Skip to content

Paper presented to the Australian Corporate Lawyers Association, 2010 Government Law Conference, Canberra
20 August 2010

The Topic

I did not choose the title for this speech, but I do know that it was chosen long before tomorrow's main event (the federal election) was announced.

So while others are "moving forward", I am constrained to "looking forward".

The virtue of prediction is that it looks to a point ahead, seeking to identify the result of movement, while the course of movement itself may be less certain.

Not only have I been asked to look forward, but I know exactly when I am looking forward to. It is 2020.

A few things are clear about 2020:

  1. It will be six years after the Constitution deemed me incompetent;

  2. Four federal elections will intervene;

  3. Efficiency dividends and cuts in public service funding will continue to stifle beneficial initiatives;

  4. The computer and its constant companion, the internet, will have moved a very long way forward - at an ever increasing pace; and

  5. Facebook and Twitter will be quaint old fashioned ideas. People will wonder what "social networking" was. The new thing will be "virtual co-relationing" or VCR for short. That acronym will be possible because no one will have heard of a video cassette recorder. Blu-ray will be a distant memory.

Well what about administrative decision-making in 2020. It is time I turned my 2020 vision on that.

Undoubtedly administrative decision-making in 2020 will be different in some respects to what it is now. What is of critical importance is that change should lead to improvement in the three essential requirements of administrative decision-making: quality, cost and speed.

The change we can predict with the greatest certainty is the increased significance of computer based decision-making. This will almost certainly increase speed. It may reduce cost. It may also reduce quality.

So what is quality? Why might the computer threaten it?

To look at this it is essential first to identify what are the characteristics of a good administrative decision.

This takes me back to our politicians "moving forward". We have been exposed to a lot of that in the last few weeks.

The scenario goes something like this:

Kerry: Do you agree that the other Party's proposal for free pizza seems to be very popular?

Interviewee: I am glad you asked that, Kerry. What I can say is that our proposal for cheap clothes pegs is what you should be concentrating on. Our proposal will provide a great comfort for the families of Australia and cost very little.

When I was privileged to practice at the bar I would then say: "Thank you for that. Now would you answer my question which was..."

Our language has even adapted to the skill of politicians in not answering questions. We say they are "singing from the song sheet" or even "on song".

But I do not think we should be too critical of the politicians. In reality, they are only doing what we all do. I am sorry to have to tell you that when you talk to me I probably will not answer your question. I may answer the question I know you should have asked. More likely, I will make a speech on a related, but different, topic I want to talk about. But I will not answer your question.

Answer the Question

So what has all this got to do with administrative decision-making. Well, a lot really. Administrative decision-making is basically question answering. The statute or regulation or policy poses the question. The decision-maker must give the right answer. We must avoid the natural tendency, encouraged by our social communications, not to answer the question precisely.

Giving the right answer to the wrong question is giving the wrong answer. I put identifying the right question at the head of my list of the qualities of good decision-making. Once the right question is identified it will, as they say, often answer itself.

What sort of decision-making am I talking about? Well, all decision-making. I suppose the emphasis is on formal primary decision-making or internal review. However, the principles also apply to ordinary decision-making. They apply to decisions whether to sue or defend and even to steps in litigation. The authority to make all these decisions will have a source and the source will usually contain regulation. The model litigant rules in the Commonwealth are an example. I think that it is right, every time a decision is made, to think about its source and apply the rules I will discuss.

I have so far isolated one problem with primary decision-making - not answering the right question. However, I would not wish you to think that I exclude higher level decision-makers, such as judges and tribunal members, from this error.

Justice Branson in the Federal Court said this concerning a decision of the Administrative Appeals Tribunal (Australian Postal Corporation v Barry (2006) 44 AAR 186 at 190; [2006] FCA 1751 at [25]):

"I observe incidentally that it is a salutary discipline for every statutory decision-maker to refer to the terms of the relevant statutory provisions and to identify each element of the statutory cause of action. Had the Tribunal in this case set out or paraphrased in its reasons for decision the terms of s 16 and s 19 [of the Safety, Rehabilitation and Compensation Act 1988 (Cth)] it is unlikely that it would have overlooked their critical elements."

The High Court recently said something similar in Shi v Migration Agents Registration Authority [2008] HCA 31; 235 CLR 286:

"As this Court has so often emphasised in recent years, [they cited 6 cases] questions presented by the application of legislation can be answered only by first giving close attention to the relevant provisions. Reference to decided cases or other secondary material must not be permitted to distract attention from the language of the applicant statute or statutes. Expressions used in decided cases to explain the operation of commonly encountered statutory provisions and their application to the facts and circumstances of a particular case may serve only to mask the nature of the task that is presented when those provisions must be applied in another case. The masking effect occurs because attention is focused upon the expression used in the decided cases, not upon the relevant statutory provisions."

Kirby J made similar observations in Shi at [25]. These passages reflect a statement made a little earlier by Callinan and Heydon JJ in McKinnon v Secretary, Department of Treasury [2006] HCA 45: (2006) 228 CLR 423 at 468 [131], when speaking of the application of freedom of information provisions. See also Australian Securities and Investments Commission v PTLZ [2008] FCAFC 164; 48 AAR 559, at 566 [34].

Administrative decision-making is nearly always authorised by legislation and must address the question posed by the legislation.

A great deal of our everyday activity is governed by perception. Contracting parties often leave their contracts in a drawer and perform the contract in accordance with their perception of its provisions. However, when a dispute arises, it is the actual provisions of the contract which will govern and to which the parties then turn. It is never appropriate for administrative decision-making to be guided by perception. The precise wording of the rule being applied must always govern.

The principle may seem obvious. However, it is often departed from. I venture to suggest that the majority of decisions of the Administrative Appeals Tribunal which are upset on appeal to the Federal Court are set aside on the ground that the AAT member did not correctly apply the governing legislation. Frequently, this is because the legislation is difficult to understand. Sometimes, however, the failure is associated with an omission to strictly observe the requirements of the legislation and to follow through the cumulative tests it contains. This can be a complex and even a tedious process, but with the application of appropriate care it is not a difficult process.

My experience is that although the proposition is obvious, it is remarkable how often its requirements are not carefully followed. Perhaps it is associated with the significance of perception - in this case misconception - of what the statute actually provides. I suspect it is also associated with the fact that familiarity can cause us to act automatically. It is very easy for a decision-maker, particularly one who makes many decisions concerning the same or similar subject-matter, to fall into the habit of assuming what the legislation requires; of proceeding on a perception of what the legislation requires. The decision of Branson J is an excellent example. Another possibility is that the domination of the facts in most cases somehow masks the importance of the rule being applied.

So, identify the question carefully and answer in precisely.

Establish the Facts

The second most important aspect of decision-making is establishing the facts. It is also often said that once the facts are established the decision will make itself.

Fact finding is especially difficult for primary decision-makers and internal reviewers because they often have no outside assistance in carrying out the task. The Administrative Appeals Tribunal has the advantage of being able to consider facts presented by both sides in an ordered way. Primary decision-makers may have no such assistance. Where the parties do present arguments, they will not usually be so well ordered and will often confuse facts and submissions. Most problematically, they will generally assert facts without offering proof.

Administrative decision-makers at every level are usually freed from the technical requirements of the rules of evidence. However, this does not mean that facts can be established merely by assertion.

Although the rules of evidence do not apply, the rules of natural justice or procedural fairness generally will apply to most administrative decision-making. In the past, these were technical rules, but since the decision of the High Court of Australia in Kioa v West (1985) 159 CLR 550; 62 ALR 321, the emphasis is no longer on technical rules but on what is fair in all the circumstances (CLR at 584-85; ALR at 346-47).

Fairness to persons affected by a decision will require the decision to be based on established facts rather than asserted facts. It will often require one party to be given the opportunity to contradict a fact which seems prima facie to have been established or to make submissions as to its relevance or significance.

Much administrative decision-making is bilateral. There are two interests. One of them is the Government interest. This is true of income tax assessments and their counterpart, social security payments. However, it is not true of child support payments. Both parents have an interest. It is not true of fisheries regulation, particularly in recent times. This is because a finite and, in some cases, contracting resource, is being managed. Every increase in quota for one fisherman is a decrease in quota for another. The same is partly true of migration decision-making. It is certainly true of decision-making concerned with environmental protection. In coming years decisions relating to allocation of water for irrigation will frequently be important areas where multiple interests are involved.

If the rules of evidence do not apply, what should guide administrative decision-makers in fact finding? The test is what is probative. What leads to reasonable satisfaction. Very often a document will provide the proof: a certificate issued under legislation, such as a birth certificate, or a notice issued under legislation, such as a rate notice.

Where evidentiary documents are not available, assertion may be sufficient where the person making the assertion knows the facts. Neither courts nor administrative decision-makers proceed as if witnesses are not telling the truth, unless there is good reason to doubt them. Sometimes an assertion might need to be verified, by its being made in a statutory declaration, before it will be accepted.

What kind of proof is required depends in every case upon the inherent likelihood or unlikelihood of the matter asserted and its importance to the decision. For example, the amount of water used by one irrigator may be an important primary fact. If an applicant's claim is challenged by someone with an opposing interest, proof of a higher order may be required. An assertion on oath may be enough. But it may not. More objective evidence may be required. The person with the opposing interest should usually be given an opportunity to challenge the claim.

The balancing of these processes will lead to the adoption of a method which will, in the end, enable the decision-maker to come to a satisfactory finding as to the facts and from that basis to make the ultimate decision.

The Role of Policy

Policy can be a very difficult area. Reviewing tribunals such as the Administrative Appeals Tribunal are not bound by policy, but they will be reluctant to depart from policy without good reason. This has been the position ever since the landmark decision in Drake v Minister for Immigration and Ethnic Affairs (1979) 24 ALR 577; 2 ALD 60.

Policy cannot alter a legislative power or the manner or basis for its exercise, but it can explain or describe the way, in general, it should be exercised. Subject to the administrative law requirement, now enshrined in s 6(2)(f) of the Administrative Decisions (Judicial Review) Act 1977 (Cth), and state equivalents, that decisions may not be taken "in accordance with a rule or policy without regard to the merits of the particular case", administrative decision-makers are entitled to take policy into account.

The complexity of modern Government administration has led to more and more policy documents which impact on decision-making. This is particularly so in the area of Government regulation of business. Policy has become so important that it is sometimes called "soft law". The contrast is with legislation or delegated legislation called "hard law".

Policy and discretion meet at the interface between uniformity and individual justice. A balance needs to be found between the two. In decision-making on review, greater currency may be given to individual justice but that is not to say that policy is the only consideration at the primary level. Indeed, such an approach will often lead to error of law.

In 1989, speaking twelve years after the Administrative Appeals Tribunal was established, Sir Anthony Mason drew five distinctions between primary administrative decision-making and decision-making on review in the Administrative Appeals Tribunal. He said this (A. Mason, Administrative Review: The Experience of the First Twelve Years (1989) 18 Fed L Rev 122 at 130):

"Experience indicates that administrative decision-making falls short of the judicial model - on which the AAT is based - in five significance respects. First, it lacks the independence of the judicial process. The administrative decision-maker is, and is thought to be, more susceptible to political, ministerial and bureaucratic influence than is a judge. Secondly, some administrative decisions are made out in the open; most are not. Thirdly, apart from statute, the administrator does not have to give reasons for his decision. Fourthly, the administrator does not always observe the standards of natural justice or procedural fairness. That is not surprising; he is not trained to do so. Finally, he is inclined to subordinate the claims of justice of the individual to the more general demands of public policy and sometimes to adventitious political and bureaucratic pressures.

The five features of administrative decision-making which I have mentioned reveal why it is that administrative decision-making has never achieved the level of acceptance of the judicial process in the mind of the public."

The five qualities are:

  1. Independence of the Tribunal;

  2. Decision-making in public;

  3. Requirement for reasons;

  4. Natural justice applies; and

  5. Individual justice will not be subordinated to public policy.

At a practical level one might add that the judicial model leads to a more thorough and detailed examination of the facts and a more rigorous consideration of the possible outcomes. I have been involved in hearings which took days where the time that was devoted by the original decision-maker must have been hours or less.

The intervening years since Sir Anthony made his remarks have, I think, brought about change. The Administrative Appeals Tribunal and its state counterparts have had a significant effect on decision-making at the primary level. Review of decisions in both the Federal Court and High Court has added to the effect. This seems to me particularly to be so with respect to the fourth and fifth matters to which Sir Anthony referred. Although natural justice issues and public policy issues may be given more thorough consideration in the Administrative Appeals Tribunal, they will nevertheless now require examination at the primary decision-making level.

In addition, reasons are becoming the rule; transparency is more and more a requirement; independence is required even of departmental decision-makers. These developments are as much due to statutory and government changes as they are to the influences of courts and tribunals. The spread of freedom of information requirements has been extended by recent legislation: Freedom of Information (Removal of Conclusive Certificates and Other Measures) Act 2009 (Cth); Australian Information Commissioner Act 2010 (Cth); Freedom of Information Amendment (Reform) Act 2010 (Cth). Reasons are now habitually required. The Department of Immigration is introducing systems of internal review which are almost tribunal-like.

All these changes themselves point to changes that will occur prior to 2020. Reactionary developments are unlikely. The steady progress towards more transparent fairness will only continue.

These changes are part of the revolution that has occurred in administrative law and administrative decision-making in consequence of the sweeping reforms introduced in Australia in the mid-1970's.

The most important advice I can offer to primary decision-makers is to exercise their functions through an awareness of these matters: to know that policy and individual justice can conflict; to know that sometimes one or other will militate a result; and to know that between the two will sometimes lie an area for discretion.

As long ago as 1979 the Federal Court of Australia established that the function of the Administrative Appeals Tribunal was to arrive at the "correct or preferable" decision (Drake ALR at 589). That description is equally apt to describe the task of all administrative decision-makers, subject to the subtle differences relating to the application of policy to which I have referred.

Primary decision-makers should therefore be looking for the "correct" decision, where only one result can obtain and the "preferable" decision where a range of possible decisions is available. Policy will play a part but must not be the only consideration and should be considered carefully where its application may deny individual justice.

So my tests of the quality of decision-making are:

  1. Identifying the question correctly.

  2. Finding the facts accurately.

  3. Exercising discretions wisely, giving proper weight to policy.


If these are important goals of decision-making, what can we say as to how well they will be achieved in 2020?

I have no doubt that in 2020 the unstoppable march of the computer will have continued. Its use in decision-making will have advanced considerably - at all levels, including tribunal and court adjudication. What is important is that this advance should not compromise good administrative decision-making and should not impede or challenge the steady march towards the greatest possible fairness and transparency in that decision-making.

The process is now much more advanced than is generally imagined. Decisions relating to veterans entitlements and social security payments are now mostly computer assisted. Immigration decisions are frequently computer aided. More and more decisions are effectively made by a computer acting on data.

I venture to suggest that all of you will be making greater use of computer technology in decision-making, in your own roles, in 2010.

This creates three potentially serious, but basic, problems which will affect my canons of good decision-making. First, the wrong data may be entered on the computer. Secondly, the right data may be wrongly entered. In both cases the absence of all the entries on paper makes verification more difficult. Thirdly, the computer may be incorrectly programmed.

All these problems were anticipated by the Administrative Review Council nearly ten years ago. It issued a far sighted report on Automated Decision-Making in 2004. I do not know how extensively the recommendations of the ARC have been introduced into the evaluation of computer decision-making. I suspect the answer is, however, not as extensively as they should have been. What I do know is that the problems present now will be significantly expanded in 2020.

The Holy Grail for computer decision-making will be the exercise of discretions. That this might be achieved by 2020 is not as fanciful as it seems. Nor is it as troubling as it might at first seem to be. But I will come to that.

The reason why computer aided decision-making is becoming so widespread is that computers are very good at applying rules - or answering questions. Indeed, a properly programmed computer will never forget to apply all the necessary statutory tests! The human failing of forgetting the need to apply the qualification in the proviso to a sub-paragraph will not occur. To that extent the computer will always answer the question.

The problem is, ensuring that the computer is correctly programmed. This is not an easy task, particularly when the question posed by a statute or regulation is complex and involves multiple layers of alternatives.

There may be computer programmers who have a good understanding of statutory construction, but I do not think there are many. Difficult problems arise when instructions are being given to the person writing the program. Even more difficult problems arise in later verifying that the program, as written, correctly records the statutory rules.

At the recent Australian Institute of Administrative Law National Forum, Professor Julian Disney spoke of discovering a programme which made incorrect calculations of statutory entitlements, contrary to the enabling legislation, because the question posed by a drop-down box was wrong. The reason. The legislation had been amended, but not the program.

This major issue is matched by an operational one - namely ensuring correct data entry. That may seem simple and computers always seem to put up a screen to enable a check, but there is nothing like doing a calculation yourself to know if the figures are correct. The more significant problem is checking data after it has been entered and verified and the calculation made. This is, of course, relevant to review, both internal and external.

So, apart from correct programming, two major problems are verifying data entry as part of the decision-making process and subsequently accessing that information.

What the problems require is systems to enable verification that data has been entered correctly and systems which will reproduce records of the processing of the data.

What the Administrative Review Council did was to address these issues and make recommendations for systems and procedures to ensure the problems are addressed. The Council made recommendations about how systems should be designed and, very importantly, how they should be maintained. They required expert systems "to provide a comprehensive audit trail".

It is, of course, one thing to have a system which performs well but another to have a system which exposes the process sufficiently to enable it to be reviewed. How does a solicitor, seeking to advise a client whether a decision is wrong, go about checking that the automated part of the decision is correct? The ARC proposed 27 principles to cover all these matters, and more. The report is available on its website. I commend it to you.

The first group of principles adopted by the ARC caution against the use of automated systems in decision-making when a discretion is involved. That seems to me to be an area to watch in 2020.

Discretionary decision-making is applying a value judgment. That may seem, at first, to be an area inappropriate for computer technology. But value judgments mostly only involve giving weight to complex factors. At least where there are only a defined range of outcomes, there may be room for computer generated discretionary decisions.

A few years ago the use of computer aided decisions was being tested in criminal sentencing in Victoria. The object was to achieve greater consistency in sentencing - a highly desirable goal. But the result of such a system would be employing technology in a most significant and serious area of discretionary decision-making.

Yet it is not easy to see how it might work. The judge sentencing a convicted criminal is marshalling a range of factors related to the crime itself and the background of the criminal. The factors are identified and weighed. Account is taken of the seriousness of the crime, the criminal's previous history, exculpatory circumstances, indications of remorse and so on. It is not a large step to ask a computer to undertake much, or even, all of this process. I have not kept in touch with the Victorian experiment but I do not doubt that proposals like this will go forward.

Indeed I suspect that some of you, may be all of you, will be able to think of areas, even in your own departments and agencies, where these systems are developing. I would be very pleased to hear of your experiences. I suggest that between now and 2020 you watch the march of the computer: not to stop it, but to ensure that it is moving forward correctly - in a way that will enhance the quality of administrative decision-making.

The ARC this week decided to revisit its report to see if it needs to be updated. Along with the Attorney-General's Department it will be seeking to bring it more before the attention of government.

I hope my remarks this morning have assisted a little in that cause.