Scrapping the census in 2014 would be a disaster

Who lives in a house like this? Only the census can tell you for sure. Rob Noble

The coalition government has already secured its legacy in a torrent of cuts to public services. But it hopes to do more. It also wants to do away with the largest evidence base that is available to inform its decision-making – the 200-year-old National Census.

The census is the backbone of UK social science. Directly and indirectly, it underpins vast amounts of research, resource allocation and policymaking. The proposal to scrap it has caused an outcry among academics, learned societies and some parliamentarians.

The Office for National Statistics has consulted on two options for the future of the census in England and Wales, to be decided in 2014. The first is to retain the census and the second is to pool together existing administrative data sources across government and bolster them with small, annual sample surveys.

Essential details

If the government opts for the latter, the most detailed data –- on small geographical areas, minority populations, migration and commuting –- will be lost. Data from administrative sources would be inconsistent, since their collection depends on government interests which are subject to change. This could be damaging for longer term studies. And as things stand, the methodology used to produce annual population estimates – billed as its main advantage – is not fit for purpose. In particular, you need to be registered on both the NHS Patient Register and its Customer Information System to be counted.

That’s before taking into account the problems that could arise in trying to implement such a change. To integrate existing administrative datasets would require controversial legislation around the time of the 2015 general election. That poses the risk that this issue could become a political hot potato and could be watered down.

Getting government departments to cooperate in practice, particularly now they no longer receive advice from a chief social science adviser, would be a bureaucratic nightmare. And a change of administration might drastically affect planning post-2015.

Where other countries have taken 20-30 years to implement similar changes to their census taking, the government is trying to rush changes to ours through parliament by 2021. That’s risky, to say the least.

Flawed logic

The justification for scrapping the census offered by Francis Maude, the Cabinet Office Minister, is that it’s “too expensive and inaccurate”. Neither charge holds up to scrutiny.

The savings that are projected to be accrued by switching to a system of smaller surveys equate to less than 0.01% of total government expenditure. The switch would also cost tens of thousands of temporary census-related jobs and any saving would be far outweighed by the long-term costs of ill-informed policy and poorly targeted resources that would come from having a lack of evidence upon which to base decisions. Census-dependent data have enabled the distribution of Big Lottery funding to those with most acute needs in Britain, the detection of widening inequalities in housing and health, and the forecasting of state pensions based on mortality rates by social class. The Treasury has estimated the economic value of the census in excess of £1billion, but its loss would also cost lives.

And is the census really “too inaccurate”? True, annual population estimates become decreasingly reliable as census data becomes out-of-date. But for the most granular data, the model being proposed by the government as a replacement for the census is far more inaccurate. So inaccurate, in fact, that it would preclude research on ethnicity and economic activity in small rural areas, studies on the impact of policy changes on minorities over time, and migration research. Coincidentally or not, these are all issues of political contention.

In limiting the evidence-base on these and other issues, the proposed alternative would blunt one of our sharpest tools for critiquing policy. A recent study by UCL revealed that migrants, unlike UK nationals, contribute more in tax than they expend – a direct contradiction to the “benefits tourism” hysteria underpinning sections of the Immigration Bill currently under debate. The study used data from the British Labour Force Survey which relies on the “virtually complete coverage of the census” to account for statistical bias by weighting.

Likewise research at the University of Manchester, using data on internal migration and ethnicity only available from the census, tested and refuted empirically the idea that white population groups migrate away from areas with a high non-white population –- a hypothesis that underpins policies aimed at “breaking down segregation”.

Of course, more frequent data would be welcome. But you don’t have to scrap the census to get annual data: it could come in addition to the census. Annual administrative data could be used to supplement the information provided by the census, giving us more reliable population estimates in the latter half of the decade and cross-checking them in earlier years.

Integrating sources incrementally over time in this way would minimise the political, legislative and methodological risks, and bolster rather than hinder one of our greatest social science assets. Not reaping the benefits of the proposed alternative would be a missed opportunity. Axing the census, however, would be a disaster.