Understanding Society User Support

Dear Marina,

In the first step, you should divide by the OECD equivalence scale rather than multiplying. Next, if you want your result to be comparable to what is calculated from the FRS, you will need to use the same methods (net/gross of housing benefit) as is used in the reports you mention.

I am just waiting for confirmation about the correct weight so will follow up with that.

Best wishes,
Stephanie

Dear Marina,

A response from our Income team:

I will try to shed some light on the issues around the data.

First, it is possible to construct different income concepts in the FRS. I am not sure which particular one is relevant to you but I am guessing you want to recreate something like the main UK measure of household net income used in official poverty estimates (before housing costs). If this is the case, I would suggest to use fihhmnnet3_dv based on reports from the individual interviews (this is the one the UK government uses in its “income dynamics” publication).

To equivalise the income, you need to divide it by ieqmoecd_dv (note, the official UK income distribution statistics are based on a normalisation of the scale so that a childless couple gets a weight of one (and not 1.5 as is standard). If you want to perform this normalisation you need to multiply the scale by 2/3).

For weighting, if you don’t intend on using any other variables from the main questionnaire, and you want to refer to individuals in the population, I suggest you use the w_indall.dta file (merge your income variable onto the w_indall.dta file) and use the weight: psnenub_xw. Then you will have eg. estimates of median share of income spent on rent for all individuals in the UK (which you could do separately by region etc.).

On the gross vs. net rent issue. This is a little less clear to me and depends on your research question. If you are interested in what proportion of household resources after taxes is devoted to rent then I think gross rent (used with fihhmnnet3_dv) is the one you want. I cant think of a good reason to use the net value. Note separately, you are averaging ratios rather than taking the ratio of averages. I think this makes sense if you are interested in the living costs of individuals rather than the macro-economic picture.

If you havnt seen this reference, it may be of help: https://www.understandingsociety.ac.uk/sites/default/files/downloads/working-papers/2019-08.pdf

And from the Weighting team:

The weight you mentioned is correct if the unit of analysis is a household, which you seem to imply, but if the unit of analysis is a person it would be an enumerated weight.

Best wishes,
Stephanie