We provide here the rationale under which our microclass scheme was developed for the United States,Japan, Germany, and Sweden. The microclass category may be defined as a grouping of technically similar jobs that is institutionalized in the labor market through such means as (a) an association or union, (b) licensing or certification requirements, or (c) widely diffused understandings … regarding efficient or otherwise preferred ways of organizing production and dividing labor. The scheme applied in our mobility research includes 82 micro-classes and captures many of the boundaries in the division of labor that are socially recognized and defended.
How much comparability? In constructing the scheme, we sought to ensure that the jobs constituting each category were comparable across countries, although inevitably some minor incomparabilities had to be tolerated because the source classification schemes were not detailed enough or because of real cross-national differences in how the division of labor is constructed. The Japanese classification was quite idiosyncratic and sometimes difficult to reconcile with the others, but for the most part the same detailed occupations could be identified even in Japan. This isomorphism, to the extent that it held, may be traced to three sources: (a) each country independently settled on the same way of dividing labor and defining occupations (perhaps owing to the "efficiency" of that shared solution); (b) a particular solution to the division of labor diffused across countries; or (c) a shared classification scheme diffused among statisticians, sociologists, and other classifiers even though it mapped only imperfectly onto the actual division of labor. While the latter, artifactual source of cross-national similarity is no doubt partly at work, there is clearly a real isomorphism in the division of labor producing many occupations that are deeply institutionalized (e.g., architect, electrician, miner). For such categories, the residual inconsistencies in coding appear to be quite small, and such cross-national differences as emerge in our data almost certainly signal real rather than artifactual variability.
Self-employment. The scheme we have developed does not distinguish self-employed and employed workers. We have, to be sure, coded storekeepers as "proprietors" and distinguished farmers from farm laborers, but otherwise the occupational affiliation takes precedence and employed and self-employed workers are combined in a single category. This raises the possibility that, for occupations with substantial self-employment, high rates of inheritance may be generated not because the occupation has unusual holding power but because of the well-known holding power of self-employment itself.
Layering effects. The distinctive feature of our analysis is that micro-class effects are layered over more conventional big-class effects. Given our suspicion that net big-class effects may be weak, it is clearly important to adopt a big-class scheme that fully captures such big-class effects as can be found, as otherwise any possible shortfall in big-class explanatory power might be attributed to a poor operationalization. We have accordingly proceeded by fitting a multiplicity of nested big-class contrasts that capture the many and varied big-class distinctions that scholars have identified. As shown in the table below, we begin by distinguishing the manual and nonmanual classes, a big-class distinction so important that early class scholars often focused on it alone. We next identify three "macro classes" in the nonmanual category (i.e., professional-managerial, proprietor, routine nonmanual) and another two macro classes in the manual category (i.e., manual, primary). Within three of these macro classes, we then allow further "meso class" distinctions to emerge: the professional-managerial class is divided into classical professions, managers and officials, and other professions; the routine nonmanual class is divided into sales workers and clerks; and the manual class is divided into craft, lower manual, and service workers. The resulting scheme, which embodies three layers of big-class distinctions (i.e., manual-nonmanual, macro class, and meso class), may be understood as a non-denominational hybrid of conventional schemes that assembles in one scheme many of the contrasts that have historically been emphasized by big-class scholars.
The coding protocol. We have sought to document our microclass coding protocol carefully enough to allow other researchers to use it. For each of our four countries, we provide below files that lay out this protocol, allowing researchers to code detailed occupations into our microclass, macroclass, and mesoclass schemes.