Victoria University of Wellington Law Review
For many years the English courts have struggled to develop a principled approach for determining when a public authority can owe a duty of care in respect of the exercise of its statutory powers. Initially, public authorities received no special treatment. Then the courts conferred an almost complete immunity on them, requiring public law irrationality to be established before considering whether a duty could arise. The English approach has not been adopted elsewhere in the Commonwealth. The High Court of Australia and the Supreme Court of Canada have developed different tests, and the New Zealand courts, while never explicitly rejecting the English position, have never followed it. This paper argues that a modified version of the Canadian Supreme Court's approach should be adopted in New Zealand. It proposes that irrationality be a precondition to the existence of a duty of care only where policy considerations are proved to have influenced the decisions of a public authority in exercising its statutory powers.
"The case law as to the duties and liabilities of a statutory body to members of the public is in a state of lamentable obscurity and confusion". So said PP Craig in 1978, quoting what MacKinnon LJ had said in 1940. One would have expected the confusion to have been resolved before the advent of the next millennium. Instead, it has been compounded.
Principally responsible, perhaps, is Parliament. Having decided that the Crown should no longer be immune from suit, it simply waived sovereign immunity completely. It further assigned governmental functions to statutory bodies and subjected them to tort liability. The statutes which effected these changes made no provision for the differences between the government and private individuals. Nor did they address the potential for the expansion of tort liability which existed following M'Alister (or Donoghue) v Stevenson. The question of whether rules limiting the government's liability in tort were to be developed was left entirely to the courts.
This paper addresses the result of the courts' struggle to provide an answer with respect to category C in Lord Browne-Wilkinson's schema of governmental liability in X (Minors) v Bedfordshire County Council. It does not address liability for breach of a statutory duty in the absence of a common law duty of care nor the question of when the government owes a duty of care not to violate the heads of judicial review. What this paper is concerned with is tort claims based on a common law duty of care arising from the failure to exercise, or the careless exercise of, a statutory power. Two difficult questions have emerged in this area. First, when decisions about whether and how to exercise statutory powers made on the basis of social and economic policy should attract liability in negligence. Secondly, when the government should be liable for pure omissions.
Only the first is addressed by this paper. Part II surveys the attempts by the main common law jurisdictions to answer this question. The House of Lords attempted to do so in Home Office v Dorset Yacht Co Ltd by conferring immunity from tortious liability wherever a statute confers a choice as to how and whether to act unless the choice made was unreasonable in the sense in which that word was used in Associated Provincial Picture Houses Ltd v Wednesbury Corporation. This approach extended immunity far beyond the policy decisions which require immunity. Mason J for the High Court of Australia attempted to address this problem in Sutherland Shire Council v Heyman. His Honour rejected Dorset Yacht's rule and proposed a rule which immunises only decisions which involve the consideration of social policy and resource allocation. Unfortunately, the other members of the High Court did not agree. Furthermore, the House of Lords in X (Minors) v Bedfordshire County Council superimposed Mason J's rule upon Dorset Yacht's rule. Instead of remedying the problem created by Dorset Yacht their Lordships thereby exacerbated it by bolstering the existing immunity. By contrast, the Supreme Court of Canada in Just v The Queen in right of British Columbia and Brown v The Queen in right of British Columbia developed and improved upon Mason J's approach by choosing a different manner of combining it with Dorset Yacht. New Zealand appears to have accepted Mason J's rule and to have ignored or implicitly rejected Dorset Yacht's rule.
Generally, the courts, and the English courts in particular, have failed to produce rules which achieve the aim of avoiding tort liability where it would interfere with governmental functions but of permitting it elsewhere. There is a notable absence of reasoning in support of some of the rules adopted. And, while most courts purport to follow precedent, they frequently depart from previous authorities without making this explicit. Confusion has resulted not least because the same terms have been assigned different meanings by different courts.
A principled reformulation of the rules governing liability is urgently required. The law in this important area should be clear. Furthermore, a rule which is intended to immunise policy decisions but which extends immunity too far has serious consequences for those harmed by government activity. A striking illustration is provided by the decision of the English Court of Appeal in Barrett v Enfield London Borough Council. The defendant council removed the plaintiff from his mother while a child and took him into its care pursuant to its statutory powers. He was relocated nine times, he was unable to develop a relationship with his family, and his psychiatric illness was not treated. The plaintiff sued in negligence, alleging that the defendant carelessly failed to exercise various statutory powers, including its power to place him for adoption. Two members of the Court applied X v Bedfordshire and struck out his claim on the basis that he could not establish Wednesbury unreasonableness. They did so although the decisions of the social workers involved raised no questions of government policy. While the House of Lords reinstated the plaintiff's action in Barrett on appeal, and although some of their Lordships clearly wished to depart from X v Bedfordshire, Barrett has not provided a principled rule.
If the New Zealand Court of Appeal wishes to discourage the Privy Council from imposing the English approach to this part of the law on New Zealand, it will have to articulate a clear position of its own and justify its rejection of Dorset Yacht's rule. This paper proposes a possible solution. Part III examines the type of rule suggested by Mason J in Heyman and the arguments for immunising the government's policy decisions from tortious liability. It proposes a policy immunity rule which confers immunity only where certain defined policy considerations are actually taken into account by a decision-maker and only where a decision is not Wednesbury unreasonable. It differs from all the tests currently employed by the courts in that it creates immunity only to the extent that it is absolutely necessary to avoid interference with the functioning of the state. Of all the rules adopted by the Commonwealth jurisdictions, the Canadian Supreme Court's approach in Brown is the closest to the proposed rule. Finally, part IV considers whether Dorset Yacht's rule should be retained if the policy immunity rule is adopted. It concludes that Dorset Yacht's rule should be rejected because it not only extends immunity too far, but also because it becomes entirely superfluous once the policy immunity rule is accepted.
It is settled law that where Parliament authorises the doing of an act, the person who does it cannot be sued in tort for any harm which is the inevitable consequence of the authorised act. This rule is known as the defence of statutory authority. No express words extinguishing tortious liability are required as the defence exists as a matter of presumed legislative intention. Of course, questions arise as to when an act is authorised and as to when a consequence is inevitable.
The first case which attempted to answer these questions was Geddis v Proprietors of Bann Reservoir, where the defendants were empowered by a local Act of Parliament to construct a reservoir in order to increase the flow of water through the River Bann for the benefit of local industry. The defendants used the River Muddock to channel water into the Bann from their reservoir. Silt accumulated in the Muddock due to the defendants using it as a channel to the Bann, but they failed to exercise their statutory power to clean the Muddock. As a result, the Muddock flooded the plaintiff's farm. The House of Lords awarded the Plaintiff damages in negligence. Their Lordships unanimously held that negligence cannot be authorised as a matter of presumed legislative intention. The rationale for this was, presumably, that negligence is always avoidable and thus cannot be the inevitable consequence of an authorised act. Thus, as a result of Geddis, any negligent conduct in the exercise of a statutory power to do some act could attract liability.
The expansion of the welfare state was accompanied by a massive increase in the delegation of policy-making by Parliament to the executive. This required the conferral of a vast array of discretionary statutory powers upon the executive. The courts were understandably reluctant to condemn the executive's policy choices as negligent, and, accordingly, the House of Lords modified the defence of statutory authority in Dorset Yacht. Parliament was now presumed to have intended to authorise negligent conduct in the exercise of discretionary statutory powers unless the conduct was also Wednesbury unreasonable. Thus no duty of care could arise in respect of the exercise of statutory powers unless irrationality could be proved first. The rationale for this was that the government often takes policy into account in exercising its discretionary statutory powers and that the courts should not interfere with such policy decisions. However, since not all discretionary decisions involve policy, Dorset Yacht extended immunity much too far. Oddly, X v Bedfordshire bolstered the executive's immunity further still, and it was not until Barrett that their Lordships exhibited any desire to retreat from Dorset Yacht. The reason may be that the until Barrett a worthy plaintiff had never been in danger of losing a case because of Dorset Yacht's irrationality requirement.
The facts of Dorset Yacht were that borstal officers had taken borstal trainees to Brownsea Island to aid their rehabilitation. The officers went to sleep instead of maintaining a watch as instructed, and the trainees escaped and damaged the plaintiff's yacht. Although the issue of whether Geddis' rule applies where the Crown's servants have a discretion to act pursuant to a statutory power did not arise for decision because the borstal officers simply breached their instructions, their Lordships considered the issue. Lord Diplock appeared to define discretion as the "right to determine the particular means within the limits laid down by the statute by which its purpose can best be fulfilled". Lords Reid and Diplock and Viscount Dilhorne all considered that where the government acts under a discretionary statutory power, the defence of statutory authority cannot be defeated by mere negligence. Viscount Dilhorne required that the public authority have acted irrationally, while Lord Reid required irrationality or an unreasonable failure to consider whether a statutory power should be exercised. Lord Diplock's "test of legality" is less clear as his Lordship at one point in his speech referred generically to the "public law concept of ultra vires" but at another went so far as to require absence of bona fides. Geddis' rule now applied only where no discretion exists. Since the borstal officers had exercised no discretion, their Lordships were able to impose a duty of care on the Home Office to take care not to permit borstal trainees to escape without irrationality being required.
The issue of whether Dorset Yacht's obiter restriction of Geddis was to become law arose in Anns v Merton London Borough Council. The plaintiffs sued a local council after they suffered loss due to its failure to exercise its statutory power to inspect the foundations of their dwellings with due care. The House of Lords refused to strike out the plaintiffs' action. Three of their Lordships concurred with Lord Wilberforce, who stated that liability could be imposed in relation to discretionary acts only if the "action taken was not within the limits of a discretion bona fide exercised". What had been obiter in Dorset Yacht thus became the law.
Lord Wilberforce further introduced the distinction between policy and operational decisions into English law. At one point in his speech he appeared to accept Lord Diplock's definition of discretion: "There may be a discretionary element in [a power's] exercise – discretionary as to the time and manner of inspection, and the techniques to be used." At another he stated that "[m]ost, indeed probably all, statutes relating to public authorities or public bodies contain in them a large area of policy. The courts call this 'discretion'". This second statement departs from Lord Diplock's definition of discretion as choice and equates discretion with the subject-matter of a decision being policy-related. However, it is submitted that Lord Wilberforce did not consider the distinction between policy-related subject-matter and discretion as defined in Dorset Yacht to be important because of his view that "[m]ost, indeed all, statutes relating to public authorities or public bodies contain in them a large area of policy". In effect, he assumed that all discretionary decisions should be afforded immunity as, in almost all cases, they relate to policy. If this interpretation is correct, then Anns did not alter the defence of statutory authority as reformulated in Dorset Yacht and simply accepted Dorset Yacht's rule. The policy/operational terminology did not reflect a new turn in the law but was merely a new label for an old rule.
A significant development occurred in Rowling v Takaro Properties Ltd, a decision of the Privy Council on appeal from New Zealand. Their Lordships recognised the need for a rule which immunised decisions involving policy rather than all discretionary decisions, stating that the policy/operations distinction was now simply "expressive of the need to exclude altogether those cases in which the decision under attack ... is unsuitable for judicial resolution". Such cases are presumably those in which social policy and resource allocation decisions are made. Thus the Privy Council's test is essentially the same as that adopted by Mason J in Heyman, and marks a significant departure from Anns, in which the distinction between policy and operations was equated with that between discretionary and non-discretionary acts. Unfortunately, their Lordships did not explain the role of Dorset Yacht's rule in the light of the new policy/operations distinction.
Clarification of Dorset Yacht's role was provided in X v Bedfordshire, where local councils were sued for carelessly exercising and failing to exercise statutory powers which allowed them to remove children from their families and to address students' learning disabilities. The case presented their Lordships with the opportunity to rationalise the law. Their Lordships could have reinterpreted Anns in the light of Rowling v Takaro Properties Ltd to hold that only decisions involving policy – as opposed to all discretionary decisions – attract immunity unless they are Wednesbury unreasonable. The English law would then have been the same as the law of Canada after Just and Brown. Unfortunately, their Lordships refused to depart from Anns and Dorset Yacht and continued to immunise all discretionary decisions – as opposed to only those involving policy – falling short of irrationality. They further refused to acknowledge that the policy/operations distinction in Rowling v Takaro Properties Ltd and Dorset Yacht's rule share the common purpose of immunising policy decisions, that the former is preferable to the latter, and that the latter is thus redundant. Instead, their Lordships superimposed Rowling v Takaro Properties Ltd upon Dorset Yacht's rule, meaning that where a discretionary decision involves policy, not even its irrationality will permit a duty to arise.
Lord Browne-Wilkinson spoke for a unanimous House of Lords in setting out four general principles. First, discretion refers to a decision made "in exercising a statutory [power] as to whether or not to do an act" and is to be contrasted with "[when] having decided to do [an] act, ... the manner in which you do it." Secondly, a discretionary decision will be immunised unless it is irrational. Lord Browne-Wilkinson expressly rejected the possibility that any of the other heads of judicial review could defeat the immunity. Thirdly, where the allocation of finite resources or the balance between competing social policy objectives must be taken into account in making a decision, the decision is non-justiciable. A finding of irrationality becomes impossible, it therefore becomes impossible to decide that the defence of statutory authority does not apply, and no common law duty can exist. It is at this third stage that the policy/operations distinction in Rowling v Takaro Properties Ltd is superimposed on Dorset Yacht.
The fourth principle is that, if an act is non-discretionary, a duty of care exists if the test for the imposition of a duty of care set out in Caparo Industries Plc v Dickman is satisfied, provided that the imposition of a duty is consistent with the statutory framework. This fourth principle implies that Geddis applies to all non-discretionary decisions. Although nothing was expressly said to this effect, it would appear that if a discretionary decision is justiciable and irrational, liability is similarly to be imposed in accordance with Caparo.
Having established a schema of liability in X v Bedfordshire, their Lordships dismantled it in part in Stovin v Wise, where a council was sued for failing to remove a bank which reduced visibility at a dangerous intersection. A majority of the House of Lords rejected the policy/operations distinction on the basis that it had produced untenable results when applied in Canada. However, their Lordships were unanimous in requiring irrationality as a precondition to liability. Nevertheless, Lord Hoffmann's reasoning may have been confined to pure omissions cases because he sought to explain liability in pure omissions cases by analogy with the test of liability for breach of statutory duty in the absence of a common law duty of care.
Dissatisfaction with Dorset Yacht and X v Bedfordshire is reflected in the House of Lords' most recent decision in this area in Barrett. In reinstating the plaintiff's claim for negligence in the exercise of a discretionary statutory power despite the absence of a pleading of irrationality, Lord Hutton stated that:
where a plaintiff claims damages for personal injuries which he alleges have been caused by decisions negligently taken in the exercise of a statutory discretion, and provided that the decisions do not involve policy which the courts are ill-equipped to adjudicate upon, it is preferable for the courts to decide the validity of the plaintiff's claim by applying directly the common law concept of negligence than by applying as a preliminary test the public law concept of Wednesbury unreasonableness ... to determine if the decision fell outside the ambit of the statutory discretion.
This statement is a clear departure from the principles set out in X v Bedfordshire, where Lord Browne-Wilkinson was emphatic in stating that "nothing which [a public] authority does within the ambit of [a] discretion can be actionable at common law". As there in no reason to suppose that Lords Hutton and Browne-Wilkinson ascribed different meanings to the expression "within the ambit of a discretion", their speeches cannot be reconciled. Lord Hutton's suggestion that Lord Browne-Wilkinson's requirement of irrationality was limited to decisions involving non-justiciable policy considerations is, with respect, incorrect. For Lord Browne-Wilkinson non-justiciability meant not that irrationality was required before a common law duty can be imposed but that "the court ... cannot reach the conclusion that the decision was outside the ambit of the statutory discretion" and therefore cannot impose a duty of care.
On the view expressed in Lord Hutton's dictum above, liability may be imposed on decision-makers if they breach a common law duty of care while exercising a discretionary statutory power without a finding that their conduct was irrational. It necessarily follows that Lord Hutton has revived Geddis and rejected Dorset Yacht as the test of when what is done under a statutory power falls within the defence of statutory authority. Indeed, his Lordship expressly described Dorset Yacht's rule as obiter. However, Lord Hutton departs from Geddis in that liability is excluded in respect of policy decisions as in Rowling v Takaro Properties Ltd. Finally, his Lordship provided no clear guidance as to whether policy decisions have retained the absolute immunity which X v Bedfordshire conferred upon them or whether, as in Brown, policy decisions can attract liability if they are irrational.
The only other Law Lord to deliver a full speech was Lord Slynn. Like Lord Hutton, his Lordship noted that Dorset Yacht's restriction of Geddis was obiter. For his Lordship, the likelihood of a decision being found to be non-justiciable increases with both the degree of discretion conferred and the degree to which competing policy considerations must be weighed. His Lordship further stated that "acts done pursuant to the lawful exercise of [a] discretion can ... be subject to a duty of care, even if some element of discretion is involved." Thus both Lords Hutton and Slynn departed from the principles set out in X v Bedfordshire.
Unfortunately, the tests proposed by Lords Hutton and Slynn differ. Furthermore, it is possible that Lord Browne-Wilkinson did not intend to depart from X v Bedfordshire and Dorset Yacht. His Lordship simply stated that the plaintiff's claim could not be struck out because it was impossible to say that the injury to the plaintiff could not have been caused by an operational decision. He probably meant that if findings of fact were made it might emerge that the social workers did not in fact exercise any discretion and could thus have owed a duty under X v Bedfordshire without having acted irrationally. If this interpretation of Lord Browne-Wilkinson's speech is correct, then because Lords Nolan and Steyn both expressed agreement with the speeches of all three of the other Law Lords, no definite rule emerges from Barrett. Whether Dorset Yacht still applies or whether Geddis has been reinstated except in respect of policy decisions is thus unclear in England.
The Supreme Court of Canada initially adopted Dorset Yacht's rule by adopting Anns. However, when it later had occasion to review Anns, it rejected Dorset Yacht. Instead of superimposing Mason J's rule in Heyman upon Dorset Yacht's rule as the House of Lords did in X v Bedfordshire, the Supreme Court in Just fused the two in the reverse order. This, it is respectfully submitted, took the Court very close to the correct approach.
Anns was adopted in City of Kamloops v Nielsen, where the city had a statutory duty to enforce a by-law which required the foundations of new buildings to meet certain standards, but a discretion as to how it enforced the by-law. Possibly because the by-law's enforcement would have been contrary to the interests of one of the city's aldermen, it failed even to consider whether it should be enforced against the builder of the plaintiff's house. Adopting Anns, the Supreme Court found the city liable. For the majority, Wilson J found that a duty existed and that the city had exceeded its discretion as to how to enforce the by-law because "inaction for no reason or inaction for an improper reason cannot be a policy decision taken in the bona fide exercise of discretion." The Supreme Court thus accepted Dorset Yacht's rule and rejected Geddis.
Just marks the Supreme Court's point of departure from Dorset Yacht's rule. The plaintiff sued the government of British Columbia in respect of its exercise of its statutory power to inspect the slopes above a road for loose rocks because a rock had struck his car. Writing for the majority, Cory J stated that ordinary common law principles were to determine liability and that special rules were to exist only in relation to "pure policy decisions". Quoting Mason J in Heyman, Cory J defined policy decisions as those "dictated by financial, economic, social or political factors". Cory J thus departed from the sense in which the term policy was used in Anns and Nielsen, where it was synonymous with discretion as defined in Dorset Yacht. Policy decisions now appear to be the subset of discretionary decisions which actually involve the consideration of policy.
Cory J then fused Dorset Yacht's and Mason J's rules. However, his approach is radically different from that in X v Bedfordshire. His Honour stated that "a policy decision is open to challenge on the basis that it is not made in the bona fide exercise of discretion". Thus, whereas in X v Bedfordshire their Lordships retained Dorset Yacht's rule and used Mason J's rule to create absolute immunity for discretionary decisions which involve policy, Cory J used Mason J's rule to confer immunity only on decisions which involve policy and not on all discretionary decisions. He then used Wednesbury unreasonableness to deny immunity to those decisions which involve policy but which are so disagreeable that they should nevertheless attract liability.
However, while repeatedly stating that the "bona fide exercise of discretion" precludes liability for policy decisions, at one point in his judgment his Honour required that the system of inspection chosen as the result of the bona fide exercise of a discretion "be a reasonable one in all the circumstances". At yet another he required "the reasonable exercise of a bona fide discretion based, for example, upon the availability of funds." From this it would appear that a reasonableness requirement of some sort has been superimposed upon the Wednesbury test which has been superimposed upon Mason J's test.
This superfluous reasonableness requirement was eliminated in Brown, whose facts were very similar to those of Just. Cory J for the majority stated that "a policy decision cannot be reviewed on a private law standard of reasonableness". Otherwise, Brown affirmed the test in Just, although it clarified that a policy decision can result in liability only if "made in bad faith or in circumstances where it was so unreasonable that it exceeds governmental discretion". Brown was itself recently affirmed in Lewis v British Columbia.
The Justices of the High Court of Australia have found themselves unable to reach agreement on their position with respect to Dorset Yacht's rule. In Heyman the High Court of Australia unanimously rejected a claim in negligence arising from facts almost identical to those in Anns. Gibbs CJ, with whom Wilson J agreed, adopted the principles set out in Anns but found that the plaintiff had failed to prove a breach of the alleged duty of care.
By contrast, Mason J, using the term "policy" not as a synonym for discretion but to refer to "decisions which involve ... financial, economic, social or political factors or constraints", granted the government an absolute immunity from negligence actions in respect of policy decisions. His Honour unequivocally rejected Dorset Yacht's rule, charging it with extending immunity too far.
Deane J did not comment on Geddis' rule but stated that a duty would be "precluded in cases where what is involved are actions taken in the exercise of policy-making powers and functions of a quasi-legislative character". Finally, Brennan J did not comment on the policy/operations distinction nor on Dorset Yacht's restriction of Geddis' rule. The fate of Geddis and of the policy/operations distinction thus remained open after Heyman.
The High Court's recent decision in Pyrenees Shire Council v Day – in which a local authority was held liable in negligence for failing to ensure that a defective chimney was attended to before it caused fire – has not clarified the Court's position. Brennan CJ required irrationality for liability. It appears, however, that his Honour saw himself as deriving a right to compensation from the language of the applicable statute or by analogy with it. Whether he considered this to have any bearing on Geddis' rule is unclear. As in Heyman, his Honour made no reference to the policy/operations distinction.
Both Toohey and McHugh JJ imposed liability without making a finding of irrationality, although McHugh J endorsed Brennan CJ's test. Neither made clear his position on the policy/operations distinction.
Finally, Kirby and Gummow JJ both rejected the requirement of irrationality as a precondition to liability, and their reasoning, unlike Brennan CJ's, was not confined to pure omissions cases. Gummow J considered public law concepts to be completely irrelevant in negligence actions, while Kirby J thought it appropriate to impose liability on the basis of Caparo's three-stage test and to ignore irrationality. Finally, Gummow J endorsed Deane J's formulation of the policy/operations distinction in Heyman, while Kirby J endorsed Mason J's.
The policy/operations distinction was affirmed as part of the law of New Zealand by the Privy Council in Rowling v Takaro Properties Ltd. This was, however, obiter, and the distinction has not been referred to in any recent New Zealand decisions. Dorset Yacht's rule has been even more unpopular. It has been referred to in only one decision, and subsequent decisions have imposed liability without requiring irrationality in circumstances in which the English courts would have required it. New Zealand, it appears, has implicitly rejected Dorset Yacht's rule and has never departed from Geddis.
The only decision of the New Zealand Court of Appeal in which Dorset Yacht's rule has been expressly considered is Takaro Properties Ltd (in receivership) v Rowling – a strike out application in the complex proceedings arising from the collapse of Takaro Properties Ltd. The Court's response to Dorset Yacht was ambiguous. Woodhouse J appeared to argue that the House of Lords in Dorset Yacht and Anns did not intend to formulate a test which would apply in all cases of alleged negligence in the exercise of statutory powers. His Honour further stated that Dorset Yacht's rule "could seem on the face of it to be potentially unjust for the aggrieved citizen". However, he then accepted as "an accurate reflection of the law" a passage from a leading text which simply restated Dorset Yacht's rule. Richardson J (as he then was) appeared to consider Anns highly persuasive and merely described Anns' adoption of Dorset Yacht's rule without offering any comment. Richmond P agreed with both Woodhouse and Richardson JJ without specifically commenting on Dorset Yacht.
In its subsequent decisions, the Court of Appeal, while enthusiastic in its acceptance of Anns' two-stage test for imposing a duty of care, has not referred to Anns' or Dorset Yacht's restriction of Geddis' rule. In Mt Albert Borough v Johnson the Court for the first time found a council liable in negligence because it had carelessly inspected a building's foundations. Despite the fact that Lord Wilberforce in Anns had expressly stated that building inspection involves discretion and that a duty could exist only if irrationality was proved, the Court imposed liability without referring to irrationality. It applied the decision of the English Court of Appeal in Dutton v Bognor Regis Urban District Council, which predated Anns and in which Dorset Yacht's rule had been ignored. The Court has taken the same approach in its most recent decisions on building inspection.
Recently, the Court of Appeal has decided two cases with facts similar to those of X v Bedfordshire. In A-G v Prince the plaintiff, who had been given up by his mother to the defendant state agency for adoption, sued the Department of Social Welfare in negligence. He alleged that he had suffered psychological injury due to the appalling parenting of his adoptive parents and that the defendant failed to prevent his injuries. The Court again found that a duty could exist without referring to irrationality, although it may not have been directly relevant because the Department was acting under statutory duties rather than powers. The same approach was taken in B v Attorney-General.
Unlike the Commonwealth Parliaments, the Congress of the United States did anticipate the difficulties which would arise in making the law of negligence into a vehicle for governmental liability. The Federal Torts Claims Act of 1946 waived sovereign immunity for torts of federal employees. However, s 2680(a) excludes liability in tort in relation to:
[a]ny claim ... based upon the exercise or performance or the failure to exercise or perform a discretionary function or duty on the part of a federal agency or an employee of the Government, whether or not the discretion involved be abused.
In Dalehite v United States the plaintiff was injured when fertiliser owned by the government exploded. The plaintiff alleged negligence in the government's decisions not to investigate the risk of explosion adequately, to bag the fertiliser at a high temperature, to use paper bagging and not to provide a warning. By a majority the Supreme Court of the United States found that these decisions all fell within s 2680(a) because they involved considerations bearing on the practicability of a cabinet decision to supply countries defeated in the Second World War with aid. It refused to define discretionary functions but said that "[w]here there is room for policy judgment and decision there is discretion".
Jackson, Black and Frankfurter JJ, dissenting, construed s 2680(a) more narrowly. Their Honours argued that the absence of evidence of a conscious decision to balance risk against the need for foreign aid meant that the exception did not apply. Any balancing of risk against the cost involved was more akin to the decisions routinely made by all private manufacturers than governmental policy decisions.
Discretion was redefined by the Supreme Court in United States v Gaubert, in which a shareholder of an investment institution sued the United States' government for the Federal Home Loan Bank Board's (FHLBB) management of the institution after taking control of it in an attempt to secure its financial position. The Court stated that the consideration of social or economic policy was not required for s 2680(a) to apply. A decision must simply be "based on the purposes that the regulatory regime seeks to accomplish." Thus, the Supreme Court has created a field of immunity similar to that which existed after Dorset Yacht and Anns, except in that s 2680(a) prevents the immunity from being defeated even by irrationality.
Given the wide divergence in the approaches taken by the courts, it may be helpful to summarise them.
Where a decision is made in the exercise of a discretion as to how to achieve a statute's purpose, a duty of care can exist only if irrationality is established. Irrationality does not suffice for liability but merely defeats the defence of statutory authority. To establish liability, a duty must also arise under ordinary common law principles and have been breached.
As for Dorset Yacht, but where a decision is non-justiciable and discretionary, liability is impossible.
Where a decision involves policy, no duty can exist. Otherwise Geddis applies.
Where a decision involves policy, a duty can exist only if the decision is irrational. Otherwise Geddis applies.
Cory J's rule in Brown and Mason J's rule in Heyman are considered here. These rules assume that different rules of liability are required for the government and private individuals because the former makes policy decisions in the public good whereas the latter do not. They therefore seek to identify a class of policy decisions and to create special rules of liability for this class. They assume that no significant differences exist between decisions made by private individuals and government decisions which do not involve policy, concluding that ordinary common law principles suffice for both. Finally, they reject Dorset Yacht's rule; the fact that public authorities may have a discretion as to how to further the purposes of a statute is not seen as a basis for differential treatment.
Arguments which have been advanced in support of the creation of special rules of liability for the government are considered below. It is submitted that these arguments show that differential treatment is necessary. And, in as far as these arguments are valid, it is submitted that the policy immunity rule developed below answers them. The rule argued for is a modified version of that adopted by Cory J in Brown. Part IV will then consider whether, in the light of the policy immunity rule formulated below, a special rule for discretionary decisions is still required.
The absence of objective standards by which courts can assess the reasonableness of the executive's policy decisions is often cited as a reason for denying them the ability to hold policy decisions negligent.
It may be difficult to assess the reasonableness of a decision to attempt to rehabilitate borstal trainees by taking them to an island and thereby to increase the risk of harm to the public. However, it seems obvious that not all policy decisions defy characterisation as negligent. For example, a decision to release all dangerous inmates from prisons so as to promote rehabilitation could without difficulty be described as careless. It seems absurd to say that an absence of objective standards would hinder a court from being able to make such a determination. Thus, while the absence of objective standards means that a court should not condemn as unreasonable every policy decision with which it disagrees, there are undeniably some extremely objectionable policy decisions which can safely be held negligent. For this reason it is submitted that even where the government can prove that it actually took policy considerations into account in making a decision as to how to exercise its statutory powers, liability should be possible, but only if the decision is unreasonable in the Wednesbury sense. The class of decisions qualifying as "policy decisions" can, drawing on Mason J's comments in Heyman, be defined as including decisions involving resource allocation, social policy and political considerations.
While it may be objected that the Wednesbury test creates uncertainty and is subjective, it is a necessary compromise, as the only alternatives are to immunise all policy decisions no matter how absurd or to immunise none. The debate concerning whether the Wednesbury test's threshold is too high is beyond the scope of this paper.
According to this argument, it is Parliament's or, when Parliament chooses to delegate, the executive's role to determine what the public good requires and to act accordingly. The courts' role is merely to apply the law as Parliament enacts it, and not to usurp the discretionary statutory powers by which Parliament delegates to the executive. Dissatisfaction with policy decisions is to be remedied via the ballot box. Judicial condemnation of executive action as negligent is therefore said to be inappropriate.
This argument is not persuasive. First, under a Westminister constitutional system the separation of powers is honoured more often in the breach than in the observance. Inherent in the common law is a judicial legislative capacity. Furthermore, the proposed policy immunity rule would restrict any intrusion by the courts into the policy realm to clearly unreasonable decisions.
Secondly, one of Montesquieu's primary reasons for propounding his famous theory was his desire that the branches of government act as checks on each other. He said that "power should be a check to power". To prevent the courts from checking executive excesses in the form of violations of individuals' common law rights on the basis of the separation of powers would be self-defeating. It would also be to ignore the fundamental requirement of the rule of law that the government act within the law.
Thirdly, the inadequacy of the ballot box as a remedy for administrative matters which are not of overwhelming public concern is reflected in the rapid development of administrative law this century. It is simply unrealistic to expect the government to fear that it will lose an election because of negligent building inspections however grave may be the injustice to those affected by them.
Finally, it is submitted that the argument that the adversarial process is unsuitable for evaluating policy decisions should be rejected. If current rules of evidence and other courtroom procedures are indeed inadequate, then it is open to the courts to modify them.
Feldthusen has argued that negligence actions, at least in respect of a public authority's failure to confer a benefit, may result in "astronomical" increases in court costs because "[b]ilateral dispute resolution is an awkward vehicle with which to assess public policy".
The costs associated with proceedings are an empirical matter in relation to which Feldthusen adduces no evidence. Furthermore, judicial review proceedings have been heard by the courts for many years without great alarm about their expense, and even a substantial increase in court costs resulting from negligence actions may be justified by a corresponding decrease in the violation of individuals' common law rights.
The most frequently advanced argument for refusing to impose a duty of care on the government is the argument that a duty would cause public servants to act defensively. This argument can take three forms.
One form of the argument from defensiveness was advanced by Lord Hoffmann in Stovin. His Lordship stated that imposing a duty of care on a county council to ensure that its roads are safe would, given its limited budget, "distort" its priorities, causing it to increase spending on road improvement and to reduce spending on education and social services. Such distortion interferes with Parliament's delegation to the decision-maker and the separation of powers, and there is force in this argument unless duties of care are also owed in respect of education and social services. However, the proposed policy immunity rule answers this objection as public servants will have no cause for concern or defensive behaviour unless their policy decisions are Wednesbury unreasonable.
Secondly, it has been argued that imposing a duty of care in respect of the exercise of statutory powers which permit the conferral of benefits instead of authorising otherwise tortious conduct may cause public authorities to decide not to exercise such powers in an attempt to avoid liability. This argument is easily answered because public authorities have a public law duty properly to consider whether to exercise a power to confer a benefit. A decision never to exercise a statutory power based solely on the avoidance of liability for negligence in the exercise of the power is likely to be quashed by a court if judicial review is sought. It is submitted that precisely this form of the defensiveness argument was raised in Anns and that Lord Wilberforce employed the very response just developed to refute it.
In its third form the defensiveness argument states that even if defensiveness in respect of policy decisions can be avoided, the imposition a duty of care will cause undesirable defensiveness in non-policy decisions. This will occur, for example, where building inspectors insist on five foot foundations where three foot foundations are adequate for fear of having overlooked the fact that five foot foundations are actually required. This argument is often overstated. While imposing a duty may cause inefficiency, it may also reduce inefficiency by encouraging the taking of due care. Whether this third form of the argument justifies denying a duty of care can be considered on a case by case basis at the policy-balancing stage of the Anns and Caparo tests.
It could be argued that the government must be treated differently from private individuals because only the government is able or permitted to do or to omit to do certain acts. For example, only the government may operate prisons.
It is submitted that this argument should be rejected. Accepting it would commit one to the odd conclusion that the government would lose its immunity from negligence in relation to prisons if ever a statute were enacted which permitted private individuals to operate prisons. Clearly, any distinction between the rules of liability for the government and private individuals must be based on differences between the policy considerations which bear on the imposition of duties of care on them and on the fact that only the government makes policy decisions.
As is argued below, the distinction between positive acts and pure omissions has no bearing on the form of the policy immunity rule.
The policy immunity rule contended for is outlined in this part of the paper. It is based upon the rule adopted by the Canadian Supreme Court in Just and Brown, but differs in significant ways. The elements of the proposed rule are defended here to the extent that they have not been argued for above.
In order to be able to invoke the policy immunity rule, the decision-maker must have a choice as to what to do in the exercise of a statutory power.
No liability can arise if a policy decision falling within the immunity rule is made, if employees are instructed to implement it and if they follow their instructions. However, when a policy decision is made and employees are instructed to do a certain act which they forget to do or do carelessly, the employees will not have made a policy decision and the government will not be able to invoke the policy immunity rule in respect of their negligence.
The view taken by some courts that, once a policy decision has been made, no subsequent decision can be a policy decision, is not adopted here. Such a rule ignores the fact a policy maker who decides on a course of action will not necessarily consider every policy aspect of the task at hand before its implementation commences. Policy matters may require consideration by other personnel once implementation has begun. Thus, any decision involving policy which meets the requirements set out here should fall within the policy immunity rule.
The government must actually have taken policy considerations into account in reaching a decision in order for it to fall within the policy immunity rule. Following Mason J in Heyman, policy decisions can be said to include those "dictated by financial, economic, social or political factors" as distinct from "administrative direction, expert or professional opinion [and] technical standards". These categories will require clarification by the courts. A detailed discussion of the decisions which should be classified as policy is beyond the scope of this paper. It is, however, obvious that a decision as to whether limited government funds should be invested in defence or in hospitals should be regarded as a policy decision. The same can be said of social policy decisions regarding the appropriate level of taxation or whether to have open rather than closed borstals.
One may wonder whether immunity should exist for government decisions which cannot involve policy or government decisions which could involve policy but in the making of which policy factors were not taken into account. It should be noted that the question here is not whether a duty of care should be owed in respect of all government activity which does not involve policy decisions. Even if a decision does not fall within the policy immunity rule, the policy or justice and reasonableness stage of the Anns or Caparo test may nevertheless prevent a duty from arising. The question is thus simply whether, where a decision is made pursuant to a statutory power, the very fact that the decision is made by the government and that it is made under a statutory power rather than by a private individual not acting under a statutory power should suffice for immunity.
It is submitted that it should not. A public authority which, without actually basing its conduct on policy considerations, acts in such a manner that a private individual in its position would be liable in negligence should be held liable. Why, for example, should a minister who has an enormous budget but who directs that foundations of inadequate thickness be used for a building due to ignorance about building safety be treated differently from a private individual who acts in the same manner and causes harm? The only justification for differential treatment is policy. When policy is absent, there should be no differential treatment because it is not justified by the separation of powers or a lack of objective criteria with which to assess the decision's reasonableness. As was said by the dissent in Dalehite, "an increased sense of caution and responsibility even at [cabinet] height would be wholesome". Furthermore, Mason J stated in Heyman that the courts are capable of assessing the reasonableness of government action where it is based on "administrative discretion, expert or professional opinion [or] technical standards" rather than policy. Overlooking this fact is Dorset Yacht's primary fault.
Curial support for requiring a public authority actually to have taken policy considerations into account can be found in X v Bedfordshire, where Lord Browne-Wilkinson referred to "the relevant factors taken into account by the authority" in explaining his non-justiciability rule. Feldthusen has also suggested such a requirement.
In Just, Cory J distinguished between "high level" policy decisions, such as a decision to build a lighthouse, and those made at a "lower level", such as a decision about how to inspect aircraft parts. It was said that a decision falling within the latter sub-category of policy decisions could attract liability unless "the government agency establishes that it was a reasonable decision in the light of the circumstances". As Feldthusen correctly points out, drawing such distinctions is "a hopeless task" as there are no obvious criteria for distinguishing among various levels of policy.
While Just's reasonableness requirement was rejected in Brown, the difference between the results in Just and Brown, unless explained by an absence of evidence that policy was actually considered in Just, appears to expose another problematic attempt to distinguish among different levels of policy. In Just a highway authority's decision as to how frequently to inspect a slope above a highway for loose rocks was found not to have been a policy decision. By contrast, a highway authority's decision as to the frequency of shifts for removing black ice was held to have been a policy decision in Brown. Such distinctions, which, in the words of Lord Hoffmann in Stovin, are "hardly visible to the naked eye", must be rejected. As his Lordship pointed out, practically every decision, no matter how trivial, can affect a public body's budget. It is therefore submitted that every decision in the making of which policy is actually considered should fall within the policy immunity rule.
While a policy decision can be made at any level, there are some government employees of whose task is no part to make policy decisions although they may have the opportunity to take policy into account in making decisions which they are required to make. Scalia J suggested the requirement proposed here in Gaubert. His Honour said of the dock workers in Dalehite that even if they had performed a careful analysis of the risks and benefits of storing explosive fertiliser in a certain manner, such a decision would not have fallen within s 2680(a) of the Federal Tort Claims Act because it was not their task to ponder such things.
It should be noted that this requirement does not reintroduce the uncertainty which rejecting the sub-categorisation of policy decisions was intended to avoid. Determining whether an employee's work involves making policy decisions is quite different from distinguishing between high and low level policy decisions.
The burden of establishing that policy was taken into account in reaching a certain decision should rest on the government. Such a burden shift is justified by the obvious difficulty which a private plaintiff could face in determining what the government took into account in reaching a decision and by the fact that, as the government routinely keeps records of its decisions for other purposes, it would not be unduly onerous. Authority for such a burden shift may be drawn, by way of analogy, from Just, where Cory J stated that "a true policy decision may be made at a lower level provided that the government agency establishes that it was a reasonable decision".
It is submitted that policy decisions should not automatically be accorded immunity from negligence liability no matter how absurd they are. Only those which are not Wednesbury unreasonable should be immune.
As was argued above, neither the absence of objective standards nor the separation of powers nor the need to avoid defensiveness mandates absolute immunity. Feldthusen, however, asserts that it is "simply incoherent" to stop short of absolute immunity because it is impossible to determine when a policy choice is negligent. He argues that courts should not demand that every beneficial programme be funded until its "net costs" exceed its "net benefits" and that it is impossible for the courts to evaluate by any other means the reasonableness of the government's resource allocation decisions. While the reasonableness of resource allocation decisions no doubt should not be assessed by means of a marginal cost/benefit analysis, there are other ways of making such an assessment. Surely, it is possible to describe a decision to close all hospitals and to invest all the money saved in ministerial limousines as negligent without a marginal cost/benefit analysis?
It is therefore submitted that New Zealand should follow the Canadian Supreme Court in Brown in rejecting an absolute immunity rule, particularly as the English courts' have adopted no definite position on this matter. While the Privy Council in Rowling v Takaro Properties Ltd suggested an absolute immunity rule, it left the matter open. The House of Lords accepted absolute immunity in X v Bedfordshire, but in Stovin the majority rejected the policy/operations distinction, while Lord Nicholls for the minority said that "an area of blanket immunity seems undesirable and unnecessary". Finally, in Barrett their Lordships did not explain whether X v Bedfordshire's absolute immunity rule had been reinstated.
It remains to explain how the policy immunity rule supplements the ordinary principles of negligence. Various possibilities exist. The rule could, as was done by the Supreme Court of Washington in Evangelical United Brethren Church of Adna v The State of Washington, simply be read into the Crown Proceedings Act 1950 and other Acts which make government bodies liable in tort. Alternatively, the policy immunity rule could be seen as part of the defence of statutory authority. As a matter of statutory interpretation, the courts could presume that Parliament, when conferring a statutory power, intended to extinguish negligence actions only to the extent that they might result in liability for policy decisions which are not Wednesbury unreasonable. It has also been suggested that a policy immunity rule could alter the standard of care required of a public defendant. Finally, the policy immunity rule could be treated as a factor negativing the existence of a duty of care under the Anns and Caparo tests.
Which of these approaches is adopted is unimportant. That based on the defence of statutory authority requires the smallest conceptual departure from Dorset Yacht and should perhaps be preferred for that reason. The necessary and sufficient conditions for liability are thus as follows:
(1) If the government has exercised a discretionary statutory power without considering policy, a duty can arise under the ordinary Caparo or Anns tests for imposing a duty. The defence of statutory authority is as in Geddis.
(2) If the government has made a policy decision, Geddis applies only if the decision does not fall within the policy immunity rule. If the decision does not fall within the policy immunity rule, a duty can arise under Caparo or Anns. The policy immunity rule applies if:(a) a policy decision is not irrational; and
(b) a statute gives the decision-maker a choice as to what to do; and
(c) the decision-maker actually took policy matters (resource allocation, social policy or political considerations) into account; and
(d) the decision-maker proves that policy was actually taken into account; and
(e) it was appropriate for the decision-maker to take policy into account.
(3) The duty was breached.
(4) The breach caused damage.
It should be noted that if a policy decision is found to be Wednesbury unreasonable, then, if a duty arises under Caparo, it should be presumed that that duty has been breached. It would be odd to say that a decision is irrational but that it was made with reasonable care. Furthermore, the policy immunity rule cannot be a "touchstone of liability". While categorisation of a decision as a policy decision precludes liability in the absence of irrationality, categorisation of a decision as not involving policy does not suffice for liability.
Section 6(1) of the Crown Proceedings Act 1950 – the provision which must generally be relied on in tort actions against the Crown – provides as follows:
Subject to the provisions of this Act and any other Act, the Crown shall be subject to all those liabilities in tort to which, if it were a private person of full age and capacity, it would be subject---(a) In respect of torts committed by its servants or agents ...
Whether the language of this provision permits the courts to create special rules governing the liability of the Crown in negligence which do not apply to individuals is unclear. Indeed, the Supreme Court of the United States in Dalehite noted that the New Zealand legislature had never enacted a provision equivalent to s 2680(a) of the Federal Torts Claims Act and had thus "left open to grave doubt how far, if at all, it ... intended ... to give the subject rights of action which in the result would seriously interfere with the ordinary administrative work of the government".
Any questions in this regard raised by the provision were, however, ignored by the Privy Council in approving the policy/operations distinction as the law of New Zealand in Rowling v Takaro Properties Ltd. Furthermore, the Supreme Court of Washington in the Evangelical United Brethren Church case simply read a rule similar to that contained in s 2680(a) into the statute which waived sovereign immunity for the State of Washington although the statute stated that immunity had been waived even in respect of acts done by the state in its "governmental ... or proprietary capacity". Faced with an argument against the policy immunity rule based on s 6(1)(a), the courts would probably reject it on the basis that Parliament did not turn its mind to the need for such a rule and that the consequences of its rejection would be unacceptable.
The fate of Dorset Yacht's rule is considered here. Dorset Yacht restricted Geddis' rule that negligence can never be implicitly authorised by a statute to non-discretionary conduct under statutory powers. It stated that where decision-makers have a choice as to how to implement Parliament's aims in conferring a statutory power their decisions, even if negligent, cannot attract liability unless they are Wednesbury unreasonable. It is argued here that Dorset Yacht's rule should be rejected for essentially two reasons; it extends immunity too far and is redundant if the policy immunity rule is accepted.
Todd has stated that "[i]t is hard to understand why the ultra vires doctrine has been introduced ... Invalidity is not the test of fault and should not be the test of liability". This criticism is misconceived. It appears incorrectly to assume that irrationality is a sufficient condition for liability under Dorset Yacht's rule. Their Lordships were, however, quite clear in stating that this is not so. They stated that if an act or omission is intra vires it is presumed to be within the margin of error which Parliament intended to permit when conferring a statutory power and that the defence of statutory authority thus prevents liability in negligence for intra vires conduct. By contrast, when an administrative decision is ultra vires, the defence of statutory authority cannot apply. However, this does not suffice for liability as their Lordships stated that liability further required the breach of a duty of care under ordinary common law principles. Dorset Yacht thus cannot be criticised for making proof of ultra vires conduct the test of liability.
Their Lordships in Dorset Yacht advanced several arguments to justify their restriction of Geddis. It is respectfully submitted that these arguments are unpersuasive.
The House of Lords in Dorset Yacht explained in detail its reasons for departing from Geddis. Lord Reid said in respect of an open borstal policy that:
the responsible authorities must weigh on the one hand the public interest of protecting neighbours and their property from the depredations of escaping trainees and on the other hand the public interest of promoting rehabilitation. Obviously there is much room here for differences of opinion and errors of judgment ... There could only be liability if the person entrusted with discretion reached a conclusion so unreasonable as ... to show a failure to do his duty.
This is simply the argument from the absence of objective standards discussed above. Lord Diplock gave the same justification, saying that in balancing the imperatives of rehabilitation and protection of the public "there [was] no criterion by which a court can assess where the balance lies between the weight to be given to one interest and that to be given to another".
In X v Bedfordshire, Lord Browne-Wilkinson retained Dorset Yacht's rule as a precondition to liability but added a separate rule of absolute immunity for non-justiciable decisions. Explaining the purpose of the non-justiciability rule, his Lordship said:
Since what are under consideration are discretionary powers conferred on public bodies for public purposes the relevant factors will often include policy matters, for example social policy, the allocation of finite financial resources between different calls made upon them or (as in Dorset Yacht) the balance between pursuing desirable social aims as against the risk to the public inherent in so doing. It is established that the courts cannot enter upon the assessment of such "policy" matters.
Strikingly, the rationale for the non-justiciability rule is identical to that for Dorset Yacht's rule. There thus exist two rules with the same function but which differ in the manner in which they seek to achieve it. The non-justiciability rule directly identifies decisions which involve policy and which the courts should not review; Dorset Yacht's rule immunises all decisions involving choice as to how a statutory aim is achieved because they may involve the consideration of policy.
It is respectfully submitted that this redundancy should be remedied by abandoning Dorset Yacht's rule. The non-justiciability rule – of which the proposed policy immunity rule is a reformulation – is better suited to the purpose which both rules serve because it confers immunity only where necessary. Lord Diplock in Dorset Yacht defined discretion as the "right to determine the particular means within the limits laid down by the statute by which its purpose can best be fulfilled". And he required irrationality as a precondition to liability for any discretionary decision. Clearly this extends the immunity beyond the policy decisions to which it is intended to apply. If a statutory power exists as to how to inspect buildings' foundations, a decision about the type of tape measure to use falls within the definition of discretion although it may have been made without any thought being given to social policy or resource allocation. As has been argued above, immunity in such cases is undesirable.
According to Lord Diplock, "over the past century the public law concept of ultra vires has replaced the civil law concept of negligence as the test of legality". However, his Lordship cited no authority in support of this proposition. Furthermore, in 1945 in Fisher v Ruislip-Northwood Urban District Council Lord Greene MR, after a wide survey of decisions on the careless exercise of statutory powers, concluded that Geddis was still correct.
Lord Diplock appeared to argue that Geddis' rule could determine when the defence of statutory authority is exceeded in respect of statutory powers which authorise otherwise tortious conduct but not in respect of powers which permit conduct which is not tortious – powers which permit the conferral of a benefit. His Lordship did not explain why this should be so, and it is respectfully submitted that the suggested distinction is unsustainable. Whether the government commits a tort by actively harming individuals or by failing to assist them, it should be liable where it breaches a common law duty of care unless a policy decision is involved. Concerns about the desirability of governmental liability for pure omissions should be addressed when deciding whether a duty arises under the Caparo or Anns test.
Even if Lord Diplock's attempt restrictively to distinguish Geddis in respect of powers to do acts which are tortious unless authorised could be justified, it would provide no support for his Lordship's conclusion that "the public law concept of ultra vires has replaced the civil law concept of negligence as the test of legality", which suggests that Geddis has been universally displaced.
Some of the decisions which are reviewed in part II would not be greatly affected by the adoption of the policy immunity rule and the absence of Dorset Yacht's rule. This is because they involved strike out applications without a finding of irrationality having been made. However, the significance of the policy immunity rule is not diminished. Its adoption would have a significant effect when cases actually go to trial and on potential plaintiffs when considering whether to sue.
As discussed above, the effect of Dorset Yacht's rule in Barrett is particularly unfortunate. If Barrett has failed to change the law in England, then if the plaintiff's injuries cannot be traced to non-discretionary acts by the social workers he will fail unless the social workers' conduct meets the notoriously high Wednesbury threshold. By contrast, the plaintiff would probably succeed if the policy immunity rule were adopted. He would lose only if no duty arose under Caparo or if the council could prove that the social workers made resource allocation or social policy decisions which were not irrational.
It is unclear whether the result in Brown would change if the policy immunity rule were applied. The decision not to switch from the summer road inspection schedule to the winter schedule may have involved financial considerations. Whether such evidence was provided by the government of British Columbia is unclear from the report of the case. If evidence that policy had actually been considered was presented, the case would fall within the policy immunity rule. The plaintiff would then have to establish that it was morally unacceptable and therefore irrational to permit the public to use the road without eliminating the risk posed by the falling rocks. This argument might succeed. If policy was not actually considered, the policy immunity rule would not apply, and, given the prime facie duty which arose in Just, liability would be likely.
Finally, Gaubert would have been decided differently had it arisen for decision under the policy immunity rule rather than s 2680(a). White J for the majority found that although no social or economic policy had been considered by the FHLBB, the s 2680(a) immunity applied because decisions about the day-to-day operation of the IASA were made by the FHLBB in furtherance of the Home Owners' Loan Act's policy of ensuring the stability of the financial system. Under the policy immunity rule, such decisions would not attract immunity because they do not fall within the class of policy decisions. The plaintiff would have succeeded if the FHLBB was negligent.
It appears clear that the law of negligence cannot, without some modification, provide satisfactory rules for holding the government liable for the careless exercise of its statutory powers. The reason, essentially, is that the government, unlike individuals, must make policy decisions in the public interest. Constitutionally it is inappropriate for the courts to create duties of care which interfere with policy decisions, and assessing whether a policy decision is unreasonable presents practical difficulties. These considerations need not, however, drive the courts to immunise all discretionary decisions or to confer absolute immunity on policy decisions.
A modified version of the Canadian Supreme Court's rule in Just and Brown suffices to address the difficulties which arise in this area of the law. Immunity should be conferred only where certain defined types of policy consideration are actually taken into account and only where the impugned decision is not Wednesbury unreasonable. Only in this manner can the courts avoid extending immunity too far and a repetition of Barrett.
What must be eschewed is the English courts' use, beginning in Dorset Yacht, of judicial review principles to immunise all discretionary decisions on the basis that the courts should not usurp the executive's decision-making powers. As Mason J said in Heyman, "[a]lthough such injunctions have compelling force in their application to policy-making decisions, their cogency is less obvious when applied to other discretionary matters".
It is thus submitted that the New Zealand courts should reject Dorset Yacht's rule. While Barrett may foreshadow the House of Lords' abandonment of Dorset Yacht, the Court of Appeal, if it wishes to avoid having X v Bedfordshire's rule imposed on New Zealand, should not remain silent. No Privy Council authorities oblige New Zealand to follow Dorset Yacht, and forceful arguments against Dorset Yacht's rule can and should be articulated.
[*] This paper was submitted in fulfilment of the LLB (Hons) requirements at Victoria University in 1999.
 PP Craig "Negligence in the Exercise of a Statutory Power"  LQR 428.
 East Suffolk Rivers Catchment Board v Kent  1 KB 319, 332.
 In New Zealand tort actions against the Crown first became possible under ss 26, 27 and 37 of The Crown Redress Act 1881. Such actions were, however, confined to damage caused by public works, which were defined by s 37(3) as such things as railways, roads and bridges. The 1881 Act was repealed and replaced by the Crown Suits Act 1908. Section 25 of the 1908 Act together with s 3(c) of the Crown Suits Amendment Act 1910 exposed the Crown to liability for all torts. The 1908 Act was in turn replaced by the Crown Proceedings Act 1950, s 6(1)(a) of which makes the Crown liable in tort but only where an individual in its place would be liable for torts committed by its servants. In England, sovereign immunity was not waived until the passing of the Crown Proceedings Act 1947.
 For example, under s 37L(4) of the Local Government Act 1974 territorial authorities are bodies corporate and are "capable ... of suing and being sued". Section 9(2) of the Securities Act 1978 makes similar provision in respect of the Securities Commission.
 This term will be used to refer to the Crown and all agencies under its control.
 M'Alister (or Donoghue) v Stevenson  AC 562.
 X (Minors) v Bedfordshire County Council  UKHL 9;  2 AC 633, 730-731.
 This question was raised in Rowling v Takaro Properties Ltd  UKPC 2;  1 AC 473.
 The second requires a voluminous analysis which is impossible in this paper.
 Home Office v Dorset Yacht Co Ltd  AC 1005 [Dorset Yacht].
 Associated Provincial Picture Houses Ltd v Wednesbury Corporation  EWCA Civ 1;  KB 223, 230. The terms Wednesbury unreasonableness and irrationality will henceforth be used interchangeably.
 Sutherland Shire Council v Heyman  HCA 41; (1985) 157 CLR 424 [Heyman].
 Heyman above n 12, 457-458, 469 per Mason J.
 X (Minors) v Bedfordshire County Council  UKHL 9;  2 AC 633, 735-739. Five cases were heard together. References to principles common to all five cases will be to X v Bedfordshire. The individual cases will be referred to by reference to the names of the parties involved in each (for example the Newham case).
 Just v The Queen in right of British Columbia (1989) 64 DLR (4th) 689 [Just].
 Brown v The Queen in right of British Columbia (1994) 112 DLR (4th) 1 [Brown].
 See below part II D: Mt Albert Borough v Johnson  NZCA 46;  2 NZLR 234; A-G v Price  1 NZLR 262.
 Barrett v Enfield London Borough Council  QB 367 [Barrett].
 Barrett above 18, 375 per Lord Woolf MR, 381 per Schiemann LJ.
 Barrett v Enfield London Borough Council  3 WLR 79.
 The rationale for the defence is that without it statutory authority would become "nugatory" because defendants could be prevented from doing what an Act permits by high damages awards or injunctions: Allen v Gulf Oil Refining Ltd  UKHL 9;  AC 1001, 1017-1018, 1023.
 Geddis v Proprietors of Bann Reservoir (1878) 3 App Cas 430 [Geddis].
 Geddis above n 22, 445, 452, 455. Their Lordships had previously, although less explicitly, accepted a similar principle in Mersey Docks and Harbour Board Trustees v Gibbs  EngR 174; (1864-1866) 11 HLC 686, 713.
 HWR Wade and CF Forsyth Administrative Law (7ed, Clarendon Press, Oxford, 1994) 771. See also Dorset Yacht, above n 10, 1066 per Lord Diplock.
 Dorset Yacht above n 10, 1030-1031 per Lord Reid, 1066-1068 per Lord Diplock.
 Below part IV B 1.
 Dorset Yacht above n 10, 1031. Lords Slynn and Hutton make this point in Barrett above n 19, 97, 103.
 Dorset Yacht above n 10, 1067.
 This rule is henceforth referred to as Dorset Yacht's rule.
 Dorset Yacht above n 10, 1049.
 Dorset Yacht above n 10, 1031.
 Dorset Yacht above n 10, 1067.
 Dorset Yacht above n 10, 1068. It is submitted that his Lordship used this term as a synonym for the term ultra vires as he appeared to use these terms interchangeably.
 Dorset Yacht above n 10, 1071.
 Anns v Merton London Borough Council  UKHL 4;  AC 728 [Anns].
 It has been suggested that by using the term "bona fide" his Lordship intended to lower the threshold governing when the defence of statutory authority has been exceeded: S Kneebone Tort Liability of Public Authorities (The Law Book Company, Sydney, 1998), 83. This cannot be correct. Lord Wilberforce adopted the expression from Lord Diplock's speech in Dorset Yacht, where it was used as a synonym for public law intra vires, and cited Dorset Yacht with approval on this point: Anns above n 35, 757. Moreover, the House of Lords and the Supreme Court of Canada both interpreted Lord Wilberforce as adopting the Dorset Yacht test: X v Bedfordshire, above n 14, 736-737; Brown, above n 16, 11-12.
 Anns above n 35, 755.
 Anns above n 35, 754.
 Anns above n 35, 755.
 Anns above n 35, 754.
 This ambiguity is pointed out by SH Bailey and MJ Bowman "The Policy/Operational Dichotomy – A Cuckoo in the Nest"  CLJ 430, 438-439.
 This view is also taken by B Feldthusen "Failure to Confer Discretionary Public Benefits: The Case for Complete Negligence Immunity"  TLR 17, 20.
 Rowling v Takaro Properties Ltd  UKPC 2;  1 AC 473.
 Rowling v Takaro Properties Ltd above n 43, 501. References to the policy/operations distinction below are to the distinction as formulated here in Rowling v Takaro Properties Ltd and not as formulated in Anns. As the policy/operations distinction as it existed in Anns was the same as Dorset Yacht's rule, both the approach in Anns and that in Dorset Yacht will henceforth be referred to as Dorset Yacht's rule. Anns will be specifically referred to where it is necessary to distinguish Lord Wilberforce's reasoning from that in Dorset Yacht.
 Below part IV B.
 X (Minors) v Bedfordshire County Council above n 14, 735.
 X (Minors) v Bedfordshire County Council above n 14, 736-737.
 X (Minors) v Bedfordshire County Council above n 14, 737-738.
 Caparo Industries Plc v Dickman  UKHL 2;  2 AC 605, 617-618. The three-stage test requires that the damage suffered by the plaintiff be reasonably foreseeable, that there be sufficient proximity between the plaintiff and defendant and that it be just and reasonable to impose a duty of care. The creation of novel duties is to be informed by precedent.
 X (Minors) v Bedfordshire County Council above n 14, 739.
 Stovin v Wise  AC 932.
 Stovin v Wise above n 51, 955-956. Their Lordships specifically referred to Brown and Just.
 Stovin v Wise above n 51, 936 per Lord Nicholls, 953 per Lord Hoffmann.
 Stovin v Wise above n 51, 953.
 It is not clear whether the plaintiff in fact sued in respect of a specific statutory power as literally understood. He may have been suing in respect of the council's breach of its statutory duties. Their Lordships however considered that the council was being sued in respect of "statutory discretions": Barrett above n 20, 82. It thus appears that Dorset Yacht's rule, which might have been thought to be confined to discretion existing under statutory powers, may also apply where there is a wide range of different courses of action which can be seen as discharging a broadly framed statutory duty.
 Barrett above n 20, 111.
 There is no reason to suppose that his Lordship intended to restrict this principle to personal injury claims as opposed to negligence claims based on property damage. It is submitted that personal injury was referred to merely because the plaintiff's claim was in respect of personal injury.
 X (Minors) v Bedfordshire County Council above n 14, 738.
 Barrett above n 20, 109.
 X (Minors) v Bedfordshire County Council above n 14, 738.
 Barrett above n 20, 111.
 Barrett above n 20, 103.
 Barrett above n 20, 97.
 Barrett above n 20, 97.
 Barrett above n 20, 82-83.
 Barrett above n 20, 100.
 City of Kamloops v Nielsen (1984) 10 DLR (4th) 461 [Kamloops].
 Kamloops above n 67, 663.
 Kamloops above n 67, 673.
 Just above n 15, 708-709.
 Just above n 15, 705-706, citing Heyman above n 12, 469.
 Just above n 15, 708.
 Just above n 15, 708.
 Just above n 15, 706.
 Just above n 15, 707.
 Brown above n 16, 16.
 Brown above n 16, 11-12.
 Lewis v British Columbia 1997 CanLII 304 (SCC);  3 SCR 1145.
 Heyman above n 12, 442.
 Heyman above n 12, 469.
 Heyman above n 12, 457-458, 468.
 Heyman above n 12, 500.
 Pyrenees Shire Council v Day  HCA 3; (1998) 192 CLR 330.
 Pyrenees Shire Council v Day above n 83, 346-347.
 Pyrenees Shire Council v Day above n 83, 347.
 Pyrenees Shire Council v Day above n 83, 363 per Toohey J, 374 per McHugh J.
 Pyrenees Shire Council v Day above n 83, 373.
 Pyrenees Shire Council v Day above n 83, 390-391.
 Pyrenees Shire Council v Day above n 83, 419.
 Pyrenees Shire Council v Day above n 83, 426.
 Pyrenees Shire Council v Day above n 83, 393-394 per Gummow J, 425-426 per Kirby J.
 Rowling v Takaro Properties Ltd above n 43, 500.
 Takaro Properties Ltd (in receivership) v Rowling  2 NZLR 314.
 Takaro Properties Ltd (in receivership) v Rowling above n 93, 326.
 Takaro Properties Ltd (in receivership) v Rowling above n 93, 327.
 Takaro Properties Ltd (in receivership) v Rowling above n 93, 327, citing HWR Wade Administrative Law (4ed, Clarendon Press, Oxford, 1977), 628-629.
 Takaro Properties Ltd (in receivership) v Rowling above n 93, 332.
 Takaro Properties Ltd (in receivership) v Rowling above n 93, 334.
 Takaro Properties Ltd (in receivership) v Rowling above n 93, 318.
 Anns above n 35, 751. Lord Wilberforce stated that whether a duty exists can be determined by answering the following two questions: "First ... whether, as between the alleged wrongdoer and the person who has suffered damage there is a sufficient degree of proximity or neighbourhood such that, in the reasonable contemplation of the former, carelessness on his part may be likely to cause damage to the latter... Secondly, ... whether there are considerations which ought to negative, or to reduce or limit the scope of the duty". The test was accepted in Brown v Heathcote County Council  1 NZLR 77, 79.
 Mt Albert Borough v Johnson  NZCA 46;  2 NZLR 234.
 Anns above n 35, 755.
 Dutton v Bognor Regis Urban District Council  1 QB 373.
 Invercargill City Council v Hamlin  3 NZLR 513. This decision was affirmed on appeal by the Privy Council, again without any reference to Dorset Yacht's rule: Invercargill City Council v Hamlin  UKPC 56;  AC 624.
 A-G v Prince  1 NZLR 262.
 B v Attorney-General (1999) 17 FRNZ 694.
 Dalehite v United States  USSC 73; (1953) 346 US 15 [Dalehite].
 Dalehite above n 107, 42.
 Dalehite above n 107, 35-36.
 Dalehite above n 107, 57-58.
 Dalehite above n 107, 60.
 United States v Gaubert  USSC 49; (1991) 499 US 315 [Gaubert].
 Gaubert above n 112, 332.
 Gaubert above n 112, 321.
 Dorset Yacht above n 10, 1031 per Lord Reid; Just above n 15, 705 per Cory J, citing Blessing v United States 447 FS 1160, 1170.
 Heyman above n 12, 469-470.
 J Sopinka "The Liability of Public Authorities: Drawing the Line"  TLR 123, 124 . See also Takaro Properties Ltd (in receivership) v Rowling, above n 93, 333 per Richardson J.
 CL Montesquieu L'esprit des lois (1748), Book XI, cited in PA Joseph Constitutional and Administrative Law in New Zealand (The Law Book Company, Sydney, 1993) 222.
 Joseph, above n 118, 222: "As long as the rule of law prevails, the principle that the executive should be subject to law overrides considerations of a separation of powers".
 Wade, above n 24, 16-21.
 Dorset Yacht above n 10, 1067 per Lord Diplock.
 Above n 42, 19.
 Stovin v Wise above n 51, 958. See also Feldthusen, above n 42, 18-19, 30; Sopinka, above n 117, 124.
 Anns above n 35, 754-755. This view is supported by the fact that his Lordship referred to the public law duty properly to consider whether a power should be exercised after summarising the defendant's argument as follows: "It is said ... that the local authority is under no duty to inspect, and this is used as the foundation for an argument ... that if it need not inspect at all, it cannot be liable for negligent inspection: if it were to be held so liable, so it is said, councils would simply decide against inspection". It is submitted that the words "no duty to inspect" refer to the absence of a public law duty on the defendant council in Anns, and not to the absence of a common law duty of care. If this is correct, then criticisms of Anns which assume that Lord Wilberforce argued that the existence of a public law duty properly to consider whether to inspect ipso facto resulted in a duty of care are misconceived. Such an argument was advanced by Mason J in Heyman, above n 12, 19.
 This "overkill" argument was advanced by Lord Keith in Rowling v Takaro Properties Ltd, above n 43, 502.
 Below part IV B 3.
 Dalehite, above n 107, 36.
 Such a suggestion is made in Just, above n 15, 707, 709. For criticism see MK Woodall "Private Law Liability of Public Authorities for Negligent Inspection and Regulation"  37 McGill LJ 83, 90.
 Heyman above n 12, 469-470.
 Dalehite above n 107, 58.
 Heyman above n 12, 469-470.
 X (Minors) v Bedfordshire County Council above n 14, 737.
 Feldthusen, above n 42, 32
 Just above n 15, 706-707.
 Feldthusen, above n 42, 22.
 Brown above n 16, 16.
 Just above n 15, 709.
 Brown above n 16, 16.
 Stovin v Wise above n 51, 955-956.
 Stovin v Wise above n 51, 951.
 Dalehite above n 107, 335.
 This is conceded by Feldthusen, above n 42, 22.
 S Todd The Law of Torts in New Zealand (1 ed, Brooker's, Wellington, 1991) 240. The author suggests that the burden of proving that a decision fell within the "policy sphere" of the policy/operations rule should rest on the government. See also Feldthusen, above n 42, 32. Compare Anns above n 35, 755.
 Just above n 15, 707.
 Feldthusen, above n 42, 30.
 Feldthusen, above n 42, 30. Perhaps the first part of Feldthusen's argument could be better expressed by saying that the government should not be required to fund a beneficial programme until its marginal benefit equals its marginal cost.
 Rowling v Takaro Properties Ltd above n 43, 500-501. Their Lordships expressly stated that they found it unnecessary to decide whether a duty of care arose.
 Stovin v Wise above n 51, 938.
 The requirements for liability being the existence of a duty of care, its breach by a failure to take reasonable care and damage caused by the breach.
 Evangelical United Brethren Church of Adna v The State of Washington (1966) 47 P 2 d 440 [Evangelical United Brethren Church].
 SH Bailey and MJ Bowman "The Policy/Operational Dichotomy – A Cuckoo in the Nest"  CLJ 430, 435; S Todd (ed) The Law of Torts in New Zealand (2ed, Brooker's, Wellington, 1997), 355-357.
 Rowling v Takaro Properties Ltd above n 44, 501.
 Dalehite above n 107, 32-33 per Reed J, citing Enever v The King (1906) 3 Com LR 969, 988.
 Rowling v Takaro Properties Ltd above n 43, 501.
 Evangelical United Brethren Church above n 150.
 Todd, above n 143, 231, citing KC Davis Administrative Law Treatise (KC Davis Publishing Co, San Diego, 1958) 487.
 Dorset Yacht above n 10, 1026-1033 per Lord Reid, 1066-1070 per Lord Diplock.
 Dorset Yacht above n 10, 1031.
 Above part III A 1.
 Dorset Yacht above n 10, 1067.
 X (Minors) v Bedfordshire County Council above n 14, 737-738.
 X (Minors) v Bedfordshire County Council above n 14, 737.
 Above part III.
 Dorset Yacht above n 10, 1067.
 Above part III B 3. See also K M Hogg "The Liability of a Public Authority for the Failure to Carry Out a Careful Exercise of its Statutory Powers: The Significance of the High Court's Decision in Sutherland Shire Council v Heyman"  MonashULawRw 12; (1991) 17 Mon ULR 285, 293.
 Dorset Yacht above n 10, 1067.
 Fisher v Ruislip-Northwood Urban District Council  1 KB 584, 592-595.
 Dorset Yacht above n 10, 1066-1067.
 Dorset Yacht above n 10, 1067.
 Heyman above n 12, 468.