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

Rapid-fire economics

We begin this #MacroMemo with a few rapid-fire economic items:

Monthly webcast

Our economic webcast for July has now been released, with the title “Iran deal, resilient economy, choppy markets.” It provides a good overview of key recent developments and some selected themes.

Bad is good

Just a few months ago, financial markets were concerned about the prospect of economic weakness resulting from the energy shock.

But the macro data, particularly in the U.S., has mostly been stubbornly good. When combined with the added inflation that very much did materialize due to the energy shock, financial markets have now pivoted: they are arguably more worried about too much economic strength (the economy overheating) than too much economic weakness.

In this spirit, when a few prominent U.S. numbers arrived below expectations over the past week, the S&P 500 actually celebrated slightly. Non-farm payrolls for June managed a below-consensus (though entirely reasonable) 57,000 net new jobs (see next chart). The Institute for Supply Management (ISM) Manufacturing Index slipped from 54.0 to a still-good 53.3. The goal at this point is steady, sustainable economic growth – not a spectacular acceleration.

U.S. job creation in June was below expectations

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

Iran peace deal

In line with our commentary in the June 16 #MacroMemo, the peace deal between the U.S. and Iran is mostly holding. This is admittedly a generous claim: there have been flare-ups, with Iran attacking a few ships transiting the Strait of Hormuz, both countries briefly launching attacks against the other and Israel still engaged in Lebanon. But neither country has abandoned the deal, both are still negotiating toward an enduring solution over the 60-day window now underway and – importantly – the Strait of Hormuz has begun to creak open.

Tanker traffic through the Strait has revived substantially and greatly exceeds any other time since the war began, though it is not yet normal (see next chart). Roughly one-quarter to one-third of the normal vessel flow has returned in recent weeks.

Tanker crossings in Strait of Hormuz start to revive

As of 07/06/2026. Sources: Bloomberg, RBC GAM

While the chart shows a recent dip in west to east traffic, we believe this is because the first queue of ships waiting to leave has now been resolved. The fact that the east to west volume continues to rise argues that energy exporters have not lost confidence in the navigability of the Strait: it would appear they believe they can fill these vessels and get them out of the region later. We look for the volume of ships in both directions to resume rising as oil and gas production recovers, as shippers grow more confident in their ability to transit the Strait and as insurance costs fall.

The oil market certainly appears to have high confidence in a constructive outcome, as per the fact that the price of oil has fully returned to its pre-war level (see next chart). Perhaps that’s a slight exaggeration since the risk of a war with Iran had been rising before the official February 28 onset. One might therefore claim that the true pre-conflict oil price is more like West Texas Intermediate (WTI) US$60 per barrel rather than the current US$68. But the fact remains that the oil market has largely priced in the return of normal oil flows. Our base-case forecast agrees with this optimistic assessment.

Crude oil prices retreat on reopening of Strait of Hormuz

As of 07/03/2026. Sources: Bloomberg, RBC GAM

However, we worry that markets are underestimating the downside risk that the peace deal flounders. As evidence of this risk, note the recent military flare-ups, the fact that the flow of ships has not fully normalized and the fact that there are still critical details to sort out over the next 45 days or so, including:

• Will the Strait be tolled by Iran?

• What will become of Iran’s nuclear materials?

• Where will the reported US$300 billion of investments into Iran come from?

• Will Israel be required to fully disengage from Lebanon?

These are sticky issues and so there is a real chance this drags on for longer.

Still, pragmatism is likely to win out and an imperfect peace deal will likely allow the oil and gas (and fertilizer, and helium, etc.) to flow. In turn, we would flag the prospect that central banks don’t need to raise rates quite as much as markets have recently been thinking, as there is scope for inflation to begin settling fairly quickly.

-EL

The economics of space

“Earth is the cradle of humanity, but one cannot live in the cradle forever.” – Konstantin Tsiolkovsky

Not since the 1960s has the public imagination been so energized about space. The past several years have brought about a space renaissance, with notable achievements that include:

• the development of reusable rockets

• a rise in space tourism

• a return to the moon

• the launch of more than 10,000 mostly communication satellites.

Private sector firms are now taking the lead in many regards, with none more prominent than SpaceX. SpaceX went public just under a month ago at an eye-watering valuation.

Given that humankind has been able to produce a remarkable US$124 trillion per year of economic output while largely constrained to the spinning rock known as planet Earth, what might be achievable if the sky is no longer the limit?

The incredible US$2.1 trillion valuation of SpaceX – constituting an almost unfathomable 58 times the company’s projected 2026 revenues – indicates that many feel very optimistic indeed about the extra-planetary economic outlook. (Though it is important to appreciate that a large fraction of SpaceX’s expected revenue will come from terrestrial AI with no direct space connection. The company is substantially more than just a bet on outer space.)

SpaceX is not the only prominent space-oriented company. Firms such as Blue Origin, Rocket Lab, Planet Labs, Redwire and Amazon are also heavily involved. So are traditional defence contractors including Lockheed Martin, Northrop Grumman, RTX and Boeing. Myriad other upstarts are specializing in satellite communications, Earth observation, propulsion technologies, and more.

Market size and viability

The World Economic Forum and McKinsey estimated in 2024 that the space economy could grow from approximately US$630 billion in 2023 to US$1.8 trillion by 2035. That’s an approximate tripling in just over a decade, representing a 9% annual compound growth rate.

Certain space technologies are already in broad use: GPS systems improve ridesharing, shipping, aviation, logistics, agriculture and more. Satellites already enhance television broadcasting, communications and defence applications. Some militaries can launch intercontinental missiles that travel through space.

Gazing to the future (and upward!), one can imagine many additional space-oriented economic applications, discussed next.

A key challenge for many of these aspirations is the high cost of launching objects into space. Slipping the surly bonds of Earth is not easy or cheap. Happily, the launch cost per kilogram has already declined tenfold over the past two decades. A new study by Google argues that it may prove technically possible to continue reducing that cost at a similar clip – a 20% improvement per year – through 2035.

By the authors’ estimation, for data centres in space to be cost-competitive with their terrestrial equivalents, the transportation cost would need to decline by another 18-fold, from US$3,600 per kilogram to US$200 per kilogram.

While possible, we caution that, in practice, this represents something of a best-case scenario: the great leap to reusable rockets has already been made, so cost-savings of that magnitude would require industrial-scale launching of SpaceX Starships (or equivalent) and the delivery of steady as-yet-undiscovered efficiency gains.

Near-term space applications: < 5 years

1) Launch research and exploration missions for national space agencies

Likelihood: Very high

Private-sector entities like SpaceX already earn government contracts launching NASA and equivalent research and exploration missions into space. There is no reason to believe this will stop. Declining costs could even encourage an increase in activity.

2) Space-oriented defence spending

Likelihood: Very high

As military spending rises around the world and satellite technology simultaneously improves, space-based defence applications are likely to proliferate. Satellites guide missiles and drones. Space-based cameras track enemy activities. Intercontinental missiles rely on rocket technology.

3) Enhance satellite communication

Likelihood: High

The number of Starlink and similar satellites in the sky has exploded to more than 10,000 over the past several years. Together, they provide telecom and internet services that theoretically span the entire world. Currently, primary access is via a bulky Starlink terminal to facilitate home internet service in remote areas, with limited low-bandwidth emergency usage in partnership with existing telecom providers direct to mobile phones.

There are now hints of a SpaceX phone that could communicate at a reasonable throughput and latency directly with the satellites, bypassing existing telecom providers. This could be quite useful in low-density and impoverished parts of the world and could – if the cost becomes competitive and throughput improves substantially – even compete with the terrestrial telecom providers.

4) Space tourism

Likelihood: High

Space tourism is a tiny business at present and is unlikely to ever be a central revenue driver. But it could grow as launch costs fall, reliability rises and destination options increase.

5) Cargo transport to the Moon

Likelihood: Medium

NASA has contracted SpaceX to develop large cargo landers for moon missions, with the first trip planned for 2028. This might become a much more important business over the medium and long run.

Medium-term space applications: 5-25 years

1) Space-based data centres

Likelihood: Medium

With the AI race at a fever pitch and data centres in high demand, some propose to construct space-based data centres. This is highly attractive from an energy standpoint. Electricity shortages on Earth could be sidestepped via giant solar panels spanning potentially hundreds of square metres per rack. Tests of the concept are expected as early as 2027 and SpaceX filed plans in January – remarkably – for an orbital data centre of up to one million satellites. But the SpaceX IPO filing acknowledges that this is an early-stage unproven technology that may or may not become commercially viable. As discussed above, the launch cost first needs to fall by 18 times to become commercially viable.

The technical demands of the data centres themselves are no less daunting. Challenges include:

• Assembling and then maintain the enormous and complicated system in space.

• Cooling the system (this is hard to do in a vacuum and proper heat dissipation might require a separate system as enormous as the solar panels).

• Shielding the system from space radiation.

• Whether these floating data centre racks could communicate sufficiently quickly and clearly with their peer data centre satellites and with the ground.

And for all of that, there is a real obsolescence risk as well.

2) Lunar resource extraction

Likelihood: Medium

To facilitate regular space travel beyond the Earth and moon, it will probably become necessary to extract resources from the moon. Fuel, oxygen and water are all especially prominent needs and the moon’s ice caps should prove capable of supplying all three. If a lunar base is to be established, extracting construction materials will also be important – lunar soil, iron and aluminum, prominently.

3) In-orbit manufacturing

Likelihood: Low

In-orbit manufacturing is an idea that sounds absurd on the surface. Why launch something all the way to space just to put it together? But it could have specific applications.

• In space, there is no sedimentation, buoyancy or convection. 

• The microgravity of space allows materials to mix more uniformly and to be applied more finely.

• Alloys and composites have fewer defects.

• Space is a better vacuum than even specialized facilities on Earth, reducing contamination.

• Crystals can grow larger and be more perfect.

• Ultra-large objects can be manipulated more easily in space (though, realistically, most would have to remain there after assembly).

There is certainly value in this, though whether it outweighs the inconvenience and cost of doing it all in space is not presently clear.

4) Point-to-point travel on Earth

Likelihood: Low

Much as an intercontinental ballistic missile ventures into space on its journey from the launching country to its destination, it is theoretically possible to travel from New York to Shanghai in under 40 minutes using a spaceship. This greatly outpaces even the fastest supersonic aircraft, potentially upending the airline industry. Of course, challenges abound.

• There are not presently launchpads near major cities.

• The exceptional noise would require the launch to occur far from human habitation (costing significant time).

• Anything other than perfect weather would be a problem.

• The cost would be quite high.

• Safety would have to improve and the passenger experience would be poor (multiple g’s of acceleration).

More practical might be military logistics, where the time-sensitivity might be great, but the cost of failure is less catastrophic.

Long-term space applications: >25 years

1) Asteroid mining

Likelihood: Low

Whereas mining the moon is a medium-term space application, mining asteroids is probably somewhat further off given their greater distance, lack of gravity and complex structures. The mining might also be motivated by the need for basics such as fuel, oxygen, water and metals. Eventually, exotic materials including precious metals might be sought, though this would risk collapsing the market for such materials on Earth if discovered on a sufficiently large scale. The SpaceX filing describes space/asteroid mining as a “multi-trillion-dollar economic opportunity”.

2) Large-scale colonies

Likelihood: Low

In contrast to small-scale bases, one might imagine humans forming permanent large-scale colonies on the moon and eventually Mars. This would require enormous space transportation, mining, construction and manufacturing capabilities. The motivation for significant colonies is hard to understand from the current vantage point – the unique challenges of space would probably render them financially reliant on the Earth and life would be hard for the residents. This probably caps the size and number of any such settlements.

3) Protecting humanity

Likelihood: Very low

Space-based technologies of various descriptions already protect humanity. Satellites are instrumental for tracking and predicting natural disasters such as hurricanes and for monitoring climate change. The telescopes deployed to space are also useful for tracking threatening asteroids, anticipating solar flares and even searching for alien communication.

Taking a page from science fiction films, the technology already exists to redirect an asteroid away from the Earth if ever necessary. But what about the ultimate protection?

Over the long, long run, continuing to develop space capabilities might allow humans to avoid extinction should some catastrophic event befall the Earth. As science fiction writer Larry Niven once said, “the dinosaurs became extinct because they didn’t have a space program.” But it is one thing to operate a colony reliant on Earth, and something else to run one that is fully self-sufficient and possibly outside of the solar system.

Conclusion

It is fascinating and exciting to think about the economics of space. There will no doubt be more to say as developments unfold. Here is our first and highly tentative conclusion:

Most of the near-term space applications cited above seem achievable, and so the industry is well positioned to grow in line with the high-single-digit expansion rate predicted by the World Economic Forum and McKinsey.

But the medium-term applications are of highly uncertain viability and scale. They require significant efficiencies to be secured, major technological leaps and entirely new industries to form. Many will probably fizzle.

Meanwhile, the long-term applications require all of that – efficiencies, new technologies, new industries – plus remarkable leaps of imagination and investments on a hard-to-fathom scale. Perhaps decades into the future we will all be living in a world of wealth and abundance that permits such visionary ideas. But more likely, the space industry will underwhelm the dreamers among us in the decades ahead.

-EL and Dominic Mulraine

Let’s get physical (AI)

Most of the discussion around artificial intelligence focuses on generative AI -- particularly large language models (LLMs) that use machine learning to create new text, images, video and computer code. Rapid improvement and rising adoption of generative AI is attracting significant amounts of capital, driving equity prices higher, fueling a data centre CapEx boom and generating concerns about economic disruption.

But LLMs aren’t the only application of artificial intelligence with potentially significant economic implications. Physical AI, which operates in and interacts with the physical world rather than purely digital environments, is in an earlier stage of development and deployment but could become another major disruptive technology. NVIDIA CEO Jensen Huang thinks physical AI is the next major wave of AI-led innovation.

Physical AI spans a number of different technologies: autonomous vehicles and drones, AI-enabled industrial robots, humanoid robots and smart cities and spaces, among others. The common thread is that physical AI must be able to perceive (using cameras, radar, sensors, etc.), reason (interpret those inputs, predict outcomes and make decisions), and act (interact with its environment) in the physical world.

The emergence of physical AI is enabled by recent breakthroughs in several key areas:

• powerful new AI models and advances in special intelligence

• new training and simulation techniques supported by cutting-edge GPUs and growing compute capacity

• hardware improvements like better sensors and actuators and lighter materials.

The technology is still developing and much remains unknown, but here is a brief overview of physical AI, where it might be heading and why we should pay attention:

1.  Physical AI is already here: Waymo’s robotaxi fleet – which operates at level 4 of 5 on the vehicle autonomy scale, just short of full automation – surpassed 500,000 paid rides per week earlier this year (see chart below). Driverless trucks are being piloted on long haul routes in North America. Retailers and supply chain companies like Amazon, FedEx and Walmart are increasingly deploying robots to automate their warehouse operations, leveraging physical AI technology like computer vision. Humanoid robots are a more nascent technology, but Barclays Research estimates 15,000 were installed last year and projects another 60,000 will be deployed in 2026, with adoption supported by a 30x decline in costs over the past decade.

Waymo robotaxis now provide 500,000 paid rides per week across 10 U.S. cities

As of 06/24/2026. Sources: Waymo, TechCrunch, RBC GAM

2.  Potential market size: PricewaterhouseCoopers (PwC) estimates the combined physical AI market could reach €430 billion (US$490 billion) by 2030, led by autonomous driving, industrial and smart infrastructure, and humanoids and service robots (see chart below). Barclays Research thinks the physical AI market could be worth between US$500 billion and US$1.4 trillion by 2035 depending on adoption, with a US$900 billion base case. For comparison, Bloomberg Intelligence recently estimated generative AI could be a US$2.3 trillion market by 2032.

Physical AI could be a US$500 billion market by 2030

As at 06/24/2026. Figures converted to USD at EUR/USD = 1.135. Sources: PricewaterhouseCoopers (PwC) Strategy &, RBC GAM

3. Blue collar labour disruption: While LLMs threaten to disrupt white collar services jobs, physical AI operates more in the domain of blue-collar jobs and manual labour. As the chart below shows, roles in installation and repair, manufacturing, construction and even health care – which are generally less exposed to generative AI – are made up of many tasks that could eventually be done by physical AI.

Humanoid robots could be particularly disruptive as they are designed to work in human environments using the same processes and tools. They perform jobs, while industrial robots generally perform mere tasks. Barclays Research suggests humanoid robots are likely to be deployed in industry first (manufacturing, logistics and warehousing) before broadening to services and consumer-facing roles.

Physical AI could disrupt jobs that are less exposed to generative AI

As at 06/24/2026. Sources: Barclays Research, U.S. Department of Labour, RBC GAM

4. New (better?) job creation: Just as LLMs will create new roles and enable entrepreneurship, physical AI could boost employment in existing jobs like skilled trade; create new roles centred around building, deploying and maintaining robots and autonomous vehicles; and also create new jobs designing, monitoring and troubleshooting physical AI systems. To the extent humans work alongside physical AI, automation-driven cost savings could boost demand for products and services and stimulate complementary human employment. As with generative AI, it's hard to say how the balance between job creation and destruction will evolve. But to the extent that physical AI replaces physically demanding, repetitive, less desirable and sometimes dangerous work with more knowledge-based, less taxing roles, there could be positive elements.

5. Training supports data centre demand: Both digital and physical AI require computationally intensive training prior to deployment, boosting demand for hyperscale data centres. Whereas LLMs are trained on massive digital data sets (text, images, video, etc.), physical AI is increasingly trained using reinforcement learning in digital twin virtual environments that mimic the physical world. This allows vehicles and robots to run through thousands of realistic simulations in a safe and controlled environment before any on-site, real-world training takes place. Foxconn, the world’s largest contract electronics manufacturer, trained its AI-powered robotic workforce in digital twin environments, reducing deployment time by 40%, error rates by 25% and cutting operational costs by 15%.

6. Inference is more local: Large scale LLMs typically rely on the cloud computing power of hyperscale data centres for inference (using trained models to make predictions on new data). For physical AI, inference involves sensing, thinking and acting nearly simultaneously in real world environments, and thus requires much faster response times (lower latency) than digital AI. Rather than cloud computing, physical AI uses on-device or on-premises edge computing to execute tasks safely and reliably. That entails different hardware and software than a data centre.

7. Broader supply chain: The supply chain for physical AI also goes well beyond the servers, networking, power and cooling equipment used in data centres. Cameras, sensors and on-device chips are the brains of physical AI. Electric motors, actuators, robotic arms, etc. enable interaction with the physical world. Batteries allow for mobility in autonomous vehicles and robots. This more manufacturing-intensive supply chain is geographically diverse compared with software and semiconductor supply chains that are central to digital AI, so more countries could benefit from the physical AI buildout. Producers of key inputs like base metals and rare earths also stand to gain.

8. China is leading the way: The U.S. is leading the AI data centre buildout, but China has a significant edge in physical AI. China accounts for more than half of industrial robot installations, which are not necessarily AI-enabled but provide a strong foundation for physical AI integration. Barclays Research estimates 85% of the 15,000 humanoid robots installed last year were in China, with most produced by Chinese firms (see chart below). China is the world’s leading drone producer and has emerged as the top auto exporter, positioning it well for autonomous vehicle and drone manufacturing. China also dominates battery and rare earth production and is a major producer of actuators.

Chinese producers dominate the humanoid robot market

As of 06/24/2026. Sources: Barclays Research, RBC GAM

Most of us are not yet interacting with physical AI in the way we are with generative AI. But the technology is increasingly being piloted and adopted behind the scenes in certain industries and gradually making its way to consumers. It seems to have significant disruptive potential, if not quite on the same scale as generative AI. Many tasks and roles that are sheltered from generative AI impacts are exposed to automation from physical AI, but new jobs are also likely to be created and there is the potential to boost productivity in a number of industries which could further stimulate demand and employment. In the near-term, we expect physical AI to have less of an impact on economic activity than generative AI, which is driving CapEx, productivity and spending via wealth effects. But it bears close monitoring given its potential as another major technological force.

-JN

Evolving corporate priorities

We are now most of the way through the latest quarterly earnings season, so it’s once again time to dig through earnings call transcripts for macro insights. The table below offers a very simple overview of the terms mentioned most often on S&P 500 earnings calls, as well as those seeing the biggest increase and decrease in mentions compared to a year ago:

Changing topics of earnings calls

As of 06/22/2026. Sources: Bloomberg, RBC GAM

A few things stand out:

• AI and machine learning became the second-most discussed topic with by far the biggest increase in mentions compared to a year ago. The tech industry unsurprisingly continues to lead but now accounts for less than half of mentions, suggesting some broadening in the theme. Financials saw the biggest increase in AI mentions compared to a year ago.

• Higher energy and other commodity prices related to the conflict in Iran resulted in increased mentions of terms like material costs, transportation costs and inflation. Material costs even made it into the top 10 of overall mentions, surpassing talk of CapEx.

• Mentions of tariffs were significantly lower than a year ago when management teams were responding to Trump’s Liberation Day tariffs. The Supreme Court struck down Trump’s IEEPA tariffs in Q1 and while a 10% baseline tariff under Section 122 replaced some of that lost revenue, the effective tariff rate on imports is now the lowest since March 2025.

• Despite concerns about rising costs and inflation, mentions of headwinds and economic slowdown declined. That seems consistent with our view that the U.S. economy is not significantly affected, on net, by the energy price shock. U.S. economic data have generally exceeded expectations in 2026.

Diving a bit deeper, we used an in-house proprietary AI system to examine a representative sample of S&P 500 earnings calls, evaluating the most common themes, changes in prevalence of those themes and how sentiment is evolving. Here are some of the findings:

• Three of the top five earnings call themes identified by the model were AI-related: AI infrastructure and compute demand, data centre power demand, and agentic AI and autonomous systems. Sentiment around these themes was strongly positive and language shifted from forward looking to present tense as management teams moved from planning to implementation. Interestingly, the model identified absence of negative sentiment on AI as a risk – perhaps a prescient warning given recent softness in some AI stocks.

• The Middle East energy shock was the dominant new macro risk identified across sectors, emerging at a speed and intensity that mirrored the energy shock following Russia’s invasion of Ukraine in 2022. Sentiment was clearly negative, but confidence was low, highlighting the degree of uncertainty around the duration of the conflict and disruption in energy markets. The model identified a “sentiment barbell” with AI enthusiasm at one end and geopolitical concern at the other.

• Memory and component cost inflation and supply constraints were emerging concerns in the latest quarter. We recently discussed how AI is driving unprecedented demand for memory, driving hyperscaler CapEx bills higher, squeezing tech companies’ margins, and pushing consumer prices higher.

• While tariff mentions were down in the latest quarter, sentiment became somewhat more negative with tariffs being treated as a persistent margin headwind and shifting tariff rates and legal authorities creating added complexity for importers. The launch of Section 301 investigations to eventually replace temporary Section 122 tariffs might have contributed to the view that tariffs are here to stay, despite some successful court challenges.

• Sentiment on consumers was mixed: management teams projected confidence when discussing higher income households but expressed ongoing concern about lower income cohorts. The energy shock was seen contributing to this bifurcation. Walmart’s CEO cited consumers filling their tanks with less than 10 gallons of gasoline as a sign of stress not seen since 2022. We recently examined the health of U.S. consumers and found spending growth is holding up relatively well for higher and lower income consumers alike, though the latter are still doing worse.

Overall, management teams were discussing many of the same themes we’ve focused on over the past quarter. Generally, their comments provide a timely read on economic conditions. However, they were rendered a bit stale this time around given recent positive developments in the Middle East.

-JN

Bank lending sends positive signal

Bank lending has accelerated across a range of developed economies over the past one to two years and is now moving at a pretty good pace.

To illustrate, U.S. commercial bank loans are now rising at an improved 4.0% annual pace (see next chart). Not long ago, bank lending was outright contracting. While there were two periods of greater strength over the last six years, these were artificial in nature:

• Companies drew down their credit lines for precautionary reasons during the early-2020 pandemic lockdown.

• They borrowed enthusiastically in 2022 as the economy rebounded from the pandemic and opportunity temporarily abounded.

The rise today appears to be somewhat more organic.

U.S. credit growth has been rising.

As of May 2026. Includes real estate loans. Shaded area represents recession. Sources: Federal Reserve, Macrobond, RBC GAM

Why do we care about this? Because strong bank lending is a promising economic signal. It conveys three happy messages:

1. First, lenders are willing to lend. This is to say, banks are well-capitalized. Their deposit base is growing. They see healthy households and businesses with whom to engage. They must anticipate favourable economic conditions ahead.

1. Second, borrowers are willing to borrow. For corporate borrowers, the interpretation is clearly positive: businesses must see opportunities for growth and apparently anticipate earning a relatively high return on capital from those opportunities.

   The interpretation is perhaps more mixed for households. In the context of the K-shaped economy narrative (high-income households doing well but low-income households faring poorly), some lower-income households may be borrowing under duress as their income fails to meet their obligations. But, in practice, household loan delinquency rates are edging lower, and the U.S. household debt-to-GDP ratio is falling. This argues such motivations should be diminishing rather than ascending. In turn, much of the household borrowing can also be interpreted positively: households feel comfortable about ramping up their spending.

1. Third, by virtue of all the added lending, more money is sloshing around the economy. The mechanics of a fractional reserve banking system means that money being lent is effectively new money being injected into the system. This then encourages the economy to grow more quickly, in much the same fashion as the Fed’s quantitative easing was instrumental in strengthening the economy at various points over the past 18 years.

As an aside, it is remarkable that credit growth is rising as it is without primary support from the residential mortgage market, which remains subdued.

A few caveats are necessary. Some of the lending growth may represent a catch-up effect after unusually soft lending when interest rates were high several years ago. Thus, it may not be sustained indefinitely. For the U.S. specifically, the regional banking stress of 2023 also drove some deposits out of the traditional banking system. As that memory fades, the deposits are trickling back and being recycled into loans.

But the acceleration of bank lending isn’t all smoke and mirrors. It is replicated in a variety of other developed markets that didn’t experience regional banking stress.

Eurozone bank loans to households and non-financial corporations have similarly accelerated (see next chart), as has UK bank lending (see subsequent chart).

Eurozone bank lending has accelerated

As of May 2026. Sources: European Central Bank (ECB), Macrobond, RBC GAM

UK bank credit growth is rising

As of May 2026. Sources: Bank of England, Macrobond, RBC GAM

Japanese bank lending to households and non-financial corporations has similarly accelerated. Its current rate of increase is unsurpassed over the past two decades, other than during a brief pandemic distortion (see next chart).

Japan’s credit growth has picked up

As of Q1 2026. Sources: Bank of Japan (BOJ), Macrobond, RBC GAM

In fairness, this bank lending narrative is not quite universal. Chinese bank lending growth is instead decelerating – though this is not necessarily a bad thing for a country that ran up a lot of debt and previously overinflated its housing market (see next chart). It is a very similar story in Canada, right down to the overheated housing market now cooling (see subsequent chart). In both cases, credit growth is not actively supporting economic growth in these countries to the extent it is elsewhere.

Credit growth in China continues to decline

As of May 2026. Sources: People’s Bank of China (PBoC), Macrobond, RBC GAM

Canadian credit growth continues to slow

As of April 2026. Sources: Statistics Canada, Macrobond, RBC GAM

Interestingly, the situation might improve in the not-too-distant future for Canada. This is in part because the housing market has recently shown a bit of life (discussed later in this #MacroMemo), and in part because Canadian bank regulations are now easing – including a capital requirement reduction implemented by the Canadian regulator just last month.

Easing bank regulations are something of a global theme right now, offering the promise that bank lending growth may remain supportive of the economy for some time to come.

-EL & VL

Fed’s Warsh era begins

There was a particularly bright spotlight on Kevin Warsh at his first meeting as Chair of the Federal Open Market Committee (FOMC) given politicization of the Fed under President Trump. The combination of a hawkish hold and no dissenting response from Trump helped to chip away at politicization concerns. Warsh’s spartan communication style could result in added volatility but might raise the profile of the rest of the committee – a useful reminder that Warsh is just 1 of 12 voters.

The Treasury yield curve flattened following the meeting as the market priced in greater odds of near-term rate hikes and less concern about inflation – the opposite of what you would expect if a politicized chair were steering the committee toward a dovish policy mistake.

Both the pared-down policy statement and updated dot plot leaned hawkish. While earlier statements under Powell emphasized the committee’s dual mandate, the latest statement focused on a commitment to deliver price stability, which Warsh clearly emphasized during his press conference. For the time being, the Fed sounds like a single-mandate central bank focused squarely on inflation.

The updated dot plot showed the committee is now evenly split on raising rates this year, a major shift from March when a majority anticipated cuts. A hawkish shift in the dots was expected given recent data and comments from Fed officials, but the extent of support for tightening was nonetheless a surprise. Those rate projections were based on a forecast for inflation to remain elevated by year end, but lower energy prices since the dots were submitted could soften that view somewhat. The latest non-farm payrolls and ISM Manufacturing releases could also subtly reduce the amount of desired tightening.

Warsh didn’t submit his own interest rate projection and doesn’t believe in offering forward guidance on future policy decisions. The most he gave away is that, outside of housing, it’s hard to say current monetary policy is restrictive. That came across as hawkish given that most FOMC members think the current policy rate is above its long-run neutral level.

Explaining his aversion to forward guidance, Warsh said financial markets function better – and give the Fed a better signal on financial conditions – when they react independently to incoming economic data rather than guessing how the Fed might react. We think markets will still do the latter, but now with less information about the Fed’s reaction function, which is likely to result in more volatility. One can argue forward guidance should only be used in extraordinary circumstances – to shape medium-term policy rate expectations and influence yields further out the curve – but failing to give any insight on how the Fed is thinking about its future policy decisions may be sub-optimal, in our view.

Volatility could be exacerbated by what sounds like an every-meeting-is-live approach under Warsh, a departure from Powell’s Fed which liked to prepare the market for any policy changes. When asked about the Fed’s reaction function and what might prompt rate hikes, Warsh noted several times that the committee will meet again in just six weeks. That caused odds of a July rate hike to jump, although a hold is still seen as more likely.

While Warsh was tightly lipped on his own policy views, other FOMC participants are still free to share theirs. Former Fed Chair Bernanke used to quip, "monetary policy is 98% talk and 2% action,” and a quieter chair arguably gives more of that policy power to other committee members.

That makes tracking individual Fed speakers even more important. Fortunately, there are a number of natural language processing tools that evaluate Fed speeches and gauge sentiment on a hawkish-dovish scale, making it easier to identify trends among a diverse group of Fed speakers. Here is one from Bloomberg, which picks up the hawkish shift heading into the June FOMC meeting:

Natural language processing suggests Fed commentary has become more hawkish

As of 06/25/2026. Sources: Bloomberg, RBC GAM

Alongside the shift away from forward guidance, Warsh is looking to put his stamp on the Fed in other ways. He announced the formation of 5 task forces to examine the following:

• The Fed’s communications, including press conferences, meeting transcripts/minutes and the dot plot.

• The Fed’s balance sheet, including the benefits and risks of the current ample reserve regime.

• The Fed’s use and reliance on existing data sources, with a view to leverage more timely/contemporaneous data sources.

• Productivity and jobs in an era of transformation, focusing particularly on the effects of AI on the economy and monetary policy.

• The Fed’s inflation frameworks, including the drivers of inflation and how to deliver price stability in a changing economy. Importantly, the Fed’s 2% inflation target is outside the scope of this review.

It should be noted that the Fed already studies many of these issues on an ongoing basis. Bringing in external experts could see some interesting proposals, but to the extent they have to be agreed upon by the FOMC – many of whom have been in their seats for years – we would lean toward any changes being at the margin rather than wholesale.

Warsh aims to get reports from these task forces by year end. That timeframe might help the Fed put off a rate hike until inflation has started to ease and there is less pressure to tighten. The review period could also reduce friction between the new chair and existing committee members as divergent views don’t have to be reconciled immediately. For instance, when Warsh was asked about his view that AI-driven productivity growth will be disinflationary and allow for rates cuts – a view that other Fed speakers have pushed back on – he simply said, “there’s a committee for that.”

Warsh favours a smaller Fed balance sheet. However, we doubt the balance sheet task force will cause the central bank to abandon the ample reserve framework it adopted during the global financial crisis, which June’s policy statement reaffirmed. Maintaining that system makes it difficult to significantly shrink the Fed’s balance sheet – reducing the risk of higher Treasury yields as the market has to absorb the Fed’s holdings. But there are other measures that could allow for some reduction in the Fed’s asset holdings over time.

Overall, the beginning of Warsh’s tenure brought more of a break from the Powell era than recent Fed leadership transitions. But that’s more about Warsh’s communication style than any dovish leanings he expressed before securing the position. Indeed, Warsh’s square focus on inflation and growing support on the FOMC for rate hikes suggest Trump won’t be getting the interest rate relief he has called for.

Our forecast assumes the Fed will remain on hold over the next year, with lower energy prices – assuming the U.S.-Iran Memo of Understanding holds – giving the Fed some scope to look through this latest wave of inflation. But it looks like there will be limited tolerance for upside surprises on inflation, and we shouldn’t count on getting much advance notice if the FOMC decides rate hikes are warranted.

-JN

Bond market breakdown

What’s been driving bond yields?

We recently estimated the 10-year bond term premium for the U.S., Japan, Germany, Italy, France, the UK and Canada, using the methodology of Adrian, Crum and Moench (ACM). With this, it is then possible to disaggregate nominal long-term yields into their component parts – the term premium, inflation expectations and the expected real short-term rate.

Long-term trend

From a multi-year standpoint, the main drive is clear: term premiums have increased substantially over the last five years, flipping from negative to positive and rising by well over 100 basis points in most examined markets (see next chart). This makes sense. The era of yield-depressing quantitative easing is over. Bond investors are demanding additional compensation to lock away their money for years at a time given the increased risk associated with sovereign borrowers. Public debt loads are high and rising. There are major question marks around critical fiscal- and economy-relevant subjects including artificial intelligence, geopolitics and climate change.

Term premiums have increased substantially

As of 07/02/2026. Term premium estimated using the Adrian, Crump and Moench (ACM) model. Sources: Federal Reserve Bank of New York, Bloomberg, RBC GAM

As per the next chart, the U.S. term premium was negative during the pandemic (and, not fully shown, quite a swath of the 2010s). It rose substantially in 2022-2024. This new, higher term premium has then proven relatively stable over the past two years.

United States term premium has risen since 2020

As of 07/02/2026. Term premium estimated using the Adrian, Crump and Moench (ACM) model. Sources: Federal Reserve Bank of New York, Bloomberg, RBC GAM

Looking at a cross-section of countries, the term premium rankings make intuitive sense (see next chart). France, Japan, U.S. and the UK are highest. Japan has the highest public debt-to-GDP ratio in the world, and the U.S., France and the UK all rank among the worst four nations in our fiscal health scorecard (see subsequent table). Bond investors have rightly demanded more compensation for this. Italy’s fiscal position isn’t quite as bad, and indeed its term premium is slightly lower than the others. At the opposite end of the spectrum, Germany and Canada look somewhat better from a fiscal standpoint and accordingly have the smallest term premiums in our analysis.

Higher term premiums align with poorer fiscal health

As of 07/02/2026. Term premium estimated using the ACM model. Sources: Federal Reserve Bank of New York, Bloomberg, RBC GAM

France, Japan, UK, and U.S. rank among the most fiscally stretched nations

Note: 2025 data for all indicators except interest payments (2024) and GDP growth (IMF forecasts for 2031 used as proxy for ‘normal’). Fiscal adjustment refers to the necessary reduction in fiscal deficit to stabilize debt-to-GDP ratio. Sources: IMF, Macrobond, RBC GAM

Alongside the higher term premiums, the past decade has also brought about a moderate increase in long-term inflation expectations as well. The pandemic was an inflection point. We saw relatively subdued 10-year-ahead inflation expectations beforehand and during the lockdown phase of the pandemic, followed by persistently more elevated inflation expectations ever since (see next chart).

This makes sense. Inflation never managed to fully settle after the post-pandemic supply chain shock. The recent energy shock has provided another upward impulse for prices.

U.S. inflation expectations have increased

As of 07/02/2026. Inflation expectations estimated using 10-year inflation swaps. Sources: Bloomberg, RBC GAM

Conversely, while nominal central bank rates are today substantially higher than the pre-pandemic norm, this is less true on an inflation-adjusted basis (see next chart). Our measure of the expected average short-term real interest rate over the next decade for the U.S. has lately moved higher – a subject tackled next – but is ultimately still fairly unremarkable compared to the level that prevailed before the pandemic.

U.S. real short-term rate remains near its decade-long average

As of 07/02/2026. Sources: Bloomberg, RBC GAM

In short, the higher yields of today are primarily a reflection of a larger term premium. They are secondarily due to higher inflation expectations and to a significantly lesser extent explained by any increase in real short-term rates.

The fact that term premiums have become substantially positive is arguably quite good news for bond investors. While they do reflect some increase in the underlying sovereign risk, the most likely scenario is that the larger term premiums simply increase the size of the coupon for those willing to take on duration in the bond market.

Further, the related implication of a relatively steep yield curve should provide modest additional capital gains for investors maintaining a fixed-duration portfolio via the roll-down effect.

Short-term trend

Turning toward the short-term trend in interest rates, what is notable is how little uniformity there is across countries (see next chart). Japanese yields have increased substantially over the past six months; the U.S., UK and Italy have delivered modestly higher yields; while France, Germany and Canada are flat to lower.

Interest rate changes have varied widely across G7 over the past six months

As of 07/02/2026. Term premium and average short-term rate estimated using the ACM model. Inflation expectations estimated using 10-year inflation swaps. Sources: Federal Reserve Bank of New York, Bloomberg, RBC GAM

Beneath the surface, the short-term drivers have also varied greatly.

Japanese yields are up primarily due to a larger term premium, while the term premium has actually shrunk slightly over the past six months in each of the other six markets.

Meanwhile, short-term rate expectations have proven a mixed influence. They have increased in three markets – most prominently, the U.S. and UK, which have trumpeted particularly hawkish messages in recent months – while decreasing slightly in four others.

The inflation expectation story is more uniform, with seven of the eight markets experiencing an increase in inflation expectations as per the energy shock that unfolded this spring. There may now be scope for this to settle back down.

The diverse performance of these bond markets over the past six months arguably presents an opportunity for savvy active investors to earn a superior return by capitalizing on such heterogeneity. This is especially true given that many of the underlying macro drivers – prominently, debt, inflation, geopolitics, AI and climate change – have significant global elements, leading to the conclusion that national bond markets should not actually be performing quite so differently.

-EL & Dominic Mulraine

Signs of life in Canadian housing

Canada’s housing market is showing signs of life after resale activity fell to a more than 2-year low in March. Home sales increased in April and May, inventories have tightened, and prices are starting to stabilize. But we’ve seen a number of false dawns during what has now been a 4-year period of sub-average resale activity (see chart). Will this time be different? We’re not entirely convinced.

Canadian home sales have been below average for four years

As at 06/19/2026. Sources: Canadian Real Estate Association (CREA), RBC GAM

Of course, there isn’t a single housing market in Canada – it’s really a collection of regional markets that can run at different speeds based on local factors. Much of the weakness in Canadian home sales is concentrated in Ontario and B.C., where affordability is particularly challenging. In contrast, resale activity is above its 10-year average in the Prairie provinces and Quebec, and only moderately lower in Atlantic Canada (see chart). Similarly, benchmark home prices are down 5% year-over-year in Ontario and B.C. but rising in most other provinces.

Subdued national home sales mask regional divergence

As at 06/22/2026. Sources: Canadian Real Estate Association (CREA), RBC GAM

That said, there are still some broad macro drivers that influence Canadian housing in aggregate. Here are some factors that might be supporting the recent uptick in home sales:

• Canada’s labour market is stabilizing: Last year, Canada’s unemployment rate rose to a post-pandemic high and Canadians were more concerned about job loss than at any point in the past decade. These are not the conditions that encourage people to buy homes. But the labour market now appears to be stabilizing with unemployment declining in the past two quarters.

• Lending conditions are no longer tightening: The Bank of Canada’s (BoC’s) Senior Loan Officer Survey showed net tightening in mortgage lending conditions throughout much of 2025. However, conditions eased late last year as banks reduced the spread they charge on top of funding costs (the price component in the chart below). Lending standards were stable in Q1 but might ease in the coming quarters. Recent changes to regulatory capital requirements by the Office of the Superintendent of Financial Institutions could unlock more than $100 billion in new lending by Canadian banks.

• Government policy is supportive: We previously discussed measures aimed at boosting home construction and easing affordability challenges, including eliminating GST on new home purchases for first-time buyers. Most of these policies don’t directly boost resales but could help reinvigorate the overall housing market.

Mortgage lending conditions are no longer tightening.

As of 06/19/2026. Price and non-price balance of opinion divided by 2 to show contributions to overall mortgage lending conditions. Sources: Bank of Canada, RBC GAM

There are, however, a number of headwinds that remain in place:

• Mortgage rates are no longer falling: Mortgage rates are down significantly from their post-pandemic highs. The BoC has cut its policy rate by 2.75% over the past two years. But 5-year fixed mortgage rates (historically the most popular term) remain north of 4%, above levels seen throughout the 2010s. If anything, mortgage rates could drift higher alongside the recent increase in market yields (see chart below).

• Affordability remains challenging: Despite some improvement in recent years, our housing affordability measure suggests Canadian home prices are still 16% above levels consistent with average affordability. While affordability is worst in Vancouver and Toronto, it is still challenging in other major markets. With mortgage rates no longer falling, some combination of declining prices and rising incomes is needed to narrow the gap further.

• Price expectations are subdued: Expectations for home price growth are just half their historical average in less affordable markets like Ontario and B.C. At the national level, home price gains over the next 12 months are only expected to marginally outpace inflation. The prospect of perpetually rising home prices previously caused households to stretch their budgets – supporting sales and prices – but that psychology appears to have broken. Similarly, the share of homes being sold and flipped within 12 months is now at its lowest level in at least a decade, weighing on sales and renovation activity.

• Rental affordability is improving: Rent inflation has slowed by more than half compared to its post-pandemic peak and is now at its lowest level in more than 4 years. Rents are declining outright in major markets like Toronto. Improving rental affordability discourages potential first-time homebuyers and is dampening investor enthusiasm for rental units. Canada’s immigration slowdown – driven by net non-permanent resident outflows, who are almost all renters – has contributed to a soft rental market.

Canadian mortgage rates have stabilized but could drift higher

As of 06/25/2025. Sources: Bloomberg, Bank of Canada, RBC GAM

Given this mix of headwinds and tailwinds, we are skeptical that the spring pickup in Canadian home sales is the start of a broader upswing. But signs of stabilization in the resale market alongside some improvement in home construction represent a welcome change for a sector that has generally been a drag on Canada’s economy in recent years (see chart). Our GDP growth forecast doesn’t assume a significant contribution from housing this year. However, simply not repeating 2025’s 0.3% drag supports a pickup in overall growth this year.

Housing has generally been a drag on Canada’s economy in the past five years

As at 06/19/2026. Sources: Statistics Canada, RBC GAM

-JN