Andy Brownback

Who they are

This page records the public professional record only. Nothing adverse is recorded and no motive is attributed.

The working-paper version, dated January 2019, lists him at the University of Arkansas, and the published record gives no affiliation in the portions read. His wider career was not surveyed for this page, which concerns one paper. That paper is the work-setting evidence in the article What Outcome Bias Does to a Good Call.

Key papers

• Brownback, A., & Kuhn, M. A. (2019). 'Understanding outcome bias.' Games and Economic Behavior, 117, 342-360. doi:10.1016/j.geb.2019.07.003

Key ideas

The experiment sets two kinds of participant against each other. An agent starts with 13 dollars and can buy up to four extra winning numbers for a principal at 50 cents each, and the number bought is the agent's effort. The principal holds one winning number and wins a 6 dollar prize if a six-sided die lands on any of the principal's numbers. The die is the luck. The principal always sees the effort, and can then pay 25 cents for every dollar of the agent's earnings removed, up to 4 dollars. The game ran for 13 periods with random anonymous rematching.

Because effort is fully observed, a principal who punishes only effort would ignore the die. The paper's first result is that principals did not. In the working-paper version, good luck reduced punishment by about a third, an effect similar in size to a rise in effort of 1.4 standard deviations. Punishing low effort did raise later effort, and punishing maximum effort lowered it, so the misdirected punishment had a cost in incentives.

The authors then tested two fixes. In one, principals could commit to view luck or not, or agents could choose whether to reveal it. Principals did not avoid the information when given the chance to commit against it, and agents used the choice strategically, revealing good luck far more often than bad. In the other, independent third parties set the punishment. They were just as biased as principals, with good luck reducing their punishment by about a quarter.

The published abstract states the conclusion that matters for this site: elicited beliefs showed that lucky agents were seen as harder workers than identical unlucky agents, so luck shifted inferences about the agent even when effort was in plain view. The authors argue that disappointment and distributional preferences alone do not explain the bias.

The article takes from the paper the three points its abstract supports: observable effort did not protect agents from luck, neither fix worked, and lucky agents were judged harder workers. The figures above come from the working-paper version and the article does not use them.

How the paper was read

The published abstract was read, and the January 2019 working paper was read for design and figures, up to the point where the text was cut. Results in the published version may differ in detail. The experiment used laboratory and online participants, and the page makes no claim about how the result carries to employment.

What the work is for

The paper belongs to a body of experimental economics that asks how pay, punishment and reputation should depend on results when results are partly luck. The standard advice is to reward effort and skill where they can be seen, and to discount luck. The experiment tests whether people follow that advice in a setting where it costs them nothing to do so, since effort is visible and the die is plainly random. They largely did not, which is why the authors treat outcome bias as a problem for incentives and not only for fairness. If effort earns the same punishment whether or not luck cooperates, the return to effort falls, and an agent has less reason to supply it.

What the paper does not show

It does not show how large the effect is in a firm, where an employer sees a record of many decisions and not one die roll. It does not show that the bias would survive a long relationship, and it does not test any remedy that succeeds. Its participants were a mix of laboratory subjects and online workers playing for small stakes.

---