Last month that the UK government is planning to follow the US and reveal how much students on each course at each university stand to earn.

At the same time, university leaders in England are finding their voices about the need to increase tuition fees. The current ?9,000 a year limit for undergraduate degree study is unsustainable for much longer, they tell us, and there are plans to permit institutions with good scores in the proposed teaching excellence framework to raise fees ¨C at least in line with inflation.

But the TEF may turn out to be a very bureaucratic exercise for little financial gain. A much better idea would be to make use of graduate earnings data ¨C from HM Revenue and Customs and other sources ¨C to institute fee caps that differ by subject and institution, and that increase (or decrease) as the evidence suggests graduate economic returns increase (or decrease).

The 2012-13 fees settlement was good for English universities: they increased their net income without incurring any of the associated risk. All that risk falls on graduates and taxpayers. And while there is nothing intrinsically wrong with graduates paying for a university education from which they benefit, that still leaves the question of what a fair fees system should look like.

ÖйúAƬ

As the expected economic return of a degree varies by subject and institution, it strikes me as fair that the contributions made by students should vary on the same basis. Such a fees system would be transparent, and this is important. Australia has differential fees for different subject bands, but the bands are decided on the basis of an opaque mixture of cost, economic return and political considerations. The banding lacks any clear rationale and has prompted widespread criticism.

Fee caps that reflect the relative economic returns of different course choices would help students make more informed decisions. We should not underestimate the problem here. Most students make their decisions before they know their A-level results; many have little understanding of the differences between universities, and careers advice is often poor.

ÖйúAƬ

Differential fees would also benefit taxpayers. The annual fee for degrees with low earning potential would drop from the current ?9,000. And fee increases above ?9,000 would be aligned with students¡¯ ability to repay, so the level of unpaid student debt would decrease.

I propose fee levels that would make students liable to repay roughly 30 per cent of the net graduate premium for each course (annual fees would take into account the fact that some courses, such as medicine, last longer than three years). In my view, it is not unfair to charge graduates less than one-third of the economic benefits they can expect from ÖйúAƬ. But there is no uniquely correct figure; the key point is that as the percentage gets nearer to 100, university becomes an increasingly bad economic bet.

Differential fees could also benefit universities. If economic returns increased, fee caps would increase too, without a need for a vote in Parliament. Fees would become depoliticised. There would, of course, be a risk associated with the likely decrease in some fees caps, but we should not overstate it. The fees for some high-premium courses, such as medicine and engineering, could increase substantially from their current levels. And universities could reduce the risk to which they are exposed by improving their graduates¡¯ employability. This would allow them to increase their fee caps.

But shouldn¡¯t teaching costs also be factored into the fee-setting equation? Only, in my view, exceptionally. After all, they are not unalterable facts of nature: they can be reduced. And if universities charge students on the basis of their costs, they have no incentive to become more efficient. Furthermore, disparities between teaching costs and economic returns are the exception. Many subjects with lower economic returns, such as English or philosophy, are cheap to teach. But there is a case for public subsidy of subjects where private returns are low but public benefits high. One example might be the creative arts.

ÖйúAƬ

Finally, while we might expect differential fees to lead to a decline in some subjects at some universities, the Western canon will survive. Some students might turn away from philosophy because it is a bad economic bet, but this is not obviously bad ¨C either for them or the subject. Philosophy will still be taught.

I suspect my proposal will be met with scepticism. Partly this will be based on universities¡¯ institutional conservatism and fear of short-term upheaval. But they should be careful. Arguments against differential fees might apply with equal force against all arguments to increase fee caps. And while universities don¡¯t want to expose themselves to financial risk, it is inevitable that the day will come when they have to. It is simply politically unsustainable that taxpayers and graduates should face all the risk of tuition fees while the university sector faces none.

Dean Machin was a teaching fellow in philosophy at University College London between 2012 and 2015, and is chair and joint head of UCL¡¯s policy commission into transport and ethics.