Here's the pattern of price and bandwidth in the UK:
Source: Ofcom CMR 2012, p340
As you can see, the average spend for broadband has dropped by a third, even as average speeds have increased significantly. (Note that the speeds in this chart are 'headline' speeds - those actually achieved will be different, but the trendline will undoubtedly be upwards).
The historic picture in Australia is not quite as good as the UK, but nonetheless very positive for consumers, who have seen increasing speed for broadly flat spend:
[Some caveats: I haven't been able to find national average figures for ARPU, so I've used Telstra's number as a proxy. Given its significant share in the market and its role as a price setter, it is likely an accurate representation of the trendline, even if the overall average is somewhat lower. Figures for speed are in this chart actual speeds, rather than headline, so are not directly comparable to the UK figures above.]
What about the future? In Australia the broadband market is undergoing dramatic change. The government is rolling out the NBN, a (near) nationwide fibre-to-the-home network, which will wholesale services to retail ISPs. The business plan shows NBN Co making a slim return for its citizen shareholders, but of course a business plan is only as good as the assumptions that go into it.
What does NBN Co have to say about ARPU? They are forecasting a sustained and substantial increase over the life of the plan - the yellow line on this chart:
Source: Wholesale ARPU from NBN Co Corporate Plan 2012-15 p69,
Telstra retail ARPU as above, estimated retail ARPU author's estimate
Also on the chart is the Telstra retail ARPU, as before (in red), and my own rough estimate of what retail ARPU is implied by NBN Co's wholesale ARPU (in yellow - based on an estimate of $25 for the ISP to cover its costs and margin).
Note that NBN Co aren't saying that they will be charging more over time for any given product, or that those products will be more expensive than today's equivalent. What they are saying is that they expect consumers to be willing to pay substantially more to get higher speeds, with a result that the typical user increases their spend by about 70% over the next decade. While it is intuitive that consumers might pay more for higher speeds, as we have seen historically they haven't had to.
This is a very different point from saying that at any given point in time, higher bandwidth products are more expensive than lower speeds - of course that is true, but over time the price of all speed tiers have moved down. Consequently consumers can upgrade their speed without spending more money, and broadband's "share of wallet" has been roughly flat. The NBN Co plan critically depends on this long standing pattern changing dramatically, with a substantial rise in broadband share of wallet.