Capital Flows and Asset Markets

This is a free preview of a paid episode. To hear more, visit www.russell-clark.com

I did a degree in Asian studies and Economics at University, and when I graduated in 2000, I knew that China was going to be the next big thing. In 2002, I tried to get a job in Asia, but no one was hiring, so I took a job as an emerging market analyst in London, which was a good second option. When I got to London, I was shocked at how little people knew about Asia. It was an afterthought at best. If you have been reading this substack, you should be aware I think JGBs lead US Treasuries.

One of the questions people almost never ask, is why is Japan a lead on the US?

This is a free preview of a paid episode. To hear more, visit www.russell-clark.com

I came of age in the 1990s. I went to High School in Japan in 1991, which was the top of the their bull market. I went to Hong Kong in 1998 during the Asian Financial Crisis, I got my first job in finance in 2000 at the top of the dot com bubble, I got my first fund in 2006 just as the housing crisis was kicking off. In all of these crisis at least 50% of the people I worked with lost their jobs. I made finding and avoiding bear markets my number one focus. Of course just as I thought I worked how to do this, the US began the most epic bull market of all time. Probably the best graph to show this is the performance of the S&P 500 relative to bonds.

Or another way to put this is equities kept getting more expensive even as bonds got cheaper.

I had originally put this divergence down to governments choosing “pro-labour” policies. But recent price action is making me think this could have also been driven by the AI boom. Now here is a wealth warning. Back in 2007, I ran an emerging market fund, which was shooting the lights out. I knew the US had some housing market weakness, and at the time I thought that would mean the Fed would cut rates, dollar would weaken, and emerging market would be fine - the so called “divergence trade”. I realised in late 2007, this was incredibly stupid, and cut the book just before emerging markets went cliff diving. When the US has a problem, everyone has a problem.

After that, I was able to pick out the steel and iron ore bear market, the shale oil bear market, and the Chinese tech bear market, and I guess the more recent bear market in bonds. But the underlying trend in stocks has been up. Unless you have been foolish enough to invest in Chinese stocks, most people have not seen a bear market.

Here are my basic guidelines to surviving a bear market:

This is a free preview of a paid episode. To hear more, visit www.russell-clark.com

I have huge love and admiration for JGBs, and by extension, the JGB investors. Basically, I never argue with JGBs. If they say inflation, I think inflation. If they say deflation, then I think deflation. I made a lot of money in Irish bonds during the Eurocrisis, but I am very proud of my purchase of 30 Year JGBS in 2015 at 1.5%, and then selling them at 0.10% a few months later.

When 30 Year JGBs did not touch new lows in the Covid sell off, I knew we were in a new world. And from there yields have surged. I always wondered what the JGB market saw. My best guess was they foresaw the huge increase in US government spending under Biden.

I also suspected they foresaw the huge budget deficits and tariffs of the Trump administration. All things I bunched under the “pro-labour” trade.

All this led to the GLD/TLT trade, which has been great, with limited volatility, until this year.

But this year, GLD/TLT has been correlated to the AI trade. Could JGBs have picked up the AI trade that early? And are they predicting the end? So time for some research.

This is a free preview of a paid episode. To hear more, visit www.russell-clark.com

The question that has been dogging me is whether the pro-labour trade, is just the AI trade - which is at its heart a supercycle in semis. Below are things that suggest end of cycle.

This is a free preview of a paid episode. To hear more, visit www.russell-clark.com

I read all the bear cases for the AI trade before. Over investment, rising cost of capital, fraud. I had seen them all before, and none of them had worked. The key to understanding the semi trade is understanding why semiconductors have turned from a deflationary product to an inflationary product. When you look at the PPI of semiconductor equipment, you can see how extraordinary it has been last few years.

Going through the mid-term results, I came across a new word - dummymandering. Basically it refers to attempts at gerrymandering, as the Republicans tried in Texas, backfiring and turning safe seats unsafe. Texas, the most Republican of Republican states has in essence turned mainly Blue.

Winning Governorship in Texas is a big deal, and a big surprise. Up until election day, it was not considered a likely outcome.

Gina Hinojosa ran on a cost of living campaign, and in particular was keen on restricting expansion of data centres. If we look at flips in governorships, they are in states where the a large amount of expansion is planned, with Texas by far the largest.

Generally speaking, mid-terms are a non-event. The party in power loses seats, and the legislative agenda slows. But the flip in Texas is a huge deal. Before the mid-terms, I would have said the change of a data centre build out slowing was relatively low, as all the main players were engaging in a capital war. But now, with the mid-terms, the chances of regulation slowing the data centre build out has increased.

Now seeing the Texas results, I can understand why investors dumped semiconductors AND sold gold, and bought treasuries. The possibility of a cyclical slowdown has increased, and possible something like a dot-com bust.

For me, the mid-terms and the win for Lula in Brazil shows that voters have tired of culture wars, and want more money. This is pro-labour trade I have been seeing for the last few years. But I can understand the weakness in this pro labour trade last month now. But on top of regulating data centres, Democrats also pushed for restoring Obamacare, increased government spending and lower taxes for households.

I cannot work out if this is truly bearish for markets. The big players in the market have been whipsawed as they announce big spending plans. Do markets then reward them for scaling it back? Seems that way today. Or do they look for work arounds? And if hyperscalers do not build data centres in the US, do they build them overseas? Or do funding markets and IPO markets close to all things AI, and we have a dot com bust? I think Japan is the best place to look. It has pro-labour politics AND big exposure to AI. If JGB yields tank - then what I thought was pro-labour, was in fact the AI trade.

I am always reluctant to trade around big political surprises. Market action has a way of bringing changes. Do Democrats comes out, and say they just want to regulate data centres and not slow them? Who knows. But at least I know what markets are looking at now.

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.russell-clark.com/subscribe

GLD/TLT has been surprisingly weak this year, admittedly after a terrific run.

It is not meant to be a economic trade, but a political trade. That is politicians will keep acting to favour labour, and wages, which means inflation should be stronger than expected. Looking at the last pro-labour era, you should end with a period where gold is strong and bonds are weak - a debt crisis, which then ushers in political change (There is a secondary trade of China swapping treasuries for gold - but we will leave that for the moment).

But you can also portray GLD/TLT as a cyclical trade. If you think the cycle is going to turn, you definitely do not want to be short bonds, or long gold. If you think unemployment is going to go up in the US, you do not want to be in cyclical trades. Looking at initial jobless claims, taking a view we must be near a cyclical top is easy to make. It has been so good for so long, there are 36 year olds who have never seen a recession in their working life.

It is hard to believe the last real recession in the US was nearly 20 years ago. The 70s, 80s, 90s and 2000s were all about recessions. It just became assumed they were unavoidable. Now my read is, that this politics of non-stop boom, or pro-labour politics will continue - until wages and asset prices have normalised. That is wages have caught up to asset prices. In a pro-labour world, that happens with inflation. In a free market world, that happens with the most enormous recession and bear market. I still think politics is pro-labour, so this seems unlikely to me. My read of the mid-terms is that the most pro-labour policy of Trump was to cut immigration. He was very successful at that. But high oil prices and cutting benefits was not pro-labour at all. At the end of the day, pro-labour is about giving workers more money, not rich people. So what do the mid-terms mean? Well I would expect the Democrats to advance a tax and spend budget, which Trump will veto, so you end up with just a spend budget. I see zero chance of the Democrats agreeing to any austerity. I also think market will start thinking about what a Democrat President will look like. Trump so thoroughly controls the Republican party, its hard to see how they can even put up a fight at the next presidential election. I would also see this as inflationary, but with the offset of tax increases. I suspect the long end of the Treasury curve should weaken (ie yields go higher).

As mentioned previously, the “Starve the Beast” strategy of the Republicans is meant to leave Democrat administrations with no money to spend, and fiscal discipline to be enforced by the bond market. My best guess is we will need to see actual bond market stress before we see any political change. So politically, I see the mid term result as bullish for GLD/TLT.

Economically the bear case is the Democrats legislated to end the data centre build out.

Which given the roll over in the residential construction market would probably be the top in the cycle.

Having looked at the plans for hyperscalers, and the way they think, I expect them to move data centres to wherever will have them, including Europe and Japan, that is until space becomes viable. Potentially a negative for the US and the US dollar. But the fear of being overtaken by China in AI should allow for negotiation on data centres - which is a political view. That is politics should keep the economic cycle going. I will keep leaning to the political view of GLD/TLT as long as JGB yields are tending higher.

And I will finish with Russell Clark’s third iron law of politics. It does not matter which party you vote for, you end up with the same policies. I still see fiscal excess, tight immigration policy, tariffs and industrial policy continuing - at least until the voters tire of it. I do not see the mid-terms as a repudiation of that.

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.russell-clark.com/subscribe

Earlier this year, I thought Softbank was worth a short. It had just raised debt at an 8% yield. Like many names in the AI space, its CDS were implying some problems.

Usually, rising CDS is associated with equity weakness - but in Softbank’s case, shares have tended higher, if extremely volatile.

This in contrast to Oracle, where equity valuation as CDS has risen. Oracle CDS has even exceeded 2008 highs.

And in Oracle’s case shares are down 60% from a peak.

Just looking at the share price above, you would assume that the dot com bust was already upon us. This is not the case, revenue growth has returned to Oracle after a decade of stagnation.

In the most recent quarter, Oracle seems to have responded to share price and debt pressure, and found ways to cut capex. I would expect this to be nominally bullish - but so far nothing.

What links Softbank and Open AI? Softbank is the biggest investor in OpenAI, and OpenAI is Oracle’s biggest customer. In other words, Softbank provides capital to OpenAI, and this capital is then used to fund the growth that Oracle is investing to supply.

The big question is whether OpenAI can actually meet the commitments that it has made to Oracle. Some rough financials for Open AI are below. We can see from most recent filings that Anthropic is seeing USD 17bn of revenue for Q3 from USD1.3bn a year ago. So the revenue forecast for OpenAI for USD 70bn in 2027 looks achievable. The question is whether they can find the capital for the USD 300bn of forecast funding.

If you are bearish, you can say there is no way that OpenAI and Anthropic can raise the cash they need. Whereas, when we look at SpaceX another company with big growth ambitions, it has raised USD100bn this year, making it the largest IPO of all time. That is OpenAI could get the capital from equity markets, as long as they stay open. In 2021, nearly USD 600bn was raised from IPO markets.

That is the key issue. Can Anthropic and Open AI raise the capital for the growth forecasts? Or in other words, can and will the IPO market stay open? With debt markets expensive, the IPO market becomes all important. And IPO’s can be all about narratives. Tesla and now SpaceX are good examples. Both listed at extremely high valuations (EV/Sales of over 20), which provided the growth capital they needed, and the financial safety to tap debt markets.

But IPO markets tend to be about narratives. In this case AI is going to take over the world, or it is just a capital destruction machine? The thing about narrative wars, is that the bears almost always lose in my experience. Tesla is probably the most relevant example. When it first listed, it was obvious that it needed more capital. Hence it attracted large short interest, but this inevitably lead to a short squeeze, which provide a more attractive level for Tesla to sell equity, which reduced the assumed dilution, which forced the valuation up, and so on and so forth. You can see short covering tended to occur with equity raises with Tesla. you can see how volume traded increased as equity was raised.

Short interest in Oracle is now elevated.

You might think Tesla was a one off occurrence - but we saw a similar dynamic with Strategy. Very high short interest led to Strategy pricing above NAV, giving the company the incentive to issues shares to buy Bitcoin, which helped created the upward surge in the cryptocurrency. As short interest has waned, crypto has struggled.

The problem with winning the narrative war, is that bearish bets create the conditions for a company to win the capital war, which ironically would not have existed without the bearishness. Every short is a confirmed buyer at some point. And every short will have borrowed from a long owner, who cannot sell. The bears are winning the narrative story - but they are unlikely to win the capital story.

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.russell-clark.com/subscribe

I have been out talking to people of late. One thing I hear a lot, is how attractive “real yields” are. My question is always, the same. What exactly is a “real yield”? And they will say something like the 10 year yield less inflation. In my mind, I always think, its 10 year yields minus what we think inflation MIGHT be. Its a little more sophisticated than that, but that’s the gist. If you go on Bloomberg or the Fed Website, you can get a graph of real yields. These yields are based on TIPS yield. TIPS (Treasury Inflation-Protected Securities) have been around since the 1990s, and are designed to protect your principal from inflation. We can use them to back out a “market based” real yield.

You can make a long term real yield graph. This is taking 10 year bond yields, less core CPI. Back in the 1990s, a 4% real yield was more common, but you had highly negative and highly positive yields in the 1970s. This is more volatile as inflation was more volatile.

According to this, real yields today are 2.6%, some of the highest seen. So they look like a buy. My problem is that TIPS were only introduced in 1997. So for most of the time, they have existed in a falling yield environment.

So what can you do if you think “REAL YIELDS” look attractive? Well you could buy TIPS. And there is an ETF to do just that, TIP US. Its share price does not scream “inflation hedge” - but this is without including dividend income. TIP US holds a collection on inflation linked bonds. Here is a screenshot of one.

If you are not familiar with TIPS, you may think a 1.875 coupon as pretty miserly. And it is, but the thing with TIPS is that they add the increase in CPI to your principal. So if CPI doubles in 10 years that a TIP is issued, you get 2 times your principal back. And the real yield, is calculated from TIPS. If you squint, you can see this one is priced at 95, which means your current yield is 2.5%.

The problem with TIPS is they are a bit of a Texas Hedge. Inflation would spin out of control if the government lost control of debt markets, so when you need TIPS the most, they are least likely to pay you (or fudge the CPI measure). We can get an idea of what real yields look like in a higher yield/inflation environment. There are various inflation linked bonds you can look at - and Bloomberg provides a useful table. Real yields are meaningfully higher in Mexico, South Africa and Brazil, all counties with higher inflation than the US.

Looking at Mexican CPI relative to the US, double the real yield looks about right. Since 2000, Mexican prices have risen at twice the rate of the US. This has been reflected in an exchange rate that has halved relative to the US dollar.

The problem I have with real yields and bonds in general, is that the move to a “pro-labour era” implies we remove the possibility of deflation. The interwar period, and post Cold War period saw outright deflation from time to time. Post War Period through to 1980 saw none, and the power of compounding saw inflation rise rapidly.

My guess is future inflation in the US will be much higher - perhaps close to Mexican levels. Not only would US treasury yields have to rise, but so would real yields. I can’t help but feel “real yields” are still pricing a past that is no longer with us. That is, they are not real at all.

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.russell-clark.com/subscribe

I learnt a long time ago that in financial markets, simplicity breeds elegance. Or, if the trade is complicated, it probably won’t work. For me, my super simple idea is that yields would keep surprising to the upside. Why? Because politics was moving to favour workers and wages, and this would be achieved through tariffs, industrial policy and fiscal dominance. When I thought about what this would mean, I actually thought equity markets would be very choppy. Why? Well the benefit of rising revenue growth would be offset by rising costs and rising yields. My loadstar for rising yields were 30 year JGBs. Before every deflationary cycle I have seen, JGBs have fallen before the US treasury market. And it picked up the inflationary turn in 2020 well before treasuries. That is, if I was wrong on yields, Japan would give me fair warning.

In September, the 30 year Treasury had finally broken out its range, and my expectation was that it would keep selling off. Instead it fell back substantially during October.

For what its worth, JGB 30 year yields also fell back to the 200MDA during October. So you could see that as deflationary. But it would break the trend of Japan pre-emptively moving.

Why did I think that yields would keep selling off? Well I thought dollar strength was coming to an end. If we use DXY Index, the dollar has been strengthening this year.

But Bessent wanted the dollar to weaken against the Yen, and both the Chinese Yuan and Korean Won have been very strong against the dollar in recent months. I thought this would allow Yen strength, and dollar weakness to come through. That is I would have expected falling US yields to ignite the inflation trade/debasement trade.

If that was the set up, the market action was very different. Gold, which was meant to react to positively to falling yields, responded poorly. GLD/TLT fell 11% during the month, with GLD down 6% and TLT up 5%.

If I look at equity markets, there was a clear rotation out of cyclicals and into defensives. That is the market was saying the economic upcycle was coming to an end. If this is true, the short GLD, long TLT WOULD be the correct trade, and I have had a version of this trade from 2011 to 2015.

So I am going to let you in on a secret - I would be delighted if I thought we were going to a deflation cycle. I have a proven track record in a deflation cycle, and most of the people I meet are terrified of a deflation cycle, and so in a deflation cycle, you could probably raise a large fund easily. On substack, I see the deflation trade has become popular, particularly for AI related themes. The problem I just see no confirmation. As I mentioned above, I believe simplicity breeds elegance. For the 2011 cycle, Chinese steel prices ended up being a great lead on that cycle, and in 2015/6 accurately called the turn.

As mentioned above, bond yields gave no preemptive sign of deflation trade. I also do not see it in copper prices.

Or even in DRAM prices. I would expect both to show some weakness if the AI trade really was over.

There is one way, I can see why equity and bond investors may be preparing for a deflation cycle - and that is spiking oil prices. But the current spot spread, makes future moves in the oil price very hard to predict.

The other more popular conclusion is that the AI trade is dead. I struggle with that conclusion, as the main players, the ones with clear access to customer acceptance, and the cutting edge ability of AI, keep spending. You don’t need to like Elon Musk or Mark Zuckerberg to recognise that they are both some of the best investors of their generation. Both have raised capital to put into the AI trade, and that is despite neither having a business currently threatened by AI. As mentioned in a previous post, I often wondered why everyone was just long bonds during the 1980 to 2020 bull market. I think this month is a good example of how hard it can be to hold on to a trade. Of course, I could be wrong about inflation vs deflation. Maybe all bond yields move together now, and maybe equities are correctly calling falling memory prices in the future. But when I look at the politics of the world, calling deflation here seems odd to me. And equities have a very poor record this year, both in the demise of SAAS and in Brazilian electoral trends. In both cases, betting against what equities were saying was the right trade.

This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.russell-clark.com/subscribe