~With contributions from Vivien Lee, Aaron Ma and Eric Savoie

Much has happened In the two weeks since we first assessed the war in Iran and the accompanying oil shock.

The situation has broadly become more problematic. The war continues and the Strait of Hormuz remains significantly blocked. Energy infrastructure has been damaged and both parties have threatened to escalate further.

Oil prices have accordingly risen even further (see next chart). However, late-breaking news about a delayed U.S. ultimatum and constructive negotiations have permitted a partial reversal that has sent Brent crude back down to approximately US$100 per barrel.

Crude oil prices soar on escalating tensions in the Middle East

As of 03/23/2026. Sources: Bloomberg, RBC GAM

The situation remains volatile, with a number of ways the war could resolve. A near-term ceasefire that occurs within the next month or so is the optimistic scenario. A ceasefire occurring in several months’ time is the middling scenario. The negative scenario is large-scale damage to energy infrastructure that turns a temporary shock into a long-lasting one.

Our bias remains slightly in favour of the optimistic scenario, though the middling scenario is nearly as probable. The negative scenario is not impossible – though less likely.

There have certainly been a range of negative and positive developments. Let us start with the negative issues and then shift to the positive.

-EL

There are several recent developments that argue for a slower resolution of the war and/or a larger impact on oil prices.

First, and in theory synthesizing the overall situation, betting markets have become less optimistic than they were in the early days of the war (see next chart). To illustrate, the likelihood of a ceasefire by March 31 was initially assigned a 44% likelihood on Polymarket. Last week it fell to just 12% by March 16. It has since rebounded to a still-low 23% as of March 23.

It has been a similar sequence for the likelihood of a ceasefire by April 30. Initially this was assigned a 67% chance. It then dropped to just 40% and has now partially rebounded to 51%. This latter figure highlights that the betting market is sitting on a knife’s edge as to whether a ceasefire will be achieved within the first two months of the war or not.

Polymarket odds point to a more prolonged conflict

As at 03/23/2026. Sources: Polymarket, Bloomberg, RBC GAM

Even after unwinding slightly, the oil futures market continues to anticipate significantly elevated (if declining) prices through to the autumn, followed by lower but still elevated oil prices after that (see next chart).

Oil futures prices are being pushed higher by the week

As of 03/23/2026. Sources: Bloomberg, RBC GAM

The Strait of Hormuz is a key chokepoint for global energy markets with about one-fifth of global oil and liquified natural gas (LNG) supply normally passing through the narrow waterway. To date, approximately 16 ships have been attacked. Six others have reported war-related incidents.

As such, the Strait remains significantly closed more than three weeks into the conflict (see chart). Over 700 vessels are effectively stuck in the Persian Gulf. If the entirety of the Strait’s normal oil supply were removed from the market and not offset elsewhere, the price of oil (WTI) might rise to around US$190/barrel based on standard elasticity assumptions.

Ship crossings in Strait of Hormuz collapse

As of 03/22/2026. Sources: Bloomberg, RBC GAM

Energy infrastructure has been swept up into the conflict. The International Energy Agency (IEA) says that “more than 40 energy assets” have been “severely or very severely damaged” in the Middle East, spread across nine countries.

As discussed earlier, a further intensification of the war in this direction could create long-lasting rather than temporary oil shortages. All parties have actively threatened to further target such installations.

For that matter, even if a ceasefire is achieved in the coming days, it will take a minimum of two weeks to a month – and potentially several months – to get idled production facilities restarted and product shipped to the appropriate destinations.

Indeed, the risk of scorched earth tactics – be it the widespread destruction of energy infrastructure, desalinization plants, airports or other civilian infrastructure – represents a key downside risk, though is arguably a lose-lose proposition for all parties. Such infrastructure cannot be quickly rebuilt. Qatar already reports that the damage to its natural gas liquefaction complex that took out 17% of its production will take three to five years to fully rebuild.

For that matter, even if a ceasefire is achieved in the coming days, it will take a minimum of two weeks to a month – and potentially several months – to get idled production facilities restarted and product shipped to the appropriate destinations. As such, the energy shortage is effectively guaranteed to last a few months even in the quick resolution scenario.

Another point of concern is the asymmetry between plentiful cheap Iranian drones and expensive quantity-constrained Israeli/U.S. interceptor missiles. Two Iranian missiles recently hit key Israeli sites over the weekend, raising questions as to whether Israel and the U.S. are losing their ability to intercept every major attack.

There is a further asymmetry, which is that Iran only needs to be able to hit the occasional ship to render passage through the Strait of Hormuz impractical. It isn’t enough for the U.S. to be able to shield the other 95% of ships from attacks.

The Iranian regime perceives this war as an existential threat to its existence. As such, Iran will be willing to tolerate greater economic and military suffering and may consider more extreme options.

Potentially increasing Iran’s leverage, the country showed in recent days that it has missiles that can travel as far as 4,000 kilometres. That brings nearly all of Europe into theoretical range.

The Iranian regime perceives this war as an existential threat to its existence. As such, Iran will be willing to tolerate greater economic and military suffering and may consider more extreme options. (Conversely, it might also be willing to accept a cease-fire if the country’s political structure is allowed to remain intact).

Certain Shia groups within the Middle East have taken Iran’s side, with Shia forces within Iraq targeting U.S. military bases and assets. This complicates the war.

-EL

There are also quite a number of positive factors and developments to weigh.

Though not a “new” development, it remains helpful that the great majority of the world is strongly incented to seek a resolution to the war. China is the greatest importer of Persian Gulf energy. India is an important customer. The bulk of Asia and Europe are very reliant on a combination of Middle Eastern oil and natural gas.

Other Gulf states are also extremely motivated to secure a resolution as their ability to sell their primary export is crimped, and the attractiveness of their cities for international businesses is being severely challenged. Many have the diplomatic connections to engage in direct communications with Iran.

The U.S. political will to avoid high gas prices at the pump and high mortgage rates (higher bond yields are hurting mortgage rates) should not be underestimated. This is of particular relevance given the White House’s focus on pocketbook issues and with the midterm elections now just seven months away. President Trump’s personal approval rating has fallen notably since the onset of the war (see next chart). The Democrats are suddenly neck and neck with the Republicans in the contest for control of the Senate (see subsequent chart). Indeed the prospect of a Democratic Party sweep across both the Senate and House of Representatives is now the single most likely outcome permutation from the election (see third chart).

Trump’s approval rating has dropped since the start of Iran war

As of 03/23/2026. Sources: Silver Bulletin, RBC GAM

Odds of the Democratic Party taking the Senate have risen since onset of Iran war

As of 03/23/2026. Sources: Polymarket, Bloomberg, RBC GAM

Probability of a Democratic sweep has risen in last few weeks

As of 03/23/2026. D for Democrats, R for Republicans. Sources: Polymarket, Bloomberg, RBC GAM

All of this is to say that the White House must urgently want a resolution that brings oil prices and bond yields back down. Note that there are two different ways to do that:

1. Agree to a near-term ceasefire.

2. Increase the force with which the U.S. is hitting Iran so as to properly “win” the war, neutralizing Iran in the Strait of Hormuz.

The near-term ceasefire recently became more likely after President Trump announced on March 23 that the U.S. had recently had “very good and productive conversations regarding a complete and total resolution of our hostilities in the Middle East.”

Simultaneously, the scenario in which Iran is neutralized via the use of additional force is also being pursued. The U.S. recently significantly degraded Iran’s missile bases via the use of 5,000 pound “bunker buster” bombs and has similarly destroyed further intelligence support sites and missile radar relays. Additional American forces are also being shifted into the region.

While Iran is likely willing to endure more economic hardship than most countries, its capacity for pain is not infinite.

While it remains a matter of conjecture, Iran’s ability to launch missiles and drones may be significantly deteriorating. The country’s missile stock is thought to have already declined by 55% to 75%, with the number of ballistic launchers thought to be down by 60% to 70%.

Conversely, the Iranian drone inventory is still quite large according to expert estimates. But evidence on the ground argues the supply of both missiles and drones may be getting pinched. With the notable exception of Saudi Arabia (where the data remains quite choppy), the daily pace of Iranian missiles and drones directed against other Gulf states – Qatar, the United Arab Emirates, Bahrain and Kuwait – has declined sharply since the early days of the war, and broadly continues to fall.

Using the United Arab Emirates as an example, Iran’s use of ballistic missiles fell sharply within a few days of the start of the war and remains low. Drone deployment was elevated through the first third of March but has since fallen by a factor of approximately five. This could mean that Iran is more limited in its drone supply than popularly imagined. Accordingly, air traffic is gradually reviving across the Gulf.

While Iran is likely willing to endure more economic hardship than most countries, its capacity for pain is not infinite. Its ability to source foreign inputs will surely be challenged by the war itself and also by the stark depreciation in the value of the country’s exchange rate since the start of the year (see next chart), alongside the pain of surging inflation (see subsequent chart).

Iranian rial has depreciated sharply since the start of 2026

As of 03/23/2026. Sources: Macrobond Financial AB, Macrobond, RBC GAM

Iran’s inflation has surged since beginning of 2025

As of January 2026. Sources: International Monetary Fund (IMF), Macrobond, RBC GAM

From a global economic perspective, it is fortunate that there was an excess of oil in the world before this war began. This has meant that roughly half of the increase in the price of oil has merely been from an unusually depressed level to a more normal level. The damage is thus more limited.

Further, as discussed in the next section, the world is importantly finding ways to offset roughly three-quarters of the theoretical impact from a blocked Strait of Hormuz.

-EL

Fortunately, there are a number of mitigants to the theoretical loss of 20 million barrels per day (mb/d) in oil and natural gas liquids that normally transit the Strait of Hormuz (see chart below):

• Saudi Arabia has a pipeline to the Red Sea with approximately 4-6 mb/d of spare capacity.

• The United Arab Emirates (UAE) has a smaller pipeline to the Gulf of Oman (bypassing the Strait) with 700 kb/d of spare capacity.

• Iran is still exporting some oil – last year it accounted for 2.4 mb/d of the 20 mb/d passing through the Strait, even if shipments have been curtailed somewhat during the current conflict.

• The IEA announced plans for a record 400 mb release of crude oil and products from its members’ stockpiles, with experts suggesting a potential 2-4 mb/d drawdown.

• Outside the IEA, China added 1.1 mb/d to its crude stockpile in 2025 (now estimated at 1,200-1,300 mb). It could draw down at a similar pace to manage supply disruptions

• The U.S. has temporarily waived sanctions on Russian oil that could unlock more than 120 mb that is currently on ships or in storage.

There is uncertainty around many of these numbers. However, one can reasonably approximate that nearly three-quarters of the theoretical supply disruption due to closure of the Strait can be made up through other routes and sources of supply. A 20 mb/d shortfall turns into a 5.5 mb/d supply gap, which might be more consistent with US$100/barrel oil – around current prices – rather than US$190/barrel.

Nearly ¾ of Strait of Hormuz shortfall offset by other routes/supply, at least temporarily

As of 03/16/2026. Sources: International Energy Agency (IEA), RBC GAM

There are other potential mitigants as well, like rising production in other regions in response to higher oil prices.

A number of non-Iranian vessels have successfully traversed the Strait and several countries are negotiating safe passage for their ships. For its part, Iran has indicated that the Strait remains open for ships not affiliated with the U.S. or Israel, though it is not entirely clear what that means for the many theoretically neutral Gulf states that host U.S. military bases.

The U.S. is also evaluating potential naval convoys and insurance schemes to get more traffic flowing, but those are operationally challenging, particularly while the conflict is ongoing.

There are also downside risks to some of the alternative routes and sources of supply. The Fujairah port at the terminus of the UAE’s pipeline is not far outside the Strait and has faced intermittent disruptions during the conflict. U.S. strikes on Iranian military bases on Kharg Island – a key export terminal – could threaten remaining Iranian shipments. Energy infrastructure was until recently being increasingly targeted by both sides.

If the conflict drags on, dwindling inventories could push prices higher.

Other supply offsets are only temporary. Russian sanctions waivers are set to last 30 days but could be extended. Drawing down IEA inventories at 3 mb/d would exhaust the record 400 mb supply release in about 4.5 months, although the IEA has suggested scope for even greater stock releases if necessary.

If the conflict drags on that long, dwindling inventories could push prices higher. As the chart below shows, lower stockpiles within the Organisation for Economic Co-operation and Development (OECD) tend to result in a non-linear increase in crude oil prices (adjusted for inflation). Drawdowns by the end of April might only be consistent with US$100/barrel oil, but a further decline in inventories might push prices higher to US$130 after another month and above US$150 by mid-year based on historical trends.

Oil prices to ramp higher as inventories drawn down

As of 03/16/2026. Assumes 3 million barrels per day (mb/d) stock draw starting 03/16/2026. Sources: IEA, RBC GAM

Of course, this assumes other factors are held constant. In reality, higher prices are likely to cause producers to increase production, mitigating some of the inventory drawdown. And demand destruction will slow the pace of inventory drawdowns. Even in a more prolonged conflict – not our base case – we don’t think oil prices would be sustained above US$150/barrel.

Unfortunately for the natural gas market, there are fewer mitigants. Around 20% of global LNG trade normally passes through the Strait. Qatar, the world’s second-largest LNG producer, exports some natural gas via pipeline to the UAE and Oman. But that pipeline has limited spare capacity and Oman’s LNG export facility is already operating close to full capacity.

Despite growing global trade in LNG, natural gas markets remain fairly regionalized with much more scope for price differentials compared with the oil market. With 90% of LNG shipped through the Strait of Hormuz going to Asia and the other 10% to Europe, the price shock is much greater in those regions than in North America where supply is more plentiful. In 2025, 27% of Asian LNG imports and 7% of European imports came through the Strait.

-JN

The current disruption to global oil supply is the worst since dual shocks in the 1970s: the 1973-74 OPEC oil embargo and the 1979-80 Iranian Revolution and Iran-Iraq war. The U.S. entered a recession following both shocks. Fortunately, much has changed since then to make the global economy less sensitive to an oil supply disruption.

It takes less oil to run the global economy than it did half a century ago. The chart below shows oil consumption per $1,000 of inflation-adjusted global economic output, which has declined by 62% relative to the 1970s.

Improving energy efficiency and more diverse sources of energy supply have been key drivers of that reduction. A growing shift toward less oil-intensive services industries has also played a role. Manufacturing’s share of global output has shrunk to 16% today from 27% in 1970.

The global economy has become much less oil intensive over the past 50 years

As of 03/18/2026. Sources: BP Statistical Review of World Energy, UN Trade and Development, Macrobond, RBC GAM

Many countries responded to previous oil shocks by established strategic reserves. This includes the U.S. Strategic Petroleum Reserve (SPR) set up in 1975, which is now the world’s largest stockpile. The IEA was formed in 1974 to coordinate energy policy among OECD countries. Members are required to maintain stockpiles equivalent to 90 days of net oil imports. As discussed above, the IEA’s record release of 400 mb of oil is a key mitigant to supply disruptions in the Middle East.

Oil production has also diversified. OPEC’s share of global output has fallen to 35% from a peak of 50% in 1973. New sources of supply in Alaska, Mexico and the North Sea emerged after the 1970s, and more recently the Canadian oil sands and U.S. shale oil have boosted North American production. The Middle East now accounts for just 30% of global supply.

Looking at the world’s two largest economies, the U.S. is now a net energy exporter, so higher prices have a more neutral economic impact. This benefits producers and their shareholders but can hurt consumers and energy intensive industries. China is a net importer but has accumulated significant reserves and diversified its sources of oil imports. Oil and LNG shipped through the Strait of Hormuz only account for 7% of China’s total energy consumption.

Central banks’ efforts to stabilize inflation expectations also help. Today, policymakers have much more leeway to look through one-time price shocks, rather than tightening monetary policy and intensifying the economic drag from higher prices, as they might have done in the past. The U.S. Federal Reserve (Fed) hiked its policy rate into the double digits during both oil shocks in the 1970s, which contributed to eventual recessions. While higher oil prices might delay rate cuts and even open the door to hikes, there is little appetite to raise rates on that scale today.

-JN

We estimate the approximate economic and inflation ramifications of the energy shock in the table below.

Economic and inflation impacts of energy shock

Permanent $40/bbl oil price increase + 50% natural gas price increase

As at 03/09/2026. Sources: Bloomberg Economics, Organisation for Economic Co-operation and Development (OECD), RBC GAM

But this is an imperfect exercise on several fronts.

1. First, we must make assumptions. These figures imagine a US$40 per barrel increase in the price of oil. In practice, however, the number has fluctuated substantially above and below that figure.

2. We imagine an average 50% natural gas price increase, with a greater rise in Asia and Europe and a lesser increase in North America. That has also fluctuated substantially in real life.

3. Admittedly unrealistically – we hope – this scenario assumes energy prices remain elevated forever. In practice, it is much more likely that they manage a significant retreat within several months.

4. These figures do not factor in other adverse forces such as the negative wealth effect from falling stocks, the negative effect of rising bond yields, or the deleterious impact of greater risk aversion.

For all of those limitations, what can we say about the results? The worst damage, as expected, is to Asia and Europe, where much of the Middle Eastern energy flows (see next chart). The damage is significantly less for the U.S. and Canada, both in terms of gross domestic product (GDP) and inflation. Canada could even benefit from a GDP standpoint given its large oil exports and the U.S. economy does not have to be weaker given its own net energy surplus.

Economic impacts of an energy shock vary by region

As at 03/09/2026. Assumes a permanent $40/barrel increase in oil prices and 50% increase in natural gas prices. Sources: Bloomberg Economics, OECD, RBC GAM

Between our slight preference for an earlier-than-consensus resolution to the conflict, the fact that these estimates are likely too pessimistic if the energy shock is not permanent, and the fact that standard economic models would argue rate cuts are the more appropriate response to a temporary supply-side oil shock, we think major central banks will manage to avoid significant monetary tightening (markets have probably become too hawkish in recent weeks – more on the central bank outlook on a country by country basis in a later section).

The risk of recession is also relatively tame, albeit higher than a month ago. Our preferred one-year-ahead U.S. recession model acknowledges that the increase in oil prices, wider credit spreads and the flatter 2-year to 10-year slope of the yield curve all point to an increased risk of recession versus the February estimate. However, the effect is relatively tame (see next chart). The risk of a U.S. recession has increased from around 6% to 12% over the next year. The risk is clearly higher for the likes of Europe and Japan given their greater exposure and relatively lower potential growth rate, but still far from automatic.

The probability of U.S. recession within a year remains relatively tame

As of 03/20/2026. Based on RBC GAM model which includes financial and macro factors. Shaded area represents recession. Sources: Haver Analytics, RBC GAM

-EL & VL

Given that the war in Iran is evolving quickly and uncertainty is extremely high, it is useful to analyze historical oil shocks to gauge the range of possible outcomes for financial markets.

We have adapted our familiar road map charts to focus specifically on oil shocks, defined as a 50% or greater year-over-year price increase in the price of WTI crude oil. T=0 on the chart is set as the first date that this threshold was reached (see next chart). The chart begins 30 trading days prior to that date, and ends 180 trading days after the oil shock. This helps to visualize historical paths for the price of oil and equities leading into and following such shocks.

Recessions are more likely when the oil shock is intense and long lasting

Oil price through oil shocks - Indexed to 0% at date of event

As of 03/17/2026. Chart captures 14 oil price spikes dating back to 1973. Identified as an oil shock when year-over-year price gain is over 50%. Sources: Bloomberg, RBC GAM

The chart considers 15 oil shocks since 1973, plus the current oil shock (which, as of March 18, had not yet reached the aforementioned 50% increase threshold, but was close). The 15 oil shocks were further sorted into ones that happened around recessions and those that did not, and oil shocks that occurred as a result of supply shortages versus those that were caused by demand surpluses.

The data reveals several important observations:

• Unsurprisingly, recessions are more likely when the oil shock is intense and long lasting.

• If there is no recession and the oil shock is supply driven, the oil price tends to start settling back down just over 60 trading days after the oil shock threshold was breached.

• The median historical scenario for the S&P 500 across all oil shocks lacks clear equity price direction until about the 75-trading-day mark. At that point stocks resume their usual upward trajectory (see next chart).

Stocks tend to rise 120 trading days after the oil shock date

S&P 500 through oil shocks - Indexed to 0% at date of event

As of 03/17/2026. Chart captures 14 oil price spikes dating back to 1973. Identified as an oil shock when year-over-year price gain is over 50%. Sources: Bloomberg, RBC GAM

It is worth noting, though, that the outcome varies substantially depending upon the nature of the oil shock. Stocks do better than average when there is no recession or the oil shock is demand- rather than supply-driven. Conversely, the maximum drawdown from t=0 is about 7% for the median of recessionary oil shocks. The maximum drawdown is approximately 6% for the median supply-driven oil shock.

Importantly, the S&P 500 is usually higher 120 trading days after the oil shock date even if the shock is associated with a recession.

In the current oil shock, the S&P 500 fell over 5% prior before t=0, which is one of the weakest such periods in history. It is ambiguous whether to view that as an unusually astute market that has already priced in the normal stock market decline ahead of schedule, or whether there could still be the normal post-threshold stock market decline to come.

-AM

While energy is the main focus, the Gulf States are also notable global suppliers of a number of petrochemicals. Thirteen percent of global seaborne chemicals trade typically passing through the Strait of Hormuz (see chart). Key products include:

• About one-third of global seaborne trade in fertilizers passes through the Strait. Three-quarters of that is urea – a nitrogen-based fertilizer produced with natural gas. The price of urea has increased by more than 50% YTD in the U.S.

• Saudi Arabia, the UAE and Qatar are among the top 10 producers of sulphur, accounting for nearly 20% of global production. Sulphur is used in fertilizer production, chemical manufacturing, and petroleum and metals refining.

• Qatar accounts for 20% of global helium supply, which has applications in electronics, fibre optics and semiconductor manufacturing. Only 15% of helium is used for “lifting,” i.e., balloons.

• The UAE and Bahrain are fifth and sixth in global aluminum production, accounting for 6% of supply.

• Saudi Arabia accounts for 5% of global plastic polymer production.

The Strait of Hormuz: Critical for global energy trade

Share of global seaborne trade volume passing through the Strait of Hormuz

As at 03/16/2026. LPG = liquefied petroleum gas. LNG = liquefied natural gas. Sources: UN Trade and Development, Clarksons Research, RBC GAM

Disruption to fertilizer supply is particularly concerning and represents a threat to food security in some of the least developed countries that rely more heavily on imports from the Gulf States. More generally, higher fertilizer prices heading into the planting season could put upward pressure on food prices globally.

If sustained, curtailed supply of other products will certainly create supply chain headaches. But we think this episode stops well short of the broadly-based supply chain disruptions seen during the pandemic. Back then, pent-up post-pandemic demand, the confluence of a particularly extraordinary appetite for goods (amid less ability to spend on services), significant fiscal and monetary policy support, pandemic-related manufacturing and shipping disruptions, and an energy price shock following Russia’s invasion of Ukraine, all contributed to widespread goods price inflation.

Aside from the energy price shock, those dynamics are not present today. Shipping costs have only increased slightly (see chart) and at this point we are seeing few signs of broader supply chain stress.

While rising food prices and some idiosyncratic supply disruptions could add a bit more to inflation beyond the immediate impact of higher energy prices, we think the effect will be modest. With only 3% of global container traffic and 2% of dry bulk trade typically passing through the Strait, even prolonged disruptions should mean a much narrower price shock than we saw during the pandemic.

Shipping costs increase amid Middle East conflict

As of the week ending 03/12/2026. Sources: Drewry Shipping Consultants Ltd., Macrobond, RBC GAM

-JN

Last week brought a flurry of monetary policy announcements. All of the G7 central banks kept rates unchanged but the Reserve Bank of Australia (RBA) hiked for a second time in a row. Policymakers noted significant uncertainty regarding the scale and duration of the current energy price shock. Their varied tones and guidance suggest reaction functions differ quite significantly across central banks based on exposure to higher energy prices, the level of inflation heading into the shock and current domestic economic conditions. Here is a brief summary of their comments:

• The Bank of Canada (BoC) was very explicit in its intention to “look through the war’s immediate impact on inflation.” It cautioned that, if energy prices remain high, it wouldn’t let inflationary pressure broaden or become more persistent – the standard inflation expectations caveat. But Governor Macklem suggested that risk appears contained given economic slack and near-target inflation. Despite Canada being a net energy exporter, the BoC noted downside risk to its growth forecasts given a run of soft domestic data.

• The Fed also espoused a look-through approach, but with more caveats. Chair Powell said before the Fed gets to the question of looking through higher energy prices, it needs more evidence that tariff-driven increases in core goods prices are not a persistent source of inflation. Beyond that, he noted that the look-through approach is contingent on inflation expectations remaining well anchored. He also suggested the Fed is mindful that inflation is already above target and consumers have recently experienced multiple price shocks. The Fed’s updated dot plot continued to point to one rate cut this year, but with fewer participants in favour of two or more cuts.

• The European Central Bank (ECB) didn’t advocate a look-through approach but said it is “well positioned to navigate this uncertainty” as it evaluates the duration, intensity and spread of the conflict. Its updated forecasts included baseline, adverse and severe scenarios. The adverse scenario, which is closest to the current oil futures curve, sees headline inflation returning to around 2% in 2027. Core inflation will only get back to target in 2028, albeit with limited spillover into wages. Lagarde didn’t offer policy guidance, but rate hikes would likely be a close call in such a scenario and are more likely if the severe scenario of a more sustained increase in energy prices appears to be playing out.

• The Bank of England (BoE) was the most hawkish of the G7 central banks given its ongoing challenges in getting inflation back to target. The Monetary Policy Committee said it is “alert to the increased risk of domestic inflationary pressures through second-round effects in wage and price-setting” and “stands ready to act as necessary” to ensure inflation gets back to target over the medium term. It seemed to open the door to hiking as soon as its next meeting in April, noting the upcoming six weeks will shed light on the scale and duration of the shock as well as early evidence on its likely course.

• The Bank of Japan (BoJ) held its policy rate steady and struck a relatively balanced tone. However, Governor Ueda said slightly more board members were concerned about upside inflation risks compared with downside growth (and inflation) risks. Among G7 economies, our models suggest Japan is the most exposed to an energy shock given its significant import dependence. The BoJ was already on a gradual hiking path and market expectations are little changed.

• The RBA increased its cash rate at a second consecutive meeting, although it was a close 5-4 decision. Another hike was already expected pre-conflict (albeit in May rather than March). Rising energy prices added to the case for a hike, with the RBA noting “there is a material risk that inflation will remain above target for longer than previously anticipated.” Governor Bullock said the starting point of excess demand in Australia’s economy makes its reaction function different from other central banks.

To sum up, central banks that were already on a tightening path (the BoJ and RBA) are likely to raise rates somewhat further. The Canadian and U.S. economies face less drag from higher energy prices, which should lower the bar to hike in response to higher inflation. The BoC and Fed seem the most inclined to look through this price shock. It helps that North America is largely spared from rising natural gas prices given significant regional supply.

The ECB and BoE face dual oil and natural gas price shocks – and thus even more inflationary pressure. They are closer to raising rates to contain any second-round effects. The 2022 energy price shock following Russia’s invasion of Ukraine clearly looms large for these central banks.

Policymakers noted significant uncertainty regarding the scale and duration of the current energy price shock. Their varied tones and guidance suggest reaction functions differ quite significantly across central banks . . .

Domestic conditions create additional nuance. The BoC sounded more dovish given a soft Canadian economy and near-target inflation. The BoE was the most hawkish given its pre-existing challenges getting inflation back to target.

For all but the BoJ, market pricing has swung significantly more hawkish in recent weeks (see chart) and even accelerated after investors had more time to digest central bank’s comments. We think some of that move is overdone, barring a more prolonged energy price disruption than the current oil futures curve indicates. But as all of these central banks noted, there is significant uncertainty regarding the duration of the conflict and its effect on energy markets. There is thus an unusual degree of uncertainty regarding the future path for monetary policy.

Iran conflict and oil shock causing markets to price out cuts, price in hikes

As at 03/20/2026. RBA latest adds back 25 bp hike delivered in March for a more direct comparison with pre-conflict. Sources: Bloomberg, RBC GAM

S&P 500 earnings season has largely wrapped up so it’s time for our quarterly look at how management commentary and analyst Q&A is evolving with respect to key economic themes. The table below provides our usual summary of the most common keywords mentioned on earnings calls and the biggest changes relative to a year ago.

Changes in top keywords mentioned on earnings calls

Sources: Bloomberg, RBC GAM

This quarter we used RBC Assist Pro, a generative AI tool, to supplement our analysis including further identifying the most relevant and emerging themes; tracking prevalence, tone and sentiment; and pulling representative quotes from earnings calls for additional context. None of this report was written by AI.

AI and machine learning

Unsurprisingly, AI and machine learning remains a key focus on earnings calls. This theme ranked fourth in overall mentions (after more generic financial terms) and saw by far the biggest increase in mentions relative to a year ago.

Similarly, our AI tool ranks this theme’s prevalence as very high and accelerating. Sentiment remains broadly positive with words like “acceleration,” “transformation” and “generational opportunity” being used alongside.

AI CapEx remains a prominent theme. Consensus expectations for 2026 spending by the big five hyperscalers have already been revised up to $677 billion from $536 billion at the start of the year on the back of strong guidance by tech companies. While the tech sector accounted for half of AI mentions in the latest quarter, 7 of 10 other sectors saw talk of AI and machine learning more than double relative to a year ago, led by financials, industrials and health care (see chart).

Mentions of AI and machine learning continue to rise, not just in tech

As at 03/18/2026. Sources: Bloomberg, RBC GAM

Our AI analysis suggests roughly half of companies mentioned AI in the latest quarter. Separate analysis of earnings calls by Bloomberg found half of industries noted strong AI CapEx or deeper integration of AI within operations. On the latter, AI adoption outside of the tech sector is driven by cost mitigation and operational efficiencies, while consumer-facing industries are using AI to boost revenue.

A regular survey by McKinsey also points to rising AI adoption and finds innovation is the most oft-cited benefit (nearly two-thirds of respondents seeing a positive impact over the past year). Improvements in cost, profitability and organic revenue growth were less widespread, reported by 33-38% of survey respondents.

Functions like software engineering, manufacturing and IT saw the most widespread cost declines while revenue gains were more prevalent in marketing and sales, strategy and corporate finance, and product development. These effects were generally modest, though. Most respondents that identified cost savings saw reductions of less than 10%, while many of those seeing revenue gains pointed to increases of 5% or less.

AI adopters could see more notable benefits to their bottom lines as the technology is more fully integrated across organizations. The McKinsey survey finds about one-third of respondents using AI were doing so in four or more business functions, more than twice as many as in early-2024. Notably, Bloomberg finds the most common AI-related theme in the latest quarter was that savings would accelerate over the next 1-2 years.

Tariffs and margins

Tariffs saw the biggest year-over-year decline in mentions. Our AI analysis finds tariffs remain a highly prevalent topic, but less so than in the first half of last year when trade policy uncertainty was at its peak. Sentiment around tariffs has become less negative. The tone has shifted from highly urgent to more routine with management using terms like “manageable headwind” and “successfully mitigated.”

Companies have adapted to tariffs through pricing, supply chain diversification and cost management strategies. Margins were the number three topic of conversation, but mentions were down slightly relative to a year ago. Bloomberg finds that while tariffs impacted margins across a number of sectors – and tariff costs are expected to remain elevated in 2026 – many firms were able to soften the blow with cost controls and finding efficiencies elsewhere.

On pricing, we are starting to see renewed evidence of tariff pass-through as businesses restock inventories and perhaps look to further mitigate margin compression. We’ve previously noted that importing firms absorbed a surprisingly large share of tariff costs (about half) with only one-third passed along to consumers and the remainder absorbed by foreign exporters (via price cuts). But that math is starting to shift with more costs being passed onto consumers and less retained by importers. We estimate little change in foreign exporters’ share.

This is evident in a renewed pickup in consumer prices of some of the most import-intensive products, after tariff impacts seemed to be fading late last year. Retail and wholesale margins, as measured by the Producer Price Index, are also seeing a renewed increase (see chart), supporting the idea that pass-through to consumers is on the rise.

Trade services margins have returned to their historical upward trend

As of February 2026. Sources: U.S. Bureau of Labor Statistics (BLS), Macrobond, RBC GAM

Affordability

Next to AI, affordability saw the biggest increase in mentions relative to a year ago. Many companies have mentioned bifurcated consumer demand consistent with the K-shaped dynamics we’ve discussed previously. Management teams noted consumers are trading down to lower price points. Bloomberg suggests some companies reported reduced pricing power with price-sensitive consumers unwilling to absorb higher prices – although this seems inconsistent with growing evidence of tariff pass-through discussed above.

Our AI analysis finds sentiment regarding consumers tilted slightly negative in the second half of last year with growing references to consumer caution. There was a slight improvement (albeit still net negative) in sentiment early this year and a less urgent tone on consumer demand. Bloomberg suggests recent management comments generally point to resilient consumer demand.

Headwinds vs. tailwinds

Relative to a year ago, mentions of tailwinds picked up in the latest quarter while talk of headwinds was down significantly (second only to tariffs). Mentions of an economic slowdown fell significantly relative to Q2/25 when recession fears were at their peak. Similarly, Bloomberg finds that two-thirds of industries reported accelerated economic activity amid improved clarity on tariffs, rising AI adoption and lower interest rates.

Most earnings calls took place prior to the escalating conflict in Iran which has emerged as a key downside risk to the economic outlook. Our AI tool suggests geopolitics remained a highly prevalent topic of discussion with elevated urgency, and found growing concern expressed on calls that took place in March, after the conflict began.

We’ve noted a number of positive tailwinds for the U.S. economy in 2026:

• less restrictive monetary policy (including the lagged effects of last year’s easing)

• fiscal support (particularly a big 2026 tax refunds for consumers)

• stock market wealth effects (which also act with a lag)

• a weaker U.S. dollar, AI CapEx and stronger productivity growth.

That said, energy prices have abruptly shifted from a tailwind to a headwind and jeopardize some of those other supportive factors – equity markets are off their highs, rate cuts look less likely with hikes potentially on the table, and the U.S. dollar has appreciated. But some of those supports remain notable (tax refunds will give consumers scope to maintain spending amid higher prices at the pump, productivity growth remains robust) or have even increased (again, AI CapEx estimates continue to be revised higher).

-JN

As discussed in a previous #MacroMemo, the leap in capabilities of recently released AI models and their variants designed specifically for coding sent shockwaves through the software sector. It also impacted other companies whose business models might be susceptible to disruption.

Recent testing illustrates this jump in performance. Model evaluator METR looks at the longest duration task (measured by human expert completion time) an AI model can complete with at least 50% accuracy. One of those new models, Anthropic’s Claude Opus 4.6, achieved a score of 12 hours, up from the previous generation’s score of less than 5 hours measured only three months earlier. See the next chart and note that exponentially rising scores show up as a linear trend when using a log scale.

AI models are becoming exponentially more capable

As at 03/20/2026. Sources: Model Evaluation and Threat Research (METR), RBC GAM

The pace of improvement shown in the chart above has accelerated relative to prior years. METR scores are rising about 6.4x per year since 2024 (the period shown). That’s up from 3.6x annually between 2020 and 2024.

While METR largely evaluates software engineering, machine learning and cybersecurity tasks, a broader measure of 37 benchmarks compiled by Epoch AI shows a similar pickup in growth. Frontier model capabilities improved nearly twice as fast over the past two years compared with the two years prior. This acceleration coincides with the growing prevalence of reasoning models and reinforcement learning techniques in model development.

That’s cause for enthusiasm (or fear), but some context is required. Again, METR generally looks at tasks related to software development. Model builders have focused their efforts on coding to further accelerate model development. Indeed, previous generations helped to train the latest Anthropic and OpenAI models. But these models’ coding capabilities might not be representative of potential proficiency in other areas.

The accelerating pace at which frontier models are mastering challenging new benchmarks is an encouraging sign that AI capabilities continue to steadily improve.

METR also notes that an 8-hour-plus time horizon (i.e., a typical workday) doesn’t suggest all white-collar jobs can be fully automated. The time horizon measurement is more representative of specific tasks a new hire or remote internet contractor might accomplish, not what a human professional can do during an 8-hour workday. Most jobs are not composed of well-specified, algorithmic tasks evaluated by such benchmarks.

Non-coding benchmarks also tend to be narrow and task-specific. They are typically set at an achievable, “within reach” level. Benchmarks that are too easy or overly difficult don’t offer useful insights for developers.

As a result, benchmarks generally “saturate” and become obsolete within a few years, with models achieving such high scores that the tests no longer differentiate top performers. Still, the accelerating pace at which frontier models are mastering challenging new benchmarks is an encouraging sign that AI capabilities continue to steadily improve.

Another leap forward in model development could further expand AI’s capabilities.

New benchmarks are being developed to better assess how models perform in real-world, non-coding tasks of economic value. But Epoch AI suggests the tasks evaluated by these benchmarks are still well-defined and relatively self-contained. Examples include providing a price quote for a prospective client or digitally rendering interior design options.

Progress on these benchmarks wouldn’t support full automation of white-collar jobs but would instead point to a shift in how those jobs will be done in the future. For example, AI could speed the manual execution of tasks but leave planning, delegation and verification to humans.

Of course, another leap forward in model development could further expand AI’s capabilities. The next step change in model performance could come from forthcoming models trained on Nvidia’s Blackwell Graphics Processing Units (GPUs). Reports suggest DeepSeek is close to releasing a model that was developed using smuggled Blackwell chips, in defiance of U.S. export bans to China.

DeepSeek made waves in early-2025 with a powerful model that was developed on a shoestring budget, although it was allegedly trained on the output of other, more expensive models. That episode raised questions about U.S. leadership in AI model development. But American frontier models still outperform Chinese models, which tend to lag by 7 months on average (see chart).

Chinese AI models are catching up vs. U.S., but still several months behind

As at 03/19/2026. The Epoch Capabilities Index (ECI) combines scores from many different AI benchmarks into a single ‘general capability’ scale. Sources: Epoch AI, RBC GAM

As last year’s DeepSeek episode highlights, there are other elements to AI performance beyond benchmark scores. These include the cost of training and running models, their computational efficiency and the energy they consume. Here is a brief rundown of recent trends on those fronts:

• The compute required to train frontier models has increased by 5x annually since 2020. Training costs are growing 3.5x. Power requirements are doubling every year. When compared to the 6.4x annual performance improvement averaged in recent years, it would appear that AI capabilities are enjoying a rising rather than diminishing return on investment. The AI models are improving more quickly than the additional compute being dedicated to them. The cost needed to generate a fixed quantum of model improvement is falling even more significantly.

• For a fixed level of model performance, inference costs (i.e., using trained models to make predictions on new data) are declining by 40x per year. For some tasks, the annual reduction is several magnitudes greater.

• Reasoning models that contributed to the performance acceleration discussed above have significantly higher inference costs, compute and energy needs relative to standard large language models.

• Smaller models are becoming more powerful. The training compute required to achieve given performance benchmarks is declining by 3x annually.

Exponential growth in compute and the significant increase in energy requirements to run frontier models speaks to hyperscalers’ push to build out additional data centre and power generation capacity, to ensure the current pace of development can be sustained or accelerated. Meanwhile, declining inference costs and improving computational and energy efficiency of smaller models should support broader diffusion of generative AI. Model capabilities are certainly a key driver of adoption but the business case for AI will also depend on how costly it is to implement.

Overall, the recent improvement in model performance is encouraging but needs to be viewed in the context of still relatively specific benchmarks and well-defined tasks. These trends seem more consistent with the “major general-purpose technology” scenario laid out in our previous report rather than the “unprecedented disruptor” path.

That said, the exponential pace at which many of these variables are shifting and the potential for further performance leaps suggests we should keep our minds open to a wide range of potential outcomes.

-JN

In March, the Trump administration launched new Section 301 investigations into unfair trade practices of foreign governments. This continues to follow the playbook we anticipated after the Supreme Court struck down the administration’s IEEPA tariffs in February, albeit with some unexpected nuance.

The first step was to temporarily impose a baseline tariff using Section 122. Trump signed off on this just hours after the Supreme Court’s ruling. The administration opted for a 10% rate and immediately threatened a higher 15% rate (the maximum allowed under Section 122) but has yet to follow through. With Section 122 tariffs only lasting 150 days and already facing legal challenges, the White House needs more durable authorities to sustain a higher tariff rate.

Section 301 gives the administration the most leeway to reimpose the country-specific tariffs that were levied under IEEPA. This is the tool Trump used to impose most tariffs on China during his first term. We expected expedited investigations into the U.S.’s top trading partners to enable Section 301 tariffs by the time Section 122 expires in late-July, but the administration appears to be going down a slightly different route.

Rather than investigating specific countries, the United States Trade Representative (USTR) will investigate broader issues. These include excess manufacturing capacity (potentially supported by governments’ industrial policies) and whether countries are doing enough to restrict imports made with forced labour. The former covers 16 countries (most of the U.S.’s top trading partners). The latter will investigate 60 countries (including the 16 in the other investigation).

The White House has suggested other Section 301 investigations could be forthcoming. This approach seems to allow for even broader tariff coverage in a short time frame. Section 301 requires an investigation and consultation process, with public hearings set to end in early-May. That appears to give the administration plenty of time to get tariffs in place by July.

It's not a foregone conclusion that Section 301 tariffs will replicate those imposed under IEEPA. The USTR’s investigations could deliver somewhat different results. But with the administration suggesting it has the tools to restore much of the tariff revenue lost due to the IEEPA ruling, some retracement of the recent decline in the U.S.’s effective tariff rate seems likely over the summer (see chart).

U.S. effective tariff rate has dropped

As at 03/12/2026. Sources: Bloomberg, RBC GAM

-JN