Via Daniel Elton , a new(ish) paper  (pdf) by economists Thomas Piketty, Emmanuel Saez and Stefanie Stantcheva makes the case for the optimal tax rate of the richest one per cent could be as high as 83 per cent.
The authors look at three possible ways in which top incomes respond to marginal tax rates.
The first is the "supply side" channel cited by Arthur Laffer when he developed his concept of the Laffer curve. Under this channel, marginal tax rates that are too high result in people doing less work than they otherwise would. Lower tax rates thus respond to more economic activity, and greater growth and higher tax revenue. Laffer's work was used by the Reagan administration to justify cutting the top rate in the US, from 70 per cent down to 28 per cent.
The second is the "tax avoidance" channel. This is the model cited by the Chancellor when he made the decision to cut the tax rate for top earners in Britain from 50 per cent to 45 per cent. The argument is that high tax rates increase tax avoidance, but the authors argue – in common with many of the Chancellors critics – that tax avoidance can be dealt with directly, writing that :
A better policy would be to first close loopholes so as to eliminate most tax avoidance opportunities and only then increase top tax rates.
The third is the "compensation bargaining" channel. The authors argue that:
While standard economic models assume that pay reflects productivity, there are strong reasons to be sceptical, especially at the top of the income distribution where the actual economic contribution of managers working in complex organisations is particularly difficult to measure.
As a result, top workers have ample opportunity to set their own pay, through bargaining harder or influencing compensation committees. The incentives for this economically wasteful activity are higher the lower tax rates are. Just as with the supply side model, tax cuts increase the wealth of the richest in society, but unlike that model they do not also increase growth; rather, the extra money for the rich comes from those poorer than them.
The authors, using their estimates of the elasticity of the various channels (that is, the magnitude of the effect), then calculate what the optimum rates would be. Under the first model, they find it to be roughly 57 per cent – 5 per cent higher than the top British tax rate is until the cut takes effect (National Insurance for top earners is 2per cent). Assessing the second model (Osborne's preferred argument), they find that the optimal tax rate would be 62 per cent, but argue that if the correct anti-avoidance and simplification measures were put in place, it would be 71 per cent.
The third model, however, outputs an optimal rate of 83 per cent. The authors still assume a minor supply-side effect, making up 40 per cent of the total elasticity, which is why it isn't higher. That is, they still assume that high tax rates will encourage the rich to work less, but they also assume that they will encourage the rich to put more of their effort into actual work, and less of it into "compensation bargaining".
How to decide between models one and three (the authors assume that once anti-avoidance measures are put in place, model two is the same as model one)? The former predicts that lower tax leads to more work, the latter that it merely leads to higher pay. As a result, tax cuts should be correlated with higher growth. Are they? Well, no:
As a result, the bargaining model can at least be said to be as realistic as the supply-side model; and if that's the case, the optimal top tax rate for most nations is likely to be far in excess of where it is set now.