Crisis season, AI and the timing of Trump's attacks on the Fed: highlights from economists' blogs

Late summer and early fall in the U.S. (and not only) is often a time of financial turmoil - it was August-October that saw the most high-profile crises of the last decades. Today, new sources of risk are being added to this season of turbulence: a boom in investment in AI infrastructure and a growing confrontation between the White House and the Federal Reserve.  What makes August especially dangerous for investors, whether the economy is in danger of a "data center bubble" and whether Donald Trump has his own strategy in monetary policy, or whether his attacks on the Fed are just a political calculation?

August traditionally ushers in the season of financial turmoil in the U.S. - almost all major crises of the past 50 years have occurred between August and October, writes Owen Lamont, senior vice president and portfolio manager at Acadian Asset Management. Among such events over the past half-century, he cites:

Only one financial crisis-causing event has occurred outside this gap - March 2020, when the COVID-19 pandemic triggered a global quarantine.

In other words, Lamont continues, over the past 50 years, five major disasters have occurred in August-October, and only one in November-July. Based on these statistics alone, the probability of a new crisis during the year is about 12%, with about 10% in August-October and only 2% in the other months.

Such seasonality is not a new phenomenon. Back in the 19th and early 20th centuries, crises were disproportionately frequent in the fall. During this period, the harvest was underway, and it was necessary to redirect significant sums from the financial centers of the East Coast to the agricultural regions of the West. Such were the panic of August 1857, caused by the collapse of banks and falling grain prices, the panic of September 1873, provoked by speculative investments in railroad companies, and the panic of October 1907,  set off by a liquidity crisis in the stock market and massive withdrawals of bank deposits, Lamont lists.

The U.S. economy has long been less dependent on agriculture, yet the fall "seasonality" of crises has persisted. Lamont attributes this to the "agrarian legacy" - the tradition of summer vacations. In August and September, liquidity and trading volumes drop in North America and Europe as market participants go on vacation, which increases vulnerability to liquidity crises, he explains.

That doesn't mean, however, that investors should sell off assets every year in May, the financier notes. "The best thing to do is to structure portfolios to withstand unfavorable events regardless of when they happen," he concludes.

While users of generative neural networks discuss the release of the new OpenAI  model and its presentation with misleading graphs, economists continue to predict a possible collapse in the industry. The boom in data center construction increasingly resembles a bubble - similar to the railroad and dot-com bubbles, writes former Bloomberg columnist Noah Smith on his blog. 

In the 19th century, capital investment in railroads reached 6% of GDP. The boom in infrastructure investment eventually led to the crash of 1873. A similar story happened relatively recently with investments in telecommunications, which led to the dot-com boom. At the peak, investment in the industry was as high as 1.2% of GDP. In both cases, companies built too much infrastructure ahead of demand growth, which led to overestimated expectations, an inability to service debt and large-scale bankruptcies, writes Noah Smith.

Today's spending on data centers is about the same - about 1.2% of GDP. But that's just the initial phase, Smith notes. In the second quarter of 2025, the combined investment of Google, Meta, Microsoft and Amazon in the construction of data centers was almost $100 billion. And for Microsoft and Meta, these capital expenditures now account for more than a third of total sales.

Moreover, these investments have become a powerful driver of U.S. economic growth. AI capital spending has contributed more to U.S. economic growth over the past two quarters than the entire consumer sector, noted Renaissance Macro Research.

It is important that in the case of both railroads and telecommunications infrastructure, investors were generally not wrong in predicting the future importance of these industries - they have indeed become important drivers of economic growth. However, market participants expected a much faster return on their investments, and this was the trigger of the crisis.

The key question today is whether tech companies will have enough equity and access to capital to continue building at scale, and whether AI industry revenues will grow fast enough to justify that investment, Smith writes. 

What is also alarming is the fact that the growing volume of borrowed funds is concentrated in one single segment of the economy - data center construction. This means that the probability of default on such loans will be highly correlated. In other words, if the industry faces a drop in revenues, debt service problems could simultaneously affect many players, causing a chain reaction in the financial system.

Trump has arguably become the fiercest critic of the Fed of any U.S. president, writes Jeffrey Frankel, a professor of capital formation and growth at Harvard Kennedy School, on his blog. The last president to put pressure on the Fed was Richard Nixon, but even he didn't do so publicly, the economist notes.

Why is Trump attacking the Fed so openly? Does he have his own vision of monetary policy, different from the approach of current Chairman Jerome Powell, or is it primarily a political calculation?

To answer these questions, Frankel analyzed Trump's social media posts and his public statements about the Fed from January 2013 to June 2025. In total, he recorded 145 such statements: in 129 cases, Trump considered the rate too high and called for its reduction, and in 16 cases, on the contrary, he argued that the rate was too low and demanded a tightening of policy.

The study, according to the professor, shows that Trump's criticism is in no way explained by macroeconomic indicators. Whatever the inflation rate, unemployment rate, federal interest rate - the president still criticized the Fed and demanded to cut rates. 

The only significant variable is whether he was in office. It explains 76% of the variability in his critical statements. This means that  whenever Trump is in the White House, he calls for lower interest rates.

This article was AI-translated and verified by a human editor