Stochastic Dominance

Investment A was the correct choice because it stochastically dominates Investment B. This is a formal way of saying that one option is unambiguously better than another because it provides an outcome that is at least as good in every possible scenario, and strictly better in at least one.

In the example, both investments gave you the same great outcome ($2,000) half the time. But in the less-great scenario, Investment A gave you $1,100 while B gave you only $1,000. Since A is never worse and sometimes better, it's the dominant choice for any rational person.

Ask this: 'Is there any possible situation where the other option would be better?' If the answer is no, you've found a dominant choice.

Stochastic dominance provides a powerful way to filter choices under uncertainty without needing to calculate your exact risk tolerance. The core idea is to compare the entire distribution of possible outcomes, not just the average.

There are two main types you'll encounter:

First-Order Stochastic Dominance (FOSD) is the most straightforward. The investment problem you just solved is a classic example. An option has FOSD if, for any target outcome, it gives you an equal or greater chance of achieving it. Any rational person, whether they love or hate risk, should choose the dominant option.

For example, when choosing between two suppliers, Supplier A delivers on time 95% of the time, while Supplier B delivers on time 90% of the time. Assuming quality is equal, Supplier A's performance has first-order dominance.

Second-Order Stochastic Dominance (SOSD) is used when two options have the same average outcome but different levels of risk. If you are risk-averse, you will always prefer the option that is less 'spread out'.

Imagine two career paths with the same expected lifetime earnings. Path A is a stable corporate job with predictable raises. Path B is founding a startup, with a small chance of extreme wealth and a large chance of modest earnings. For most risk-averse people, the stable corporate job (Path A) has second-order dominance.

Stochastic dominance is a cornerstone of decision theory, providing rules for ranking uncertain prospects without full knowledge of an individual's utility function. It establishes a partial ordering over probability distributions based on general properties of preferences, such as 'more is better' (first-order) or risk aversion (second-order).

• Hadar & Russell (1969) and Hanoch & Levy (1969) independently established the foundational theorems, formally linking first-order dominance to all non-satiated decision-makers and second-order dominance to all risk-averse ones. [1, 2]
• Levy (2015) provides a modern, comprehensive overview of the theory and its applications in finance, demonstrating its use in portfolio selection, option pricing, and capital budgeting under uncertainty. [3]
• This framework allows economists to make robust welfare comparisons; for example, if one economic policy leads to an income distribution that stochastically dominates another, it can be judged as superior without needing to agree on a specific social welfare function. [3]

Stochastic dominance is a powerful but not universally applicable tool.

• Incomplete Ordering: Its biggest limitation is that it often fails to rank options. In many cases, neither option A nor option B will dominate the other, particularly if one has a higher upside and the other has a less severe downside.

• Data Intensity: Applying the rules formally requires knowledge of the entire probability distribution of outcomes for each choice. In the real world, this information is rarely available and must be estimated.

• Behavioral Blind Spots: The theory assumes rational actors. Research from behavioral economics shows that factors like framing effects, loss aversion, and probability weighting can lead people to choose stochastically dominated options.