Manufacturing: How we collected the data
Thursday, June 14th, 2012
Two years ago, I asked students in a business journalism class at American University to pull together an unusually broad and deep dataset that tracked manufacturing employment and total employment by state over the past 40 years. It was their work that provided most of the data we present here.
On the surface, it sounds simple, but it wasn’t.
First, there was the issue of finding a source for the material. Eventually, we discovered that the Statistical Abstract of the United States contains a table with the data we were seeking. The Statistical Abstract is one of the wonders of the modern world; it has some data that goes back to 1789. To make it even better, the Census Bureau, which collects and assembles the information in the Statistical Abstract, has scanned every edition and made it available on line. Unfortunately, these are scanned PDFs, organized by year and not searchable. So the student researchers, each of whom had about five states to work on, had to go through each volume by hand and enter the numbers into a spreadsheet.
They had to do that for about 20 years' worth of data. Beginning in the early 1990s, the Bureau of Labor Statistics had the numbers available online, which at least made finding the data somewhat easier, once the students figured out how the BLS had organized the information and which tables to use.
The numbers we are using are the annual average of manufacturing workers by state as calculated by the Bureau of Labor Statistics in its State and Area Employment, Hours, and Earnings report. These are not seasonally adjusted. Even though there have been significant changes to the types of firms that are considered “manufacturing,” the data do not reflect those changes. It is likely that the decline in “manufacturing” employment is partly a result of these reclassifications.
The students who did the basic research were Jennifer Calantone, Joshua Cook, Kevin Eng, Alicia Guidi, Kim Ha, Fahima Haque, James Hascup, Gabrielle Jones, Anel Ramazanova, Jessica Rybka, Kristin Tangel and Alyssa Wolice. Without their effort and energy, we never could have done this project.
Inevitably, there are gaps in any dataset like this. So Workshop researcher and graduate assistant Carrie McCloud picked up the project in the fall of 2010 to continue gathering and perfecting the data. In the fall of 2011, Monica Arpino, a Workshop researcher and graduate assistant, took over the dataset and saw it to completion. We are grateful for their invaluable assistance.