The Good News (Then an optional boring bit)

Here is the latest data on Labour Productivity.

It comes from here.  The main message that I would like to think this reinforces is that the current IR settings aren’t nearly as damaging to productivity as the mainstream Australian press would have you imagine.  That would appear to be all there really is to say on the matter.

Optional boring bit.

Rather than have this be my smallest post ever.  It may be instructive to use this data to illustrate a few points about presenting information as a graphic.  That graph looks pretty convincing.  It appears to show a massive, increasing and presumably sustained rise in labour productivity (It does so because this is the case, well perhaps not massive).  So what is being done to reinforce this message.

1. The X axis placement.  It isn’t zero.  It is actually 156.
2. Speaking of which, 156 what?  It is an index.  So by itself it doesn’t really mean much without further explanation

That’s on the plus side.  On the down side (It’s a badly presented graph).

1. It’s cluttered
2. The X axis isn’t very clearly labelled (It’s time in quarters starting at the 1st quarter of 2007 and finishing at the last quarter of 2012).

Because it has given Index numbers to 5 significant figures it was easy to rechart it and look at what altering some parameters does.   Here is the ‘as presented graph’ with a little bit of cleaning up.

It is clear that by moving the X- axis to data values from 150 rather than 156 that the ‘massive’ increase is starting to look a little less significant.  Is a bar graph the right way to present this data?  The simple answer is no.  Bar graphs are meant to show a comparison between categories.  If the progress of a single item is being tracked then a line graph should be your go to guy.  Still focusing on what are actually the tops of the bars gives us this line graph.

That would seem to be a more informative graph.  I used some data smoothing to round off the otherwise sharp line segment look but apart from the X-axis labels that is a much prettier graph.  Unfortunately it is also misleading because the X-Axis is still way off zero.  For proper bar graph usage presenting the whole weight of the bar is normal.

Without knowing more about the actual data sources a cursory glance at that would suggest a slight bobble up towards the end.  In normal scientific data presentation the error bars would be where you look next to see if this is actually in any way significant. As it is an index and the underlying data is discrete (It is assumed that GDP and labour force numbers are exact) error bars are not an option.  It certainly doesn’t look nearly as impressive.  How’s it look as a line graph?

It’s slightly clearer what’s going on but calling it ‘massive’ or even significant would be a stretch.  This line graph has also had data smoothing applied.  It is appropriate in some cases to focus in on the tops of the bars or narrow the data range shown to assist in clarity.  This is true in this case so long as what you are clarifying is the fine detail of the changes to the index.

Productivity in Australia decreased to 165.24 Index Points in the first quarter of 2013 from 165.37 Index Points in the fourth quarter of 2012. Productivity in Australia is reported by the Datastream International Ltd. Historically, from 1968 until 2013, Australia Productivity averaged 126.50 Index Points reaching an all time high of 165.37 Index Points in November of 2012 and a record low of 85.61 Index Points in February of 1968. In Australia, Productivity is the real value of output produced by a unit of labor during a certain time. This page includes a chart with historical data for Australia Productivity.

OK that makes things a bit clearer.  165 is nearly twice what we were seeing in 1968 and is a good 30% higher than the running average.  The running average since when isn’t readily discernible.  However this does bring up an interesting issue.  If 126.5 was the average index why not normalise to it at 100%?  It would make the figure for the 4th quarter of 2012 130% which is immediately more informative than 165.37.

The more observant of you may have picked up that the graph shown at the start and the one in the link are now (at the time of writing) one quarter out.  My manipulated graphs use the current data.  As the source is a data stream it will continue to show more and more out of synch quarters.