Federal Court enlivens transfer pricing "reconstruction" powers: Commissioner wins SingTel case

Written by Jerome Tse, Calum Sargeant

The Federal Court on Friday handed down its decision in Singapore Telecom Australia Investments Pty Ltd v Commissioner of Taxation [2021] FCA 1597 (SingTel), which adds to the evolving body of case law considering Australia’s transfer pricing rules.

The single judge decision (Moshinsky J) found in favour of the Commissioner, reversing the trend in the earlier decisions in Glencore, which had previously clarified and confined the scope of the Commissioner’s reconstruction powers under the transfer pricing rules. The decision in SingTel, like the earlier decisions in Chevron, highlights the evidentiary difficulties that taxpayers may face in discharging their onus of proof in transfer pricing cases.

The dispute between the Commissioner and SingTel concerned the application of former Division 13 of the Income Tax Assessment Act 1936 (Division 13) and Subdivision 815-A of the Income Tax Assessment Act 1997 (Subdivision 815-A) to the terms and conditions of related-party borrowings. Although these provisions no longer apply, many legacy disputes under these provisions remain to be resolved, and the principles developed in relation to Division 13 and Subdivision 815-A are likely to be of continuing relevance to the interpretation and application of the current transfer pricing laws in Subdivision 815-B of the Income Tax Assessment Act 1997 (a provision which is yet to be considered by any court).

Background

In June 2002, Singapore Telecom Australia Investments Pty Ltd (STAI) acquired 100% of the issued capital in Singtel Optus Pty Ltd (Optus) for $14.2 billion from its immediate parent company, Singtel Australia Investment Ltd (SAI), a Singapore tax resident. The acquisition was funded by STAI through the issuance of ordinary shares as well as the issuance of $5.2 billion of loan notes to SAI (Loan Notes). The acquisition followed SAI’s earlier acquisition of Optus from third party shareholders in 2001.

The Loan Notes were denominated in AUD and were issued in June 2002 for a tenor of approximately 10 years, bearing interest at a floating rate of 1 year BBSW + 1.00%. The Loan Notes also provided for a withholding tax “gross-up” (i.e. STAI was required to gross-up the interest payable under the Loan Notes so as to indemnify the noteholder for interest withholding tax payable to the Commissioner at the rate of 10%).

The Loan Notes were amended in March 2003, purportedly with retrospective effect from the issuance date (2003 Amendment). The effect of the 2003 Amendment was to make the accrual and payment of interest (including the interest that had notionally accrued since June 2002) contingent on certain financial benchmarks being met, after which an interest rate premium of 4.552% would apply to the face value of the Loan Notes to “recoup” the interest that would otherwise have accrued during the interest-free period. STAI contended that the 2003 Amendment was economically equivalent to a loan that provided for the deferral and capitalisation of interest, with a credit spread to maturity of 1.44% based on an estimate that the financial benchmarks would be met within 3.5 years of issuance (in the event, the benchmark was reached within 3 years). The reason for the 2003 Amendment was a capital expenditure programme implemented by the Optus business, which meant that the business was expected to have insufficient cash flows in the short to medium term to service the interest expense).

The Loan Notes were further amended in March 2009 by replacing the 1 year BBSW floating rate with a fixed rate of 6.835% (2009 Amendment). The effective rate after the 1.00% margin, the 4.552% “interest rate premium” and the withholding tax gross-up was 13.2575%.

The terms of the Loan Notes (as amended) resulted in a total interest expense to maturity of $4.9 billion (giving rise to deductions of the same amount).

The Commissioner issued transfer pricing determinations to STAI under Division 13 and Subdivision 815-A denying approximately $895 million of the interest deductions claimed by STAI in the 2010 to 2013 income years, on the basis that the amounts of interest payable by STAI exceeded the amounts that might be expected to have been paid if the parties had been dealing at arm’s length. The Commissioner issued amended assessments to give effect to the determinations (Amended Assessments) and STAI objected to those Amended Assessments.

The transfer pricing hypotheticals

The transfer pricing provisions in Division 13 and Subdivision 815-A require the consideration of “hypotheticals” to assess whether amounts paid were arm’s length (and to assess what transfer pricing adjustments should be made if they are not).

In the case of Division 13, the relevant hypothetical is “the consideration that might reasonably be expected to have been given or agreed to be given … if the property [in this case, rights under the Loan Notes] had been acquired under an agreement between independent parties dealing at arm’s length with each other”.
In the case of Subdivision 815-A (which relevantly adopts the language of the applicable tax treaty, in this case the Singapore DTA), the relevant test is whether “conditions operate … which differ from those which might be expected to operate between independent enterprises dealing wholly independently with one another”. If so, then it is necessary to work out whether any profits would have accrued to STAI “but for” those non-arm’s length conditions.

The dispute was whether the amounts of interest payable by STAI under the Loan Notes exceeded the “consideration that might reasonably be expected to have been given” (in the case of Division 13) or the amount payable had the “conditions which might be expected to operate” replaced the actual conditions that operated in respect of the Loan Notes (in the case of Subdivision 815-A).

The different positions advanced by STAI and the Commissioner were as follows:

STAI acknowledged that the actual structure of the Loan Notes taking into account the 2003 Amendment (i.e. the interest-free period followed by an interest rate premium) was not typically observed in the markets, but was economically equivalent to a traditional debt capital market instrument with deferred and capitalised interest. On that basis, STAI’s debt capital markets expert (Mr Charles Chigas) determined that such an instrument (i.e. a 10 year loan note with deferral and capitalisation of interest) issued in the debt capital markets between arm’s length parties might be expected to have a credit spread of 4.00%. This is compared with Mr Chigas’ calculation of the effective credit spread to maturity on the Loan Notes of 1.44% plus the withholding tax gross-up. The interest deducted by STAI was therefore, according to Mr Chigas, less than the arm’s length amount of interest.
The position advanced by the Commissioner in his Economist Practice (EP) report (and which was relied on in making the transfer pricing determinations and Amended Assessments) also determined that STAI might be expected to have issued a 10 year loan note with deferral and capitalisation of interest. However, the EP report determined that such an instrument might be expected to have a credit spread of 1.00% over the AUD swap rate, and would not have included a withholding tax gross-up. The EP report also determined that the 2003 Amendment and the 2009 Amendment would not be expected to have occurred.
The Commissioner advanced a different position in the appeal (Commissioner’s Hypothesis). He relied on the report of his debt capital markets expert (Mr Gregory Johnson). Mr Johnson hypothesised that an arm’s length financing structure would have involved STAI issuing a bridge loan for the first 9 months of the term (with interest payable at termination), to be refinanced by two loans with a term of 9 years and 3 months – one loan of $5.2 billion to refinance the principal amount and another loan of $1.5 billion to cover interest that would otherwise be deferred (including interest accrued under the bridge loan and interest expected to accrue and be deferred under the refinancing). Mr Johnson also hypothesised that the loans would be 50% fixed and 50% floating, with no withholding tax gross-up. The interest rate payable under the Commissioner’s Hypothesis was based on a credit rating uplifted by 6-8 notches from STAI’s standalone credit rating, to reflect implicit support from STAI’s parent and indirectly from the government of Singapore. Alternatively, the Commissioner posited that SingTel (the ultimate parent of STAI) would reasonably be expected to have provided a parental guarantee.

The key issues in dispute

On appeal, STAI challenged the Commissioner’s position on several legal and factual grounds:

The relevant statutory tests require a consideration of what “might reasonably be expected” rather than what “would have” or “must have” happened. As such, the tests contemplate a range of arm’s length outcomes based on rational commercial behaviour and judgment. On that basis, STAI argued that the Loan Notes (or an equivalent instrument with traditional deferral and capitalisation features) were within that arm’s length range.
Relying on the decision in Chevron and Glencore, the hypothetical transaction must remain as close as possible to the actual transaction (subject to commercial rationality). STAI argued that its hypothesis of a 10 year debt capital market issuance was closer to the actual transaction than the Commissioner’s Hypothesis which posited a bridge loan refinanced into two separate loans 9 months later.
The Commissioner’s Hypothesis should be rejected because:
1. STAI might reasonably be expected to have borrowed on terms which allowed for the deferral and capitalisation of interest given the capital expenditure programme. The hypothetical bridge loan did not allow for interest deferral.
2. An independent party might not be expected to assume the risk and uncertainty of a bridge loan.
3. The quantum of the hypothesised second loan ($1.5 billion) to cover interest could only be determined with hindsight as the deferral period was based on benchmarks being met.
The existence of implicit support was a product of the non-arm’s length relationship between STAI and the lender (its parent) and so should be excluded from the statutory hypothesis. Even if implicit support was relevant, STAI contended that an uplift of 6-8 notches was excessive. Following the opinions expressed in the Chevron case, STAI argued that implicit support had very little real-world impact on pricing by lenders.
STAI rejected the Commissioner’s factual assertion that a borrower might not reasonably be expected to have made a decision to fix the interest rate on the Loan Notes in October 2008 (with effect from March 2009), during the unfolding global financial crisis at that time.
The Commissioner contended that a withholding tax gross-up might not reasonably be expected between independent parties, because independent parties might be expected to have structured the arrangement so as to attract the withholding tax exemption in section 128F (public offer test). STAI disputed this argument because:
1. STAI’s expert gave evidence that loan agreements typically have such a gross-up clause.
2. It is impermissible for the hypothesis to change the characteristics of the lender or substitute multiple lenders (so as to satisfy the public offer test).
3. Even if it is permissible to hypothesise that the public offer test would be satisfied, the Commissioner should make a consequential adjustment to refund the withholding tax actually paid on the Loan Notes.

The court’s decision

The Court ultimately found in favour of the Commissioner on the basis that STAI had failed to demonstrate that the Amended Assessments were excessive.

The decision of the Court, however, was based on a different hypothesis to the primary cases advanced by the parties in argument. In summary:

Moshinsky J disapproved of STAI’s hypothesis of a debt capital market bond issuance on the basis that it departed too far from the actual transaction between STAI and SAI who were relevantly the vendor and purchaser of shares in connection with the Loan Note. In particular, the adoption of the financial benchmarks, “interest free” and “premium interest rate” period concepts were not consistent with a DCM issuance.
Rather, independent parties in the positions of STAI and SAI might have been expected to agree in June 2002 to issue loan notes with an interest rate of 1 year BBSW + 1.00% with interest deferral and capitalisation and a withholding tax gross-up, subject to a parent company guarantee. In this regard:
1. The arm’s length condition relating to the interest rate was consistent with the actual interest rate on the Loan Notes originally issued in June 2002. Mr Chigas gave evidence that an arm’s length rate based on his debt capital markets analysis would have been 4.00%, but this was rejected by the Court because it was based on STAI’s credit rating without a parental guarantee and took into account the 2003 Amendment (both of which were found not to be arm’s length conditions, as explained below).
2. The conclusion that an independent party in the position of STAI might reasonably be expected to have borrowed subject to a parental guarantee was informed by a number of factual considerations. These included:
  1. the size of the borrowing;
  2. the fact that a lender providing vendor financing would have expected security of some description;
  3. the material difference between the guaranteed and unguaranteed credit ratings (resulting in a large differential in interest rate that would have made an unguaranteed borrowing unappealing to STAI and its hypothetical parent); and
  4. the fact that SingTel had in fact provided a guarantee of its subsidiary’s bank facility in May 2002.
3. The Court accepted that the withholding tax gross-up was an arm’s length condition as it “appears to be common in international borrowings”.
4. The Court accepted that independent parties might reasonably be expected to have deferred and capitalised interest. This was based on the projected cashflow position of Optus.
The Court preferred the evidence of STAI’s credit rating expert (Mr William Chambers), who opined that a ratings uplift of 2-3 notches for implicit support was more appropriate than the 6-8 notches determined by the Commissioner’s expert (Mr Robert Weiss). However, this issue was not of particular relevance given that the Court determined that the relevant hypothesis would have included a parental guarantee (which would have equalised the credit rating with that of SingTel).
Moreover, independent parties would not have agreed to the 2003 Amendment, which introduced the financial benchmarks, “interest free” and “premium interest rate” period concepts. The Court was of the view that the 2003 Amendment made the accrual of interest contingent on financial benchmarks and exposed both parties to significant risks – (i) in the case of SAI, the risk that the benchmarks would never be met and that it would receive no interest; and (ii) in the case STAI, the risk that if the benchmarks were met earlier than the estimated 3.5 years (as in fact occurred), STAI would be required to pay the premium interest rate for a longer period of time. Although it was suggested that the 2003 Amendment was needed to address the cashflow position of Optus, this was already addressed by the original terms of the Loan Notes which allowed STAI to defer the payment of interest. His Honour said that it was “very difficult to see why independent enterprises would agree to such terms” and that there was no commercial rationale for them.
Additionally, independent parties would not have agreed to the 2009 Amendment. The Court found that there was no commercial rationale to fix interest rates at that time, based on the fact that STAI had committed funds under the Loan Notes for a further 3.5 years (so there was no immediate need to secure refinancing), interest rates were dropping at the time and the fact that STAI’s subsidiary Optus had increased its exposure to external floating rate borrowings around the same time.
As such, the key differences impacting price between the Court’s hypothesis and STAI’s position were: (i) the use of a traditional deferral and capitalisation mechanism rather that the “interest-free” and “premium” periods per the 2003 Amendment; (ii) the existence of a parental guarantee; and (iii) the fact that a floating rate of BBSW + 1.00% operated rather than a fixed rate of interest from March 2009 until maturity.

In reaching its conclusion, the Court primarily considered the operation of Subdivision 815-A as the parties had focussed on those provisions in their submissions. Moshinsky J gave extensive consideration to and sought to discern and apply the principles emerging from the Chevron and Glencore cases. Importantly, the Court expressed the following opinions:

The hypothetical parties must generally have the characteristics and attributes of the actual parties to the transaction. In this regard, Moshinsky J held that the hypothetical independent borrower was a member of a multinational group like the SingTel group. It follows that such an approach to the attributes of the parties permits the imputation of a parental guarantee, where the circumstances dictate that such a guarantee would reasonably be expected between independent parties.
However, in finding that a parental guarantee would reasonably have been expected, his Honour did not place weight on the group policies of SingTel as the Full Federal Court had done in Chevron. Group policy was not, according to the Court, essential to such an hypothesis. Rather, his Honour preferred to rely on the objective factors going to the “logic of the situation”, including the actions of the SingTel group in practice.
In Chevron, Pagone J had alluded to the possibility of the arm’s length conditions including the payment of a guarantee fee, where a parental guarantee formed part of the arm’s length hypothesis. The issue was expressly considered by Moshinsky J but ultimately rejected on the basis of there being no probative evidence to support the inclusion of a guarantee fee. Although Mr Chigas gave evidence that independent parties might reasonably be expected to have paid a guarantee fee, that evidence was “very general” and “speculative”. Moreover, there was no evidence of SingTel charging a guarantee fee to its subsidiaries in practice.
His Honour confirmed that the identification of the “conditions” to which Subdivision 815-A applies “permits a broad and wide-ranging enquiry” and is “unconfined”. In this regard, his Honour identified non-arm’s length conditions that operated between STAI and SAI which included some of the terms of the Loan Notes as well as features of the relationship between STAI and SAI (such as their common ownership and directorship).
Although weight must be given to the actual transaction entered into, that may be altered if necessary to permit the hypothesis of dealings between independent parties to operate. However, the hypothetical should be as close as possible to the actual transaction.
The extent to which the Commissioner may substitute the actual terms of the transaction remained an open issue after Chevron and Glencore. Whilst the Full Court in Chevron arguably gave licence to the substitution of non-price terms (such as the imputation of a parental guarantee), the majority decision in Glencore expressed hesitation and left open the issue of whether the Commissioner could substitute conditions that bear only indirectly on price. As the majority in Glencore had left the issue open, Moshinsky J found that it was open to him to agree (and did agree) with the opinion of Thawley J in Glencore that Subdivision 815-A permitted the substitution of non-price conditions.

Observations and implications

Reconstruction

A key area of concern for taxpayers has been the extent to which the transfer pricing rules allow the Commissioner to change “conditions” of a related-party transaction other than price. The concern is that, if the Commissioner is able to change such conditions, the form and structure of the transaction could be reconstructed into something entirely different to the actual transaction entered into by the taxpayer. The Commissioner would be able to substitute key terms of a transaction (such as the quantum or tenor) to conform the transaction to the Commissioner’s own assessment of the risk appetite of hypothetical parties, rather than the risk profile actually assumed by the parties to the transaction.

The decision in SingTel appears to demonstrate cautious endorsement of the “reconstruction” principle subject to limitations. The majority judgment of Middleton and Steward JJ in Glencore accepted that the transfer pricing rules in Division 13 and Subdivision 815-A allowed the substitution of “price-like” conditions – such as elements of a formula used to calculate price or remuneration – but left open the question of whether other conditions, which bear only indirectly on price, could permissibly be substituted under those rules. Thawley J, in his separate reasons, determined that he could and, in SingTel, Moshinsky J agreed with the decision of Thawley J in Glencore. The extent to which the law permits reconstruction of non-price terms now awaits consideration by the Full Federal Court (or High Court).

However, the actual application of the reconstruction principle in SingTel demonstrates that it may have quite limited scope. The Court largely accepted that the terms of the Loan Notes, as originally agreed between the parties, were within the arm’s length range of conditions that might reasonably be expected between independent parties. It endorsed the principle emerging from Chevron and Glencore that the arm’s length conditions must comport as closely as possible with the actual transactions. In particular, the Court accepted relevant structural features of the Loan Notes, including the 10 year tenor and the deferral and capitalisation of interest. The Court did not adopt key elements of the Commissioner’s Hypothesis, which included the substitution of the actual 10 year borrowing with a shorter bridge loan refinanced into two separate loans of 9 years and 3 months, even though the Commissioner’s expert (Mr Johnson) opined that such a borrowing structure would have resulted in a lower overall cost of funds.

The Court did, however, reconstruct the actual transaction by determining that the 2003 Amendment and the 2009 Amendment might not reasonably be expected to have been agreed between independent parties. In both cases, the Court made express findings that such amendments lacked commercial rationale. Although no express reference was made to the OECD Guidelines in this regard, the outcome is consistent with the OECD’s guidance that reconstruction is permitted only in “exceptional circumstances” where the actual transaction agreed between the parties is “commercially irrational”.

Taxpayers may take some comfort that they may be able to avoid a reconstruction outcome provided they can demonstrate that the features of their transaction have commercial rationale. This would appear to respect a broad range of risk appetites (which may impact price), provided that the risks assumed by the parties are within the range of what is commercially observed in the market or it can otherwise be shown that the risk profile is rational. In the case of the 2003 Amendment, the Court disapproved of the financial benchmark concept because it exposed both parties to significant risks and there was no evidence that commercially rational borrowers or lenders would have accepted those risks.

Parental guarantees and the relevance of group policy

However, the judgment in SingTel demonstrates that Courts may take a more relaxed approach to imputing a parental guarantee as part of the arm’s length conditions in related party borrowings where the borrower is a member of a multinational group.

Moshinsky J did not make an express finding that it would have been commercially irrational for an independent party in the position of STAI to have borrowed without a parental guarantee, however he determined that the “logic of the situation” dictated that a parental guarantee would be reasonably expected.

In Chevron, the Full Federal Court imputed a guarantee, largely on the basis that to do so was in conformity with the policy of the Chevron group to borrow in such a way as to achieve the lowest cost of funds for the group overall. The Court in SingTel avoided reliance on group policy, preferring objective evidence including the significant difference in interest rate between guaranteed and unguaranteed borrowings and the actual practice of the SingTel group. The approach of the Court is arguably more consistent with the comments of the Full Federal Court in Glencore (which were quoted by Moshinsky J) which eschewed reliance on features such as policies relating to intragroup dealings (on the basis that they are the product of the non-arm’s length relation between group members).

As such, SingTel leaves open the possibility that taxpayers could establish the commercial rationality of unguaranteed borrowing – for example through probative expert evidence or actual examples of borrowers and lenders transacting on an unguaranteed basis in comparable circumstances.

Guarantee fees

The decision in SingTel demonstrates the difficulties that taxpayers are likely to face in arguing that the arm’s length conditions would have included the payment of a guarantee fee (for example, where the Commissioner imputes a parental guarantee as part of the arm’s length conditions). Whilst the Court appears to have accepted the theoretical possibility of a guarantee fee, the taxpayer was subjected to a high burden of proof in establishing that such a fee might reasonably be expected or the quantum of such a fee. Mr Chigas’ evidence in this regard was rejected as “very general” and “speculative”.

There remains a legal question, not addressed in SingTel, whether an imputed guarantee fee must itself be subjected to the transfer pricing rules – on the basis that it is a cross-border dealing between related parties – such that an arm’s length price must be determined on the hypothesis of a guarantee between independent parties (for example, guarantees provided by monoline insurers). This may need to be addressed in future cases.

Discharging the evidentiary burden

SingTel once again highlights the evidentiary difficulties that taxpayers may face in discharging their onus of proof in transfer pricing cases. That is particularly the case where the Court determines arm’s length conditions that diverge from the primary cases of both the taxpayer and the Commissioner, or where the Court’s hypothesis is based on a selection of conditions from each. In such a case, the Court may ultimately substitute conditions that did not attract significant attention in the submissions or arguments of either party to the dispute.

In the present case, the Court determined that the taxpayer might be expected to have borrowed subject to a parental guarantee. On that basis, the Court determined that a credit spread of 1.00% was arm’s length, having regard to the actual terms of the Loan Notes and the contemporaneous correspondence of the taxpayer. This evidence, although subjective, was preferred to the expert evidence led by the taxpayer’s expert (Mr Chigas) that an arm’s length credit spread on a guaranteed borrowing might be expected to have been in the range of 2.05% to 2.15% (which, if accepted, would have significantly reduced the transfer pricing benefit obtained by STAI). Mr Chigas’ evidence of the guaranteed credit spread was rejected because it was “based on the actual [Loan Notes], rather than on an hypothesis which excludes from consideration the conditions [arising from the 2003 Amendment” and “it is unclear to what extent this affects the credit spread he calculated”. In addition, the Court rejected Mr Chigas’ assumption that SingTel would have been placed on negative watch.

In addition, and as noted above, the Court imposed a high evidentiary burden in relation to the arm’s length nature and quantum of a hypothetical guarantee fee.

These evidentiary “defects” demonstrate the need for taxpayers to present expert evidence that is robust and responsive to a wide range of permutations, given the wide ambit the Court has in determining the arm’s length conditions. In practice, this may place an unreasonably onerous burden on taxpayers in transfer pricing cases. In Glencore, Middleton and Steward JJ said:

Whilst the onus remains on the taxpayer to discharge its onus of proof of demonstrating excessiveness in the amended assessments, one should not apply Div. 13, or indeed Subdiv. 815-A, narrowly. … The Court should acknowledge, and take into account, the practical difficulties faced by both the taxpayer and the Commissioner in finding evidence that grounds what is sufficiently reliable, or which demonstrates that something is insufficiently reliable. The answer is not always to be found in overly lengthy and complex expert reports. Common sense is required.

Taxpayers should also consider that impressionistic features will invariably influence the Court and evidentiary burden that it may bring to bear. For example, in the present case, the 2003 Amendment and the 2009 Amendment gave rise to an overall interest rate in excess of 13%. In addition, there was some allusion in the judgment to tax-related purposes for some of the features of the Loan Notes – i.e. the Commissioner implied that the 2003 Amendment was designed to ensure that interest did not accrue for withholding tax purposes during the deferral period, and that the accrual of interest (giving rise to deductions) was designed to align with the time when the STAI group’s carry forward losses were expected to have been fully utilised. Whilst the Court correctly acknowledged that these factors were strictly irrelevant to the transfer pricing exercise, the inference that the conditions of related-party dealings are influenced by tax considerations unique to the taxpayer may strengthen a view that they are not conditions that would be expected between independent parties.