There's one point raised by sterling's fall this year which deserves more attention than it's had - it's that sometimes (often!) quite powerful correlations can break down.

My chart shows what I mean. It plots the share of profits in GDP against the inverse of sterling's trade-weighted index. You can see that from the mid-70s to the mid-90s, there was a close link between the two. A weak pound raised the profit share, and a strong pound cut it. There are good reasons why this should be:

- a fall in the pound helps insulate domestic firms from foreign competition, thus raising their monopoly power and hence allowing them to mark up prices more.

- higher import costs are seen as a fair (pdf) reason for a firm to raise prices.

- exporters who price to market must cut sterling prices of exports when sterling rises, thus suffering a margin squeeze, but can raise margins when sterling falls.

However, this relationship broke down in the late 90s. Sterling's rise then did not squeeze profits as much as you'd expect from past behaviour.

This might be partly because the supply of labour from emerging markets shifted the balance of class power in favour of capital then. But this is unlikely to be the whole story, because sterling's slump in 2008 did not much raise profit margins. Instead, I suspect what's going on is a change in the nature of international trade. A lot of trade now occurs within (pdf) firms, as multinationals (pdf) ship in components from their overseas branches, sometimes at shadow rather than market prices.This makes prices, and hence income shares, less responsive to exchange rates.

The precise reasons for the change, however, aren't really my point. Instead, I merely want to point out that even strong empirical relationships with sound theoretical support can sometimes break down.This is why the study of economics must be a study of mechanisms - which are often local and temporary - rather than of apparent statistical regularities.

* Strictly speaking, I should be using the real rather than nominal exchange rate, but I doubt doing so would make much difference.