James Bloodworth says we need stronger unions. Unions, he says, give us higher pay and greater equality.

But what do they do to productivity? The standard Tory line for years has been that militant unions are wreckers. The facts, though, tell a different story.

Academic research based upon comparisons of unionized and non-unionized workplaces has found little impact. One meta-analysis (pdf), for example, concludes that the link between unions and productivity is "near zero": it's slightly negative in the UK but positive in the US.

A cross-country comparison, though tells a different story. My chart plots GDP per worker in 2013 against union density for 33 OECD countries. Across all 33, the correlation is positive, at 0.4. Eyeball econometrics should tell you this finding is robust to outliers.

The US - which has high productivity and low union membership - is an exception. More generally, union membership is associated with higher productivity across countries.

You might wonder how the cross-country evidence shows a positive correlation whereas the cross-workplace evidence generally doesn't. Here's a theory. People will always want better pay and conditions. This is simply because they are human. If they can't achieve these through unions they will try to get them through the ballot box, in the form of legislation.

In this sense, as Philippe Aghion and colleagues show, there can be good and bad equilibria; a good equilibrium in which there are strong unions and litte legislation, and a bad one in which there are weak unions and much regulation.

The UK fits their story. Whereas in the 70s businessmen complained about unions, they now whine about minimum wage laws and red tape.

But here's the thing. Regulation is a bad substitute for unions. Regulation is inflexible. Whereas collective bargaining - when done intelligently - can respond to different local conditions.

This might seem like an odd argument, because lefties tend to want both regulation and unions whereas righties want neither. But insofar as there is a trade-off, strong unions are to be preferred.

Which poses the question: if unions are good for productivity, why have bosses traditionally been opposed to them? The answer, I suspect, lies in this paper, which finds that unionization "is significantly associated with lower levels of total CEO compensation."

Trades unions, then, promote the national interest, whereas hostility to them is founded upon narrow sectional interests.