Yesterday at 11pm the UK left the European Union.

In the period leading up to the referendum and in the three years of political crisis that followed it there were repeated warnings that Brexit would be catastrophic for employment. Britain Stronger in Europe claimed that :

If we leave the EU experts predict that the economic hit would mean up to 950,000 UK jobs could be lost (Source: Confederation of British Industry), meaning less security for you and your family.

But the closer Brexit came, the more optimistic was the actual news about employment. In February 2019 the Independent, a pro remain paper asked:

How is UK unemployment so low despite Brexit turmoil?

In March 2019 a Guardian headline read:

UK unemployment falls to 44-year low despite Brexit fears

But should this have been any surprise?

Have the 47 years of EU membership actually been good for employment?

Was unemployment unusually low when we were in the EU?

The Bank of England has recently published a very interesting long term database about the British economy entitle A Millennium of Economic Data. Using this we can compare  the EU period with other periods in the country’s history. In particular we can look at unemployment.

I suspected before I looked at the figures that the EU membership period would compare badly with the two decades that preceded it. What came as a surprise to me was just how bad it was in long term perspective of British history.

The EU period not only compared badly with the period 1945 to 1973, it compared badly with every period except 1919 to 1939.

unempovertime

The graph above shows in the red line the decade moving average of unemployment. The dotted blue line shows the fraction of the previous decade that the country has been in the EU. (I have counted the individual years since the 2016 referendum as 50% membership since the knowledge that we were leaving had immediate effects on migration from the EU.)

It is clear that the EU period corresponded to a historically high rate of unemployment.

There are 3 periods of elevated unemployment in the time series.

  1. The first is the late 19th century long depression which started around 1870 and lasted for the rest of the Victorian period.
  2. Next is the inter war recession which, conventional economic historians tell us, was produced by the breakup of the pre 1914 free trade system and the growth of protectionism world wide.
  3. Finally come the great EU recession.

Of these the first was the least severe, and the EU the most severe. The interwar downturn was severe, but shorter than the other two.

There are clear drops in unemployment corresponding to the two world wars. It is universally acknowledged that war reduced unemployment.

For completeness I have plotted, in addition, the fraction of the decade moving average that the country was at war, including the Crimean, Boer, Korean and Gulf wars.

The correlations are fairly evident to the naked eye. We can see that EU was correlated with high unemployment, the wars with lower unemployment. When I computed it numerically I get the following:

correlations annual basis 10 year basis
Unemployment to EU membership 0.37 0.47
Unemployment to war -0.38 -0.43

As we expect unemployment was negatively correlated with war. But the correlation between unemployment and the EU was just as strong, or stronger in the opposite direction. The correlations on a decade basis are better because doing it on a 10 year basis eliminates the noise of the business cycle whose period is usually 7 to 10 years.

We can show this correlation between low unemployment and war graphically thus:

unempversuswar

We plot the fraction of the decade spent at war against the average unemployment during the decade. The negative correlation is visible in the clear downward trendline.

Because the figures are decade averages, the crosses bunch along particular vertical lines, but the trend calculation takes this into account.

Let us repeat this type of diagram with the EU figures:

unempversuseu

In this case an upward trend is now evident.

The experience of older working class voters in England who could compare conditions before and after joining the EU led them to believe that something had gone very wrong. The long term figures from the Bank of England tend to back this up.

Of course correlation is not causation. It could just be a matter of chance that wars coincided with low unemployment and EU membership coincided with high unemployment. But purely on the evidence of the historical data, the onus is on the defenders of the EU to show that there was no causation.  On the basis of the statistics, it is no more reasonable to deny that the EU brought recession than it is to deny that wars ended recessions.

Nor, on the basis of the historical evidence, is there any grounds to fear that leaving the EU will usher in a period of high unemployment.