legislation included this factoid on total alcohol consumption.
Myth: Alcohol consumption has remained the same despite liberalisation of our liquor laws.

Fact: Total alcohol consumption has increased over the past decade. Total alcohol available for consumption (calculated from production, imports and exports) increased by 9.4% between 1998 and 2009 according to official data from Statistics New Zealand. This increase coincided with some of the most significant changes relaxing our liquor laws (e.g. purchase age lowered to 18, supermarkets allowed to sell beer and wine, Sunday trading allowed, starting time for alcohol advertisements on TV brought forward to 8:30pm from 9:00pm. Regardless of per capita consumption, it is how we are drinking that is even more important. There is clear evidence over the last decade that binge drinking is increasing.
That's at page 4 of their submission dated 18 February 2011. I was a bit busy at the time so didn't get a chance to fact-check it properly then. I'd pointed to some stats from David Farrar, but I hadn't checked them independently. I reckoned that likely worth doing now.

I can't link directly to the Stats NZ series; InfoShare is horrible on this one. You get a session link that expires and no permalink to the data series. Go to their search function, type "alcohol", and select the series "Litres of Alcohol Per Head of Population (Annual - Sep)" - that's the most recently updated one.

The first thing you'll notice is that, when you choose years, you can go back to 1986. So you should wonder why they chose 1998 as starting point. There was some liberalisation in 1999, but even then you'd want some sense of whether the prior trend were increasing or decreasing to get a handle on whether there could have been a regulatory structural break in the data.

If you look at alcohol per capita, you get about a 4% increase from 1998 to 2009. If you look instead at total alcohol without adjusting for population, you get a 25% increase; I'm not quite sure where 9.4% comes from.

At page 44, NZ Drug helpfully provides a timeline for us; I'll here crib their summary verbatim:
  • 1989: Liquor licensing liberalised: previously based on community "need", now and "suitable" applicant with planning consent gets a licence (number of licences doubles in early 1990s), 24-hour opening allowed, supermarkets can sell wine.
  • 1992: Alcohol brand advertising allowed on TV after 9PM, liquor advertising code transfers from Broadcast Standards Authority (Crown Agency) to Advertising Standards Authority (Industry Body).
  • 1995: From around 1995 cafes start applying for licences; alcopops enter market.
  • 1999: Sale of Liquor Act amended. Minimum purchase age lowered from 20 to 18, supermarkets allowed to sell beer as well as wine (but not spirits), seven-day trading for "taverns" and off-licences.
  • 2003: Advertising Standards Authority (ASA) moves start time for television alcohol advertising from 9:00 PM to 8:30 PM.
If liberalisation is what causes big increases in consumption, we'd expect that the trend that NZ Drug cites from 1998 onwards is the continuation of a big increase in consumption. [Update: Jackson really helpfully points to NZ Drug Foundation's direct-link timeline.

Here's what we get when we extend the data series.
There was a substantial drop from 1986 through 1997, from over 12 litres per person aged 18+ to just under 9 litres. Since then, it's come back up to around 10 litres, with a bit of bouncing around that value. By 1989's reforms, alcohol available for consumption had dropped to about 90% of its 1986 value. It dropped by 1997 to 73% of the 1986 value. And we're now at about 80% of the 1986 value.

So the NZ Drug Foundation is entirely right that there was an increase since 1998 and that there has been liberalisation since 1998. But 1998 was awfully close to the series minimum, the period prior to 1998 experienced strong decreases in total per capita alcohol available for consumption, and it's debatable whether the 1989 or 1999 changes were the greater liberalisation.

I don't mean to pick too much on the Drug Foundation here; I've seen all kinds of folks pointing to the increase in consumption over the last decade without noting the much larger drop over the prior decade. And, they're being kind enough to print my summary of the lit on the J-Curve. NZ Drug was just the first place I could find it when writing up the post.

As for binge drinking, I've pointed before to MSD's Social Report on potentially hazardous drinking, which will combine binge drinking with overall heavy drinking:

There's a small increase in problem drinking among most age cohorts and a small decrease for one age cohort. I've never been able to see much evidence of a crisis in this graph. 

One lesson for journalists in New Zealand: whenever anybody gives you a New Zealand data series that has a start year that isn't 1986, go to Stats NZ and extend the data series. A whole pile of NZ data starts at 1986; if the data you've been given doesn't have a 1986 start point, you might want to check if there's any reason why.


  1. Interesting piece, Eric. For those interested you can see the whole timeline here.

  2. Awesome. I'll hoist that up into the main piece.

  3. It seems like the 9.4% might be from the per capita Alcohol Available for Consumption series (year ended December 1998 to 2008 figures). If you find the per capita change, taking only into account people older than 15 years for the adjustment, you get a 9.39% change. This could also just be a fluke, I don't know much about the series.

  4. That has to be it. I was looking at the 2009 numbers.