The MSM reports that yesterday's figures show that unemployment fell to just over 2.5 million. This greatly understates the true level of joblessness.

The figures also show that there are 2.32 million who are economically inactive but who would like a job. Adding these to the official unemployment count gives us a total of over 4.8 million who are not working but would like to. That's 12% of the working age population.

One common objection to this calculation is that the inactive's claim that they would like to work is just an idle preference.Such people haven't actively sought work in the last four weeks and/or are unavailable to start in the next two; if they were, they'd be measured as unemployed.

However, other figures yesterday - on labour market flows - show that this objection is false. They show that, in Q4, 423,000 "inactive" people became employed. That's equivalent to 18.1% of all the inactive who'd like a job. By comparison, 595,000 moved from unemployment to work - 23.7% of all the unemployed.What's more, of the 871,000 who lost their jobs in Q4, more moved into inactivity than unemployment (453,000 vs 418,000).

These numbers tell us that, for practical purposes, the distinction between "unemployed" and "inactive" is blurred. In practice, there's not much difference between being "inactive" and being unemployed.

You might wonder how it can be that so many people can find work if they are not actively seeking it. I suspect much hinges on the vagueness of "actively". Consider two people. One distributes his CVs to firms speculatively. The other asks family and friends to listen out for vacancies. You might think the former is "actively seeking" work whilst the latter isn't. But given the importance of social networks for getting jobs, the latter could have as good a chance of finding work as the former.

My point here is not just that the official figures greatly understate employment. There's worse, shown in my chart.It plots the unemployment plus inactivity rate since 1993 (when data began) against real wages four quarters later*.It's a variant of the wage curve (pdf).

This shows that, between 1993Q2 and 2007Q4, there was a humungous negative correlation between the two - of minus 0.94. A greater excess supply of labour led to lower real wages, and less excess supply to higher ones. That's Econ 101.

However, since the recession began, the wage curve seems to have shifted rightwards. The excess supply of labour has not depressed real wages as much as you'd expect. In fact, if the 1993-2007 relationship had continued to hold, real wages would now be almost 20% lower than they actually are.

In this sense, the puzzle is not: why have real wages been squeezed? It's: why have they not fallen much more?

And herein lies a danger. If the wage curve does return to its 1993-2007 pattern, real wages could fall a lot. I don't know how likely this is. But you could interpret the fact that real wages have fallen recently at the same time as the jobless rate has also fallen as a sign that this risk might be materializing.

Perhaps, then, the downward pressures on real wages are much greater than generally thought.

* I define real wages as total compensation of employees, divided by the number of employees, deflated by the CPI.