A new study was posted to the National Bureau of Economic Research's "NBER Working Paper Series," titled, "When Cash Wakes Up: Inertia in Institutional Money Market Funds and the Potential Effects of AI Disruption." Written by The University of Chicago's Ali Hortacsu and Younggeun Yoo, the Abstract states, "We study institutional cash management by constructing a novel dataset that records the specific money market funds (MMFs) held by each mutual fund in each month from 2010 to 2024. We document remarkably high inertia: more than half of mutual funds never switched their MMF choice throughout the sample period, despite non-negligible yield dispersion. To decompose the sources of this inertia, we develop a two-stage structural model that separately identifies the determinants of portfolio managers' attentiveness ('waking up') and their conditional preferences once attentive. We estimate the model using a methodology that combines BLP-style contraction mapping with maximum likelihood."
It continues, "We establish three main findings. First, MMFs with less attentive investor bases engage in greater liquidity transformation while offering lower yields. As a result, they earn higher profits and exhibit significantly higher survival rates, especially during periods of low interest rates. Second, we find that aggregate flows appear more yield-sensitive during low interest rate periods (the well-documented 'reaching for yield' phenomenon) not because individual investors become more yield-seeking, but because a larger fraction of investors becomes attentive -- a composition effect rather than a preference shift."
They explain, "Third, counterfactual analysis reveals a potential tension between market efficiency and financial stability if AI-driven automation reduces investor inattention. In the absence of inattention, fees fall by approximately half, yield dispersion collapses to near zero, and market concentration declines by roughly 20%, but monthly fund flows increase by about three times and the incidence of fund-months in which redemptions exceed a fund's weekly liquid assets rises by about six times, raising the risk of fire sales and MMF runs."
The Introduction claims, "Artificial intelligence is rapidly transforming industries across the economy. From software development, where AI-generated code now accounts for a substantial share of new output, to financial services, where autonomous agents are beginning to execute complex workflows with minimal human oversight, the pace of adoption has accelerated markedly in recent years. While AI affects different industries in different ways, one consequence that cuts across sectors is the erosion of consumer and investor inertia. As AI tools automate the search, comparison, and switching decisions that humans previously found too costly or tedious to undertake, the frictions that historically kept customers locked into suboptimal products are rapidly diminishing."
It continues, "This erosion of inertia could be particularly consequential for industries whose profitability has long depended on it. The banking and cash management industries are leading examples. As highlighted in McKinsey's Global Banking Annual Review (2025), approximately $23 trillion of the global $70 trillion in consumer deposits sits in checking accounts earning near-zero interest rates. The report estimates that 'if just 5 to 10 percent of checking balances migrated to top-of-market rates, an action that might be prompted by AI agents, that could reduce the banking industry's total deposit profits by 20 percent or more.' Recent articles in The Wall Street Journal ('AI Will Be a Boon for Savers') and Reuters ('Bankers Are Sitting on a $4 Trillion Rotten Nest Egg') similarly suggest that AI-driven reductions in depositor inertia could substantially erode bank profitability."
They write, "Money market funds (MMFs), which function as close substitutes for bank deposits in cash management for both retail and institutional investors, could face analogous pressures. MMFs currently charge around 20–30 basis points in fees on roughly $7.80 trillion in assets. The fees are notably high given that they are constrained to invest in a narrow universe of short-term, low-risk securities that deliver largely homogeneous returns, compared to, for example, the largest S&P 500 index funds that charge as little as 2–3 basis points. If such high fees are sustained in large part by investor inertia, they may too be vulnerable to erosion if portfolio management becomes increasingly automated by the adoption of AI."
The paper comments, "The end of inertia in banking and money markets deserves special attention not only because these industries have traditionally relied heavily on inertia, but because of a fundamental tradeoff that arises when inertia is reduced. In most markets, reducing inertia unambiguously improves welfare: consumers switch to superior alternatives, prices fall, and market efficiency increases. In banking and MMFs, however, the welfare implications are far less straightforward. These institutions perform liquidity transformation -- investing short-term deposits in longer-term or less liquid assets -- and this function depends critically on the stability of their funding base."
It says, "When depositors and MMF investors are inertial, funding is sticky and predictable, enabling intermediaries to perform liquidity transformation with limited run risk. When inertia disappears and investors continuously re-optimize, funding becomes volatile, amplifying the risk of sudden, correlated outflows and destabilizing runs. Given this tension between market efficiency and financial stability, it is increasingly important to understand the sources and roles of inertia, quantify the associated tradeoffs, and assess the potential consequences of AI-driven reductions in inertia."
NBER summarizes, "In this paper, we study mutual fund portfolio managers' selection of money market funds for cash management. We focus on this setting for several reasons. First, portfolio managers are among the fastest to adopt AI-driven tools for portfolio management, making them a natural population in which to study the potential effects of AI on cash markets. Second, MMFs share many of the key features of bank deposits such as serving as cash-equivalent vehicles, performing liquidity transformation, and being susceptible to runs. Therefore, studying MMFs allows us to draw broader implications for the banking industry, where analogous data on individual institutional depositors' account-level choices at a high frequency are difficult to obtain. Third, mutual fund managers are sophisticated, face low search and switching costs, and are financially incentivized to optimize returns, all of which are the characteristics that the literature typically associates with low inertia. Therefore, the estimated effects of eliminating it could provide a plausible lower bound on the disruptions that would arise among less sophisticated investor populations."
They add, "To be clear, this paper does not seek to predict whether or when advances in AI will eliminate investor inertia. Instead, our goal is to assess what would happen if AI-driven automation were to substantially reduce inattention, by identifying the sources and role of inertia in institutional MMF market and performing counterfactuals with a structural model. Although institutional investors are increasingly adopting AI, its current applications primarily involve information processing and research rather than automated portfolio management."
The piece also tells us, "Investors often devote limited attention to MMF choices not because identifying MMF attributes, such as yields and portfolio maturities, requires extensive research or analysis, but because they prioritize other portfolio decisions that offer greater opportunities to generate alpha. AI could therefore substantially reduce inattention in MMF choices if AI agents were to eventually automate portfolio management on a broad scale, although whether and when such automation will become widespread remain uncertain. Rather than forecasting this transition, our analysis quantifies the potential consequences if investor inattention were substantially reduced or eliminated, and highlights their implications for financial stability, industrial organization, and liquidity provision. These findings can help policymakers prepare for the associated risks and tradeoffs should widespread AI-driven automation materialize, as many have conjectured."
Finally, the paper says, "Using this dataset, we document remarkably high inertia in portfolio managers' cash management: 52.81% of mutual funds never switched their choice of MMF over fourteen years, and in a typical month, more than 95% of mutual funds remained with their current choice. This inertia persists despite non-negligible yield dispersion across MMFs. What explains such strong inertia? Two main explanations emerge. The first is rational inattention. MMFs are constrained to hold similar types of short-term securities and therefore exhibit very similar risk profiles -- only two funds have ever broken the buck throughout the entire history of MMFs -- so mutual funds may view most MMFs as close substitutes for which switching appears unnecessary.... The second explanation is stable preferences for non-yield attributes, particularly a strong tendency to hold MMFs sponsored by the same parent company."
