A recent research paper by Sorapop Kiatpongsan and Michael Norton investigates the difference between people’s perception, actual and ideal wage differentials of CEOs and other groups to unskilled workers (hat tip to HBR’s blog and Jia Lyng). It is quite impressive how far reality diverges from people’s perception. Here is one of the main graphs from the paper:
It is very interesting to see how pay ratios of CEOs to unskilled workers differ over the world. According to neoclassical theory, these difference can be explained in different ways:
- CEOs in Australia, Chile, South Korea, the US and other countries are much more productive than elsewhere and therefore the pay ratios reflect marginal productivity. This, however, would imply that Sweden’s and Norway’s managers must be pretty inefficient, even compensated for exchange rates. That is hard to believe, given that Scandinavian countries are relatively rich.
- Those countries which are relatively rich in capital will gain much more from good CEOs – hence their relatively high wages. Again, I find this implausible.
- The difference can be explained by market power of the CEOs in each country, but then the market should take care of it – where CEOs are relatively expensive they will be outsourced, replaced by machines or unskilled labour. For the case of the US that would mean that there are or will be less managers than elsewhere. Again, this is not helpful. That is not how it works.
A political economy explanation would be needed here to explain the power structure that has produced these results. In a 2006 paper by Gomez and Tsioumis, part of the explanation is the non-existence of unions:
We estimate the relation between union presence and executive compensation using a unique panel of executives in publicly listed US firms during the period 1992-2001. We find evidence that union presence is associated with lower levels of total executive compensation. We find this union effect to be primarily the result of substantially lower stock option awards, and to a lesser extent due to lower cash pay. Moreover, the negative relation between unionization and executive remuneration becomes larger at the higher end of the conditional distribution of executive remuneration.
Correlation is not causation, but it seems quite straightforward which way causality runs in this case. Neoclassical theory ignores institutions and focuses on markets exclusively. In some cases this makes sense, but here I would argue that the problem is mostly institutional.