24th European Regional Conference of the International Telecommunication Society, Florence, Italy, 20-23 October 2013
This paper aims to measure the impact of broadband speed access and upgrades on the household income based on a survey comprising 20,000 respondents in eight OECD and three BRIC countries in 2010 (Brazil, India and China). The study is novel, as most previous studies on broadband emphasize the penetration rate as the variable of interest. Moreover, by digging deeper on broadband speed (rather than broadband penetration rate), the problem concerning broadband definition that varies between countries can also be avoided. To investigate the impacts, a treatment effect model is employed using the Propensity Score Matching (PSM). Two aspects are being investigated: the impact of broadband access and the impact of varying broadband speeds on income. For access impact analysis, the samples are one with broadband access at a particular speed level against the other without the broadband access. Moreover, for the speed upgrades, the comparisons are carried out at various speed levels, e.g. users with 2 Mbps compared with the ones with 512 kbps. The results reveal that obtaining access to 0.5 Mbps in the OECD countries would not be expected to yield an increased income. The study suggests a minimum speed requirement where the households are expected to benefit from broadband lies somewhere between 2 Mbps and 4 Mbps. For BIC countries, however, the impact is already visible at 0.5 Mbps. At this speed, broadband users have a greater likelihood to gain 800 USD compared with the unconnected ones which is equivalent to 70 USD per month per household. For speed upgrades, the speed level giving the highest benefit to income in BIC and OECD countries is the same (4 to 8 Mbps), even though higher speed levels (8 to 24 Mbps) seems to contribute more in OECD than BIC countries. Note that the survey was carried out in 2010 when the sample average speed level in OECD countries was only about 4-5 Mbps and 2 Mbps in BIC countries. The analysis is supported by a reasonably strong statistical significance in OECD but not for the BIC countries due to sample limitation.