AI, taxes and markets: three signals worth watching

A few different things have caught our attention this week. First, there’s the ongoing debate around the decision of the US Treasury to increase its purchases of long-dated government bonds. Second, there’s the announcement of a US$16.7 billion settlement between Meta and various US states on the question of social media harm. Finally, Bill Gates has released an essay highlighting the significant potential risks and opportunities from Artificial Intelligence – highlighting the need, in his view, for significant adjustments to tax regimes around the world. We wanted to explore how these various points might be related and what they could mean for portfolios.

We think there are a few points worth highlighting. First, the fiscal picture in Europe and the US is generally quite challenging. Debt to GDP is generally rising, notably in the US, and societies are ageing. As we discussed last week, that might be part of the reason why long-dated yields in the US have been drifting higher – prompting some intervention from the Treasury. The current consensus view is that intervening in the Treasury market isn’t a long-term solution for rising rates. Most investors would argue that having a lower fiscal deficit is the correct answer. But there’s currently little political will to achieve that, in the US and elsewhere.

Rising bond yields provide an important context for the ongoing debate about the future of work in the age of AI. Many believe that AI will usher in an age of sustainably higher unemployment – potentially reducing tax revenues and increasing costs. By way of background, in the UK income tax accounts for around 28% of total government tax revenue. National insurance (another tax on labour) accounts for a further 18%. Value-added-tax (VAT) – call it a tax on how people spend the money they earn – accounts for another 17%. That’s a pretty big percentage coming from labour, one way or another. If employment, and possibly household spending, is going to be structurally lower (and it’s still a big if), governments will need to find some alternative sources of revenue. Hiking taxes on households and employers even further probably won’t do the trick.

That brings us to the settlement between Meta and the US states. Whatever the merits of the case against Meta may be, these days, if you’re looking for money, you can find it in large US corporates, particularly in tech. Or you could before they started to spend all of it on data centres and chips. US corporate profits as a percentage of GDP are at a seventy year high, while wages and salaries as a percentage of GDP – starting at a much higher level than corporate profits – have steadily drifted lower since the 1970s.

What does all this mean for markets and portfolios? We’d make a few points. It highlights again the complexities of the current environment – as businesses, governments and workers try to make sense of the potential changes from AI. These questions won’t get resolved in a month or a quarter. These are long-term considerations.

If AI does damage employment, then that could have a significant impact on the tax base for many governments. In that sense, AI isn’t just about equities, private markets or even private credit. It can have an impact on government bond markets as well and that could increase the focus on the potential winners from the AI revolution. As we saw bank levies in the wake of the Global Financial Crisis, we could eventually see more targeted levies on the beneficiaries of AI efficiency – not just tech businesses. That’s not necessarily a bad outcome for investors – it would likely reflect a scenario where AI has helped drive stronger productivity growth. We think we’re in a period of great potential but also considerable uncertainty, and that argues for maintaining a well-diversified portfolio.