Standardizing BLS CES NAICS to IPUMS ACS NAICS codes

This repository attempts to join BLS CES data to 2014-18 IPUMS ACS data for the purposes of estimating PUMA level and higher job losses, and other estimates using ACS data to do so.

Data are organized in the data directory and programs in the scripts directory. Extracts from IPUMS should be sure to include the INDNAICS variable, and both the DDI file usa_xxxxx.xml and the data file should be placed in the data/ipums folder.

Standardizing CES and IPUMS ACS NAICS codes

We went through a manual process to standardize the codes. You can find our decisions and notes on the decisions in this Excel document: https://github.com/UI-Research/ipums-acs-naics-standardization/blob/master/data/manual-files/2017-industry-code-list-ces-crosswalk-manual.xlsx. You can find the machine readable CSV used for analysis here: https://github.com/UI-Research/ipums-acs-naics-standardization/blob/master/data/processed-data/2017-ind-ces-crosswalk.csv.

Be careful when using this dataset. It can be thought of as an instruction manual, where each row is a unique ces_code, ind combination. Values are meant to be operated on at the ces_code level, summarized up to the IND level. The variables produced are as follows:

• IND: 4-digit ACS Industry classification.
• naics: NAICS classification per ACS provided crosswalk.
• ces_code: The 8-digit CES series code from BLS.
• led_code: The 2-digit NAICS code.
• formula_type:
  • single denotes no formula, just one ces_code for the IND
  • formula denotes a formula for using multiple ces_code to summarize at the IND level, with the formula provided by operator
• operator: + or - values denote addition or subtraction necessary to summarize at the IND level. E.g., 32311100 + and 32311200 + for IND = 1070 means that to summarize to IND = 1070, you should add ces_code 32311100 to 32311200.
• series_id: The series_id column from the CES data corresponding to ces_code that should be pulled for seasonally adjusted total employment.
• recency: The number of months this particular series_id lags behind the most recent CES data release. 1 means that, for example, when April data are released, that series_id is just releasing new March data, so it is 1 month behind.
• parent: The parent ces_code for rows where recency = 1 that provide industry categories that contain the ces_code in question, but are more broad, and have recency = 0. This column is for helping adjust ces_code with recency = 1 to the current month using imputation from the broader category.
• parent_series_id: The series_id column from the CES data corresponding to parent that should be pulled for seasonally adjusted total employment.