Abstract:
We use more than 63,000 datapoints from a German used car market website to document systematic and substantial price drops at vintage (= year of first registration) thresholds and 10,000 km odometer marks. The latter finding replicates the findings in Lacetera et al. (2012), whereas the first dimension cannot be analyzed with their US data because only German cars have such legally mandated and regulated “birthdates”. Hence we have the unique opportunity to study the presence of coarse information processing within the same dataset and decision problem but across two separate domains. We document that discontinuities in these two domains are of comparable size. While Lacetera et al. (2012) explain
their result with a left-digit bias in the processing of numerical information, vintage discontinuities cannot be explained by this. We propose a slightly more general model of information prominence and availability bias to accommodate our findings.
We continue to show evidence from a natural field experiment in another online market to disentangle the effects of inattention and "standard" search costs to such discontinuities. We finally document similar discontinuities in German wage data and show that these disontinuities are linked to economically important outcomes like job satisfaction and job search.