February 1, 2004
Danny Kahneman’s experimental work is justifiably famous. Individuals report their experienced utility at various points in time throughout a number of different events. Their choices of which experiences to repeat do not always maximise total enjoyment or minimise total pain. Decisions are better described by the average of the peak experience and the final experience. Decision utility is thus a transformation of experienced (momentary) utility, rather than its sum, as might have been supposed. This phenomenon, known as peak-end theory, has been verified in a number of different experimental settings. It has not been applied in large-scale, long-run settings. We apply peakend theory to the decision to quit a job. We use job spell data from the BHPS and the GSOEP, in which there are a number of different observations on the same job spell. Our results show that the peak-end transformation of job satisfaction is the best predictor of quits. Job satisfaction at time t is therefore best thought of as experienced rather than decision utility.