Reading The Trillion Dollar Baby

• Norway becomes an independent nation in 1905
• In 1909 - After three years of political turmoil, parliament passes the final version of the ‘concession law, which ensures that Norway’s natural resources, principally hydro, would revert to the state when developed by foreign investors.
• In 1973 - Arab oil producers had formed the cartel known as the Organization of Petroleum Exporting Countries (OPEC), which slashed production and triggered a quadrupling of oil prices, thus ending a halcyon era of abundant fossil fuel.
• Norway population in 1974 - 3.85 million people
• 1974 - Lots of oilfields around Norway being discovered since 1969.  Firm and wise stewardship of this endowment was called for.

In the five years since oil was discovered in 1969, Norway had learned a great deal about the powerful oil industry and it was now ready to extract the maximum possible share of revenue. It had set up a national oil company and given it a mission to generate direct ownership and profits for the state, and to drive the involvement of locals in the development of the industry. It had created a policy that allowed the state to take a direct equity stake in oilfields should they prove to be very profitable, without having to invest in the exploration costs. It was what might be known as a cake-andeat- it policy. But on the revenue front, the existing tax regime still meant that any super profits would be collected by the companies, not the government of Norway.

So in 1974 - finance officials explained that companies would now pay a special tax of 40 per cent on top of Norway’s already high corporate tax rate of 50 per cent. 

Despite Norway’s boldness, Big Oil kept its rigs in Norway’s part of the North Sea, known as the  Norwegian Continental Shelf (NCS), because the region remained an attractive place to invest and because it could gain no support from Norway's conservative party. Even though conservatives are traditionally great advocates of lower taxes, Norway’s conservatives agreed with the higher rate and refrained from political opportunism. This is remarkably different to what has transpired in many other countries, where oil companies have been able to win support by dividing and conquering the political class. Even to this day, both sides of Norwegian politics remain firmly in favour of the tax regime.

Unlike other resource-rich countries, Norway was showing how to become the master rather than the servant of resource multinationals. That November 1974 meeting was a decisive step along the way towards reaping the maximum national benefit from the country’s resource wealth.

As a society, Norwegians have remained true to the sentiment put to visiting oil executives in the late 1960s of wanting to eschew the euphoria of oil riches. This record is all the more remarkable because many other countries, and not just poor ones, have blown their boom-time windfalls on largesse, whether in the form of grand edifices, military hardware, white elephant ee or cash splashes for swinging voters.

• Oil has been a problem for middle and low income countries. But other developed countries have also not managed it very well. Examples of UK bailout. US, Canada, Australia - can't be said to have astutely managed it.
• Norway is alone in the developed world in having successfully kept most of the windfall from its oil and gas wealth — 90 per cent of cash flow now accrues to the Norwegian state.* And only Norway has set up sound institutions to manage this boom-time bonus for generations to come
• Written in 2016/17:  At the start of its oil era, Norway was a middle-income country with a per capita GDP just ahead of Greece, and by the end of the 1970s it was heavily indebted to the rest of the world. Its net foreign liabilities as a share of GDP peaked at 40 per cent. But its effective taxation and disciplined savings strategy have since turned it from a net debtor into one of the world’s biggest creditors. Its net foreign assets are worth 185 per cent of GDP, or around $760 billion, and they’ve risen even more sharply when converted back into local currency as a result of Norway’s flexible exchange rate (one advantage of not being part _ of the euro). During a twenty-year period of relatively high oil prices, Norway salted away $870 billion in a long-term sovereign wealth fund. Despite lower oil prices, the fund is still growing and is on tratoc hikt $1 trilliin o20n20 .
• Other developed countries have not fared thus. Even after resource boom, they have ended up with increased debt numbers.  
• Norway offers the best example in the world today of how to govern extraction of non-renewable resources, but it is by no means perfect. The country made mistakes in the early years by allowing development to run ahead of safety, leading to the loss of many young lives in the 1970s and 1980s. The focus on onshore development involved high risks that pushed the technological frontier and led to massive cost blow-outs. The country is now licensing new development in the Arctic region, which is questionable on both technical and moral grounds. Norway can also be criticised for relying too heavily on oil production, which has meant that about one-quarter of its national income has been directly or indirectly tied to a single commodity. But there’s a great deal that can be learned from the Norwegian experience. Its long-term strategy and firm management,
• Norway is the small country that’s become the big exception to the resource curse thesis.

--

In the second chapter, the book dives into the history of Norway. How for 400 years it was either under Danish or Swedish rule before 1905 independence. How it was poor (Munch as an artist from poor Norway). How it began ship building, to diversify its basic economy. Then 1940s, Germany, and the resistance movement. Hauge. Defence Capability. Post war, 18 years of labour rule. The desire to save for future. Until the 1970s when the discovery of oil. 

Instead of unleashing an oil boom, Norway adopted a policy of controlled development and moderation so that the country could develop private-sector capacity and government institutions to benefit fully from its new-found wealth. As Hauge and his team of confidants saw it, Norway needed to do much more than produce another commodity — the oil should be used to help the country to industrialise and generate lasting returns. This was the genesis of a strategy that would handsomely reward Norway for decades to come.

--

Leonid Vitaliyevich Kantorovich

At the end of the 1930s Kantorovich was faced by a concrete planning problem – how to combine the available productive resources in factory in such a way that production was maximized. He solved this problem by inventing a new type of analysis, later called linear programming. This is a technique for finding the maximum value of a linear function under constraints consisting of linear inequalities. A characteristic feature of this technique is that the calculations give as by-products some expressions, called shadow prices, which possess certain qualities that make them useful as accounting prices.

• Speech given in 1975 Nobel lecture
• optimization models and their use in the control of the economy for the purpose of the best use of resources for gaining the best results.
• October Revolution. Then for the first time in history all main means of production passed into the possession of the people and there arose the need for the centralized and unified control of the economy of the vast country.
• The specific and peculiar problems
• The main purpose of economic theory changed from observation to control. Needed detailed planning. Since planning at such scale appeared for the first time, the solution needed to be found afresh
• Not just for the national economy but for enterprises and projects as well. Goals and Objectives. Precise accounting methods needed
• And then, the thing about indices. The way price, rent, interest exists in capitalist economy, in a centrally planned economy, its significance and existence itself gets questioned
• Together with material flows and funds in capitalist economies there are also studied and directly observed such important economic indices as prices, rents, interest rate in their static and dynamic properties. The indices mentioned serve as the background for all economic calculations, for aggregation, for the construction of the synthetic indices. It became clear that a, consistently planned economy cannot do without indices characterizing the analogous aspects. They could not be observed here and were given as normatives. The problem of their calculation was however not restricted only by technical aspects of calculation and statistics. It is important that in the new conditions similar indices received a quite different sense and significance, and some problems as to their nature, role and structure arose. For example, it was unclear and open to discussion whether a land rent should exist in a society where land is in the possession of the people or whether such an index as the interest rate has a right to exist.
• Centralization upto a level. Some control left for lower levels.
• The problem is to construct a system of information, accounting, economic indices and stimuli which permit local decision-making organs to valuate the advantage of their decisions from the point of view of the whole economy. In other words to make profitable for them the decisions profitable for the system, give a possibility to check the validity of the work of local organs activity also from the point of view of the whole economy.
• Question of the most efficient structural form of the control organisation.
• The problem of prediction and control in large shifts of weights of different branches of economy,  of rapid changes in tech development.
• I mean the problems of prediction and control in conditions of large shifts in the weights of different branches, of the rapid changes in production and technology, national economy. The problems of estimating technical innovations and the general effect of technical progress. The problems of ecology connected with the deep changes of the natural environment under the influence of human activity, the prospects of exhausting the natural resources. The prediction of social changes and their influence on the economy. The changes in presence of contemporary computational technique, means of communication, managerial devices and so on.
• These problems exist in capitalist countries too, but the control problem was different level and peculiar. Since nothing existed at scale of this nature, "The economic theory of Karl Marx became the methodological background of the new created Soviet economic science and of the new control system."
• I just wish to point out that the system of planning organs was created on the initiative of the founder of our state V. Lenin and simultaneously on the same initiative a system of economic accounting (hozraschet) was introduced which gave a certain financial form of balance and control of separate economic activities.
• Although it worked well, there were inefficiencies/ possibilities of further improvements hence qualitative mathematical methods
• The new methods: The optimizing approach is here a matter of prime importance. The treatment of the economy as a single system, to be controlled toward a consistent goal, allowed the efficient systematization of enormous information material, its deep analysis for valid decision-making. It is interesting that many inferences remain valid even in cases when this consistent goal could not be formulated, either for the reason that it was not quite clear or for the reason that it was made up of multiple goals, each of which to be taken into account.  For the present the multi-product linear optimizing models seems to be mostly used. I suppose that now it is spread in economic science not less than for instance Lagrange equations of motion in mechanics.
• It is well-known that the choice of optimal program i.e. of the set of intensities of these activities under some resource and plan restriction gives us a problem to maximize a linear function of many variables satisfying some linear restrictions.
• Properties which allow for wide and varied use:
• Universality and flexibility - For many branches of economy and levels of control. Additionally, in complex situations, becomes a take-off point for generalizations. It is possible to consider a series of models where necessary conditions and restrictions are introduced step-by-step while the needed descriptive precision is not reached.
• Simplicity - simple linear algebra
• Efficient Computability. Simplex method. Algorithms for computers - 1975 - nowadays modern variants of the methods on modern computers can rapidly resolve problems with hundreds and thousands of restrictions, with tens and hundreds of thousands of variables
• Qualitative analysis, indices. Together with the optimal planning solution the model gives valuable devices of qualitative analysis of concrete tasks and of the whole problem. This possibility is given by a system of indices for activities and limiting factors which is found simultaneously with the optimal solution and is in accordance with it.
• Concordance of the means with the problems - Both in capitalist economies and socialist economies. He feels the spirit corresponds closer to the problem of socialist economies. (I though wonder that efficiency and maximisation is a very capitalist objective as well)
• A problem that needs to be pointed out especially is that of decentralized decisions. The investigation of a two-level model complex leads us to the conclusion that in principle the decentralization of decisions with observance of the total object of the complex is possible by the means of a correct construction of objects in local models.
• Difficulties  Difficulties arise both from the specific features of the object under investigation and from defects in the researches and their practical realization.
• It is especially important to test the influence of the difference between the model and reality on the obtained result and to correct the result or the model itself. This part of work is not often observed.
• Complication in the correct choice of the objective function.
• A solution of newly appearing economic problems, and in particular those connected with the scientific-technical revolution often cannot be based on existing methods but needs new ideas and approaches. Such one is the problem of the protection of nature. 
• Accounting method is just one part. Control strucutre as well.
• Education of people. psychological barrier, to shift from many-years-used routines to new ones.
• 
  

Notes from Space X IPO document (1)

• Space Exploration Technologies Corp.
• Filed May 20, 2026
• 1 Rocket RoadStarbase, Texas 78521
• CEO - Elon Musk
• Elon Musk, our founder, Chief Executive Officer, Chief Technical Officer and Chairman of our board
• 2 classes of shares A and B. A offered. B 10 times voting rights.
• Controlled Company (hence). Seeks exemption from certain corp governance requirements
• Lots of joint book running managers - GSach, MS, BoFA, Citi, JPM, and many others
• About SpaceX (CY2025 - ~$19 billion revenue) 
• Space
• Revenue FY25 - $4.1 billion. Loss making. Capex Fy25 - $3.8 billion
• 2002
• % of 2025 mass to orbit: 80%+
• 78 crewmembers flown
• Since 2023, we have launched more than 80% of mass to orbit for the world each year with an over 99% mission success rate with Falcon rockets
• extreme vertical integration
• Space flight that historically cost billions per launch now costs in the tens of millions, fundamentally reducing the cost of space access and providing the opportunity to build new enterprises in space
• Connectivity
• Revenue FY25 - $11.4 billion. Profitable.Capex Fy25 - $4.1 billion
• 2020
• 9600+ starlink broadband and mobile satellites
• 10.3 million starlink subscribers
• 164 countries
•  a high-speed, low-latency global broadband data and communications network
• 650 V1 Mobile satellites provides satellite-to-mobile data, (7.4 million monthly unique devices across 30 countries)
• AI
• Revenue FY25 (CY) - $3.2 billion. Loss making.Capex Fy25 - $12.7 billion
• 2023 
• 1GW+ Nameplate compute draw
• First coherent Gigawatt scale AI training cluster
• 4 major model versions 
• ~550 M+ monthly active users (1.3 billion supported accounts)
• 350 million daily posts
• xAI, which was founded in 2023 and acquired by SpaceX in early 2026, is now an integral pillar of our vertically integrated company.
• Our infrastructure supports training and inference for Grok, which has emerged as one of the world’s most advanced frontier models. Grok is designed as a truth-seeking AI model, built on our founder Elon Musk’s mission to enable humanity to understand the universe
• deep integration with X
• Some terms from its glossary:
• AI” or “artificial intelligence” refers to advanced computational technologies and systems enabling machines to learn, comprehend reality, solve complex problems, exhibit creativity, make critical decisions, and function with growing autonomy.
• “The Algorithm” refers to our five-step iterative process that we use to rapidly innovate and optimize, emphasizing making the requirements less dumb, deleting unnecessary processes or parts, optimizing the necessary processes or parts, accelerating cycle timesteps, and automating only proven processes after the first four steps are completed
• “Kardashev Type II” refers to a civilization that harnesses the full energy output of its local star, like our Sun, to power unprecedented growth and sustain the civilization’s existence.
• “mass to orbit” refers to the total kilograms of payload deployed to orbit in a given period, and is a key indicator of our capacity and scalability that supports Space revenue and drives expansion across our Connectivity and AI segments.
• GPQA Diamond score, an industry benchmark that evaluates AI models on a standardized set of questions written and validated by experts,
• Strategy
• Specifically, we believe SpaceX’s reusable rockets, scaled satellite manufacturing, and operational expertise can enable the cost-effective and rapid deployment of massive AI compute satellite constellations—with potentially millions of satellites—for orbital data centers. We believe these AI compute satellites in Sun-synchronous orbit will be able to handle energy-intensive AI workloads, such as inference demand, at far greater scale and efficiency than terrestrial alternatives, with Starlink providing low-latency, global connectivity linking these orbital AI systems to people around the world and delivering real-time intelligence. We expect to begin deploying our orbital AI compute satellites as early as 2028
•  Space-based solar arrays can generate more than five times the energy per unit area of terrestrial solar due to continuous illumination, lack of atmospheric interference, and optimal orientation. 
• P&L 
• In 2025, we generated revenue on a consolidated basis of $18,674 million, loss from operations of $(2,589) million and Adjusted EBITDA of $6,584 million.

ETFs

• Launched in 1993
• AUM- As of early 2026, USD 19.85 trillion (US listed ETFs 10.5 trillion). 
• Compared to Mutual Funds?
• As a % of market - the total global market cap is $x trillion. So 

"Exchange-traded funds (ETFs) debuted in 1993, marking a major shift in how investors could access the markets. Originally making up less than 1% of all fund trading, they've now grown into a powerhouse, representing about 26% to 30% of daily U.S. trading volume"

History - around 3 decades.

 

Older chart - with old data but good to see the global landscape.

Appendix - Chat with Claude to understand some things

I am exploring ETFs - history, impact etc. One of the things that stands out is their growth in the last three decades. Perhaps it has reshaped the equity market participation, perhaps it still does. One of the things that I wonder about how do passive mutual funds still maintain their edge in this world? Wouldn't it converge more to ETFs eventually? What are the differences between the two? The second thought is more about the philosophical impact of ETFs on equity participation. Earlier, people built portfolio by taking calls on companies. Now perhaps people build portfolios taking calls on themes?

09:29

This is a rich question worth unpacking in two distinct threads, as you've correctly identified. Let me work through them both.

---

ETFs vs. Passive Mutual Funds — Why Both Exist, and Why Convergence Is Incomplete

The structural differences are more durable than they might appear. They're not the same product in different packaging.

How they differ mechanically. A passive mutual fund (think Vanguard's index funds) is priced once daily at NAV — you transact with the fund company directly at end-of-day. An ETF trades on an exchange continuously, like a stock, with a bid-ask spread and an intraday market price that can deviate slightly from the underlying NAV. The ETF uses an "authorised participant" (AP) mechanism — large banks that can create or redeem large blocks of ETF shares by delivering or receiving the underlying basket of stocks — to keep that deviation minimal. This creation/redemption mechanism is structurally elegant and is actually one reason ETFs can be more tax-efficient than mutual funds in US structures (capital gains are flushed through in-kind redemptions rather than taxed within the fund).

Why passive mutual funds haven't been displaced. A few structural reasons:

First, 401(k) and superannuation-equivalent plan architecture. Many employer-sponsored retirement plans in the US (and similar structures globally) are set up to transact in mutual fund units, not ETF shares. Changing the plumbing is slow and expensive. Vanguard's index mutual funds still hold enormous AUM simply because that's how the institutional 401(k) wrapper was built.

Second, fractional investment without a brokerage. Mutual funds accept any dollar amount — you invest $500 and get fractional units at NAV. ETFs traditionally required whole-share purchases (though fractional ETF share trading has grown). For automatic monthly investment plans (think direct debit from payroll), the mutual fund wrapper is cleaner.

Third, cost competition has largely converged anyway. Vanguard's S&P 500 index fund and its ETF equivalent now both sit at ~3 basis points. The fee gap that once made ETFs compelling has largely closed in the passive space. The competitive advantage of ETFs is sharper for active and thematic strategies — not the vanilla index.

Will they converge further? Probably yes, at the margin. The SEC has allowed ETF-of-mutual-fund structures, and there are active ETFs now that approximate the old mutual fund model with intraday liquidity. But complete convergence is unlikely because the distribution infrastructure — retirement plans, adviser platforms, direct-to-consumer savings products — was built around the mutual fund unit and re-plumbing it has costs that exceed the marginal benefit for most participants.

---

The Philosophical Question — From Stock-Picking to Theme-Picking

This is the more interesting question, and I think you've identified something genuinely significant in the architecture of contemporary investing.

The original mental model. When Graham wrote The Intelligent Investor (1949), the implicit model was: analyse a company, form a view on intrinsic value, build a position. The portfolio was an aggregation of discrete judgements on discrete entities. You were betting on companies.

What ETFs changed. ETFs didn't just lower costs — they shifted the unit of bet. The first generation of index ETFs (SPY launched in 1993) replaced stock selection with market selection: you're no longer asking "is Microsoft undervalued?" but "do I want US large-cap exposure?" The second generation — sector ETFs — refined this to "do I want technology exposure?" The third generation is where the philosophical shift becomes acute: thematic ETFs let you bet on narratives. "Clean energy transition." "Artificial intelligence." "Longevity." "Deglobalisation."

This is a meaningful change in how equity participation works. You are now buying a story, a structural trend, a thesis about the shape of the future — and the index provider translates that thesis into a basket of companies on your behalf. The investor's analytical work shifts from "what is this company worth?" to "do I believe in this theme, and has this ETF captured it well?"

Some second-order effects worth thinking about:

Price discovery becomes noisier at the stock level. If capital flows into an AI ETF, every company in that basket gets bought proportionally to its weight — regardless of individual merit. The marginal investor in Nvidia via a thematic ETF has done no analysis of Nvidia. At scale, this could push valuations in popular themes above fundamentals — and potentially create crowded reversals when sentiment shifts.

The index provider becomes an enormous power in the market. MSCI and S&P Dow Jones decide which companies are "clean energy" and which aren't. When they include or exclude a stock from a major index, billions of dollars respond automatically. The index provider has become a gatekeeper without really being recognised as one.

Sector rotation has accelerated. It's now trivially cheap to move from "financials" to "healthcare" to "commodities" via ETF trades. This may have increased the volatility of sector-level factor moves even as stock-specific volatility has been diversified away.

But individual stock-pickers haven't disappeared — they've become the price-setters for everyone else. This is Grossman-Stiglitz's famous paradox applied to the ETF era: if no one analyses stocks, prices become inefficient, which creates an opportunity for someone to analyse stocks. The passive investors free-ride on the price-setting work done by the active investors who remain. There's an equilibrium somewhere — we don't know where it is — where the market is "just inefficient enough" to sustain the analysts needed to keep it approximately efficient.

For family offices and sophisticated allocators specifically. The thematic ETF world creates a new due diligence problem: not just "is this theme real?" but "has this index captured the theme accurately?" Many thematic ETFs are surprisingly poor representations of their stated thesis — the "clean energy" ETF that has significant fossil fuel exposure through diversified utilities; the "AI" ETF that's 30% Nvidia and then a long tail of loosely related companies. The sophistication has moved from stock analysis to index construction analysis.

The deeper question you're raising — whether the move from companies to themes represents a regression in the quality of capital allocation — is genuinely open. It democratises participation (anyone can bet on deglobalisation now), but it also collectivises the analytical process in a way that may distort prices over medium-term cycles in ways we haven't fully observed yet, because the thematic ETF era is still quite young.

To your last point, hence, in a portfolio, where does an ETF sit? Alongside companies, if someone wants a thematic exposure? Within the equity bucket?

09:35

Yes, and the placement is actually more nuanced than it first appears. Let me work through how a thoughtful allocator would think about this.

The straightforward answer first. A broad market ETF — S&P 500, MSCI World, ASX 200 — sits squarely within the equity bucket. It is equity. You're buying the same cash flows, the same economic exposure, just packaged differently. No portfolio construction debate there.

Where it gets interesting is with thematic ETFs. Here the question isn't "which bucket?" but "what am I actually buying, and does it complement or duplicate what I already hold?"

Consider a family office that owns direct positions in Macquarie, NextEra, and a handful of other infrastructure-adjacent businesses, and then adds a "clean energy transition" ETF alongside them. There's likely meaningful overlap — the ETF basket probably holds some of the same names, or companies whose returns are driven by the same macro variables. The thematic ETF isn't adding a new exposure; it's concentrating an existing one, often less efficiently (because the ETF has internal costs and index construction compromises).

So the more useful framework isn't where does it sit but what marginal exposure does it add?

---

Three distinct roles a thematic ETF can play in a portfolio:

Pure completion. You have a view on a theme — say, Indian domestic consumption — but you don't have the capacity or jurisdiction to pick individual Indian mid-cap companies. The ETF delivers the exposure efficiently where direct ownership is impractical. This is the strongest use case. The ETF is the only viable vehicle; it sits in the equity bucket alongside your direct holdings as a geographic or sector completion tool.

A placeholder pending conviction. You want semiconductor exposure but you haven't done the work to choose between TSMC, ASML, and Applied Materials. The ETF holds all of them proportionally while you build that view. This is a legitimate transitional use — but it should have a defined lifespan. Leaving it as a permanent holding means you're paying (even if modestly) for someone else's index construction decisions indefinitely.

A deliberate diversifier within equities. If your direct equity book is dominated by Australian financials and resources — which a home-biased family office often is — a global technology ETF or a healthcare ETF is genuinely adding factor diversification within the equity sleeve. The thematic ETF is doing the work of reducing the idiosyncratic concentration of the direct book. This is intellectually clean.

---

Where things get murkier. Some thematic ETFs blur asset class lines. A "real assets" ETF might blend listed REITs, infrastructure stocks, and commodity producers — things that belong in three different allocation buckets in a traditional framework. An "alternative income" ETF might hold a mix of preferred shares, REITs, and senior loans. If you drop these into the equity bucket because they're listed, you're likely misunderstanding the risk you're carrying — particularly their behaviour in an equity drawdown.

The practical discipline is to decompose the ETF's holdings and ask: what would I call these individually? If the answer is "mostly equity with some real asset flavour," it belongs in equity but you should note the bleed. If the answer is "this is really a bond proxy," it doesn't belong in the equity bucket at all, regardless of how it's packaged.

---

For a family office specifically, the cleanest portfolio architecture treats ETFs as tools, not categories. The equity sleeve has a target allocation — say 40% of the portfolio. Within that sleeve, you might have direct equities (Australian and international), one or two broad passive ETFs for efficient market exposure (US large-cap, MSCI Emerging Markets), and perhaps a thematic ETF or two where direct construction isn't practical. The ETFs don't occupy a separate bucket; they're just delivery mechanisms for equity beta and factor exposures that you've decided you want.

The discipline is periodic review: is this ETF still doing the job I hired it for, or has it drifted (through index rebalancing) into something I no longer recognise?

Appendix 2 - Ownership of companies

On the regional split, in the United States institutions hold 69% of listed equity — their dominance is largely what drives the global 47% figure. That share falls to 20% in Asia excluding China and Japan, 9% in China, and 20% in Latin America. Corporate cross-holdings are significant in Asia (24% of listed equity in Asia ex-China/Japan), while the public sector dominates in China at over one-third of market cap. Strategic individuals are large owners in Asia, China, and Latin America. OECD

On where ETFs and mutual funds nest within this. The OECD data doesn't disaggregate the institutional 47% further — that bucket contains everything from Vanguard index funds to Norwegian sovereign wealth to Japanese life insurance companies to university endowments. But we can reason through it with what we know:

The 47% institutional bucket, decomposed roughly, contains equity mutual funds (estimate ~18–22% of global equity market cap, derived from $27–33T mutual fund equity AUM against $152T market cap), equity ETFs (~8–9% as we established), and the remainder — pension funds, insurance companies, sovereign wealth funds, hedge funds, and other institutional direct holdings — probably another 17–21%.

The "other free float" 27% is not pure retail. This is an important methodological note: when flow-of-funds data aggregates household sector holdings, it includes a lot that is not individual mom-and-pop investors — smaller funds, unattributed positions, and miscellaneous intermediaries all get swept into that category. ICI True retail direct stock-picking is probably closer to 10–15% of global equity, concentrated heavily in the US, Australia, and parts of East Asia.

The thing that stands out when you look at the regional data as an allocator: the "institutional" framing of equity markets is almost entirely a developed-market phenomenon. In China, the state and strategic families own the majority. In Latin America and Southeast Asia, corporate group structures and founder families still dominate. The liquid, institutionally-held, analyst-covered equity market that we tend to think of as "the market" is largely a US and OECD construct — which is exactly why passive indexing and ETFs work best there, and why the efficiency arguments for active management are so much stronger outside those markets.

Appendix 3 - The institutional ownership

What does it mean philosophically that in the US, close to two/thirds of ownership is institutional? Does it change the way decision making happens in companies?

09:56

This is one of the more consequential and underappreciated structural shifts in capitalism over the last fifty years. Let me think through it carefully.

---

The original theory of the public corporation

When Berle and Means wrote The Modern Corporation and Private Property in 1932, they identified what they called the separation of ownership and control as the defining feature of the modern firm. Shareholders own, managers control — and because shareholders are dispersed and disorganised, managers effectively run companies in their own interest. The principal-agent problem was born.

The implicit assumption was that shareholders were individuals — thousands of them, atomised, unable to coordinate, rationally ignorant about any particular company because their stake was too small to justify the cost of attention. This assumption held reasonably well for decades.

Institutional ownership changes the premise entirely. When two-thirds of US equity is held by institutions, you no longer have dispersed, disorganised shareholders. You have a relatively small number of highly concentrated, highly informed, professionally managed entities sitting on the other side of the table from corporate management. The principal is no longer weak.

---

The "Big Three" concentration problem

The institutional 69% isn't spread evenly. A meaningful fraction of it flows through a very small number of hands. BlackRock, Vanguard, and State Street — the passive index giants — are the largest single shareholders in most S&P 500 companies simultaneously. They own approximately 20–25% of the average S&P 500 company collectively, often as the single largest shareholder bloc.

This creates a structural novelty that has no historical precedent: a small number of permanent, universal shareholders who cannot sell. Because index funds must hold every stock in the index, they cannot exit a poorly-governed company. They cannot vote with their feet. Their only lever is voice — proxy voting, engagement, governance pressure. This transforms the governance dynamic fundamentally. The old market discipline (bad management → selling → share price falls → discipline) doesn't operate for passive holders. They must govern through engagement rather than exit.

The consequence is that three asset managers — whose combined assets exceed the GDP of every country except the US and China — effectively have a say in the governance of nearly every significant publicly traded company in the world. The political economist John Coates called this "The Problem of Twelve" — roughly twelve individuals at the major index managers make proxy voting decisions that effectively determine corporate governance norms across the entire economy.

---

Does it change decision-making in companies? Substantially yes, in three directions.

Short-termism — the conventional worry that probably overstates the case. The common argument is that institutional pressure forces CEOs to prioritise quarterly earnings over long-term investment. There is something to this — the activist hedge fund that accumulates a 5% stake and demands a buyback or a spinoff is a real force. But the passive giants, by contrast, have explicitly long-term horizons. Vanguard's investment horizon is, in a literal sense, infinite — they will hold a company forever. The governance pressure they apply is not "hit this quarter's number" but "build durable, long-term value." Larry Fink's annual letters to CEOs over the last decade were explicitly demanding long-term thinking, stakeholder capitalism, and resistance to short-termism. Whether you agree with the content or not, it is structurally opposite to the short-termism critique.

Homogenisation of governance norms. When the same three shareholders are simultaneously the largest holder in Apple, Microsoft, Google, JPMorgan, and ExxonMobil, they necessarily apply a common governance template across all of them. They cannot develop bespoke governance views for each of the three thousand companies they hold. They codify — proxy voting guidelines, governance scorecards, ESG frameworks — and apply them universally. This creates enormous pressure toward conformity. Companies converge toward the same board composition, the same executive pay structures, the same disclosure frameworks, the same sustainability reporting, because they are all being evaluated against the same institutional checklist. The diversity of corporate governance models that might otherwise evolve through competition is compressed.

The rise of stakeholder language. When BlackRock and Vanguard tell CEOs to think about employees, communities, and climate alongside shareholders, it changes what is legitimate to say in a boardroom. The Business Roundtable's 2019 statement redefining corporate purpose away from shareholder primacy — signed by 181 CEOs — was partly a response to this institutional pressure. Whether stakeholder capitalism is a genuine philosophical shift or a sophisticated form of reputational risk management for the institutions themselves (who face political pressure to justify their enormous concentrated power) is a separate and very live debate. But the mechanism is clear: institutional owners now set the terms of what counts as responsible corporate behaviour, more than regulators or market competition.

---

The deeper philosophical question — who does the company serve?

The shift to institutional dominance reopens, in a more complex way, the very question Berle and Means raised. In the old dispersed-ownership model, the answer was theoretically "shareholders" but practically "managers." In the institutional model, the answer is theoretically "shareholders" (who now have actual power), but the shareholders themselves are intermediaries whose ultimate principals are pension beneficiaries, index fund holders, insurance policyholders — the whole working population, essentially.

Vanguard's ultimate beneficial owners are, in a significant sense, the American middle class saving for retirement. So when Vanguard votes on executive pay at a Fortune 500 company, it is nominally a capitalist governance act — one shareholder checking management. But in substance, it is something closer to a diffuse form of social ownership expressing preferences through a financial intermediary. The philosophical gap between "institutional shareholder capitalism" and "market socialism" is narrower than either side of that debate typically acknowledges.

This produces a genuinely novel governance structure: dispersed beneficial ownership (millions of retirement savers), concentrated formal ownership (three asset managers), exercised through standardised governance templates, over managers who are themselves institutional actors with their own principal-agent dynamics. There is no clean theory of corporate governance that maps onto this reality. The old models — shareholder primacy, stakeholder theory, managerial capitalism — were all built for a world that no longer exists.

---

The one thing that probably hasn't changed as much as people think

For all the structural novelty, the actual daily decision-making inside most companies still belongs to the management team and the board. Institutional shareholders vote once a year, publish proxy guidelines, and occasionally engage directly with management — but they don't set strategy, approve capital allocation decisions, or choose customers. The CEO of a large public company still has enormous discretion in practice, constrained by board oversight and quarterly market feedback, but not meaningfully directed by any individual shareholder. The power of institutional ownership is most visible at the margin — in contested votes, in board composition, in compensation structure, in responding to activist pressure — rather than in the day-to-day.

What has changed is the legitimacy framework. The question "who do we serve?" is now asked differently, answered differently, and enforced differently than it was when shareholders were individuals. The definition of good corporate governance is now written largely by institutions who are themselves answerable to a diffuse beneficial ownership that spans most of the population. That is a profound, quiet, and largely unnoticed revolution in the theory of the firm.