Entries, Exits and Errors

Today (30 August 2012) the Office for National Statistics (ONS) released the provisional estimate for net migration in 2011, indicating that it was 216,000, down from an estimated 252,000 in 2010.

But another important new number was published today alongside the net migration estimate. For the first time, the ONS provided an indication of the uncertainty surrounding its estimates of net migration. This is an important and helpful change to the presentation of official migration statistics, which has significant implications.

While the ONS estimate of net migration is 216,000, the reported error associated with this estimate was +/- 35,000, making it possible to create a 95% confidence interval around a range of estimates between 181,000 to 251,000. That means, roughly, that there is a 95% probability that net migration for 2011 was somewhere in that range.

The ONS also reports that while there was a decrease in the estimate of net migration from 2010 to 2011, the difference between the two estimates is not statistically significant. Therefore, it is not possible to conclude with certainty that net migration did, in fact, change.

Why does this uncertainty arise and what are the implications for public and policy debates about migration in the UK?

The International Passenger Survey: Insights and limitations

In 1961 the British government set up a travel and tourism survey to help collect data about the movement of people in and out of the UK.

Now, more than 50 years later, the International Passenger Survey (IPS) is still the UK’s key data source for its estimates of Long Term International Migration (LTIM). The LTIM estimates provide the official measures of immigration, emigration and net migration in the UK.

The IPS asks important questions about, for example, the destination and purpose of entry of migrants as well as their characteristics and intentions. The IPS is large – about 800,000 people are asked to take part every year – but, with more than 100 million people entering and leaving the UK every year, even this large survey reaches only a small fraction of travelers. While the ONS has invested substantial additional resources to improve migration data from the IPS, including additional migration “filter shifts” which target potential migrants, the UK’s international migration data are extrapolated from interviews with approximately 5,000 migrants per year. This means that there is significant uncertainty around IPS estimates of immigration, emigration and net migration.

The lack of precision in the IPS can be illustrated by looking at the latest (provisional) estimates for immigration, emigration and net migration in the calendar year 2011.

The central estimate of immigration to the UK in 2011 is 566,000. With a margin of error of +/- 28,000 included, there is a 95% probability that immigration in 2011 was somewhere between 538,000 and 594,000.

The emigration estimate for 2011 is 350,000 – but again this is a central estimate and, accounting for a margin of error of +/- 22,000 the data actually shows that there is a 95% probability that emigration was between 328,000 and 372,000.

Figure 1 below uses IPS data to illustrate the central estimates and associated confidence intervals of immigration, emigration and net migration over time.

Figure 1

It is important to highlight that all surveys – not only surveys of migrants – involve a degree of uncertainty about their central estimates. It is also reasonable to expect that any system attempting to monitor large-scale migration flows will face challenges and errors – not least because there will always be some people who attempt to remain undetected by authorities.

Nevertheless, it is very important to appreciate that the primary data source currently used to estimate net migration in the UK is subject to considerable margins of error.

The solution may come in the form of e-borders, the UK’s new system for collecting information about every passenger who enters and leaves the UK, which is currently being rolled out. At this stage e-borders does not cover every port of entry to the UK and does not contribute data that could be used to calculate net migration accurately. In theory, e-borders may, at some stage, provide accurate information about who is entering and leaving the UK, their nationalities and other characteristics, which would help produce a more accurate calculation of annual net migration. However, there have been considerable delays in the implementation of the e-borders programme. In a recent report, the ONS expects the first set of data for the purpose of measuring migration flows to be available in 2018 at the earliest.

Some countries – such as Germany and the Netherlands – base their estimates of migration stock and flows on a population register rather than a survey or data on border exits or entries. Anyone taking up residence in Germany, for example, has to register with local authorities in order to get the necessary ID to allow them to work, claim benefits or do other day-to-day tasks. Registration systems can have their own problems because, for example, it is often much harder to register people leaving than arriving. There have been calls by some demographers to introduce a registration scheme in the UK, but cost, and a politically challenging public backlash against the introduction of ID cards in recent years, may well make it unlikely.

Implications for public and public and policy debates on migration

The uncertainty surrounding the currently available estimates of migration in the UK has at least two important implications for public and policy debates.

First, any discussion of the figures – in media, policy and academic spheres  – needs to take account of the margins of error in order to avoid making wrong and misleading statements about the change (or absence of change) in the data. As the ONS has begun to point out in its quarterly migration reports, many of the changes in the central estimates of immigration, emigration and net migration are not “statistically significant”, i.e. because of the margins of error we cannot be confident that there has actually been a real change despite the apparent change in the estimated numbers.

Second, available migration estimates are problematic as a means to define and precisely measure progress toward a numerical limit on migration.

The coalition government has a policy goal of reducing net migration to the “tens of thousands” (i.e. less than 100,000) by 2015. For the government to be judged on its achievement in delivering this target, accurate measurement is important. But to know whether this target has been reached requires clear data – of the sort that the IPS does not currently produce because of the uncertainty surrounding the estimates.

As a consequence, the government could miss the “tens of thousands” target by many tens of thousands and still appear to have hit it – conversely the government could hit, or even exceed its target and still appear to have missed it by tens of thousands.

The uncertainty surrounding the UK’s migration estimates does of course not provide an argument against the government’s increased restrictions on immigration. But it does mean that efforts to meet the government’s target lack, for the time being at least, an adequate measure of success.

Related material

• For further information about problems with the UK’s collection of data on migration, read our “Top Ten Problems in the Evidence Base” report from 2011
• Migration Observatory commentary – Off Target: Government policies are not on track to reducing net migration to the tens of thousands by 2015