The Global Supply Chain Part II – How Did Computer Chips Throw a Wrench in the Global Economy?

This transcript has been generated by an AI tool. Please excuse any typos.

00:04 - 00:47

Hi, everybody, and welcome to The Pulse, where we cover trends in the economy, markets, and asset allocation for long-term investors. I'm Matt Palazzolo, a Senior Investment Strategist at Bernstein and Head of our Investment Insights Team. Those of you who listen to our last podcast will recall that we've been talking about supply chain issues and their impacts, both on economic growth and on inflation. Well, today we're going to get to the heart of the matter. We're going to discuss semiconductors. Semis have been causing problems in a range of industries over the past few months, most notably in the shortage of cars. And to help us understand these issues, we're happy to have one of Bernstein Research top analysts on the show today, Stacy Rasgon. So, Stacy, welcome to The Pulse.

00:48 - 00:49

Good to be here. Thank you.

00:49 - 01:01

So before we get to all of these issues in the semiconductor market, Stacy, can we just start with the very basics for those of our listeners who don't know, what is a semiconductor, what are they used for? How important are they?

01:02 - 01:52

So, I mean, a semiconductor, I mean, it's a chip. I'm trying to think how complex I want to get here. Let's try to keep it simple. Semiconductor is a chip. It is a computing engine. It takes in, you know, information and commands and it does functions. It can be computing functions. It can be power management, voltage regulation for your cell phone charger. There's hundreds or thousands of different things that these chips can do. And frankly, yes, where they're used, they are used everywhere, anything that you can look at, anything that is within your gaze, probably wherever you're sitting, probably has some kind of semiconductors in it, some kind of computer chips. It is safe to say that today's world does not exist without semiconductors. And the dependence that everybody has on these is getting more and more as time goes on. So they are incredibly important for everything that we do every day.

01:52 - 02:07

And I guess that's the point, right, it's because they are so ubiquitous and because they are almost everywhere, because so much of what we do is technology based or electronics based, when we ran into problems with the supply chain, and semiconductors in particular, it caused ripple effects.

02:07 - 02:19

Before we get there, if we, maybe we step back, Stacy, to the earliest days of the pandemic. What happened with chip production and orders for chips back in February and March of 2020?

02:20 - 02:49

Yeah, you bet. So that was when the pandemic first started. That was when the first lockdowns began. And so there was an issue with production in many cases because of those lockdowns, factories were shut down and everything else, that caused problems. But at the same time, the companies were worried that demand would go away because the consumers themselves were also locked up. And by the way, I'll get to it. As it turns out, this was entirely not the case. Demand in many cases was not impacted. If anything, it was exacerbated. Yeah,

02:49 - 02:51

Was just going to say, it was the opposite,

02:51 - 03:09

It was the opposite. But we didn't know that, and neither did the companies. And the industry, like many industries, tends to be terrible at forecasting right off the bat. And so what happened is, we had supply shutdowns. We had at least initially there was demand, you know, collapse or at least worries of demand collapse. And there were order cancellations because of this.

03:09 - 03:14

Sorry to interrupt. Just so we know, where are these factories? Are they all over the world? Are they concentrated in one area?

03:14 - 03:52

Many of them are in Asia, places I [...] in Korea. But there are factories everywhere in the world. So, well, when it happened effectively, like I said. So there were supply shutdowns, there were order cancellations. And that was probably, you know, sensible at least for a couple of months in the beginning. But as you know, the lockdowns continued. The industry went from, you know, from working in the office to working at home, from going to school to home-schooling from home. Demand in many cases did not collapse, it actually spiked. And so you have this massive surge in demand, right when, you know, in many cases the supply had been shut down, so that had to reverse.

03:52 - 04:42

And some of these end markets, you mentioned automotive, and automotive is kind of the poster child for this. The auto industry was probably the earliest in canceling their orders and their capacity immediately got backfilled from other industries that did not cancel. As far as, for example, you know, PCs or anything else where demand surged. And ultimately the auto guys like had to go back hat in hand and sort of beg for capacity. And then the semi guys were like, well, okay, great. You know, we didn't start your wafers, like call us in six months, maybe we'll have some for you by then. The other issue with automotive, by the way, was they tend to run very lean, like they talk about like running with a, just in time. They don't run, they sort of didn't run with any buffer inventory. So there was absolutely no slack in that system. And because of their order, their own order cancellations and the supply chain, which was they actually got hit much more than any other industry.

04:42 - 05:20

We've had further shortages, though, because because of demand. And you have to remember, everybody's talking about semi shortages. And I think people are getting the impression that people were not selling semiconductors. Quite the opposite, in 2020, obviously, the beginning of the year, people thought the industry was going to go into a downturn, and it actually didn't. Semiconductors grew 6% in 2020. They did $440 billion, was up from, I think, 412 billion in 2019, and in 2021, amid all of these shortages, we did $556 billion, the industry grew 26% year over year, it was the strongest year we've had since 2010, which is the recovery of the financial crisis.

05:20 - 05:31

So how is that even possible? Were they just running, were they running more hours per day? Did they build out new, you can't build out new plants in six months? How did that all happen? How do you get to a new high in production?

05:31 - 05:54

Because the factories weren't full, right, when they were running at $440 billion. Now they're absolutely full, like they're running out. Right. This is one reason we have these shortages because the industry didn't really have the capacity in place to really support a run rate of $556 billion, because they would have had to put that capacity in two years ago and shareholders would have shut them. Right. They wouldn't let it. Now they're putting it in. And so we'll see.

05:54 - 06:35

And this gets to the other worry is, is of this really, really struggling, we actually don't know how much of it is real and how much of it isn't. So that there's a phenomenon in this industry. It's called double ordering. And many of you listening may be familiar with this. This is the toilet paper situation from early in the pandemic writ large. What happens when customers can't get the parts that they need is they order more. And I'll give you a contrived example. Let's say I'm making widgets and I need 100 parts, 100 semiconductors to make my widgets. And my supplier says, okay, great, I can give you ten now and I give the other 90 in a year. And by the way, we're in that situation where the lead times, which is the time it takes between placing your order and getting your parts, in many cases are stretching out to a year or more right now.

06:35 - 07:09

So what is your next step as a customer? Do you hear that? Your next step is order 1000 semiconductors from anywhere you can possibly source them, on the hope you can cobble together the 100 that you need, maybe cancel all the other orders. But this is normal. This is a normal phenomenon of what happens, it's part of what drives these cycles. And so even of all of this $556 billion, we don't know how much of it is actually real and going into end products and how much of it is phantom and going on the shelf. We have no idea. We won't know anything until lead times pull back in, because when lead times were a year, your revenues are not being gated by demand, they've been gated by supply.

07:10 - 07:27

You mentioned the auto industry as running generally pretty lean. How about some other big users of semiconductors in the production of their widgets? Do they run lean, is it fairly variable from industry to industry? Where were they going into March of 2020?

07:27 - 07:49

So this is one reason that they give... 2019 was a downturn. It wasn't a massive downturn. I think we're sending revenues globally in 2019, we're down 12%. So it was a modest downturn and it's fine, but channels may have been somewhat lean because of that, because we were in a downturn already. And then you had to sort of spring back into the COVID situation.

07:49 - 07:56

Help our listeners understand why down 12%, 2019, I think about the economy was doing fairly well. There were no issues.

07:56 - 08:36

Let's step back. I'll give you some tips and let's go back to, let's go back to the financial crisis. So I started this job in April of 2008, about three weeks after Bear Stearns failed, and 2008 I don't know how many of your listeners were around during the financial crisis, looking at this stuff, but it was kind of the end of '08 when everything hit a wall, right? This was the financial crisis. People thought the world was coming to an end. And most of the forecasts for 2009, in the beginning of the year had the industry down anywhere from 20 to 30% year over year. By the time we exited 2009, it actually was, we had a massive rebound in demand; in 2009 we wound up being down 10%. 2010 was up 30.

08:36 - 09:05

And it was this, it was the same kind of thing. Right? We had this massive...what looked like a cataclysm at one point, this massive disruption. People thought the world was coming to an end. They drained out their inventories. Turns out it wasn't. And they were all caught flat-footed. They had to massively order. So 2010 was a big recovery year, and we crossed $300 billion for the first time globally to the industry. We didn't really grow for like six years after that, kind of plus or minus. We were within 10% of that number, plus or minus through like, until 2016.

09:05 - 09:09

And what drives growth? What it was just GDP plus? Or is there something else?

09:09 - 10:01

The only correlation is the GDP or high. There's also content increase. Over the cycle, it's probably GDP plus kind of grower. But in any given year, it can be a lot, a lot more or a lot less. It's cyclical. It's a cyclical. Cycles tend to happen because, think about this, like supply takes a long time, and it can take years to add. Demand can change on a dime, so you tend to overshoot in both directions. Again, it's not abnormal, but over the cycle it's probably mid-single digits. 2019 was a downturn. A lot of that was because of, this was when the trade war started under the Trump administration. And if you remember, all the trade and tariffs and everything got put on middle of the end of 2018. 2019, it started in fact, all of the stocks did very, very well in 2019 because everybody was playing for a trade deal. But even with the downturn, we still did 412 billion. So we're still sitting well above that $400 billion. In 2020, 2021, We've been growing and now we'll see where we go from here.

10:01 - 10:11

Yeah, and I want to get to that. But I didn't appreciate how cyclical of an industry this is. I generally think about, you know, construction and, you know, cranes and things like that as heavily cyclical. This is true, isn't it?

10:12 - 11:04

Well, there's two types of cycles, by the way. So there are short-term cycles. These are inventory cycles. So semis are the back of the supply chain, and small fluctuations in demand can propagate backwards or have correspondingly larger impact on the semis. And what typically happens is you maybe there's a change in and then maybe things get worse and your customers say, well, I've already got enough inventory in my shop. I don't need any more. They've got too much. So they reduce their purchases of semi, upstream stop, and they sell off of their books. So by definition at that point the semiconductor companies are under shipping demand and so these tend to show up as a reset, like you go into some order and all the companies will guide five or ten points below seasonal, and that can reverse the other way when all of a sudden the customers are now they're too lean and they have to rebuild inventory. They buy more than they need. At that point, the semiconductor companies by definition are over-shipping. So these sorts of things tend to show up as resets and they tend to be shorter. Maybe they last a few quarters.

11:05 - 11:49

There's another type of cycle, the supply cycle, so stop me if this starts to sound familiar, like supply is tight, capacity is tight, pricing goes up because customers are fighting for the available capacity. Companies are leaving money on the table. So they start to build and it can take time. It could take 18 to 24 months between making the decision at capacity to actually bring it on line. Usually like just from sheer dumb luck, you know, it usually comes online, right when the business cycle starts to turn, your factories go from full to empty, pricing starts to fall because now you're priced to fill the factory. Because as long as you're covering your cash cost, it's always better to fill it with something rather than nothing. Or so you get units and pricing moving in together and you get these these big swings.

11:49 - 12:19

So these are so your stereotypical like four-year peak to trough the peak kind of cycles, supply cycles. We get these in memory quite often because memory is a commodity. We haven't really had a broad based supply cycle outside of memory since the tech bubble, like in the early 2000. We're in one now right now, the world of a global fashion supply cycle. We've got an inventory cycle on top of it as well. We've never had a cycle quite like this before. We've also never had one that impacted every single end market in every single geography simultaneously before, because we never had one that was checked off by a global catastrophe. That's where we are now.

12:20 - 12:30

So I don't want to bury the lead here, Stacy, where are we? Give us a sense for the lay of the land and you know, when does it get better and what does better look like? What is normal in your mind? And when do we get there?

12:30 - 12:33

Yeah. So we won't know for a while until lead times pull in.

12:33 - 12:55

So let me step back. There's a broad controversy in the space going on right now, this kind of peak cycle versus stronger for longer. And from a fundamental standpoint, I would say the stronger for longer has been cracked. It's already lasted longer than I would have thought. And frankly, it's funny, the more disruptions we get, the longer in theory this can last. Like we've got new COVID shutdowns in China right now, perversely, that could actually make it stretch even longer.

12:55 - 13:47

Right. I'd say from a sentiment standpoint, people who tend to be firmly into the peak cycle camp and they're worried that all of the demand that we're seeing is not sustainable. Again, we won't really know anything until lead times pull in because like I said, when lead times are really long, the revenues are being gated by supply not demand, and people are just shipping as much as they can make and customers are buying as much as they can get. And I'll talk about some of the signs that are getting people worried. But in general, we won't know, like if lead times pull in, I'm going to make up the numbers, but if lead times pull in from 52 weeks to 15 weeks or whatever is normal, if the demand trajectory was still there, then at that point you could probably underwrite it. Then you could feel confident that you're actually now being gated by demand rather than supply. What investors, and again, I won't mention any names, but broadly, what investors will tend to do when lead times start to pull in, that's when they tend to hit the panic button, because usually what follows when lead times pull in, that's when the order cancellations start. That's when people realize they already have enough stuff.

13:47 - 14:18

And just simplistically, I mean, here's the way to think about it. If I'm a customer and I'm buying semis and it takes me three months to do on weekends, I'm going to I'm going to grossly simplify. But roughly, I'd probably want to have about three months worth of inventory on the shelf because it takes me that long to replace the part once I use it. If all of a sudden the lead times are a year, I probably will have a year's worth of inventory on the shelf. So now if lead times pull back to six months, I've already got enough. And there are some signs that not all of the demand is going into the end product because for a lot of these end markets, you can track end market shipments or revenues and you can track the semis.

14:19 - 14:51

So I will give you some examples. Let's start with PCs. PCs is one of the markets where you can do this quite cleanly because I get good data on PC shipments and I get good data on CPUs and I know how to subtract. And PC have been very strong. For your listeners who don't know, PCs were running 250 million units a year, give or take, for several years pre-COVID. And they'd come down from the peak, which was 350 million in 2011 and structurally, they'd come down, but they'd sort of stabilize here for a while, 250 million units. We did 300 million units in 2020. We did 350 million in 2021.

14:51 - 14:54

And they're buying computers for their kids who are doing homeschool. Just personally,

14:54 - 15:21

I have four kids. I bought four notebooks like last year. They were home from school for 18 months. Right. So you could argue that the steady- state run rate going forward should be higher than it was because now the installed base is bigger with more to be replaced. But it's still an unknown what that natural level is going to be. We don't know yet. Right. But PC CPUs were much, much higher. And so when you look at notebooks specifically, for three or four quarters, as strong as like notebooks were, notebooks CPUs were over-shipping notebooks by 30%.

15:21 - 15:22

What is

15:22 - 15:38

Stacy? I'm sorry. The processors, the central processing unit that goes into... there's two companies that sell them, and I won't mention the name, but they sell those CPUs and they were over-shipping holdings by 30%. They started to correct in Q3 of last year, but there was a massive amount of over-shipment in that market.

15:38 - 16:10

Let's take automotive. So again, like all of these auto guys are complaining and there are clearly production shortfalls because people use the term SAAR sometimes, seasonally adjusted annual rate. When you auto SAAR, that just means how many cars get sold. So the auto SAAR pre-COVID was like 95 million cards, give or take, like light vehicles. We did about 76 million last year. So they were clearly having production shortages because they're having production trouble. At the same time, you can look at the auto semis, so the auto shipments were 20% lower than they were pre-COVID and auto semi shipments were like 30% higher. So massive. Yes.

16:10 - 16:45

So if you do the division and you say, what's the content, over time, that content grows. That's one of the nice things about automotive. It's content that goes up, but we're over-shipping content by like 40%. And so you said, well, how can that be when they're having shortages like the issue, especially in cartridges, they've got a ton of semiconductors. You need like a thousand different semis to make a car. But if you can't get one of them, you may not ship your car. So they're clearly up in shortages, right, because they're out production shortfalls. But I also suspect they're buying as much of everything else as they can possibly get their hands on hand over fist. And so you can see those kinds of dynamics in many, many end markets.

16:45 - 16:52

Where are we on pricing? You know, given these the shortfalls, given the demand for semis, a price has gone up. Do they have pricing power?

16:52 - 17:28

They have gone up. They do broadly. And so I cover quite a few companies. I have not had a single company complain about margin compression because of input costs. And for all the semi guys, all of their input costs are going up. What most of them are doing, are passing along those input costs and their customers, by and large, are accepting those increases. Most semiconductor companies are not trying to be terribly transactional about it. They're not really trying to extract more because for a lot of these guys, especially in markets like automotive, for example, they're very long-term relationships. So you don't really want to try to squeeze your customers, but they are passing along input cost increase and pricing is going up, but it's not like 40%.

17:28 - 17:36

I know you've talked about Neon. I want to get to that in a second. But just to clear this, what are the input costs that are going up? But what goes into a semiconductor?

17:36 - 18:27

I mean, it's everything. So it's some, if you make them yourself, there's a couple of different models that do semiconductors. Some companies make them themselves. That model is called IDM, Integrated Device Manufacturer. These are companies like Intel, for example, that both design and manufacture their own chips. Intel is a good example of an IDM. The other model is called a Fabulous Foundry. These factories are very expensive. There aren't a lot of companies that can afford them. And so what happened decades ago is you split the supply chain into companies that only do chip design. They're called Fabulous companies. They don't own FaBs. The FaB, that is the name of, what the factory's called. Companies like in video or Qualcomm are fabulous companies. They only design chips. They don't manufacture them. And then they outsource the manufacturing to what's called a foundry who agglomerate demand together from many different customers and they can build up the revenue scale that's needed to support the investors.

18:27 - 19:08

TSMC is your classic foundry, they're the biggest foundry, for example. And so if you're making chips yourself, you've got your raw material, your raw silicon wafer costs go up and your energy costs are going up and the prices of gases and chemicals and all that stuff is going up. If you're a fabulous company, you know your foundry quotes are going up. The foundries are charging more because their input costs are going up. You're making them think they have to buy equipment. Right. There's a whole separate industry. It's called semiconductor capital equipment. They just make the tools that make semiconductors. You know, those prices are going up; as you're moving to more and more advanced products, those costs go up, right? So there's lots of inflationary pressures, but so far the industry has done a pretty good job of passing those costs along.

19:08 - 19:14

Is it fair to assume that the production of a semiconductor is fairly light on labor and heavy on capital equipment?

19:15 - 19:59

It depends. And if you're looking at a leading edge manufacturing facility, the labor's probably, ooh, it's probably ten or 15% of the cost structure, like very roughly, but depreciation could be 50%. So they're more capital intensive. But, but there's still a lot of labor and it's very skilled labor. So this is actually another whole thing we could get into. Obviously, there's a lot of interest in building more factories and we haven't even talked about it yet. But there's obvious governmental investments and interest in doing this. Labor is a huge issue because even if you have the money to build these factories, how are you going to staff them? These aren't just like people you hire off the street, even the technicians and the operators need to be skilled and the engineers, I mean, they don't grow on trees. They're not like labor intensive in the sense that, you know, like digging ditches is labor intensive, but it's highly skilled labor that you need.

20:00 - 20:05

Talk to our listeners about Neon. You've mentioned that a couple of times. Why is that relevant?

20:06 - 20:31

Yeah, you bet. So there's a number of... so semis are, by the way, a global industry. And I know there's a lot of talk right now about bringing manufacturing onshore and [...]. But there's lots of materials and other things that come from all over the world. And it turns out there are a number of key materials that are sourced from the Russia-Ukraine region. Neon is one of them. What is a neon, neon is a noble gas. And it turns out to be very critical for manufacturing semiconductors. When you make semiconductors,

20:32 - 21:19

I'm going to grossly simplify here again. When you're building up your circuits on your silicon wafer, you're doing three things. You're putting material down on your wafer. You pattern that material because you only want material in some spots and not in others, and then you take material away. That patterning process is called lithography, and you basically use a laser to transfer a pattern of your circuitry on under the wafer and use a laser light to do this. In this krypton fluoride in our guide fluoride eczema, lasers like 95 to 98% of the gas and the gas is actually neon. And so it turns out without neon, semiconductor manufacturing comes to a screeching halt. You must have neon. You must have it. And as it turns out, Russia, Ukraine is responsible for maybe 20 to 25% of the world's neon.

21:19 - 21:21

And where is it? It's in the ground?

21:21 - 22:00

No, no, no. So it's a constituent of the atmosphere. It turns out when you have large facilities...that's usually done in large old steel plants, they have these big air separation units that are used as part of the steel process, steel manufacturing process, where they're basically like, cooling oxygen out of the air. And it turns out there are rare gases, krypton and xenon in Oregon and some others that come out along with this. And so crude neon is basically made as a byproduct of these large air separation units in these steel plants. And then you ship that crude neon to purification plants to further purify it. So as it turns out, Russia and Ukraine from back in the Cold War have a lot of these like old dirty steel plants.

22:00 - 22:07

I was going to say, why, why do they have such a monopoly? It's because they have, first of all, they have a lot of them and they're poorly run or they're dirty?

22:07 - 22:28

No, that's not a monopoly. Like you said, they've got 20 to 25%. So it's not like 100%, but that's a good chunk of it. If you go back five or ten years, it probably was 60 or 70% of the neon came from these regions. So it's still a lot. But yeah, but they've got a bunch of old steel plants that have been there forever, which by the way are probably getting bombed right now. Like there's no neon coming out of Ukraine. So this is potentially a problem.

22:29 - 22:56

I think we're okay for now, but this is not a transparent market, unfortunately. It's not big. The market may be a couple of hundred million dollars. In the context of a $550 billion semiconductor industry, it's tiny. But all of a sudden, if you lose 20% of your neon capacity, it could become a problem. The industry's got several months of inventory. There are other sources, again, there are sources in China and Europe and the US. I do suspect some of these other sources are exploring now adding capacity a little bit at a time.

22:56 - 23:31

I think pricing is likely to go up, but we saw something similar to this back in 2015 during the Crimea crisis. The industry got through it then. I think we're okay probably for six months. I think beyond six months we'll start to see some disruptions. I think we'll see pricing going up if it's not already going up now. I mean, you have to remember that the semi industry can afford to pay whatever it takes probably. It's not, it's a tiny piece of the cost structure. If, however, you need Lasik eye surgery or something, I would go get it now. Those lasers use neon as well. I think if this goes more than a year, we may start to have more problems, although by then you could start to see some other capacity come online. So we'll probably muddle through.

23:31 - 23:36

For all you eye surgery listeners. Don't say we don't add value in this podcast.

23:37 - 23:49

Stacy, I want to start to pull it together and we've got just a couple of minutes. But you know, what are your conversations like with people on the street now? The investors that call you up and they say, Stacy, what's your take on semis and the broader industry now? What are you telling them?

23:49 - 24:12

Yeah, I mean, like I said, it's where are we in the cycle? What do you think about the cycle? Is the first question, which makes sense. We're at that point of the cycle now. And what are we seeing? I told you, sentiment is firmly in the peak cycle camp, you can see that, estimates numbers are going up and the stock prices are going down and multiples are contracting. But that's why; that is not atypical in this part of the cycle. That's just where we are. So that becomes the question at what point do we price in and out.

24:12 - 24:57

If you look at sector valuations, they've come in, they were trading, I think, at the... let me just take a look at like the stocks index, it's just a proxy for the industry. I think at the peak in November, the stock was trading at something like a 22 forward price to earnings and it was a something like a ten or 15% premium to the S&P in the market, which is high. Right now, at least as of a week or two ago, when last I looked, sector was trading about 16 times, which actually was dead on the average, like from the five or six years pre-COVID. And we were trading like at 16% discount, which is low, kind of on the low end relative to this. You're still kind of on the low end of normal, but it's kind of low. So I would argue that valuations have kind of normalized, but we probably haven't seen panic yet, you know, because the stocks tend to move before the estimates get cut.

24:57 - 25:21

So like what are we pricing in and like how bad can things get and when are we going to start to see it? I wish I had a better idea on timing. I just don't. All I could do is do the math. The math is making me somewhat nervous. Like I said, you can look at these sort of over-shipments, and now we've got a few new things, unfortunately. is There actually, we've obviously got Russia-Ukraine, which probably isn't going to be great for demand, at least in Europe. We have new China COVID lockdowns, further supply chain disruptions and everything.

25:22 - 25:44

And then, frankly, they are starting to see some some incremental signs of end market weakness, at least in consumer focus, things like PCs, China, Android, Chinese smartphones all look pretty weak. Enterprise in other areas still look pretty good. Looks like it's holding and data centers all look pretty good. We're starting to see some incremental signs of weakness, at least on some of the consumer markets. So that starts to get people nervous. People always worry that it's just sort of dominos, right? One market goes in the next one follows.

25:44 - 25:54

Yeah, I got to say, it's a fascinating industry with a lot of different inputs. So hats off to you for doing a good job covering for so many years. And I want to thank you for coming on our podcast today as well.

25:54 - 25:58

My pleasure. Any time anytime you want me back, just let me know, I'm happy to.

25:58 - 26:31

And thanks to all of our listeners out there for joining us, as you can tell, semiconductors are a fascinating industry with a lot of moving pieces at any point in time. And as always, we'll continue to monitor the key trends and the risks in the economy and share them with you here on The Pulse. If you've enjoyed this podcast, please subscribe and rate us on Apple Podcasts or Google Play or Spotify, wherever you listen to podcasts, and e-mail us with your thoughts or questions or any feedback that you have to insights@Bernstein.com and be sure to find us on Instagram and on Twitter at BernsteinPWM. Thanks and be well.