Changes in Micro-Scale ETF Investment: From Full Replication to Sampling Replication

Recently, a large public fund company in South China announced that, in accordance with the fund contract and prospectus, and considering the target index and market conditions, it has decided to change the replication method of its CSI 2000 ETF. Starting from March 18, 2026, the fund will adopt a sampling replication strategy.

The announcement also notes that during the initial phase of the method change, there is a risk that the secondary market trading price of fund shares may trade at a premium or discount. The IOPV (Indicative Optimized Portfolio Value) and the real-time net asset value of the fund shares may differ, and the actual settlement prices for investor subscriptions and redemptions may also vary. Investors are advised to pay attention.

According to the fund contract of this CSI 2000 ETF, as a passive index fund, it primarily uses full replication, constructing an index-based portfolio by buying all the constituent stocks in proportion to their weights in the target index, and making adjustments based on changes in the index components and their weights. However, the contract also states that, to better track the target index, the fund manager will consider factors such as liquidity, investment restrictions, and transaction costs of the index components when deciding whether to adopt a sampling replication strategy. The change will be announced in advance following proper procedures, without the need to convene a shareholder meeting.

Qianhai PaiPaiWang Fund Sales Co., Ltd. public product operations officer Zeng Fangfang told Securities Daily that: “Full replication means buying all the index constituents and maintaining their weights, which results in high tracking accuracy and low error, but also involves higher trading and management costs. Sampling replication only selects some representative stocks from the index, which significantly reduces costs and increases operational flexibility, but may lead to larger tracking errors.”

Zou Zhuoyu, Director of Quantitative Investment Research at Yingmi Fund Institute, told Securities Daily that: “Switching from full to sampling replication can reduce trading costs associated with index tracking and address issues caused by insufficient liquidity of some stocks. However, it also requires higher management standards; without a good sampling strategy, tracking error may increase.”

The underlying index of this CSI 2000 ETF is the CSI 2000 Index. According to the official CSI Index website, the CSI 2000 Index selects 2,000 securities with relatively small market capitalization and good liquidity as its sample. The largest free float market value among the samples is 23.687 billion yuan, and the smallest is 553 million yuan.

Wind data shows that as of March 16, the CSI 2000 Index has gained 8.71% year-to-date; there are 14 ETFs tracking the CSI 2000 Index (including enhanced strategy ETFs), with a total scale of 7.07 billion yuan, and a net fund inflow of 1.879 billion yuan this year.

The CSI 2000 Index, representing small and micro-cap stocks, generally features a large number of constituents with relatively small individual market values. If the number of products linked to small and micro-cap indices continues to grow, will sampling replication become a standard industry practice? Zou Zhuoyu responded: “For small and micro-cap indices, if there are many constituents and some lack liquidity, sampling replication may be increasingly adopted. Full replication is more suitable for a few institutional customized products.”

Zeng Fangfang believes that sampling replication is highly suitable for small and micro-cap indices. “Indices like CSI 2000 have many constituents with varying liquidity, making full replication costly and difficult to operate. Sampling replication can effectively control costs and improve operational flexibility, aligning well with the market demand for index product expansion and increased quantitative strategies. In the future, different replication strategies may coexist for indices with different liquidity characteristics, forming an ecosystem.”

(Edited by: Xu Nannan)

Keywords: