Side Lesson 18: Global Markets — The Case For and Against International Diversification

---

1. Why This Is Important

Open any introductory portfolio textbook and you will find the same sentence: *the U.S. is roughly 60% of global market capitalization, so a properly diversified equity portfolio should hold the other 40% overseas.* It sounds obvious, almost a tautology. This course disagrees with the conclusion. Side lesson 18 explains the disagreement and gives you the rule that the rest of the curriculum quietly relies on.

There are four reasons this is worth a full lesson rather than a footnote:

**The home-bias debate is the single biggest allocation question

most retail investors will face.** It is bigger than which factor tilt to adopt, bigger than active vs passive — because it controls roughly forty percentage points of your equity exposure. Getting the right answer matters more than almost anything else.

**The "rebalance into international" reflex is built on a survivor

sample.** Modern Portfolio Theory was canonised in the 1950s-1980s, when U.S., U.K., German, and Japanese markets all looked roughly comparable on the dimensions a textbook cared about (free float, accounting, rule of law). That is not the world investors live in today. Russia 2022 and China 2021 are not edge cases — they are the modal foreign-equity outcome over a long enough horizon.

**The U.S.-versus-world performance gap from 2010 to 2024 was not

noise.** It was a structural divergence in earnings growth, capital formation, and corporate governance. The textbook models that recommend 40% international were calibrated on a regime that no longer exists.

The course's investable universe rule originates here. Every other lesson — from week 3's risk-return

premia to week 24's institutional sleeve template — implicitly restricts itself to U.S.-listed securities. This side lesson explains why, so the rule does not feel arbitrary when it shows up in later weeks.

The textbook position is not wrong; it is incomplete. The honest answer is that international diversification gives you a real mathematical benefit if and only if the foreign markets you buy share four properties with the U.S.: rule of law, accounting transparency, minority-shareholder rights, and a deep secondary market. Roughly half of "international" by index weight does not clear that bar.

---

2. What You Need to Know

2.1 The Map: How Big Is "the World" Outside the U.S.?

The MSCI All Country World Investable Market Index (ACWI IMI) is the broadest investable equity benchmark on the planet. As of April 2026 its weights look approximately like this:

• United States — ~60%. Roughly $55T market cap, with the

Magnificent-7 alone equal to the entire EAFE index by float.

• Developed ex-U.S. (EAFE) — ~25%. Japan ~6%, U.K. ~4%,

France ~3%, Canada ~3%, Switzerland ~2.5%, plus Germany, Australia, Netherlands, Sweden, Hong Kong, Singapore, Spain, Italy, Denmark, Belgium, Finland, Norway, Israel, Ireland, Portugal, Austria, New Zealand.

• Emerging markets — ~12%. China ~25-30% of the EM bucket,

India ~17%, Taiwan ~16%, Korea ~12%, Brazil ~5%, Saudi Arabia ~4%, South Africa ~3%, Mexico ~3%.

• Frontier — ~3%. Vietnam, Nigeria, Kenya, Bangladesh, Sri Lanka,

Romania, Kuwait (until promoted), Pakistan, Morocco. Frontier is so small and illiquid that even institutional managers rarely treat it as a real allocation.

The key intuition the chart should leave you with: *global market cap is dominated by one country.* This was not always true — in 1989, at the peak of the Japanese bubble, Japan was briefly the largest equity market in the world and the U.S. was a hair behind. Forty years later, the gap is roughly five-to-one. That regime change matters more than any backtest.

2.2 The Performance Tape, 1990-2024: Two Decades, Two Stories

The textbook case for international diversification leans heavily on the 2000-2009 "lost decade" — the period when SPY went sideways and EAFE / EM significantly outperformed. The textbook is right that this period existed. It is wrong to extrapolate from it.

A cleaner way to look at the same data: the period **1990 through April 2026** decomposes into roughly three regimes.

• 1990-1999: U.S. dominance. SPY compounded at about 18% per

year, EAFE at about 7%, EM at about 11% with the 1997 Asia crisis taking a chunk out late. The U.S. tech run made every other equity market look slow.

• 2000-2009: International dominance. SPY -1%/yr (the dot-com

unwind plus the GFC). EAFE +1%/yr unhedged (helped by a falling dollar). EM +10%/yr (the BRIC commodity supercycle). For the only decade in modern history, holding ex-U.S. stocks paid.

• 2010-2024: U.S. dominance, hard. SPY about +14%/yr.

EAFE about +6%/yr. EM about +4%/yr. The gap was the largest in modern data — wider than the 1990s — and was driven by three things that may or may not persist: software-eats-world (favours the U.S. tech complex), shale (favours U.S. energy independence), and dollar strength (penalises unhedged ex-U.S. holdings).

The honest reading of this chart: **leadership has rotated, and nobody knows where it goes next.** That is also Horace's first principle: *alpha is rare; the toolkit is portfolio construction.* If you cannot predict which region wins the next decade, the standard textbook response is *own all of them in proportion to size.* This course's response is different and is explained in §2.4.

2.3 The Currency Layer

International returns have two components: the foreign-currency return of the underlying stock, plus the change in the foreign-currency-to-dollar exchange rate. Over a single year these two pieces are roughly the same size in volatility — about 7-9% each — and the FX piece can flip a positive local return into a negative dollar return or vice versa. 2014-2016 (strong dollar) and 2022 (very strong dollar) hurt unhedged ex-U.S. holdings; 2002-2007 (weak dollar) and 2017 (weak dollar) helped.

You can hedge currency. Vehicles like HEFA (hedged EAFE) and DBEF buy a rolling forward to neutralise the FX leg. The hedge costs roughly the interest rate differential — about 1% per year against the euro, more against the yen when U.S. rates are higher than foreign rates. Empirically, over 10-year windows the *currency effect averages close to zero* but adds about 30-40% of the volatility of unhedged international equity. So the trade-off is clean: hedge if you care about year-to-year noise, leave unhedged if you have a 10+ year horizon and want the diversification of holding non-dollar assets. Most retail-grade research lands at "leave it unhedged for the equity sleeve."

This course leaves the question moot, because the U.S.-only rule side-steps it entirely.

2.4 The Political-Risk Premium and the Investable-Universe Rule

The argument the textbook does not make is the one that determines this course's allocation rule. International equities — especially emerging-market equities — carry a class of risk that does not show up in any of the standard sigma / VaR / drawdown statistics: the risk of expropriation. The textbook implicitly assumes you can always sell.

A short list of times that assumption broke in the last decade:

• China 2021. The CCP's "common prosperity" campaign wiped

about $1.5T off Chinese tech stocks in six months. DiDi was delisted from the NYSE within months of its IPO. The for-profit tutoring industry was made illegal by regulatory fiat overnight. Retail holders of FXI / KWEB watched 50-70% drawdowns with no fundamental catalyst — only a policy change.

• Russia 2022. After the Ukraine invasion the Moscow Exchange

was closed to foreign sellers for several months. By the time trading reopened, Western sanctions had effectively rendered every Russia-listed share worthless to a U.S. holder. Index providers marked Russian weights to zero. Roughly $50B of foreign-held Russian equity was wiped out — not in a market-driven drawdown, but by political fiat. This was the largest single-country expropriation event since the Cuban revolution in 1959.

• Hong Kong 2020-2024. The National Security Law and subsequent

capital-control tightening have eroded the rule-of-law premium Hong Kong used to carry. Stocks listed there are now pricing in mainland-China policy risk that did not exist five years ago.

• Turkey 2018-2024. Erdogan's unconventional monetary regime

destroyed about 90% of the lira's purchasing power. A Turkish stock that doubled in lira terms still lost two-thirds in dollars.

The pattern in all four cases is the same: *an event the U.S.-listed investor could not hedge against, did not show up in the historical risk model, and could not be unwound at any price.* This is the political-risk premium, and the standard equity-premium math does not capture it.

This course's response — the "investable universe" rule — is to **restrict the entire recommended portfolio to U.S.-listed equities.** The rule has four pillars:

Rule of law. U.S. courts enforce contracts, including against

the government. The Fifth Amendment takings clause has been tested, and it works.

Accounting transparency. SEC-registered issuers file 10-K

and 10-Q statements audited under PCAOB oversight. Foreign-private issuers reporting under their home regime do not get the same audit guarantee.

Minority-shareholder rights. Delaware corporate law has

200 years of case law protecting minority holders against controlling-shareholder self-dealing. China has none. Russia has none. Korea is improving but is not there yet.

Deep secondary markets. You can sell a U.S.-listed stock at

a tight bid-ask spread on any trading day in any regime. You cannot say that about Vietnam, Nigeria, or Argentina.

The four pillars are substitutes for the diversification benefit. You give up about 40% of the global investable opportunity set; in exchange you get a portfolio you can actually liquidate in a crisis and that nobody can confiscate by political decision.

2.5 The Compromise Lane: U.S.-Listed ADRs

Refusing to own EFA / VXUS / VWO does not mean refusing to own non-U.S. companies. The compromise — the one the rest of the course quietly assumes — is **U.S.-listed American Depository Receipts (ADRs).**

An ADR is a U.S.-listed share of a foreign company, registered with the SEC, custodied by a U.S. bank, and subject to U.S. rule of law. Buying TSM (Taiwan Semiconductor) on the NYSE is materially safer than buying 2330.TW on the Taipei exchange — same underlying business, but the U.S. listing forces SEC reporting, U.S. courts have jurisdiction, and the shares clear through the same DTC infrastructure as AAPL.

The shortlist of high-quality ADRs that earn a place in the course's investable universe:

• TSM — Taiwan Semiconductor. Effectively a monopoly on

leading-edge logic fabrication. The single most important semiconductor business on Earth.

• ASML — Dutch maker of EUV lithography tools. The only

vendor of the machines that print every leading-edge chip. Listed on NASDAQ since 1995.

• SAP — German enterprise-software incumbent.
• NVO — Novo Nordisk, Danish maker of Ozempic.
• TM / SONY / HMC — high-quality Japanese exporters.
• SHOP — Shopify, Canada-domiciled but NYSE-listed.

A second tier is worth caveating because of the political-risk overlay: BABA, JD, PDD, BIDU, NIO are U.S.-listed Chinese ADRs. They share a U.S. wrapper but the underlying assets are inside mainland China and inside a Variable Interest Entity (VIE) structure that has never been tested in a full Chinese expropriation event. Position-size them like venture bets, not like core equity.

The mechanic to internalise: *the investable universe is defined by the listing venue, not by where the company does business.* TSM is in. 2330.TW is out. Both are claims on the same assembly line.

---

3. Common Misconceptions

**"60% U.S. of world cap means I should own 60% U.S. in my

portfolio."** This is the textbook recipe. It assumes all 100 cents on the dollar of global cap are equally investable. They are not.

**"International is a free diversification benefit because of

low correlation."** Correlations between SPY and EFA have run 0.85 in the last decade — high enough that the diversification math gives you maybe 1-2% volatility reduction, not the 5%+ the textbook suggests.

**"Emerging markets must outperform because they're growing

faster."** Decades of data show GDP growth and equity returns are essentially uncorrelated. China's GDP grew 10x from 2000-2020 and the MSCI China index returned about 5%/yr in that span.

"I'm under-diversified if I don't hold VXUS." You are

concentrated in U.S. legal jurisdiction. That is a deliberate choice, not an accident. It is the trade-off, not a mistake.

"Currency hedging is too complicated to bother with." It is

bundled into ETFs. HEFA charges 35bp. That is the cost of the hedge — usable if you want it.

"Russia 2022 was a one-off." It was the largest single

event in the modern era, but China-2021 and Hong Kong's slow re-rating since 2020 are the same family of risk. The base rate is higher than it looks.

"ADRs are no different from local-listed shares." They have

identical economic exposure but materially different legal exposure. The U.S. listing is a real protection layer.

"China A-shares are now investable because MSCI added them."

MSCI's index decisions are about benchmark coverage; they do not confer rule-of-law. The 2017 A-share inclusion did not change what happened to retail holders of YY, DiDi, or NTES in the 2021-2022 crackdown.

"Buying EAFE is safe — it's all developed markets." EAFE is

safer than EM but still carries currency risk and concentrated single-country risk (Japan was 70% of EAFE in 1989; the U.K. was 30% in 2000). Diversification within EAFE is uneven.

"International diversification protects me from a U.S. crash."

The 2008 GFC saw EAFE -43% and EM -53% versus SPY -37% — the worst drawdowns of the three were ex-U.S. The diversification benefit fails precisely when you most need it.

---

4. Q&A Section

**Q: If I follow the U.S.-only rule, what about TSM, ASML, BABA? Are those allowed?** A: TSM and ASML are yes — they are SEC-registered, U.S.-listed, and you have a Delaware-style claim on the depository shares. BABA is a maybe — the wrapper is U.S. but the underlying VIE structure sits inside mainland China and has never been tested by a full expropriation. Size BABA-class names like opportunistic positions, not like core equity.

**Q: What about VXUS or VEU as a core holding? Vanguard sells them precisely for retail.** A: They are perfectly serviceable products. The course rule is not that they are bad; it is that the course's risk-return math deliberately excludes them. If you choose to hold 10-20% in VXUS, your CAGR will look approximately like the lesson's 1990-2024 chart — i.e. lower than U.S.-only over the last 15 years, higher in the early 2000s. You are taking the textbook trade.

**Q: What's the right hedge for U.S.-only concentration risk if I believe the U.S. could underperform for a decade like 2000-2009?** A: Three answers. First, the four-tranche framework already includes a stores-of-value sleeve — gold, real assets — that is currency-agnostic. Second, U.S. multinationals (AAPL, MSFT, PG, KO) earn 50%+ of revenue overseas, so SPY itself is already half-international by economic exposure. Third, if you really want the full FX/region diversification, the cleanest expression is GLD plus a small commodities sleeve, not VXUS.

**Q: Do U.S.-listed ADRs of foreign companies pay foreign withholding tax?** A: Yes. The custodian bank withholds at the foreign-source rate (15% under most treaties, 30% in some cases). In a taxable account you can claim a foreign tax credit on Form 1116. In an IRA the withholding is permanent and not recoverable — which is one reason the course recommends holding international-exposed names in taxable, not tax-deferred accounts. The tax-location lesson covers this in detail.

Q: Is VWO different from EEM? Both are emerging markets ETFs. A: Same asset class, different index providers. VWO follows FTSE (includes Korea), EEM follows MSCI (excludes Korea). VWO's expense ratio is 0.08% vs EEM's 0.70%, which is enough cost difference to matter. If you do hold EM, VWO is the correct vehicle.

**Q: What about the China A-shares MSCI added in 2018-2019? Are those any safer than Hong Kong-listed Chinese shares?** A: They are less safe. A-shares trade on the Shanghai and Shenzhen exchanges and are subject to mainland Chinese law in a way that even Hong Kong listings are not. The 2017 inclusion was a benchmark-coverage decision by MSCI, not an accessibility upgrade. Capital controls limit your ability to sell in a crisis. Hard pass.

Q: How does the lesson square with Horace's barbell principle? A: The barbell says *combine cheap-and-safe + lottery- ticket + nothing in between.* The U.S.-only rule applies to the cheap-and-safe leg: VTI / SCHD / TLT — all U.S.-listed. The lottery-ticket leg is allowed to include non-U.S. names (small TSM or ASML position) because those are sized like options, not like core. The middle of the barbell — international index funds at 40% weight — is exactly what the framework rules out.

Q: What single number would change my mind? A: A persistent reduction in the political-risk premium between the U.S. and the next-largest market (call it Japan or the U.K.). Concretely: if U.K. corporate-governance ratings, accounting transparency, and contract enforcement converged with U.S. levels, plus FX hedging costs collapsed, plus a sustained 20%+ valuation discount opened, then the trade becomes interesting. None of those conditions hold today.

**Q: Doesn't the U.S. itself carry political risk? FTC, antitrust, crypto crackdown, tariff policy?** A: Yes — the U.S. is not zero-political-risk. But the magnitude gap is large. Antitrust unwinds are litigated for years and bounded by the Sherman Act and Delaware case law; they do not look like the China 2021 crackdown. The course's claim is not "U.S. is safe" — it is "U.S. is materially safer on the rule-of-law axis than the alternatives."

**Q: I already own VXUS in my 401(k) and can't easily get rid of it. What now?** A: Don't fight your plan. The 401(k) match (the tax wrapper beats security selection) is worth 50%+ on the dollar; do not sacrifice that to enforce a 5-percentage-point allocation rule. If your plan menu offers VXUS but not a clean U.S.-only balanced fund, hold VXUS in the 401(k) and tilt the taxable account U.S.-only to bring the household total to the target.

**Q: Final question — if I run the global blender in the interactive panel, what should I expect to see?** A: Three things. First, almost any plausible U.S.-heavy mix (70-100% VTI) beats the cap-weighted (60% VTI / 28% VXUS / 12% VWO) on CAGR over the 1990-2024 backtest, by roughly 1-2% per year. Second, the Sharpe gap is narrower than the CAGR gap — international does damp some volatility. Third, the maximum- drawdown line moves around less than you'd expect: the GFC and COVID drawdowns hit everywhere. The data argues for more U.S., not less, even before the political-risk overlay.