Alexiou v Australia and New Zealand Banking Group Limited (Subpoena)

This Federal Court decision sits inside a larger dispute between Etienne Alexiou and ANZ. The judgment is not about who should ultimately win that broader case. It is about a late procedural fight over one witness, Mr Nigel Henry Murray Williams.

Mr Williams was not a peripheral figure. He had been ANZ's Group Chief Risk Officer from December 2011 to March 2018. The court said he took part in the decision to claw back Mr Alexiou's deferred but unvested equity awarded in 2012 and 2013. He also had knowledge about ANZ's response to ASIC's investigation into the Bank Bill Swap Rate setting process, ANZ's own investigation, and the circumstances of a 19 November 2014 press release. On any practical view, he was a potentially important witness.

The difficulty was that ANZ could not get his cooperation earlier. After leaving ANZ, Mr Williams joined Commonwealth Bank of Australia as Group Chief Risk Officer in November 2018. While employed there, he maintained that he did not wish to assist ANZ in the litigation because of what he saw as a conflict of interest. ANZ's lawyers therefore could not meet with him.

That explains why ANZ served its lay affidavit evidence in November 2023 without an affidavit from Mr Williams. The judge described that as unsurprising. At that stage, ANZ had to prepare its case on the basis that this witness was not available to assist voluntarily.

Mr Williams' employment at CBA ended in February 2025. Although his departure had been publicly reported in October 2024, ANZ's solicitor gave evidence that he and his team did not know until July 2025 that Mr Williams had ceased employment with CBA. The judge accepted that evidence as plausible in the context of hard-fought litigation and extensive discovery disputes.

Once ANZ's lawyers understood that the position had changed, they moved close to trial. A subpoena to give evidence was issued on 22 September 2025. Mr Alexiou applied on 25 September 2025 to set it aside. ANZ then filed its own interlocutory application on 28 September 2025 seeking leave to rely on an affidavit from Mr Williams dated 25 September 2025. Both applications were argued on 29 September 2025, the day the trial was due to start.

The central question was not whether Mr Williams had useful evidence. The judge accepted that he did. The real issue was whether ANZ should be allowed, so close to trial, to call him by subpoena and rely on his affidavit despite the likely effect on the hearing timetable and the other side's preparation.

That required the court to balance several competing considerations. On one side, the evidence was relevant and ANZ would suffer some prejudice if it could not call a witness involved in important decisions under challenge. On the other side, the trial timetable was already tight, there were many witnesses, and ANZ accepted that Mr Williams would need to be called towards the end of the trial because substantial preparation for cross-examination would be required and the discovery would need to be revisited for that purpose.

The judgment also dealt with an argument about whether ANZ should be criticised for not calling evidence from its own officers about when the bank itself became aware that Mr Williams had left, or was to leave, CBA. Counsel for Mr Alexiou relied on the principle discussed in Tamaya Resources Ltd (in liq) v Deloitte Touche Tohmatsu. Perram J accepted that the principle could apply in this kind of situation, even though the immediate issue was the late calling of a witness rather than some other evidentiary step.

However, the judge declined to criticise ANZ on that basis in the circumstances. The chronology mattered. The application to set aside the subpoena came on very quickly. The public reporting about Mr Williams' departure was only raised in affidavit material filed late on 28 September 2025, shortly before the hearing on 29 September 2025. At the time ANZ's solicitor prepared his affidavit, he did not know that this public information would be relied on. The judge therefore considered it unfair to criticise ANZ for not having adduced evidence from its officers on that point.

Perram J set aside the subpoena to Mr Williams, dismissed ANZ's interlocutory application seeking leave to rely on his affidavit, and otherwise dismissed Mr Alexiou's interlocutory application. In practical terms, ANZ was prevented from calling Mr Williams as a witness in the trial.

The judge's reasoning is important because it was not based on irrelevance or wrongdoing. The court accepted that Mr Williams' evidence was relevant. It also accepted that ANZ was blameless in the circumstances that led to the late attempt to call him. The judge further accepted that ANZ would suffer real prejudice from not being able to use his evidence, especially because he had been involved in the decisions to commence a disciplinary inquiry into Mr Alexiou and to claw back his performance bonuses.

Even so, the application failed because of the risk to the orderly conduct of the trial. The case had originally been listed for six weeks from 15 September 2025, but because of the size of ANZ's discovery the applicant could not be ready and the trial was pushed back two weeks to commence on 29 September 2025. It was then due to run until 14 November 2025, with the week commencing 27 October 2025 not being a sitting week.

The judge said the timetable was tight and there were many witnesses. ANZ accepted that if Mr Williams were called, he would need to be called towards the end of the trial because his cross-examination would require preparation and the voluminous discovery would need to be revisited. ANZ suggested his cross-examination might fit on Friday, 7 November 2025. The judge did not think it was realistic to finish Mr Williams' evidence, hear another witness, Ms Babani, and then move into closing submissions the next day. Nor did the judge accept that any free time earlier in the trial solved the problem, because Mr Williams could only sensibly be called near the end.

The decisive concern was what would happen if the hearing ran over. Perram J said there was a significant risk that allowing Mr Williams to be called would cause the trial to go part-heard after Friday, 14 November 2025. Because of the judge's Full Court commitments, other obligations, leave in December 2025, and a full 2026 calendar, the next available dates to resume a part-heard matter would be at the beginning of 2027. The judge regarded that as very undesirable, particularly in a case where credit was likely to be important.

The court also noted that ANZ had already had to confront the reality that Mr Williams might not give evidence at all. Until July 2025, that had effectively been the bank's position. The judge said the prejudice from losing the witness was real, but it was a kind of prejudice with which ANZ had already reconciled itself. The recent and fortunate change in availability did not justify running the risk of the case becoming part-heard for over a year.

One useful feature of the judgment is the way it separates witness relevance from proof strategy. The court recognised that Mr Williams had direct knowledge and that his intentions could matter, especially given the reference to section 361 of the Fair Work Act 2009 (Cth). But the judge also said his intentions were not provable only through his own testimony.

That point matters for businesses. In many disputes, parties become overly dependent on one former executive or employee. If that person refuses to cooperate, changes jobs, claims a conflict, or becomes available only very late, the case can become vulnerable. Courts may then ask whether the same issues can be addressed through contemporaneous documents, internal communications, board or committee papers, investigation materials, policies, emails, meeting notes and other witnesses.

The judgment also records that no Jones v Dunkel issue would arise against ANZ for not calling Mr Williams, because the course of calling him had been successfully opposed by Mr Alexiou. That is a narrow procedural point, but it reinforces a broader lesson. Businesses should not assume that the absence of a key witness will automatically lead to an adverse inference. The surrounding procedural history matters.

From a practical perspective, this means record-keeping and evidence mapping are not secondary tasks. They are part of risk control. If your business can prove a decision-making process through documents and multiple witnesses, it is less exposed when one important person becomes unavailable or only appears at the last minute.

Businesses should read this case as a trial management decision with a clear operational message. The court was willing to accept three things at once: the witness mattered, the party seeking him was blameless, and the application still had to fail. That combination is what makes the case useful.

In practice, courts expect parties to prepare witness evidence early and to avoid steps that create unfairness or major timetable disruption. If a former executive has been unavailable for years and then suddenly becomes available shortly before trial, that does not mean the court will reopen the evidentiary landscape. The later the step, the more the court will focus on preparation burden, fairness to the other side, and whether the hearing can still finish on time.

This is especially relevant in larger commercial, employment and regulatory disputes where discovery is extensive and witness preparation is document-heavy. A late witness can trigger a chain reaction: fresh conferences, new affidavit material, revisiting discovery, expanded cross-examination preparation, timetable pressure and possible adjournment risk. The court may decide that the system cost is too high even if the evidence itself is useful.

Another practical lesson is to monitor changes in witness availability. ANZ succeeded in persuading the judge that its solicitors did not know until July 2025 that Mr Williams had left CBA, despite public reporting in October 2024. But even with that acceptance, the late step still failed. So businesses should not rely on blamelessness as a safety net. The better approach is to monitor key witness developments actively and move as soon as circumstances change.

For a business owner or in-house team, the safest reading of this case is procedural rather than doctrinal. It does not establish a broad rule about unfair contract terms or the merits of employment clawback decisions. It shows how a court may respond when a party tries to add an important witness too late.

The practical response is to build a witness plan early, revisit it often, and make sure the case can still be run if one important person never appears. That means combining legal strategy with project management: witness availability, document preservation, hearing dates, preparation time and fallback proof all need to be considered together.

This page is based on the Federal Court of Australia decision in Alexiou v Australia and New Zealand Banking Group Limited (Subpoena) [2025] FCA 1238. The orders were made on 29 September 2025 and the reasons were published on 9 October 2025.

The judgment concerns an interlocutory subpoena dispute. It should not be read as deciding the final merits of the underlying proceeding between the parties.