Task Variation Within Occupations

• Date: 01 July 2017
• Published date: 01 July 2017
• DOI: http://doi.org/10.1111/irel.12179
• Author: Hugh Cassidy

Task Variation Within Occupations

*

HUGH CASSIDY

This paper investigates trends in task usages within and between occupations, and

the validity of assigning job task usages based on occupation code alone. Using

data from the German Qualification and Career Survey, I assign analytical, inter-

active, and manual task usage levels to workers independent of their occupation.

Task usages within occupation shifted nontrivially between the surveys consid-

ered, and individual characteristics such as age, education, gender, and hierarchi-

cal level differentially affect individual task usages and occupation-mean task

usages. Individual task usages are predictors of income, even when controlling

for mean occupational task usages as well as worker and job characteristics.

Introduction

Understanding the nature and specificity of work is central to the study of

labor economics. Recent literature has expanded the notion of human capital

beyond the firm-specific, occupation-specific, industry-specific, and general

categories, and conceives of human capital as being in part specific to tasks

performed on the job. For instance, mathematicians and economists perform

job tasks that are similar, whereas a mover would perfect job tasks that are

quite different. Thus, human capital accumulated as a mathematician is likely

more transferable if a worker became an economist compared to if a worker

became a mover.

An important aspect of this literature involves how to measure the tasks

workers perform on the job. Unfortunately, few labor-market surveys contain

rich information regarding the tasks performed on the job. As a result, work in

this area, including Poletaev and Robinson (2008), Ingram and Neumann

(2006), Yamaguchi (2010, 2012), and Acemoglu and Autor (2011) has instead

relied on a secondary source of data to assign tasks. Two commonly used

*The author’saffiliation is Kansas State University, Manhattan, Kansas. Email: hughcassidy@ksu.edu.

The author would like to thank Audra Bowlus, Chris Robinson, and Jed DeVaro for helpful comments and

guidance; Janice McGregor for assistance in translating the questionnaire text from German into English;

and Uta Sch€

onberg for providing code for performing the task assignments. He would also like to thank the

editor, Steven Raphael, and two anonymous referees for helpful suggestions. He is grateful for the data used

in the analysis provided by the Federal Institute for Vocational Education and Training in Germany. All

code used in preparing the results is available upon request. All errors are the author’s own.

INDUSTRIAL RELATIONS, Vol. 56, No. 3 (July 2017). ©2017 Regents of the University of California

Published by Wiley Periodicals, Inc., 350 Main Street, Malden, MA 02148, USA, and 9600 Garsington

Road, Oxford, OX4 2DQ, UK.

393

sources of job task information include the Dictionary of Occupational Titles

(DOT) and the Occupational Information Network (O*NET) survey. In these

papers, each occupation is assigned a task usage vector based on either the

DOT or O*NET, and the task usages of workers in another data set (e.g., the

National Longitudinal Survey of Youth or the Displaced Worker Survey) are

assigned based on the worker’s occupation. As a result, this task assignment

approach does not allow for task usage variation within occupations.

In this paper, I investigate the extent to which tasks performed by workers

on the job vary within occupations. I use the German Qualification and Career

Survey (GQCS), which is a worker survey conducted by the German Federal

Institute for Vocational Training (Bundesinstitut f€

ur Berufsbildung) and the

Research Institute of the Federal Employment Service (Institut f€

ur Arbeits-

markt- und Berufsforschung). This survey contains rich task usage information

reported by individual workers, and it enables me to assign task usage vectors

at both the individual and occupation level, which allows for within-occupa-

tion variation in task usages. Furthermore, because I use surveys from 1986

and 1992 that are closely comparable in terms of task usage questions, I am

able to assess the degree to which task usages changed between the survey

periods, and how much of this change can be attributed to changes within

occupations as opposed to a reallocation of workers between occupations.

Three tasks are calculated: analytical, interactive, and manual. These task

usages are calculated at both the individual level, as well as at the occupation

level by taking the mean across all individual task usages of workers within a

particular occupation. Using an analysis of variance, I find that while occupa-

tion code can explain a large portion of the variation in individual task usage,

much remains unexplained. Conducting a shift-share analysis, I find that large

changes in task usages occurred between 1986 and 1992, notably an increase

in analytical task usage and a decline in manual task usage. The large majority

of the changes in analytical and manual task usage occurred within the occu-

pation, while the changes in interactive tasks were primarily the result of a

reallocation of workers to different occupations.

Estimating a series of regressions, I show that worker and job characteristics

tend to impact individual task usages differently than occupation-mean task

usages. Worker characteristics such as age and education tend to have a

greater impact on individual task usage than occupation-mean task usage. Hier-

archical level—which is a relatively novel feature of the GQSC—also has a

much more important impact on individual job tasks than occupation-mean job

tasks. These results suggest that assigning task usage by occupation alone

misses a potentially significant amount of task usage information.

I also perform several income regressions, where I control for tasks both at

the individual and occupation-mean levels. I find that both individual and

394 / HUGH CASSIDY