Another Hidden Cost for Index Funds: Index Share Changes

When S&P decided to add Tesla to the S&P 500 in late 2020, Tesla’s stock price surged on the announcement and continued to go up before it was officially added a month later, a prime example of the index reconstitution effect revealing the hidden costs of indexing. Since then, the number of Tesla shares in the S&P 500 has not remained constant, and such index share changes, like index reconstitutions, can also come with costs for those who need to rigidly match index changes.

For example, on September 16, 2022, the number of Tesla shares in the S&P 500 increased by around 5% due to S&P making a positive float adjustment.¹ On that day, Tesla saw a 17-fold increase in trading volume during the closing auction, when share changes become effective, relative to its median closing auction volume over the previous 30 days. In terms of price movement, Tesla increased by over 4 basis points (bps) from the end of continuous trading to the closing auction (a short span of 10 seconds or less), and reversed downward by over 3.5 bps by market open the following trading day.² Therefore, Tesla’s share increase in the S&P 500 created a similar scenario to what we saw with Tesla’s addition: For index managers who were forced to buy additional shares at the close to match index changes, they were buying at a higher price.

Index share changes, and their associated costs, are not unique to Tesla. Such changes occur frequently across indices. Reasons include float share adjustments, as seen in the Tesla example, as well as corporate actions that change the number of shares outstanding, such as a seasoned equity offering, stock-based compensation, and share buybacks.³

In Panel A of Exhibit 1, we present the average annual index weights based on market capitalization for the different types of rebalancing events across 10 US equity indices: migrating adds and deletes (gray), nonmigrating adds and deletes (teal), and index share changes (blue). Our sample focuses on discretionary changes and excludes events such as mergers, acquisitions, spin-offs, and share splits. Index share changes make up a meaningful portion by weight, ranging from 3% for the Russell 1000 Growth Index to nearly 10% for the CRSP US Mid Cap Index. However, an index share change event for a stock is typically a much smaller weight adjustment than an addition or deletion. Accordingly, when we examine the number of individual stocks involved in different rebalancing events, shown in Panel B, index share changes stand out as the most common type by a wide margin for almost all indices. For example, the CRSP US Mid Cap Index’s 8% weight for migrating adds and deletes involved just 67 stocks, while the index’s 10% weight for index share changes involved 359 stocks. Panel C provides further information on the typical magnitude of index share changes, plotting the distribution of the absolute percent change in the number of index shares for each index. The median ranges from less than 1% for S&P and CRSP indices to roughly 2% for Russell indices, while the 75th percentile can be as large as 7%.

Just as they are required to rigidly follow index additions and deletions, index fund managers also need to match index share changes in order to minimize tracking error. This lack of flexibility is evident from trading volume during closing auctions, when share changes become effective. Exhibit 2 presents the average trading volume in stocks with index share changes during the closing auction on rebalancing day, reported as a multiple of the stock’s median closing auction volume over the previous 30 days. Across all indices examined, we see abnormally high trading volume during the closing auction on rebalancing day, ranging from an average eightfold spike for the CRSP US Large Cap Value Index to 27-fold for the Russell 2000.⁴ Due to CRSP’s five-day transition schedule, in which share changes are evenly spread out over five trading days, we report multiples for all five days and see the largest spike on the third day.

How does this demand for immediacy from index fund managers impact the prices of stocks undergoing index share changes? In Exhibit 3, we quantify this impact by regressing the returns of stocks around market close on the percentage change in number of shares. We measure returns both from the last midpoint price of the continuous trading session on rebalancing day to the closing auction price (blue bar) and from that closing auction price to the market open price on the following day (green bar). On average, a stock with a 2% index share increase sees its price increase by 0.15 bps relative to nonrebalanced stocks from the end of the continuous trading session to the closing auction on rebalancing day, meaning index fund managers must buy those additional shares at a higher price to match the price reflected in the index. By market open on the day following the rebalance, that same stock would see a downward reversal in price of 0.55 bps relative to nonrebalanced stocks. The opposite would be true for a stock with a 2% index shares decrease: a 0.15 bps price decrease leading into market close with an upward reversal of 0.55 bps by market open the following day.

Even though these impact estimates are relatively small, the substantial number of index share change events that occur each year means that these costs can add up for investors. Indeed, the cumulative impact of the reversal in stock prices for the S&P 500 over the five-year period 2019-2023 is 0.33 bps, which would have resulted in over $150 million dollars “left on the table,” assuming that $5 trillion dollars tracked the S&P 500 on average over that period.⁵ Additionally, these estimates do not capture the ripple effects of needing to sell other stocks proportionally to fund purchases matching share increases, and vice versa.

While index additions and deletions typically get the limelight, we find similar volume and price pressure patterns for index share changes. Given the large number of adjustments to stocks’ shares in indices, even small average costs can add up for investors over the long haul. To avoid such costs, a better approach would be a daily, flexible investment process that spreads turnover across all trading days in the year and uses information in market prices to target stocks with higher expected returns.