MSCI World Index: what it is, how it works, what it includes

The ETFs that track the MSCI World: costs and tracking difference

It is the index European portfolios replicate the most, yet almost nobody has read the rules that govern it. We read them all — from the 192-page construction manual to the committee policies — and this guide tells you what actually sits behind the name: who decides what gets in, when, and why the numbers you see around never quite match.

What the MSCI World really is

The MSCI World is an equity index: a list of stocks, each with a weight, meant to represent the stock markets of developed countries. It is calculated and governed by MSCI, a listed US company whose business is producing indexes and licensing them to product manufacturers — for instance the ETFs that track it.

The single most important thing to understand: the MSCI World is not a selection. No manager picks “the world’s best stocks”. The methodology defines it as the plain sum of the indexes of every market classified as “developed”: pass your market’s filters and you are in; fail them and you are out. As of 31 July 2026 it held 1,282 stocks.

The index has existed since 31 March 1986. An honest detail the official factsheet states and almost nobody repeats: the “since 1969” data in long-run charts is partly back-tested — a reconstruction of how the index might have behaved had it existed.

What it includes and what it leaves out

“World” is indexing’s most misleading name: the MSCI World covers 23 developed markets — not the world. They are: Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Hong Kong, Ireland, Israel, Italy, Japan, the Netherlands, New Zealand, Norway, Portugal, Singapore, Spain, Sweden, Switzerland, the United Kingdom and the United States.

Three things many believe are inside actually are not:

• Emerging markets: China, India, Brazil, Taiwan — and even South Korea, which MSCI classifies as “emerging” despite being one of the most advanced economies on Earth (we will see why: being rich is not enough, the market must be accessible under 18 criteria).
• Small companies: the MSCI World takes only large and mid caps, roughly 85% of each country’s investable market cap. If you want small caps too, look for the IMI suffix.
• Two “too small” developed countries: Luxembourg and Cyprus are classified as developed by the methodology but excluded from the index for their modest size. A gem you only find in a footnote of the manual.

How it is built: a rule, not a choice

Every quarter MSCI starts again from the universe of every listed company across the 23 countries and applies a chain of filters written down in the GIMI methodology (the 192-page master rulebook):

• Size: the company must be large enough. The threshold is global, not per country — which is why Hungary’s Standard Index has 4 stocks and New Zealand stays in only thanks to the 5-constituent minimum rule.
• Free float: only the shares genuinely buyable on the market count. Founders’, states’ and strategic holders’ stakes carry no weight. At least 15% of the company must float to qualify, and index weight is computed on float, not on total market cap.
• Liquidity: the stock must be genuinely buyable and sellable. Two measures: it must have traded on at least 9 trading days out of 10 (many minor stocks sit still for whole days), and over a year a traded value of at least one fifth of its float must change hands — the measure is called ATVR. The point: a billion-euro fund tracking the index must be able to get in and out without moving the price. A curiosity from the manual: a stock priced above USD 10,000 is deemed de facto illiquid and cannot get in — the reason Berkshire Hathaway is in through its class B shares, not the A.

This is also where the famous US weight comes from: the index has no country caps, weights follow float-adjusted market cap. If American companies are worth more than two thirds of the developed world’s market, the index weighs them accordingly — not MSCI’s choice, just a photograph of the market. Same for concentration: as of 31 July 2026 the top 10 stocks weighed 26.41% of the whole index (NVIDIA 5.18%, Apple 5.07%, Microsoft 3.66%…), out of 1,282 stocks.

The obvious question: shouldn’t weight follow GDP? No — and the difference is the heart of the index. Market cap measures the stock-market value of listed companies: the earnings the market expects (wherever they are generated) times what it is willing to pay for them, counting floating shares only. GDP measures something else — a territory’s output. The two diverge for three reasons: not all of an economy is listed (Germany is full of large family firms outside the market, and state holdings do not float); multinationals “weigh” entirely in their listing country, wherever they sell (an iPhone bought in Milan lands in the US weight); and valuation multiples differ across markets. So the 72% United States does not say the American economy is 72% of the developed world: it says the American stock market gathers that share of the listed, buyable value.

Today’s snapshot: countries, sectors, valuations

Here is what the index looks like right now, from the official factsheet. Countries first — the chart that explains on its own why “World” needs an asterisk:

Then the sectors, under the GICS classification (the 11-sector taxonomy MSCI runs together with S&P — another machine with rules of its own): technology alone is worth more than a quarter of the index, and the top three sectors exceed half.

Finally the aggregate valuations: what the index “costs” relative to the earnings, book value and dividends of the companies it holds. These are numbers to compare over time rather than read in absolute terms — which is exactly what this page will let you do, month after month.

How a stock gets in (and out)

Four times a year — February, May, August and November — the index gets its scheduled service: MSCI regenerates the universe, re-applies the filters and announces the changes at least two weeks ahead; they take effect at the close of the month’s last day. Funds tracking the index therefore know in advance what they will need to buy and sell.

To stop borderline stocks from hopping in and out on every swing, the doors are asymmetric: moving up a segment requires beating 150% of the threshold, dropping out requires falling below 66.7%. That buffer is what keeps churn low: over the last 12 months the MSCI World’s turnover was just 2.95%.

Between services the index does not sleep: large IPOs can join after as little as 10 trading days if they clear the size bars, bankrupt companies leave the same day, and since 2021 there is even an “anti-bubble” filter blocking stocks with extreme, sudden price spikes versus their peers. All of this machinery — thresholds, buffers, snapshots, exits — gets its own instalment, part two: how a stock gets into (and out of) an index.

Price, Gross, Net: three numbers, one index

When two sources show different returns “for the same index”, they are almost always looking at two of its three official flavours. The Price version counts prices only; the Gross one reinvests gross dividends; the Net one reinvests dividends after deducting the highest possible withholding tax for a foreign investor with no tax treaties — the worst case by construction. The methodology says it plainly: Gross approximates the maximum possible reinvestment, Net the minimum.

The assumed withholding varies a lot by country: 30% on US dividends, 35% on Swiss ones, zero on British ones. Which is why an Ireland-domiciled ETF can beat its own Net index: thanks to the Ireland–US tax treaty it pays less on American dividends than the index assumes. No manager magic — a structural margin written into the calculation rules. European ETFs almost always state the Net version as their benchmark. How closely a fund actually tracks it (or beats it) also depends on how it replicates the index: we cover that in our guide to the differences between physical and synthetic ETFs.

Who governs it (and who profits)

The rules above do not write themselves. Inside MSCI, proposals come from research teams, but every decision goes through a committee: the Equity Index Committee approves methodologies, corrections and exceptional cases, with a higher body (the Index Policy Committee) for the most delicate questions. Individual discretion is banned by policy; committee discretion exists, and is documented case by case.

Before changing an important rule MSCI opens a public consultation: anyone may respond, but the decision stays with MSCI, which can even go against the majority of respondents (it must then say so). Since 2018 MSCI has also been an authorised benchmark administrator under the EU benchmarks regulation — born of the benchmark-manipulation cases such as LIBOR cited in its first recital — listed on the supervisors’ public registers.

Two things are worth knowing, and MSCI itself writes them in its documents. First: its revenues include fees computed on the assets of products linked to its indexes — the more the ETFs grow, the more the index’s administrator earns. Second: if MSCI makes a calculation error whose impact stays below a declared threshold (50 basis points on a country index), history is not rewritten; and errors found after 12 months are generally never fixed. These are public rules, not scandals — but knowing them changes how you read an index. This is the subject of part three: who governs MSCI indexes.

How it has done: a tale of two currencies

The MSCI World’s return depends on which flavour you look at (Net or Gross) and which currency you use. 2025 is the perfect example: +21.09% in dollars, +6.77% in euros — same index, same Net flavour: the entire gap is the exchange rate. For a euro-area investor the return that matters is the euro one, which is why our charts start there.

Risk changes with currency too: the Net series’ worst fall in dollars was 57.82% (October 2007 to March 2009), while in euros the worst peak-to-trough is a different period and a different number (53.6%, from 2001 to 2009). Over 15 years (July 2011 → July 2026), 100 invested in the Net version in euros became 606; in the Gross version 655: the gap is the cumulated effect of dividend withholding. As always: past returns say nothing about future ones — and for this index, pre-1986 data is a reconstruction.