Falling between two stools? The risk in New Zealand’s approach to the assessment of grave risk in Hague Convention cases

This article is co-authored by Kezia Milne (LLB (Hons) student at the University of Canterbury), Alex Summerlee (Partner, Parry Field Lawyers) and Stephen van Bohemen (Barrister). Parry Field Lawyers were the solicitors for the mother “Ms Creswell” and Alex Summerlee and Stephen van Bohemen were counsel for her in the Family Court, High Court and Court of Appeal.

Grave risk defence cases under the Hague Convention require the courts to take a precautionary approach to evidence. Intentionally or unintentionally, the Court of Appeal decision of Roberts v Cresswell means that the court will impose an evidentiary burden on the taking parent to prove a likelihood of harm, rather than adopt the precautionary approach of requiring such a parent to prove a risk which, if true, cannot be discounted.¹ The court’s approach in Roberts is contrary to a line of United Kingdom authorities, contrary to the interests of children, and contrary to the Convention itself.

Background

The Hague Convention on the Civil Aspects of International Child Abduction (the Convention) was adopted by the Hague Conference on Private International Law on 25 October 1980.² The objects of the Convention are to secure the prompt return of children wrongfully removed from or retained in Contracting States, and to ensure respect for rights of custody and access under the law of Contracting States.³

The rationale of the Convention is summarised in its preamble. Contracting States affirm that the interests of children are of “paramount importance in matters relating to their custody”. The Convention desires to “protect children internationally from the harmful effects of their wrongful removal or retention”.⁴

Competing policies

Two policies are built into the Convention:

1. prima facie, if a removal is “wrongful” it is harmful and contrary to children’s welfare and interests and, in the absence of proof of one of the defences, there must be an order for prompt return; and
2. sometimes return will not be in children’s welfare and interests.

In terms of the first policy, the Convention provides for the return of children who have been wrongfully removed from the state of their habitual residence.⁵ The left-behind parent can apply to the Central Authority of the state of habitual residence, who will in turn transmit that application to the state in which they believe the child can be found.⁶ The Central Authority will then facilitate proceedings on behalf of the left-behind parent.⁷

Under the Convention the court must issue a return order unless any of the prescribed exceptions are made out.⁸ In New Zealand the courts must, as far as practicable, give priority to Hague Convention proceedings and there is an expectation that they be resolved within six weeks from the date of application.⁹ To meet those expectations typically these proceedings are dealt with summarily, with limited evidence and no cross-examination.

Articles 12 and 13(b) of the Convention provide:¹⁰

Article 12

Where a child has been wrongfully removed or retained in terms of Article 3 and, at the date of the commencement of the proceedings before the judicial or administrative authority of the Contracting State where the child is, a period of less than one year has elapsed from the date of the wrongful removal or retention, the authority concerned shall order the return of the child forthwith.

The judicial or administrative authority, even where the proceedings have been commenced after the expiration of the period of one year referred to in the preceding paragraph, shall also order the return of the child, unless it is demonstrated that the child is now settled in its new environment. ...

Article 13

Notwithstanding the provisions of the preceding Article, the judicial or administrative authority of the requested State is not bound to order the return of the child if the person, institution or other body which opposes its return establishes that—
...

b) there is a grave risk that his or her return would expose the child to physical or psychological harm or otherwise place the child in an intolerable situation.
...

In New Zealand, the grave risk defence is found in s 106 of the Care of Children Act 2004.

The principle of prompt return is central to the operation of the Convention. The importance, but also its limitation, is reflected in art 12 of the Convention and s 106(1)(a) of the Care of Children Act which provides that if more than a year has elapsed since the wrongful removal or retention the court need not issue a return order if the child is “settled”. If return is not prompt, it becomes difficult to see how it can undo the prima facie harm of wrongful removal or retention. As Hale LJ observed in Re M (Children) (Abduction: Rights of Custody), “the further away one gets from the speedy return envisaged by the Convention, the less weighty those general Convention considerations must be”.¹¹

In view of this, the grave risk defence in art 13 creates a significant challenge for judges. The court must balance proper enquiry into the defence (which, if true, means a child faces a grave risk if returned) on the one hand, against the prompt determination of the proceeding (which, if not done, will defeat the benefits that the Convention offers). This tension creates a practical challenge for courts in the assessment of evidence. The nature of the grave risk defence means that evidence from witnesses of fact, and potentially from experts, is often required, with little (if any) opportunity for it to be challenged.

The stakes are high for the court and the children at the centre of the proceedings: cut corners on evidence assessment, and a child may be unduly exposed to a grave risk of harm; overly engage in evidence assessment, and a matter may take so long to come to trial that the Convention becomes an instrument of harm itself by requiring the return of a child who has become settled into their new environment. This dilemma was identified by the Court of Appeal in LRR v COL, which recognised that:¹²

... there will be cases, of which this is one, where the parties give conflicting evidence about issues that go to the heart of the question that the court must answer. Each challenges the credibility of the other. It may be impossible to resolve these conflicts without oral evidence and cross-examination.

Having identified the tension the Court of Appeal properly asked: “What is the court to do?”¹³ — but it did not provide an answer.

Approach to evidence in Hague proceedings in the United Kingdom where grave risk is raised as a defence

In Re E (Children) (International Abduction),¹⁴ the United Kingdom Supreme Court laid down an approach to evidence in grave risk defence cases, which (correctly, it is submitted) resolved this tension in a manner that is consistent with the Convention’s overarching object of protecting children. The court established a two-step enquiry:¹⁵

• first, where the facts giving rise to the alleged grave risk are disputed the court must consider whether the allegations, if true, would establish the required grave risk — ie, evaluate the risk as alleged and ask whether or not it can be discounted; and
• second, if satisfied that the risk as alleged cannot be discounted, the court must ask whether the child can be protected so as to avoid that grave risk.

How this test is applied in practice is illustrated by the 2021 decision from the English Court of Appeal in Re A (Children)(Abduction: Article 13(b)).¹⁶ This was an appeal from the High Court ordering that the appellant mother’s two children be returned to the United States where they were habitually resident.¹⁷ The mother had alleged that the father had been physically and verbally abusive to her and to her child by a previous marriage (child C).¹⁸ Accordingly, she said she could not return with the children if an order was made.¹⁹ The mother said this gave risk to a grave risk by reason of:²⁰

• the children being separated from her, their primary carer;
• the children being separated from their half-sibling, C, with whom they were close; and
• the children being placed in the care of their father, against whom serious allegations of domestic abuse had been made.

The approach taken by the High Court began in line with the Re E test: the court asked whether the allegations, if true, would amount to a grave risk. It characterised the mother’s allegations as ones of serious domestic violence.²¹ However, then the High Court proceeded to discount and minimise the allegations of abuse by relying on potential mitigating factors or weaknesses in the mother’s allegations, such as:²²

• the fact that none of the mother’s messages to her friends mentioned the abuse;
• the fact that there was “no mention of smacking”; and
• the context of the father being “under considerable stress”.

On appeal, the Court of Appeal said that a summary hearing (which was the case) did not justify such discounting.²³ That is to say, the discounting factors were not ones of a magnitude that enabled the Court to “confidently discount” the allegations. Rather, the court held, in the context of a summary hearing under the Hague Convention, it was wrong to take into account these factors to diminish the seriousness of the allegations.²⁴ The court observed that a judge should not discount allegations of physical or emotional abuse merely because “he or she has doubts as to their validity or cogency”.²⁵ The risks to the children if they were true would be simply too great.

By way of example, in Re A, the allegations were of physical abuse of the children. If ordered to return, the children were to be living with the father and “no protective measures [could] mitigate or address the grave risk arising from that very situation”.²⁶ As the Court of Appeal observed, the children would have been returned with no protections in place, based solely on a determination, at a summary hearing, that the risk was not likely to manifest at the highest level at which it was pleaded.²⁷

The Court of Appeal noted that the approach by the High Court meant that it failed to consider how the children could be protected from the alleged risk, if true.²⁸ It treated the allegations less seriously than they deserved in the context of a summary proceeding. To put it differently, rather than ask whether the allegations, if true, would amount to a grave risk, the High Court had effectively asked whether the allegations, once tested, would create a grave risk — ie it had in effect required the mother to prove a likelihood of harm, not a risk of harm. That is precisely what is not appropriate in a summary hearing.²⁹

The Court of Appeal concluded that the High Court failed to apply the Re E test correctly: namely to ask whether the allegations, if true, would create a grave risk.³⁰ Applying the approach to the specific allegations of abuse in Re A, the Court of Appeal held that “they were allegations of a nature and of sufficient detail and substance to warrant a careful analysis, applying the Re E approach”.³¹

As to the second limb of Re E, on the facts, the court then concluded that there was no proper way to protect the children from the grave risk the allegations posed if true.³² Consequently the appeal was allowed, and the application for the return order dismissed.³³

In overturning the result in the High Court, the Court of Appeal in Re A was explicit about the fundamental problem if a precautionary approach to evidence was not adopted. The Court of Appeal described any approach, other than taking the risk at its highest, as creating an inevitable prospect of the evaluation “falling between two stools”.³⁴ This creates an undue risk to children because:³⁵

... The court’s “process of reasoning” ... would not include either (a) considering the risks to the child or children if the allegations were true; nor (b) confidently discounting the possibility that the allegations gave rise to an Art 13(b) risk. The court would, rather, by adopting something of a middle course, be likely to be distracted from considering the second element of the Re E approach, namely “how the child can be protected against the risk” which the allegations, if true, would potentially establish.

In other words, the danger in adopting a middle course, involving an assessment of the veracity of the allegations against discounting factors, is that a court “will be treating the allegations less seriously than they deserve, if true.³⁶ In doing so, it risks exposing a child to grave harm with insufficient protections in place.

This approach begs the question: is a judge to take any allegation from a taking parent at face value, particularly when such a parent has every incentive to overstate an allegation to justify the unilateral removal of the child from their state of habitual residence?

Simply put, no. Citing with approval of Black LJ in Re K (A Child) (1980 Hague Convention) (Lithuania),³⁷ the Court of Appeal held that the evidential burden on the taking parent is discharged if the evidence before the court is such that the court cannot “confidently ... discount the possibility that the allegations give rise to an Article 13(b) risk”.³⁸ As the language “confidently discount” suggests, it is a low evidential threshold for the taking parent to meet.

In terms of assessing whether an allegation can be confidently discounted or not, the court cited the 1980 Child Abduction Convention—Guide to Good Practice Part VI Article 13(1)(b)(Good Practice Guide), published post-Re E, by the Hague Secretariat. The guide adopts the same approach as Re E and suggests that the court “consider whether the assertions are of such a nature, and of sufficient detail and substance, that they could constitute a grave risk”.³⁹ Allegations will be of sufficient detail and substance if they enable the court to make “reasonable and reasoned” assumptions about the maximum risk they pose to the child if true.⁴⁰ As in Re E, the Good Practice Guide provides that if it is satisfied that the grave risk has been established, the next step is to evaluate evidence from the left-behind parent as to the measures which might protect the child from the risk.⁴¹

Whether following the approach of Black LJ, or the Good Practice Guide, the result is the same: when a court cannot confidently discount the possibility that the allegations give rise to a grave risk, it must take those allegations at their highest, treat them as if they were true, and move to the second limb of the test by asking how the child can be protected from that risk, if true.

This does not mean that the court is making a finding of fact that the risk is, in fact, true (ie, it is not making a finding of likely harm) — rather, it is finding that, in the context of a summary hearing, the risk is there and the policy of the Convention is to now assume it is true so that a child is not exposed to risk. It is a fair application of rules of evidence to a summary procedure that works best for children caught up in the middle of international custody disputes.

The court was blunt about why this precautionary approach to evidence is necessary in Convention proceedings. It emphasised the summary nature of the process, meaning that matters are normally decided on the papers.⁴² The court stressed, citing Re E, that this summary process in a case involving the interests of children obviously creates:⁴³

... a tension between the inability of the court to resolve factual disputes between the parties and the risks that the child will face if the allegations are in fact true.

In light of this tension, a precautionary approach is crucial, appropriate and, most importantly, in the best interests of the child. This is because this approach shifts the burden to the left-behind parent to establish that protective measures are in place so that the court can be satisfied that the risk is obviated. In other words, this approach to evidence encourages measures to be put in place to ensure children are not exposed to further harm through the Convention’s otherwise blunt return mechanism. It places an evidential burden of proving safety measures in the country of habitual residence on the left-behind parent, the party best able to prove those measures.

Once the burden of proof is shifted to the left-behind parent, that parent must satisfy the court that measures adequate to protect the child from the identified grave risk are in place. In terms of what constitutes adequate measures, the court in Re E noted “appropriate protective measures and their efficacy will obviously vary from case to case and from country to country”.⁴⁴ In some cases that evidence could be as simple as what the actual living arrangements will be once the children return. In other cases, expert evidence might be required as to what protective measures are necessary. In Re E the trial judge accepted evidence about post-return living arrangements and from a psychiatrist. The court found that the children could be protected from the grave risk posed to them by their mother’s adjustment disorder (and the chance it would render her suicidal if she returned to Norway with them) by:⁴⁵

• the mother receiving ongoing psychological interventions, such as counselling or cognitive behavioural therapy (CBT);
• a court order preventing the father from knowing her address and from physically approaching her; and
• support from close family, in particular her mother, and statutory agencies.

In that case, the trial judge was satisfied with these protective measures on the basis that such measures could, and would be, put in place in Norway.⁴⁶ The New Zealand Court of Appeal in LRR has taken a consistent approach, requiring effective protective measures.⁴⁷ In LRR, the court noted that protective measures might take the form of orders made by the courts of the requesting state, or undertakings made by the left-behind parent.⁴⁸ However, each case must be considered on its own facts — so where, for example, a parent has previously breached court orders it is inappropriate to assume such orders can mitigate a grave risk.⁴⁹ In LRR, the grave risk that return to Tasmania would cause the mother’s mental health to relapse could not be obviated by family violence orders in her favour where the father had a history of breaching those orders.⁵⁰

Policy rationale

What Re E and Re A demonstrate is that in summary hearings under the Hague Convention, the evaluation of evidence is focused on the potential risk as alleged, to the maximum degree at which it cannot be confidently discounted. In these cases the evaluation of risk is not focused (as is usual for non-summary trials) on the most probable outcome that arises from the allegation or the likelihood of harm, that would be derived by taking into account, say, unexplained gaps in the evidence.

The rationale is simple: in a summary hearing (which is necessarily so to achieve promptness), there is insufficient opportunity to give the parties adequate opportunity to bring evidence to the standard required by an ordinary civil trial. This is because this is not an ordinary civil trial. The usual rigours of trial are not available: larger timeframes to assemble evidence, cross-examination of witnesses, and the like, are luxuries which are not generally available to the parties.

The solution? A redefinition of the evidential burden on the taking parent: the taking parent need not prove, on balance, that the risk will eventuate. Rather, the taking parent must prove, on balance, that there is a grave risk that the court “cannot confidently discount” or, in other words, that there is a risk that the court is unable to put it out of its mind, or dismiss out of hand. As the New Zealand Court of Appeal said in LRR:⁵¹

[108] The apparent tension between speed and informed decision-making is mitigated to some extent if one bears in mind that in the context of s 106(1)(c) the Court is concerned with risks, not with certainties or even probabilities.

This policy is prudent because it achieves an imperative: an allegation of a grave risk which is not spurious should be taken seriously, and the court should be slow to expose a child to such a risk. Conversely, if an allegation can be discounted and deemed spurious it will not provide the foundation for a finding of grave risk. This interpretation achieves safer returns for children because, once that burden is discharged, that is not the end of the matter. A new burden arises on the left-behind parent to lead evidence of protective measures that obviate the risk. Any other approach would mean the court was abdicating its responsibility to children. As the Court of Appeal of England and Wales described it, it would be falling between two stools.⁵²

Another way to think of this is to take the risks that comes with not wearing a seatbelt. The risk, at its highest, which cannot be confidently discounted, is death. However, death is not the most likely outcome of driving without a seatbelt. Less serious outcomes are more probable, ranging from serious injury, minor injury, to no incident occasioning any injury at all. But when assessing public safety measures for cars, Parliament took the risk at its highest, and required all occupants to wear seatbelts.

The policy for the Hague Convention is the same: when it comes to whether, in a summary procedure, a child should be returned, the court should take the alleged risk that cannot be confidently discounted at its highest. If that is a grave risk, then the court should act accordingly and not order a return unless satisfied that protective measures are in place that protect the child from that risk.

This approach to evidence in Hague Convention proceedings creates a challenge. Judges are accustomed to weighing evidence on the balance of probabilities and determining likely outcomes. But, as far as the grave risk defence is concerned, that is not the correct approach.

Hague proceedings in New Zealand

It is settled law that signatory states should endeavour to apply international treaties uniformly.⁵³ One, therefore, would expect consistency between the New Zealand approach to evidence in grave risk cases, and the approach in Re E. Until very recently, this was the position.

The two-step test in Re E was expressly approved by the New Zealand Court of Appeal in LRR.⁵⁴ This decision represented a major development in New Zealand’s Hague jurisprudence. In LRR the Court of Appeal recognised that courts must be alert to not applying the wrong burden of proof, at the expense of children who are subject to (properly) rushed court procedures.⁵⁵

After LRR came the Re A decision, which put the Re E principles into practice. One therefore would have expected the logic of Re A to be adopted in New Zealand.

However, the recent decision of the New Zealand Court of Appeal in Roberts v Cresswell indicates a divergence in approach to the assessment of evidence in grave risk cases between the New Zealand courts and the courts in the United Kingdom.⁵⁶ In our view, while the court in Roberts professed itself to be following Re A, its actual approach was that of the High Court in Re A, which was overturned by the English Court of Appeal.

In Roberts, the respondent mother had removed her two daughters, Amelia and Brigitte, from France, their country of habitual residence.⁵⁷ The father made an application under s 105 of the Care of Children Act, and proceedings were bought in the New Zealand Family Court.⁵⁸ A return order was issued in the Family Court, but quashed in the High Court who found that return would expose the children to a grave risk.⁵⁹ Leave was granted to the father to appeal.

In the Court of Appeal (as in the courts below) the mother relied on several grounds to establish that return would expose the children to a grave risk within the meaning of art 13b and s 106(c). These grounds included the allegation that she was suffering from post-traumatic stress disorder (PTSD) as a result of family violence by the father, and that a return to France would trigger the PTSD resulting in grave risk to the girls due to its effect on her parenting.⁶⁰

The Court of Appeal began by noting that it apprehended no material difference between the LRR decision and the Re A approach to evidence.⁶¹ It commenced step one, considering whether there was an allegation of a grave risk that it could not put out of its mind. Adopting the language of Re A, the court made findings that:⁶²

Adopting the approach outlined in LRR v COL, we consider that the assertions made by the mother in relation to family violence are of such a nature and of sufficient detail and substance that they cannot be discounted.

...

We cannot confidently discount the possibility that the mother suffers from PTSD, or that a return to France will involve psychosocial stressors that will trigger a recurrence of PTSD.

Thus, on the Re A approach to evidence, the court’s next task should have been to ask whether those risks, taken at its highest, if true, were sufficiently serious to amount to a grave risk. If it did, the court would then move to the second step in the Re E test — and ask if and how the girls could be protected from the risk. However, the court deviated from the Re A approach. It did not ask if those risks amounted to a grave risk or, if so, what protective measures could be put in place. Instead, it embarked on a probability assessment of the risk occurring, with resultant actual harm to the children. It assessed the most probable effects of the mother returning to France on her mental health and concluded that a PTSD breakdown would likely not occur.

The court accepted that return to France would involve “significant stress for the mother, and a level of risk to her mental well-being”⁶³ and accepted that “if she is under stress, that will have some adverse effects on the children for whom she is the primary attachment and primary carer”.⁶⁴ However, just as was done by the High Court in Re A, the court went on to discount the likelihood of the risk by applying countervailing factors:

• Against the risk of a significant deterioration in her mental health, the court applied a discount:⁶⁵
  
  Even when the stressors on the mother were at their highest, while she was living with the father, she continued to be an effective and competent parent.
• Against the risk of financial hardship impacting on her mental health, which it accepted as a factor which could exacerbate PTSD, the court applied a discount:⁶⁶
  
  ... She will have some financial support from the father. There is evidence that she will also be entitled to certain welfare benefits.
• Against the risk of being unable to access suitable counselling services in a language she spoke, to reduce the risk of deterioration in her mental health, the court applied a discount:⁶⁷
  
  ... [the] suggestion [that counselling services in English] will not be available in the area in which she would be living seems speculative. And one option would be for her to continue counselling online with her existing counsellor in New Zealand.
• Against the risk presented by a recurrence in PTSD, the court applied a discount on its likely severity:⁶⁸
  
  ... [The] more likely outcome if the mother’s PTSD deteriorates [would be an] adverse impact on the quality and sensitivity of her parenting. That would be unfortunate. But it would not amount to an intolerable situation for the children.

The court concluded that the possibility of the PTSD break-down was “in our view, too speculative to be described as a grave risk”.⁶⁹ Notwithstanding its decision that the mother’s allegations could not be discounted the court failed to follow its own guidance from LRR and to remind itself it was “concerned with risks, not with certainties or even probabilities”.⁷⁰ The court never assessed whether the allegation, at its highest, was a grave risk. It fell between two stools: the court assessed what it considered to be a more probable iteration of the risk, rather than the alleged risk at its highest. In doing so, the court declined to describe it as “grave” by reason of the risk, at its highest, being less probable.⁷¹

Probability of the risk occurring (other than where the allegation can be dismissed out of hand) is not the test. The test in Re A (which the Court of Appeal said it was applying) is whether the risk can be confidently discounted or not. On that question, the court had already concluded that it could not be confidently discounted. Thus, following Re A, the next step should have been to consider whether the PTSD breakdown would pose a grave risk to the children — not to consider the most likely impact that a return to France would have on the mother’s mental health — and then to consider if any protections could be put in place to mitigate the risk.⁷²

The Court of Appeal in Roberts committed the error that its English counterpart had been at pains to warn against. It fell “between two stools”. It took the allegations less seriously than they merited, if true. Consequently, there was no discussion of how the children could be protected from the grave risk of the mother’s PTSD breakdown if it eventuated. The result of this is that the leading decision from the New Zealand courts as to how to assess evidence in grave risk defence cases has set the stage for a deviation from the approach to evidence in the United Kingdom. To add to the confusion, the New Zealand judgment expresses itself as following the approach of the English Court of Appeal in Re A — when it did the opposite.

Conclusion

The Supreme Court declined leave for Creswell v Roberts to be appealed.⁷³ This leaves first instance courts in a bind: the Court of Appeal has both said that Re A is correct, and then applied a test which is not the Re A test. The correct approach to evidence is therefore likely to be back before the Court of Appeal for clarification.

It is submitted that this confusion should be resolved in favour of the precautionary approach set out in Re A. Not only is this approach sound as a matter of policy for children, it better reflects the Convention’s policy objectives. The probability-based approach taken in Roberts fails to protect the interests of children subject to decisions made in summary hearings. That is not in the interests of children as a class, or of any specific child, in any case involving allegations of grave risk.

The article was written for New Zealand Family Law Journal. To enquire about New Zealand Family Law Journal subscription, submit the form below: