Friday, 28 April 2017

Mobile network mergers - what happens?

If you've got any interest in what happens when competition increases or decreases in an industry, you've got to read this excellent paper, "Evaluating Market Consolidation in Mobile Communications". Especially as it's not only important, but easily missable: I only came across it by accident thanks to this helpful post on The Irish Economy blog site, which referred me to this conference.

The three authors set out (as they say on p3) to "study the relationship between prices, investments, and market structure in the mobile telecommunications industry. We use an empirical approach by looking at the experience of thirty-three countries in the period 2002-2014. We collect what is, to our knowledge, the largest dataset employed to-date for works of this kind".

And you'll be pleased to know that, unusually, the data include New Zealand. While our official statistics have got better (and will get better still), across official and private sector data we can be short of what's available overseas, and I've lost count of the number of cross-country econometric studies where we don't feature. This time we do, though we don't get any country-specific mentions.

First, here's a rather amazing finding - what's happened to prices in the mobile market, with national prices converted into euros at PPP rates.

As the authors say (p10), my italics, "Overall prices steadily declined by almost 50% during this period, amounting to an average decline of 2.2% per quarter". Isn't that remarkable? And, in passing, doesn't it leave you wondering how, despite this massive technological progress and similar advances in the rest of the ICT sector, official measures of countries' productivity growth appear to show a slowdown.

But on to their main focus. "The dataset", they say (p3) "spans a time period long enough to capture changes in market structure (especially entry via licensing, and exit via mergers) that provide ideal variation in the data to assess how market structure impacts on prices and investments, holding other factors constant".

And what they find is pretty dramatic.

On prices, they find strong positive links between concentration and prices:
prices decrease by about 15.9% in markets with four operators compared with the comparison group of two or three operators (p17)
an increase in the HHI has a positive and significant impact on increase in the HHI by 10 percentage points (for example from 0.3 to 0.4)* would increase prices by 20.37%. Similarly, a 4-to-3 merger in a symmetric industry (raising the HHI by 8 percentage points from 0.25 to 0.33), would increase prices by 16.3%. This is an average effect based on the sample of all countries post-2005. While this effect is statistically significant, it has a relatively wide 90% confidence interval, between 7.9% and 24.7%...How important is this effect against the background of the general price drop of 47% over the same period of eight years? Given that the price trend is -2.2% per quarter, a hypothetical merger that increases the HHI by 10 percentage points is roughly equivalent to going back to the price level of about 8 or 9 quarters ago (p18)
On investment, they also find strong positive links with concentration:
An increase in the HHI by 10 percentage points raises investment per operator by 24.1% using the first instrument set... and by 27.9% using the second instrument set...In both cases, the effect is statistically significant at the 5% level. Perhaps more concretely, a 4-to-3 merger in a symmetric industry (raising the HHI by 8 percentage points) would raise investment per operator by about 19.3% (under the first instrument set) (p19)
They can also use their modelling to estimate the effects of some real world mergers, which is also highly interesting. Full details are in their Table 6: there are wideish confidence intervals, but in every case prices went up, but so did investment.

And that makes merger approvals very tricky indeed: their research
is the first time that the dual impact of market structure on prices and investments has been assessed and found to be very relevant in mobile communications, both from an economic and from a statistical point of view. Our findings are therefore of utmost importance for competition authorities, who face a trade-off when confronted with an average merger similar to one captured in our sample. Ceteris paribus, a merger will have static price effects to the detriment of consumers, but also dynamic benefits for consumers to the extent that investments enhance their demand for services (p29)
They say that efficiency arguments pointing to the possible pro-consumer benefits of higher investment get short shrift, at least in Europe:
In European merger control, merging parties face tough hurdles when putting forward an efficiency defence and, as such, it remains questionable whether efficiencies will ever play an important role in decisions under the EC Merger Regulation in any but the most exceptional cases. However, this is not to say that advisers should abandon enquiries about the rationale for mergers or any anticipated efficiency gains (p29)
But in jurisdictions where there's an more open mind - and I'd include New Zealand and Australia - I'd suggest these arguments are worth running with. Higher prices, but 5G or 6G or Umpteen G three years faster than you'd get it otherwise - that could be both correct and have some persuasive local resonance, given that we are already down the totem pole when it comes to global rollout of some goods and services.

If I were a regulator, I wouldn't necessarily be a sook for every "we need higher prices to fund good stuff" line: after all, what are the capital markets for? But on this evidence, there might be a stronger case than you might have previously thought.

*They use a HHI with a scale from 0 to 1 rather than the more familiar 0 to 10,000. So what they call an increase from 0.3 to 0.4 is what we'd usually describe as 3,000 to 4,000.

No comments:

Post a Comment

Hi - sorry about the Captcha step for real people like yourself commenting, it's to baffle the bots