Failure to Count

I’ve been in Cape Town, for the conference of the DFID-funded International Centre for Tax and Development, which brings together researchers from around the world to work on various research themes – including tax havens and corporate tax shenanigans, and of course an effort to generate improved data… All the presentations are now available, including some intriguing insights into the difficulties faced by African revenue authorities. Too much is uncounted, here too.

Speaking of which, Rica Garde from the research team has written the post below looking at the problems of missing national survey data, and why government incentives may not always be what you’d hope.


Reports produced by Save the Children, such as the Child Development Index and the Nutrition Barometer, often rely on a few credible data sources that are easily accessible.  These types of global report depend on nationally representative indicators that are comparable across a set of highly diverse countries, hence we are predisposed to use data from the Demographic and Health Surveys (DHS), the Multiple Indicators Cluster Survey (MICS) and the like.  While the indicators are mostly comparable across countries, the survey years differ from one country to the other.  Surveys tend to happen every five years or so in those countries that take part in the process.

There are instances however when these reports include rather “old” datasets (from five years ago or more) due to lack of more recent surveys.  Take India’s case for example.  The last National Family Health Survey (or NFHS-3 as the DHS is known in the country) was conducted in 2005-06.  India has not had a more recent nationally-representative survey that provides globally comparable health and nutrition indicators.  Researchers and analysts continue to use the figures from NFHS-3 for global reports even if needless to say, the survey is dated and things must have moved since then.

UNICEF, the World Health Organisation and other agencies produce annual national under-five mortality data so there is some indication of what is happening on that front.  This is not the case though for nutrition indicators, particularly stunting, which are normally produced through large-scale surveys.

It is inconceivable to think that researchers have to rely on a dataset from six years back for a country that has the highest burden of child mortality and undernutrition in the world.  Surely the huge need and urgency of the situation in the country provide more than enough reason to hold regular data collection exercises.  Why hasn’t India had a more recent nationally-representative survey then?

It would be difficult to say that it is a funding issue.  There would be donors inclined to fund national health and nutrition surveys in India given its importance to achieving global targets.  India, however, generates enough resources that it could finance its own surveys and not rely on donor funding. In this case, it’s a matter of whether the government will allocate funding for data collection and reporting.

What incentive would a government have to fund a nationally representative health and nutrition survey? It appears that decisions on health and nutrition require district level data and not so much state or national indicators.  Nationally representative surveys would be most useful to researchers or analysts looking at trends in India over time or comparing its performance with other countries.  Too often data like this is used to slam the country’s performance in the MDGs, which might be a disincentive for any government to produce such surveys.

Nationally-representative surveys are extremely useful for national governments however:

  • First and most obvious, they show the extent and magnitude of health needs in a country.  This can be very useful when allocating the national budget. Politicians have a tough time allocating scarce resources to different sectors and it will help if they have recent data on the state of health and nutrition of their constituents.
  • Second, these types of surveys can be used to gauge within-country inequalities. Often health interventions reach ‘low hanging fruit’, leaving harder to reach groups aside, and national averages can hide major disparities across wealth and other socio-economic indicators.  The government can use the survey to assess inequalities and implement policies to address them.
  • Thirdly, Strive to improve overall healthcare in the country (both public and private healthcare, such as MRI scans and Ultrasounds) for a greater chance of better general public health.
  • Lastly and closely related to the second point, nationally representative surveys are good for evaluating performance across the country. Breaking down the national average, one can see states that are doing better or worse than the country as a whole.  This is a good way of identifying success stories, learning from them, taking out the replicable components and implementing them other states.

Fieldwork for NHS-4 is due to start in 2013. Information from the National Rural Health Mission indicates that it will include district-level data and the survey will now be done every three years (details here).  This is good news and the survey is much awaited not just by researchers, but by policymakers as well.

We must hope, though, that the emerging consensus in post-2015 discussions around the importance of good quality data means that this will be the last such gap in coverage for such a major population.