David Birch is best known for his seminal work The Job Generation Process, and in the first section this report will be presented in more detail. The report had an enormous impact on the research community as well as among politicians and policymakers. David Birch not only took an interest in which companies created new jobs but also in where these jobs were created, i.e., the regional effects of enterprise. After the first section the book Job Creation in America published in 1987 is presented. David Birch was among the first to emphasize the importance of rapidly growing firms, and he introduced the felicitous term “gazelles” to describe them.
The Job Generation Process (1979)
The Job Generation Process represents, as already mentioned, a genuine groundbreaking report and is one of Birch’s most cited works. It is a research report comprising a mere 52 pages, published within the framework of the MIT Program Neighborhood and Regional Change. In the report Birch aimed at developing an “economic microscope” that could reach beyond aggregate statistics in order to explain how the behavior of individual firms caused employment changes in the US. One of the main problems for Birch was to obtain adequate data; existing databases were not well equipped to cope with large longitudinal data. Birch used the Dun & Bradstreet database originally developed for credit ratings. The research group acquired the complete files for the US as per 31 December 1969, 1972, 1974 and 1976 – containing about 12 million records and over 100 reels of magnetic tape. Considerable efforts were made to reduce the files into a compact set, with all four years merged together, which made it possible to analyze changes in each firm between years. Each establishment was assigned a unique identification number, and the files for the four years were matched on a case-by-case basis.
The database had its limitations. Even if Dun & Bradstreet had a strong incentive to ensure high quality information, the database was not developed for the purpose of studying economic change. Firstly, it was never intended to be a census of the corporate population in the US, i.e., it was a sample, and it made no pretence of covering all businesses. For a variety of reasons, Dun & Bradstreet had concentrated their efforts on manufacturing firms, but improvements were made between 1969 and 1976 to expand coverage in the trade and service sectors as well. Contrary to what could be expected, smaller firms were not underrepresented in the database – smaller firms usually pose greater credit risks than larger ones, and therefore they were well represented in the database. Secondly, the data were collected for credit rating purposes. For example, new firms entered the database as credit information was required of them, which probably created a bias in terms of underreporting new firm births. In addition, there were difficulties separating “branches” – branches are an inherent part of the corporation – and, therefore, were less interesting from a credit point of view. Thus, the Dun & Bradstreet data file could be regarded as a unique resource – it contained a large sample of firms that could be analyzed over time on an individual firm basis. But the file was not designed for analytical purposes and, of course, there were several biases in the database.