This month was another large-cap, growth driven market where MSCI Japan Large/Growth outperformed MSCI Japan Small/Value by +3.2%. I would note that it was even more pronounced in the US where the outperformance of the respective indices was a massive +9.7% and the respective MSCI World spread was +8.0%! So, the Japan market’s price action seems much broader than the high concentration of the US market drivers. This was true both for the month and the quarter (Japan spread +0.9%, US spread +15.3%, World spread +11.1%). By the way, the Magnificent Seven also accounts for 23% of MSCI World with the other top 10 names being Eli Lilly and Broadcom (who probably deserves to be included in the list, but “Magnificent Eight” just doesn’t sound very cool; besides, most people probably don’t know what Broadcom does nor does it get a fancy buzzword directly associated with it). By factor, Japan was driven by Financials despite Ueda-san’s resistance to raise interest rates, or at least not at the speed the market participants want him to, and Industrials whose order book appears to have bottomed out and are looking for a second half turnaround led by a renewed tech cycle.
 
Despite these factor headwinds, we’ve managed to outperform for the last 3 months although still far from recovering our underperformance for the year. One primary reason is the beta factor (or lack thereof). During CY2023, our benchmark, despite being “small/midcap” was up +23.2%, essentially the same as TOPIX. During Q1 of this year, the benchmark was up another +13.9%. While not as concentrated as the S&P500, (relatively) larger cap tech and cyclicals led the rally during these 15 months even in our MSCI Japan SMID benchmark. For example, the Top 10 drivers [of our benchmark] were Disco (Tech), MHI (Industrials), Advantest (Tech), NEC (IT services), Lasertec (Tech), Nippon Steel, TDK (Tech), Nomura Holdings (Financials), SCREEN (Tech), and Japan Exchange (Financials) with an average market cap of 4 trillion yen, 10x our portfolio average market cap (and, yes, I find it odd that Nippon Steel and Nomura Securities is in our “small/mid cap” benchmark, but it is what it is). However, the activity of the broader market in the last 3 months has noticeably calmed with our benchmark down -0.4% for the quarter and TOPIX up only +1.5%. Note, volumes have not changed and has, in fact, risen. Volumes have not been this high since 2017, though still off its peak during the early years of Abenomics (value traded is, of course, well above). So, the interest level remains high, but I suspect (hope?) that it is becoming more selective.
 
If that is the case, it would not surprise me. As I probably mentioned a few months ago, I believe 2024 to be the year that will be remembered when Japan finally escapes disinflation and began its march toward higher wages, rising productivity, sustainable inflation, and maybe even positive real interest rates. However, I also believed that the road will be extremely bumpy with major consolidations, bankruptcies, and perhaps even a little social unrest as the income gap would likely widen even in a semi-socialist country like ours. But this is happening despite the yen at ridiculously cheap levels, global political uncertainty, geopolitical conflicts, rising US-China tensions, and an inverted yield curve in many Western markets. By the way, the Tokyo governor elections this past weekend maintained the status quo, unlike our elected officials in the federal government. This is important given that this city, whose nominal GDP would rank about #17 globally if it were a country, above the Netherlands and below Indonesia, has major implications for the country as a whole. While the global situation makes predicting events difficult in the short run, I truly believe the trend has been set. The efforts by the FSA and JPX are further helping to accelerate this change (though, as I said in 2013 at the start of Abenomics, I would caution that “accelerate” is a very relative term).
 
Back to our portfolio, the other reason for the recent outperformance is that I believe our persistence (and irritation toward some of our longer-held stocks) is finally paying off. In the last 6~9 months, we’ve spent an extraordinary amount of time doing something I had hoped, when I launched the fund, that we would avoid – aggressive activism. I’m happy to engage with management to provide our views on product cycles, marketing, pricing, strategy, and, most importantly, capital allocation and IR. And we’ve been doing so regardless of market cap or industry. But I hadn’t thought that I would need to twist any arms at times. But having done so (behind closed doors, of course), our largest position announced how they will manage their balance sheet, showing how business ROIC is different from the mathematical formula one uses via Bloomberg or Excel and RONTA (return on net tangible assets) is above 25%, that business investments will be made with those figures in mind, and that reinvestment of capital has and will continue to be highly accretive. Furthermore, they “educated” their shareholders (even some of the sell-side analysts who use simple PER multiples to calculate target prices) that accounting earnings are different from free cash flows and thus has plenty of room to raise dividends. They announced all of this in their fiscal Q3 results, an odd time to make such statements. Needless to say, the market liked these comments. Another of our larger holdings will hopefully do the same, if our continued cajoling works. And at yet another large holding, we strongly “encouraged” a revamp of their IR materials and explanations on their progress of their midterm plan given that, despite good intentions, their stock price had fallen twice after the results meeting, even when the stock first rose after the earnings results were made public. And at another company, we highlighted to senior management that their midterm plans have been meaningless because they rarely hit their own annual targets (above and below) and, therefore, the market had lost faith in their comments. Apparently, the CFO has been passing that one-pager to all levels of management from the board to business unit managers and has been adamant that they need to hit targets at all costs, regardless of market conditions.
 
Now, of course, I have no idea if these engagements are going to help the stock. What I do know is that they are all high quality businesses that are not being appreciated by the market and, in all cases, is due to weak IR and lack of financial-savvy (at least public market savviness; coincidentally, these 4 cases were all privately owned in the past). And, therefore, it is up to minority shareholders like us to engage with companies to help lift the stock whether it be through education or better balance sheet management or, as is usually the case, both.
 
Given the rising uncertainty of the markets, I don’t know if this alpha will be enough to break through any market-wide headwinds that we may face. But I do know that we have no intention of standing idly by and let them bring down the value of our companies. I hope we can continue this momentum for the remainder of the year (and then some).

​

Masaki Gotoh

​​

===

​​​

“Words of the jargon sound as if they said something higher than what they mean.” – from The Jargon of Authenticity, Theodor Adorno, German philosopher.
 
“Our business is infested with idiots who try to impress by using pretentious jargon.” – David Ogilvy, the “Father of Advertising”.
 
My weekly weekend project continues as I clear out my house of endless “stuff” to make room for Leo’s own bedroom. I’ve never been very good at throwing things away. I always think, “I might want to reread this book” or “I’ll join a gym so I can fit into this T-shirt from college” or “Maybe I’ll buy something that’ll need this cable”. I’d keep collecting so many items so much that, until I moved to my current house 15 years ago, my rented storage space was always larger than my living space. That may be another reason why I always lived in the suburbs as I needed to be close to storage and I couldn’t afford both an apartment and storage space unless I commuted an hour away by train unlike most of my friends who were close enough to work to take a taxi every day. But my son has helped me “DanShaRi”, a trademarked term meant to mean “decluttering” the home through “dan” (refuse), “sha” (dispose), and “ri” (separate). Ms. Hideko Yamashita, a “clutter consultant” who coined the term, lectures around the world to teach its philosophy and the differences vs minimalism … or something to that effect. I hear she has quite a global following; I’m just trying to find space for my son.
 
And so, every weekend, my garage fills up with bags of garbage. While it depends on local municipality, there are different days of garbage collection. For example, in my neighborhood, they collect combustibles twice a week, non-combustibles twice a month, PET bottles twice a month, and other recyclables (glass bottles, cans, and paper products all in separate bins) once a week. Large-sized items (over 30 cm) are collected for a fee by reservation only (although recently, I’ve found that if I sneak some large electronics goods at night, someone steals them before the non-combustible garbage collection arrives, probably to sell on Mercari, a popular C2C site, or for the parts, notably the copper). Now, this garbage collection system might sound complicated, but Tokyo has modern garbage incinerators, so the definition of “combustibles” is quite wide. Where my parents live, for example, there are very specific definitions of types of trash and differing collection days for food scraps, combustibles, non-combustibles, recyclables (each with a different date), used linens and clothing, ignitable items, plastic containers, and natural trash like branches and leaves. Each have designated, pre-paid bags for collection whereas I can put my trash in any translucent bag. But because what can be taken out is different by day, I must keep the rest of the garbage in my garage until the day of collection.
 
But among all the things I sort through to determine what stays and what goes, the books take the longest time. I actually don’t want to throw any of them away and, to be honest, very few make it to the OUT pile to my wife’s annoyance (although I would say the same about her clothes). As I was going through my endless stack of books, I came across my old college yearbook. And between the pages, I found my syllabus for my 4 years of college.
 
As you probably know, I was a Computer Science major. I started Cornell at the age of 15, in retrospect much too young to attend college alone. And so, I’d waste time messing around with little studying and spent most of my time at Ithaca College nearby which was a liberal arts school with students who were much more fun than the Engineering campus at Cornell. In 1990, the final year of the Japanese bubble, I received an employment offer from one of the largest advertising companies in Japan with still over 2 years left in school. Yes, it was a crazy time. Inevitably, I’d start to ditch class, leading to a sub-2.0 GPA during my second semester as a sophomore. I received a letter from the school that one more semester below the minimum would get me expelled. My job offer was, obviously, contingent on graduation so I started to take school more seriously. During my junior year, my offer was unexpectedly pulled as the company entered a hiring freeze and, with no job offer and dismal grades, I doubled my efforts at school. During my senior year, I started taking graduate level courses, which is where I met my future employer, Hitachi, who was sending professionals to study for their Master of Engineering degree. And with that, despite my mediocre 4-year average GPA, I completed my senior year with honors for the final semester. During that semester, I took graduate-level electives in Text Processing and Retrieval (a forerunner to Natural Language Processing), Distributed Operating Systems and Parallel Networks (the origins of Neural Networks), and Computer Graphics.
 
While I didn’t take any machine learning classes, I knew the concept behind it. But machine learning is simply part of the algorithm where individual processes (or neurons) make predictions regarding a subset of the point being learned in parallel with other neurons doing the same with other subsets. The system “learns” by being provided a solution and backpropagation is used as a feedback mechanism to adjust for any errors that those predictions by each neuron made, thus making the solution better the next time around. My three electives are essentially the building blocks of what people call AI today (I’m stretching a bit with Computer Graphics but, as we all know, NVIDIA was (is) a GPU company). Back in 1992, we had to connect a classroom of UNIX workstations to work in parallel to find a solution to the problem (or, rather, we learned to develop the operating system to do so). But now, NVIDIA can do it on a single chip in some insane order of magnitude faster, I presume. But the concept is the same. The machine doesn’t actually “know” anything but is just processing data based on pattern recognition reinforced through the feedback loop running in the background to generate a result. It doesn’t actually understand what it’s doing. It is, in our parlance, using past performance to estimate future results. It just does lots and lots of it really, really fast (I wonder what disclaimer they will put on AI-based investing strategies …)
 
And so, it amuses me that, wherever I look, there is AI. I can’t remember how many times I saw “AI” in a presentation this past fiscal year end. If I do a Bloomberg search among listed Japanese equities for “Artificial Intelligence”, I get over 500 hits (Bloomberg limits the sorting process to 500; note that there are over 4000 listed companies in Japan so at least 10% portends to do some sort of business in AI). As an example, here’s the first one in the list: Trial Holdings, listed just this year. “The Company provides foods, daily necessities, hard goods, and other products. Trial Holdings also develops retail artificial intelligence, real estate development, restaurant management, and other businesses.” As of last fiscal year, 99.7% of their business is from their discount stores and 0.1% of sales comes from “Retail AI” (while the other businesses, collectively 0.2%, fall into “other”, so clearly this 0.1% is special). It says, “we provide various solutions utilizing IoT devices and software services.” (By the way, “IoT” and “DX” are 2 other buzzwords in Japan that used to be on every presentation before “AI” became the thing). Clicking through their website, this retail AI solution looks like a shopping cart connected with a tablet. I’m not sure where the “AI” part comes in, although it does provide personalized coupons on recommended items based on, I’m guessing, your historical shopping patterns, not too different from what Amazon has been doing since, at least, 2000 when I placed my first order on Amazon, nor is it that different from what POS registers used to do before the Internet.
 
Another company I’d mention is GMO Research & AI. They are a small competitor to one of our holdings in online marketing research. They used to be called GMO Research but just recently tacked on the “& AI”. They even changed their URL to “gmo-research.ai” (their parent, GMO Internet Group, started as a domain provider). Note, they haven’t actually started any AI service (at least, their website doesn’t highlight any new product based on AI). I’m sure, like our company, that they use a lot of machine learning to analyze past surveys to provide questionnaire templates. But, reminiscent of the dot-com bubble, the stock opened +3.5% on the day after they announced that they would be changing their corporate name. It reminds me of companies like “furniture.com” and “pets.com”.
 
The concept of “AI” isn’t all that revolutionary. The fact that the hardware has caught up to the concept IS (and why Nvidia deserves quite a lot of respect for that). But AI itself is just enhanced probability and statistics. Similarly, “IoT” is simply anything other than your PC or phone that is connected to the Internet to gather and analyze some information. And “DX” is … well, I’m still not sure what that is. Sounds like “software using different types of data to analyze stuff”. But to think we are on the verge of singularity sounds a little like hogwash to me – a little too much Matrix and Terminator. 
 
Now don’t get me wrong, I am thankful for IoT and AI (and DX, I suppose). I can turn on my A/C a few hours before I get home so I don’t have to walk into a sauna. I don’t have to worry about my boys losing their keys since I installed a cheap card key system on my traditional door handle key that I can disable remotely if they drop it. I have Google cams around my house so I can see when my boys came home, and it notifies me if someone other than my family is near the house through face recognition. All of this became available in the last few years at extremely reasonable prices.
 
But AI hasn’t learned enough to think for me on my behalf. As an example, Bloomberg recently limited the number of daily research reports that I can download. I complained as to why and they told me that many funds are downloading as much research as possible to scrape the data. Therefore, the report providers (the sell side brokerage houses) asked to place a limit on daily downloads. I’m assuming that AI is trying to learn how specific stocks react to certain phrases that are written by sell-side analysts. And I have no doubt that it will figure it out such that they can send a buy/sell order instantaneously while old-fashioned people like me actually download the report (if I notice it was released in the first place), read it, think about it, and then, at least in our case, do nothing … except maybe to call the analyst and/or the company if it’s really important. Most times, we’d probably ignore it or, at most, mention it to the company when we see them a month later. I’m fairly certain that, whatever the stock is doing by then, it’d have little to do with the report, if at all. But when Bloomberg asked me why I would need to download more than 300 reports, I said that when I research a new company, I would download about 10 years’ worth of research and glance through most of them. They said they hadn’t heard of users complaining about this limit except quants funds. I guess they found it quaint that someone would actually pull reports to read it rather than to scrape it.
 
We, as part of our daily lives, do a lot of monotonous tasks, most of which can be automated. Some of these tasks require a little decision-making. AI can possibly do some of that for me as well. AI will, undoubtedly, help augment my life to make it much better, both professionally and personally. I will be able to make better informed decisions. But I can’t imagine that those decisions will be made for me. And particularly in our strategy where daily stock prices and reactions to earnings reports or news items is often times taken as noise, I’m not convinced that AI can figure out what drives long-term returns. If it does, will it learn to ignore short-term information? If learning is about taking information and triangulating to “true” information, can AI learn what to ignore and what to take as actual input, particularly when the target time horizon is undefined? And with markets changing constantly as long as the majority of participants are humans, will it learn that what was ignored before may be of utmost importance now? And what about the people aspect? Can it figure out that the new management team is good or bad? How is it going to obtain that information? Will it provide a questionnaire to a human to ask (or perhaps it asks itself)? In fact, will AI replace management teams? If all of this became true and singularity were to occur, there would be no need for public markets since, by definition, public markets are a mass of contradicting views which could not happen if we achieve singularity. In which case, our entire industry will be pointless. Taken to the extreme, how could any asset provide anything more than a fixed return which would, by definition, become the risk-free rate?
 
I respect AI very much. But it is, like so many buzzwords, used mistakenly very frequently. And so, one should take caution on assuming everything AI is awesome. To think that it can replace human judgement seems a little far-fetched.
 
The first thing AI needs to figure out is tomorrow’s weather. That would be incredible progress.

​​​​