I have a belief that interest rates will be structurally higher for the entire world over the next decade, higher than the market currently expects. World population is declining in the countries that generate most current GDP, and are likely to in the future. China, in particular. Trade is fracturing, world trade isn't going to end, but it's going to compartmentalize, and that's going to make it less efficient at the margin. Energy is going to get more expensive as we address climate change, making everything else more expensive along with it.
All of these things cause structural inflation, and thereby necessitate structurally higher interest rates. Not to mention the fact that at the very least, the tailwind of globalization is over, which means that the ultra low rate regime we have enjoyed over the past few decades absolutely cannot be sustained. Inflation is back as a problem for the fed unless and until we got another sustained source of cheap growth.
There is one, and only one plausible such source imo, and that is powerful AI labor substitution. Hence the solution: Go long high interest rates, and make a levered long term bet on AI. These two things hedge each other. You don't have to believe AI labor substitution is going to happen (I'm not totally confident it will), all you have to believe is that it's our only out for structurally higher interest rates.
A portfolio constructed to benefit in the right proportions from these two things is currently cheaper than it ought to be, in my opinion. I expect this bet to play out roughly over the next 10-15 years, and I am still not decided on exactly how I want to construct it in terms of specific assets. But broadly I think it's the right move.
EDIT:
Totally separately, i'd like to quibble slightly with this bit of analysis:
> This suggests that, if interest and tax expenses had not declined as a share of EBIT (as shown in Figure 1), then the real growth rate of corporate profits would have been almost 2 percentage points lower each year (5.4 – 3.6 = 1.8 percentage points). In other words, the relative decline in interest and tax expenses is responsible for a full one-third of all profit growth for S&P 500 nonfinancial firms over the past two decades (1.8 / 5.4 = 1/3). This is a very substantial contribution
This isn't quite accurate. Cheap debt decreases the hurdle rate for capital investment. In other words, if rates weren't so low, much of this debt wouldn't have been taken out, and only the higher returning projects on average would have been invested in. Stated another way, as interest rates decline, the profitability of the marginal debt-financed investment declines along with it. It can also lead to anti-productive debt-financed market share wars between companies, etc (think of the vc battles between uber and lyft). This complicates the picture they're painting a bit, and likely reduces the true number here somewhat, but doesn't alter the broader story.
"Go long high interest rates, and make a levered long term bet on AI."
The challenge with levered long bets on technology is that most foundational technologies undergo waves of innovation, growth, replacement, and bankruptcy. You could've been totally right about social networking by betting on Xanga in 1999, and still gone bankrupt because they were long gone by the time Facebook ultimately won the market. Same with betting on Altavista for Search in 1995, WebVan for e-commerce in 1999, BlackBerry for smartphones in 2002, etc.
If you're looking to make a levered long-term bet on technology, often the smartest thing is to bet your career on it. The skills you personally hold will survive multiple generations of companies, and then you can seek out whoever happens to be hiring at a given point in time for the technology that you think will be huge.
> The challenge with levered long bets on technology is that most foundational technologies undergo waves of innovation, growth, replacement, and bankruptcy. You could've been totally right about social networking by betting on Xanga in 1999, and still gone bankrupt because they were long gone by the time Facebook ultimately won the market. Same with betting on Altavista for Search in 1995, WebVan for e-commerce in 1999, BlackBerry for smartphones in 2002, etc.
Ya, that's why it's a hedge, not an isolated bet. The idea is that one of two things has to happen: high interest rates, or strong AI labor substitution. You construct a portfolio to benefit from either outcome.
My expectation would be that the AI bets would lose money and the interest rates bets would make money, on average. However, in the case where AI really does undergo a miracle in the next decade sufficiently profound to lower interest rates, my portfolio would be hedged to that.
The challenge is that the eventual winner in the AI space might not be a company that exists today. Say that you build a portfolio that's 60% interest rate swaps and 10% each into DeepMind & 3 other hot AI companies of the moment. In 10 years, we've solved the labor shortage, but the eventual winner is a company named ConvBrain, founded 2028. All the capital you allocated to the AI space is tied up in DeepMind and the three other incumbents, so unless you're watching the space like a hawk (in which case you might as well be employed in it), you would've missed the ascendancy of ConvBrain and picked the losers despite the space as a whole being a winner.
Agree. You would have to keep up with changes in the space, and accept the possibility that you even if you're right about it strategically, you aren't able to make the appropriate bet, because e.g. the company that succeeds is private.
However, you could try to think about ways to hedge that risk, by thinking about who benefits in ancillary ways. E.g. for instance, which firms are likely to benefit most from AI labor substitution (whether developed in house or not). There are no certainties here, of course.
Betting on higher rates can be done using interest rate futures: https://www.cmegroup.com/markets/interest-rates.html These are fairly complicated instruments though so you'll want to do a bit of reading in order to be sure you're making the bet you want to make.
Betting on AI is more complicated: Obviously chip makers are likely beneficiaries, along with tech giants, in particular Google via Deepmind appears to be a leader in this area. But there are other ways for this sort of thing to play out, a lot of companies HNers probably never think about are investing in AI automation for their factories and automation in various ways. One possible future is that Deepmind builds The General AI Solution, and then licenses it to everyone. Another possible future is that the tech becomes so easy to build that each company just builds its own tailored solution for its particular problem.
I think I find the latter solution somewhat more plausible, but both are definitely in play. How you invest to benefit from that is tricky. It's likely in that scenario that some big companies will develop solutions in house, and others will buy startups that you'll never have a chance to invest in. Your job as an investor at this stage would be to figure out which big companies that you can invest in are likely to be the ones who do this successfully.
And when I talk about big companies here I don't necessarily mean the big tech companies. I mean companies like Tyson that makes chicken, or chemical companies, or firms like Accenture, etc. This is the harder side of the bet, because it's going to play out over quite a while, and there is a lot of uncertainty about exactly how.
Moreso than AI, ML will continue to integrate into business and improve overall productivity. I don’t worry so much about monopolies in AI, because compared to the business solutions from ML — which are often much smaller, boutique models that perform more efficiently than large, general purpose models.
I believe the soon to be realized lesson is that solving one specific problem is significantly cheaper than trying to leverage much more expensive models built to solve a great many problems.
There’s still a lot of runway left in “small” models (eg BERT) that are still being researched and augmented to solve common business problems. 10 years from now, I believe some form of the current models will become industry standards as methods for solving specific problems.
Yep, I totally agree, as a median forecast. But the future is always uncertain, and it is possible the things we're seeing now will hit some sort of wall, where only large companies will be able to build highly general models and everyone else will be left licensing from them. I think that's unlikely, but I just wanted to enumerate it as a possibility. And when you're constructing portfolios for long time scales, you want to think about all the likely paths things might take and try to roughly assign a weight to them and construct your portfolio accordingly.
All of these things cause structural inflation, and thereby necessitate structurally higher interest rates. Not to mention the fact that at the very least, the tailwind of globalization is over, which means that the ultra low rate regime we have enjoyed over the past few decades absolutely cannot be sustained. Inflation is back as a problem for the fed unless and until we got another sustained source of cheap growth.
There is one, and only one plausible such source imo, and that is powerful AI labor substitution. Hence the solution: Go long high interest rates, and make a levered long term bet on AI. These two things hedge each other. You don't have to believe AI labor substitution is going to happen (I'm not totally confident it will), all you have to believe is that it's our only out for structurally higher interest rates.
A portfolio constructed to benefit in the right proportions from these two things is currently cheaper than it ought to be, in my opinion. I expect this bet to play out roughly over the next 10-15 years, and I am still not decided on exactly how I want to construct it in terms of specific assets. But broadly I think it's the right move.
EDIT:
Totally separately, i'd like to quibble slightly with this bit of analysis:
> This suggests that, if interest and tax expenses had not declined as a share of EBIT (as shown in Figure 1), then the real growth rate of corporate profits would have been almost 2 percentage points lower each year (5.4 – 3.6 = 1.8 percentage points). In other words, the relative decline in interest and tax expenses is responsible for a full one-third of all profit growth for S&P 500 nonfinancial firms over the past two decades (1.8 / 5.4 = 1/3). This is a very substantial contribution
This isn't quite accurate. Cheap debt decreases the hurdle rate for capital investment. In other words, if rates weren't so low, much of this debt wouldn't have been taken out, and only the higher returning projects on average would have been invested in. Stated another way, as interest rates decline, the profitability of the marginal debt-financed investment declines along with it. It can also lead to anti-productive debt-financed market share wars between companies, etc (think of the vc battles between uber and lyft). This complicates the picture they're painting a bit, and likely reduces the true number here somewhat, but doesn't alter the broader story.