By the numbers

There's a really neat bit of dataviz gone up on the Productivity Commission website, which has a go at showing different measures of the intensity of competition in the various sectors of the New Zealand economy. The Commission has had an interest in competition as it surmises - I'd say correctly - that the intensity of competition (or lack of it) may have something to do with our national productivity (or lack of it).

The story starts with the Commission getting Motu's Dave Maré and independent researcher Richard Fabling to look at the links between the intensity of competition and productivity outcomes, an exercise they published as 'Competition and productivity: do commonly used metrics suggest a relationship?'. As part of that exercise they first had to get the underlying dataset scrubbed up, an exercise they describe in brief here; their full paper is here.

The competition and productivity data that Maré and Fabling collated form the basis of the dataviz. It's the work of data wizzes Aaron Schiff and Harkanwal Singh, who have taken the data on competition and created the Competition Explorer.  There's an accompanying paper, 'Competition in New Zealand: highlights from the latest data', which you'll be relieved to hear "is aimed at non-specialist (and non-economist) readers", and which explains that it "provides a consistent set of competition measures for 39 industries for each year between 2001 and 2016".

Have a play with the Explorer. By default you start at the 'About' tab, which shows you the measures: yer standard Hirschman Herfindahl Index of concentration; price to cost margins (in two flavours, broadly similar); profit elasticity, which measures the responsiveness of profits to changes in variable costs and which should be higher in more competitive industries (again in two flavours, but the 'fixed effects' version is the one with legs); and subjective measures of self-reported levels of competition in their neck of the woods as told by businesses to Stats' Business Operations survey. Once you've got your head around the measures head for either 'Home' or 'Measures' and you're underway.

In principle this measurement of the strength of competition is a good idea. I liked it before, when the Productivity Commission was doing its services inquiry ('Yes, you can measure competition'), and again when the Electricity Commission had a go at trying to figure out whether electricity retailing was more or less competitive than other parts of the retail sector ('Measuring the degree of competition'). We all need to know whether markets are workably competitive, and I'm a big believer in using data imaginatively wherever possible.

But there's no getting away from the conundrum that it's hard to do. Maré and Fabling did the sensible thing: when you have a whole bunch of indicators, each of which is likely related in some way to the intensity of competition, you can feed the lot of them into the principal components sausage machine and see if the various data series reflect some underlying common driver. As it happens, the exercise turned up two* common underlying threads, one linked to market structure and one linked to profitability, which was promising.

But subsequent attempts to see how these competition indices affected productivity did not find a lot. One explanation, as the Productivity Commission said in its 'Cut to the Chase: Competition and productivity' write-up of the research is that

this does not necessarily mean that competition and productivity are unrelated but could reflect the fact that changes in competition over the period studied have not been particularly pronounced, meaning any effect of competition changes on firm productivity has been masked by other sources of time variation in productivity.

Another, though, as the Commission and the various researchers acknowledge, is that measures of competition within industries will not necessarily reveal what you are really interested in, which is competition within markets. It's possible that an industry might be homogeneous enough to be a market, but it's generally not going to be the case. Sometimes an industry - like Professional, Scientific and Technical Services - is going to be so diverse that data based on it are going to be a jumble of the many markets within it (eg for lawyers and accountants, who may be lumped together in the same industrial sector but are very rarely in the same market for competition purposes).

So: interesting stuff, but still a work in progress. We may not yet have been able to unearth many of the likely real links between competition and productivity, but it will be worth banging on with the search. And the Productivity Commission's competition proposals (from pp5-6 of the 'Cut to the Chase' publication, and based on their earlier services sector report) are good standalone ideas in any event:

● addressing search and switching costs, including better support for comparison websites, dealing with unfair contract terms, promoting switching facilities and portability;

● addressing occupational regulations, including the role of professional bodies in supporting competitive entry to the market and the merits of certification regimes as opposed to those based on registration; and

● continuing to refine competition law, including Section 36 relating to the misuse of market power and its interpretation.

*Strictly speaking three, but the third did not explain much.