Expected shortfall, also called conditional value-at-risk (CVaR), measures the average loss an investor would face in the worst cases beyond a chosen threshold. Where value-at-risk marks a line and asks how often losses cross it, expected shortfall looks past that line and asks a sharper question: if losses do cross it, how bad do they get on average?

The distinction matters because two portfolios can share the same value-at-risk yet behave very differently once that point is breached. One might lose a little more than the threshold; the other might lose a great deal more. Value-at-risk treats both the same, while expected shortfall separates them by averaging the depth of the losses in the tail. This focus on the severity of bad outcomes, not just their frequency, is the reason expected shortfall has become a standard companion to value-at-risk in risk reporting.

Definition

Expected shortfall at a given confidence level is the average of all losses that fall in the worst tail of the outcome distribution beyond the value-at-risk (VaR) point. If the value-at-risk at the 95% level is the loss that is exceeded only 5% of the time, then the 95% expected shortfall is the average size of the losses within that worst 5%. The measure therefore reads the entire shape of the tail rather than a single cutoff.

Because it averages over the tail, expected shortfall is always at least as large as the value-at-risk at the same confidence level, and usually larger. The gap between the two reflects how heavy or stretched the tail is. A distribution with fat tails (more frequent extreme outcomes than a normal bell curve predicts) tends to show a wide gap, because the rare losses that do occur can be far deeper than the threshold itself.

How It Works

Estimating expected shortfall starts with a distribution of possible portfolio outcomes, which can come from historical returns, a fitted statistical model, or a Monte Carlo simulation (a method that generates many random scenarios). The analyst then locates the value-at-risk threshold for the chosen confidence level and averages every loss that lands beyond it. The result is a single number expressing the typical depth of a bad-case loss.

The confidence level and the time horizon both shape the figure. A 99% expected shortfall reaches deeper into the tail than a 95% figure, so it produces a larger and more conservative number. Likewise, a longer horizon allows losses to compound, which widens the tail. Because the measure depends heavily on how the tail is modeled, the same portfolio can produce different expected shortfall estimates under different distributional assumptions, an issue that connects directly to its limitations.

Expected Shortfall Compared to Value-at-Risk

Applications

Expected shortfall is used wherever the severity of rare losses matters more than their frequency. Risk teams apply it in stress testing and scenario analysis to understand how a portfolio behaves under adverse conditions, not just how often those conditions arise. It also serves as an input to position sizing decisions, since constraining the average tail loss can help shape how much capital is committed to risk-bearing positions.

The measure has gained ground in regulatory and institutional practice as well. International banking frameworks moved toward expected shortfall as the basis for market risk capital, citing its coherence and its sensitivity to the depth of losses. For an investor or analyst, the practical appeal is the same: a number that distinguishes between a portfolio that loses a little past the threshold and one that can lose a great deal, even when their value-at-risk figures look identical.

Known Limitations

Academic Origin

Expected shortfall emerged from a broader effort to define what makes a risk measure sound. In 1999, Artzner, Delbaen, Eber, and Heath set out four properties that a sensible risk measure should satisfy, introducing the idea of a coherent risk measure and showing that value-at-risk fails the sub-additivity requirement. This work framed the central weakness that expected shortfall was designed to address.

Rockafellar and Uryasev then gave conditional value-at-risk a practical foundation in 2000, showing how it could be computed and optimized efficiently, even within portfolio construction problems. Their formulation made the measure usable in practice, not just appealing in theory, and helped expected shortfall move from an academic preference into mainstream risk management and regulation.

Further Reading

• Artzner, P., Delbaen, F., Eber, J.-M. and Heath, D. (1999). "Coherent Measures of Risk." Mathematical Finance, 9(3), 203–228.
• Rockafellar, R.T. and Uryasev, S. (2000). "Optimization of Conditional Value-at-Risk." Journal of Risk, 2(3), 21–41.
• McNeil, A.J., Frey, R. and Embrechts, P. (2015). Quantitative Risk Management: Concepts, Techniques and Tools. Revised ed. Princeton University Press.

Related Terms