nomis - official labour market statistics
By Carl

On 22/08/2011 at 13:51

JSA Onflows: Cohort Survival Rate Analysis

Does anybody have any suggestions of how to derive on-flows, cohort based, survival rates form the data on NOMIS accurately? (by age and gender & time series if possible)

So far, I have tried reverse engineering form CC and Off-flow durations, but it is not as accurate as I would have wanteo assign. From the duration bands, it is not always possible to asign with perfect confidence to the correct monthly on-flow cohort.

Any ideas? Thanks.


By ONS Labour Market

On 25/08/2011 at 15:16
Could you provide us with more details on the anaysis you are trying to perform?
By ONS Labour Market

On 31/08/2011 at 14:00
You are probably not going to be able to do this - Nomis is very good, but it does have its limits on what you can do.

I suspect that to achieve what you are hoping to achieve you will need to track individual cases. Nomis does not store any 'unique identifier' information that would allow a case to be tracked. The underlying data stored within Nomis is just counts, with descriptives (age, sex, durations etc). Nothing is stored with a unique indentifier. Therefore you would not be able to trace these cases through.

At ONS we do have a cohort file with identifier, that might make some of this possible. However, it is only a 5% sample of individuals, which means that the kind of detail that you are looking at is likely to have significant error attached.

I would suggest that to get anything on this, you will need to get in touch with DWP and see whether they can help you.
By paulbivand

On 12/09/2011 at 17:45
There are several ways of doing this, with more or less degree of accuracy. We've used a proportion-at-risk method for estimating the risks of becoming long-term unemployed. DWP has used a 'threshold flows' method (if you google for "threshold flows" you will find something in various places.

These methods then get a little confused because the length of a JSA month or quarter is not uniform. You can standardise off ad on-flows to a 4 1/3 week month, but your stock changes will still be affected by this. There are strong seasonal patterns as well.

So 'accuracy' is within a range determined by the impact of the 4/5 week problem as well as other public holidays etc issues.
By David Walker

On 19/09/2011 at 13:37
You might want to take a look at a recent analysis by GLA Economics that uses longitudinal data from the Labour Force Survey LFS to show flows between employment, unemployment and inactivity from one quarter to the next.
By David Walker

On 19/09/2011 at 13:41
I should have said ONS in partnership with the GLA Intellligence Unit (but brought to my attention by GLA Economics).
By ONS Labour Market

On 19/09/2011 at 14:06
I know it wasn't what Carl was actually after, but if people are interested in the LFS flows information, it is actually being published every quarter as part of the ONS Labour Market Release.

Here is the Labour Market Flows Article for the April-June 2011 period.
By ONS Labour Market

On 19/09/2011 at 14:09
My last link doesn't work, because the link hasn't made the "aprilto-june-2011.html" bit part of the link.

If you go to the Labour Market Release page for September, you will see the Flows article as one of the publications within the Release.

(Hopefully this one will work!)

Create a new reply

Log in to post a reply
About us | News | Release dates