Eric Lascelles: Managing Director, Chief Economist and Head of Investment Strategy Research

Hello and welcome to our latest video #MacroMemo. As always, there’s plenty to talk about, with a very heavy skew toward developments in this war with Iran and oil and so on. And so we'll talk about some of the major thinking on that subject. We'll talk a bit about some of the negative developments, some of the risks that exist on that front.

We will equally address some of the positive developments, some of the good news there and ultimately summarize into something aligning with some macro, some economic and market ramifications. And at the very end of this video, we will also talk a little bit about AI. It's such an important subject and I don't want to lose sight of that as other events transpire.

And so we've done some work on the rate at which AI is getting better. And that sometimes is kind of lost in terms of how much better it is than a year ago. And so we'll talk through that.

But let's start on Iran. Certainly, now, just over three weeks into the war as I'm recording this, and the scope for things to change on a day-to-day basis.

And so, do be gentle, but nevertheless, as it stands right now, the war continues. The price of oil is in the approximate vicinity of $100 or so a barrel. The situation remains quite volatile.

We've seen quite large market swings in oil price swings in recent days. And I would say there are three approximate stylized scenarios. There is an optimistic one in which something approximating a cease fire is reached, perhaps by the end of April.

So within, let's say, two months of the onset of the war. The slightly more pessimistic take is that the ceasefire takes something longer than that. So I suppose more than two months. And then there is perhaps a worst-case scenario or a very negative scenario, which is that there is sufficient damage done to energy infrastructure, that the oil shock isn't just temporary.

And we're not just debating whether it lasts a month and a half or four months or something like that. Instead, it's long lasting and there isn't an ability to restart production and it proves highly problematic. And so that's the worst-case scenario. We're a little bit more sympathetic to the optimistic versus the pessimistic scenario.

So we do think this is a temporary shock. We think that it's slightly more likely it could be resolved within two weeks as opposed to longer.

I should acknowledge a slightly more pessimistic one, in which it takes more than a few months is also very much on the table. We just have a slight preference for the other. We think it's relatively unlikely to get to that worst case scenario in which there is large-scale infrastructure damage – though it should be acknowledged there has been some smaller scale damage done already, so it's not a zero risk.

Looking at some of the negative developments in recent weeks since we last recorded one of these calls. Well, there's a few things I think that we can say. One would be, as I just mentioned, some energy infrastructure has been hit in Qatar and in a few other places. So there has been some enduring diminishment to the ability to produce, though it is still, on the grand scale, relatively small.

It is still very consequential that the Strait of Hormuz is significantly closed. There are more than 700 vessels stuck in the Persian Gulf, and there is still this asymmetric problem, perhaps two asymmetric problems. One is Iran has plentiful and relatively cheap drones it can use versus quite expensive interceptors the U.S. and Israel are being forced to use to stop those drones in particular.

And, the other asymmetry is that Iran only needs to be able to hit the very occasional ship, not all of them, to effectively plug the Strait of Hormuz. And so that is very much the issue. You need to almost completely neutralize Iran to feel comfortable with that shipping lane and to get oil flowing again across the Strait of Hormuz.

The other negative development or concern, perhaps, is just that even if there were to be a ceasefire and it happened tomorrow, it takes weeks to potentially months to fully restart some of the stopped production . . . to get the ships where they need to be . . . essentially to get product to market. And so, again, as much as it could well be a one-month event, in terms of the war itself, it then takes another several weeks to months to actually get things going.

So there was a real impact there as well. So that's the negative. And it is significant and perhaps worth flagging that the market has become, if anything, more negative than positive in recent weeks, though it has ebbed and flowed, and there was a bit of positivity in the last few days, but not enough to offset some of the earlier negativity.

On the positive side though, and again, I did already tip my hand, suggesting that I'm a little bit more optimistic than some. There are still very strong incentives to resolve this war. And I wouldn't want to underestimate that importance in terms of getting it resolved.

And so China is the number one consumer of the energy products coming from the Persian Gulf.

It cares a great deal about this. India also cares a lot, as does much of Asia, as does Europe, particularly from a natural gas perspective, as does the U.S. And so of course, the U.S. was the initiator, or at least among the initiators. The U.S. cares about pocketbook issues right now. There's a midterm election coming in just over seven months.

Mortgage rates are thought to be in focus, and the price of gas at the pump is in focus. And actually, both of those things have been hurt by this war. The price of gas, quite obviously. But also, it should be mentioned, mortgage rates are higher because bond yields have gone up. And we're a little bit surprised that's the direction bonds opted to go.

Nevertheless, that's hurt mortgage rates as well. As you look at the personal popularity of President Trump, it's notably down since this war began. You look at betting markets and what they think happens in the mid-term election. They now give the Democrats equal odds for capturing the Senate, quite good odds for picking up the House of Representatives. The single most likely two-by-two matrix permutation of who holds the House and Senate coming out of the midterms, the most likely single outcome is now that it is a Democratic Party sweep, which certainly wouldn't be to the liking of the White House.

And so, again, they're strongly incented to find a resolution. We've heard some talk recently about perhaps stepping down and de-intensifying and bringing the war to an end, perhaps without a clear victory. Nevertheless, reducing the pressure on oil and on the Strait of Hormuz.

Stepping back for a moment, it is useful to reflect on the fact that there are two ways to resolve this war.

You can win it. The U.S. and Israel certainly have been winning by conventional definitions. The damage they've inflicted and so on has been greater. And they've ousted leadership and so on. Saudi Arabia may be joining. That is the rumor right now. More forces are being sent by the U.S. to the region.

So it's quite possible there is just a clean and traditional military victory that incapacitates Iran. Or the other way that this could be resolved, relatively quickly, is via ceasefire. And the U.S. has signaled recently that perhaps its appetite for war is diminishing. And, we gather negotiations are happening, possibly via Turkey as an intermediary.

And then just looking at it from a different standpoint, Iran's capacity to limit the Strait of Hormuz and to attack its Gulf neighbors may be declining. We are tracking the missile supply and usage more precisely. And the same for drones. The Iranian missile supply is thought to be down by about 55-75% based on the missiles that have been used and destroyed.

The number of launchers in Iran are thought to be down about 60-70%. So a significant diminishment. Not clear about drones. That is the risk. There may still be a lot of drones. But when you actually look at the number of attacks Iran is conducting on other Gulf states, it is visible that there has been a decline over time, a particularly sharp drop in recent weeks.

It's not just a decline in missile strikes. It's also a decline in drone strikes. It suggests, without much precision, that perhaps Iran is running out of some of these weapons. And so that would of course be a positive sign in terms of finding a resolution of the war and really getting oil and gas moving again.

On that front, in the Strait of Hormuz, you have 20% of the world's oil and natural gas liquids transit through that area. So 20 million barrels a day, loosely speaking. It should be noted that, based on the efforts of other parties and Saudi Arabia's alternative pipeline, it released a strategic petroleum reserve, and allowing Russia to produce more and so on.

The thinking is that instead of losing 20 million barrels, you're losing more like 5.5 million barrels a day. That's still very significant. It helps to explain why oil prices are in low triple digits as I record this. But ultimately, it's about a third to a quarter as problematic as it first looks.

I guess the other angle is – and this is perhaps a segue into the economic implications – but the world is approximately 62% less sensitive to the price of oil and therefore to an oil shock than it was in the 1970s.

The answer is just that cars are more fuel efficient and the economy is more service oriented as opposed to manufacturing and goods oriented. And so an oil shock just doesn't have the same pop as it did in the 70s. It is still consequential, though. And so when we talk about the economy, we see a hit to GDP (gross domestic product).

Our modeling would suggest if oil prices stay here permanently and if natural gas prices stay here permanently, some major economies could lose up to a percentage point of GDP growth, which is significant. Some countries could see inflation temporarily of up to one and a half percentage points more than normal, which is also very significant. I would flag, though, that modeling – if this is permanent – and so unless you think energy infrastructure will be so damaged that production can't revive later when a cease fire or resolution occurs – probably those numbers are too negative or are too large relative to the reality. We look at central banks and markets have priced in more rate hikes. And it does depend, you can be more sympathetic to that, let's say, in Canada where the economy could benefit from higher oil prices.

And inflation, of course, is also higher. In a lot of markets, though, we would express some skepticism. We think the market has perhaps over-priced. Rate hikes modeling would actually suggest you should be cutting rates if you're economy is hurt and there was a temporary inflation shock, a stagflationary shock like this, essentially. Our recession models are focused on the U.S. which isn't perfect, but it's nevertheless a bellwether economy.

I would say the risk of recession over the next year has gone up. It makes sense. Oil prices are higher, credit spreads are wider, the stock market is lower. Those are all inputs into the recession model. The risk has only risen from a 6% chance to about a 12% chance. And so, certainly not.

When we look at past oil shocks, and there have been quite a number, we looked at 15 dating back a good 50 plus years, and we tried to break them into demand side shocks and supply side shocks and so on.

This is a supply side shock. Historically, supply side shocks have resulted in about a 6% decline in the stock market, peak to trough. That's about where we are right now. And so not making any promise that the market can't go down further. This is a pretty serious supply side shock. And maybe it merits a larger move than the average.

But nevertheless, it is useful to reflect on that fact. Gives you a bit of a sense for the normal magnitude. And equally, again, to the extent we think this is a temporary shock, that does suggest there should be buying opportunities in here, and there should be some measure of normalization that happens later, whether it's within a month or whether it takes longer than that.

Okay. I'm going to finish on artificial intelligence. And so, as you may have noticed, we have been tackling this very, very important subject from a variety of angles. I suspect if we were to fast forward a year, we'll look back and hopefully say that situation with oil wasn't great, but it was ultimately temporary.

Meanwhile, artificial intelligence is this profound force and something that's relevant not just for the year, but for the decade and perhaps beyond.

So we need to pay awfully close attention to this, as the market was, up until this war began. And so we've tackled, as you may recall, various angles and the CapEx outlook, which is quite extraordinary, and electricity constraints, which are an issue, and how AI might change the world and utopian and dystopian scenarios and so on.

And this time we looked at a more basic question, but I think a very important one, which is just how much better are AI models getting? We just didn't have a sense, we realized, are they twice as good as a year ago? Are they better than that? Are they worse than that? And it's hard to measure.

There are some benchmarks that try to measure. I'm sure we'll get better at this over time. But based on what we've seen, we can say that since 2024 – so over the last, let's say one and a half or two years, there has been a 6.4 times per year annual improvement. So the rate of improvement is quite impressive.

Well, more than doubling every year, if that makes sense. If anything, it seems to be accelerating as opposed to decelerating. It was a bit slower prior to 2024.

And interestingly, we weren't sure what we'd find here. You can argue that there was a rising return on investment for AI. And so that is to say, the amount of compute required to train the models is rising about five times annually, which is a huge increase.

But recall that the output, what they're able to achieve is rising more than that – by 6.4 times. And so you're getting kind of more bang for your buck.

Meanwhile the training cost is rising. And ‘just’ may be not the perfect word here, but the training cost is rising ‘just’ three and a half times annually. So to frame that, again, the cost is going up three and a half times a year.

The amount of compute you're using is going up five times a year. So again I guess compute getting a bit cheaper as better chips come out. And the actual capabilities of the AI are going up 6.4 times per year. So AI is getting better faster than the amount of money you're having to spend on it to make it better, if that makes sense.

More AI bang for the buck. So that's quite promising. And just to bring it right back to economics, we are looking for faster productivity growth ahead. And as we talked about in the last one of these videos, we think the most likely scenario for AI is it's a major general-purpose technology that does very significantly drive productivity growth over a period of years and perhaps even decades.

Okay, I'll stop there. Hopefully you found this useful and interesting. I appreciate your time and please consider tuning in again next time.