Choice Bracketing
Choice bracketing is the name Daniel Read, George Loewenstein and Matthew Rabin gave in 1999 to a question that every decision raises: which other decisions are being considered at the same time? Choices made one at a time are bracketed narrowly. Choices made together are bracketed broadly. The vocabulary is the economists' sibling of narrow framing and broad framing, the terms Daniel Kahneman and Dan Lovallo used in 1993.
What it is
The 1999 paper, "Choice Bracketing", appeared in the Journal of Risk and Uncertainty (volume 19, 171 to 197). Its subject is the grouping of decisions. A person facing several choices can treat each as its own problem, or treat the set as one problem and choose the best combination. The first is narrow bracketing and the second is broad bracketing.
The paper's first author is Daniel Read, and the paper is often credited to Rabin or to Kahneman in passing. This page uses it for its vocabulary.
Narrow framing and narrow bracketing describe the same habit from two literatures. Kahneman and Lovallo (1993) wrote of narrow decision frames and isolation errors. Read, Loewenstein and Rabin wrote of bracketing. Thinking, Fast and Slow uses the framing words, and the later experimental work uses the bracketing words.
In effect
The cleanest test of narrow bracketing is Rabin and Weizsäcker's 2009 paper in the American Economic Review, Narrow Bracketing and Dominated Choices (volume 99, issue 4, 1508 to 1543). The authors show that a decision maker who evaluates decisions separately will sometimes choose an option that is first-order stochastically dominated, which means a combination that is no better in any outcome and worse in some, provided the person's risk preferences are not of one particular kind, constant absolute risk aversion. The result holds even for a simple pair of independent binary decisions.
They tested it in a real-stakes laboratory experiment that replicates Tversky and Kahneman's 1981 pair of decisions. In that experiment 28 per cent of participants chose dominated combinations. A representative survey of hypothetical large-stakes choices produced higher rates, and the rates varied little with personal characteristics. From average preferences that look like those of prospect theory, the authors estimate that 89 per cent of people bracket narrowly. That figure comes out of their model and is not a count of anyone's choices.
The pair of decisions itself is described in the article What is narrow framing?: a sure gain of $240 against a 25 per cent chance of $1,000, and a sure loss of $750 against a 75 per cent chance of losing $1,000. Narrowly bracketed, most people take the sure gain and the gamble. Broadly bracketed, the combination they reject pays $10 more in both branches.
What it does not say
It does not say people always bracket narrowly. The authors' own estimate is a share of people, taken from a model, and the laboratory rate of dominated choices was 28 per cent, not all participants.
It does not say a broad bracket is always the better one. Kahneman's account of the remedy, a risk policy, carries three limits: the gambles must be independent, a loss must not threaten total wealth, and long shots are excluded. The same limits apply to any advice to bracket broadly.
It does not rest on a full reading of the 1999 paper here. The paper's own treatment of how bracketing varies by situation is not reported on this page.
Sources
- Read, D., Loewenstein, G., & Rabin, M. (1999). "Choice bracketing." Journal of Risk and Uncertainty, 19(1 to 3), 171 to 197. doi:10.1023/A:1007879411489.
- Rabin, M., & Weizsäcker, G. (2009). "Narrow bracketing and dominated choices." American Economic Review, 99(4), 1508 to 1543. doi:10.1257/aer.99.4.1508.
- Tversky, A., & Kahneman, D. (1981). "The framing of decisions and the psychology of choice." Science, 211(4481), 453 to 458.
- Kahneman, D. (2011). Thinking, Fast and Slow.