Early SPDR Creation-Unit Arbitrage
When the first modern ETF launched in 1993, its creation-and-redemption mechanism was a genuinely new arbitrage tool, and being one of the few authorized participants who understood how to use it was a real, if short-lived, edge.
Prerequisites: ETF Creation and Redemption
In January 1993, the SPDR — the SPY ETF tracking the S&P 500 — began trading as one of the first products of its kind: a security that trades continuously on an exchange all day, like a stock, but whose price is kept anchored to the value of an underlying basket of stocks through a mechanism called creation and redemption. That mechanism was new enough, and understood by few enough market participants, that the handful of firms authorized to use it directly had a real arbitrage edge for the first several years of the product's existence, before the structure became widely understood and heavily competed.
An ETF's price can only drift meaningfully away from the value of its underlying basket if arbitrageurs aren't actively closing the gap through creation and redemption — in SPY's earliest years, before every desk understood and had built the operational plumbing for that arbitrage, the gaps were wider and more persistent than they later became.
How creation-unit arbitrage worked
An authorized participant — a large broker-dealer with a direct relationship to the fund — can hand over a specified basket of the underlying stocks in the exact index weights and receive a large block of new ETF shares in exchange, "creation," or do the reverse, handing in ETF shares to receive the underlying stocks, "redemption." If the ETF's market price on the exchange trades above the value of its underlying basket, an arbitrageur buys the basket of stocks, delivers them to create new ETF shares, and sells those shares on the exchange at the higher price, pocketing the difference — and the extra supply of ETF shares created this way pushes the ETF's price back down toward fair value. The mechanism is symmetric in the other direction when the ETF trades below its basket value. In 1993, few firms had the operational infrastructure — the basket-construction systems, the relationship with the fund sponsor, the capital — to actually execute this quickly, so the arbitrage that later became fast and thin was, in the product's earliest days, slower to close and correspondingly more profitable for whoever could do it.
Worked example
Suppose the underlying basket of S&P 500 stocks is worth $44.00 per SPDR share, but strong retail demand on launch day pushes SPY's exchange price to $44.30, a 30-cent premium. An authorized participant buys the full underlying basket in the exact index weights for $44.00 a share equivalent, in a 50,000-share creation-unit block (worth roughly $2.2 million at basket value), delivers it to the fund, and receives 50,000 new SPDR shares in return. Selling those shares on the exchange at $44.30 nets $44.30 × 50,000 = $2,215,000, versus the $2,200,000 cost of the basket, a gross profit of $15,000 on the block before transaction costs — a trade only available to firms with an authorized participant agreement and the operational ability to buy an entire 500-stock basket in the correct weights within the trading day, a real barrier to entry in 1993 that thinned out considerably over the following years as more firms built the same capability.
What this means in practice
Early SPDR arbitrage is worth understanding as the founding case of a mechanism that now underlies trillions of dollars in ETF assets globally: the reason a modern ETF's price rarely drifts far from its net asset value isn't luck, it's a standing arbitrage incentive that keeps a competitive set of authorized participants closing any gap within minutes. The edge available to early participants came from being one of the few who had built that plumbing before it became table stakes for the entire industry.
It's a common mistake to think ETF arbitrage guarantees the ETF price never deviates from its basket value. The mechanism only works as fast as authorized participants can execute it — for baskets of illiquid or hard-to-borrow underlying securities, or during periods when trading in the underlying is halted, ETF premiums and discounts can persist and widen well beyond what a liquid, all-stock ETF like SPY typically shows.
Related concepts
Practice in interviews
Further reading
- Gastineau, The Exchange-Traded Funds Manual