Tuesday, January 11, 2011

JOLTS data

The JOLTS data from which an earlier post was based has been an invaluable source of data and information. There are several areas in which I think it can be improved.

It's sampling frame of 16,000 businesses per month should be enlarged. I say this because the data only allow analysis at region levels rather than state levels. I suspect that the data is not released at the state level because the sampling error would be too large for it to be useful. State level data would also be of interest to state representatives and government agencies.

Which brings me to the sampling error. It needs to be published. I wasn't able to find what the error would be - I'm expecting maybe 5% error? What is worrisome is that with the small sample some firm types might not be fully represented and that the data is biased.

No comments: