Discussions

Hi,

We are using the Job Records data to compute the vacancy duration of each job posting. Currently, we define vacancy duration as DELETE_DATE (the most recent time the site was scraped and the posting was no longer found) minus CREATED (the first time the posting was observed and scraped).

However, we observe a noticeable downward trend in average vacancy duration from 2015 to 2025. We are trying to determine whether this pattern reflects actual changes in posting duration or is instead related to the data collection process.

In particular, we were wondering whether the frequency of scraping changed over time. For example, if postings were scraped more frequently in recent years, the measured difference between DELETE_DATE and CREATED could mechanically become shorter. In that case, the longer durations observed in earlier years might partly reflect less frequent scraping rather than true differences in vacancy duration.

We would appreciate any clarification you could provide on this issue, especially regarding whether scraping frequency or parsing procedures changed over the sample period.