Today's figures showed that unemployment fell slightly in the third quarter, pushing the unemployment rate down to 3.8%, its lowest since 1974. This, however, vastly under-estimates the amount of slack in the labour market.

My chart shows the point. It shows that, in addition to the 1.3m officially unemployed there are also almost 1.9m who are out of the labour force who would like a job. We know this is no idle preference; labour market flows data show that, in Q3, 547,000 people moved from inactivity into employment (and 574,000 in the opposite direction).

But there's much more. Other data today show that, in addition to these, there are 3.3m people who are under-employed - who are working fewer hours than they'd like.

Put together, this adds up to 6.45 million people - 695,000 more that at the low-point for labour market slack seen as far back as 2005.

A big reason why there is so much underemployment is that wages aren't high enough to provide a decent income without long hours; this is also why more people in the same household are working. As the Resolution Foundation says:

Why are so many of us working? Because we're a lot poorer than we expected to be

From this perspective, one fact becomes explicable - that real wages are still lower than they were in 2007*, and that nominal wage growth has slipped back slightly recently, to 3.6% in Q3 after 3.9% in the three months to July.

This is because the under-employed put significant downward pressure on wages. Put yourself in a boss's shoes. If the labour market is really tight, you'll have to raise pay to get people to work more. But if you already have an employee who wants longer hours, you can get them to work more instead. You don't need to raise pay.

Of course, the under-employed are not the only ones unhappy with their hours. ONS data show that there are 3.4 million people who'd like fewer hours even if this means low pay: this suffices to refute the silly idea that you can think of the labour market in terms of a "representative agent".

Such over-employment is, however, also a sign of labour market slack. It tells us that there aren't enough jobs for people to find one with hours that suit them.

If we add together the unemployed, under-employed and over-employed we get almost ten million people who are unhappy with the world of work. That's almost a quarter of the working age population. And this is before we mention those who are over-qualified for the jobs they are doing or who are in what David Graeber calls bullshit jobs.

This is surely a colossal systemic failure. In his book Not Working, David Blanchflower says one solution to this is for macro policy to put the pedal to the metal - to be much more expansionary. But I wonder: is this sufficient?

* OK, depending on how you measure price inflation.