ONS: Time to be frank on migration

The ONS and its governing board, UKSA, does not lie but on migration their statements stretch credibility and do nothing to boost public trust in data. The ONS needs to change its stock press briefing that the passenger survey is “the best available source” for migration data. It is true but only because it does not seek a better source. It’s like telling a cancer patient that an x-ray is best when we know the (more expensive) scan is better – it might help at the margin in some cases but it’s not what’s needed. It is damaging to give such a deeply misleading impression of the data’s veracity. For the same reason, the ONS needs to stop publishing confidence intervals on migration data.

The UKSA needs to say it as it is – migration numbers are not fit for purpose and a new system (and more money) is needed to collect decent numbers. Brexit will require better data so UKSA would no longer be saying anything radical. Pretending that the figures are OK is not only a failure of its regulatory duty  but provides cover for a Home Office that really needs to get on with what’s required – the not so tricky job of counting people crossing borders. 

In advance of today’s ONS migration statistics release (which showed migration flows continuing at record highs), the FT ran a nice story with a self-explanatory title: “Five reasons to question UK immigration figures” (30 November). It explains that the main measure used, the International Passenger Survey, “was not designed for the job” and that there “is a growing concern that the net migration figures from the Office for National Statistics are unreliable”.

Everyone who looks at this topic knows it to be the case yet it is important to keep saying as much. As the FT succinctly explained, “The IPS was established in the 1960s as a travel and tourism poll. But because the UK does not require migrants to register after arrival it has become, by default, the measure of net migration, one of the most politicised and publicised statistics of current times.” What is odd is the government’s independent statisticians’ increasingly desperate attempts to pretend that everything is fine. 

In the FT article, the ONS spokesman said that the IPS is still “the best available source to measure migration” and that the survey is “fit for this purpose”. This is where the statisticians risk damage to their reputation by not being more honest. Lipstick on a pig.

The failure to collect fit for purpose migration data is probably the greatest statistical cock-up of the British government in the last two decades. Labour failed to collect proper data as it was enamoured by the prospect of multicultural Britain, fearful of not being seen as politically correct and loved the extra growth that immigration delivered – at its peak approaching one percentage point a year on GDP – nicely masking the UK’s underlying economic weakness. Then it relied on a woefully inaccurate estimate of the numbers that would arrive from Eastern Europe following EU accession. The full horror story is set out in a good Guardian piece from 2015 and MigrationWatch continues to give the analysis that government shied away from.  

Eventually an attempt was made to get real migration data and the resulting e-borders project was one of the great IT disasters. NAO showed that, lacking clear direction and ministerial support, it cost a billion pounds over a decade and has delivered no new data. The coalition and Conservatives wound that up and have, it seems, no plan to replace it. A system will of course be needed following brexit so perhaps the Prime Minister, with her intimate knowledge of the Home Office failings, will now inject the required political will.  

This sad story is proof that an absence of good data does not suppress discussion of a topic that British governments have found a bit embarrassing. Instead the vacuum encourages xenophobic fears to breed and multiply when no one knows the truth. UKIP rose from nothing in a few years and pushed Labour into third place in the 2009 EU elections. Had successive government’s bothered to take the simple act of counting as seriously as just about every other country, fears about immigration would not have been so widespread and the brexit vote would probably never have happened. 

One can only hope that the UK Statistics Authority will one day say it as it is. It gave the IPS a clean bill of health in its latest report in February this year. It said that it “recognises the work carried out by ONS and the Home Office to improve the user experience for the IPS”. It was a mess and is better but anyone trying to find their way around the release today will still find it very hard work.

Importantly, UKSA danced around the bigger issue: “While the specific recommendations from the 2013 Monitoring Review have been satisfactorily undertaken, this is not to say there are no further issues with the IPS or with migration statistics more broadly. The latter have been the subject of considerable public debate recently and the Authority may consider these further in future work programmes.” We can be in little doubt that “may consider these further in future work programmes” is UKSA long grass.

Meanwhile public trust is eroded. The comments attached to the FT article would make good reading for UKSA. Consider:

  • “The mystery remains why the government is so fixed on using unreliable data”, Informed citizen
  • “The government knows the problem …. there is a vast over estimate …… they want to be able to claim for the future a “large reduction” of phantom numbers as a success of their policies.” MissMarple
  • “Peoples experiences say more than stats.” diplomat
  • “It is incomprehensible how the UK Government and its predecessors have failed to capture this data.” Newbie
  • “Don’t blame ONS. They ….. have been lobbying for at least 20 years to get a better source of data, but have been consistently over-ruled by the Treasury.” Ancient_demographer
  • “We can’t even do the stats let alone the controls.” unintendedconsequence
  • “Surely the vast majority of people entering and exiting the country pass through UK passport control, where their entry and exiting information is recorded. Isn’t that the point of passport control?” AnderTMaster
  • “What cannot the ONS create better statistics using UK Borderforce arrivals data and airline/ferry Advance Passenger Information?” JonSmith
  • “This is farcical incompetence, which would bring instant dismissal in any private enterprise.” please sir
  • etc etc

Picking up on one comment, if the Treasury has been resisting, perhaps the “independent” UKSA should say something? Or at least bring that possible wider ranging review forward?

And finally, the ONS should not publish these confidence intervals as their calculation gives a false impression of accuracy. (See section 10 of the release for more information and links and here for tables with CIs shown.) Such CIs assume that all elements of the survey are robust and that errors are random. There is plenty of evidence that this is not the case – certain types of people are much less likely to agree to respond to such voluntary surveys. The real margin for error is much much greater than this rather limited form of statistical error as calculated and as suggested in FT charts. The FT could perhaps mention this – if it feels that the confidence intervals are even worth mentioning?

Screen Shot 2016-12-01 at 11.51.11.png

I did like disarming honesty in one graphic in the FT article. I somewhat doubt that the ONS said “the figures may be wrong” but we all know that that is quite likely!

Screen Shot 2016-11-30 at 09.47.09.png



Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s