With contributions from Vivien Lee, Aaron Ma and Eric Savoie
Monthly webcast
The monthly economic webcast for March is now available: Iran in focus / AI worries surface. (Note it was recorded on March 2 and so contains only preliminary comments on the war in Iran.)
War with Iran
The U.S. and Israel struck Iran in a coordinated effort on February 28, delivering on an outcome that we had believed was even more likely than the market was pricing in.
However, the magnitude of the action exceeded virtually all expectations. It has ignited a full-blown war that bears little resemblance to the targeted strikes of June 2025. A wide range of military and political targets were struck, and a significant fraction of Iran’s political leadership were killed – including the country’s Supreme Leader and the head of the Revolutionary Guard. Intense missile strikes continue more than a week later and seem set to stretch into the future.
The motivations for the attack are now being debated. They arguably include several goals:
to quash Iran’s nuclear weapons aspirations once and for all
to halt the country’s missile-development progress
to protect other Middle Eastern countries
to stop Iran from projecting power across the region via its support of militias
to punish Iran for its recent suppression of dissidents
to achieve regime change so that Iran becomes a more democratic and West-friendly government.
Sensing an existential threat, Iran’s government has responded ferociously. It targeted not just Israel and regional U.S. military sites, but also the energy and civilian infrastructure of neighbouring Gulf states.
Oil market
Iran is of greatest consequence to global markets via the price of oil. The country itself produces about three million barrels of oil per day. The pressure it exerts on the Strait of Hormuz puts a whopping 20% of the world’s oil and gas output in doubt.
It introduces some major variables into global markets:
the potential loss of Iranian production
the damage Iranian rockets have done or could do to extraction, refinement and shipping facilities elsewhere in the Persian Gulf
the threat to tanker ships attempting to transit the area.
In practice, the volume of ships travelling through the Strait of Hormuz has fallen to almost nil, signaling that the worst-case scenario is so far playing out (see chart below).
Ship crossings in Strait of Hormuz come to a halt
As of 03/08/2026. Sources: Bloomberg, RBC GAM
Given that the demand and supply of oil are both highly inelastic in the short run, it is unsurprising that the price of oil has surged (see chart below) and remains in flux. Over just the past day (March 9, 2026) the price of West Texas Intermediate oil started at US$89 per barrel, rose to US$119, and then tumbled back to US$95 per barrel. Regardless of the exact price, the cost of oil is much higher than the approximately US$65 per barrel that prevailed a month ago.
Crude oil prices rebound on geopolitical tensions
As of 03/06/2026. Sources: Microband, RBC GAM
Middle Eastern oil is consumed disproportionately by China and other Asian nations (approximately 70% of the total), increasing the theoretical damage to those markets.
Natural gas production and exporting has also been curtailed – a severe negative for the Asian and European markets that import this product.
Duration of disruption
The severity of the oil and natural gas disruption is reasonably well understood. Middle Eastern energy is simply not getting out in large quantities.
The other key dimension is the expected duration of the disruption. This is not yet clear: it might be two weeks, two months or even two years. A short-lived event can be shrugged off easily by markets, the economy and inflation. A long-lived event would be a big problem.
The oil futures market argues that the oil spike should be temporary, but not resolved overnight. The price of oil is expected to trend steadily lower over the next eight months. At that point it will be only modestly higher than the pre-war futures market had assumed.
Oil futures suggest price spike will be temporary, some risk premium persists
As of 03/09/2026. Sources: Bloomberg, RBC GAM
Betting markets give just a 35% chance that the war is over by March 31, but a substantial 73% chance it is done by May 15. A separate market assigns a 59% chance that the Strait of Hormuz traffic will return to normal by April 30.
These are not unreasonable expectations. As to where the risks lie, we believe it is more likely that the energy blockage is resolved sooner than this, rather than later. This is for several reasons:
Interests are strongly aligned, with the possible exception of Russia – most of the world cares deeply about getting this energy flowing again.
China is the most reliant on Persian Gulf oil, and as a notional ally of Iran, has reportedly been pressuring the country to allow oil to flow uninterrupted.
Even though the U.S. does not import significant quantities of Persian Gulf oil, the White House cares about the stock market, the economy and pocketbook issues like the cost of gasoline in the run-up to the midterm elections in November. It is also presumably sensitive to the fact that President Trump’s political base does not like military adventurism.
The rest of Asia and Europe are deeply reliant on Persian Gulf energy and also want a solution.
Gulf states are doubly incented to resolve the current situation. First, they are unable to sell their energy. Second, their diversification plan is being challenged: low-tax, low-regulation business hubs are much less appealing when they are under physical threat.
The U.S. has offered to insure and perhaps even protect vessels transiting the Strait of Hormuz, picking up the slack from private insurers that are unable to take that level of risk or offer protection.
Iran’s military capabilities are presumably declining as the U.S. and Israel bombard it. A question mark is its supply of drones, which may be considerable, and can be launched from almost anywhere.
Iran’s ability to finance its side of the war is limited without the ability to sell its oil to the international market.
The rest of the world is already pursuing workarounds. While hardly replacing Middle Eastern oil altogether, they do make the loss a bit less painful. India is being allowed to import more Russian oil. China is tapping its ample reserves. The G7 nations are now talking of a coordinated release of their petroleum reserves. OPEC (mainly Saudi Arabia) agreed to produce incrementally more oil.
There are obvious parallels to the war between Russia and Ukraine, as in March 2022 that event also triggered a spike in oil prices to around $120 per barrel. While prices remained volatile and quite elevated for several months, oil prices ended the year at about $80 per barrel – approximately where they had started the year. This happened even though the war in Ukraine did not end.
While risk assets were quite unhappy during the first half of 2022, this had more to do with rapidly tightening monetary policy and spiking inflation. This inflation was partially a function of oil prices, but more so the result of overly strong post-pandemic demand and gnarled supply chains.
Efforts to draw parallels to the highly damaging and long-lasting oil embargos of the 1970s are flawed. In those cases a sizable number of oil producers were united in their desire to punish the western market. No such coordination or desire exists today – indeed, it is the opposite. In turn, the dislocation in markets should be less severe and shorter-lived.
Conversely, there are risks to this relatively constructive view:
The barrage of Iranian drones may be difficult to stop.
A ceasefire may be elusive even if the acute phase of military strikes winds down.
Iran may do lasting damage to the Middle East’s oil infrastructure.
Restarting oil and gas production at a later date may prove difficult and time-consuming.
The U.S. and Israel may stubbornly insist on regime change before any cease-fire.
End-game scenarios
The odds of regime change in Iran have clearly increased given the scale of the U.S. and Israeli attacks, the removal of Iran’s head of state, and U.S. comments that regime change is indeed a goal. If a regime change succeeded, it might bring Iran into better alignment with the rest of the world and allow sanctions to be lifted and Iran to produce more oil. The world would be a notably safer and more prosperous place.
However, despite becoming more conceivable, regime change remains relatively unlikely, for several reasons:
The existing theocracy is deeply entrenched.
The opposition is weak.
Regime change is extremely difficult without “boots on the ground” – the deployment of American or Israeli soldiers into Iran. Neither party seems to want this.
Lasting regime change is hard even with boots on the ground, as per the examples in Iraq and Afghanistan.
The odds of an enduring period of political chaos in Iran have also gone up. The longstanding clerical/security regime is on its back foot and the country has an unusually diverse mix of ethnicities (Persian, Arab, Kurd, etc.). It contains a wide range of political factions (the incumbent clerical/security regime, reformers/moderates, a secular/democratic faction, and another with primarily economic grievances). This could create a messy outcome akin to Iraq or Syria with parts of the country held by different groups. It would be an environment ripe for the formation of terrorist groups.
Finally, the odds of a status-quo outcome have gone down given the intensity of the conflict. However, this remains the single most likely outcome.
Thus, whenever a ceasefire occurs, Iran will probably still be in opposition to the West, will still count Israel as an enemy, and will still be subjected to harsh sanctions. An important and positive difference is that the country’s nuclear ambitions and military capabilities should be set significantly back. This may reduce the risk to the rest of the world for a period of time -- although this conflict also risks stirring up intense resentments that manifest negatively at a later date.
Economic implications
The economic implications are difficult to nail down with any precision given that oil prices themselves are gyrating by up to US$40 per day, and given vagaries around how long the high oil prices will last.
As a thought exercise, a permanent increase in oil prices by US$40 per barrel (so from the ~US$65 per barrel that persisted before the war to ~US$105 per barrel) should theoretically subtract about 0.4% from Eurozone GDP and a bit less for other markets like the UK, Japan and China.
However, loosely tacking on an additional hit from the equally large jump in natural gas prices that have befallen much of Asia and Europe, and the theoretical GDP impact would be approximately doubled. Thus, there is a scenario in which economic output in the Eurozone is up to a percentage point weaker than otherwise.
The U.S. economy suffers only a small hit given that it does not directly consume much oil that flows through the Strait of Hormuz, and the country is a net energy exporter. The damage might come instead through a negative stock market wealth effect channel, and via higher uncertainty discouraging risk-taking.
For Canada, higher oil and gas prices are theoretically a net positive for the economy. However, we would not expect a big jump in capital expenditures since the improved outlook is presumably only temporary.
Tracking the same scenario, eurozone inflation might be on the order of 0.8% higher due to the oil price increase. That number might double when natural gas effects are also factored in (this is much harder to do given the inclination of regulators to smooth out price changes).
Other countries would mostly see a slightly smaller inflation jump, though with less variation since the consumers of oil-exporting countries would also face the same approximate increase in their cost of living.
Of course, the fatal flaw in this scenario is that it assumes the oil and gas price increases are permanent. Whether they are resolved in short order or not for several months, most of the damage to the level of output should be recovered at a later date. For every month of elevated inflation there should be a month of depressed inflation later.
On central banks, theoretical models would say that stagflationary energy shocks should elicit rate cuts rather than rate hikes. The damage to economic output should attract more attention than a one-time price level increase. (It’s the same idea as with tariffs.)
Conversely, financial markets have so far priced in a more hawkish monetary policy. For the moment, central banks are likely to watch carefully and see how the war goes, where oil prices settle, and then how much spillover there is into the labour market and the cost of other products.
Market implications
The recent strength in the dollar has proven that its safe-haven status, while diminished, is not entirely gone.
Bond yields have increased, choosing to focus on the inflationary implications and to imagine that central banks must raise rates, rather than to prioritize the safe-haven bid that so often sends government bond yields lower during adverse events.
Credit spreads are wider, but only slightly
Stock markets are down, but with significant variation. Those with a greater exposure – in Asia and Europe – are notably softer than those in North America. As the subsequent article in this #MacroMemo details, past geopolitical events have tended to elicit relatively small and short-lived stock market responses. This argues that markets could return to their prior levels before too long.
Furthermore, if the war with Iran is resolved sooner than the market expects or the Strait of Hormuz is unblocked more quickly – a possibility we are highlighting – that would present another reason for markets to reverse their recent moves.
Longer-term musings
Presuming a cease-fire is achieved within the coming months, what long-run legacy might this conflict leave?
Iran’s ability to pursue nuclear weapons could be permanently snuffed by recent events, dialing down that particular risk to the world.
This war could bring the non-Iran Middle East closer together, helping to forge future alliances. Conversely, the goal of making the Middle East a global commercial hub has been dealt a potentially lasting blow.
The newfound willingness of the current U.S. Administration to engage in military interventions – think not just Iran but also Venezuela – argues that Cuba, Panama, Colombia and Greenland must be on particular watch given recent U.S. comments about them. Canada is more far-fetched. To the extent that the U.S. pattern in recent months has been to pursue “bad actors,” one might also posit that North Korea or Russia could be in focus, though these seem unlikely given their capacity to counterattack.
Even after the dust settles, and notwithstanding a reduced nuclear weapon risk, it would make sense for there to be a slightly larger geopolitical risk premium in markets for the foreseeable future. This war is further proof that these are volatile, dangerous times.
As such, metrics of expected volatility like the CBOE Volatility Index (VIX) could remain a hair higher than normal. Likewise oil prices might remain a bit higher than otherwise. The dollar – while expected to decline over the coming years – could fall by a bit less over that period. It would also make sense for credit spreads to a bit higher; for bond yields to be a bit lower; and for stock valuations to also be a bit lower.
-EL
The stock market during military actions
The war in Iran is causing significant volatility in financial assets. Investors are becoming increasingly concerned that the military conflict could persist for an extended period. A prolonged war could push energy prices higher and weigh on risk assets as investors consider the risks to economic growth and inflation.
While stocks don’t typically suffer significant or lasting downturns as a result of acts of war, the range of historical outcomes is wide. So far, it appears that investors believe the path to de-escalation in Iran may be further away than initially imagined. This is being reflected in larger-than-usual declines in the stock market.
To gauge how the current market experience compares to past military events, we’ve constructed a road-map chart that captures the S&P 500’s reaction to 42 acts of war dating back to World War II (see chart below).
S&P 500 has tended to recover quickly after acts of war since World War II
As of March 6, 2026. Chart captures 42 acts of war dating back to 1941. Source: Bloomberg, Macrobond, Ned Davis Research, RBC GAM
The data includes events such as the Hiroshima bombing in 1945, Iraq’s invasion of Kuwait in 1990, Russia’s invasion of Ukraine in 2022 and many others. T=0 on the chart marks the date of the event. We’ve plotted the S&P 500 Index 30 days leading up to the event and 60 days afterward.
The events are grouped into categories such as U.S. aggressor, U.S. target, assassinations, external events, and acts of terrorism. The median experience for each of these categories as well as the median of all events is plotted in the coloured lines on the chart.
As per the median, the historical stock-market reaction to acts of war has been muted. Stocks have continued to march higher about a dozen days after the event. Since medians don’t reflect the extremes, though, we’ve also added a grey shading to the chart that represents the 10th and 90th percentile of all outcomes as a representation of the full historical range.
The war in Iran remains highly fluid. The situation can change rapidly. But for now the current experience is tracking toward the lower end of the historic range in terms of stock-market reaction.
Even though the S&P 500 has pulled back since the end of February, its decline pales in comparison to the drop in international stocks. These may be more exposed to the war in Iran, given their geographical proximity. As of March 6, 2026, the S&P 500 has fallen 2.0%, but ETFs tracking European, Japanese and emerging market (EM) equities sank 6.6%, 8.2% and 8.4%, respectively, in U.S. dollars over the same time frame (see table below).
But it is also worth noting that non-U.S. equities significantly outperformed in the first two months of the year. As a result, even with the steep drops in March, most international markets remain ahead of U.S. markets on the year.
Stock markets in North America have reacted less than other markets closer to Iran
In U.S. dollars, as of March 06, 2026
Sorted best to worst since 02/27/2026. Sources: Bloomberg, RBC GAM
We note a similar trend in sectors, where many of the hardest hit sectors in the past week remain the outperformers year to date.
Materials, Consumer Staples and Industrials suffered large declines since February 27, but remain up close to 10% so far this year.
Information Technology, Consumer Discretionary and Financials, which fared better in the past week, are still down in the mid-single-digits for the year.
The one big outlier is Energy, which is the only sector that gained in the past week. Helped by surging oil prices, it is up a staggering 25% so far this year.
Many of the hardest hit sectors remain outperformers
In U.S. dollars, as of March 06, 2026
Sorted best to worst since 02/27/2026. Sources: Bloomberg, RBC GAM
The underlying trends within markets beyond the past week may be a reflection of the secular artificial-intelligence (AI) theme. While users interact with AI through a software interface, substantial hardware systems are required behind the scenes to generate the computing power needed to run large language models at scale. And the largest technology companies are expected to accelerate their annual spending, to the tune of hundreds of billions of dollars in capital expenditures to build out AI capacity. That is, more data centres, which require computer chips, energy and real infrastructure to function effectively.
As a result, investors appear to be favouring companies with hard assets rather than intangible assets such as software services which can easily be disrupted by new AI-based solutions. The chart below shows that asset-heavy and goods businesses have outperformed asset-light and services companies by 30% and 19%, respectively, since the start of 2025.
With AI-related capital expenditures expected to continue rising for the next several years, these trends could have further room to run. Of course, the Middle East conflict is dominating headlines and is front of mind for investors more recently. But that situation may eventually pass or lose its ability to shock investors, while durable trends related to artificial intelligence are likely to remain.
Asset-heavy, goods-producing S&P 500 companies have outperformed
As of 03/06/2026. Sources: Goldman Sachs, Bloomberg, RBC GAM
-ES
The promise and peril of artificial intelligence
Up until recently, the artificial intelligence (AI) focus had been the new technology’s great promise for the economy and investors. This is still very important.
Chipmakers are earning a great deal of money. American tech giants are again set to deploy a large and growing sum into data centre expansion in 2026 as they build out their models – a material support to near-term economic growth.
In theory, artificial intelligence should also deliver substantial productivity gains, even if the scale is debated (see chart). That’s a substantial helping hand, and one that we think is already starting to flatter the economic numbers. It is easy to lose sight of the fact that the main theoretical beneficiaries of AI are not the companies developing the models, but instead their many potential users as the technology diffuses across sectors and around the world.
Estimates of potential AI productivity gains vary widely
As at 12/11/2025. Dotted bars indicate range of projections. Sources: OECD, Brookings, Penn Wharton Budget Model, RBC GAM
Fresh AI concerns
For all of that promise, the dark side of this technological revolution is increasingly coming under scrutiny. These include concerns that artificial intelligence is disrupting the software sector, worries about worker displacement and even fears that AI will induce a race to the bottom that in the final accounting impoverishes all parties.
Sector disruption
Some business sectors are undeniably at particular risk of being disrupted. Software companies have been rerated substantially lower on the view that new AI models are so proficient at coding that they can replace a substantial portion of what legacy software makers do (see chart below).
Software companies are underperforming relative to S&P 500 as a whole
As of 03/06/2026. Sources: Bloomberg, Macrobond, RBC GAM
Financial markets have even coined a new term for companies that will hopefully prove resilient to AI: “HALO,” for Hard Asset / Low Obsolescence. A few examples are helpful. The core definition is companies that tick both boxes: they own or construct physical assets that are unlikely to be displaced by AI. The transportation sector is a good example – railroads and so on. Even better is when the companies are also set to directly benefit from the growth of AI, such as power grids and data centres.
Hard Asset, Limited Obsolescence companies show strength relative to S&P 500
As of 03/06/2026. Sources: Morgan Stanley, Bloomberg, RBC GAM
A broader definition of HALO would include companies that tick one box or the other. Businesses that extract natural resources have hard assets that are unlikely to be displaced by AI. Conversely, some businesses enjoy low obsolescence by virtue of their barriers to entry, plus deep entrenchment into the economic system. Payment networks, financial exchanges and certification/standards authorities are examples.
But the very existence of a basket of companies that are resistant to AI highlights the fact that many others are theoretically exposed. These businesses include software, call centres, localization work, financial research and tax preparation.
Worker displacement
There is also tentative evidence that AI is indeed displacing some workers, particularly recent graduates in coding-related fields.
But it is not unusual for new technologies to eliminate a subset of the workforce. The concern is if those workers are then unable to find gainful employment elsewhere. So far, that doesn’t appear to be a problem: U.S. initial jobless claims remain quite low and the unemployment rate is normal.
Dystopian fears
The most intense fear about AI is that a worst-case dystopian scenario might play out. In this case, job displacement by AI metastasizes to the point of eliminating a large fraction of white-collar jobs. Consumer demand collapses and the economy circles the drain.
Scenario analysis
What to make of this dystopian picture? It is a possibility, but hardly the only way this could play out over the medium and long run (see flowchart).
How might the impact of AI play out over the medium and long run?
As of 03/08/2026. Source: RBC GAM
We identify three main AI scenarios:
AI proves to be a minor general-purpose technology. A general-purpose technology is an innovation with broad applications across society and across businesses that enables a sustained period of faster productivity growth.
AI proves to be a major general-purpose technology.
AI becomes an unprecedented disruptor.
1 - Minor general-purpose technology
At the cautious end of the spectrum, perhaps large language models and their ilk prove “merely” to be a minor general-purpose technology, akin to the laser or the lithium-ion battery. This is still useful, but it is not revolutionary. We assign a 25% chance to this outcome.
The rate of AI improvement slows significantly, or even hits a developmental dead end. Perhaps glitches prove difficult to overcome, or the quality of information on the internet being used to train the models is sufficiently low that there is an element of garbage in-garbage out. Human ingenuity and institutional knowledge are still significant differentiators relative to what large language models can accomplish.
Productivity growth picks up even in this conservative scenario, but perhaps by only 0.25 percentage points per year. The tech giants that spent massive sums end up with a poor return on their investment, AI adopters are a little better off and the world is not wildly different than today.
2 - Major general-purpose technology
Alternatively, and most likely in our view (with a 45% probability), perhaps AI becomes a major general-purpose technology. This would put it on par with past massive advancements such as the invention of electricity, the internal combustion engine or the computer. AI continues to improve and its applications are both broad and profound.
Under this scenario, there is notable job displacement and some sectors suffer. But the prosperity resulting from AI also creates new sectors and job opportunities.
There is ample historical precedent for this. Improvements in farm efficiency caused the U.S. agricultural share of employment to collapse from 41% in 1900 to 21.5% in 1930 and then to just 12% in 1950, all while food production rose.
Yet unemployment did not go up. Those farmhands moved to cities and broadly got better jobs than they had had before. If a person can do something useful, that has value and historically has found a place in the economy.
Furthermore, it is often forgotten that demand rises when products become less expensive. If AI were to displace 10% of lawyers, a significant fraction of the cost savings would likely be passed onto customers given the realities of a competitive marketplace. Those customers, in turn, would increase their consumption of legal services, with the implication that legal sector employment might “only” decline by 6% instead of 10%.
Some sectors including management consulting, architecture, investment banking, advertising and marketing have such elastic demand that it is theoretically possible that AI does a lot of the work and yet the need for humans goes up rather than down.
Notably, past predictions of technology-driven doom have reliably been exaggerated:
Online shopping was going to kill brick-and-mortar stores due to superior choice and convenience. Yet 30 years later 84% of retail spending still takes place within physical stores.
Video conferencing and other digital tools were going to kill the office. Yet the fraction of American workers who are fully remote merely increased from 6% in 2019 to about 11% in 2026. If anything, that figure may now be in retreat.
Cryptocurrencies were going to disrupt banks in a major way but have barely put a dent in them.
Elsewhere, new technologies have significantly diminished old ways of doing things – but, critically, not wiped them out altogether. Streaming has not completely replaced television. TV did not completely replace radio. Airplanes did not completely replace the train.
China’s cheap labour was supposed to undercut U.S. workers and ultimately did in certain sectors. But, at an aggregate level, U.S. real wages kept rising and the unemployment rate fell over subsequent decades.
And so it may be the case that while AI would seem theoretically capable of disrupting a large fraction of employment and radically recomposing the world, it simply may fall short of that, as per these examples.
In this scenario, AI accelerates productivity growth by a substantial 0.5 to 2.5 percentage points per year, unleashing a golden age of growth. This would parallel what took place in the 1990s through early 2000s with the computer boom. The tech giants earn a solid return on their investment, AI adopters benefit substantially and as with prior technological leaps, unemployment does not permanently increase.
3 - Unprecedented disruptor
Finally, perhaps AI is instead an unprecedented disruptor, completely different from past innovations in the speed of its disruption, the scale of its impact, and the way that it eliminates the high-skill jobs that past technologies tended to create more of. We assign this a 30% likelihood.
This could be really bad, but it could also be really good. Let us evaluate three sub-scenarios that span the range of possibilities.
3A – Dystopian scenario
The dystopian AI scenario is capturing a great deal of attention right now.
Broadly, it envisions that many companies will find that they can replace a significant fraction of their workforce with AI. At the level of individual corporations, this delivers productivity gains and profit margin improvements.
However, at the economy-wide level, a tragedy results. The many unemployed workers must curtail their spending and all businesses accordingly suffer a sharp decline in demand. The decline in demand is sufficiently severe that it swamps the productivity gains delivered by the AI. In the end, both workers and businesses worse off.
A deep recession results. This then necessitates further layoffs, encourages further automation and ultimately creates a vicious loop that ends with the economy in ruins and potentially even AI makers impoverished.
White-collar workers bear the brunt of the initial layoffs given that large language models are most disruptive of knowledge work. This is already a big problem by itself, as the top 10% of U.S. households by income – disproportionately white-collar workers – generate nearly half of all consumer spending. But the goods sector would also reel from declining consumer demand and greater competition from displaced workers, as would the trades.
Human work long ago de-prioritized physical strength and endurance as powerful machines came to dominate such activities. It is unclear how labour would be sorted if intelligence also came to be de-prioritized.
Why might this time be different for workers?
1. Past technological advances have normally reduced difficult, dangerous, tedious and low-paying jobs. Workers have often been glad to move on from them. This time, AI is more likely to replace the interesting, engaging (and perhaps most importantly) high-paying jobs. Workers will be disrupted into jobs down the income hierarchy, rather than up as was often the case in the past.
2. AI is coming at an incredible speed and is seemingly capable of disrupting many different sectors all at once. At the most basic level, whereas computers have long been better than humans at math and at remembering things, they are suddenly threatening to surpass people at sensing (think self-driving, speech recognition) and evaluating complicated and sometimes contradictory information (that’s what the large language models do). Humans don’t have a lot of advantage left.
Thus, whether other jobs do eventually prove available to displaced workers over the long run, it’s unlikely the rest of the economy can absorb them quickly enough or that they are of a high enough quality to avoid significant pain.
3. Unlike most of the prior technologies, AI is cheap once developed and infinitely replicable. The advantage to using a computer rather than a human may be vast, incentivizing a particularly profound shift by companies.
In conclusion, if AI is radically cheaper than workers, disrupts a large number of sectors all at once, and leaves less attractive jobs in its wake, that’s quite a problem for workers. In this scenario, humans are the horses that never found another purpose after the automobile was popularized. The number of horses in the U.S. fell by a factor of approximately five in the 30 years from their 1920 peak. We assign a 3% chance to this scenario.
3B – Utopian scenario
But there are other ways the “unprecedented disruptor” scenario could go.
The utopian version is rooted in the idea that the productivity gains from the unprecedented technological change associated with AI are so remarkable that they create a world of abundance. If productivity gains are coming at 5% or 10% or even more per year, the global economy is doubling in size every decade. By 2050, the economy is something like eight times larger than it is today.
The resulting positive supply shock lowers costs for businesses, inducing falling prices and surging corporate profits. This in turn bleeds into higher real wages. Government coffers are supercharged. Policymakers can then fully compensate the large numbers of displaced workers via a basic income or unemployment transfers.
In this scenario, John Maynard Keynes’ famous 1930 prediction that technological advances will allow people to work a mere 15 hours per week proves correct, and right on schedule. Displaced workers effectively get to retire early or work part-time, enjoying life as they see fit. (In a more pessimistic version, people are rudderless without the purpose and discipline of a job and are not as happy as one would imagine).
People also enjoy access to an unprecedented amount of high-quality knowledge and advice – essentially, the world’s smartest person on every subject. This naturally improves their quality of life. Artificial agents make consumers more efficient, sussing out the best value for discretionary goods, flights, real estate transactions and financial advice.
At the societal level, AI makes breakthroughs in drug development and medical treatments, materials science (battery technology, superconductors, lightweight alloys) and nuclear fusion. Human life expectancy soars and climate change slows.
It is, however, a utopian vision. We assign a 3% chance to this scenario.
3C - Mixed AI outcome
The “Unprecedented Disruption” scenario doesn’t have to be black or white. In fact, by far the most likely of the three sub-scenarios is this one: a mixed outcome. We assign it a 24% chance.
If the world finds itself steering toward the dystopian sub-scenario, governments would seek to put a stop to it. At the risk of oversimplifying, a tax on AI’s “compute” would accomplish this.
This would make using AI relatively more expensive for businesses and render human labour comparatively more attractive than otherwise. It would also generate tax revenue to retrain and at least partially compensate structurally displaced workers.
It isn’t a perfect outcome, but neither is it apocalyptic. Productivity growth would still be quite rapid in this scenario. AI would enable rising prosperity and some of the exciting possibilities discussed in the utopian scenario – despite being somewhat dimmed by the AI tax. Businesses would be disproportionate beneficiaries.
However, displaced workers would likely be somewhat worse off, though this is an inexact science. If the productivity growth is fast enough and government coffers sufficiently rewarded, they wouldn’t necessarily fare poorly.
A significant complication: if some countries fail to tax AI, they could undermine the countries that do via AI-enabled cheap exports. Resolving this might require a measure of protectionism. Or – as the EU is exploring doing with regard to climate taxes – it could lead to import taxes on products that enjoy a more favourable tax regime than the domestic market.
Conclusion
To review, there are quite a number of ways AI could go, as detailed in the following table. The most likely outcome, with a 45% probability, is Scenario 2, a “Major General-Purpose Technology.” To recap, AI materially accelerates the rate of productivity growth, does displace some workers, but doesn’t permanently increase the unemployment rate.
Alternatively, there is a chance AI underwhelms as per “Minor General-Purpose Technology” (25% probability), but also a chance it blows expectations out of the water via the “Unprecedented Disruption Scenario” (30% probability).
Here the greatest likelihood, with a 24% probability, is a more mixed outcome.
Of all scenarios, AI is most likely to emerge as a major general-purpose technology
As of 03/09/2026. Productivity shock is initial increase in annual productivity growth. Source: RBC GAM
-EL
Trump pivots after IEEPA tariffs struck down
In a much-anticipated ruling, the U.S. Supreme Court (SCOTUS) struck down the Trump administration’s broad IEEPA tariffs on February 20. As expected, the White House quickly pivoted to a separate tariff authority to re-impose a global 10% baseline tariff (with important exemptions maintained).
The result? The average tariff rate on U.S. imports has come down (see chart), at least temporarily, and emerging market (EM) trading partners are the main beneficiaries. The new tariffs already face their own legal challenges, and other key questions remain unresolved.
IEEPA ruling and Section 122 replacement nets out to a lower effective tariff rate
As of 03/02/2026. Sources: Bloomberg, RBC GAM
While the timing was unknown, the decision itself was widely expected. Market odds of a ruling in the White House’s favour fell to around 20% when SCOTUS began hearing the case and conservative justices sounded skeptical of arguments from the Department of Justice (DoJ). Indeed, the 6-3 ruling saw three conservatives side with the Court’s three liberals, although they differed in their reasoning for the ruling.
Ultimately, they agreed that the International Emergency Economic Powers Act (IEEPA) does not authorize the use of broad tariffs. The decision took effect immediately.
Recall, IEEPA was the legal foundation of the administration’s Liberation Day reciprocal tariffs (establishing country-level baseline rates), and fentanyl tariffs levied on imports from Canada, Mexico and China. Those tariffs accounted for about half of overall tariff revenue in recent months and nearly three-quarters of the new tariffs imposed during Trump’s second term.
Section 232 sectoral tariffs and pre-existing Section 301 tariffs on some Chinese imports were not affected by the decision. Without IEEPA, the effective tariff rate on U.S. imports falls to 7.6% from 13.6% previously (prior to any replacement tariffs).
We’ve been cautioning that the administration has other tools to maintain a relatively high effective tariff rate if the IEEPA tariff were struck down. That’s exactly what happened. Trump signed an executive order just hours after the decision, imposing a 10% global tariff under Section 122 of the Trade Act of 1974. That legislation allows for a tariff of up to 15% for up to 150 days to address balance of payments issues. Congressional approval will be needed for any extension.
There was some initial relief that the administration opted for a baseline tariff of just 10%, but that subsided the following day when Trump posted on social media that the rate would be increased to the maximum 15%. But there has been no executive order imposing that higher rate, and a 10% tariff went into effect on February 24. In early March, Treasury Secretary Bessent suggested the administration plans to increase the rate to 15% soon.
Importantly, the executive order maintained key exemptions that had also applied to IEEPA tariffs:
USMCA-compliant goods
products subject to separate Section 232 tariffs
more than 1,500 products.
All are spared from Section 122 tariffs. The result is an effective tariff rate of 10.6% if the 10% baseline is maintained, or a 12.1% rate if a 15% tariff is eventually imposed.
Again, that’s down from 13.6% under IEEPA. The 10% rate would represent a roughly $8 billion reduction in monthly tariff revenue (from a run rate of around $30 billion under IEEPA). The 15% rate would cause a $4 billion shortfall.
Replacing country-specific IEEPA tariffs with a flat baseline tariff generally levels the playing field across trading partners, although exposure to Section 232 tariffs and exemptions still causes some differentiation. Broadly speaking, emerging market economies that generally faced higher rates before will benefit the most from a flatter tariff structure (see chart). Brazil sees the most relief, as it was subject to a 50% tariff on some products under IEEPA. Other EM trading partners that faced 19-25% rates also disproportionately benefit.
IEEPA ruling and 10% replacement lowers U.S. effective tariff rate, helps emerging markets
As of 03/02/2026. Sources: Evercore ISI, RBC GAM
Having already conducted a Section 301 investigation into China – used to impose tariffs during the first Trump administration – the White House could theoretically re-impose IEEPA-level tariffs on Chinese imports, but hasn’t done so yet. It’s speculated that the administration will hold off ahead of an April meeting between Trump and Xi.
At a 15% rate, not much would change for major developed market trading partners like the European Union (EU) and Japan that had agreed to a similar baseline rate in framework deals last year. They could face a slightly higher effective rate if Section 122 tariffs aren’t de-stacked from pre-existing ‘most-favoured nations’ (MFN) tariffs (their trade deal rates de-stacked under IEEPA).
However, we think a solution is likely as the U.S. seeks to keep up its end of those bargains. If a 15% baseline is implemented, the UK would need special treatment to maintain its 10% deal rate.
Without domestic legislation imposing higher tariffs, trading partners can’t voluntarily agree to pay higher tariffs under the framework agreements negotiated last year. And losing IEEPA as a tariff tool limits the administration’s ability to threaten higher tariffs and retaliate against non-compliance.
Yet the White House is likely to undertake Section 301 investigations that would eventually restore the power to impose higher country-level tariffs – as well as ongoing authority to impose Section 232 sectoral tariffs. So we think other countries will stick to their commitments to lower import barriers and invest more in the U.S.
Beyond the question of whether Section 122 tariffs are raised to 15%, there are a few other outstanding issues we’re looking for clarity on:
Refunds
SCOTUS didn’t rule on whether importers that paid IEEPA tariffs might be eligible for refunds, leaving that decision to the lower courts. The Court of International Trade (CIT), which initially struck down the IEEPA tariffs, acted quickly. On March 4, a CIT judge ordered refunds for all importers that paid IEEPA tariffs. The DoJ asked for a stay of that decision pending appeal but was denied. The DoJ can still appeal the refund order.
In a subsequent court filing, U.S. Customs and Border Protection (CBP) said it can’t immediately comply with the order due to technological, process and manpower constraints. But it said it could begin issuing refunds as soon as late April after updating its systems.
The filing noted that CBP collected $166 billion in tariff revenue from around 330,000 importers. Some estimates suggest the biggest 1,000 importers paid 80% of those tariffs. In some cases, shipping companies paid the customs duties and subsequently charged their customers. FedEx is now being sued to ensure it will ultimately reimburse its customers if it receives refunds.
While CBP appears to be making efforts to streamline the refund process, there are still hoops to jump through – particularly in cases where tariffs were paid early last year and entries have already been liquidated (i.e., duties finalized). Some smaller importers might not go to the trouble.
Nonetheless, refunds would amount to a sizeable injection of cash into the U.S. economy, worth about 0.5% of GDP. While that’s helpful for importers’ bottom lines, we wouldn’t expect a boost to growth of that magnitude. We don’t expect much of the refunds will flow directly to consumers (who haven’t borne the brunt of tariff costs in any case).
If the full $166 billion had to be repaid, it would theoretically add around 9% to the U.S.’s budget deficit this year – and that’s before adding the temporary loss of future tariff revenue under Section 122 vs. IEEPA. That would modestly worsen the country’s already challenging fiscal position. But we think the refunds could be covered by bill issuance without too much disruption to the Treasury market.
Section 122 legality
Section 122 seems to stand on more solid legal ground than IEEPA tariffs. It clearly allows for temporary tariffs to address “large and serious United States balance-of-payments deficits.” But it has never been used and is already facing its own legal challenges from 24 states.
Trade deficits and balance of payments deficits are conceptually distinct – even White House lawyers argued as much during the IEEPA case. The latter historically refers to a persistent outflow of reserves in a fixed exchange rate system. With a floating exchange rate, the U.S. can’t actually run a balance of payments deficit as long as there is sufficient demand for American financial assets, which there is.
But when hearing the IEEPA case, the CIT seemed to have a slightly different interpretation. As the balance of payments must always equal zero (by its current definition in a floating exchange rate regime) the court suggested balance of payments deficits must refer to deficits within the various accounts comprising the balance of payments, including the trade of goods. That seems to open the door to using Section 122 to address trade deficits.
It will be interesting to see how this plays out in court, but it might be a moot point if it takes more than 150 days to reach a final decision, as seems likely.
After 150 days
Section 122 clearly limits tariffs to up to 150 days unless extended by Congress. Given lack of enthusiasm for tariffs among lawmakers, that seems unlikely. It is possible that, after the initial 150 days, the administration simply claims balance of payments issues remain unresolved and re-imposes Section 122 tariffs.
That would surely be challenged in court. But unless an immediate injunction is granted or Congress fights back, it could temporarily extend the life of Section 122 tariffs.
But that’s not our base case. We think the administration will pursue a more durable and legally sound alternative to the Section 122 tariffs. Trump said additional Section 232 sectoral tariffs are on the way. Investigations into products like batteries and industrial chemicals are reportedly forthcoming. He also said additional Section 301 investigations will be undertaken. These would allow the re-imposition of higher and differentiated country-level tariffs.
Those investigations into other countries’ trading practices typically take half a year or more, but the administration might seek to accelerate that process and ensure it can impose Section 301 tariffs on major trading partners by the time Section 122 expires in late-July. The U.S. government already publishes a 400-page annual report detailing the foreign trade barriers of 59 countries and trading blocs. This could form the foundation of expedited Section 301 investigations.
Still, there is some risk that Section 122 tariffs roll off before they can be sufficiently replaced by alternatives, and the U.S. effective tariff rate drops further. That might be convenient timing politically, with businesses and consumers getting an additional tariff break in the months before midterms.
Overall, the IEEPA ruling was widely expected. Trump appears to be following the playbook (Section 122 tariffs followed by further 232 and 301 investigations) that we and many others anticipated. Some modest, temporary relief from higher tariffs is welcome but we see scope for most IEEPA tariffs to be eventually replaced by more durable authorities.
If nothing else, the bipartisan 6-3 SCOTUS ruling against the administration is a nice reminder of checks and balances within the system, and that executive branch power is not unlimited.
-JN
Gauging tariff impacts to date
We’ve spent plenty of time reviewing country-level tariff rates, analyzing how dependent trading partners are on exports to the U.S. and examining framework agreements struck with the Trump administration last year. Those all get to the potential impact of U.S. protectionism and remain an important aspect of cross-country comparisons.
But tariffs have now been in place for long enough that we can add actual outcomes to our analysis. Did trading partners gain or lose market share in the U.S.? Did they diversify their exports to other countries? How are their manufacturing sectors faring? To what extent have currency moves softened the blow?
We’ve developed a simple scorecard that combines these potential impacts and actual outcomes for the U.S.’s top 20 trading partners. Each country is evaluated across the following eight metrics to come up with an overall score:
1. Tariff rate: the change in the country’s effective tariff rate, measured by actual tariff collection in Q4/25. This does not incorporate the replacement of IEEPA tariffs with a Section 122 baseline rate.
2. Tariff exposure: the dollar value of tariffs levied on a country’s exports as a share of that country’s GDP. This combines the country’s effective tariff rate with its intensity of trade with the U.S.
3. Trade deal: whether the country has struck a framework agreement with the U.S., and the tariff rate negotiated. We think trade deals provide a measure of certainty in addition to any negotiated tariff reductions captured in the two indicators above.
4. Market share: whether the country gained or lost market share in the U.S. after tariffs took effect.
5. Trade diversion: the extent to which the country increased its non-U.S. exports after tariffs took effect.
6. Manufacturing PMI: the level of a country’s purchasing managers’ index (PMI) relative to both its peers and its own history.
7. PMI change: the change in the country’s manufacturing PMI after tariffs took effect.
8. Foreign exchange offset: whether the country’s currency appreciated or depreciated after tariffs took effect. Currency depreciation helps offset the impact of tariffs.
Here’s what the scorecard looks like, with the most impacted countries at the top.
Brazil, China, Malaysia, India, Mexico and Canada most affected by current U.S. tariffs
As of 02/06/2026. Sources: U.S. Census Bureau, S&P Global, International Monetary Fund (IMF), Bank for International Settlements (BIS), Bloomberg, Singapore Institute of Purchasing & Materials Management, Taiwan Ministry of Finance, RBC GAM
Perhaps unsurprisingly, EM economies with higher tariff rates are worse off while many European countries with relatively low tariff rates and limited trade with the U.S. fare better in our scorecard. But there is some interesting nuance that emerges from this analysis:
Vietnam’s high potential impact – reflecting a relatively high tariff rate and significant trade with the U.S. – hasn’t translated into particularly bad outcomes. It gained market share in the U.S. and substantially increased exports to other countries. Currency depreciation helped soften the blow of higher tariffs.
Despite facing relatively low 10% baseline tariff rates, the UK and Singapore haven’t had particularly good outcomes. They failed to take advantage of their tariff edge and actually lost some market share in the U.S.
Even with relatively low effective tariff rates, Canada and Mexico’s significant trade exposure and absence of framework agreements limits their potential. That has translated into relatively poor outcomes, with weak manufacturing PMIs reflecting uncertainty around USMCA re-negotiation.
Canada in particular has lost significant market share in the U.S., likely due to its high exposure to sectoral tariffs. While Canada and Mexico sit next to one another in the scorecard, Canada ranks last among developed markets while Mexico is toward the middle of the pack among EM countries.
China and Brazil faced the biggest tariff hikes and are the most impacted per our scorecard, but it’s not all bad. Neither economy is heavily reliant on exports to the U.S., and both managed to increase exports to other markets by about 10%.
These scores are not static, of course. As discussed earlier, replacing country-specific IEEPA tariff rates (as used in this scorecard) with a flat Section 122 rate levels the playing field and disproportionately benefits EM trading partners. We’ll have to see whether that potential shift translates into better outcomes.
Also keep in mind that if the administration pursues Section 301 investigations against its major trading partners (durably replacing temporary Section 122 tariffs) an IEEPA-like differentiated tariff schedule could be restored.
We think Mexico and Canada, currently sitting #6 and #7, have the potential to move down the list (becoming less negatively impacted) if USMCA re-negotiation successfully resolves trade policy uncertainty. Mexico has already gained some market share in the U.S. and could pick up more if a renewed USMCA gives companies the confidence to invest in North American supply chains.
For Canada, reducing the impact of damaging sectoral tariffs is key. If punishingly high auto, steel and aluminum tariffs are maintained without some form of quota exemption, it’s hard to see Canada recovering all of the market share it lost last year.
-JN
