Who Moves the Stock Market? Return Chasers, 60/40 Funds, and Inelastic Prices

If every investor priced stocks off expected cash flows, the stock market would be calmer than the earnings it is built on. In fact it swings far more. Robert Shiller showed in 1981 that the variance of U.S. stock prices was 5 to 13 times larger than the upper bound implied by the dividends that followed, and a large share of asset pricing research since then has tried to explain why.¹

A new working paper by Victor Haghani, Vladimir Ragulin, Jeffrey Rosenbluth and James White, “Who Killed the Random Walk?”, offers one answer: look at who owns the market and how each group decides how much stock to hold.² Haghani walked through it on episode 429 of the Rational Reminder podcast.³ The paper is a simulation, so most of its numbers are model outputs. The authors also test their rules on 30 years of real returns, and the gap between the two sets of results matters for anyone who holds index funds.

Three investors and one formula

Every investor in the model sizes their stock position with the same rule, the Merton share: hold stocks in proportion to the expected excess return, divided by risk aversion times variance.

Risk aversion ($\gamma \approx 3$) and the volatility estimate are the same for everyone and never change. The investor types differ in one input only: how they estimate the expected return on stocks.

• Valuation investors use the market’s earnings yield. When prices rise faster than earnings, their expected return falls and they sell. They are the rational investor of textbook finance.
• Static investors assume a constant risk premium, so they always hold the same mix and rebalance back to it. Target-date funds, balanced funds and a DIY three-fund portfolio with a fixed target all behave this way. The authors call this group “arguably the largest investor group in today’s marketplace.”
• Extrapolators estimate future returns from past returns: a weighted average of the last five years, with 33% on the most recent year and each earlier year weighted 25% less than the one after it. Their estimate is squeezed so they never borrow or go short, and they move toward their new target gradually.

The extrapolator is grounded in survey data. Greenwood and Shleifer compiled six investor-expectation surveys from 1963 to 2011 and found that expected returns rise after the market has risen and are “strongly negatively correlated with model-based expected returns.”⁴ People expect the most just when valuations imply the least.

The base case gives each of the three types one-third of starting wealth. The paper also runs four smaller types that do not move prices in the base case: a binary momentum investor, a buy-the-dip investor, a value-plus-momentum hybrid, and an “anticipatory” trader who knows the model and trades ahead of everyone else.

A market of fixed-mix funds amplifies every flow

Haghani’s example on the podcast needs no equations. Suppose the only investors are static 50/50 investors holding $50 of stock and $50 of bonds, and companies buy back 10% of all shares. The static investors are the only sellers, so they end up with $45 of stock and $55 of cash. They want to be 50/50 again, and trading with each other cannot change their combined holdings. The only way back is for the price to rise until $45 of stock is worth $55, a 22% rise from a 10% reduction in shares.

The general version comes from Xavier Gabaix and Ralph Koijen’s “inelastic markets hypothesis.” A fund that keeps a fixed share $\theta$ in stocks has a demand elasticity of

where $M$ is the price multiplier: the percentage rise in prices for each 1% of market value that a new buyer absorbs.⁵ A 50/50 fund gives M = 2, which is why a 1% buyback lifts prices about 2%. A 60/40 fund gives 2.5. An 80/20 fund gives 5, the example Gabaix and Koijen use. Adding investors who buy more when stocks get cheaper raises the elasticity; their formula adds a valuation term, $\zeta = 1-\theta+\kappa\delta$.

A static fund still rebalances, buying after a fall and selling after a rise, so it leans against price moves more than a buy-and-hold investor does. It leans much less than a valuation investor. In a market made only of static investors, a change in earnings would not move prices at all, while a change in anyone’s savings would move them a lot.

Try it: the flow multiplier

Set the stock share of the static investors and the size of the buyback or new purchase. With valuation sensitivity at zero the result is exact. With it above zero the calculator uses the linear form from Gabaix and Koijen.

How far $1 of buying moves the market

Estimates of $M$ span two orders of magnitude.

The survey asked: “If a fund buys $1 billion worth of US equities (permanently; it sells bonds to finance that position), slowly over a quarter, how much does the aggregate market value of equities change?” In both samples, 192 responses from economists on social media and 102 from attendees of an online finance seminar, the median answer was zero.⁵ The Haghani et al. estimate sits more than three times below Gabaix and Koijen’s, and far above the survey. Both papers agree on the main point: flows move prices, and the more of the market sits in fixed-mix and return-chasing hands, the more they move.

This framing explains part of the past decade in a way that does not rely on earnings. Net equity issuance by U.S. nonfinancial companies was negative $611 billion in 2023, $407 billion in 2024 and $333 billion in 2025, according to the Federal Reserve’s Financial Accounts.⁶ S&P 500 buybacks set a record of $1.02 trillion in the 12 months to September 2025.⁷ With fewer shares on offer each year and steady retirement-plan contributions buying them, prices have to rise to keep fixed-mix investors at their targets.

Higher prices do not create higher returns. Elm Wealth’s own analysis of buybacks makes this point: “if buybacks raise stock prices, they must also lower forward-looking expected returns.”⁸ The same flows that lifted past returns lower future ones. The process also has a limit. When stocks become cheap to issue relative to debt, companies sell shares. Japanese nonfinancial companies raised ¥10.3 trillion of new equity in 1989, five times the ¥2.1 trillion of 1985, at the top of their bubble.⁹

Return chasers supply the excess volatility

In the model, earnings have 10% annual volatility. A market made only of valuation investors, with earnings as the only shock, would have stock volatility of 7.4%, less than the earnings themselves. The full base case, with static investors, extrapolators, random savings flows and a moving real interest rate, produces 17% stock volatility, close to the U.S. figure for the past 30 years. By the paper’s calculation, roughly 80% of the variance comes from things other than earnings (1 − (7.4/17)²). Remove the extrapolators and volatility drops below the earnings level again.²

The same interaction produces fat tails, even though every random shock in the model is drawn from a normal distribution. Over 30 years of daily returns, a normal distribution would produce about half of a single 4-standard-deviation day. The model produces 38. U.S. stocks from 1997 to 2026 produced 49.

Source: Haghani, Ragulin, Rosenbluth and White (2026), Table 6.

Real-world flows look like the model’s. Patrick Adams of MIT Sloan used more than 25 years of U.S. tax returns and found that high-income households’ flows into and out of stocks “are strongly procyclical,” with large outflows from taxable accounts around the early-2000s and 2008 crashes and steady inflows during expansions.¹⁰ His explanation differs from return chasing, though. Those households lost wage and business income in the crashes and drew on their savings to cover it. Selling at the bottom does the same damage whatever the reason.

None of this mechanism is new. De Long, Shleifer, Summers and Waldmann showed in 1990 that when some traders chase returns, rational speculators can profit by buying ahead of them, which increases volatility.¹¹ The authors say so directly; their contribution is a richer simulation that matches more market facts at once.

Return chasing and momentum look alike and pay opposite

Time-series momentum, holding more of an asset after it has gone up over the past year, has worked in every one of 58 futures markets Moskowitz, Ooi and Pedersen studied.¹² Rob Arnott has called performance chasing “arguably the single-largest cause of tears in the investment arena.”¹³ Both describe buying what has gone up.

The usual reconciliation is time horizon. Cliff Asness has described return chasers as “momentum chasing but at a value time horizon,” acting on three- to five-year returns, where prices tend to reverse.¹⁴ (Our review of Asness’s pet peeves covers that argument.) The paper tested that explanation and found it incomplete. Shortening the extrapolators’ lookback to one year, the same as the momentum investor’s, made them do worse. Momentum investors given three- and five-year lookbacks did not do nearly as badly as extrapolators. The authors name three differences instead:

1. Shape of response. Momentum in the model is binary: 33% or 67% in stocks, depending on whether the past year beat a threshold. Extrapolators adjust continuously, adding a little more stock after each good stretch and selling a little more after each bad one. That gradual buying and selling is what keeps the feedback loop going.
2. Size. Extrapolators hold one-third of the money and move prices. Momentum investors start with almost none and take the trend the extrapolators create.
3. Signal. Extrapolators react to past returns minus the risk-free rate, squeezed into a range. Momentum investors react to raw past returns. The signals differ enough that momentum traders get ahead of the crowd.

The second point matters for anyone buying a trend fund. In one variant, the authors gave momentum investors one-third of starting wealth and removed the extrapolators. The momentum Sharpe ratio fell from 0.41 to −0.01. In the model, momentum pays because it is small and has a large crowd of return chasers to trade against.

Haghani offered a second reason momentum works in stocks: it overlaps with volatility targeting, because markets tend to be calmer when they rise and more volatile when they fall. Moreira and Muir found that cutting exposure when volatility is high raised Sharpe ratios across several factors.¹⁵ Elm tested a bank’s suggestion to buy stocks when the VIX exceeds 30 and found the opposite. From 1990, buying on high VIX produced a Sharpe ratio of 0.47, against 0.50 for a static mix, 0.54 for reducing exposure as volatility rose, and 0.59 for one-year momentum.¹⁶ Haghani’s summary on the podcast separates two forces: higher volatility argues for less stock, and a lower price, which means a higher expected return, argues for more.

What the model and 30 years of data say about each investor

The paper reports Sharpe ratios for each investor type in the simulation, then applies the same rules to actual stock and T-bill returns from 1997 to 2026, the period for which TIPS yields exist.

In the simulation, the ordering is clear: extrapolators 0.14, buy-the-dip 0.20, static 0.25, valuation 0.28, momentum 0.41, value plus momentum 0.53. The anticipatory trader, who knows how everyone else will trade, earns 1.14. Valuation timing alone improves on a fixed mix by only about 12% in the base case. Combining value with momentum is the version that stands out, consistent with Asness, Moskowitz and Pedersen’s finding that the two signals are negatively correlated in nearly every market.¹⁷

The real data are much less clear. Over 1997–2026 in the U.S., the static investor’s Sharpe ratio was 0.479, extrapolators 0.497, momentum 0.520, valuation 0.536 and value plus momentum 0.605. Extrapolators beat the static mix. Outside the U.S. the model’s ordering held better: extrapolators 0.210, static 0.290, value plus momentum 0.429. The authors attribute the extrapolators’ U.S. showing to “sampling variability,” noting that a 30-year Sharpe ratio has a standard error near 0.18. The same caution applies to the strategies that did well: the gap between value plus momentum and the static mix, 0.13, is smaller than that figure. A paired test on the strategies’ return differences would be more precise, since the strategies share most of their risk, and the paper does not report one. The authors describe their tests as establishing “plausibility rather than uniqueness.”

Where we disagree with the paper

The paper closes with a section titled “The passive investing illusion.” It argues that a static 60/40 allocation “embeds the assumption that the equity risk premium is constant,” that it is a sub-optimal approach, and that the growth of balanced and target-date funds makes markets more fragile. Elm Wealth runs a value-and-momentum allocation strategy, and the authors disclose that they have a large share of their own wealth in it.

The fragility argument follows from the model, and the inelastic markets evidence supports it. The investment advice is weaker than the paper presents it, for four reasons:

• The edge is a model output. The clean Sharpe ordering comes from simulations whose investor mix the authors chose. Their preferred mix for today, 40% static, 20% valuation and 40% extrapolators, rests on “anecdotal conversations with investors and our reading of practitioner research.” On real data the gaps are smaller than the paper’s standard error, with no paired test reported.
• The model has no taxes or trading costs. The paper assumes no frictions, and its investors rebalance daily. A taxable investor moving between stocks and bonds on value and momentum signals realizes gains that a fixed mix rebalanced with new savings mostly avoids.
• The edge depends on staying small. The same paper shows momentum turning negative when it holds a third of the money. A strategy whose returns depend on most investors not adopting it cannot be the default recommendation for everyone.
• Behavior is the binding constraint. A fixed target gives an investor a rule to follow after a 40% decline. A dynamic strategy asks them to trust a signal at the moment they are most likely to abandon it, which is how someone ends up acting like the extrapolator in the model.

Our Merton share guide audits Elm’s published backtests of dynamic allocation and finds them real but sensitive to method and smaller after taxes. The paper’s 30-year table fits that reading.

What to do with this as a DIY investor

• Check whether you have been the extrapolator. Pull your stock allocation at the end of each of the past ten years and compare it with the market’s return over the prior few years. If your allocation rose after good stretches and fell after bad ones by more than market drift explains, you have been trading like the group with the worst results in the model.
• Write down a target and a rebalancing band. A fixed mix rebalanced at, say, five percentage points from target is the static investor in the paper. It earned the market’s risk-adjusted return in every test.
• Size any trend or dynamic sleeve modestly. The evidence for trend following is strong across markets. The model adds a reason to keep the sleeve small: it profits from crowd behavior that a bigger crowd of trend followers would erase.
• Ignore the news for allocation decisions. Haghani said on the podcast that an investor needs to follow the news “kind of close to zero.” His firm’s experiment of giving traders the next day’s Wall Street Journal found that nearly half still lost money, mainly by betting too much. See our position sizing guide for the results.
• Expect lower returns after strong flows. If buybacks and retirement contributions have pushed prices up faster than earnings, the starting valuation is higher and the expected return from here is lower. Plan with a forward-looking estimate rather than the trailing ten-year return.

Key Takeaways

• Haghani, Ragulin, Rosenbluth and White model the stock market as valuation investors, fixed-mix investors and return chasers who all size positions with the Merton share and differ only in how they estimate expected returns.
• In their simulated base case, a market of only valuation investors would be less volatile than earnings (7.4% versus 10%). Adding fixed-mix investors, return chasers, savings flows and rate changes produces 17%.
• A fixed-mix fund with stock share θ has a price multiplier of 1/(1−θ): about 2 for a 50/50 fund and 2.5 for 60/40. Published estimates of the market’s multiplier range from 0 (median economist) to 5 (Gabaix and Koijen).
• In the simulation, return chasers earned a Sharpe ratio of 0.14 against 0.25 for a static mix and 0.41 for momentum. Momentum fell to −0.01 when it held a third of the money.
• On U.S. data from 1997 to 2026 the strategies’ Sharpe ratios ranged from 0.48 to 0.61, gaps smaller than the paper’s estimate of sampling error. A written fixed target remains a sound default.

Frequently Asked Questions

What is an extrapolator in investing?

An investor who forms expectations of future returns from recent past returns, and so holds more stock after the market has risen and less after it has fallen. Survey data from Greenwood and Shleifer show that investor expectations behave this way and move opposite to valuation-based forecasts.

Is a 60/40 or target-date fund bad for the market?

It makes the market less elastic: each dollar of new buying or selling moves prices more than it would in a market of valuation investors. That can amplify booms and busts. It is still a reasonable choice for the person holding it, because the fixed rule keeps them from acting like a return chaser, and in the paper the static investor earns the market’s risk-adjusted return.

Why does momentum work if return chasing does not?

In this model, return chasers move gradually and control a large share of the money, so their buying creates trends. A small momentum investor who switches between overweight and underweight on a one-year signal rides those trends without creating them. If momentum investors controlled as much money as return chasers, the model says their edge would disappear.

What is the inelastic markets hypothesis?

Gabaix and Koijen’s finding that aggregate stock demand responds weakly to price, with an elasticity near 0.2. That implies $1 of new buying raises the market’s value by about $5. Much of the inelasticity comes from funds with fixed equity mandates.

Do stock buybacks raise stock prices?

In a market dominated by fixed-mix investors they can, because those investors receive cash and must buy stock back to reach their target. Haghani and White estimate a 3% buyback could lift prices about 7% if a third of investors chase returns. Whatever the size, a higher price means a lower expected return going forward.

Should I switch to a dynamic asset allocation strategy?

Not on the strength of this paper alone. Its strongest results are simulated, the real-data differences are smaller than the paper’s estimate of sampling error, and the model ignores taxes and trading costs. If you want a dynamic element, keep it a defined sleeve with written rules, and keep the core allocation fixed.

Related Guides

• The Merton Share for the sizing formula every investor in this model uses, and an audit of the dynamic version.
• Do You Need Managed Futures? for the evidence on trend following and how to hold it.
• Is the S&P 500 a Momentum Trade? for the four meanings of momentum and why cap weighting is not one of them.
• Bernstein’s Four Signs of a Bubble for extrapolative expectations in bubbles.
• Cliff Asness’s 10 Investing Pet Peeves, Revisited for the case against judging strategies on three- to five-year records.
• Position Sizing for the crystal ball experiment and the Kelly criterion.
• Missing the Best Days for the evidence on trend and volatility timing.

Sources

1. Robert J. Shiller, “Do Stock Prices Move Too Much to be Justified by Subsequent Changes in Dividends?” American Economic Review 71(3), 1981, 421–436. Working-paper version: NBER 456.
2. Victor Haghani, Vladimir Ragulin, Jeffrey Rosenbluth and James White, “Who Killed the Random Walk? How Extrapolators Create Booms, Busts, Trends, and Opportunity,” working paper v0.45, June 15, 2026, forthcoming in the Journal of Investment Management. SSRN. Figures cited: Tables 1–6 and pp. 6–29.
3. Rational Reminder, “Who Causes Stock Market Anomalies? (w/ Victor Haghani),” episode 429, October 1, 2026. rationalreminder.ca.
4. Robin Greenwood and Andrei Shleifer, “Expectations of Returns and Expected Returns,” Review of Financial Studies 27(3), 2014, 714–746. doi.org.
5. Xavier Gabaix and Ralph S. J. Koijen, “In Search of the Origins of Financial Fluctuations: The Inelastic Markets Hypothesis,” NBER Working Paper 28967, 2021. Equations 4 and 9. nber.org.
6. Board of Governors of the Federal Reserve System, Financial Accounts of the United States (Z.1), March 19, 2026 release, Table F.103, nonfinancial corporate business, net equity issuance. federalreserve.gov.
7. S&P Dow Jones Indices, S&P 500 buybacks press release, December 18, 2025 (12 months to September 2025: $1.020 trillion). spglobal.com.
8. Victor Haghani and James White, “The Impact of U.S. Stock Buybacks: Theory vs Practice,” Elm Wealth, October 6, 2025. elmwealth.com.
9. Steven M. Fries, “Japanese Banks and the Asset Price ‘Bubble’,” IMF Working Paper 93/85, November 1993. imf.org.
10. Patrick Adams, “Stocks for the Long Run or Liquidity? Tax Data Evidence and Portfolio Choice Implications,” MIT Sloan working paper, January 7, 2026. patrick-adams.com.
11. J. Bradford De Long, Andrei Shleifer, Lawrence H. Summers and Robert J. Waldmann, “Positive Feedback Investment Strategies and Destabilizing Rational Speculation,” Journal of Finance 45(2), 1990, 379–395. doi.org.
12. Tobias J. Moskowitz, Yao Hua Ooi and Lasse Heje Pedersen, “Time Series Momentum,” Journal of Financial Economics 104(2), 2012, 228–250. doi.org.
13. Dan Weil, “Arnott: Performance Chasing Is Investors’ Greatest Sin,” WealthManagement.com, August 8, 2017. wealthmanagement.com.
14. Cliff Asness, “A Gut Punch,” AQR Cliff’s Perspective, December 22, 2020, footnote 21. aqr.com.
15. Alan Moreira and Tyler Muir, “Volatility-Managed Portfolios,” Journal of Finance 72(4), 2017, 1611–1644. doi.org.
16. Victor Haghani and James White, “When Fear Spikes, Should You Buy?” Elm Wealth, June 30, 2026. elmwealth.com.
17. Clifford S. Asness, Tobias J. Moskowitz and Lasse Heje Pedersen, “Value and Momentum Everywhere,” Journal of Finance 68(3), 2013, 929–985. doi.org.