I know the real reasons behind fears of a dotcom-style AI bubble

That increase in value of global shares has been an eyewatering 31% in just six months. In modern stock market history this has only happened three times before. In 1987, in 1999, and in 2009. These years are seared into investor’s brains as year’s where market bubbles burst - or in the case of 2009, the world economy exited from life support in the aftermath of the Global Financial Crisis. This has led to a series of high-profile investors, bank executives, and regulators sounding the alarm. How credible are these warnings? And how impactful could sharp decline in stock market indices be for the world economy?

The first thing to note is that warnings from those active in financial markets come with several motivations, and health warnings. It is always worth considering, at the outset, the motives of those urging caution. For regulators like the US Federal Reserve and Bank of England - tasked with maintaining financial stability in the economy - warnings are part and parcel of the role. Sounding the alarm is an exercise in trying to ensure that unchecked enthusiasm doesn’t impair the allocation of capital and create undue investor risk. It is true to say that uttering warnings in dusty reports and press conferences are the easy bits - pre-emptive actions to try and deflate any bubble are far harder, and controversial. But it is also true that impactful communications are part of any regulator’s toolkit. Last week the International Monetary Fund (IMF) added its voice to these warnings, drawing uncomfortable parallels to the dot.com bubble, and subsequent bust of 1999 and 2000. 

The motivations for warnings from investment bank executives and economists are subtly different. Cynically, I would suggest a large number are motivated by having their warnings out there in public just in case the market does crash, and they can then flag a previous podcast, Substack, or LinkedIn post to reinforce their sage-like credentials. I am mindful that this very column could be interpreted as my version of that exercise. But it is also valuable to hear these warnings from market participants who can identify pattern repetition from lengthy careers of observing markets and investor behaviour. There is a rich history of research identifying that financial market crashes are more likely when there is a waning in the number of participants who have lived through previous crashes. The number of active investors were active through the dot.com bubble shrinks with each passing year.

The challenge facing investors trying to draw their own conclusions are there are arguments on both sides. The fears of a bubble have strong empirical roots. The overall valuation of the pre-eminent US stock market - where AI hyperscalers have driven the addition of market value - is now more than two standard deviations above its long-term average. The last time valuations were that rich was in 1999. A widely-followed measure of how expensive the US stock market is – the Shiller Price/Earnings ratio - is hovering around forty times earnings. February 1999 to September 2000 are the only other periods in market history that this measure has been above forty. To add fuel to the fire, non-profitable technology companies have seen their valuations double since April. These companies represent the closest parallel to the dot.com era, so this trend represents the loudest echo from history.

But there are, as always, credible counterarguments from investors and advisers that remain bullish. In particular, the companies that are fuelling enormous capital investment in AI are already hugely cash generative from a range of their cloud, commerce, and chip businesses. This existing flow of earnings to companies like Amazon, Microsoft, Google, Meta and Apple underpins the attractions of owning these companies even if investors struggle to calculate the long-term payback of their huge investments in AI.

There is also the financing argument for further stock market gains, in particular that the Federal Reserve looks poised to continue cutting US interest rates. Investors are expecting two more interest rate cuts by the end of the year. Traditionally a path to lower US borrowing costs has been supportive for stock market valuations. Finally, whilst a relatively small number of huge US technology companies look expensive, markets across the world - including in the UK - are far less expensive even after adjusting for lower earnings potential. These provide something of a safe haven if the hopes and dreams of AI fall short.   

For my part, what gets widely underappreciated is how share ownership - and direct wealth exposure - has ballooned in recent years. US savers now have 30% of their wealth in shares. This equates to almost $42 trillion. The ratio of almost a third of US wealth held in stocks has, also, recently exceeded its 1999-2000 high. The fastest growth of share ownership since the COVID-19 pandemic has been amongst low income US households which is undoubtedly a good thing for spreading the proceeds of growth. But it comes with the accompanying risk that less affluent households spending behaviour is likely to be more highly sensitive to the impact of falling share prices. 

Another area of higher economic volatility may also stem from the changing face of pensions. In the UK, as well as in Japan and a number of EU countries, there has been a rapid transition away from defined benefit pensions - where the investment risk is held by employers and investment companies - towards defined contribution schemes where the saver/ retiree holds the investment risk. That risks amplifying the impact of any stock market correction through to household spending in a way that was simply not the case twenty-five years ago.

This kind of fundamental analysis may also miss the fact that an entirely unrelated event like a surge in geopolitical risk, or a lack of transparency in the private credit market, may yet trigger a reassessment of value in an increasingly passive investment market where an initial negative catalyst can feed off itself in a way that is hard to stop.

Overall, as I noted in these pages a couple of weeks ago, there are understandable hopes that AI-related productivity could be a silver bullet for government debt, and for investors fearing where the next stage of economic growth comes from. The problem is that if current shareholder hopes prove illusory the blowback to the real economy could be far larger than anything we saw after the dot.com bubble burst.