Discretionary Investing by ‘Passive’ S&P 500 Funds

So-called passive index funds—investment funds that are designed to track a pre-specified underlying index—have become a dominant force in the investing landscape, collectively controlling over $12 trillion in assets. It is widely assumed that these funds are obligated to follow their underlying index, and that fund managers cannot, or do not, select portfolios that deviate from the index’s holdings. As a result, various critics have attacked these funds, raising concerns about their corporate governance incentives and their influence on market efficiency.

We show this assumption is overly simplistic. To do so, we examine funds that track the most prominent index, the S&P 500. S&P 500 index funds do not typically commit to holding even a representative sample of the underlying index, nor do they commit to replicating the returns of that index. Managers have the legal flexibility to depart substantially from the underlying index’s holdings. We also show that these departures are commonplace: S&P 500 index funds routinely depart from the underlying index by meaningful amounts. While these departures are largest among smaller funds, they are also present among megafunds: even among the largest S&P 500 funds, holdings differed from the index by a total of between 1.7% and 7.5% in the fourth quarter of 2022. Across all S&P 500 funds, these deviations amounted to almost $61.5 billion in discretionary investment decisions. Moreover, at least within observed ranges, we find no meaningful relationship between these deviations and investment flows.

In sum, S&P 500 index funds have substantial investment discretion, which they exercise to an extent not previously recognized. Our findingscomplicate the narrative around index funds and weaken many of the criticisms levied against them. At the same time, to the extent that investors—and particularly retail investors—fail to recognize this discretion, our findings suggest they may not be getting what they expect.

Replication code for this paper is available on JREG’s Dataverse account.