Bias comparison · evidence reviewed

Loss Aversion vs Sunk Cost Effect

Both can make stopping, switching or giving something up feel difficult, but they ask different questions. Loss aversion concerns how prospective losses are valued relative to comparable gains around a reference point. Sunk cost effect concerns whether irrecoverable past investment is influencing the current choice.

Loss Aversion – when a loss can carry more weight than an equivalent gainvsSunk Cost Effect – when past costs influence the next choice

The shortest distinction

Ask which information is doing the work: a possible loss from the current reference point, or resources that have already been spent and cannot be recovered?

Question	Loss Aversion	Sunk Cost Effect
Time direction	The judgment concerns prospective gains and losses relative to a reference point.	The decision is influenced by investment that is already in the past and irrecoverable.
Core comparison	Would an equal-sized loss receive more subjective weight than a comparable gain under the same conditions?	Would the current choice change if the past cost had never been incurred but the future options were identical?
Typical project example	Stopping is experienced as accepting or realising a loss, which can make continuation feel more attractive.	The team argues for another investment because too much money or time has already been spent.
Best check	State the reference point and compare matched gain/loss descriptions or absolute outcomes.	Remove irrecoverable past costs from the forward-looking comparison and evaluate only remaining costs, benefits and alternatives.

Decision diagnostic

Review the decision without letting the result rewrite the past.

1
Write the decision from today forward, including realistic future outcomes and alternatives.
2
Mark any past cost that cannot be recovered. If removing that history changes the choice, sunk-cost reasoning is relevant.
3
State the reference point that makes an outcome feel like a gain or loss. Check whether an equivalent gain and loss are being valued differently.
4
Keep both effects separate. A project can contain a sunk cost and also make stopping feel like a prospective loss, but evidence for one does not prove the other.

Same outcome, different bias

Three examples where the distinction matters.

Software migration

Loss AversionStopping the migration is framed as losing the expected future capability that the team had already started treating as theirs.

Sunk Cost EffectThe migration continues mainly because six months of work has already been spent, even though the remaining business case is weak.

Subscription

Loss AversionCancelling a service feels like giving up a benefit that has become the reference point, even when a cheaper alternative offers similar value.

Sunk Cost EffectA customer renews because they already paid a setup fee that cannot be recovered, although that fee does not change next year's value.

Investment review

Loss AversionA possible reduction from the current portfolio value receives more weight than an equal possible increase when comparing future choices.

Sunk Cost EffectNew money is committed because the investor wants to justify or recover an earlier investment that is already lost.

Retrospective review protocol

Separate forecast quality, decision quality, and learning.

Separate past, current and future values before discussing the recommendation.
Remove sunk costs from the forward-looking calculation, while preserving any real future consequences of stopping or switching.
Name the reference point and express equivalent changes as both gains and losses when possible.
Check whether the decision survives when written in absolute outcomes rather than only relative gains or losses.
Describe only the pattern the evidence supports instead of calling any reluctance to stop both loss aversion and sunk cost.

Evidence

Reviewed sources behind this comparison.

Prospect Theory: An Analysis of Decision under Risk1979 · DOI 10.2307/1914185
The psychology of sunk cost1985 · DOI 10.1016/0749-5978(85)90049-4
Meta-analysis of Empirical Estimates of Loss Aversion2024 · DOI 10.1257/jel.20221698
Loss aversion is not robust: A re-meta-analysis2025 · DOI 10.1016/j.joep.2025.102801

Comparison reviewed 2026-08-18. The individual bias pages contain their full evidence status and boundary conditions.