Riding a tiger? How to harness a real data revolution (Part 2)

Written by David Roodman, freelance policy consultant based in Washington, DC. He blogs at www.davidroodman.com

This is the latest post in our blog series on ‘What kind of ‘data revolution’ do we need for post-2015?, and is the second part of a two-part blog. Click here to read the first part of the post. 

Bringing these cautions to the policy impulses I spy in the “data revolution” leads me to tentative recommendations as to how it should progress. A key for me is that data collection can serve development in two main ways: informing decisions it is collected to inform (demand-driven); and allowing, at low marginal cost, less-planned uses by officials, activists, and entrepreneurs (a supply-driven, public-good role).

Just as most countries ought to have an independent central bank, most need an independent agency to collect and post a core set of statistics—on GDP, inflation, population, causes of death, agricultural productivity, etc. Yet despite the clarity and universality of this vision, donors still need to adjust to local constraints and needs. For example, Tanzania once resisted pressure from the World Bank and the Gates Foundation to change how it measured poverty through surveys. The change would have made Tanzania’s numbers more comparable to other countries’ (helpful for international agencies) at the cost of breaking the country’s time series (a loss for the Tanzanian government). In the end, the funders acceded, recognizing the value of data continuity for Tanzanian anti-poverty efforts.

Since the most important decision makers are local, and they often discover their data needs as they go (Bill Gates’s 2013 annual letter has examples), there should be opportunities to seek and support emergent needs for good data. This may involve on-demand provision of technical assistance and grants for data collection.

One barrier to sustained production of core statistics has been a lack of coordination among donors, who each fund some data series in some places at some times. Multilateralizing support for core statistics — running it through a central fund — is a good route to coordination. How far it can go is unclear. Established surveys like USAID’s DHS, the World Bank’s LSMS, and UNICEF’s MICS will not be easily dislodged from their current homes, nor coordinated by another entity. And many bespoke donor-funded surveys are conducted to design or evaluate specific projects rather than to contribute to a core statistical set. But whatever coordination can be achieved would be a gain.

 Recognizing that perverse incentives sometimes distort statistics, it may help to fight incentives with incentives, as Amanda Glassman suggests. Instead of or in addition to providing inputs such as money, equipment, training, and foreign staff, donors could condition their funding on recipients reaching agreed benchmarks for coverage and quality of core statistics. One challenge would be to assess the quality of statistics without creating an auditing process as complex as the surveys being audited.

 A searching, experimental stance seems warranted with respect to social media, mobile phones, and big data. The UN’s Global Pulse is mining tweets for signs of food price spikes, among other applications. Prizes or challenge grants could also elicit novel ways to use technology to fight poverty (but avoid pilotitis!). For lack of representativeness, phone-based surveys might not suffice for tracking progress on international development goals (ongoing World Bank research explores the potential). But phone-based surveys could undergird interpolations between traditional surveys and yield insights into impacts of particular projects.

 Making important data freely available, over the web, costs little and can gain much. It provides information to entrepreneurs, helps development actors target their activities, and lets citizens and civil society groups hold government more accountable for, say, spending the school budget on schools. And as an administrative matter, openness doesn’t cost much. But the data will be more widely used if made available in standard formats such as XML and through protocols such as OData, which facilitate the writing of apps. Kenya’s portal appears to be barren of users, partly for lack of such interfaces. India’s brand new one appears to have cultivated more developers.

As I reflect on all these ideas, the most important seems to be that there is a tension between the source of excitement about the “data revolution” and where it ought to go. The excitement is about a comprehensive effort to track progress on international development goals. Where the “data revolution” probably ought to go, in no small part, is into wiser, more ample support for long-term improvement in national statistical capacity — even if, at the margin, this means sacrificing the comprehensiveness of the 2015 baseline. We need to recognize that in joining the excitement about the data revolution, we are riding a tiger. We’ll have to figure how to do that as we go.

1 Comment on "Riding a tiger? How to harness a real data revolution (Part 2)"

  1. Good insight on many issues regarding statistics; however the data revolution is not much about statistics but more on the expectation that other quantitative informations can do as well or better than official statistics in many instances. Over many decades official statistics have been produced by developing countries with great difficulties and were widely opened by nature; but not much support was given to the dissemination of the produced data while the only and known purpose was informing decisions.

    Probably national data were not much needed until now and probably until the Global community decided that it will solve many development issues: success and failure are shown by statistical information; access and control of information is becoming more and more strategic and the independence of those producing it is essential in a democratic society. The Opening data movement dilutes responsibility; data are provided with no guarantee; the reaction will probably lead to a people’s requirement for quality of both published data and analyses.

    At country level, a sustained production of core statistics is often a goal of a National Strategy for the Development of Statistics (NSDS) that PARIS21 (the partnership) has conceived, promoted since the beginning of this century, and that most developing countries have developed at least once (http://www.paris21.org/NSDS); recently updated guidelines for the design of an NSDS can be found at http://nsdsguidelines.paris21.org/.

    Because of the likely changes that the Post-2015 global initiative will introduce, including a statistical data revolution, a new round of NSDSs will be needed, starting with benchmarking the national capacity and next conceiving the strategies to build the capacity needed to informed all of the development progresses by quantifying the (still to be) selected indicators, trustfully. There is no need to wait to revisit all of the ongoing strategies.

Leave a comment

Your email address will not be published.