You don’t have to be a Luddite to worry about an AI bubble. I am a big fan of AI. I confess that I now use an LLM every day to answer questions that otherwise might take a lot of time and effort. I am more than grateful for the time this frees up. But I try to use this time for reading and thinking. When it comes to bubble-related implications of AI, all you really need to read is one paper, “Financing the AI Buildout,” by Professor Stijn Van Nieuwerburgh (SVN), just published in the Brookings Papers on Economic Activity. It hits what could well be a bubble-pricking nail right on the head.

The SVN paper makes two key points. First, scale: In the long history of the US economy, there has never been a capital spending boom on the order of magnitude like the one we appear to be going through right now. Data from SVN plotted in the first chart show capital spending shares of GDP associated with five earlier

technological revolutions, from canals and railroads of the 19^th century to electrification, interstate highways, and telecom and fiber optic of the 20^th century. On average, the capex that drove these earlier breakthroughs absorbed 1.1% of GDP for an average of nearly 14 years. By contrast, SVN’s tally of the cumulative AI buildout comes in at $10 trillion over the seven years ending in 2032; as shown in the red bar that works out to a 3.6% share of GDP, fully three times the norm of the other historic capital spending booms. The construction of the transcontinental railroad system over the 20-year period of 1870-90 is the only earlier boom that comes close. But the current AI buildout is more than 60% larger as a share of GDP than that one.

The second point is leverage — the lethal debt-intensive characteristic of all major bubbles The chart below from SVN indicates that the AI bubble has now crossed the Rubicon, from a buildout supported initially by the internal generation of positive cash flow by the five largest hyperscalers to what is now a negative cash flow position of these same companies. So often you hear the excuse that unlike the Dotcom bubble over a quarter of a century ago that involved phantom companies, today’s AI behemoths are real companies generating real earnings. That is true, especially from 2020 to 2025; but now, starting this year, their capex buildouts are no longer covered by internally generated cash flow — i.e., according to SVN, they are now in a negative cash flow position, with the shaded bars below the solid bars in the chart below.

That, in fact, is the point of the Google/ Alphabet anecdote that I noted above, where the unintended consequences of rapidly expanding AI-enabled queries are now cannibalizing cash flow from core search activity and forcing the company increasingly into the debt market to square the financing circle. [Note: Alphabet’s net debt issuance (gross less repayments) was in excess of $75 billion in the first nine months of 2026 versus a little more than $11 billion over the comparable period of 2025]. And, of course, this expansion of debt is happening at a time of rising interest rates, making the financing of this massive AI capex boom all the more disconcerting. We heard whispers of these concerns when problems at CoreWeave first surfaced; the case of Alphabet is far more than a whisper.

The basic idea of a bubble is that it distorts the real economy and the financial markets. The historical comparison in the first chart above has massive distortion written all over it. And the second chart speaks to the ominous onset of the leverage cycle. As seen together, these two charts above are painfully reminiscent of the intersection between asset price and leverage bubbles that has made the music stop in the past. The final shoe in any bubble invariably comes from investor speculation: As Bob Shiller has long stressed, bubbles encourage speculators to buy stocks for the simple idea that they are going up in perpetuity — the FOMO mania. While the US equity market has been choppy for the past five months, it remains less than 3% below its all-time high. Mag 7 stocks still account for some 33% of the market capitalization of the total S&P500 index, underscoring the broad market’s lingering vulnerability to the extremes of concentration risk. That last thing we need is to have the AI bubble keep inflating — the post-bubble carnage will then be all the worse. The sooner it bursts, the better.

Yes, I am drawing most of my conclusions from a new peer-reviewed paper by Columbia Business School Professor Stijn Van Nieuwerburgh. In my opinion, it is a superb piece of research that rests on a careful assessment of AI data center campus costs that are then run through the comprehensive Cleanview AI project database. The historical comparisons with earlier capex cycles represent state-of-the-art research. Ignore this paper at great peril.

SHANGHAI—My search for truth inside China has been stymied. Banned from the China Development Forum (CDF), after participating for 25 years as the longest-attending foreign delegate, I feel like an outsider in what was once the country’s most prestigious international gathering. Deteriorating relations with former friends in Hong Kong, some of whom avoid eye contact with me now, have not helped matters. In both cases, I was cut off after offering constructive criticism—highlighting structural malaise in the mainland economy and a loss of autonomous identity in Hong Kong — the essence of what we call debate in the West.

Fortunately, there are other avenues for engagement. One is the Bund Summit in Shanghai, which I attended last week. It reminds me of the CDF in the early days, when, under the leadership of the late premier Zhu Rongji, the exchange of views was focused, vigorous, and open. As one well-connected Chinese friend conceded, “The CDF is now a government conference. The Bund Summit is the real thing—the best conference in China today.”

This year’s agenda, organized around the theme of “Technology, Geopolitics, and the Global Economy,” demonstrated the summit’s breadth and depth: there were panels on monetary policy, fiscal and debt challenges, global imbalances, new energy sources, AI, currencies, and trade. Discussions were structured to reflect strong opinions both inside and outside of China.

Not surprisingly, I was most interested in two sessions—a dinner that I spoke at and a follow-up panel the next day—on Sino-American ties. Both sessions considered whether the relationship between the United States and China is entering a new, less contentious chapter following the first two of this year’s four summits between US President Donald Trump and Chinese President Xi Jinping. The possibility of a new era hinges on the notion of a “constructive strategic stability” that Xi and Trump are now stressing.

To be sure, constructive strategic stability is a classic example of the word salad that diplomats often offer as sustenance in bilateral discussions. But, based as it is on the principles of “fairness, reciprocity, and respect” that were underscored in the readouts of the recent Trump-Xi summit in Washington, this new chapter appears to come as welcome relief, especially when compared to the “reciprocal” tariff war that Trump launched on last year’s “Liberation Day.” After raising the tariff rate on China to 145%—an absurdly high level—the US administration backed down only after Xi played his rare-earths card, underscoring Trump’s TACO reputation (“Trump always chickens out).

But a new label for the bilateral relationship does not constitute a new power-sharing initiative. Some have argued that this year’s Trump-Xi summits are paving the way for a G2 system, in which the US and China set global economic strategy and address geopolitical pressures like the wars in Ukraine and the Middle East. While rumors of a trilateral summit between Trump, Xi, and Russian President Vladimir Putin at next month’s Asia-Pacific Economic Cooperation forum in Shenzhen lend some credence to this view, the consensus at the Bund Summit was that a G2 is not in the cards.

Lastly, it was widely agreed at the Shanghai gathering that the biggest takeaway of the recent meetings between Trump and Xi is their willingness to buy time. Both leaders are preoccupied with domestic concerns. The last thing they need is another dispute over trade, let alone a technological or military conflict.

This aligns with my own assessment of the current state of the US-China relationship: stable for now, but with an uncertain future. The Chinese participants in this year’s Bund Summit were just as realistically cautious: like me, they do not trust Trump’s newfound constructive perspective on China. Abrupt policy shifts are the norm for this US administration. Most recently, Trump fiddled with Canadian tariffs to boost Republican congressional candidates in Maine, Michigan, and New Hampshire; he also issued an executive order to countermand newly enacted legislation (which he signed into law) under which he can impose secondary sanctions on China for purchasing oil to support the military aggression of Iran and Russia.

I went further in my intervention, arguing that whatever Trump says today has no bearing on America’s long-term strategic commitments to any nation. His mendacity and treachery know no bounds—just ask Canada, Ukraine, and European NATO members. He would not think twice about reversing course on foreign policy, even on something as critical as maintaining stable relations with China. In line with this perspective, the Chinese participants in the Bund Summit were more interested in what the post-Trump era means for Sino-American ties. They were particularly interested in whether that era could begin shortly on November 4 of this year—the day after the midterms—rather than after the presidential election in 2028.

The honest exchange of views in Shanghai was a heartening demonstration of what truth telling in China can still be about. The deep insights shared by knowledgeable experts from both sides of the Sino-American divide gave me hope for what the world’s most consequential relationship can look like in a post-Trump future. Between now and then, I’m afraid the best we can hope for is a tenuous stability.

Sponsored by CF-40, a prominent nongovernmental think tank that focuses on macroeconomic policy (and funded by Chinese and western financial institutions), the Bund Summit provides a four-day platform of engagement for around 200 practitioners, split evenly between Chinese and foreign experts. I have attended a few times in the past, most notably in both 2025 and 2026. It reminds me of the CDF in the early days, when under the leadership of Zhu Rongji, the debate was focused, vigorous, and open. Sadly, that is no longer the case for the CDF. As one of my well-connected Chinese friends conceded this week in Shanghai, “The CDF is a government conference. The Bund Summit is the real thing — the best conference in China today.”

This year’s agenda, organized around the broad theme of “Technology, Geopolitics and the Global Economy,” is indicative of the quality of this gathering. There were a number of panels on monetary policy, fiscal and debt problems, artificial intelligence, global imbalances, new energy, currencies, and trade. Discussions were all structured to straddle the debate both inside and outside of China. Unsurprisingly, I was particularly drawn to two sessions on the US-China relationship — at a dinner that I spoke at and in a follow-up panel session the next day.

Both sessions addressed the intriguing possibility of whether a long contentious relationship between the United States and China is truly entering a “new chapter” following the first two of this year’s four summits between Donald Trump and Xi Jinping. The idea, of course, hinges on the notion of “constructive strategic stability” that both Xi Jinping and Donald Trump are now embracing as a descriptor of this new chapter.

I took away three points from this rich discussion. First, the concept of constructive strategic stability is a classic example of the diplomatic word salad that is often offered as food for thought in esoteric bilateral relationship discussions. Based on the combined principles of “fairness, reciprocity, and respect” that were underscored in in the mutual readouts of the recent Washington Summit, the next-chapter framing idea certainly comes as welcome relief. That is especially the case following the extreme tensions of Trump’s so-called Liberation Day gambit, which at one point raised US tariffs on China to an absurd rate of 145%. As per his TACO mindset, Trump chickened out when Xi Jinping played his rare earth card, and the two leaders agreed to a ceasefire when they met in Busan, South Korea a year ago. Stability obviously looks good by comparison.

Second, mutual agreement by the two superpowers on a new label for the state of their bilateral relationship does not constitute a new global power sharing initiative. Some have argued that the summits of 2026 are paving the way for the long-discussed G-2 like arrangement between the United States and China as equal partners in setting global strategy — not just for the world economy, but also for broader geopolitical concerns, like the wars in Ukraine or the Middle East. Recent rumors of a rare trilateral summit between Xi Jinping, Donald Trump, and Vladimir Putin at the upcoming APEC leader’s summit in Shenzhen in November have lent some credence to this view. This new G-2 strategic idea was not endorsed by the consensus at the Bund Summit.

Third, there was general agreement that the most constructive thing to take away from this year’s Sino-American summitry is the willingness of both the US and Chinese governments to buy time. After the extremes of last year’s high-volatility tariff war, anything would be an improvement. Both nations are viewed as having plenty on their plates in dealing with domestic economic and political issues. The last thing they need at this point in time is another contentious relationship dispute over trade, artificial intelligence, and military conflict.

All this aligns with my assessment of the current status of the US-China relationship: Stability for now — and still considerable uncertainty for the future. The Chinese participants in this year’s Bund Summit were realistic in expressing their caution about what was likely to come. Like me, they don’t trust Donald Trump’s newfound constructive perspective on China — for example, fiddling with Canadian tariffs to address some mid-term political pressures in New Hampshire, Maine, and Michigan, or issuing an executive order to countermand recent legislation (that Trump, himself recently signed) that would impose secondary sanctions on China for supporting Iran and Russia through outsize oil purchases. The Chinese delegates at this year’s Bund Summit were much more focused on what the post-Trump era means for the Sino-American policy debate. And they were especially intrigued with the notion that the post-Trump era might actually begin on November 4 of this year — the day after the mid-term elections — rather than coincide with the presidential election two years from now in November 2028.

I deliberately avoided naming names regarding who said what at this week’s Bund Summit. The Chinese government still has a heavy hand in coming down hard on anyone who doesn’t endorse the party line — namely, the so-called good stories of China as per Xi Jinping’s explicit preferences. But having been involved in many important conferences inside of China for more than a quarter of a century, I was heartened by what I heard in Shanghai this week. There was an honest sharing of deep insights from extremely knowledgeable experts on both sides of the Sino-American divide. In the end, that is what truth telling is all about. That gives me hope for a more constructive post-Trump future for the world’s most important relationship.

First the numbers. Shown the solid line in the chart below is the trajectory of the Consumer Price Index in the first 21 months of Donald Trump’s second administration (from January 2005 through September 2026). It is indexed to equal 100 in January 2025, the month that Donald Trump was sworn into office. The dashed line is the same metric for the first 21 months of the Biden Administration, in this case indexed to equal 100 in January 2021, the month that Joe Biden was sworn into office. By the 21^st month of the Biden era, the cumulative increase in the CPI index was 13.5%; by contrast, in the first 21 months of Trump 2.0, the CPI had risen by 5.4%. By these simple metrics, the affordability problem became considerably worse in the first 21 months of the Biden era than was the case in the same time span of Trump 2.0.

Alas, the verdict is not that simple. With the benefit of hindsight, most credible economic research attributes the Biden affordability shock to the direct and indirect impacts of Covid-19, especially the major disruption of global supply chains during and after the shutdown of the global economy in 2020-21 that was subsequently exacerbated by lasting shifts in the composition of aggregate demand traceable to newfound stay-at-home buying patterns of American consumers. The affordability shock of Trump 2.0 is very different. Empirical research traces it to two conscious shifts in economic policy — sharp increases in tariffs and Trump’s war of choice in Iran that has boosted prices throughout the energy complex, from oil and gasoline to diesel fuel and fertilizer. Biden’s affordability problem was an outgrowth of a classic once in a century exogenous shock. Trump’s affordability problem reflects his “own goals” of tariff blunders and a war of choice.

Three important caveats about the numbers in the chart above: First, the calculations are for the total, or headline, CPI. So-called “core’ measures of inflation are important for policymakers and some economists; they were created in the early 1970s to remove volatile food and energy prices, whose mean-reverting ups and downs tend to obscure underlying trend in inflation. Working as a staff economist at the Federal Reserve under Arthurs Burns at the time, I am very familiar with the creation of this construct. But consumers could care less about core metrics; they are pressured by all the things they buy — not just the ones that pass a statistical filtering test. I am still haunted by the “ghost of Arthur Burns,” a Fed chairman who failed to understand the risks of “core creep” — the spillovers between energy and food prices to other segments of the economy. Think, for example, about today’s critical role of diesel fuel in shaping agricultural and grocery prices. A narrow fixation on core inflation dismisses one of the most of important aspects of today’s affordability problem.

Second, the economic narrative on affordability is framed far too much through a focus on inflation rather than on price levels. Inflation is the rate of change in the price level. When a price shock works its way through the system and the inflation rate goes from high to low, we celebrate disinflation — technically, a deceleration in the rate of inflation. But the benefits of such a disinflation ring hollow for consumers, whose aggregate price level continues to rise, albeit at a slower rate. For example, after the CPI headline inflation rate peaked at 9.1% (y/y) in June 2022, over the subsequent 30 months, there were only 4 months when the CPI actually declined (fractionally in all cases) during the ensuing disinflation, which took inflation down to 2.5% in the summer of 2024; in other words, fully 26 months of disinflation occurred when the aggregate price level was still rising.

Third, and related to the second point above, today’s affordability concerns reflect a cumulative problem that spans both the Biden and Trump 2.0 administrations. Prices, as measured in the aggregate level of the CPI, are “sticky” — they continually rise but almost never go down. A deflationary Japan is a notable exception to that rule, with nearly 19 consecutive years of a declining aggregate price level (i.e., a GDP deflator that dropped by 16% from early 1994 to early 2013). But the US is not Japan, and despite all the predictable volatility in food and energy prices, the steady, unrelenting uptrend in the CPI has been largely unbroken for nearly six years. As of September 2026, the headline CPI is up fully 29% since Biden’s first day in office. Not only has Trump failed to address an affordability problem of his own making, but he has failed to purge the residue of affordability pressures left over from the Biden era. No wonder public opinion polls cite affordability as America’s number one economic problem heading into the midterm elections in just over a couple of weeks.

It is tough to tell the affordability story without getting ensnared in America’s intense political debate. I have tried very hard to be apolitical in discussing this issue. During the Biden era and, especially, in the months leading up to the 2024 presidential election, I repeatedly stressed affordability as “The Defining Economic Issue of the US Election.” I didn’t blame Joe Biden for the Covid-related aftershocks that were complicating his political predicament. I simply noted that he owned this tough issue, whether he liked it or not. In the same sense, Trump also owns the affordability issue, but largely for choices of his own — whether he and his MAGA loyalists like it or not. Once again, American voters are about to decide who deserves the blame for this wrenching problem.

Unfortunately, both sides appear headed in this direction. The two presidents have both seemingly embraced the deep friendship that now supposedly binds them to share special insights into the Sino-American relationship. They have left it to their principal underlings on economic policy — He Lifeng (China’s Vice Premier in charge of the economy) and Scott Bessent (US Treasury Secretary) — to do their bidding in hammering out the details of a likely deal by the end of this year. Unsurprisingly, Bessent has taken the bait, apparently canceling his long-planned appearance at an upcoming APEC finance ministers’ meeting in Hong Kong to prepare Trump for his possible trilateral November 18-19 summit with Xi Jinping and Vladamir Putin in Shenzhen. Yet neither Scott Bessent nor He Lifeng can be considered independent thinkers capable of coming up with new and creative solutions to deep relationship problems. They both take their cues from their masters.

The principal problem lies with the vacuous strategy of the Trump-Bessent team. They possess no overarching view of what a successful vision of Sino-American engagement might look like. That’s true on trade, where the US at last month’s Washington summit was only able to cut a two-month extension of the tariff truce that was initially negotiated in South Korea nearly a year ago. A two-month extension accomplished next to nothing — it fell far short of America’s timeworn tactics of “kicking the can down the road.” Moreover, it failed to address the deep concerns of Chinese industrial policies favoring new productive forces in advanced technologies. Nor did it tackle Trump’s vengeful, illegal, and analytically barren increases in US tariffs that only end up exacerbating the affordability pressures bearing down on American consumers.

The lack of US strategy was even more evident in the failure to make any meaningful mutual headway on the existential issues of artificial intelligence. Two loose AI actions came out the recently held Washington summit — the possibility of a hotline and a vague commitment to some type of dialog. But the whole issue was upended by a typically whimsical reaction from Donald Trump in favor of a “hands off” approach to America’s purported AI leadership, despite concerns recently expressed by leading AI insiders of the non-zero probabilities of an “extinction accident.” Relying on his own IQ to render instant judgement on such a profound risk is only the latest demonstration of Donald Trump’s arrogant strategic ineptitude. The President, who just switched sides on “mail-in voting” — one of his longest and most deeply held political positions — could just as easily wake up one morning and decide to nationalize Open AI and Anthropic.

What does China have to gain by negotiating with such a loose cannon? Whatever Donald Trump stresses today cannot be trusted as a meaningful and lasting statement of America’s strategy toward China or anyone else. Just as he used the extremes of post-Liberation Day tariff escalation to threaten China in early 2025, he is just as capable of reversing course on any proposal that he and his leading economic sycophant, Scott Bessent, offer to Xi Jinping and He Lifeng on trade, AI, and a host of other issues. Unless the Chinese team comes to this realization, the greater the likelihood of another meaningless Trump deal.

It seems reasonable to presume that China knows all the above. Yet Xi Jinping persists in playing Trump’s summit-focused deal game. We can only speculate on why. The friendship angle can be easily dismissed as yet another Trump fantasy. Xi has taken down many of his closest long-time Chinese associates — including General Zhang Youxia and Zhong Shaojun — as part of his anti-corruption campaign and is unlikely to offer special dispensation for an unpredictable American president. It is possible that XI is acting to neutralize his exaggerated fears of US containment as a major threat to sagging Chinese economic growth. It is also possible that Xi’s all-important global power aspirations need a stronger ally than Russia, Iran, or North Korea. I suspect that a combination of the two latter considerations is partially at work.

Whatever Xi’s motives, I am not suggesting that China return to playing hardball with the United States. As my next book will stress, I am fully in favor of conflict resolution. I do not, however, believe that there is any meaningful chance that this will occur under the unstable US leadership of Donald Trump. The Chinese would be better advised to recharacterize “constructive strategic stability” as nothing more than “relationship stability.” There is nothing strategic about negotiating with a transaction-fixated US president who doesn’t have a strategic bone in his body. Nor is there anything constructive that can be counted on from a US president whose views on most any issue blow aimlessly with America’s unpredictable political winds. Moreover, Trump has an uncanny knack of offending US sensibilities of civil discourse with his constant lies, an unprecedented vengeance campaign, and his ego-driven renaming of national monuments (the Kennedy Center, Denali, and possibly the Ford Theater) international bodies of water (the Gulf of Mexico, Lake Ontario, and most recently, the Atlantic and Pacific Oceans); now, he is even insisting that artificial intelligence be renamed super-intelligence.

The media and the punditry are guilty of making far too much out of the purported progress that so far has been accomplished by the first two summits of this year. Eager to extrapolate more progress into the next couple of summits, there is a growing sense that the US and China have now addressed the big issues that have come between them, from trade and AI to Taiwan and other geostrategic risks (like Russia, Iran, and Ukraine). Nothing could be further from the truth. It is not in China’s best interests to buy into America’s Trumpian spin. Hopefully, a much better opportunity will come after 2028.

Previously, I have underscored the dramatic narrowing of the innovation gap between the United States and China. This has, of course, been going on for a number of years, and not much has changed in 2026. China held on to 10^th place in the GII universe — about the same as its ranking in 2025 but up dramatically from 43^rd place in 2010. Over that same period, the United States moved from the 11^th position in 2010 to 3^rd place in 2026; since 2019, the US ranking has fluctuated in a tight range between 2^nd and 3^rd. Needless to say, China’s ascendancy by 33 positions since 2010, dwarfs that of any other nation in the GII sample, including, of course, the United States.

The GII is not designed to pinpoint the potential for any specific innovation. It covers a broad range of innovation inputs — institutions, human capital and research, infrastructure, market and business sophistication — that could go into any one of a number of innovative activities. The output side of the GII, which attempts to approximate that range of innovative activities, is equally broad; it stresses the “production” of innovative inputs into the all-encompassing categories of knowledge, technology, and creative outputs.

What are we to make of such an abstract vision of a nation’s innovative potential? My own suspicion is that the GII is best at capturing the rolling cycles of different innovations over time. In recent years, that could range from computer hardware and wireless telecommunications to alternative energy generation and electric vehicles. But now, of course, it seems reasonable to conclude that the GII currently bears most critically on the greatest innovation of our time — Artificial Intelligence.

Of all the 54 detailed subcomponents on the input side of the Global Innovations Index, I would argue that test scores of high-school student matter most in assessing a nation’s AI potential. Sure, financing capacity, data-based infrastructure, venture capital cultures, university-centric technology clusters, and regulatory oversight — just to name a few of the other prominent GII components that are related to AI development — all matter a great deal. But at the end of the day, secondary schools that produce super-smart hard-driving young entrepreneurs focused on the theory and applications of the deep science of artificial intelligence could well matter the most.

On that count, there is a compelling case to argue that China is far better placed than the United States. This shows up in the so-called PISA scores of 15-year-old students, the most comprehensive comparative assessment of aptitude in reading, mathematics and science. Conducted every three years under the auspices of the Program for International Student Assessment, PISA scores have been a key component of the GII since its inception in 2007. As the chart below shows, with the exception of the three years from 2017-19, Chinese students have ranked number one in the world on PISA scores since 2011; by contrast, 15-year old high-school students in the United States have ranked in 24^th place, on average, over this same period. Putting it another way, the raw scores in the PISA examination for Chinese students were 15% higher than they were in the United States over the 2011-26 interval; in the latest PISA cycle (centered on 2025), that gap stood at 18%.

According to documentation on the PISA website, the assessment is aimed to capture student aptitude near the end of compulsory education. It is coordinated by the Organization for Economic Cooperation and Development (OECD), and since 2000 has been conducted every three years (with a one-year delay in 2022 due to COVID-19). Starting in 2025, PISA has moved to a four-year testing cycle. During each PISA cycle, a major testing domain is emphasized on a rotating basis; in 2025, science was the major domain, which followed mathematics (2022), and reading (2018). For China, the students tested since 2015 are only from four provinces — Beijing, Shanghai, Jiangsu and Guangdong (2015), and Zhejiang (2018 and 2025); in 2025, these four provinces accounted for 13% of the total Chinese population (which equates to 56% of the total US population)..

I am not attempting to use PISA scores to render the definitive verdict on the US-China AI race. There are obviously innumerable other considerations that will bear on the outcome. But unlike other major breakthroughs that were hugely dependent on innovations in production, the logistics of scale, the emphasis on quality, the deep pockets of venture capital, and thriving start-up cultures, AI seems far more dependent on the raw talent of an exceptionally bright younger generation of talent. At #10 in the latest GII, an ascendant China still remains behind the third-place US. But with China well positioned to dominate the most knowledge-intensive innovation of our lifetime, that may end up being bitter consolation

CH was tenacious in his commitment to free and open debate. He welcomed and encouraged outside views on China and Hong Kong from those of us from the West. He was utterly determined to find a constructive path for Sino-American re-engagement. And he was inherently optimistic that such a resolution was not just essential, but both possible and likely.

The following was published in “China-US Focus,” a flagship effort of the China-US Exchange Foundation founded by Tung Chee Hwa in 2008.

Tung Chee Hwa’s Strategic Insights

My first in-depth interaction with Tung Chee Hwa came at a time of crisis. In April 2001, an American EP-3 spy plane had just collided with a Chinese military jet and was forced to make an emergency landing in Hainan Island. At the time, I was running a Morgan Stanley investment tour of China for about fifteen of the world’s most prominent and deep-pocketed investors, most of whom had never been to China before.

Our carefully orchestrated field trip was quickly in shambles. Many of our meetings on the Mainland were downgraded or canceled as a demonstration of protest by the Chinese government over the spy plane incident. Fortunately, that was not the case at our final stop in Hong Kong, where we were warmly greeted by Chief Executive Tung.

He quickly focused on three aspects of this crisis: He apologized for any inconveniences that had disrupted our carefully planned schedule. He also underscored his deep admiration for the United States, not just for its values as the bastion of Western democracy but also the birthplace of his three children. Lastly, he noted that while this incident will surely blow over, it should also be taken as a warning of more trouble to come if Washington and Beijing were not more attentive to the complexities of their vitally important relationship.

It was an uplifting end to our trip and as I shepherded the investors out of the room, Tung Chee Hwa asked me to stay behind. He encouraged me, and by inference our influential group, to play a more active role in resolving this crisis. I was a bit aghast over this suggestion but asked him if he had any advice on a course of action. His quick response was a name — Paul Wolfowitz, newly installed US Deputy Secretary of Defense and one of America’s leading hardline neocons. Mr. Tung knew this was more of a military than a diplomatic issue and suggested we should get in touch with Wolfowitz to press our case. He urged me to fill him in on the outcome. He also insisted that henceforth I call him “CH.”

I drafted a letter to Wolfowitz on the plane trip back to the US, and much to my surprise he quickly accepted our meeting request. In the end, there was a last-minute conflict that forced him to send a substitute. That turned out to be a most fortuitous adjustment. We ended up meeting with the legendary Andy Marshall, Director of the Pentagon’s Office of Net Assessment who ran a renowned civilian strategic think tank for the US military. When I chatted with CH a few days later and told him of this meeting, without revealing the substance of the discussion, he was very encouraged. My notes say he said, “You were able to get this issue in front of the Pentagon’s most important strategic thinker, exactly what was needed in this case.”

This meeting was pivotal for me. It opened the door for a long-standing relationship with Andy Marshall that over the years was of immense help to me in grappling with geopolitical tensions, especially insofar as China was concerned. But it was also a demonstration of CH Tung’s uncanny knack for grasping the critical strategic implications of the US-China relationship. For the next twenty years, only cut short by his failing health after 2020, CH and I had frequent meetings and late-night telephone conversations, grappling with the very strategic issues he warned of during our initial encounter in 2001.

In his heart, CH was a strong believer in the overarching concept of “grand strategy” as a unifying framework to think about relationships between major powers, especially the Sino-American relationship. To some extent that was an outgrowth of his long-standing relationship with Henry Kissinger who viewed grand strategy through a balance-of-power lens. CH constantly urged me to think about US-China relationship problems in that same strategic construct. He was most emphatic in stressing the critical importance of this relationship between the two superpowers. Many times, he ended our conversations by saying something like, “We have to do everything in our power to get this relationship right.”

By establishing the China-US Exchange Foundation (CUSEF) in 2008 at the ripe young age of 71, CH put much of his energy into that very objective. Its focus was on dialogue and exchange between two powerful nations that were running into trouble in managing the relationship between them. Sino-American engagement had fallen out of favor, and through CUSEF, CH Tung was determined to do his best to correct that problem. And he certainly found a kindred spirit in me. Over the years, I have been privileged to have participated in many of CUSEF’s activities and fora.

But typical of CH, when thinking about the mission of CUSEF he asked me the same question he posed at the end of our first meeting in 2001, “What are you going to do about it?” By “it” he meant the US-China relationship. By then, I was at Yale and had entered the teaching and book-writing phase of my career. My focus was — and still is — on the codependency of America and China. This forced me to deepen my research on the close ties that both nations and their economies have with each other.

CH, given his own focus on the relationship aspects of Sino-American engagement, was an enthusiastic supporter of my initial approach to codependency. But I must be perfectly honest, he was not thrilled with my subsequent prognosis of a conflicted codependency. It made for some contentious discussions between us in his later years. He urged me, once again, to think of a way out. One book turned into two, and I have just completed a third book on the Sino-American relationship that lays out a detailed plan for conflict resolution. I am reasonably certain that my dear friend, CH, would be quite pleased.

To me, strategy was one of CH’s greatest strengths. But in the spirit of realpolitik, he was always pragmatic in recognizing the hard work that it takes to implement any strategy. He leaves us with a legacy of principled determination to do everything in our power to put a still troubled Sino-American relationship in a much better place. With the world (including Hong Kong) suffering from a serious leadership deficit, his warm smile will always motivate me to think of what is possible.

NEW HAVEN—Long on ceremony but short on substance, the eagerly anticipated summit between US President Donald Trump and Chinese President Xi Jinping was a disappointment. Perhaps not a major disappointment, as expectations were generally low and as some talk is certainly better than no talk. But on the major issues of trade, AI, Taiwan, and addressing China’s role in two theaters of war, there was little or no progress. Instead, ceremonial flourishes substituted for substance, unmasking the summit as yet another Trump vanity project.

On trade, the two-month extension of the current truce—initially negotiated nearly a year ago in South Korea—was surprisingly short, implying that tariff uncertainty remains high. Pre-summit chatter hinted at an extension of 6–12 months. Moreover, there were expectations of another round of Chinese purchases of US agricultural exports. Instead, the two nations made a joint post-summit announcement to cut tariffs on a total of $60 billion of “non-sensitive” goods, including trivial products such as US imports of electric blankets and magic tricks and Chinese imports of eels and emus.

On AI, the biggest issue facing civilization, there was no meeting of the minds. The two superpowers committed to a loosely structured dialogue and dangled the possibility of an unspecified “hotline” for AI-related national-security issues. This bare-minimum achievement is far from the major breakthrough on AI safety many had hoped for—especially in the aftermath of the recent Hugging Face hacking incident and reports of other intrusions that have led some AI insiders to raise human extinction concerns. The disappointing outcome of the summit — which merely paid lip service to such risks — is hardly surprising considering that Trump has doubled down on his hands-off approach to AI regulation.

On Taiwan, Xi attempted to press the advantage he gained after the Beijing summit in May, when Trump described the pending $14 billion US arms package to Taiwan as a “good negotiating chip.” In view of America’s normally staunch commitment to arming Taiwan, the Chinese side saw this as a major concession, especially given that the deal already had received congressional approval. In his opening speech to last week’s summit, Xi urged the US to “adhere to the correct position of opposing Taiwan independence.” But doing so would be a stunning about-face on the “strategic ambiguity” that has long shaped US strategy toward Taiwanese sovereignty.

Finally, China’s role in the wars in Ukraine and Iran escaped scrutiny. As the buyer of some 90% of Iran’s oil exports and about 50% of Russia’s, China is a major funder of both countries’ military efforts. A new law gives the US president broad authority to address this activity through secondary sanctions and tariffs on those countries like China, that indirectly support these illegal wars. Don’t count on it. Fearful of spoiling the summit’s optics, Trump made no mention of this political hot potato.

By making little effort on the big issues, both in terms of staff preparation and bilateral talks, Trump was painfully transparent about what mattered most to him: impressing Xi and his entourage. Ironically, while the tightly controlled Chinese press delivered enthusiastic messaging back home of Xi’s great successes at the summit, the US broadcast media was muted by Trump’s White House ban on three major news outlets, a shocking crackdown on free speech.

As was the case in Beijing, Trump once again abandoned any semblance of protocol in his cringe-worthy declarations of heartfelt admiration for his “truly great” friend’s leadership skills. The hyperbole was embellished by physical demonstrations of Trump’s affection. Washington’s profusion of red carpets would put even China to shame, while the glitzy state dinner with Big Tech “broligarchs” was missing only the grand setting of a Chinese-style ballroom.

Trump attempted to compensate for that by offering Xi a personal tour of his new vanity projects in the White House: the new granite helipad, the ballroom construction project, and his rogues’ gallery of annotated presidential portraits. Xi was shown in a rare state of laughter when he came to Trump’s crude and derisive depiction of former US President Joe Biden as an “autopen”—suggesting that the two leaders may share more personality traits than we have been led to believe.

I am hardly the first to bemoan the autocratic symbolism of political leaders’ vanity projects. But there is something considerably more vulgar about Trump’s approach. It consumes him as he approaches the end of his presidency. By slapping his name on monuments, programs, and US financial instruments, he routinely violates America’s sense of propriety, which has long held that naming rights are to honor and memorialize the deceased. Not so for Trump, who audaciously claimed “I am the code” when apparently clashing with the ballroom architect over safety issues.

Unfortunately, this has everything to do with Trump’s approach to high-level meetings with America’s most challenging counterpart, China. The heavy lifting of Sino-American engagement has never been needed more to address a host of bilateral and global challenges. Historically, at critical moments like this, leaders in both nations have risen to the occasion. That was not the case on September 24.

There is an important asymmetry on how the two leaders see this moment in time. Xi, with his constant references to a world that is “undergoing major changes unseen in a century,” underscores the gravity of today’s circumstances. Trump couldn’t care less. His focus on befriending and impressing autocrats — not just Xi but also Russian President Vladimir Putin and North Korean leader Kim Jong-un— and insulting and alienating long-standing allies is part of his twisted sense that the trappings of power are power itself. America’s first-ever vanity summit, while disappointing, sadly comes as no surprise.

On trade, a two-month extension of the truce — negotiated nearly a year ago in Busan, South Korea — was surprisingly short. Pre-summit chatter on the extension had been in the range of six to twelve months. Moreover, expectations had focused on another round of Chinese purchases of US exports. Normally those negotiations are completed before a summit even begins. Not so this time. US Trade Representative Jamieson Greer has hinted we may see something on incremental purchases in a few days. But don’t hold your breath for successful implementation, even if that happens. The Phase I trade deal failed miserably on that count and recent reports of farm goods purchases are tracking well below earlier Chinese promises. The bottom line on trade: Limited truce extension implies that tariff uncertainty remains high.

On AI, the two superpowers have done the minimum. They committed to ongoing discussions, rather than a formally structured dialog. And they dangled the possibility of an unspecified AI communications “hot line.” The likelihood of a major breakthrough on AI safety, as a result, is lower than many had hoped for — especially in the aftermath of the recent Hugging Face hacking incident and reports of other AI intrusions. However, in light of President Trump’s recent “hands-off” statement on US intervention in domestic AI security, this week’s summit outcome on AI cannot be considered a major disappointment.

On Taiwan, Xi gets the nod from the standpoint of power projection. He continued to press the advantage he gained from Trump last May, when on the plane ride back home from the Beijing summit, the US president described the pending $14 billion arms sale to Taiwan as a good “negotiating chip.” The Chinese side considered this as a major concession from America’s previous unwavering commitment to delivering on defensive arms packages that had already been approved by the Congress. In his opening speech to this week’s summit, Xi was very clear in stressing that the US should “adhere to the correct position of opposing Taiwan independence.” This would be a stunning backtracking from the “strategic ambiguity” that has long shaped US strategy toward Taiwanese sovereignty.

Finally, there appears to have been no attention addressed to China’s key role in supporting the serious wars in Ukraine and Iran. As the buyer of 90% of Iran’s oil exports and about 40% of the oil sent abroad by Russia, China has played a major role in funding the aggressive military operations of both nations. Passage of a recent bill on secondary sanctions now gives the US president broad authority to address this activity through highly punitive tariffs. Don’t count on it. There appears to have been no mention of this “hot potato” at yesterday’s summit.

By doing the minimum on the big issues, both in terms of staff preparation and in short bilateral chats, Donald Trump was painfully transparent on what mattered most to him in this summit. He went beyond any semblance of normal protocol in declaring heartfelt admiration for his “truly great” friend. Add to that, a multiplicity of red carpets that would even put China to shame, the glitzy State dinner with all the tech bros, and his personal tour of Xi to show him all the new vanity projects in the White House — the granite helipad, the ballroom construction project, and a walkthrough of his rogues’ gallery of annotated presidential portraits — there can be little doubt of how the Trump-Xi Summit of September 24, 2026, summit will be remembered. It was America’s first vanity summit.

It’s not just the legacy effect. I also stressed three unique sources of ongoing superpower conflict that are at work today: a protracted trade war that continues to place intensified US tariff and sanctions pressures on China; the likelihood of China being targeted by the US for secondary sanctions due its role as the major foreign customer for Iranian and Russian oil; and the crescendo of heightened concerns over the existential extinction potential of artificial intelligence.

I do not expect meaningful breakthroughs in any of these three areas. On trade, a one-year extension of the existing truce is the most likely outcome, with average statutory US tariff rates on Chinese imports holding at 47.5%, more than double the 20.7% average when Trump returned to office in January 2025 and nearly seven times the approximately 7% statutory tariff rate imposed on goods imported from the rest of the world. Secondly, President Trump just signed into law the “Lindsey O. Graham Sanctioning Russia and Iran Act of 2026,” which gives him broad power to impose steep secondary sanctions in the form of 100% tariffs on China as the major purchaser of Russian and Iranian energy products. Lastly, while many proposals have been floated on AI, it is hard to get encouraged when the US side is initially represented by Scott Bessent, wannabe market manipulator and the most politically compromised Treasury Secretary in modern history.

While it is always possible there may be more progress on these three issues than noted above, that doesn’t detract from my long-standing critique of the personalized approach to leader-to-leader summitry. Post-summit assessments are invariably couched in terms of reading between the lines of leader body language, statements, and official communiqués. Yet with two untrustworthy leaders, that is patently absurd. As Donald Trump has demonstrated repeatedly — especially in his second term — literally nothing he says can be trusted as a truthful representation of US policy. He has shredded historic alliances, broken his promises on foreign wars and affordability, and has orchestrated countless attempts to crush America’s once cherished rule of law. Xi Jinping, while less inclined than Trump toward constant repetitive actions of deceit and distrust, has among other things lied about China’s nonmilitary intentions in the manmade “islands” of the South China Sea, misrepresented the intent of his “unlimited partnership” with Vladamir Putin, abrogated the 50-year timeline of Hong Kong’s assimilation in the PRC, and tightened the noose on censorship and information control.

Notwithstanding the serious character flaws of Trump and Xi, post-summit commentary is likely to take the implied trust of leader-based messaging on “constructive strategic stability” far more seriously than it deserves. That raises an important point on the meaning of trust in the current climate. Trust is essentially a voluntary state of acceptance in another person, nation, or idea. It reflects largely a psychological state of confidence that can be expressed through a nod of the head, a hug, a handshake, or even a signature on a document. That makes trust a soft connection, a temporary state with no guarantee that it won’t change on a whim as circumstances or leader personalities dictate.

Sino-American conflict resolution doesn’t need another superficial demonstration of trust between two great friends, as Trump is wont to say about Xi. I would argue, instead, that it needs commitment, more of a firm and lasting promise, a self-renewing hard pledge. As per Thomas Schelling, behavioral economics views commitment largely through institutional arrangements that increase the credibility and durability of future cooperative behavior. Commitment is harder wired than the looser presumptions of trust; it constrains the temptation to defect from agreements when short-term situational conditions may change. Commitment needs assurances in the confidence of a relationship but also requires formalized institutional mechanisms of exchange and engagement between nations to validate such assurances.

Commitment is the antithesis of deal making. The “deal” purportedly offers a short-term fix to a tough problem. It is a transaction based, myopic substitute for longer-term strategy. Most importantly, the deal offers no guarantee of continuity or follow-up. The so-called Phase I US-China trade deal of 2020 is a recent and important case in point. In what President Trump boasted was a “historical trade deal,” China agreed to purchase $200 billion of additional US exports over a two-year time frame ending in 2021; at the same time, the concept of a first phase alluded to a succession of more phases to come. The Phase I deal failed on both counts — it did not deliver new, incremental purchases and lacked any next-phase follow ups. Deals commit to nothing beyond the signing ceremony.

Nor is commitment equivalent to the permutation of “trust and verify” — a popular construct in modern diplomacy that relies on the assurances of compliance metrics to guarantee trust. Since the late 1980s, verification through surveillance and monitoring has become a central feature of most major arms agreements. That includes existing and potential stockpiles of nuclear weapons, strategic and anti-ballistic weapons, and chemical weapons. Onsite inspections were built into verification mechanisms with the former Soviet Union and were a key aspect of the now abandoned nuclear deal with Iran. However, the need to verify trust reflects an inherent state of suspicion between adversaries. Adding hard verifications to a soft trust does not provide the lasting assurances of commitment.

Commitment is a more positive and lasting condition that draws on institutional arrangements which increase the credibility and durability of future cooperative behavior. Commitment is harder wired than the looser presumptions of trust; it constrains the temptation to back away from agreements when short-term situational conditions may change. Commitment needs assurances in the confidence of a relationship but also requires formalized institutional mechanisms of exchange and engagement between nations to validate such assurances. Trust is a necessary, but not sufficient requirement for the more rigorous, disciplined, and binding condition of commitment.

I have just completed a new book on a commitment-based approach to sustained Sino-American conflict resolution. I would be shocked if the upcoming Xi-Trump summit takes a meaningful step in that direction.

In this April’s issue of the World Economic Outlook, the IMF presented a “resilience assessment” of the global economy in 2025 (chart below). This news-driven assessment estimated that the build-up of downside risks initially amounted to about -0.7 percentage points off global growth — a hit to world GDP that was subsequently completely unwound by the events after April 2025, especially, a TACO-related retreat in US tariffs, sharply improved financial conditions, and an emerging AI boom. My warning that “global recession risks are high and rising” coincided with this trough in the current “resilience cycle.”

In sounding the alarm, I relied on a time-tested framework that has generally served me well in my four decades as a forecaster. It hinges on the notion of an economic “stall speed” — a subpar pace of economic activity that lacks the cushion of resilience needed to withstand the blow of a significant exogenous shock. I argued that with baseline global GDP growth of just 3% for 2026 — a slowing that was on the cusp of the upper reaches of the 2.5% to 3.0% stall speed danger zone — the world was but one major shock away from outright global recession. I concluded that the lagged effects of conflict and energy-related supply bottlenecks were enough of a shock to warn of global recession. What happened?

The defensive answer is to be mindful of the lags — give the shocks time to work their way through the system. This may still be the correct answer, but I must concede that it is not an entirely satisfactory explanation of how things have played out thus far. Some shocks, like wars, can hit quickly. If anything, that would tend to argue for shorter lags — especially if energy shocks reflect the abrupt disruption of the supply of crude petroleum and related petrochemical feedstocks that, in turn, take a toll on the worldwide agricultural complex.

The sharp contraction of oil tanker traffic due to six months of Iranian pressure on the Strait of Hormuz (which used to handle about 20% of world oil consumption), is now being reinforced by (Iranian-backed) Houthi pressure on the Red Sea chokepoint of the Bab-al-Mandab Strait ( which had handled another 10% of world oil consumption). Together with recent closure of a key Saudi pipeline due to drone attacks from Iraq, the risks of a full-blown global energy crisis are now high and rising.

The one possibility that might be limiting likely energy-related impacts (both on quantity as well as prices) is the buffer of strategic petroleum reserves. But with these petroleum reserves now severely depleted by the 400-million-barrel emergency release of these stockpiles coordinated by the IEA, it still seems to be only a matter of time before the lags of increased geopolitical risks kick in. All in all, taking both the downside risks and the upside offsets into consideration, the IMF’s model-based calculations suggest that the current array of such risks is likely to reduce world GDP by around 0.5% within two quarters, then by double that amount (to -1.0%) over the following four quarters. As per the stall-speed framework noted above, that qualifies as a potentially decisive shock — one that could well reduce a baseline global growth estimate of 3% down to, or below, the global recession threshold of 2.5%.

To be sure, there are other possible complications that bear on global resilience. The most obvious is an offset stemming from the AI surge. Ballpark estimates of this possibility are all over the place — from a modest annual increment of 0.5 percentage point on global growth to a tongue-and cheek hypothetical increment of 20% to 30% per year posed by the always cheeky Economist predicated on the “singularity” of AGI (artificial general intelligence). However, with the warnings of AI-related threats to civilization now reaching a crescendo in the aftermath of the “Hunting Face hack” by seemingly autonomous actions of clustered AI agents, the possibility of a downside AI shock has suddenly gotten considerable attention; needless to say, with the surge in data-center construction expected to boost economic activity sharply for the next 5-10 years, a negative AI shock could pull this rug out from under this widely expected powerful assist to global economic growth.

Other structural shifts in the composition of the global economy might also be an upside offset that my old framework doesn’t adequately capture. Emerging and developing countries currently account for 61% of world output — fully twenty percentage points greater than their average share of 40% over the 1980 to 2010 period when my forecasting framework performed so brilliantly. To the extent that the composition of the global economy has shifted away from crisis-battered and relatively sluggish growth in the developed world into more resilient and relatively rapid growing economies in the Global South and elsewhere in the developing world, there may be a need to recast the very concept of global resilience. How that conclusion holds up in the face of a protracted slowing in the Chinese economy remains to be seen.

Lastly, we tend to think of globalization as an amplification mechanism that compounds global shocks in an increasingly integrated global economy and in well-connected financial markets. That has generally been great news for supply-chain efficiencies and connected global capital markets, both of which tend to increase global resilience. But in an era of trade shocks, supply-chain disruptions, and homeshoring, de-globalization has led to a fragmentation of those forces; over time — those lags, again — these developments would tend to diminish the amplification and resilience of cross-border and cross-market connectivity. Moreover, it is a stretch to consider America’s TACO-related moderation of severe tariff hikes relative to the bluster of Liberation Day as the functional source of resilience as per the IMF news-accounting framework illustrated in the chart above. I certainly don’t buy the notion that the remaining residue of tariffs and trade disruptions is cause for celebration.

My bottom line on global resilience remains pretty much the same as it was last April. It’s always hard to be patient with lags, especially in the case of the most disruptive of shocks, like conflict and energy shortages. But I still suspect that global economy is headed for its stall speed. That means it will take another sharp upleg in the resilience cycle to avoid global recession in 2027 — a little later than I initially thought but a decisive downturn, nevertheless.

Don’t hold your breath for a breakthrough. Apart from two historic summits in the 1970s — Nixon-Mao in 1972 and Carter-Deng in 1979 — summits between the two leaders of the United States and China have generally produced little of great substance As I have noted before, and as can be seen in the table below, since the 1970s, most of the subsequent 21 leader-to-leader summits fall in the place-holding or dashed-expectation categories in the taxonomy below. That includes the two Trump-Xi summits of 2017 that were followed almost immediately by the onset of the US-China trade war in 2018, as well as the Beijing summit held in May of this year that made vague, and still unfilled, promises to establish new bilateral trade and investment mechanisms.

The Trump-Xi Summit of late 2025, held on the sidelines of an APEC gathering in Busan, South Korea, did succeed in bringing the post-Liberation Day escalatory phase of the US-China trade war back from the brink. The US reduced sharply elevated tariffs on China to a still excessive 47% and China lifted its stranglehold on rare-earth exports to the US. There were ancillary agreements on soybeans and fentanyl precursor chemicals, but the goal was more damage control than meaningful progress on conflict resolution.

On one level, it is surprising that a summit is even occurring this month. That’s because Trump’s latest pivot on Iran, the “economic D-Day” strategy, takes dead aim on China. In Treasury Secretary Scott Bessent’s words an “economic onslaught” is now being aimed at crushing the Iranian economy by severing its links with the rest of the world “until Tehran stands alone.” With exports accounting for slightly more than 25% of Iranian GDP and about 43% of that made up of crude oil (and a considerably larger share if petrochemicals and other hydrocarbon products are included), there can be no mistaking the implications of this strategy. With China currently buying fully 90% of Iranian crude oil exports, it is obvious that that there can be no D-Day landing without hitting China hard with onerous secondary sanctions.

A similar inference can be drawn from Trump’s Ukrainian strategy. If he was at all serious about putting pressure on Russia to end its barbaric and illegal assault on innocent Ukrainian civilians, comprehensive secondary sanctions on Russia’s trading partners would be an equally effective threat to do grave damage to an export-led Russian economy. Trump has done that with India, but not with others, especially China, who purchases over 40% of Russia’s crude oil exports. Of course, that would be particularly upsetting to Xi Jinping who has gone out of his way since early 2022 to endorse China’s “unlimited partnership” with Russia. Here, as well, there appears to be no desire by the Trump Administration to get tough with Russia and, by implication with its largest patron, China. Meanwhile, a tragic war in Ukraine rages on.

If President Trump were to take aggressive action against Iran or Russia, China would immediately cancel the upcoming summit. Bessent claims the Trump Administration prefers “quiet diplomacy” rather than stating the obvious about China’s culpability in providing economic support to Iran. That would be the equivalent of then General Eisenhower ,as Supreme Commander, ordering the original D-Day invasion armada of some 6,900 vessels to stop in the middle of the English Channel while Allied forces negotiated with Germany. A similar conclusion can be drawn from Trump’s strategy to have sent his inexperienced negotiating team of son-in-law Jared Kushner and real estate buddy Steve Witkoff on a recent mission of quite shuttle diplomacy between Moscow and Kyiv.

The big question is, why? Why doesn’t Donald Trump execute an aggressive strategy of secondary sanctions on Iran and Russia to move the needle on serious and potentially devastating conflicts in the Middle East and Russia? The answer is painfully obvious. He does not want to unsettle what he believes is his most powerful source of leverage — a great personal relationship with Xi Jinping. If his good friend were to cancel the summit, he fears his already tainted reputation as the world’s greatest deal maker would be in tatters.

Don’t kid yourself, Xi Jinping is very much aware of the importance that Trump puts on the stagecraft of summitry and the questionable claims of great personal friendship with Xi. That was certainly the case with the two summits in 2017 and very much the case today. But for Xi, the current circumstances are very different than they were nine years ago. Since then, he has placed considerable emphasis on China’s global strategy. He has succeeded by diversifying China’s export markets away from the US, by focusing on strong alliances with Iran, Russia, and North Korea — ironically, the same relationships that Trump covets — and by emphasizing his intent to challenge the long-established US-centric western rules-based order. If Trump thinks he can prevail on the basis of another glitzy summit with his so-called good friend, that is precisely what Xi Jinping wants him to believe. Some might even call that a trap.

Meanwhile, there has been little of the way in serious planning for the upcoming Washington summit. The White House is more focused on the guest list and seating plan for the upcoming State dinner than it is on a meaningful strategic agenda. Nor has there been much in the way of advance planning for a long awaited critical AI summit that is also supposed to take place this month. All this underscores my long-standing criticism of the current architecture of US-China engagement – too much reliance on the personalization of diplomacy to guide and manage an increasingly complex relationship between two deeply conflicted superpowers.

Following the Busan Summit of October 2025 both the US and China were quick to claim that a contentious trade war had given way to a new era of “constructive strategic stability. “ It looks to be more like a charade of strategic ambiguity, with China clearly having the upper hand.

Amid all the extraneous noises of geopolitics, war, domestic political tensions, Warsh spoke of one over-arching strategic consideration: It is high time to declare an end to the emergency state of monetary policy. That means that it is no longer appropriate for the Fed to maintain the emergency settings that were established 17 years ago in the depths of the Global Financial Crisis. That is true of the Fed’s balance sheet strategy, as Warsh has stressed for some time, and it is also the case with respect to his views on forward guidance of a pre-designated policy trajectory

This is a long overdue development. Like Japan, balance sheet leverage was put into play as the Fed ran out of basis points in the face of mounting deflationary risks. Forward guidance became an instrument of determination and commitment to prospective policy actions in the depths of crisis. Crisis-torn economies have a special need for such actions, especially when pre-crisis policy interest rate settings are close to the zero-bound as they were in 2002-04. Market participants need the assurance that while the central bank may be close to running out of traditional policy options, there are plenty of non-traditional, or unconventional, actions that could be deployed. Ben Bernanke made this case in a series of speeches in 2002 after the bursting of the dotcom bubble.

Until Kevin Warsh, however, no Fed Chair has spoken forcefully of postcrisis policy normalization — even though the crisis in question , as well as its precarious aftermath, ended over fifteen years ago. The longer the Fed maintained its crisis-dependent stance, the harder it became to break the financial markets of old habits.

Warsh’s Jackson Hole speech emphasized the heroic assumptions that are required to maintain a credible guidance of future policy intentions. Not only does such an approach require accurate forecasts of the dual-mandate objectives of inflation and employment, but it also needs model-based linkages back to the policy interest rate. As the new Fed Chair implies, there is no mechanistic Taylor-like rule that “could be rigorously relied upon” to provide a provide a policy reaction function to guide markets with any degree of precision.

Warsh concludes his conceptual case for normalization with the “hall-of mirrors” paradox — of markets focusing too much on a Fed guidance that draws its support from market and asset prices that were themselves outcomes of the Fed’s own prior messaging. As he put it, such circularity can blind the central bank to unexpected developments, leaving it unprepared for exogenous shocks in the future and, as a result, prone to policy blunders. The inflation problem of 2021-22 was an especially egregious example of that tendency. These are all inarguably powerful points in support of the case for ending crisis-determined emergency policies and communication strategies as critical first steps on the road to normalization.

Strategy is one thing; practical application of that strategy is another. Warsh is very clear in sending the operational message that elevated inflation should be the Fed’s “predominant focus right now.” This is consistent with his third key principle of monetary policy — the least ambiguous of the seven principles he elucidated in his Jackson Hole Speech — that “… there should be no misunderstanding … it is the Fed’s job to deliver stable prices.” By going on to stress that recent trends in underlying inflation have not meaningfully improved, there can be no mistaking the policy bias of monetary tightening.

There was far more in the Warsh speech than the points I have stressed above. Market commentary focused on his hawkish assessment of inflation, and there was an upfront discussion of AI-related complications to the interplay between economic growth, inflation, and interest rates. But, in my view, his effort to clarify and simplify the approach to monetary policy in a non-emergency setting deserves special emphasis. At the same time, it is equally responsible for the Fed to admit that it no longer feels compelled to make future promises to financial markets as to what and when comes next.

To be honest, the biggest concern I had with Kevin Warsh as Fed Chair had to do with the politics of his appointment. President Trump has long been vocal in expressing his views on the need for lower US interest rates. He demands blind obedience from those who serve at his pleasure. He attacked Jerome Powell unmercifully and repeatedly for his failures to deliver the interest rates that he wanted. I was concerned that Warsh had cut some type of political deal with the President in order to secure his appointment. His reassuring message at Jackson Hole not only dispels that concern but it sets out a perfectly reasonable path for a long overdue normalization of Fed policy. Kevin Warsh deserves great credit for establishing his mark as a serious and independent policymaker. He was very much on point in doing so.

Intervening in financial markets is a fool’s game. And yet many seem incapable of internalizing that lesson. So it is with US Treasury Secretary Scott Bessent, as he seeks to manipulate currency and bond markets.

Bessent has the world’s most consequential markets in his crosshairs. In targeting the foreign-exchange market, he is going after the largest and most liquid financial market in the world; according to the Bank for International Settlements’ latest triennial survey, over-the-counter foreign-exchange turnover reached $9.6 trillion per day in April 2025. Daily trading volume in Bessent’s other target, the $31.5 trillion US Treasury market (the world’s largest individual security market), currently averages about $1.2 trillion.

Bessent somehow believes that he knows better than markets. He thinks that the Japanese yen should be stronger (and, by inference, the dollar weaker) and that long-term US interest rates should be lower. Accordingly, in partnership with Japan’s finance ministry, Bessent intervened to support a weakening yen on July 31. While the scale of the yen purchases was not disclosed, a photograph of Bessent’s to-do list suggests that the US spent $5–10 billion, with Japan reportedly chipping in another $53 billion. A few weeks later, Bessent announced an increased buyback of long-dated US Treasuries that potentially works out to about $32 billion per quarter.

These are puny amounts relative to the trading volume in these vast markets. The combined yen intervention represents about 0.6% of daily foreign-exchange turnover (and less than 4% of daily net yen trading, while the support measures for long-dated bonds is equivalent to only about 0.1% of the market per quarter. Bessent has hinted that there could be more to come but without meaningful policy changes by the Fed or the Bank of Japan, the impact of tiny interventions in huge markets will likely be negligible.

There is a myth that, in the 1980s, coordinated currency intervention was decisive in addressing extreme swings in the value of the dollar, which was too strong in the mid-1980s and too weak a few years later. But the Plaza Accord, the agreement reached by the then-G5 in September 1985 to weaken the dollar, came fully seven months after the greenback peaked. The next joint intervention, the Louvre Accord, followed in February 1987, aimed at stabilizing international currency markets after the dollar’s subsequent plunge in 1985–86. But it came about a year and half before the dollar started bottoming out in 1989–90.

In other words, joint currency interventions by the major advanced nations did not match up well with major shifts in foreign-exchange markets. While the Plaza Accord piled onto a trend that was already underway, the Louvre Accord struggled to have any impact. Significantly, these actions did not break the back of currency speculators, as George Soros famously did when he bet against the British pound in 1992.

Bessent, who worked for Soros at the time of his famous attack on sterling and was part of the team that planned it, may be guilty of conflating today’s circumstances with those more than 30 years ago. Fond memories, however, often mislead today’s compasses.

As for Bessent’s intervention in the bond market, there is a slight twist. The Treasury has not only pledged to increase buybacks of long-dated bonds, but also has indicated a willingness to issue more short-term securities. This is not a formal “Operation Twist” like those orchestrated by the US Federal Reserve in 1961 and 2011–12 as part of a conscious effort to flatten or tilt down the yield curve. Again, the best that can be said of the current operation is that it may be an attempt to signal a change in debt management that might lead to a reduction in longer-term financing costs.

The early returns in the currency and bond markets offer little encouragement that Bessent’s interventions have made much of a difference. Any momentary blips immediately following the measures have subsequently been reversed. The yen remains weak, and rates on long-term bonds are basically back to where they were before the buyback announcement. Meanwhile, the dollar has weakened, and gold prices—possibly a new safe haven—have risen sharply.

If Bessent is following former European Central Bank President Mario Draghi’s 2012 motto and doing “whatever it takes” (originally referring to the preservation of the euro), he clearly has a lot more “whatever” to do.

What strikes me about Bessent’s role in Trump’s administration of sycophants is how dramatically he has changed from the man I first met in the mid-1990s. For a few years, Bessent was a regular attendee at Morgan Stanley’s fabled Lyford Cay investment conference. He moved on, and so did I. We reconnected at Yale, where in 2011 he invited me to speak at an undergraduate seminar on financial crises that he was teaching at one of the residential colleges.

Scott Bessent’s character has been transformed in the political arena. A year ago, I questioned his willingness to support many of Trump’s most outrageous economic positions, on everything from tariffs and growth to the integrity of the US Bureau of Labor Statistics and the veracity of the Yale Budget Lab.

Trump’s mendacity reminds me of Zhao Gao, a famous eunuch of the early Qin Dynasty (221–206 BC). According to legend, Zhao was a ruthless master of factual distortion. He even convinced the emperor at the time, Qin Er Shi, and his court that a deer was a horse. The 2026 version of Bessent seems more than willing to embrace the deer as a horse. Markets don’t like fables.

Bessent doesn’t think these markets are providing a ”fair” signal of their underlying fundamentals — that long-term US interest rates should be lower, the yen should be stronger, and by inference, the dollar weaker. Accordingly, in partnership with Japan’s Ministry of Finance, he committed the US Treasury to between $5 and $10 billion in support of a long weakening Japanese yen, while Japan has chipped with small change of another $53 billion. A few weeks later, the treasury secretary announced an increased buyback of long-dated US Treasury securities that works out to about $32 billion per quarter.

These are puny numbers relative to the vast size of both markets. The foreign exchange market is the largest and most liquid financial market in the world. In 2025, global FX turnover averaged about $9.5 trillion per day. By contrast, in the $31.5 trillion US Treasury market, the largest individual security market in the world, daily trading volume currently averages about $1.2 trillion, or 13% of that in FX markets.

The yen support move of the US and Japan, combined, works out to about to about 0.6% of total daily FX turnover (or nearly 4% of daily yen trading) while the Treasury support action is equivalent to only about 0.1% of the market per quarter. Bessent has suggested that the Treasury buyback initiative could be expanded and there is a related possibility of additional joint yen intervention as well. Either way, these actions are tiny, at best, a signaling mechanism of more market intervention to come.

There is a myth that coordinated currency intervention was decisive in the 1980s in addressing the extremes of US dollar valuations — too strong in the mid-1980s and too weak a few years later. It turns out that that the Plaza Accord, aimed at weakening an overvalued US dollar, came in September 1985, fully seven months after the dollar peaked in February of that year. A different result followed the joint intervention of the Louvre Accord in February 1987 that was aimed at stabilizing international currency markets after the dollar’s subsequent plunge of 1985-86; this intervention came about a year and half before the dollar started bottoming in 1989-90. (BIS data on the dollar’s real effective exchange rate against a basket of 27 major economies.)

In other words, the joint currency intervention of major industrial powers does not match up well with major shifts in global FX markets. The Plaza Accord piled on to a trend that was already unfolding and the Louvre Accord struggled to have any impact. Significantly, these joint actions did not break the back of currency speculators, as George Soros famously did in betting against the British pound in 1992. Bessent, working for Soros at the time, and part of the team planning that famous attack on sterling, may be guilty of conflating today’s circumstances with those he was involved with nearly 34 years ago. Fond memories do not make for great forecasts all these years later. I can certainly attest to that myself!

The early returns in the currency and bond markets offer little encouragement that the latest interventions have made much of a difference. Any momentary blips at the time of the moves have subsequently been reversed. The yen continues to weaken and rates on long Treasures are back to where they were before the buyback announcement. Meanwhile, the dollar has weakened and gold prices — possibly, a new safe haven — have moved sharply higher. If Mario Draghi’s 2012 motto of doing “whatever it takes” to support a then beleaguered euro holds true this time, Bessent clearly has a lot more of the whatever to do. A lot more.

I first met Scott Bessent in the mid-1990s. For a few years, he was a regular attendee at Morgan Stanley’s fabled Lyford Cay investment conference. He moved on and so did I. We reconnected at Yale, and in 2011 he invited me to speak at a seminar on financial crises that he was teaching at one of the residential colleges.

He went on to bigger and better things, and I am still at Yale. No complaints but just a few observations: Back then, I was struck by Scott’s modest demeaner and humility. I don’t see that today in his current high-profile position as chief economist of the Trump Administration. A year ago, I raised questions about his new role as sycophant that bordered on mendacity in supporting many of President Trump’s most outrageous positions on the economy — tariffs, jobs, growth, and the integrity of the Yale Budget Lab and the US Bureau of Labor Statistics.

Bessent, the politician, took me back to the ancient Chinese tales of Zhao Gao, the most famous eunuch of the early Qin Dynasty (200 BC). According to legend, Zhao was a ruthless master of factual distortion in providing advice to the emperor of the time, Qin Er Shi. He even convinced the emperor and his court that a deer was a horse. In his attempts to manipulate markets, Treasury Secretary Bessent is most assuredly not a horse.

With the world media overflowing in well-deserved tributes, let me add a few first-hand experiences to try and bring the legend to life. My most memorable encounters with Zhu Rogji center on the China Development Forum — an event he established in 2000. It was a brilliant concept. Zhu believed that the ministers of China’s State Council — the government he headed as Premier — should have to defend and debate the policies that had just been enacted at the National People’s Congress. He established the CDF to convene what was initially a small group of outside experts from academia, think tanks, and the business community to engage State leaders a couple of weeks after the conclusion of China’s Two Sessions.

Engagement is exactly what he got — the vigorous give and take of debate between inside (Chinese) and outside (foreign) experts, the latter group of which I was proud to be a part of. I was invited to the CDF from the start — conflicted from the first meeting but beginning in 2001, I attended the event for 25 consecutive years. Before being banned in 2026 by a CDF that now discourages the debate originally envisioned by Zhu Rongji — another story — I was the longest attending foreign delegate to what was China’s first and, eventually, its most prominent international conference.

CDF 2001 was especially memorable. First, I was astonished with the vigorous participation of the home team. In one of my panel sessions, Professor Wu Jinglian, a well-known pro-reform economist, and something as an intellectual mentor to Zhu Rongji, was sitting prominently in the front row; Wu, who I had met earlier, was brimming over with enthusiastic reactions and questions at the mere mention of market-based corporate restructuring . I was relatively new on the China circuit at the time and contrary to my expectations of a tightly controlled discourse, the early years of CDF were delivering precisely the spirited engagement that Zhu Rogji was seeking. Sadly, those days are now over.

Second, I was offered a spot as the keynote lunch speaker for CDF 2001. I warned of a likely downturn in the world economy, triggered by the bursting of the US dotcom bubble a year earlier and the collateral damage it had taken on other segments of the equity market. The Chinese jumped in with vigorous questions, but what I remember the most was being challenged by a fellow American, C. Fred Bergsten founder and President of what was then called the Institute for International Economics, a Washington DC think tank. Fred, who I knew well at the time, took the other side of the outlook and was far more optimistic than me.

All this is preamble to the main event of the CDF — a closing session at the Great Hall of the People with the host, Premier Zhu Rongji. It started with Sir John Bond, Chairman of HSBC and co-chair of CDF 2001, reading a lengthy report of our three days of meetings. Being jet-lagged, I started to nod off in my soft chair. Zhu quickly cut Bond off and said something to the effect of, “Send me the report, I will read it later. I heard there was an interesting debate about the economy between Roach and Bergsten. I would like to hear about that.”

I jumped and was simultaneously startled by a sharp elbow in my side, urging me to stand up and offer a summary of my view, which somehow, I did. Bergsten did the same. The Premier had a few questions and then adjourned the meeting. On the way out, he came over to me and said in perfect English, “Roach, I hope you are wrong, but we will plan as if you are right.”

A year later at my second CDF, Premier Zhu reviewed the state of the global economy, where growth had slowed in 2001 to about 2.5%, right on the threshold of the global recession I had warned of. He turned to me and simply said, “thank you.” China had, indeed, adopted pro-active fiscal measures to shield its economy from the tough global climate I had predicted; it ended up growing a little more than 8% in 2001. Zhu added, “You are a good friend of China.” He repeated those comments at a smaller meeting at Zhongnanhai with senior Morgan Stanley executives later that year, and with a characteristic sense of humor added, “That’s why we didn’t ask Bergsten back.” He also invited me to speak at his alma mater, Tsinghua University.

I have told parts of this story several times over the years. That single comment from Zhu Rongji about being a good friend of China subsequently opened many a door for me in official Chinese policy circles. That included building strong relationships with several of Zhu’s protégés who went on to hold senior policy positions in China — like central banker Zhou Xiaochuan, Finance Minister Lou Jiwei, Vice Premier Wang Qishan, and senior financial regulator Guo Shuqing. But the endorsement of the Premier was more than just a door-opener — it was a sign of trust from the top that added considerable credibility to my reputation inside of China, especially regarding analysis of the Chinese and global economy.

There were several other occasions when I had the opportunity to see Zhu Rongji in action. An especially notable encounter was an April 1999 speech he gave to the Economic Club of New York. After a disappointing meeting with President Clinton in Washington, Zhu mixed both English and Mandarin together in making a compelling case for tying Chinese reforms to WTO accession. It was an amazing speech. The air was electric with enthusiasm — you honestly did not need a translator to understand his message. That speech, along with pressure from the US business community, turned out to be pivotal in getting Clinton to change his mind on China’s entry into the rules-based global trading system.

A more personal encounter came at a conference in Beijing in late 2000. A large group of attendees was standing around the perimeter of a reception room, and Zhu made the rounds personally greeting all of us. When he came to me, the Premier glanced at my name badge and called me “Stanley Morgan.” He smiled when I corrected him and then we chatted a bit about his memorable 1999 speech in New York. There was an unfortunate (unrelated) security incident shortly after that, but the conference was quick to resume, and an unruffled premier delivered another strong message.

Several years later in late 2006, I followed up on Zhu’s earlier invitation and gave a lecture at Tsinghua University’s School of Economics and Management (SEM). It was the first of many lectures I have since given at leading Chinese universities. With a demanding day job, Zhu had stepped down a few years earlier from his position as Founding Dean of SEM but still chaired the Advisory Board. An overflow audience of extraordinarily bright, well-informed students peppered me with tough questions for at least an hour after my formal remarks. Zhu Rongji was not in the room that night, but his spirit of engagement and the talent he attracted was on full display.

No leader is perfect, and that includes Zhu Rongji. Like others before him, as well as his senior leadership partner, Jiang Zemin, he endorsed hyper growth as the antidote to much that was still ailing China at the time. It fell to his successor, Wen Jiabao, to raise questions about a growth recipe that was increasingly “unstable, unbalanced, uncoordinated, and unsustainable.” Those questions, of course, remain unanswered today. Zhu is often cited as expressing regrets that he had only one five-year term to serve as premier. Count me as one who shares that regret. Zhu Rongji’s strength was at its greatest in answering China’s toughest questions.

Yet the Chinese leadership continues to march on, fixated on a growth model that draws unsustainable support from innovation, new technologies, and other trappings of what they now call “new quality productive forces.” Xi Jinping is the champion of this approach, paying lip service to Chinese consumption but unwilling to take the big steps required of consumer-led rebalancing. He warns , for example, that aggressive social safety net reforms — health care and retirement — will backfire; the excesses of fear-driven precautionary saving might decline and boost discretionary consumer demand, goes the argument, but at the cost of raising the corrosive risks of a western-style welfare state. Egad, who wants that?

Let me try a different angle that proved very accurate in foretelling the Japanese disease. Shown in the chart below are capital-output ratios for Japan and China. When the ratio rises, the economy experiences what economists call “capital deepening” — meaning that it takes ever larger increments of its capital stock to generate a unit of output (GDP). This, of course, is associated with progressively lower returns on the capital stock that presumably would set a self-corrective process in motion that would lead to a reduction of capital-intensive economic growth.

That was not the case for Japan in its first lost decade, when the capital-output ratio (shown in the green line) increased sharply by 24% from 1990 to 2000. As the stock market surged, Corporate Japan became convinced that capital deepening was the recipe for higher returns to shareholders. The bet ignored the bursting of the Japanese equity bubble in the early 1990s. Like all stock market implosions, denial was powerful in the early post-bubble period and investment-led capital deepening continued. For Fumio Hayashi and Edward Prescott, two of the earliest scholars of the Japanese disease, this was a powerful signal of the coming stagnation in the Japanese real economy; with one lost decade eventually turning into three, that forecast was very much on the mark.

By comparison, a more recent strain of Chinese capital deepening is very disturbing. From 2008 to 2023 (last data point available in the Penn World Tables), China’s capital-output ratio soared by 62% — over two and a half times the 24% increase in Japan during the 1990s. Moreover, by the end of this fifteen-year period, the Chinese ratio had basically converged on its Japanese counterpart. Unlike Japan where capital deepening continued in the face of post-bubble carnage in its equity market, China’s capital deepening took place in a more volatile equity market climate — three sharp run-ups in 2014-15, 2019-21, and again in 2024-26. Chinese capital deepening was driven less by the attraction of equity returns and more by the strong guidance of its leader, Xi Jinping, and his steadfast fixation on new quality productive forces.

(As an aside, I would venture the guess that China’s capital deepening has intensified in recent years beyond the period covered by the chart above (which ends in 2023). That’s because the AI-related data center construction frenzy is raising the capital stock at a time when Chinese GDP growth is slipping further, falling recently below its 4.5% to 5% target for 2026.)

There is a great read by Evan Osnos in the August 3 issue of The New Yorker that speaks to China’s glorious technology-led future if it stays its recent course. Entitled “The Future, Made in China,” his latest piece is loaded with anecdotal tales of wonder about China’s robots, “dark” (i.e., completely automated) factories, high-speed EVs, and everything you want to know about AI. Osnos is a wonderful writer in The New Yorker tradition, but I was left wondering if he hasn’t taken some of the CCP’s bait too far — hook, line, and sinker.

The subtext of this article is America’s looming competitive battle with a rising superpower, a topic I have also addressed repeatedly in my own work. Like Osnos, I worry about this a good deal, but less so if China stays its course and more if the US falls into a MAGA sinkhole dominated by underinvestment in basic research and higher education, along with an arrogant disdain for expert advice. In another sense, I have to confess that the Osmos piece has a familiar ring in similar tales we heard about the Japanese techno miracle in the late 1980s — books and articles by Sony founder Akio Morita, Ezra Vogel, and Clyde Prestowitz. It’s not that they under-estimated America — they all missed what was to befall Japan.

Michael Pettis and I are kindred spirits in this debate. We have both focused for years on the imperatives of Chinese rebalancing. Our frameworks are a little different, but our bottom-line conclusions are very much the same. We agree on Japan where the problem was not so much its technological successes but the sustainability of its growth model. We both share the view that the same lesson might be very much be applicable to China. My point on capital deepening is offered in that same spirit — the sustainability of the Chinese growth model, which is now showing unmistakble signs of sputtering.

Recently, I wrote a Project Syndicate op-ed, “Yes, the Hong Kong of Old is Over,” and also posted it on LinkedIn. It was a follow-up to a controversial piece I had published about two and a half years ago in the Financial Times (“It pains me to say that Hong Kong is over”). In the past few days, my LinkedIn pages have overflowed with largely one-sided critical responses. Several were laced with distortions and ad hominem attacks on me and my distant (presumably disqualifying) residency in Connecticut. But they were the exceptions rather than the rule.

The bulk of the feedback has been exactly what I am looking for — strongly worded, focused criticism that picked apart my arguments and made me think about the main assumptions of my thesis. The strength of the responses was indicative of the depth of conviction that most of these (presumably) Hong Kongers feel in defense of their city. I respect and understand that. As a former resident and still frequent visitor, I, too, have openly confessed my love for the Hong Kong of old — on the pages of the South China Morning Post, as well as in a 2024 speech before Hong Kong’s distinguished Foreign Correspondent’s Club. But I digress.

Several readers took the time to write lengthy, thoughtful articles challenging my views on Hong Kong. Three of them were especially impressive — one by Viginia Lee ( lawyer/ solicitor from the Greater Bay area and a China-appointed attesting officer), another by Adam Clermont (an American-born lawyer living and practicing law in Hong Kong), and a third by Michael Wong (Deputy Financial Secretary of Hong Kong). Ms. Lee’s article was published as an opinion piece by China Daily; Mr. Clermont’s is available on LinkedIn; and Mr. Wong’s op-ed was published in traditional Chinese by Sing Tao.

The thrust of these articles took issue with my core point on Hong Kong’s transformation from old to new. Wong wants us still to view Hong Kong through the “One Country, Two Systems” lens. Lee and Clermont speak eloquently and more directly to my key point — the changes that occur in any city; they argue that, in this respect, Hong Kong is no different than other dynamic financial centers such as New York, London, or Singapore. Their key premise is that Hong Kong is just as much a part of China as New York is of the US, or London is of Britain.

Fair point — but not the one I am making. Conveniently overlooked is that the timeline of a 50-year handover to China has basically been cut in half by Beijing’s aggressive takeover in 2020 following massive public demonstrations. Ergo, the transformation of Hong Kong has occurred in the middle of the period when the city was still largely guaranteed a “high degree of autonomy” under its Basic Law. That autonomy gave it the freedom, at least on paper, to operate independently outside the legal and political structures of the mother country, the PRC. As a Pew Research survey conducted in mid-2023 shows, Hong Kongers still overwhelmingly supported those democratic rights after Beijing stepped in.

That is not a trivial detail. Beijing’s actions, as condoned by a pro-Beijing Hong Kong government, resulted in a forceful takeover of the city rather than a spontaneous, organic evolution of a dynamic urban center. Would the large outflow of expats have occurred had there not been such a forced development? Would six foreign judges have not resigned? Would bookstores not have been closed? Newspapers not shut down? More so-called dissidents not rounded up?

It’s one thing to love your city for all its history and traditions. It’s another thing to conveniently look the other way and ignore a much bigger force that connects the dots. The questions I raise are not about normal occurrences, nor spontaneous coincidences, for any city. They all happened after Beijing’s takeover. Right before your eyes. Resilience is Hong Kong’s timeworn battle cry. To me, it looks more like forced capitulation.

The other points in my thesis — especially IPO leadership and economic linkages to China — speak for themselves and, despite the howls of protests from some social media commentators, are not worth re-litigating again in print. But there is one overarching aspect of the debate that I want to conclude with — the character of the city. As the questions above imply, the character of Hong Kong is now under assault.

Yes, the city looks the same, and to many, especially those who focus on fine restaurants and the sleek, efficient MTR, it feels the same. But it’s not. Steeped in denial, the questions I raise anger Hong Kongers, and their fear of Beijing-sponsored retribution silences the free and open debate that was once a core attribute of the character of Hong Kong.

You get what you ask for on social media. There is no way to control the quality of the feedback, which ranges from highly constructive to vitriolic. Over a year and a half ago, I pulled the plug on Twitter/ X when the balance shifted toward anonymous (often bot-driven) guttural soundbites. I struggled with Bluesky and Threads but then settled on LinkedIn as a higher-quality platform — more limited distribution but connectivity with peer-level counterparts who generally delivered the type of feedback I was looking for.

That approach worked this time. Many thanks to those of you who responded forcefully and in most cases with dignity and clarity. You may not like what you read, and I may not agree with your feedback. But in the end, respectful dialogue is what a thriving marketplace of ideas is all about. I will continue my best to try and uphold that lofty standard.

He can’t help himself in playing the tariff card and, as a duly certified economist, neither can I. As long as the President continues to act out his McKinleyesque infatuation with what he has dubbed the “most beautiful word” in the dictionary, you can expect more of the same from me.

Why bother? Three reasons: Firstly, the economics I practice continues to judge this as a policy blunder of historic proportions. Secondly, Trump’s political MO is repetition; he tries to beat us down with a steady drumbeat of misguided policies on a multitude of fronts — from tariffs and alliances to Iran and the “rigged” 2020 presidential election. I am just as stubborn as him. Lastly, the title of this Substack is “Conflict.” It still focuses mainly on the on the US-China conflict, but thanks largely to President Trump there are plenty of other opportunities to broaden the scope.

Most of the tariff debate has focused on the first point, impact — or lack thereof. The Golden Age spiel of the President claims the impact is zero. By his reckoning, a low unemployment rate and sustained GDP growth, to say nothing of record stock market froth, only add to the gilt that has now transformed the Oval Office. Never mind that inflation and long-term interest rates have moved higher as most conventional tariff modeling would suggest. In Trump’s view, those are nothing more than the blips of a temporary “excursion.” As Maggie Haberman and Jonathan Swan document in Regime Change, it infuriates him to hear anything else.

Academics have written numerous serious papers on the impacts — both theoretical and empirical — of Trump’s tariffs. Believe it or not, an LLM search of peer-reviewed journal articles and working papers (NBER and CEPR) puts the number of these papers since January 2025 also at 18. (For what it’s worth, in Chinese culture, the number 18 symbolizes great wealth and prosperity — or, alternatively, the 18 levels of hell! Take your pick on the hidden message from Trump’s tariff fixation.)

The research shreds Trump’s view that tariffs are paid by other nations. It generally concludes that about 95% of the costs of Trump’s tariffs are borne by US consumers and businesses, with the bulk of that falling on consumers. For a CPI inflation rate running 3.5% in the year ending this June, the annualized tariff impact is about 0.7 percentage point, costing the average American family about $1,100 per year, according to the latest (July 24) estimates of the Yale Budget Lab.

These are meaningful but not devastating impacts. They are significant enough to influence monetary policy and boost the inflation premium embedded in the term structure of longer-term interest rates. Moreover, with tariffs having been changed more than 50 times under Trump, the uncertainty factor weighs heavily on the decision making of businesses and consumers, alike. And, of course, they have an impact on affordability, the big issue in the upcoming midterm elections.

So far, the high-flying AI-driven equity market has been immune to any resulting valuation pressures arising from the recent run-up in bond yields. Moreover, the lack of serious retaliation from most of America’s trading partners has been important in tempering Smoot-Hawley-like comparisons with the global trade war of the 1930s. On that point, I plead guilty … at least, for the time being. There are, however, still good reasons to worry about the increasingly global risks of protectionism; while my recent concerns have focused more on China and the massive export-led imbalances of its producer economy, the latest round of Trump tariffs is also ground for concern.

That brings me to the second reason for writing yet another piece on tariffs. While July 23 was not another April 2, 2025, Liberation Day fiasco, it is a clear sign that the Trump Administration is far from giving up on its signature international economic policy gambit. True to its promise after losing the emergency IEEPA tariff battle in the US Supreme Court five months ago, the Trump team has opted for a Section 301 back-up; the US Trade Representative investigated over 60 economies for unfair trading trade practices related to their alleged failure “to impose and effectively enforce a prohibition on the importation of goods produced with forced labor.” Finding all 60 countries guilty as charged — a rather remarkable coincidence — the USTR has concluded that a new tariff of 10% to 12.5% is appropriate for the full sample of investigated economies.

By Yale Budget Lab calculations, this will bring America’s average statutory tariff up to 11.1% immediately and to 11.8% by the end of this year. This is close to six times the average tariffs rates relative to those prevailing at the end of the Biden Administration in late 2024 and, yes, the highest since the 20% effective tariff rates in the Smoot Hawley days of the early 1930s. The latest announcements do, however, fall well short of the initial Liberation Day threat of 22.5%.

We learned two important lessons from Liberation Day: One, Trump’s trade policy pronouncements are hardly set in stone. As the markets swooned in early April 2025, his TACO instincts took over and he chickened out; the combination of a tariff pause and partial rollback limited overall tariff rates to about 10%, less than half the original “bluster rate” of Liberation Day. The actions of last week, which replace the temporary (and now expired) Section 122 tariffs that were put in place after the SCOTUS rejection of IEEPA tariffs, ups the ante only slightly from overall tariff rates prevailing at year-end 2025. Two, his latest Section 301 actions are more serious than the ludicrous emergency attempts under IEEPA. While there will undoubtedly be court cases testing the slippery legality of the forced-labor violation that all 60 investigated countries are being charged with, the US judicial system has normally granted presidents broad authority on Section 301 actions.

Finally, my Substack speaks for itself. We live in a world in conflict — whether we want to admit it or not. It is not just the United States and China. The Uppsala Conflict Data Program indicates that in 2025 there were 65 active state-based kinetic conflicts in the world, the highest reading since 1946 (see chart above). That is not exactly a heart-warming comparison. There is more than enough food for thought to keep me going for a long time.

Hong Kong deserves credit for putting up a fight. In the nearly two and a half years since I expressed the apparently controversial opinion that Hong Kong is “over,” the city’s boosters have orchestrated a determined campaign to prove otherwise.

At first glance, this heroic defense seems to have some merit. Hong Kong’s economy has been relatively resilient, and a resurgence of the Hang Seng index has erased memories of the 2018–20 collapse stemming from political turmoil. Moreover, the city has reclaimed its position as the global IPO leader, giving it bragging rights as Asia’s premier financial center. Paul Chan, Hong Kong’s financial secretary, and other leading government officials consistently sing its praises.

Beneath the surface, however, the story is very different. Beijing’s hostile takeover following the pro-democracy demonstrations of 2019–20 has transformed Hong Kong into just another big Chinese city. Today, its Mandarin name, Xiānggǎng (香港, “fragrant harbor”), which celebrates the city’s pre-colonial Chinese heritage, seems more apt than Hong Kong, a phonetic translation from Cantonese, the city’s long dominant dialect.

This dramatic transformation into Xiānggǎng is the real story. Hong Kong’s defenders are in denial about this new chameleon-like identity, instead viewing it as another example the city’s inherent resilience. Nice try.

IPO leadership, long the most precious jewel in the city’s financial-services crown, is a glaring case in point. The city’s return to the top spot in 2025 was driven by listings of mainland Chinese companies such as CATL, Luxshare Precision, Z.ai, Momenta, and Montage Technology. By one estimate, they account for as much as 90–95% of recent funds raised. Hong Kong is less a thriving global IPO market than a major platform for Chinese issuers.

The rule of law, long regarded as one of Hong Kong’s greatest institutional advantages, has been severely compromised. In 2020, China bypassed the Hong Kong legislature and imposed a sweeping national security law. Four years later, Hong Kong’s lawmakers rushed through Article 23, an expansion of the 2020 security law that integrated China’s national security framework into the city’s legal system. These changes have stifled open debate, resulted in new arrests and harsh sentences for previously imprisoned activists, led to closures of independent bookstores, and destroyed any semblance of a free press.

Moreover, six foreign judges have resigned from Hong Kong’s Court of Final Appeal, calling the independence of the city’s highest judicial authority into serious question. Among the departures was Jonathan Sumption, one of England’s most brilliant jurists, who warned in the Financial Times that a Mainland strain of “judicial ‘patriotism’” was poisoning the city’s legal system.

Lastly, there are the people. While Hong Kongers remain understandably proud of their heritage, and the city’s overall population has held steady at around 7.5 million since 2020, there has been a dramatic shift in the composition of the workforce. The Hong Kong government does not publish official emigration statistics by nationality, but research points to a significant outflow of expats, including many mid- and senior-level professionals, since 2020, followed by a surge of Chinese workers from the mainland. Walking the streets of Hong Kong today, one is just as likely to hear Mandarin as Cantonese.

Taken together, this represents a fundamental transformation in Hong Kong’s character. The city’s stunning outward appearance remains. But that hardly justifies the tales of competitive innovation that the pro-China camp spins. Conveniently left out of this narrative is that this resilience has been largely made in China, not in Hong Kong. China’s stock-market stimulus of September 2024, for example, propelled equity markets upward, unleashing a torrent of Chinese IPOs that were consciously steered to Hong Kong’s sophisticated equity-distribution platform. And now the PRC’s “national team” is attempting to do it again.

It is obvious that Hong Kong’s economic performance is closely correlated with China’s. Thus, sluggish Chinese GDP growth, which slid below target to 4.3% in the second quarter of 2026, will undoubtedly weigh down Hong Kong’s growth.

The Hong Kong government seems to think that Chinese-style central planning offers new hope for its economy and recently launched its first five-year plan. But, as China is painfully learning, such plans can often over-promise and under-deliver. This suggests that until there is a meaningful test of the downside—for the economy or the equity market—any conclusions drawn about Hong Kong’s newfound resilience are at best premature.

To some extent, I have become a lightning rod in the “Hong Kong is over” debate. Hong Kong and Chinese officials are quick to pounce on my comments. Local journalists often ask if I am ready to admit that I was wrong (particularly after I conceded that a managed stock market has risen more than I had expected). Just the other day, a leading reporter messaged me: “‘Hong Kong is over’ seems over. What would you say about it?” Well, let me be clear: the Hong Kong of old is, indeed, over. Go to Xiānggǎng and see for yourself.