The GAAR applied: Deans Knight Income Corp v Canada

In Deans Knight Income Corp. v Canada, 2023 SCC 16, (“Deans Knight”) the Supreme Court of Canada (“SCC” or “the Court”) dismissed the taxpayer’s appeal, holding that the General Anti-Avoidance Rule (“GAAR”) under s. 245 of the Income Tax Act, RSC 1985, c 1 (5th supp) (“ITA” or “the Act”), applied to a series of transactions that were technically compliant with s. 111(5) ITA, but amounted to an abuse of the underlying rationale. 

The GAAR invariably injects some uncertainty into a tax avoidance analysis, but Deans Knight confirms that if a taxpayer achieves a result functionally equivalent to that which a provision aims to prevent, it’s likely the GAAR will apply.


Legislative Scheme and Facts

Legislative Scheme

Section 111(1)(a) ITA permits offsetting previous years’ losses against taxable income, reducing the amount subject to taxation (Deans Knight, para 77). 

Section 111(5) ITA serves as an anti-avoidance measure tied to s. 111(1)(a), restricting loss carryover when a change in control of a corporation occurs unless there’s continuity in its business (Deans Knight, para 77). An acquisition of control occurs when specific share-related rights are obtained (ITA, s. 256(8)), with “control” recognized as de jure control. De jure control constitutes ownership with a majority of votes to elect the board.

The determination of corporate “control” is pivotal, as change in control is the sole trigger for the loss-trading restriction in s. 111(5) (Deans Knight, para 175). 

The GAAR (s. 245 ITA) is a broad measure designed to prevent undesirable tax avoidance without creating specific rules that could complicate matters or create loopholes (Deans Knight, para 44). The GAAR analysis involves three questions: (i) Was there a tax benefit? (ii) Was the transaction an avoidance transaction? (iii) Was the avoidance transaction abusive? To determine if a transaction was abusive, the purportedly abused provison’s underlying rationale is characterised by its “object, spirit, and purpose” and the transaction is assessed against that rationale (Deans Knight, paras 56–57).  

With the first two questions undisputed in this case, the Court’s focus turned to the third (Deans Knight, para 75). 


In 2007, the appellant taxpayer, formerly known as Forbes Medi-Tech (“Forbes”), faced financial challenges but had accumulated nearly $90-million in unused tax attributes (Deans Knight, para 7). To capitalise on these, Forbes entered into an agreement with Matco, a venture capital company, to find a way to take advantage of the s. 111(1)(a) carryover benefit without triggering the s. 111(5) limitation (Deans Knight, paras 3, 8–10).

The appellant board orchestrated a strategic reorganisation through a takeover whereby the appellant retained its tax attributes but all its assets and liabilities were transferred to a newly established entity (“Newco”) (Deans Knight, para 9). 

Newco and the appellant entered into an Investment Agreement with Matco, in which Matco agreed to find a business that could utilise the tax attributes. The Investment Agreement included the following:

  • The appellant issued a $3-million convertible debenture to Matco, giving Matco 35 percent of voting shares and 100 percent of non-voting shares (total of 79 percent of equity shares) (Deans Knight, para 11). 
  • Matco guaranteed that Newco would be able to sell its remaining shares in the appellant for a minimum $800,000 (Deans Knight, para 11). 
  • The appellant and Newco agreed to not undertake a host of financial, structural, and strategic decisions without Matco’s consent, effectively prohibiting the Appellant from acting without Matco’s approval (Deans Knight, para 14). 
  • The Agreement was carefully structured to not be a unanimous shareholder agreement (Deans Knight, para 15). 

Deans Knight Capital Management (“DKCM”) is a mutual fund management company that was looking to invest in high-yield debt instruments. In collaboration with DKCM, Matco proposed an initial public offering (“IPO”) facilitated by the appellant. This strategic plan involved leveraging the appellant’s accumulated tax attributes to shield income and capital gains (Deans Knight, para 17). 

Matco orchestrated a $100-million IPO for the appellant, managed by DKCM, utilising the appellant’s $90-million tax losses to earn income from corporate debt securities, sheltered from tax. 

Matco converted the debenture, acquiring the appellant’s shares for the agreed $800,000. Newco received $3.8-million for appellant’s shares, and Matco’s publicly-traded shares in the appellant were valued at $5-million post-IPO (Deans Knight, paras 11, 18, 20–22). 

Following execution of the plan, the appellant deducted $65-million from its tax attributes to offset tax liability (Deans Knight, para 24). 

However, the Minister of National Revenue reassessed the claimed losses, invoking the GAAR to deny the deductions due to abusive tax avoidance (Deans Knight, para 15). 


Judicial History

Tax Court of Canada

The Tax Court of Canada (“TCC”) was tasked with two issues: (i) Did Matco acquire control pursuant to ss. 256(8) and 251(5)(b) of the Act; and (ii) does the GAAR apply to deny the deduction of the Tax Attributes (Deans Knight Income Corporation v The Queen, 2019 TCC 76 (“Deans Knight TCC”). The primary issue was the GAAR question. 

The TCC found that the object, spirit, and purpose of s. 111(5) was to “target manipulation of losses of a corporation by a new person or group of persons, through effective control over the corporation’s actions” (Deans Knight TCC, para 134). Justice Paris held that Matco had not obtained control; therefore, the s. 111(5) limitations had not triggered and the GAAR did not apply. 

The Crown appealed to the Federal Court of Appeal (“FCA”). 

Federal Court of Appeal

The FCA reversed the TCC’s decision, finding that its “object, spirit, and purpose articulation” was largely correct, but replaced the term “effective control” with “actual control” (para 35), though it did not elaborate much on the contours of “actual control” other than to say it extended beyond a strict de jure test (Canada v Deans Knight Income Corporation, 2021 FCA 160, para 73 (“Deans Knight FCA”)).

Justice Woods held that object, spirit, and purpose of s. 111(5) was not fully expressed by its text (Deans Knight FCA, paras 76–76). Looking at ITA’s general policy purpose—i.e. to prevent loss transfers between taxpayers—s. 111(5) and the GAAR responded to unintended use of loss carryovers (Deans Knight FCA, para 81).

The FCA concluded that the Investment Agreement gave Matco actual control over the appellant’s actions and that the avoidance transactions resulted in an abuse of s. 111(5). The GAAR conditions were met, and the tax benefit should be denied (Deans Knight FCA, paras 105, 114).

Supreme Court of Canada 

The sole issue before the Court was whether the tax avoidance transactions were abusive under the GAAR given the first two steps of the GAAR analysis were undisputed. 

The GAAR Analysis

Writing for the majority, Rowe J. clarified that the third question amounts to determining if the GAAR applies to s. 111(5), requiring (i) understanding the “object, spirit, and purpose” of s. 111(5), and (ii) assessing if the transactions frustrated this underlying rationale (Deans Knight, paras 4, 75). In essence, the analysis asks if the tax avoidance transaction, when evaluated against the provision allowing tax avoidance, abused the provision’s underlying rationale (Deans Knight, para 117).

Object, Spirit, and Purpose

The object, spirit, and purpose is a concise statement of a provision’s rationale, but the provision’s text itself may not give a full answer to its rationale (Deans Knight, para 83–84). Justice Rowe held that a court must consider the Act as a whole, related sections, legislative history, and extrinsic evidence to discern Parliament’s intentions through s. 111(5) to determine what Parliament sought to achieve. 

The Court found that the object, spirit, and purpose of s. 111(5) is “to prevent corporations from being acquired by unrelated parties in order to deduct their unused losses against income from another business for the benefit of new shareholders” (Deans Knight, para 78).

The appellant argued that Parliament’s use of a de jure control test within s. 111(5) indicated the object, spirit, and purpose of the provision was to target tax loss trading where acquisition is within actual de jure control of the tax attribute-bearing company, which Matco had technically avoided through its manoeuvre (Deans Knight, para 91). The Court rejected this reasoning, noting that while the text of a provision alone may explain its rationale in some cases, here, the de jure control test was merely a means of achieving Parliament’s intent, but did not capture it fully (Deans Knight, para 116).

Subsection 111(5) essentially encompasses three elements: control, acquisition, and continuity, permitting deductible losses if the acquired corporation remains in the same or similar business post-acquisition (Deans Knight, paras 79–82). Sections 111(5) and 111(1)(a) work together to provide benefits aligning with the Act’s overarching purpose—linking tax benefits to those who endured losses (Deans Knight, para 86). 

The provisions work in concert to offer the benefit of mitigating the financial disparities of varying income years while adhering to an overarching principle of the Act that tax benefits should correlate with the losses suffered (Deans Knight, paras 86–90).

Abusive transactions

Upon reviewing the impugned transactions, Rowe J. determined that their outcome undermined the purpose of s. 111(5) (Deans Knight, para 122). The manoeuvres led to a significant transformation, transferring the taxpayer’s assets and liabilities to Newco, leaving only tax attributes (Deans Knight, paras 6, 124). These tax attributes were then utilised by an unrelated party to shield profits from a business entirely different from the one generating the tax attributes (Deans Knight, paras 124, 127, 140).

The impugned transactions resulted in a “near-total transformation” of the appellant which was “gutted” of virtually everything except its tax attributes (Deans Knight, para 124). The Investment Agreement was made to benefit from the tax attributes (Deans Knight, paras 126–127). The appellant became “in practice, a company with new assets and liabilities, new shareholders and a new business” (Deans Knight, para 127).

Despite not triggering a de jure “acquisition of control,” Matco achieved what Rowe J. dubbed a “functional equivalent” to such an acquisition (Deans Knight, para 128). This was evident in the Investment Agreement, which dictated director appointments, imposed restrictions on the board’s powers, required written consent from Matco for taxpayer actions, and allowed Matco significant financial benefits (Deans Knight, para 129–133).

Thus, the series of transactions was designed to achieve the very mischief that subsection 111(5) aimed to prevent. Consequently, the obtained result contradicted the rationale of subsection 111(5) and constituted abuse.


Justice Côté diverged from both the majority and FCA’s interpretation of s. 111(5), asserting that both of their applications of the GAAR overlooked Parliament’s intent (Deans Knight, para 143).

The sole dissenting judge emphasised the conclusive nature of a provision’s text, particularly in specific anti-avoidance rules like s. 111(5), contending that the object, spirit, and purpose of s. 111(5) is to limit the use of tax attributes when accessed through an acquisition of de jure control (Deans Knight, para 171). 

Justice Côté accordingly pointed out that mechanisms used in the tax avoidance transactions did not constitute de jure control because they lacked formal direct control or influence over the corporation’s voting shareholders (Deans Knight, para 164). 

Challenging the majority’s approach, Côté J. argued that looking beyond the means to give effect to a provision implies that “Parliament is unable to translate its objectives into effective means” (Deans Knight, para 152). She highlighted the significance of Parliament’s “bright-line” distinction between the de facto and de jure control tests as fundamental to the provision’s rationale (Deans Knight, para 168). Looking beyond the text and finding de jure control not determinative “completely ignores how Parliament actually defined what constitutes acquisition of control” (Deans Knight, para 176).

While she agreed with Rowe J.’s characterization of the GAAR as targeting unforeseen tax strategies, Côté J. emphasised that it should not be wielded to penalise taxpayers for exploiting intentionally open avenues (Deans Knight, para 154).

Justice Côté criticised Rowe J.’s introduction of “functional equivalence” for treating an investment agreement as a constating document, arguing that this approach ignores the distinct enforcement methods of such agreements (Deans Knight, paras 178–179). 



A key divergence between the majority and dissent rested on whether a provision’s text alone can determine its rationale, a question involving Parliamentary intent, which raises the subsidiary issue of certainty and predictability.

These were concerns raised by the appellant and intervenors, questioning whether tax planners can anticipate the explicit boundaries of a provision to define themselves. As Iacobucci J. observed in Duha Printers (Western) Ltd. v. Canada, [1998] 1 SCR 795 (“Duha”), “taxpayers rely heavily on whatever certainty and predictability can be gleaned from the Income Tax Act” (para 51).

While courts usurping the role of legislatures is an ailment Rowe J. and the SCC are aware of, I think Rowe J. does resolve this issue, even if not head-on. 

Relying on Canada Trustco Mortgage Co. v Canada, 2005 SCC 54, (“Canada Trustco) Côté J. argued that the GAAR simply “cannot override Parliament’s specific intent regarding particular provisions of the [ITA]” which she says is a critical oversight of the majority (Deans Knight, para 146). 

Canada Trustco held that courts are restricted to interpreting the ITA through a unified, textual, contextual, and purposive analysis of specific provisions (para 41). Searching for an overarching policy not rooted in this interpretation—which undermines the certainty and fairness goals of Parliament—contradicts legislative intent (Canada Trustco, para 42). While the GAAR aims to prevent abusive tax avoidance, Parliament also seeks consistency, predictability, and fairness in tax law (Canada Trustco, para 42). 

That said, I would think that even Canada Trustco’s “consistency, predictability, and fairness in tax law” cannot reasonably be taken to mean ignoring the ITA’s overarching purposes (para 42). Further, as Rothstein J. noted in Copthorne Holdings Ltd. v. Canada, 2011 SCC 63, “if the Court is confined to a consideration of the language of the provisions in question, without regard to their underlying rationale, it would seem inevitable that the GAAR would be rendered meaningless” (para 111). 

If the GAAR can’t override Parliament’s “specific intent” for s. 111(5), as per Côté J., but the s. 111(5) is subject to the general anti-avoidance rule, the issue is harmonisation. Though not referred to in any of the courts’ reasons, the modern principle of statutory interpretation seems instructive: i.e. “…read in their entire context…harmoniously with the scheme of the Act, the object of the Act, and the intention of Parliament” (Elmer Driedger in Construction of Statutes (2nd ed. 1983), cited in Rizzo & Rizzo Shoes Ltd. (Re), [1998] 1 SCR 27, para 21). 

Granted, that the intention of Parliament was not agreed upon by the majority and dissent may result in some question begging. Nevertheless, to me, statutory harmonisation ultimately emerged through Rowe J.’s comprehensive overview of the history of the provisions, the ITA as a whole, and Parliamentary records (Deans Knight, para 40–52). 

Did Parliament intend for the use of de jure control to allow for an all but technically de jure acquisition of control to be? For the reasons the majority gives, it seems reasonable that they did not. Yet this conflicts with what might be considered the most (or more accurately, simplest) predictable resolution, which would be to allow for the loophole found in a plain reading. Even still, on a plausible meaning reading within the entire scheme, I am much more easily convinced by Rowe J.’s conclusion.

As Osgoode professor Jinyan Li observes, the GAAR necessarily creates some level of uncertainty as a “standard-based” rather than “rule-based” provision, unlike most of the ITA. The majority acknowledged this, stating that in creating a general anti-avoidance rule, Parliament was aware of the GAAR’s implications for the respective level of certainty in tax planning (Deans Knight, para 48). Ultimately, Deans Knight underscores that principle that the GAAR applies to transactions contrary to the ITA, and it and s. 111(5) both operate within the overarching purpose of the regime.

Leave a Reply

Your email address will not be published. Required fields are marked *