Limits to Arbitrage: Why Market Mispricings Can Persist Indefinitely

The Trade That Was Right and Still Lost

In the spring of 1998, Long-Term Capital Management held one of the most diversified convergence portfolios ever assembled. Its positions spanned sovereign bonds, equity volatility, swap spreads, and mortgage-backed securities across dozens of countries. The firm's principals included a former Federal Reserve vice chairman and two future Nobel laureates in economics. The trades themselves were textbook arbitrage: buy the cheap security, sell the expensive one, and wait for the gap to close.

By September 1998, LTCM had lost $4.6 billion. Not because the fundamental analysis was wrong — most of the convergence bets eventually proved correct — but because the firm ran out of capital before the mispricings corrected. As Russia defaulted and global markets seized, spreads that should have narrowed instead blew out to unprecedented levels. Counterparties demanded more collateral. Investors panicked. The very act of unwinding positions to meet margin calls pushed prices further against the fund, widening the dislocations that the positions were designed to exploit.

One year earlier, Andrei Shleifer and Robert Vishny had published a paper that explained exactly why this could happen. Their 1997 article in the Journal of Finance, "The Limits of Arbitrage," dismantled a foundational assumption of financial economics: that rational arbitrageurs with unlimited capital and infinite patience would rapidly correct any deviation from fair value. In reality, Shleifer and Vishny argued, the people best equipped to eliminate mispricings operate under constraints that can prevent them from doing so — and those constraints bind hardest at the worst possible moments.

The Efficient Market Assumption and Its Weak Link

The efficient market hypothesis rests on a compelling chain of logic. If a security trades below its fundamental value, arbitrageurs will buy it. Their buying pressure will push the price back toward fair value. In equilibrium, no persistent mispricings should exist because the profit motive ensures they are corrected almost as soon as they appear.

Shleifer and Vishny identified the crucial gap in this reasoning. The argument assumes that arbitrageurs are unconstrained: they can deploy unlimited capital, they can hold positions indefinitely, and they are not subject to external evaluation. None of these assumptions holds in practice. Most arbitrage capital in real financial markets is managed by professionals — hedge fund managers, proprietary trading desks, specialist fund managers — who invest other people's money. And the providers of that money evaluate their managers based on recent returns.

This observation, seemingly obvious once stated, has radical implications.

The Shleifer-Vishny Model: How Capital Flows Undermine Correction

The core mechanism of the Shleifer-Vishny framework can be stated in three steps.

First, specialized arbitrageurs identify a mispricing and take a position against it. They might buy an underpriced security, short an overpriced one, or construct a relative-value trade that profits when the gap closes.

Second, the mispricing temporarily worsens rather than correcting. This can happen for any number of reasons: noise trader sentiment shifts, correlated selling by other distressed investors, or macroeconomic shocks that have nothing to do with the fundamental value of the position.

Third, the arbitrageur's investors observe the short-term losses. Unable to distinguish between a manager who is wrong and a manager who is early, they withdraw capital. The arbitrageur must reduce positions — selling into weakness, buying back shorts at elevated prices — precisely when the expected return on the trade is at its maximum.

The result is what Shleifer and Vishny called performance-based arbitrage: a system in which capital allocation to corrective trading is a decreasing function of the magnitude of the mispricing. The worse the mispricing gets, the less capital is available to correct it.

This mechanism is not a quirk of poorly designed funds. It reflects a fundamental agency problem in delegated portfolio management. Investors cannot perfectly observe the quality of an arbitrageur's positions. When they see losses, the rational response given their informational disadvantage is to reduce exposure. From the perspective of any individual investor, withdrawing after losses is prudent. In aggregate, these individually rational decisions produce a collectively irrational outcome: the evaporation of corrective capital when it is most needed.

Noise Trader Risk: The Amplifier

Shleifer and Vishny's model built on earlier theoretical work by De Long, Shleifer, Summers, and Waldmann (1990), who formalized the concept of noise trader risk. In their framework, markets contain two types of participants: rational arbitrageurs who trade on fundamentals, and noise traders who trade on sentiment, momentum signals, or flawed beliefs.

The key insight is that noise traders are not merely a source of static in an otherwise efficient market. Their collective behavior creates a genuine risk for arbitrageurs. If sentiment-driven selling pushes a stock below fair value, a rational arbitrageur who buys will profit if prices eventually revert. But "eventually" is the operative word. Before convergence, noise traders might become even more bearish, pushing the price lower still. An arbitrageur with a one-year horizon cannot afford to hold a position that loses 30 percent before it recovers, even if the five-year expected return is strongly positive.

Noise trader risk is distinct from fundamental risk. An arbitrageur shorting an overpriced stock faces fundamental risk: the company might genuinely be worth more than estimated. Noise trader risk operates even when the fundamental analysis is unambiguously correct. The danger is not that the arbitrageur is wrong about value. The danger is that irrational participants will make the price even more wrong before it becomes right.

Empirical Evidence: From Theory to Observed Markets

The Shleifer-Vishny framework gained credibility not through abstract elegance but through its ability to explain real-world phenomena that the standard efficient-market framework could not.

Pontiff (1996) provided early empirical support using closed-end funds. These funds trade on exchanges at prices that frequently deviate from their net asset values. In a frictionless market, arbitrageurs would buy discounted funds and short the underlying portfolios, driving discounts to zero. Pontiff showed that closed-end fund discounts were larger and more persistent when the costs of arbitrage — specifically, the idiosyncratic volatility of the fund relative to its assets — were higher. In other words, the harder it was to arbitrage, the wider the mispricings grew.

Pontiff (2006) extended this logic across the broader anomaly literature. He demonstrated that well-known return anomalies, the predictable patterns that appear to violate market efficiency, generate their largest profits among stocks with high idiosyncratic volatility. This finding is precisely what the limits-to-arbitrage framework predicts: mispricings persist where arbitrage is most costly and risky. If anomaly returns were driven by risk compensation rather than mispricing, there would be no reason for them to concentrate in hard-to-arbitrage securities.

Mitchell, Pulvino, and Stafford (2002) documented a particularly striking violation of the law of one price. During the technology bubble, several companies traded at prices that implied negative value for their non-internet subsidiaries. In the most famous case, 3Com's market capitalization, after subtracting the value of its Palm stake following the Palm IPO, implied that the rest of 3Com's business was worth negative $22 billion. This was not a subtle mispricing requiring sophisticated models to detect. It was an arithmetic absurdity visible to anyone with a calculator. Yet the mispricing persisted for months because short-selling constraints prevented arbitrageurs from establishing the positions needed to correct it.

The Anatomy of a Limits-to-Arbitrage Episode

The LTCM collapse exemplifies the Shleifer-Vishny mechanism in compressed form, but the pattern has repeated across multiple episodes.

During the 2007-2008 financial crisis, quantitative equity hedge funds experienced a similar dynamic. In August 2007, several multi-billion-dollar quant funds suffered sudden, correlated losses over a period of days. The losses were concentrated in their most crowded factor positions. As one fund began deleveraging, the selling pressure hit positions shared with other funds, generating losses that forced further deleveraging across the industry. Liu and Longstaff (2004) had shown theoretically that even textbook arbitrage opportunities can generate interim losses large enough to bankrupt an arbitrageur before convergence occurs.

The pattern has a consistent structure across episodes:

An external shock (Russian default, subprime losses, pandemic) triggers initial losses on arbitrage positions. Investors in arbitrage funds observe the losses and begin redeeming capital. Fund managers sell positions to meet redemptions, generating additional selling pressure. The selling pressure widens the very mispricings the positions were designed to exploit. Wider mispricings generate further losses, triggering more redemptions.

This feedback loop between performance, capital flows, and market prices is the core prediction of the Shleifer-Vishny model. It transforms what should be a stabilizing force (arbitrage) into an amplifying one.

The Role of Funding and Leverage

The original Shleifer-Vishny model focused on the investor-manager relationship. Subsequent work integrated their insights with funding-market dynamics, creating a richer picture of how arbitrage constraints operate.

Gromb and Vayanos (2010) surveyed the theoretical literature on limits to arbitrage and identified several categories of constraint: fundamental risk (uncertainty about true value), noise trader risk (sentiment-driven price movements), transaction costs, short-selling constraints, funding liquidity constraints (difficulty borrowing to finance positions), and the agency problems that Shleifer and Vishny originally emphasized.

These constraints interact in nonlinear ways. A fund using leverage faces both performance-based capital risk from investors and margin-based capital risk from prime brokers. During periods of market stress, both channels tighten simultaneously. Investors redeem because past performance has deteriorated. Brokers raise margin requirements because asset volatility has increased. The fund faces a two-front squeeze that neither constraint alone would produce.

Implications for Market Efficiency

The limits-to-arbitrage framework does not claim that markets are entirely inefficient. It offers a more precise statement: markets are efficient to the degree that well-capitalized, patient, unconstrained investors operate within them. In markets dominated by such investors, mispricings will be small and short-lived. In markets where arbitrage capital is scarce, constrained, or volatile, mispricings can persist for extended periods.

This perspective resolves an apparent paradox in the empirical literature. Many anomalies, including value, momentum, and various accounting-based signals, have been documented for decades. Under the strict efficient market hypothesis, these patterns should have disappeared once they were published and investors could exploit them. McLean and Pontiff (2016) found that anomaly returns do decline after publication — but they do not disappear entirely. The partial decay is consistent with limits to arbitrage: publication attracts some corrective capital, but not enough to fully eliminate the mispricing, especially in the hard-to-arbitrage segments of the market where anomalies are most concentrated.

The framework also explains cross-sectional variation in anomaly strength. Mispricings should be larger and more persistent among securities that are difficult to arbitrage: small-cap stocks with low liquidity, stocks with high borrowing costs for short sellers, and stocks with high idiosyncratic volatility that make positions risky to hold. The empirical evidence broadly confirms these predictions.

What This Means for Portfolio Construction

For individual investors and institutional allocators, the limits-to-arbitrage framework carries several practical lessons.

The first is about position sizing and time horizons. Even a correct view on valuation can generate significant interim losses. Position sizes should be calibrated not to the expected return but to the maximum drawdown the investor can tolerate without being forced to liquidate. The question is not just whether a trade will work, but whether you can survive the path to profitability.

The second is about liquidity and leverage. The Shleifer-Vishny mechanism is most devastating for investors who are leveraged and illiquid. A fully funded position in a mispriced security carries noise trader risk but not forced-liquidation risk. An investor who can add to positions when mispricings widen, rather than being forced to cut, holds a structural advantage.

The third is about the source of returns in factor strategies. Many factor premia may persist partly because of limits to arbitrage. The value premium, for instance, requires buying unpopular stocks that may underperform for years before reverting. The carrying costs, reputational risk, and career risk of maintaining such positions deter professional managers, allowing the premium to persist. Investors with long horizons and no agency problems, such as endowments or patient individual investors, are better positioned to capture these premia precisely because they face fewer limits to arbitrage.

Unresolved Questions

Several open debates remain within the limits-to-arbitrage literature. One concerns the distinction between risk-based and mispricing-based explanations for anomaly returns. Limits to arbitrage explains why mispricings persist, but it is often difficult to determine empirically whether a given return pattern reflects a mispricing that arbitrageurs cannot correct or a fair compensation for risk that arbitrageurs are bearing.

Another question concerns the stability of the framework over time. As markets have evolved — with more capital in quantitative strategies, lower transaction costs, and faster information dissemination — have limits to arbitrage diminished? The evidence is mixed. Some anomalies have weakened, consistent with reduced arbitrage constraints. But new forms of limits have emerged, including crowding risk (where many funds exploit the same opportunities) and complexity risk (where strategies require sophisticated models that may themselves contain errors).

The events of 2020 and 2021, when retail investor activity through platforms like Robinhood created enormous short-term dislocations in heavily shorted stocks, demonstrated that noise trader risk remains a potent force even in an era of algorithmic trading and abundant institutional capital.

A Framework for Navigating Imperfect Markets

Shleifer and Vishny did not prove that markets are inefficient. They proved something more useful: they identified the specific conditions under which inefficiency is most likely to persist. Mispricings survive when arbitrage is costly, when capital is constrained, when noise traders are active, and when the agents best positioned to correct prices face evaluation based on short-term results rather than fundamental accuracy.

For investors, the practical takeaway is that identifying a mispricing is necessary but not sufficient for profitable trading. The equally important question is whether you possess the capital structure, time horizon, and psychological resilience to maintain the position through the period — possibly extended — during which the market may disagree with you.