Can I merge basic monthly data with March supplement?

#1

Hello,

I am trying to merge the labor force information from Apr-Jun into the March supplement dataset for observations who are not in their ORG month in March supplement.

For example, Observations with MIS=3 or 7 in March will be in their ORG month in April. However, after writing the program (and using h_id, h_hhnum a_lineno, h_mis, a_sex and a_age to make observations in both mster and using file uniqely identified), I see no matching between two data sets (i.e., _merge=1 or _merge=2).

I want to know what is the problem?

In advance, I appreciate you for your help.

Rasoul

Following shows a general structure of my code.

*STEP 1: Load data for April for h_mis=4 or 8

use e:\data\cpsapr94, clear

do e:\data\Lfvars

keep if (h_mis==4| h_mis==8)

assert h_mis==4 | h_mis==8

sort h_idnum h_hhnum h_mis a_lineno a_age a_sex h_size

destring h_id, replace

egen count=count(1), by(h_idnum h_mis h_hhnum a_lineno a_age a_sex h_size)

tab count

drop if count > 1

saveold e:\data\apr94B,replace

… * STEP 2: Load data for March for Mis = 3 or 7

use e:\data\cpsmar94, clear

do e:\data\Lfvars

keep if (h_mis==3|h_mis==7)

replace h_mis=h_mis+1 if (h_mis==3 |h_mis==7)

assert h_mis==4 | h_mis==8

sort h_idnum h_hhnum a_lineno h_mis a_age a_sex h_size

destring h_id, replace

egen count=count(1), by(h_idnum h_hhnum h_mis a_lineno a_age a_sex h_size )

tab count

drop if count > 1

* STEP 3: Merge April-ORG labor force data to March

merge 1:1 h_idnum h_hhnum a_lineno a_age a_sex h_size using e:\data\apr94B, update

Thsi is my result:

Result # of obs.


not matched 73,779

from master 35,011 (_merge==1)

from using 38,768 (_merge==2)

matched 0

not updated 0 (_merge==3)

missing updated 0 (_merge==4)

nonmissing conflict 0 (_merge==5)


#2

Hi Rasoul,

You’re essentially dealing with two problems:

  1. Merging the “March Supplement” (the ASEC) with the March basic/monthly

  2. Merging the March basic with the April basic

Regarding problem #1, it’s a notoriously tricky problem. MPC researchers Jose Pacas & Sarah Flood are working on it, and from 1989 onward, you can follow their algorithm described here. A number of people (such as this thread and this thread) have asked when this variable will be available in the standard IPUMS release, but as of yet, there seems to be no word.

Regarding problem #2, it depends on which years you’re using since the specific set of variables you need for cross-month merging changes over time. You should have a look at some of the references in Pacas & Flood, particularly Drew, Flood, and Warren (2014). They discuss this issue in depth. But IPUMS has created a variable to do this merging (CPSIDP).

As a general comment, your variable names make it clear that you’re not using IPUMS data. Usually, MPC researchers don’t respond to general CPS questions, particularly because different data providers use different variables and it’s impossible to keep track of them all. I suggest using IPUMS, or figuring out a way to ask your question in terms of IPUMS variables.

Hope this helps,

Mitch