On August 17th, the forum held a seminar titled "Global Financial Market Volatility: Challenges and Responses," where Xing Ziqiang, Chief Economist of Morgan Stanley China, delivered a speech. Xing Ziqiang believes that the recent turmoil in the global financial markets has sparked a reflection on three mindsets:
1. Whether the US economy can remain strong indefinitely. Recent data indicates that the US economy is far from recession and is heading towards a soft landing, but the market remains vigilant for signs of economic weakness. It is expected that the sustained decline in inflation will drive a rate-cutting cycle, with rate cuts beginning at the Federal Open Market Committee (FOMC) meeting in September, and three 25 basis point rate cuts to be carried out within 2024.
2. The increase in financial market uncertainty. The valuation of US stocks has reached a very high level. Even if the prediction of a soft landing for the US economy is confirmed, there is no room for upward movement in the main US stock indices in the next year or so. The high uncertainty of the US election results and the lack of precedent for market behavior before the election mean that cross-asset allocation is an effective strategy to navigate market cycles. In an environment of a soft economic landing, stock market allocation should also be more defensive.
Advertisement
3. Patience with the AI technology revolution. Although AI technology has brought about a new technological revolution, its enhancement of productivity is still in its early stages. AI technology is currently concentrated in enterprises that provide equipment and tools, and it still requires patient waiting for its maturity and popularization.
In the past month, the roller coaster-like turmoil in the global financial markets has sparked a reflection on the three major mindsets in recent years: Can the US economy maintain its resilience and not land? Is AI really a new round of technological revolution, or is it a speculative bubble in the capital market? Is the yen carry trade on hold or completely reversed? I believe that the deceleration of the US economy is a soft landing, but the uncertainty in the capital market has significantly increased. Although I am optimistic about the AI technology revolution, it is still in its early stages at this stage, and the release of its productivity dividends still has some time.
The global financial market shock in August had a wonderful omen: the Bank of Japan (BOJ) website temporarily crashed before the release of its policy announcement at the end of July. This black humor-like omen foreshadowed the subsequent turmoil in the market. The BOJ's decision to raise interest rates in July, although slightly earlier than expected, was still within the bearable range. However, at the subsequent press conference, Governor Ueda talked about the future interest rate hiking cycle, and this hawkish stance caught the market off guard. The hawkish stance of Japan's interest rate hike is an important background for the release of the US non-farm employment data two days later. The employment data for the United States in July was about 114,000: although this is far from a disaster, it is lower than expected, and the market reaction is excessive. Ueda's tone and the unexpected decline in US employment data triggered two risk concerns: the landing of the US economy and the closing of yen carry trades.
Reflection one: Can the US economy remain strong indefinitely?
Our analysis is that the US economy will decline, which is a "soft landing." But the market is concerned about the inflection point, or the second derivative.
In this cycle, we emphasize that it is not only important to look at the changes but also at the levels. The GDP growth rate of the United States in the second quarter was 2.6%, and consumer spending grew by 2.3%. Even after the disappointing data in July, the average employment data for three months was still 170,000. These numbers are not terrible. The unemployment rate in the United States is 4.3%, which is still very low. The increase of 17 basis points in the unemployment rate last month came from the rise in labor force participation rate. This is not a sign of a significant deterioration in the labor market. In the past, an increase in the unemployment rate has always been a harbinger of economic recession because it not only means a decrease in labor demand but also a reduction in job positions. However, in the current cycle, labor demand has obviously slowed down from an unsustainable speed, but layoffs remain at a low level. In addition, in the past cycles, due to the decline in labor supply during economic recessions, the role of the unemployment rate as an economic signal has become less obvious. This time, due to the increase in labor supply, the rise in the unemployment rate is more obvious. In other words, the unemployment rate of 4.3% is still at a low level, and the economic signal conveyed by its upward trend is far less serious than in the past.Consumer spending, which accounts for about 70% of the US economy, is another prime example. US consumer spending has far exceeded its trend level, and a pullback to a level more in line with the fundamentals of income is necessary. Tight monetary policy will only reinforce this trend. We believe this process is underway, and the August retail sales report indicates that US consumers remain active.
In summary, these data are far from a recession and are heading towards a soft landing, but the market remains vigilant for any signs of more severe weakness.
The upcoming Jackson Hole Global Central Bank Annual Meeting is one of the most closely watched events in the near term. The theme of this conference is "Reassessing the Effectiveness and Transmission Mechanisms of Monetary Policy," a discussion that is timely. We expect Federal Reserve Chairman Powell to outline the Fed's medium-term strategy at this meeting, emphasizing that even if inflation continues to slow, the Fed has the ability to maintain economic growth while also achieving the goal of reducing the inflation rate to 2%. The Federal Open Market Committee has already signaled a rate cut, but Powell may indicate that even with a rate cut, monetary policy will still be tight.
It is crucial to distinguish between current levels and future trend changes. The market often trades based on this dynamic change - that is, the market focuses on inflection points, or second derivatives. We believe Powell's speech will at least implicitly emphasize this distinction: the economy can slow down from unsustainable rapid growth but still be healthy enough to avoid a recession. We expect the continued decline in inflation to drive the rate-cutting cycle, with rate cuts starting at the September Federal Open Market Committee (FOMC) meeting and three 25 basis point rate cuts expected within 2024.
Of course, both we and the Fed could be wrong. State-level data suggest that July employment data were affected by hurricanes and declined, but if August employment data do not improve, we expect the Fed to have a more significant rate cut, but if this situation does occur, it will become a turning point in the trend.
As for Japanese monetary policy, we expect the Bank of Japan to cautiously raise interest rates, with the next one in January. This means that Japan's real interest rates will remain negative until the end of 2025. In other words, the normalization of Japan's inflation and interest rates may take years rather than months to gradually achieve. Therefore, the convergence speed of the interest rate differential between the US and Japan will continue to fluctuate. Concerns about the appreciation of the yen and the unwinding of carry trades will also continue to exist.
Reflection II: Increased Financial Market Uncertainty
The main indices of the global stock market have broken the strong upward trend since last fall and have recently shown more volatility. Many people attribute this to the aforementioned Fed's decision to keep interest rates unchanged in July despite weak employment data. Others emphasize the technical adjustment of the yen carry trade. However, if we look back, the consolidation process of the main stock indices began in April, which was the first major adjustment since the low point in October last year. Even though many stocks and indices rebounded to new highs this summer, the best-performing sectors in the market have gradually shifted to defensive sectors such as utilities, consumer goods, and real estate, which have relatively outperformed. This shift in market structure reflects expectations of a softening economy. The question now is, has this adjustment become a good buying opportunity?
In tracing back to the source, a simple fact cannot be ignored: the valuation of US stocks has reached a very high level. In fact, even assuming that our prediction of a soft economic landing is confirmed, our asset allocation team still believes that there is no room for upward movement in the US main stock indices in about a year. In other words, current stock prices have already anticipated the most perfect situation, and reality is hard to exceed expectations. So far, the forward valuation of the S&P 500 index is still as high as more than 20 times PE. Assuming a soft economic landing, we believe the reasonable valuation is close to 19 times, which means the market is not cheap until the stock price adjusts by more than 10% to reach a valuation of 17-18 times PE.
At the same time, our asset allocation team continues to believe that defensive stocks in sectors such as utilities, healthcare, consumer goods, and some real estate may have more advantages. On the contrary, we are not optimistic about small-cap cyclical stocks that are relatively vulnerable under the current slowdown in growth.In this context, how will the results of the US election impact the market? We typically analyze this by presetting several scenarios. However, a series of recent events have greatly exceeded expectations: First, on June 27th, during the election debate, the incumbent President Biden unexpectedly performed poorly, leading to a decline in his poll ratings. Subsequently, on the eve of the Republican National Convention, former President Trump encountered an "attempted assassination." Just recently, Biden rarely announced his withdrawal from the race. Then, the current Vice President Harris quickly announced her candidacy and gained support within the Democratic Party. Several high-quality polls show that the competition between the two parties in this election appears more intense than last month: National polls show that Harris and Trump's support rates are neck and neck, with a close race. The current poll results in swing states have a small sample size, but the current data shows that Harris can still compete with the Republicans. Therefore, despite twists and turns, the election situation is evenly matched between the two parties, and the market needs to adapt to the uncertainty of the results and the impact of various scenarios again.
The high uncertainty of the election results and the lack of precedent for market behavior before the election mean that cross-asset allocation is an effective strategy to navigate market cycles. In an environment of economic soft landing, stock market allocation should also be more defensive. If the United States raises tariffs after the election, the pressure on the economy may also intensify.
Reflection three: The AI technology revolution cannot be achieved overnight and requires patience.
In the past few months, the US market has been led by AI giants, which has also attracted scholars and Wall Street to reflect, worrying that AI may have been over-expected and bubbled. I believe that the adjustment and reflection on the AI industry are healthy, in line with the development law of new things, and are adjustments after the extreme trend trading of the capital market. This does not deny that the world is still at the beginning of a new round of technological revolution. In the next 5-10 years, AI artificial intelligence will be widely promoted; humanoid robots will rise and be widely used, which will profoundly affect the global economy and society. It is a major change that has not been seen in 30 years since the end of the 1990s Internet technology revolution and the globalization of the industrial chain. However, it is still in the initial stage of the AI technology revolution, and there is still a long way to go before it is widely used in the whole society, and its role in promoting productivity may not be apparent until several years later.
First, let's look at AI artificial intelligence. At the end of 2022, the emergence of ChatGPT ignited the topic of AI and detonated the next round of technological revolution. Although the market enthusiasm is high, it should be recognized that, referring to the context of global technological innovations, including the Internet revolution from the 1990s to the early 2000s, the impact of the AI revolution on the enterprise level will be in turn: the first wave of beneficiaries are enterprises that provide equipment and tools for AI, such as large models and GPU-related providers. The second stage is enterprises that manufacture devices equipped with AI technology, such as consumer electronics like mobile phones. The third stage is enterprises that use AI, including traditional enterprises, benefiting from the efficiency improvement brought by AI. Currently, it is still in the initial stage of the AI technology revolution, and the beneficiaries of the first two stages have emerged one after another, while the third category requires widespread use in the whole society, and there is still a long way to go, and its role in promoting productivity may not be apparent until several years later.
Over the past year, the AI trend has generally belonged to the first stage, with beneficiaries including well-known leaders in GPU chips such as NVIDIA, and pioneers in large language models such as OpenAI. However, the current AI revolution is gradually entering the second stage, from Apple to Samsung, from Bloomberg terminals to autonomous driving cars, which have successively carried AI new technology and become new selling points for ordinary consumers. The pull of these two stages on the development tools of AI itself and the capital goods GPU it uses has already emerged. We estimate that in the next few years, US businesses will invest $3 trillion in AI. The third stage is the widespread application of AI by enterprises in various scenarios to improve enterprise efficiency. This stage has not yet been reached, leading many people to worry whether the huge investment in AI will be a castle in the air? I think this is related to the curve of technological progress.
The enhancement of productivity by AI tools can be divided into three stages: 1) General language tools to achieve information retrieval and answer questions. 2) Custom models that can access company and private domain data, tailored to specific needs. 3) Custom models that can take action according to commands.
At present, we are still in the first stage, and most companies basically use general AI tools, such as ChatGPT, for information retrieval and sorting. From 2024, investment in custom tools and models will increase. Our technology team's survey report on CIOs (Chief Information Officers) of US listed companies shows that 1/7 of the surveyed companies have identified AI as the company's top priority, showing that the trend of large-scale adoption of AI technology in the medium and long term is gradually becoming a trend. However, due to investment returns, data volume, regulatory and technical obstacles, this stage will be relatively slow, and it will not be until the end of 2025 that a broader increase in productivity can be seen. To reach the third stage, it will take several more years, as further technological breakthroughs are needed to attract more enterprises to use them on a large scale.
From some early empirical studies and company feedback, AI has already improved enterprise productivity to varying degrees, such as:
• Companies that have adopted AI-assisted programming tools, mainly large Internet companies in Silicon Valley, such as Microsoft, Amazon AWS, Airbnb, PayPal, and other companies have indicated that their programming productivity has increased by about 20-50%.**AI's Impact on Various Industries:**
- **Drafting Documents:** AI tools can reduce the average writing time by at least 40% and improve output quality by about 20% (a study by MIT involving 453 professionals).
- **Consultants' Productivity:** Consultants using AI can complete 12% more work than before, with work quality improving by at least 40% (a study by Harvard involving 758 consultants from Boston Consulting Group).
- **Financial Industry:** AI tools integrated into company databases achieve a comprehensive understanding of research reports, macro models, corporate earnings, and asset allocation plans, thereby achieving intelligent investment advisory functions.
- **Art Creation and Entertainment Content Generation:** Unity and Adobe's feedback indicates that the time to create 3D content can be significantly reduced with AI tools, increasing creators' productivity by 20% and saving an average of 8 hours per person per week.
These are some examples of industries currently benefiting from AI. As AI technology spreads, it will further support productivity growth. From an industry perspective, AI has the most direct impact on productive services. The future improvement of service industry productivity is expected to become the main driver of Total Factor Productivity (TFP).
One important industry judgment is that AI will drive the rise of humanoid robots. The four major bottlenecks faced by previous robots were limited data volume, insufficient learning and observation capabilities, lack of reasoning and prediction capabilities, and limited human-robot interaction capabilities. The evolution of Large Language Models (LLMs) and General AI (GenAI) helps to overcome these obstacles. It can enable robots to observe and imitate behaviors in the physical world, communicate in natural language, and iterate in data centers. AI algorithms can automate repetitive requests, enhance prediction capabilities, enable virtual simulation, and significantly shorten the R&D cycle of robots.
By conducting top-down and bottom-up analysis of the labor market, considering the repetitiveness, danger, and labor intensity of jobs in different industries, we have sorted out the possibility of humanoid robots being implemented at various industry levels overseas. The industries with the best adaptability include agriculture, construction, mining, building and ground cleaning and maintenance, and catering. However, in management, art design, and entertainment industries, the adaptability of humanoid robots is relatively low. In terms of cost, taking Tesla's Optimus Gen2 as an example, considering quotes from various components, we estimate the material cost is currently $50,000-$60,000 per unit. However, if considering future economies of scale and cooperation with efficient suppliers including China, the manufacturing cost is expected to be reduced to a target of $20,000 per unit, thereby achieving a broader scale effect: we calculate that by 2040, humanoid robots in the US market will be applied to 8 million job positions, generating revenue exceeding $300 billion.
In summary: The trillions of dollars of massive investment in AI in the coming years will inevitably be accompanied by concerns about its returns, which is a normal state at the beginning of each technology cycle. The widespread adoption of AI still requires a longer time and can ultimately promote the improvement of productivity across society. This is a reflection on the three major trends after the global market shock: the slowdown of the US economy is a soft landing, but the uncertainty in the capital market has significantly increased. The AI technology revolution is still in its early stages and cannot be achieved overnight.