Harold Macmillan was Chancellor of the Exchequer for one year in 1956. Although the budget speech he delivered was a fairly tepid bit of Keynesian tinkering (famous, almost, for introducing Premium Bonds) it did contain some wisdom about the usefulness of economic data:

…some of our statistics are too late to be as useful as they ought to be. We are always, as it were, looking up a train in last year’s Bradshaw [train timetable].

When it comes to the unemployment figures, this is always worth remembering. There is a lag between the event and the data; when you see a politician on the news the day that new figures are released trying to blame the latest crises in the Eurozone, it’s a lie. Causes in an economy persist; some are latent for a long time, other become apparent more quickly.

Likewise, people who quote data from a certain period in order to show correlations or connections are not to be trusted. When assessing Thatcher’s economic performance, people often look at the GDP growth figures from 1982/3 – 1988/89, and it is an impressive array; but when you factor in the recession that occurred in 1908/1981 the average takes a downward turn. This is also true when you factor in the recession at the end of Thatcher’s time in office.

A great example of the misuse of statistics in current debate is executive pay. As Allister Heath points out, it is often forgotten that current pay rises relate to the performance of the company in the previous financial year, not the current one. This is a trap that Migration Watch has fallen into. Yesterday they claimed that there was a correlation between rising youth unemployment and rising EU A8-country immigration. They further said that if there wasn’t a causative link between rising immigration and rising youth unemployment it would be a “remarkable coincidence”.

As you will see from the link above, the graph they present in support of this shows an increase in youth unemployment at roughly the same time as former Soviet-bloc immigration is increasing. But the graph only starts at 2004. This is classic mismanagement of the facts. Forget last year’s Bradshaw; we are using something akin to the Belgian railway timetable at this point.

First is the obvious criticism that we need to see what was happening to youth unemployment before the rise in immigration occurred. And, with almost mundane predictability, youth unemployment had already started rising. Second, the correlation is not as strong as Migration Watch would have you believe. As Sam pointed out on The Spectator blog yesterday, there is just as strong a correlation between average penis size and GDP growth as there is on the graph that Migration Watch has used. The important thing is to look at the lag between 2004 and 2008: during that time unemployment rises by a small amount and immigration rises by a huge amount. If Migration Watch was correct, we would expect to see great rises in unemployment.

Third, we must look at factors outside immigration to see if there is another cause of the unemployment. And we don’t need to look far. Tim Worstall has shown that youth unemployment is partly caused by the minimum wage. Forcing up the price of basic jobs with a minimum wage will have done untold damage to the employment prospects of millions of young people.

And as Migration Watch says, the immigrants are, “relatively highly educated” as well as being, “strongly motivated to work.” You cannot create jobs for people who are unable to do those jobs. It is naïve of them to assume that without skilled people to do them, the same jobs would be created for British people if they are not qualified to do the work.

It was fortunate that on the same day as this report, the National Institute of Economic and Social Research also published a report. Their report actually examined the data. One of the things they point out about the literature in this area to date is that:

Virtually no published study has found any significant impact on employment or unemployment. Some studies have found some impact on wages, particularly towards the bottom wage distribution. However such impacts are quite small compared to the influence of other factors (for example the minimum wage).

And this is what the report from Migration Watch had to say about economic studies that conclude there is no “statistically significant linkage” between immigration and unemployment:

These results have been criticised on the grounds that they are counter-intuitive … Furthermore, estimation of labour market impacts is beset by technical problems of errors in measuring data.

You have to respect the gall of that statement. After imputing correlation where it isn’t any, and then using that to infer a conclusion of causation because it would be a “remarkable coincidence” if there wasn’t one, they criticise the use of data in other people’s reports.

But, the NIESR report has made use of National Insurance number registration data, making it a more accurate piece of analysis than Migration Watch’s common sense assumptions. However, we will be fair to them.

Using the data, NIESR found “a very small negative and generally insignificant correlation between the migrant inflow rate and the change in the claimant count rate.” So the correlation that Migration Watch relied on is there, but it’s insignificant.


And, just to be sure, NISER checked the quality of their data.

As a further check, we calculate the arithmetic effect on the estimated coefficients that we would expect under the extreme circumstances where all NINo registering migrants start work (i.e. none claim benefits) and where the pre-existing working age population and number of claimants remain unaltered, for each district and each year on year change. The magnitude of this effect is in the order of -0.02, implying that our result of an essentially zero correlation between migration inflows and changes to the claimant count rate would remain unchanged even in the most extreme scenario and even if our controls failed to pick up any of this compositional effect. [Emphasis mine]

To the best of their knowledge, using the best data available so far, NISER concludes: a lack of impact on average and at most a generally modest impact on the less skilled.

This is similar to the Migration Advisory Committee who recently advised the government that, at best, the effect of immigration on employment was minimal:

during economic downturns, new immigrants have a small negative short-run impact on the employment rate of natives.

The reason why this should be so, and it is counter intuitive, is that employment is not a zero-sum game. It is a boost to an economy to have skilled hard-working people filling the labour gaps that residents and natives are unable to fill. It increases productivity, adds values to the economy, and helps to create work.

I like to give an example from a restaurant.

If the chefs had to order supplies, lay tables, serve plates, look after wine, wash up AND do the cooking then the food would be bad and the chefs exhausted. By employing waiters and cleaners, and using food suppliers, the restaurant becomes more productive – it can do more things in the same amount of time, and the chefs are able to spend more of their time cooking and improving their skills. And as the restaurant becomes more productive it will have more need for extra labour. And at the end of the day there is still enough time for them all to go home and relax. Or revise for the exams that will get them their next better-paid job.

As Sam said in his Spectator piece, “Immigrants don’t create unemployment any more than women entering the workforce in the 1950s and ’60s did.”