Options for reforming the Serious Fraud Office are currently being considered by the Attorney-General.

The latest 'failure' by the SFO, in the Brent Walker case, will not make his decision any easier.

It is further proof, on top of many previous 'failures', say the SFO's detractors, that the experiment proposed by the Roskill report in 1986 has simply not worked: it is time for wholesale changes.

But is this right? And if so, what changes should be made?Statistics show that the SFO has been a success: it has a conviction rate of 75%.

The problem with this statistical fact is that the 25% acquittal rate includes nearly all the high profile cases, and it is in relation to those cases that public opinion has been formed.Whatever the rights and wrongs of this reaction on the part of the public, the uncomfortable truth for the SFO is that public perception and confidence is at least as important as any statistical analysis.The SFO was set up as part of a package of measures in the mid to late 1980s aimed at reassuring the business community, both nationally and internationally, that the UK was a safe and honest place in which to conduct business.In addition, in the new era of the share-owning democracy, it was important that members of the public believed that they could join in the capitalist dream with the confidence that they would not fall foul of the unacceptable face of capitalism.The major scandals that have rocked the financial world since 1988 seem to have exposed this new regulation regime as a paper tiger, and public confidence has been further sapped by a number of long and expensive trials which have resulted in acquittals.What has gone wrong? Three comments may be made.

First, no jurisdiction in the world has found a satisfactory method of dealing with allegations of complex fraud within a standard criminal justice system.

The UK is one of the very few countries that has made a concerted effort.

No country, not even the USA, has done better in the very big cases.Secondly, the system brought in by the Criminal Justice Act 1987 is, by and large, sound.

It provides sufficient investigative powers and trial procedures to allow cases to be investigated and tried satisfactorily.

There are no fundamental flaws.

It is the use of the system, by all participants, which has too often failed.The SFO has sometimes chosen the wrong cases, sometimes included too much unnecessary detail, sometimes lacked the courage of its convictions.

Defence lawyers have exploited the weaknesses of the system.

Judges have not always exercised sufficient control over trials.Thirdly, in this discussion we are talking about half a dozen very difficult cases over the last six years, out of several hundred cases involving fraud charges over the same period which have caused no substantial concern.

It is important not to make hasty decisions based on so small a sample.If changes are to be made, what are the options? Frankly they are limited.

Either trials of fraud charges must be maintained within the criminal justice system, recognising that the system was not created to deal with very complex cases; or they should be removed from the system, recognising that a two-tier approach is being adopted towards criminal conduct.One option is to adopt the Roskill proposal to abolish the jury for serious and complex allegations of fraud, and replace it with a fraud charges tribunal.

Roskill found that, as a matter of common sense (he was not allowed to conduct proper research), the average jury would find it difficult to grapple with complex factual and legal arguments spread over many months.There are, of course, substantial difficulties in the way of removing jury trial, but if an efficient way of trying fraud charges is required, with a relatively short trial, the jury must probably go.If, however, it is not possible, or desirable, to grasp that nettle, the time has surely come when research must be undertaken, so that it can be determined whether the way in which cases are presented to juries is sensible, and whether it is likely that juries are returning verdicts, based on a proper analysis of the evidence, which are fair to both prosecution and defence.Another option, which was considered by the working party looking into the future of the SFO, is that it should be absorbed into the Crown Prosecution Service.

Apart from economies of scale, this is unlikely, in itself (bearing in mind that the CPS is not short of criticism) to bring improvements in the handling of complex fraud allegations.

In addition, it would surely send a signal to the financial community that suspected fraud investigation was in some way being down-graded.Some experts believe that a US-style Securities and Exchange Commission, capable of imposing civil penalties, reaching settlements and striking fear into the hearts of wrong-doers, is the answer.

Those who have had any experience of the US legal system would probably not welcome an SEC, but a unified body with jurisdiction over a wide area of corporate regulation, from rule infringement to dishonesty, might have advantages.All these proposals have their attractions, but it is difficult to believe that they would solve more problems than they would create.

It is idle to believe that there is any quick fix solution to such an intractable problem as fraud trials.

Some refinement of the current system may be necessary, better training possibly, better use of resources; but it is time to adopt the good old British custom of leaving well alone, rather than the new custom of being incapable of leaving anything alone for long enough to allow it to mature and develop.