• Corpus ID: 17241886

Costly Information Processing and Income Expectations

@inproceedings{Gutknecht2014CostlyIP,
  title={Costly Information Processing and Income Expectations},
  author={Daniel Gutknecht and Stefan Hoderlein and Michael Peters},
  year={2014}
}

• Daniel Gutknecht, Stefan Hoderlein, Michael Peters
• Published 1 October 2014
• Economics, Computer Science

Do individuals use all information at their disposal when forming expectations about future events? In this paper we present an econometric framework to answer this question. We show how individual information sets can be characterized by simple nonparametric exclusion restrictions and provide a quantile based test for costly information processing. In particular, our methodology does not require individuals' expectations to be rational, and we explicitly allow for individuals to have access to sources of information which the econometrician cannot observe. As an application, we use microdata on individual income expectations to study the information agents employ when forecasting future earnings. Consistent with models where information processing is costly, we find that individuals' information sets are coarse in that valuable information is discarded. To quantify the utility costs, we calibrate a standard consumption life-cycle model. Consumers would be willing to pay 0.04% of their permanent income to incorporate the econometrician’s information set in their forecasts. This represents a lower bound on the costs of information processing.