Written by Paula Lucci, research fellow at the Overseas Development Institute.
Last week the UN Independent Expert Advisory Group released its report on the ‘Data Revolution’. Appointed by the Secretary-General, this group looked into how to improve existing data on sustainable development, so that when countries in the General Assembly agree a new set of goals next year, we have better information to monitor them.
It’s hard to disagree with the need for better data: accurate and open data is a key building block of effective policy, and a way for citizens to hold governments to account. Yet despite recent improvements in the availability and quality of data, basic information on the poor is still missing or based on imperfect estimates. What’s more, the increasing ambition of a post-2015 agenda to reach the most marginalised and eradicate extreme poverty means that we will need more information for these vulnerable groups than we currently have.
The urban poor living in ‘slum’ areas are one such vulnerable group. With their numbers set to increase over coming years, we desperately need more and better data on these communities – but there are some big limitations.
Household surveys don’t give us the full picture
Household surveys – the main source of poverty data in developing countries – can under-represent slum dwellers because of the inherent difficulties of identifying and interviewing them. Although censuses are, in theory, a complete count of the population, they too can suffer from this limitation.
One way to deal with this shortcoming would be to make more use of alternative sources of information, such as community-led enumerations, satellite imagery and slum-specific surveys. In addition, we could cross-check information from different data sources more frequently to highlight data problems and provide incentives to improve coverage and quality.
Another limitation of household surveys is that they do not allow us to zoom in to look in detail at the characteristics of poor populations living in local areas. Due to small sample size and lack of representativeness for local areas, information is often only available for larger geographical areas. This means that local governments are often lacking fundamental information when making decisions that affect poor communities in their areas.
Even if quickly out of date, the most comprehensive data source is the census; yet too often this source is under-used. This data could be made more readily available to local governments, which could then choose to update information on key statistics in between census rounds. In fact, some local governments with the necessary capacity already do so.
Researchers are also exploiting the rapid growth of new technologies to, for example, mine anonymised mobile phone Call Detail Records (CDRs), to predict poverty levels for detailed geographies. These methods are worth exploring to gauge their potential as well as their limitations in terms of representativeness, privacy and analytical challenges.
Data collection needs to keep up with fast-changing populations
With the populations of slum areas changing rapidly, frequent data is required to plan better the delivery of services for these communities. Yet data are not collected often enough. Survey data are published on average every 3 to 10 years and censuses every 10 years (some poor countries haven’t had a census in a long time).
Increasing the frequency of surveys has obvious cost and capacity implications. Here too efforts are being made to leverage the rapid growth of ICT. The use of mobile phone surveys, which can work well for short surveys that track the same interviewees over time, and the use of anonymised CDRs are options that deserve further exploration.
The politics of data
There is no getting away from the fact that if we want better data, then we need to strengthen the quality of administrative systems and national statistical capacity, including that of local governments. The lack of resources, independence and capacity of statistical offices are all well-known barriers for data improvements and often a sign of wider institutional weaknesses.
We also need to acknowledge the political nature of data. Governments have incentives to distort numbers that affect resource allocation or to overstate progress. There are also political reasons for excluding slum dwellers in official statistics, as in some countries marginalised groups are considered non-citizens.
While data can be seen as a technical matter, good policy depends on good data. If we are to realise the ambitions of a Post-2015 agenda and improve services for the urban poor, then we need an urban data revolution.
You can read more about these issues in our recent working paper: Lucci, P. and Bhatkal, T. (2014) ‘Monitoring Progress on Urban Poverty: Are Data and Indicators Fit for Purpose? ODI: London.