The Obama administration is expected today to propose a reorganisation of the way we regulate financial markets. I am not an advocate of too much regulation. Having gone too far in deregulating – which contributed to the current crisis – we must resist the temptation to go too far in the opposite direction. While markets are imperfect, regulators are even more so. Not only are they human, they are also bureaucratic and subject to political influences, therefore regulations should be kept to a minimum.

Three principles should guide reform. First, since markets are bubble-prone, regulators must accept responsibility for preventing bubbles from growing too big. Alan Greenspan, the former chairman of the Federal Reserve, and others have expressly refused that responsibility. If markets cannot recognise bubbles, they argued, neither can regulators. They were right and yet the authorities must accept the assignment, even knowing that they are bound to be wrong. They will, however, have the benefit of feedback from the markets so they can and must continually re-calibrate to correct their mistakes.

Second, to control asset bubbles it is not enough to control the money supply; we must also control the availability of credit. This cannot be done with monetary tools alone – we must also use credit controls such as margin requirements and minimum capital requirements. Currently these tend to be fixed irrespective of the market’s mood. Part of the authorities’ job is to counteract these moods. Margin and minimum capital requirements should be adjusted to suit market conditions. Regulators should vary the loan-to-value ratio on commercial and residential mortgages for risk-weighting purposes to forestall real estate bubbles.

Third, we must reconceptualise the meaning of market risk. The efficient market hypothesis postulates that markets tend towards equilibrium and deviations occur in a random fashion; moreover, markets are supposed to function without any discontinuity in the sequence of prices. Under these conditions market risks can be equated with the risks affecting individual market participants. As long as they manage their risks properly, regulators ought to be happy.

But the efficient market hypothesis is unrealistic. Markets are subject to imbalances that individual participants may ignore if they think they can liquidate their positions. Regulators cannot ignore these imbalances. If too many participants are on the same side, positions cannot be liquidated without causing a discontinuity or, worse, a collapse. In that case the authorities may have to come to the rescue. That means that there is systemic risk in the market in addition to the risks most market participants perceived prior to the crisis.

The securitisation of mortgages added a new dimension of systemic risk. Financial engineers claimed they were reducing risks through geographic diversification: in fact they were increasing them by creating an agency problem. The agents were more interested in maximising fee income than in protecting the interests of bondholders. That is the verity that was ignored by regulators and market participants alike.

To avert a repetition, the agents must have “skin in the game” but the five per cent proposed by the administration is more symbolic than substantive. I would consider ten per cent as the minimum requirement. To allow for possible discontinuities in markets securities held by banks should carry a higher risk rating than they do under the Basel Accords. Banks should pay for the implicit guarantee they enjoy by using less leverage and accepting restrictions on how they invest depositors’ money; they should not be allowed to speculate for their own account with other people’s money.

It is probably impractical to separate investment banking from commercial banking as the US did with the Glass Steagull Act of 1933. But there has to be an internal firewall that separates proprietary trading from commercial banking. Proprietary trading ought to be financed out of a bank’s own capital. If a bank is too big to fail, regulators must go even further to protect its capital from undue risk. They must regulate the compensation packages of proprietary traders so that risks and rewards are properly aligned. This may push proprietary trading out of banks into hedge funds. That is where it properly belongs. Hedge funds and other large investors must also be closely monitored to ensure that they do not build up dangerous imbalances.

Finally, I have strong views on the regulation of derivatives. The prevailing opinion is that they ought to be traded on regulated exchanges. That is not enough. The issuance and trading of derivatives ought to be as strictly regulated as stocks. Regulators ought to insist that derivatives be homogenous, standardised and transparent.

Custom made derivatives only serve to improve the profit margin of the financial engineers designing them. In fact, some derivatives ought not to be traded at all. I have in mind credit default swaps. Consider the recent bankruptcy of AbitibiBowater and thatof General Motors. In both cases, some bondholders owned CDS and stood to gain more by bankruptcy than by reorganisation. It is like buying life insurance on someone else’s life and owning a licence to kill him. CDS are instruments of destruction that ought to be outlawed.

The writer is chair of Soros Fund Management and author of ‘The Crash of 2008’.