Canada’s general anti-avoidance rule has come under the spotlight in recent years: as the Canada Revenue Agency becomes more aggressive about challenging complex tax planning, more taxpayers are pushing back. But the litigation that’s bloomed as a result has done little to create more clarity around a section in the Income Tax Act that many view as far too vague.

Of course, tax planners thrive on certainty; a non-specific rule such as the GAAR is likely to be viewed with skepticism by tax professionals at the best of times.

A growing number of industry experts are calling for clarification of the law, as well as for specific guidelines on issues ranging from interest deductibility to the treatment of losses.

Peter Megoudis, senior manager in the Toronto office of Deloitte & Touche LLP, says that the GAAR is very difficult for tax practitioners to deal with. He argues that a practitioner will assume the ITA means what it says, while the courts seem to follow a different set of criteria.

“The application of the GAAR is incoherent and does not fit within the Income Tax Act,” says Megoudis. “[The act] becomes irrelevant when analyzing GAAR cases.”

The GAAR states that three conditions must be satisfied before the rule is applied to disallow a transaction:

> there must be a tax benefit from one or more of the transactions implemented;

> the transaction must be an avoidance transaction — meaning that its only purpose is to obtain the tax benefit and that there is no other non-tax purpose for the transaction; and

> the transaction or series of transactions implemented must be abusive — meaning that the avoidance transaction is inconsistent with the object, spirit or purpose of the ITA.

These concepts seem simple enough at face value; but, in reality, the application of the GAAR is painfully difficult — mainly because of the third condition. What exactly does “abusive” mean in this context?

Brian Carr, a partner with Toronto-based Moskowitz & Meredith LLP, a law firm affiliated with KPMG LLP in Canada, says that the application of the GAAR is, in fact, highly subjective: “It is difficult to predict what decision the court would make, even if you satisfy the provisions of the Income Tax Act.”

The Supreme Court of Canada, for its part, has openly acknowledged the difficulty; the SCC has also stated, however, that this type of uncertainty is typical of difficult legal situations and, thus, to some extent, must be expected.

As the SCC noted in Lipson v. The Queen, the widely noted 2009 decision dealing with mortgage swaps and the attribution rules: “To the extent that it may not always be obvious whether the purpose of a provision is frustrated by an avoidance transaction, the GAAR may introduce a degree of uncertainty into tax planning, but such uncertainty is inherent in all situations in which the law must be applied to unique facts.”

In other words, when it comes to the GAAR, a certain amount of uncertainty goes with the territory. (For more on Lipson, see page B12.)

Although the GAAR has been around for two decades, application of the rule was not considered by the SCC until the 2005 decision in the Canada Trustco Mortgage Co. case. (Canada, which is one of only a handful of countries with such a rule, enacted the legislation in 1988).

The SCC, in the Canada Trustco case, signalled much of what was to come from the jurisprudence by noting that: “The line between legitimate tax minimization and abusive tax avoidance is far from bright.”

On the other hand, the SCC also stated that the GAAR was enacted as a provision of last resort in order to address abusive tax avoidance; it was not meant to introduce uncertainty in tax planning.

Be that as it may, the legal terrain of the GAAR is now full of potholes, with a clutch of new cases suggesting that the legal ground is likely to become even more uneven before greater certainty emerges. At the same time, it’s helpful to keep a classic legal principle in mind when it comes to reviewing GAAR-related decisions: every case turns on its own particular facts and circumstances.

Although any one legal decision may provide guidance, it’s always possible that another similar —- but somewhat different — situation can arise that ultimately leads to the opposite result.

@page_break@A number of recent judgments make the point. The decision in Collins & Aikman Products Co. , released this past July, dealt with a group of three companies, one of which was a non-resident parent company.

In a complex series of transactions, Collins & Aikman, a multinational car parts manufacturer with significant Canadian operations, engaged in a reorganization that resulted in its paid-up capital leaping from $475,000 to $167 million. Following the reorganization, $104 million was paid out as a return on capital to the non-resident parent company — with no Canadian taxes being paid by that company.

The CRA reassessed the transactions under the GAAR. The agency reduced the paid-up capital to its original amount — from $167 million back to $475,000; reassessed Collins & Aikman for withholding taxes on the deemed dividend on the difference between the $104 million paid out and the paid-up capital of $475,000; and reassessed Collins & Aikman for failing to withhold taxes.

Collins & Aikman appealed to the Tax Court of Canada, which ruled in its favour. The company agreed that the transactions were aimed at tax avoidance (a common admission in such cases; tax avoidance, in and of itself, is lawful), leaving the TCC to determine whether the transactions constituted “abusive” tax avoidance.

The TCC found that the results of each of the transactions were appropriate and not abusive. The TCC emphasized that the GAAR cannot be used to fill in any apparent “gaps” in the ITA: if a provision inadvertently appears to allow for transactions such as the one in question that are highly advantageous to the taxpayer but not consistent with the expectations of the CRA, that is for the government to resolve. The CRA will be appealing Collins & Aikman to the SCC.

The treatment of losses appears to be a particularly unpredictable area when it comes to the application of the GAAR. A recent example is the April 2009 decision in the Gary Landrus case.

Landrus was a limited partner in two partnerships that had acquired and owned separate residential condominium buildings. Following a general downturn in the real estate market, the value of the condos declined significantly. The partners decided to restructure the two partnerships into a single new limited partnership that now owned both buildings. As a result of this restructuring, terminal losses were triggered in the two original partnerships, and the losses were allocated to the investors.

The CRA argued that there was no change in the beneficial ownership of the new partnership, as the investors were the same; the CRA applied the GAAR to disallow the transfer of the losses to the inves-tors. The TCC disagreed with the CRA and found that there was no misuse or abuse of the terminal loss provisions of the ITA.

The CRA appealed to the Federal Court of Appeal, which also ruled in favour of the investors: the FCA said that if one looked to the overall result of the transactions, the object, spirit and purpose (the factors listed in the legislation) of the terminal loss rules were not frustrated.

However, in the 2005 Mackay case, a partnership sought to acquire a shopping mall from a bank that was in the process of acquiring the mall by foreclosing on an existing mortgage. The transaction was structured in such a way that the bank transferred the mortgage to the partnership, which then foreclosed on the property. The partnership then claimed a loss that was allocated to members of the partnership.

The FCA held that the GAAR did apply — that this transaction amounted to a misuse or abuse of the ITA provision that was designed to prevent artificial loss transfers to third parties. The SCC then denied Mackay leave to appeal.

The FCA’s decision in Mackay was similar to that of the SCC in the 2005 decision known as “Kaulius,” in which investors acquired interests in a partnership to gain access to partnership losses. The CRA and the FCA applied the GAAR to the transaction in both cases.

The FCA held that there was a general policy in the ITA against the transfer of losses between arm’s-length taxpayers, such that the transactions in both Mackay and Kaulius resulted in abuses with regard to the provisions of the ITA when read as a whole.

But in the Lipson case, the SCC seemed to go in two different directions at once, allowing one highly artificial transaction but disallowing another — in the latter instance, the transfer of a loss.

The April 2009 Lehigh Cement Ltd. case was decided on similar grounds to Lipson and Landrus; in Lehigh, the TCC focused on the object, spirit and purpose of the relevant statutory provisions and the “overall results” of a series of transactions in determining that the transactions were abusive for purposes of the GAAR.

In Lehigh, interest payments made by Lehigh on debt held by a related non-resident corporation were restructured as interest payments to an arm’s-length non-resident bank in order to access the domestic withholding tax exemption for interest on certain arm’s-length corporate debt.

The TCC ruled in favour of the CRA, holding that Lehigh abused the withholding tax exemption provisions through a series of transactions.

It is evident that there is little predictability on how and when the GAAR will be applied by the courts. In a strongly worded dissent in the Lipson case, commercial law expert Justice Ian Binnie stated that: “The GAAR is a weapon that, unless contained by jurisprudence, could have a widespread, serious and unpredictable effect on legitimate tax planning.”

Yet the SCC established in the Canada Trustco case that if the existence of abusive tax avoidance is unclear, the benefit of doubt goes to the taxpayer.

William Innes, counsellor in the Toronto office of Fraser Milner & Casgrain LLP, takes a more defensive position: “If you look at how the GAAR has been applied, it has been pretty much of an even playing field.”

Innes admits, however that there is conflict on how the GAAR should be interpreted — and on the techniques of interpretation. IE