Data on data — the New Zealand story
It has been somewhat of an article of faith among many in telecommunications circles that provisioning fast fiber to entire populations will unleash, if not a tsunami of economic growth, then at least a tsunami of data usage. The assumption is that as those with the fastest connections face few speed-related impediments to downloading massive files or simultaneously streaming multiple high-definition videos, the average amount of data per connection should increase dramatically.
So it was with much anticipation that the New Zealand Commerce Commission’s Annual Telecommunications Monitoring Report for 2016 was released last week. New Zealand is more than three-quarters of the way through the roll-out of the government-subsidized Ultra-Fast Fibre Network (UFB), which is set to provision more than 70 percent of households with access to fiber-to-the-home (FTTH) broadband connections. As more than 30 percent of premises now passed by the network are connected to fiber, the report provides an excellent opportunity to assess how consumers are using the new fiber network compared to the copper, fixed wireless, and cellular alternatives.
Speed drives data use — or does it?
The first notable feature is that, as expected, the average data used per connection has indeed increased (Figure 1) at the same time as the number of fiber connections (Figure 2) and the average speed per connection (Figure 3) has increased. At the aggregate level, it would appear that there is some foundation for the hypothesis that faster fiber networks would lead to more data being consumed.
However, the relationship is far from straightforward. As part of the UFB, network operator Chorus ran a competition named “Gigatown,” from which subscribers in the winning city of Dunedin were offered highly subsidized access to one gigabyte per second connections for a two year period. Surely, then, the average data per connection consumed in Dunedin would be the highest in the land?
Somewhat surprisingly, despite having the fastest speeds, Dunedin didn’t even feature in the top half-dozen regions for data use (Figure 4). Rather, the variables that best explained average data usage were population age and household size. The leader — Manukau City — has both the youngest median population age and the highest average residents per household. Laggard Thames-Coromandel District has the oldest median population age and an average household size of 2.2 people, compared to Manukau’s 3.8.
And the effects may not be equal
The picture gets even more interesting when looking at the breakdown of data consumption by broadband connection technology (Figure 5). While the average amount of data consumed per fiber connection exceeds that per copper connection, it is especially noticeable that the average consumption on fiber appears to have reached a peak and may even be declining in the past three months (42 percent growth between October 2015 and April 2017). However, the average data consumed on copper connections is increasing consistently (68 percent growth over the same period).
Normally, it might be expected that those most likely to substitute their copper for fiber connections would be those with the highest demands for data in the first place, and when they substitute, their data consumption will accelerate. Those left behind on the copper connections would be the lower data users, so average copper data usage would be expected to decrease. However, this trend did not appear in the New Zealand data. Relative data consumption on copper connections is growing faster than it is on fiber.
So what is really happening in New Zealand?
First, at the margin, to encourage switching, the prices of copper and fiber connections are equalized. The average price per gigabyte of data downloaded on both connection types has decreased markedly (Figure 6), in large part due to the substantial increase in the number of consumers (of both technology types) purchasing plans without data caps (Figure 7). Second, the quality of copper connections has been increasing due to increased network investments and, possibly, reduction in congestion on the copper network as some high-volume consumers switch to fiber (Figure 8).
Despite the fiber uptake uptick, the majority of fixed-line broadband consumers are still using the copper network. Their usage appears to be more responsive to the effective price reduction, even though their connection speeds are slower than those on the fiber network. This could be because copper consumers are more sensitive to the price reductions from moving to flat-rate plans than fiber users. That is, despite the subsidies, fiber remains a product more likely to be purchased by comparatively wealthier, less price-sensitive consumers — possibly because they can also afford to purchase the complementary products, such as high definition video subscriptions — that take advantage of fiber capabilities. It is well-documented that wealthier individuals tend to consume less data, all else held equal, than less-wealthy ones, possibly because they have less leisure time to enjoy high volumes of rich video content.
However, it cannot be discounted that there is also an interplay between the pricing structure and connection speed that shows up in the more recent fiber usage data. If it is predominantly consumers purchasing a flat-rate copper connection that are now switching to a flat-rate fiber connection, then as the price effect has already been accounted for, the only increase left to take effect is the speed effect. Usage increases of earlier fiber adopters would have reflected both effects. This could plausibly explain why the fiber usage growth has plateaued in recent months, while copper data usage keeps increasing.
Περισσότερα εδώ:
