Following July's sharp volatility in the AI-driven market, global equities have been gradually recovering since August. The South Korean Kospi index and the Philadelphia Semiconductor Index, both proxies for the AI theme, have rebounded 33% and 19% from their recent lows, respectively. The S&P 500 has even broken through 7,800 points to quietly set a new all-time high, while the Nasdaq is a mere 1.4% away from its record. This aligns with our mid-year outlook, where we raised our S&P 500 target to 7,800-8,000 points.
Our analysis of the dot-com bubble and the seven corrections during the current AI rally shows they were all driven by the same three factors: high micro-level valuations and crowding, macro-level external disturbances, and industrial development bottlenecks. At this stage of the rebound, high leverage and crowding have largely stabilised and cleared. For instance, the combined total of margin debt and leveraged ETF size in the South Korean market has contracted by nearly 60% from its peak before stabilising, while other markets have also seen varying degrees of deleveraging. Our proprietary AI bubble pressure index has improved, but the market is still waiting for a new breakthrough to break the current ceiling. Against this backdrop, liquidity has become a key focal point for market movements, reflected in concerns about a potential reversal of the yen carry trade and increased US Treasury issuance.
Fed Policy: Easing Rate Hike Pressure, Inflation is Not the Core Issue
Since Governor Warsh took office in late May, market expectations for Fed rate hikes have swung back and forth. The vague rhetoric at the July FOMC meeting failed to alleviate rate hike concerns and instead pushed long-term bond yields above 4.7%. Recently, positive news has emerged, with both July's non-farm payrolls and CPI/PPI data falling short of expectations, pushing the CME FedWatch tool's implied rate hike timing to December. As we noted in our mid-year outlook, the US economy is experiencing a K-shaped divergence, with traditional cyclical sectors like housing still constrained by high rates. Therefore, the Fed did not need to hike aggressively. Inflation is not the primary contradiction; the rate hike concerns stem mainly from AI-driven growth and Governor Warsh's style.
Inflation is likely to continue its downward trend and is not the main contradiction. July CPI fell further from 3.5% to 3.4%. Despite a rebound in some core goods and services components, we believe the decline is probable, given the high base from last year's tariffs and the increased likelihood of Trump adopting a more accommodating stance on Iran-US tensions ahead of the midterm elections. The key question is the speed of the decline, not the direction. Our calculations suggest that if oil prices gradually fall to $70 per barrel by year-end, the year-on-year CPI could drop to around 2.7%.
On the growth front, although the overall picture is K-shaped, US traditional demand, excluding housing, is not weak and even shows signs of spreading thanks to AI. AI-related growth remains robust. While Q2 real GDP growth was 1.5% annualised, AI-related investment contribution fell from 1.11 percentage points in Q1 to 0.23 percentage points, it remained positive. The drop in inventories and increase in exports also reflect the pull from AI demand. The ISM Manufacturing PMI rose from 47.9 at the end of last year to 55.6 in July. Private consumption, though not spectacular, has been stable, contributing 2.12 percentage points to Q2 GDP as a core driver. However, July retail sales unexpectedly fell 0.6% month-on-month. While July's non-farm payrolls were significantly negative, seasonal factors and potential data adjustments mean this cannot yet be taken as a sign of a major labour market cooldown. The rise in US real interest rates from 1.9% in early May to 2.4% currently, which accounts for most of the increase in nominal rates, underscores this point.
Governor Warsh's style is a key factor behind the recent volatility in rate hike expectations. His ambiguous and shifting stance on rate hikes, such as alternating between focusing on the trimmed mean inflation gauge and AI-driven inflation, and recently suggesting that rising market rates have had a similar effect to a rate hike, has left the market confused. His first two FOMC meetings in June and July failed to provide clear signals on rate hikes, and the "hide-and-seek" communication at the July meeting actually fuelled concerns that the Fed cannot effectively control long-term inflation expectations, pushing long-term bond yields higher. At this point, we see two potential scenarios: if growth and inflation continue to weaken, the Fed will have no reason to hike, alleviating market concerns. Recent economic data is trending in this direction. However, if the August non-farm payrolls and inflation data show renewed strength, a "pre-emptive rate hike" in September would not be a bad thing. It would help stabilise long-term inflation expectations and bolster the Fed's credibility. A rate hike could also mean that the bad news is out of the way, similar to the 1997 pre-emptive rate hike when the Fed acted despite strong growth and industry trends without runaway inflation. After that hike, bond yields peaked and soon fell, and the stock market rebounded quickly. Conversely, if the market expects a hike but the Fed stands pat in September, and a hike becomes more difficult closer to the midterm elections in October, it could amplify market anxiety.
US Treasury Supply: Q3 Issuance Increases, but Impact May Be Manageable
On the supply side, the net issuance of US Treasuries in the third quarter is set to rise to $739 billion, a factor that could push up bond yields and affect liquidity. According to the Treasury Borrowing Advisory Committee's (TBAC) financing plan, net issuance will be $739 billion in Q3 2026 and $628 billion in Q4, a significant increase compared to the $577.3 billion and $189.8 billion net issued in Q1 and Q2, respectively. Since July, supply pressure and concerns about long-term inflation have pushed the 10-year term premium from 50 basis points to 80 basis points, causing the 10-year yield to rise from 4.47% to 4.7% despite a drop in rate expectations.
However, a closer look suggests the impact is less severe than feared. The bulk of Q3's supply pressure was concentrated in July, with nearly half of the planned $739 billion already issued. This means only $384 billion remains for August and September, suggesting a smaller disturbance than in July. Furthermore, the issuance structure is heavily skewed towards short-term bills, limiting the supply pressure on long-term bonds. Over the past three years, short-term bills have consistently accounted for over 80% of total issuance. In July 2026, they made up 86% of gross issuance and 84% of net issuance, having a smaller impact on long-term yields. Looking back at the two peak issuance periods in Q3 2023 and Q3 2025, supply pressure impacts bond yields by pushing up the term premium. Without the help of rate cut expectations, the impact could be greater, as was the case in 2023. In Q3 2023, a surge in issuance pushed long-term yields sharply higher. The net issuance was $1.01 trillion, and the 10-year term premium rose nearly 120 basis points in three months, driving the long-term yield from 3.8% in June to a peak of 5% in October. In Q3 2025, despite a larger issuance of long-term bonds, rate cuts helped push bond yields lower. Net issuance was $1.058 trillion, and long-term bonds made up 41% of the net issuance. However, two consecutive months of cold non-farm payrolls and a slowdown in the pace of inflation increases fuelled rate cut expectations, which the Fed delivered in September 2025. The 10-year term premium oscillated in the 60-80bp range during this period, while rate expectations drove the long-term yield from 4.5% in mid-July to 4%.
Overall, the Q3 2026 issuance is smaller than those two peaks. Without the tailwind of rate cut expectations seen in 2025, its impact on the market is likely to fall between that of Q3 2023 and Q3 2025.
Bank of Japan and Yen Carry Trade: A Repeat of 2024's Unwinding is Less Likely
With Fed rate hike pressure easing but costs still high, and minor supply-side disturbances, the stability of demand is crucial for the US Treasury market to avoid increased volatility. As of May, foreign investors held $9.37 trillion in US Treasuries, or 24% of the outstanding stock. Japan remains the largest foreign holder, with $1.14 trillion, or 12% of foreign holdings. This makes Japan's official sector a key focus, as the yen carry trade is one of the largest macro trades for buying US Treasuries.
Japan has a need to intervene in the forex market, but doing so requires selling US Treasuries, which is why the US stepped in to "help" prevent a sell-off. When the yen depreciated to a near 164 historical low in late July, the Japanese Ministry of Finance conducted a forex intervention, causing the yen to appreciate by up to 3.4% in an hour. Subsequent volatility saw the yen appreciate by as much as 5.5%. According to our estimates, this intervention was the largest single round in the 1990s, exceeding 12 trillion yen (approximately $750 billion). On August 2nd, US Treasury Secretary Bessent confirmed a joint US-Japan exchange rate intervention, a statement later corroborated by the Japanese Ministry of Finance, which also stated its intention to use the Fed's FIMA Repo Facility in the future. Japan's forex intervention mainly involves selling dollar assets (dollar deposits and US Treasuries) from its reserves or slowing the pace of reinvesting maturing bonds to buy yen. To avoid a large-scale sell-off of US Treasuries following excessive yen depreciation, the US Treasury used verbal intervention and sold euros to buy yen. This included verbal guidance to manage expectations, with Bessent stating that the yen is undervalued and that the US would not hesitate to intervene again, pledging to support Japan in stabilising the yen. The US also sold euros and bought yen, but the scale was small. The euro-yen pair depreciated by up to 4.3% from July 29th to August 3rd before recovering. As of August 12th, the FIMA Repo Facility has not been used. This facility, established in 2020 and made permanent in 2021, is a Fed tool for providing dollar liquidity to foreign central banks in times of crisis. The current single counterparty limit is $60 billion, and Bessent has recently indicated a potential expansion of this limit. If the intervention proves ineffective and forces the BoJ to raise rates, narrowing the US-Japan interest rate differential, it could increase the pressure for a carry trade unwinding. The appreciation of the yen and a contraction in risk appetite could also contribute to this. Following the intervention, BoJ rate hike expectations have risen sharply. Historical experience shows that forex intervention can only stabilise the exchange rate in the short term, and a rate hike may be needed to secure long-term stability. After the July 30th intervention, the BoJ's language at its July meeting turned notably hawkish, with market probability of a September rate hike surging from 21% on July 30th to over 80% currently. In this scenario, the yen exchange rate becomes a critical variable. If the intervention successfully leads to yen appreciation, the pressure on the BoJ to hike will ease. Otherwise, a BoJ rate hike combined with forex intervention-led yen appreciation could be a catalyst for unwinding the yen carry trade, leading to private sector selling of US Treasuries and creating greater demand-side pressure. The yen's recent depreciation from 155.2 to 159.32 on August 3rd warrants close monitoring.
Is a repeat of the August 2024 yen carry trade unwinding storm likely? Under static conditions, it is less probable, but the risk of a "perfect storm" of multiple factors cannot be ignored. The 2024 unwinding required several conditions, which are currently not all present. Yen short positions are significantly lower than before the intervention. CFTC data shows net non-commercial short positions in the yen, which peaked at 170,000 contracts in late July 2024, have since been reduced to 40,000 contracts. The market has already priced in a BoJ rate hike in September, with an 80% probability. The key driver of the 2024 unwinding was the narrowing of the US-Japan interest rate differential, triggered by a sharp drop in US bond yields following weak ISM and non-farm payroll data that sparked recession fears. This caused the carry trade to suffer losses on both the liability and asset sides, leading to the unwinding and a liquidity shock. The most significant difference between now and 2024 is the absence of a US recession narrative. The US economy remains strong. Therefore, while a repeat of the 2024 shock is difficult under static conditions, we must remain vigilant about the amplification effect of multiple negative factors coinciding, as liquidity shocks are often unexpected and "black swan" events. The yen exchange rate will be particularly critical. However, it is important to note that even if a liquidity shock occurs, it does not change the trend and often presents a better buying opportunity. Historical liquidity crises, such as the UK pension fund crisis, Silicon Valley Bank, and Credit Suisse, show that while short-term impacts are severe, central bank intervention to provide liquidity is effective, often creating better entry points.
Market Implications: Liquidity Likely to Stay Stable, September is a Key Hurdle
Current US dollar liquidity has not tightened significantly. Price indicators like the SOFR-OIS spread have widened but are manageable. Corporate bond spreads and credit default swaps for major tech names have narrowed, suggesting easing corporate financing pressure. Onshore and offshore dollar liquidity conditions are not showing signs of acute stress. On the quantitative side, the Fed has stopped expanding its balance sheet, and its August-September reserve management purchases have been paused. However, if the Fed uses the FIMA Repo Facility, it could create a short-term balance sheet expansion effect. The ratio of reserves to bank assets is hovering around 11.7%, which is not excessively abundant but not tight either. Our financial liquidity indicator (Fed liabilities - TGA - ONRRP) suggests that if the Fed does not start quantitative tightening this year, US financial liquidity will likely remain in a stable range without a significant contraction.
Overall, US dollar liquidity is expected to remain broadly stable, with no major contraction. The key focus should be on the yen exchange rate, and September could be a critical juncture. For the broader market, more significant upside will come from breakouts in earnings on the numerator side. If investors are concerned about disturbances on the denominator side (liquidity), they could consider slightly reducing positions in the short term. However, any pullback driven by liquidity concerns may present a good entry opportunity. For specific assets: We maintain a positive view on US equities. Our mid-year S&P 500 target of 7,800-8,000 is being realised. Future upside will come from catalysts in the industrial trend, such as breakthroughs in AI (like the coding breakthrough from Anthropic in Q1), which could drive further earnings upgrades. If investors worry about liquidity volatility, they could slightly reduce positions or use consumer staples or gold to hedge against tech volatility. However, any significant market correction due to liquidity issues should be viewed as a buying opportunity. If the Fed does not hike rates this year, the 10-year yield fair value could be around 4.1-4.5%, with supply pressure concentrated on short-term bills, having a limited impact on long-term bonds. Under our baseline scenario, the US dollar is likely to continue oscillating and is unlikely to weaken significantly. The bottom for gold is well-defined, with current interest rates and the dollar supporting a price around $4,500/oz. Larger upside will depend on macro narratives. In our baseline, if the dollar stays around 98-100 and real rates, uncertainty, and momentum remain at their post-Iran-conflict averages, the year-end gold price could be around $4,500-$4,800/oz. The key events to watch are the US August non-farm payrolls and CPI data, the September FOMC meeting (September 16th), the Bank of Japan's September rate decision (September 18th), and any further US-Japan forex intervention.
Comments