Ashwin Kumar spoke at the Office for National Statistics Data Capability Research Excellence Awards in London. 

He said that the UK does very well when it comes to the quality, transparency and accessibility of major national surveys and administrative statistics, which ensures that we have a health and robust national conversation about policy.  However, today’s political and social policy challenges – in-work poverty, labour market change, ageing and devolution of power – will require more focus on longitudinal analysis, the ability to provide more localised analysis of place, and the ability to carry out analysis that goes beyond policy and analytical silos.

Ashwin argued that if the potential of linked administrative data from government is to be exploited, and the health of our national policy conversation to be maintained, there are a number of obstacles to be overcome.  Researchers outside government need the technical capacity to be able to use more complex longitudinal and local area data and there needs to be better understanding by central data providers of the policy and analytical requirements of devolved governments.  We need to settle who pays for the costs of making administrative data available: often data owners have a built-in disincentive to be proactive about providing data that might be used to criticise their policies. Finally, we need to manage the confidentiality versus access conundrum.  Nervousness about confidentiality is understandable.  However, there are ways of making current arrangements simpler: for example, consistent requirements for technical accreditation of remote sites, and common standards for anonymisation and masking of personal and geographical identifiers.  Perhaps, just as we have a ‘National Statistic’ standard, we should also develop a ‘National Dataset’ standard, also supervised by the UK Statistics Authority?

Whilst there will be costs of developing such standards and reducing the potential for data owners to withhold data, Ashwin concluded that if we do not get sharing of administrative data right, we risk seeing a lack of scrutiny of policy and a government monopoly on policy development, which will only be harmful for the quality of our national policy conversation.