EM ETFs compared: MSCI versus FTSE, ex-China and IMI variants, withholding-tax leakage and sampling

What you will learn

• How the Korea difference between MSCI and FTSE creates gaps or double counting when you mix index families.
• What "IMI" and "ex-China" variants change in an emerging-market (EM) portfolio.
• How to estimate dividend withholding-tax leakage inside an EM ETF.
• Why most EM ETFs use optimised sampling and how to check whether it hurts.

Explanation

ETF structure, replication methods, total expense ratio and tracking difference are explained in the course "ETFs, Funds & Long-Term Investing". Here we apply them to the concrete choice between EM ETFs.

MSCI versus FTSE: the Korea question

As of 2026, MSCI still classifies South Korea as an emerging market (its June 2026 review kept Korea in the EM index), while FTSE Russell has treated Korea as developed since 2009. The practical consequence: an "MSCI World + MSCI EM" pair and an "FTSE Developed + FTSE Emerging" pair both cover Korea once. Mixing families does not. MSCI World plus FTSE Emerging leaves Korea out entirely; FTSE Developed plus MSCI EM counts Korea twice. Poland is a smaller version of the same issue. Rule: stay within one index family for your developed and EM building blocks, or use one all-world ETF.

IMI and ex-China variants

A standard MSCI index covers about 85 % of each country's free-float market capitalisation (large and mid caps). An IMI (Investable Market Index) adds small caps and reaches about 99 %. The EM IMI holds several thousand stocks instead of roughly 1,200 to 1,400, adds a small-cap tilt and is usually replicated by sampling.

An ex-China index removes China and rescales the remaining countries. Investors use it to control China's weight separately or because of policy, sanctions or ethical concerns. The side effect is concentration: with China removed, India and Taiwan each become roughly a quarter or more of the index, and the technology-hardware weight rises.

Withholding-tax leakage

Dividends paid by EM companies are taxed at source before they reach the fund. Typical statutory rates for foreign investors as of 2026 are around 10 % in China, 20 % in India (plus surcharges), 21 % in Taiwan, 22 % in Korea including the local surtax, and 20 % in South Africa. Brazil, long at 0 %, introduced a 10 % withholding on dividends paid to non-residents from 2026 (Law 15,270/2025). Rates change; check the current rule. Irish- and Luxembourg-domiciled funds get some treaty relief, but in most EM countries little. The result is a permanent drag equal to the index dividend yield multiplied by the average withholding rate. Net-return indices assume the maximum rate, so an ETF slightly beating its net index is usually recovering treaty relief. German fund taxation is covered in the ETF and taxes courses.

Sampling

With 1,200 or more constituents in many currencies, most EM ETFs hold a representative subset matched on country, sector and factor exposures (optimised sampling). Check the number of holdings against the index, and compare the five-year tracking difference with the total expense ratio: a tracking difference larger than the fee, in a bad direction, suggests sampling or tax costs.

Worked example

Hypothetical example: all figures invented and rounded; order costs, spreads and German taxes excluded.

Step 1 – Korea gap. An investor holds 88 % in a FTSE Developed ETF and 12 % in an MSCI EM ETF. Assume Korea weighs 1.8 % in FTSE Developed and 10 % in MSCI EM. Korea exposure = 0.88 × 1.8 % + 0.12 × 10 % = 1.58 % + 1.20 % = 2.78 %, roughly double the intended 1.6 %. With MSCI World (Korea 0 %) and FTSE Emerging (Korea 0 %) the exposure would be zero.

Step 2 – withholding leakage. The EM index yields 3.0 % gross. The weighted average withholding rate the fund actually bears is 12 %. Annual leakage = 3.0 % × 0.12 = 0.36 % of fund value. On a €20,000 position that is 20,000 × 0.0036 = €72 per year, about €720 over ten years before compounding.

Step 3 – ex-China rescaling. Assume China is 30 % of the standard index, India 20 % and Taiwan 18 %. In the ex-China version India becomes 20 ÷ 70 = 28.6 % and Taiwan 18 ÷ 70 = 25.7 %. Two countries now make up more than half of the fund.

Common mistake

Practice activity

On paper: you hold €50,000 in equities, 12 % of it (€6,000) in an MSCI EM ETF and the rest in an MSCI World ETF. Assume Korea is 10 % of MSCI EM. (a) How many euros of Korea exposure do you have, and what would change if you swapped the EM ETF for a FTSE Emerging ETF? (b) The EM ETF's index yields 2.8 % gross and the average withholding rate borne by the fund is 10 %. Estimate the annual leakage in euros. (c) The factsheet shows a total expense ratio of 0.18 % and an average annual tracking difference of −0.45 % over five years against the net-return index. What does the gap suggest?

Key takeaways

• Stay within one index family: MSCI keeps Korea in EM (confirmed June 2026), FTSE treats it as developed; mixing creates a gap or double count.
• IMI adds small caps and sampling; ex-China concentrates the fund in India and Taiwan.
• Withholding leakage ≈ gross dividend yield × average withholding rate; it usually exceeds the fee difference between funds.
• Judge a sampled EM ETF by its multi-year tracking difference, not by the total expense ratio alone.

Next step

The next lesson goes to the edge of the investable world: frontier markets, single-country funds and the discounts of closed-end funds.