Lai’Latif & Co Advocates
A click-to-learn guide
Step 1 / 9
Tax avoidance · the toolkit

Five tools that stop clever tax schemes

Some arrangements obey the letter of a tax law while breaking its spirit. These five mechanisms are how tax systems answer back. Tap each circle to meet them.

THE DOOR
Tap a circle above to see what it stands for.

None of them works alone. Their strength is the way they cover each other. Press Next to take them one at a time.

Mechanism One

GAAR

The catch-all rule for schemes nobody saw coming.

General Anti-Avoidance Rule. Tap to open each part.

A broad law that lets the tax authority cancel or rework an arrangement whose main purpose is an unintended tax saving.

Three things must usually be shown:

  • An arrangement exists.
  • A tax benefit flows from it.
  • Purpose: getting that benefit was a main purpose.

Many GAARs add a fourth filter: the scheme must misuse the law or be artificial. Genuine business deals are safe.

What the company does. AlphaCo owns a building that cost it $6 million and is now worth $10 million. A genuine outside buyer wants to buy it. Sold directly, the $4 million profit would be fully taxed in Zimbabwe.

How it tries to avoid tax. Instead of selling directly, AlphaCo first sells the building to BetaCo, a company it controls in a low-tax country, for only $6 million, so no profit appears in Zimbabwe. BetaCo then immediately resells the building to the real buyer for $10 million. The $4 million profit now sits in the low-tax country.

Why the GAAR steps in. The first sale had no commercial point. BetaCo never used the building and carried no real risk, and the two sales happened almost together. The only reason for the extra step was to move the profit out of Zimbabwe.

How the GAAR is applied. The authority shows there was an arrangement, that it produced a tax benefit, and that obtaining the benefit was its main purpose. It then disregards the artificial first sale and taxes AlphaCo as if it had sold the building straight to the real buyer for $10 million. The $4 million is taxed in Zimbabwe after all.

It is the backstop. Because it is built on purpose, not on a list of banned structures, it catches the scheme no specific rule predicted.

Mechanism Two

SAAR vs GAAR

Specific rules for known tricks, beside the broad rule.

Two ways to draft an anti-avoidance rule. Tap to compare.

Precise, mechanical rules aimed at one known abuse, such as thin capitalisation, transfer pricing or interest limitation rules. Certain and easy to apply, but narrow.

Broad rules built on purpose. They reach the new and unexpected, but with less certainty until they are applied to real facts.

They are partners, not rivals. SAARs are the front line for known risks. The GAAR is the safety net behind them. A strong system uses both.

The set-up. A country has a thin capitalisation rule, a SAAR, that disallows interest on related-party debt above a debt-to-equity ratio of 3 to 1.

The clever response. A group studies the rule and keeps the borrowing of its Zimbabwean company carefully at 2.9 to 1, just inside the limit. On its face, the specific rule is satisfied.

Where the GAAR comes in. But that loan is only one part of a wider arrangement whose steps, taken together, have no commercial purpose beyond stripping profit out of the country. The specific rule has been respected to the letter, yet the scheme as a whole is still abusive.

The result. The GAAR can still apply to the arrangement as a whole, because its dominant purpose is the tax benefit. The SAAR draws one hard line; the GAAR judges the whole scheme. This is why a system keeps both.

Mechanism Three

LOB

Limitation on Benefits: who may use a tax treaty.

Limitation on Benefits. Tap to open each part.

A clause in a tax treaty that limits the treaty’s benefits to qualified persons: residents who pass a set of objective tests.

From United States treaty practice, designed to stop treaty shopping: routing income through an intermediary in a treaty country just to grab benefits meant for someone else. It is now part of the global standard.

A qualified person is a treaty resident with a real economic connection to its own country, strong enough that it genuinely deserves the treaty’s benefits. The LOB clause will not simply trust a company’s registered address. It makes the company prove that connection by passing at least one objective test:

  • Publicly traded test: its shares are listed on a recognised stock exchange in that country.
  • Ownership test: it is owned mostly by residents of that same country.
  • Active business test: its income is connected to a real trade or business actually carried on there.

Pass one of these and you are a qualified person, entitled to the treaty. Pass none and you are treated as a mere conduit, and shut out.

What the company does. DeltaCorp is resident in Country A, which has no tax treaty with Zimbabwe. Dividends from its Zimbabwean investment would suffer the full 15% withholding tax.

How it tries to avoid tax. DeltaCorp sets up HoldCo in Country B only because Country B’s treaty with Zimbabwe lowers that rate to 5%. HoldCo has one nominee director, no staff and no office, and does nothing but receive the dividends and pass them up to DeltaCorp.

Why the LOB steps in. When HoldCo claims the 5% rate, it must prove it is a qualified person. It is not listed, so it fails the publicly traded test. It is owned from Country A, not Country B, so it fails the ownership test. It runs no real business, so it fails the active business test.

How the LOB is applied. Passing no test, HoldCo is not a qualified person. The treaty’s 5% rate is refused, and the full 15% applies.

It screens out treaty shoppers by a checklist. It asks what you are, not why you came.

Mechanism Four

PPT

Principal Purpose Test: why are you really here?

Principal Purpose Test. Tap to open each part.

A treaty rule that denies a benefit if obtaining it was one of the principal purposes of an arrangement.

The threshold is low: the benefit need only be one of the main purposes, not the only one. But if granting it fits the treaty’s real intent, the benefit is kept.

  • LOB is a checklist. It asks what are you?
  • PPT is a purpose test. It asks why did you do this?

Many modern treaties now use both together.

The set-up. Take HoldCo from the LOB example, but now Country B’s treaty uses a Principal Purpose Test. Suppose HoldCo was arranged a little more carefully, holding a few genuine assets, so that it just scrapes past one objective LOB test.

Why a checklist is not enough. Under a pure checklist, HoldCo might now slip through. The PPT asks a different question: looking at all the facts, was obtaining the reduced 5% treaty rate one of the principal purposes of creating HoldCo?

How the PPT is applied. The facts answer it. HoldCo was incorporated only after the Zimbabwean investment was planned, it has almost no activity, and the structure makes no sense except for the tax saving. The authority concludes, reasonably, that the treaty benefit was a principal purpose.

The result. Unless HoldCo can show that giving it the benefit would fit what the treaty was actually designed to achieve, the 5% rate is denied.

It is the treaty-level backstop: the GAAR of the treaty world.

Mechanism Five

APA

A deal struck in advance, so the dispute never happens.

Advance Pricing Agreement. Tap to open each part.

An agreement made before the fact, between a taxpayer and one or more tax authorities, fixing the transfer-pricing method for future related-party transactions.

Transfer pricing, the pricing of deals inside one group, is the biggest source of cross-border tax disputes. An APA settles the method in advance.

  • Unilateral: one tax authority. Quick, but no cover abroad.
  • Bilateral: two authorities. Removes double taxation.
  • Multilateral: more than two authorities.

Pre-filing chat → formal application → review → negotiation → agreement → yearly reports → renewal.

The situation. EpsilonManufacturing is a Zimbabwean company that makes components and sells them only to its own parent company abroad. Because the buyer and the seller are in the same group, there is no open-market price.

The risk. Years later, the tax authorities of both countries could each argue the price was wrong, each wanting more of the profit taxed at home. The same profit could even end up taxed twice.

What is agreed in advance. The group applies for a bilateral APA. Over about two years, the company and both tax authorities study what Epsilon really does and agree a method: ‘cost plus’, under which Epsilon charges its parent its costs plus a fixed 8% margin, for the next five years.

The result. For those five years, as long as the facts hold and Epsilon files its yearly confirmation, neither authority will reopen the pricing of those sales, and double taxation cannot arise.

It is preventive. It removes the dispute before it can even form.

Drawing it together

How the five fit together

Line them up and a pattern appears. Tap each group to open it.

Both work by objective tests. An objective test is one with a fixed, factual answer that does not depend on anyone’s intention: a debt ratio is met or it is not; a company is listed or it is not; ownership is above a set percentage or it is not.

This makes them fast and certain, because a taxpayer and an official will reach the same answer. The weakness is the other side of the same coin: because the answer is purely factual, a scheme designed carefully around the exact wording can satisfy the test and still be abusive.

Instead of a fixed factual test, both ask why an arrangement was entered into, judged reasonably from all the facts.

This lets them reach schemes a checklist never imagined, including brand-new ones, because they look at substance over form. The cost is some loss of certainty: until the rule meets real facts, the outcome involves a judgement.

It stands alone. Instead of catching abuse afterwards, an APA settles matters before the transactions happen. What can be agreed in advance includes:

  • the transfer-pricing method, for example cost plus, or a comparable open-market price;
  • the profit margin or mark-up that will be accepted as fair;
  • exactly which transactions, and which years, the agreement covers;
  • the assumptions it depends on, and how it is reviewed if business conditions change.

With these fixed up front, there is nothing left to dispute later.

Known tricks, new tricks, treaty abuse by structure, treaty abuse by purpose, and mispricing. Together, the five leave very little room to hide.
Your turn

Try five scenarios

Read the story, tap the mechanisms you think apply, then reveal the answer. You can pick more than one.

Scenario One
A company in Country X wants dividends from a Zimbabwean company, but the two countries have no tax treaty. So it sets up an empty holding company in Country Y, which does have a favourable treaty with Zimbabwe. The Country Y company has no staff, no office, and does nothing but hold the shares.
Which mechanisms catch this?

This is treaty shopping. The answer is LOB and PPT.

LOB: the empty Country Y company passes none of the qualifying tests, so it is not a qualified person and is refused the treaty rate.

PPT: the treaty benefit was plainly a principal purpose of inserting that company, and it has no real substance.

Not the others: GAAR and SAARs are domestic rules; an APA is about pricing. The abuse here is the misuse of a treaty.

Scenario Two
A profitable Zimbabwean subsidiary is restructured by its multinational group. It takes a huge loan from a related company and pays a big royalty to another, wiping out its taxable profit. The scheme is new, no single specific rule clearly stops it, and its only real purpose is the tax saving.
Which mechanisms apply, or would have prevented this?

This is domestic profit-stripping. The answer is GAAR, SAAR and APA.

GAAR: the scheme is new and its only real purpose is the tax saving, the classic job of the backstop.

SAAR: thin capitalisation or interest rules may cap the loan deduction; transfer pricing rules may attack the royalty.

APA: a bilateral APA could have fixed an arm’s-length price for the loan and royalty in advance.

Not the others: LOB and PPT are treaty rules. This scheme abuses domestic law, not a treaty.

Scenario Three
ZenithCo, a Zimbabwean company, is funded by its foreign parent almost entirely through a very large loan rather than share capital. The interest ZenithCo pays on that loan each year is so big that it wipes out nearly all of its taxable profit. Loading a company with related-party debt like this is a long-known tax trick.
Which mechanism deals with this most directly?

A known trick, so the answer is SAAR.

SAAR: a thin capitalisation rule caps the deductible interest on related-party debt at a fixed ratio. The excess interest is added back to ZenithCo’s taxable profit, automatically, with no need to prove anyone’s purpose.

Because a precise rule already covers this exact mischief, the broad GAAR backstop is not needed here.

Not the others: LOB and PPT are treaty rules, and an APA is about agreed pricing. None is the tool for a domestic debt-loading trick.

Scenario Four
OrionParts is a Zimbabwean manufacturer that sells all of its output to its parent company overseas. The business is completely genuine, but because the buyer is a group member, there is no open-market price. OrionParts wants certainty: no surprise audit in five years arguing the price was wrong, and no risk of the same profit being taxed in two countries.
Which mechanism gives OrionParts what it wants?

There is no abuse here at all, and that is the lesson. The answer is APA.

APA: OrionParts and the tax authorities agree the transfer-pricing method in advance, for a fixed number of years. A bilateral APA, involving both countries, also removes the risk of the same profit being taxed twice.

Not the others: GAAR, SAAR, LOB and PPT all exist to catch or deny abuse. There is no abuse here, only a genuine business that wants certainty.

Scenario Five
A group wants to take a large one-off profit out of its Zimbabwean company while paying as little tax as possible. It creates a brand-new subsidiary, transfers a valuable profitable contract into it for no payment, lets the profit build up inside it, then liquidates the subsidiary so the value returns as a lightly taxed capital distribution rather than ordinary income. Each step is lawful on its own, the combination is unusual, and no single specific rule squarely covers it.
Which mechanism is built for exactly this?

A novel, multi-step scheme with no purpose but tax. The answer is GAAR.

GAAR: the authority looks at the steps as one composite arrangement, not separately, shows the arrangement produced a tax benefit that was its main purpose, and then taxes the profit as ordinary income, as if the artificial steps had not happened.

Not the others: the SAARs were deliberately designed around; no treaty is in use, so LOB and PPT do not apply; and an APA concerns transfer pricing, not a one-off restructuring.

One picture to remember

It is a night at the club

Think of the tax system as the door of a club. Tap each role to see who is who.

SAAR
The printed list of specific bans
No trainers, no under-21s. It stops the trouble someone already named, but the clever guest simply dresses around it.
GAAR
The bouncer’s own judgement
No list needed. If he can see your real reason for coming is trouble, you are turned away.
LOB
The guest list
You get in only if your name is on it, or you meet a set rule. Quick and certain, but a wrong name can slip through.
PPT
The question at the door
Why are you really here? Even a name on the list is refused if the true purpose is wrong.
APA
The booking made ahead
You phone, agree the terms, reserve the table. There is no argument on the night.

Tap any row to open or close it.

A good door needs all five. So does a tax system.
Anti-Avoidance Mechanisms in Tax Law
Created by Lyla Latif (PhD) · Lai’Latif & Co Advocates
www.lai-latif.com
Lai’Latif & Co Advocates  ·  www.lai-latif.com