A while ago I wrote that while unions could increase the wages of workers in the short run, the wage growth we have seen over the last two centuries and will probably continue to see for the foreseeable future, was due entirely long run productivity growth. Compared with long run improvements in technology and education, unions had only a very small impact on the economic welfare of people.

Some data over at Super Economy suggests maybe I was still being too generous:

I have plotted the share of workers that are covered by collective bargaining agreements. This is better than the share of workers that are members of unions. In some countries, such as France, few workers are actually members of unions, but unions have the power to determine contracts for workers that are not members. There are huge differences across countries. In Sweden and France 90% of workers are covered by collective union agreements. In the U.S and Japan only 15% of workers have their wages determined by unions.

I also plot the share of factor costs that goes to labor (as wages and compensations). What you will notice is that this does not vary across countries.

The way the economy gets divided between capitalists and workers is virtually identical in weak union countries such as the U.S as it is in countries with powerful unions, such as France and Sweden.

If anything, American workers get a bigger share of the cake compared to European workers, that have to some extent been replaced by machines.

Italian workers get slightly less than other countries. The reason I believe is that they have a high share of self-employed, that do not fit nice in the capital-labor divide.

So are unions useless? Not entirely. Unions can raise the wage of some workers, but not at the expense of capital, but at the expense of consumers (other workers), or perhaps single “rent” earning industries (such as mining). This is good and well for those few workers, but does not work as a large scale program, since workers are just transferring resources from other workers.

Second, Unions seem to be better able to extract rent within the class of workers, from high skilled to low skilled workers, than they can do with capital, which is responsive. Whereas I.Q. and schooling are less elastic in supply, if capital earnings go down, investments in capital goes does, or perhaps re-located to other countries.

This simple graph can tell us a lot about the world, and is a powerful argument against the world-view of the left.

Because it was hard to tell how much of the union wage premium was coming at the expense of  other workers compared to capital owners and I was trying to be as generous to unions as possible, my analysis assumed that the union wage premium came at the expense of no one. Judging by the figure above, any union wage premium in developed countries must be coming entirely at the expense of other workers. The most unions can then do is redistribute income from high to low income workers within a country. In Australia the wage premium for high income workers was 6% and for low income workers was 12%. Continuing to assume unions induce no economic inefficiency at all (unlikely), their impact then at most be to increase by 6% the incomes of low income earners. If high and low income workers each took home half of wages to start with, their true increase in buying potential for low income earners would actually be 2.7%. In countries where unions only cover some workers, they won’t even be achieving that – because much of the wage increase will be coming expense of other poor non-unioned workers.

This greatly reduces our assessment of the relative importance of unions relative to variations in productivity growth. It also deals with the objection that unions are necessary for workers to be paid the marginal product of their labour and so for wages to track productivity growth; were this true, we would expect countries with strong unions to have a much higher share of wage income, which they don’t.

Added: Is there even a union wage premium in most countries?