Dear IPUMS officers,
I am using IPUMS International data to study the evolution of economies’ industry structure (for example, agriculture -> manufacturing -> services). Could I use INDGEN for the cross-country case? Is it a time-consistent variable?
Thanks!
Xiaofei
Yes, to the extent possible from the original source microdata, INDGEN codes are consistent accross time.
However, before analyzing INDGEN (or any other IPUMS variable), users should study the metadata, particularly the variable description and comparability pages. To access these pages, click the variable nemonic (e.g., INDGEN) on the “Select Data” or “Select Variables” pages. Then, click the tab “description”, “comparability”, etc.
For INDGEN, the description reads:
Description
INDGEN recodes the industrial classifications of the various samples into twelve groups that can be fairly consistently identified across all available samples. The groupings roughly conform to the International Standard Industrial Classification (ISIC). The third digit of INDGEN retains important detail among the service industries that could not be consistently distinguished in all samples.
“Industry” refers to the activity or product of the establishment or sector in which a person worked.
<end>
The comparability metadata includes a statement for each country. For INDGEN the total number of words is 1,825 for the samples currently in the database. The text begins:
Comparability — General
The industry codes are relatively comparable across samples, but there were numerous judgment calls in making the differing industry classifications conform to INDGEN. Not all categories are available in all samples because of the nature of the underlying classifications. The full detail of each original census industry classification is retained un-recorded in the IND variable.
Some service categories were not separately identifiable in all samples, and it may be necessary to combine all the service categories to make certain comparisons. The “other” service group includes miscellaneous personal and community services that are not identified elsewhere in the particular sample in question. This may include such things as sanitary services and entertainment activities.
Repair activities are inconsistently coded. Automobile repair is included with wholesale and retail trade where it is separately identified. Repairs that are clearly for business functions (e.g., office machinery) are coded with “real estate and business services”. All other repairs and those that remain undifferentiated – and may include the functions listed above in some samples – are included with “other services”. Rental activities are also included with “other services” unless they are explicitly of a business nature, in which case they are coded to “real estate and business services”.
The age of the persons to whom the question applies varies across countries. In some samples, the question was asked only of those who were employed at the time of the census. In others, persons were asked to report their last industry, even if they were no longer employed or in the labor force.
Note: An alternative weight variable must be used to yield accurate statistics for 1991 Argentina.
<end>
The “Questionnaire Text” tab displays the wording of the text in English from the original form along with whatever instructions the field workers were given for coding responses to the question.
For each item, there are links to view the entire form either as English text or as an image in the official language.
“With great power comes great responsibility.” Study the metadata thoroughly.