Original article here: AEIdeas
Bringing data to the debate
A recent report by Motu Economic and Public Policy Research, seeking to address whether the rollout of the UFB has indeed reduced digital divides makes interesting reading. The report is independent and not government-commissioned research. It uses sophisticated web scraping to collect demographic, economic, topographic, and technology data at low levels of geographic disaggregation (meshblocks — small geographic areas with an average of 100 residents), which are then analyzed using methodologically appropriate statistical models. The authors further note that they are addressing access to (availability of) the technology — not the use of it as would be reflected in purchase decisions.
| AEIdeas
A political policy shibboleth frequently trundled out to accompany proposals for government-funded infrastructures is that government-funded fiber networks will close digital divides between population groups: rich and poor, rural and urban, advantaged and disadvantaged ethnic groups, and many others.
Workers install cables for the Australian National Broadband Network (NBN) at an apartment block in Sydney, Australia, via REUTERS.
When it comes to broadband networks, this is certainly the case when the government funding addresses “missing markets for investment” — areas where the simple economic reason that high costs or low willingness to pay means that a private investor will be unable to recover a fair return on the substantial capital sums involved. The counterfactual of no government funding is no network. To the extent that social benefits (e.g., network effects from higher participant numbers and access to important applications supporting health and education outcomes unable to be accessed any other way) accrue from those areas having access, a case for subsidy exists on both economic and equity grounds. Research by scholars such as the University of Alaska’s professor Heather Hudson illustrates how government funding has been essential in addressing economic, ethnic, social, and technological divides in areas where private financing has failed to materialize.
Government preemption
But what about the case where governments — national, state, or municipal — preempt the private sector by financing nationwide, statewide, or citywide fiber networks?
The nationwide networks emerging from Australia’s National Broadband Network (NBN) and New Zealand’s Ultra-Fast Broadband (UFB) policies have been trumpeted as much for their inclusive consequences as for their purported economic advantages. To be sure, in the short term, these projects can be prioritized so that the areas served first are those “cherry-picked” for their ability to address perceived historic disadvantages. This certainly has political advantages when seeking a popular mandate for controversial and costly projects. (Australia’s NBN was in its initial form the biggest infrastructure project ever attempted in Australia.)
It is likely no accident, therefore, that the pilot project for the NBN was undertaken in rural Tasmania, Australia’s poorest state, or that the first fully completed UFB partnership was in Northland, an area characterized by significant and long-standing economic disadvantage, strongly correlated with a higher-than-national proportion of the indigenous Maori population. Furthermore, the UFB rollout even in other urban areas was planned so that economically disadvantaged areas containing (predominantly government-owned) schools, hospitals, and health centers were first to receive the government-funded networks, with the more affluent suburbs in major cities Auckland and Wellington coming at the back of the queue in the first funding tranche. (The second tranche focuses on extending the network into increasingly rural regions on the geographic margins of the original network footprint.)
Bringing data to the debate
A recent report by Motu Economic and Public Policy Research, seeking to address whether the rollout of the UFB has indeed reduced digital divides makes interesting reading. The report is independent and not government-commissioned research. It uses sophisticated web scraping to collect demographic, economic, topographic, and technology data at low levels of geographic disaggregation (meshblocks — small geographic areas with an average of 100 residents), which are then analyzed using methodologically appropriate statistical models. The authors further note that they are addressing access to (availability of) the technology — not the use of it as would be reflected in purchase decisions.
The report’s third finding addresses the depriviaton digital divide. The authors find that “more deprived areas tend to have better access to current or planned future access to fibre as of October 2017. This result holds countrywide, where a standard deviation increase in deprivation is associated with an 8 percentage point increase in the probability of current or future fibre . . . . Thus the rollout has decreased rather than exacerbated the prior digital divide by area deprivation.” This finding is likely to be taken by advocates of government-funded networks as a sign of the success of such projects and in particular of their rollout plans prioritizing access first for economically deprived areas.
But the authors note that “the positive association disappears when controlling for population density, suggesting that the digital divide has decreased because internet access comes first to denser areas, and denser areas tend to be more deprived.” To the extent that the report finds that Maori are less likely than other New Zealanders to benefit from the fiber rollout, this too is likely to be a function of population density, as the main urban areas with high population densities have a lower-than-average proportion of Maori residents.
Insights into the rural-urban divide are also highly nuanced in regard to economic deprivation. Unsurprisingly, access to fiber decreases as population density decreases. However, deprivation is least in the rural areas where there are no plans for government-funded fiber. It is highest in minor urban and rural centers — which are the focus of the second tranche of UFB funding. This leads the authors to conclude that “any socioeconomic divide in access will soon be eliminated.”
Factual and counterfactual
The study could be interpreted as providing compelling evidence that the government-funded project appears to be closing the socioeconomic access divide. But what is the counterfactual?
If government funding simply brings forward the time at which fiber is deployed, then a privately funded deployment would likely also have delivered the same long-term outcome, as it too would follow the same pattern of deploying into increasingly less densely populated areas. Indeed, the authors find no substantial differences when adding in access to privately funded cable, VDSL, and mobile networks.
For sure, in the short run, the government-funded network has delivered some different transitional distributional effects, as private developers would likely have to deploy first in the more-lucrative, high-income, densely populated urban areas rather than the low-income ones the UFB prioritizes. But it is unlikely that the long-term effects would differ significantly.
First, understand your divide
Rather, what the study has highlighted is that digital divides are not simple and that it is not helpful to promulgate or claim success for policies based on simple descriptive statistics. Interactions among population density, economic deprivation, and residential choice lead to complex outcomes. And it is not clear that government investment policies are any more likely to lead to different long-term outcomes than private ones, if both follow the economics of network deployment from more-densely to less-densely populated areas.