I have been looking at job search methods in the ASEC. While most job finding methods seem rather smooth from 1968 to 1993, the fraction of unemployed workers who report using a public employment agency has an odd discrete jump upwards in the year 1976 and downwards in 1987 (as do the levels).More importantly, the size of the jumps are exactly the same, so that if you subtracted the size of the jumps from the data, the data woul dlook totally smooth. These discrete jumps remain even after dropping allocated values.
Does anyone have a potential explanation for this? It seems almost certainly like a record keeping issue. What I cannot figure out is why it would only impact this one method and only during these years. There is documentation noting a slight change in data collection in 1976 but not in 1987 (if it had occured in both years, I would say that was the culprit).
Any suggestions or guidance would be greatly appreciated!