This is a bit less ambitious than my plan to reform capitalism, but the principles are the same.

Too big to fail is too big to exist

Big banks create too much systemic risk, and are far too much of a strain on government finances when they need to be bailed out.

The British government is going the right way by breaking up big-banks that have been part-nationalised. Why not go further and prevent any bank of having too much market share in the first place? No bank would be allowed to take more than a set percentage of deposits within a country.

Given that most big banks have are formed through mergers it would not even need much legislative change to block this: a slight tightening of competition law or an extension of the bank regulator’s power would be enough. It would be better to break up all the existing big banks as well.

The main economies of scale that would be lost would be the need to maintain the over-lapping branch networks, but, given outcry that often greets the closure of bank branches, there are obviously many people who think there is a public benefit to maintaining the networks.

Let banks go bust, then nationalise the assets and deposits

The way in which the British authorities deal with Northern Rock left the shareholders with nothing. This is the best approach to bailouts as it avoids creating moral hazard.Shareholders will know what they will suffer as much of the consequences of bank failure as possible and will have a greater to restrain management who are taking excessive risks.

This should be accompanied by better corporate governance so that the management cannot take risk regardless of the shareholders. This requires the break up of big banks as the economic disruption caused by the failure of a big bank is not acceptable: RBS and Lloyds could not have been dealt with in the same as Northern Rock.

Force banks to issue more bonds

It looks likely that the crisis is likely to lead to higher capital adequacy requirements, so banks will need more equity funding. Banks should also be required to issue more bonds, and it should be made absolutely clear that bond-holders will not benefit from any government bailouts.

The market price of the bonds will adjust to reflect the level of risk of default, providing a market driven measure of risk. A high yield on the bonds would be a clear signal that the market expects problems.

If the banks are required to fix maturity dates so that a fair amount of debt needs to be raised each year to replaced year year, excessive risk taking will be quickly translated into high interest paid, penalising it fairly quickly.

It will also provide another layer of capital to pay depositors out of. The Cocos that Lloyds is issuing also do this, and their prices will also measure the market’s view of risk: but they will not penalise banks that the market sees as risk taking. They are also more complex and therefore harder to value.

The crisis has undermined faith in the market, given that CDOs were clearly over-priced — but market mechanisms usually work, and simple bonds are far easier to analyse than the complex securities than caused the problems.

Separate investment banks from commercial banks

This ought to be a no-brainer, but banks lobby furiously every time the idea is raised. It insulates the big deposit takers from the risks of investment banking. There is not real reason to allow the two types of bank to be combined: its main advantage is that it allows investment banks to fund their risky activities with the cheap money available to a commercial bank.

The point of the separation is to prevent the government underwriting risky activities, the profits of which go to shareholders (and in bonuses!). This amounts to a governments subsidising investment banking.

What about bonuses?

Politicians are very fond of talking about taking action on the huge bonuses paid to bankers. They are not that effective at actually doing anything about it. The problem with direct government intervention is that is not obvious what the “correct” pay for bankers (or any one else) actually is.

I regard the high remuneration of bankers as a symptom, not a cause. Lack of control by shareholders, implicit government guarantees that under-write high risk businesses, and too many (profitable to investment banks) large takeovers and mergers (most of which ought to be blocked because they undermine competition) are the root causes.

Rather than capping bankers pay, we need to ask what has changed. Banking was not so spectacularly well paid a few decades ago and if we look at the reasons for the changes we can make a more sensible assessment of which are bad.

It seems to be to be inevitable that in a capitalist society dealing with money will be well paid: it will always be worth paying people well to manage large amounts of money just a little bit more efficiently. If you think that the pay is excessive, find where the market failure is and directly address that.

Keep it simple, stupid regulator

One of the big mistakes that contributed to the crisis was a regulatory change, and, as before the regulators are proposing changes that could make things worse.

The Basel 2 accord was intended to provide a more accurate measure of the risks that banks had taken. One of the problems with its predecessor was that there were ways in which banks could manipulate it: for example by doing more lending at the riskier end of each “risk bucket”. Part of the solution to the banks manipulating the system was to allow them to use their own internal risk models for regulatory purposes, giving them a lot more discretion. Could no one really see a problem with that at the time?

Basel 2 also relied on credit ratings: those ratings issued by agencies that were paid by the people they were rating? I do not know why that ever sounded like a good idea (although regulators were not alone in relying on it, almost everyone did). I was once offered a job at a rating agency, and turned it down partly because I had doubts about whether they way they worked was sustainable. I feel almost as smug about that as at getting rejected by Enron.

Regulators are still not being very sensible about capital adequacy. One popular proposal is to weaken mark to market rules, so that banks balance sheets will not all weaken at once when markets fall. In other words, we will all feel more confident because we will not know how how much the banks had lost. Is this going to encourage bankers to be more cautious next time round? We need more transparency not less.

Keep it simple (no complex risk models), use the market (which will only work if you avoid moral hazard), and rely on caution rather than perfect measurements.