Why are there so many missing values in 'HRSWORKORG' IPUMS variable within The outgoing rotation group?


I recently attained all the 2010 basic monthly CPS surveys from IPUMS and selected out of these only households which are part of the earner studay. ‘Earnweek’ has no missing values once this is done and since I’m looking for measures on earnings from curent occupation this measure seems to fit. However, to complete the picture I also need a measure of ‘hours worked in a usual week in current occupation’. If I’m not mistaken the only such measured offered in CPS basic monthly surveys (or in general in fact) is the variable called ‘HRSWORKORG’. The problem is that a large per cent of the sample gets coded ‘NIU’ (around 92,300 cases when looking at 16+). I understand that only those who are paid by the hour get a value on this measure in general (although that still leaves us with an additional ~25,000 cases that are missing for other reasons) but that is a big chunk of the sample to lose.

My questions are:

  1. Other than using ‘HOURWAGE’ to revearse engineer weekly working hours (solves the same 25000 cases mentioned), is there any way to get some of these missing values filled (without using Multiple Imputation and the such)?

  2. I saw that in the parallel NBER extracts of the famous CPS Annual Earnings File the amount of missing values in the ‘hours of work’ variable (called ‘uhourse’) is much lower (only around 10,660 cases). Is this the same variable as ‘HRSWORKORG’ (weekly hours of work in current job)?

  3. If so, how did they do that? Are these imputed values or are they using additional data IPUMS does not include in the data to deduce hours of work? Are there any documents explaining the process?


The IPUMS-CPS HRSWORKORG variable corresponds to the CPS basic monthly variable “PEERNHRO”, which records the usual hours worked for those who reported a pay periodicity of hourly (PEERNPER=1). The CPS basic monthly variable PEHRUSL1 records the usual hours worked for non-hourly workers. Unfortunately, IPUMS-CPS does not currently provide an equivalent for PEHRUSL1. We are expanding our basic monthly offerings, however, and usual hours worked for non-hourly workers should eventually be included in a future data release. In the meantime, I recommend linking your IPUMS extract to the original CPS data. Because the NBER and IPUMS-CPS files share a common sort order, you can sequentially merge the data files (i.e. first NBER observation to first CPS observation, second to second, etc). I recommend checking values of sex, age, and race before and after merging the data.

I hope this helps, and I apologize for any inconvenience this may have caused.