We investigate whether two heuristics, the peak-end rule and herding, lead to cognitive biases in the index of consumer sentiment published by the University of Michigan. Both affect respondents' assessment of changes in their financial position over the past year. Consistent with the peak-end...

We introduce a novel method for training computer algorithms to measure news sentiment. Our approach leverages human-coded sentiment scores from over 200,000 newspaper articles to teach the computer to select words, word combinations, and their linear weights. In an out-of-sample test, examining...