The Task Approach to Labor Markets: An Overview (David H. Autor, 2013)

Summary

A J Labour Market Res. article (2013) 46:185-199, DOI 10.1007/s12651-013-0128-z, The "task approach" to labor markets: an overview, by David H. Autor, accepted 11 January 2013 and published online 14 February 2013. The abstract states that an emerging literature argues that changes in the allocation of workplace tasks between capital and labor, and between domestic and foreign workers, have altered the structure of labor demand in industrialized countries and fostered employment polarization. It says the essay sketches an alternative model of the assignment of skills to tasks based on comparative advantage, reviews conceptual and practical challenges, and cautions against two common pitfalls. The article also carries an abstract in another language.

Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used

Issuer: Source documents
Document type: The Task Approach
Case: The Task Approach

Cite this document

“The Task Approach to Labor Markets: An Overview (David H. Autor, 2013).” Original at link.springer.com. Archived at Pandemic Darlings, https://pandemicdarlings.com/source-documents/the-task-approach/, SHA-256 602be604d3a6288604a6424f599e4a087fb5ca85d0bc9fb66516591d77b27783.

Full text

J Labour Market Res. (2013) 46:185–199
DOI 10.1007/s12651-013-0128-z

A RT I C L E

The “task approach” to labor markets: an overview
David H. Autor

Accepted: 11 January 2013 / Published online: 14 February 2013
© Institut für Arbeitsmarkt- und Berufsforschung 2013

Abstract An emerging literature argues that changes in the Struktur der Arbeitskräftenachfrage in den Industrieländern
allocation of workplace “tasks” between capital and labor, verändert und eine Polarisierung von Beschäftigung geför-
and between domestic and foreign workers, has altered the dert hat – d.h. steigende Beschäftigungszahlen in den best-
structure of labor demand in industrialized countries and und schlechtestbezahlten Berufen. Eine Analyse dieses Phä-
fostered employment polarization—that is, rising employ- nomens im Rahmen der gängigen Produktionsfunktion ist
ment in the highest and lowest paid occupations. Analyzing jedoch schwierig, da die Zuweisung von Aufgaben zu Arbeit
this phenomenon within the canonical production function und Kapital in diesem Modell im Wesentlichen statisch ist.
framework is challenging, however, because the assignment Dieses Essay skizziert ein Alternativmodell zur Zuweisung
of tasks to labor and capital in the canonical model is essen- von Kompetenzen zu Aufgaben basierend auf dem kompa-
tially static. This essay sketches an alternative model of the rativem Vorteil, bespricht wichtige konzeptionelle und prak-
assignment of skills to tasks based upon comparative advan- tische Schwierigkeiten, vor denen Forscher stehen, wenn
tage, reviews key conceptual and practical challenges that sie den „TASKS-Ansatz“ in Daten übertragen wollen und
researchers face in bringing the “task approach” to the data, warnt vor zwei gängigen Fallstricken, die die immer stär-
and cautions against two common pitfalls that pervade the ker wachsende Literatur zu diesem Thema durchziehen. Ich
growing task literature. I conclude with a cautiously opti- schließe mit einer vorsichtig optimistischen Vorhersage für
mistic forecast for the potential of the task approach to illu- die Möglichkeiten des TASKS-Ansatzes zur Erläuterung der
minate the interactions among skill supplies, technological Interaktionen zwischen Angebot an Kompetenzen, techno-
capabilities, and trade and offshoring opportunities, in shap- logischem Potenzial und Handels- und Offshoringmöglich-
ing the aggregate demand for skills, the assignment of skills keiten in der Gestaltung der aggregierten Nachfrage nach
to tasks, and the evolution of wages. Kompetenzen, der Zuweisung von Kompetenzen zu Aufga-
ben und der Entwicklung von Löhnen.
Der „TASKS-Ansatz“ für Arbeitsmärkte: ein Überblick

Zusammenfassung In einer wachsenden Literatur wird die 1 Introduction
Auffassung vertreten, dass Veränderungen in der Zuweisung
von Arbeitsplatzaufgaben (tasks) zwischen Kapital und Ar- The canonical production function found in economic mod-
beit und zwischen in- und ausländischen Arbeitskräften die els is anthropomorphic—it’s built to resemble humans. It
features a role for labor and a role for machinery (capi-
tal), and, in general, these roles are distinct. This is not
D.H. Autor () illogical—after all, these are separate inputs. But what pre-
MIT Department of Economics, 50 Memorial Drive, E52-371, cisely is distinctive about the role of each input is left
Cambridge, MA 02142, USA opaque, and the nature of the interactions among them is
e-mail: dautor@mit.edu
highly constrained. In particular, capital is either a comple-
D.H. Autor ment or a substitute for labor, different types of labor are
NBER, IZA and IAB, Cambridge, MA, USA either complements or substitutes for one another, and these
186 D.H. Autor

roles are essentially fixed. Changes in the supply of each supplied by domestic labor, foreign labor, or capital, the ca-
input can of course affect marginal products, but each fac- pabilities of which may change over time. Consequently, the
tor’s “purpose” in the production function is both distinct assignment of factors to tasks in this model is determined by
and static. comparative advantage; in equilibrium, only the least-cost
These restrictions stem from the fact that the canonical factor is assigned to any given task.
production function implicitly equates two distinct aspects The second part of the essay briefly discusses some of
of production. One aspect is which factors are used as in- the key conceptual and practical challenges that researchers
puts, e.g., capital, high skill labor, low skill labor. The other face in bringing the “task approach” to the data. The lit-
is what services these factors provide. In the canonical setup, erature is already grappling with these issues, as I note
a factor’s identity and its role in the production function are below, and I’m cautiously optimistic that progress will be
synonymous. In reality, however, the boundary between “la- rapid.
bor tasks” and “capital tasks” in production is permeable I end by highlighting some of the pitfalls that researchers
and shifting. We observe numerous instances in which tech- may encounter in applying the task approach, most promi-
nological advances allow machinery to substitute for labor nently, loose and inconsistent task definitions that tend to
in performing a specific set of tasks—for example, dispens- sow confusion, and the beguiling temptation for researchers
ing boarding passes and assigning seats at airport check-in to confidently forecast the “future of work” based only on
counters in lieu of gate agents—while simultaneously com- the very modest understanding that we have developed to
plementing labor in other tasks, for example, allowing gate date.
agents to rapidly identify alternate flights and issue new tick-
ets. This evolving division of labor has a clear economic
logic: novel tasks—those demanded by new products, tech- 2 Tasks in production1
niques, or services—are often assigned first to workers be-
cause workers are flexible and adaptive. As these tasks are For the purposes of this discussion, it is essential to draw
formalized and codified, they become fallow for automation a distinction between skills and tasks. A task is a unit of
since machinery typically has a cost advantage over human work activity that produces output. A skill is a worker’s
labor in rote execution of repetitive tasks. stock of capabilities for performing various tasks. Workers
A growing body of literature argues that the shifting al- apply their skills to tasks in exchange for wages. Canonical
location of tasks between capital and labor—and between production functions draw an implicit equivalence between
domestic and foreign labor—has played a key role in reshap- workers’ skills and their job tasks, as noted above. Here, we
ing the structure of labor demand in industrialized countries emphasize instead that skills are applied to tasks to produce
in recent decades. Analyzing this phenomenon is difficult output—skills do not directly produce output. This distinc-
within the canonical production function framework, how- tion is of course inconsequential if workers of a given skill
ever, because the assignment of tasks to labor and capital is, always perform the same set of tasks. It is relevant, however,
for most conceptual purposes, static. To overcome this con- when the assignment of skills to tasks is subject to change,
ceptual hurdle, it is valuable to consider a setting where the either because shifts in market prices mandate reallocation
equilibrium mapping between production tasks and produc- of skills to tasks or because the set of tasks demanded in the
tive factors responds dynamically to shifting economic and economy is altered by technological developments, trade, or
technological forces. The “task approach” to labor markets offshoring. In my view, we are currently in such an era.
offers one promising, though far from complete, set of con- If the boundary between “labor tasks” and “capital tasks”
ceptual tools for studying these forces. is fluid, what determines the division of labor—or, more pre-
In this brief essay, I first sketch a formal model, devel- cisely, the allocation of tasks—between these factors? At
oped by Acemoglu and Autor (2011), that offers a flexible least two forces are central, one technological, the other eco-
and tractable approach for analyzing the interactions among nomic. On the technological front, the boundary between la-
skill supplies, technological capabilities, and potential trade bor and capital shifts primarily in one direction: capital typ-
and offshoring opportunities in shaping the demand for fac- ically takes over tasks formerly performed by labor; simul-
tors, the assignment of factors to tasks, and the productivity taneously, workers are typically assigned novel tasks before
and hence the wage of each factor. This model removes what they are automated. This sequence of task allocation makes
we view as an artificial set of distinctions typically made be- intuitive sense: when a task is unfamiliar or poses unex-
tween labor, capital, and trade (or offshoring) in the canoni- pected obstacles, workers can often draw on outside knowl-
cal production model, where different factors of production edge and problem-solving skills to devise work-arounds. By
play distinct and often incommensurate roles. In the this al-
ternative model, the fundamental units of production are job 1 This section draws extensively on Acemoglu and Autor (2011), and

tasks, which are combined to produce output. Tasks can be some sections of text are directly reproduced from that article.
The “task approach” to labor markets: an overview

187

contrast, few machines can improvise. Consequently, au-
tomating a task requires attaining a level of mastery beyond
what is required for a worker to simply perform the task;
it must be codified to the point where a relatively inflexible
machine can perform the work semi-autonomously.

Even when a task is fully codified, however, this does
not mean it will be automated. When Nissan Motor Com-
pany builds cars in Japan, it makes extensive use of indus-
trial robots to reduce labor costs. When it assembles cars in
India, it uses robots far more sparingly.” The key difference
between production in India and Japan is not technology but
cost: labor is comparatively cheap in India.

At the intersection of these two forces—technological
feasibility and economic cost—lies the principle of com-
parative advantage. Comparative advantage in production
means that the factor with the lowest economic cost of per-
forming a task is assigned that task. Economic cost in turn
reflects both a factor’s technological capability and its op-
portunity cost. It’s likely that the technological capabilities
and economic costs of robots in Japan and India are quite
similar. The opportunity cost of labor in India is far lower,
however, and hence Nissan hires Indian workers to perform
assembly tasks that are roboticized in Japan.

Though I’ve drawn examples from the information age so
far—electronic check-in counters, industrial robots—the on-
going reallocation of tasks between labor and capital is in no
sense unique to the current era. One could argue broadly, for
example, that the Industrial Revolution was an era in which
artisanal tasks were rapidly reallocated from labor to capital
while, simultaneously, new cognitive engineering and repair
tasks were demanded of labor.* The era in which we cur-
rently live also offers a particularly salient and economically
important source of task reallocation: the rapid diffusion of
computer technology, spurred by precipitous real price de-
clines, has likely altered the tasks performed by workers at
their jobs and ultimately the demand for human skills. In a
2003 paper, Frank Levy, Richard Murnane and I argued that:

“’.. (1) computer capital substitutes for workers in
carrying out a limited and well-defined set of cognitive
and manual activities, those that can be accomplished
by following explicit rules (what we term “routine
tasks”); and (2) that computer capital complements

*Fackler (2008). Fackler writes, “With Indian wages only about a tenth
of wages in Japan, the Chennai [India] factory will rely more on human
labor than Oppama [Japan], where robots staff many assembly lines.”

3See Goldin and Katz (1998), Katz and Goldin (2008) and Bessen
(2011). Katz and Margo (2012) present evidence that the occupational
distribution of U.S. manufacturing (though not the broader economy)
“hollowed out” between 1850 and 1910 as new manufacturing capi-
tal replaced skilled artisans, thus increasing the occupational shares of
both high skill professional, technical, managerial, clerical and sales
workers, and, simultaneously, low skill operatives and laborers.

workers in carrying out problem-solving and com-
plex communication activities (“nonroutine” tasks).
Provided that routine and nonroutine tasks are imper-
fect substitutes, these observations imply measurable
changes in the task composition of jobs” (Autor et al.
2003).

The 2003 paper also offered a simple production function
framework that showed in barebones terms how information
technology might simultaneously displace labor from rou-
tine tasks while complementing labor in non-routine tasks.*
Subsequent work has greatly enriched this framework.

In this section, I lay out a simple task-based framework
proposed in Acemoglu and Autor (2011), which in turn
builds on Autor et al. (2003), Acemoglu and Zilbotti (2001),
Autor (2006, 2008) and Costinot and Vogel (2010), among
others.° Distinct from the canonical model, this frame-
work explicitly incorporates the potential for technological
change or international trade and offshoring to cause tasks
that were previously performed by domestic workers to ei-
ther be replaced by capital or substituted by workers abroad.
This provides a natural mechanism through which techno-
logical advances can lead to real non-monotone changes in
the structure of employment by occupation as well as de-
clines in the wages for certain groups of workers.

2.1 A task assignment model

Consider a static environment with a unique final good. For
now, assume that the economy is closed and there is no trade
in tasks (a possibility we allow for later). The unique final
good is produced by combining a continuum of tasks repre-
sented by the unit interval, [0, 1]. Suppose, to simplify the
analysis, that the technology combining the services of tasks
is a constant elasticity of substitution aggregator, so that the
final good is produced as

1 n=l a
y= / nna | . (1)
0

where Y denotes the output of a unique final good, y(Z) is the
“service” or production level of task i, and 77 is the elasticity
of substitution between tasks.°

Suppose that there are three types of labor, high, medium
and low skill workers, who inelastically supply H, M and
L units of labor respectively. At any given point in time, a

4In The New Division of Labor, Levy and Murnane (2004) provide
a rich and nuanced discussion of the interaction between information
technology and human labor in accomplishing workplace tasks.

5Goldin and Katz also explore related ideas in many recent writings.
See, for example, Katz and Goldin (2008).

n = 1, or equivalently, InY = Io In y(i)di. I also focus on this case
here.

a Springer
188

D.H. Autor

subset Z C [0, 1] of the potentially feasible tasks is avail-
able (the remaining tasks cannot be produced). Each of the
available tasks has the following production function

y@) = Aza, @I@ + Amay()m(i) + Apa h(i)
+ Axag (i)k(i), (2)

where A terms represent factor-augmenting technology, and
az.(i), ay (i) and a7, (i) are the task productivity schedules,
designating the productivity of low, medium and high skill
workers in different tasks. In particular, a7 (7) is the produc-
tivity of low skill workers in task 7, and /(7) is the number
of low skill workers allocated to task i. The remaining terms
are defined analogously. Similarly, ax (i) is the task produc-
tivity schedule of capital and k(i) is the amount of capital
allocated to task i. Analogously to the canonical model, we
can think of Ay, Ay and Ay as low, medium and high skill
factor-augmenting technological changes. Distinct from the
canonical model, however, a factor-augmenting technologi-
cal change need not increase the wages of all factors in this
setup, as discussed below.

Though each task can be performed by low, medium or
high skill workers or by capital, the comparative advantage
of skill groups differs across tasks, as captured by the a
terms. These differences in comparative advantage are cen-
tral to understanding the interplay of tasks and skills. In par-
ticular, the model imposes the following simple structure of
comparative advantage: a, (i)/ay(i) and ay (i)/ay (i) are
(strictly) decreasing. This assumption can be interpreted as
stating that higher indices correspond to “more complex”
tasks in which high skill workers are more productive than
medium skill workers and medium skill workers are more
productive than low skill workers. Though not very restric-
tive, this assumption ensures a particularly simple and tight
characterization of equilibrium in this economy. We initially
set ax = 0 so that capital is not a competing source of task
supply, and hence all tasks are supplied by labor.’

Factor market clearing requires

1 1
[ toast. [ modi sm and
7 0 (3)
[ noai sn.
0

The structure of the model’s equilibrium is derived in
Acemoglu and Autor (2011), and I provide only basic details
here. Because of the simple nature of comparative advantage
outlined above, the equilibrium of the model involves a par-
tition of the continuum of tasks into three adjacent sets: the

7The canonical model can be nested within this framework trivially
by ignoring middle skill workers and assuming that a, (i) = az > 0
and ay (i) = O for all i <7 and ay (i) = 0 and ay (i) = ay > O for
all i > 7. Acemoglu and Autor (2011) discuss other ways in which the
model admits special cases similar to the canonical model.

DQ Springer

least complex set of tasks, those on the interval 0 <i < Iz,
will be supplied by Z workers; an intermediate set of tasks
on the interval J; <i < I will be supplied by M workers;
and the remaining highest skill set on the interval Iy <i <1
will be supplied by H workers. Crucially, 7, and I, which
are the cut-points of the task partition, are endogenously de-
termined in the model.

Competitive labor markets require that the law of one
price for skill applies. Each unit of Z labor will receive a
wage W_, and similarly for units of M and H labor. Addi-
tionally, since each task can potentially be performed by any
skill group, the allocation of tasks to skill groups is governed
by a no arbitrage condition: for the marginal task located
at I,, the cost of performing this task must be identical in
equilibrium whether it is supplied by L or M workers. Sim-
ilarly, the cost of performing task J} must, in equilibrium,
be equated between M and H workers. For tasks on the in-
terior of these sets (i < I,, I, <i < Iy, andi > Jy), how-
ever, workers of the relevant skill groups hold strict compar-
ative advantage. Using these conditions, it is straightforward
to demonstrate that there will exist a unique J; and J;, that
jointly satisfy the law of one price, the no arbitrage condi-
tion, and the market clearing condition in (3).

2.2 Implications for interpreting patterns of wages and
employment

A central virtue of the model above is that it facilitates a
more nuanced view of the nature of technological shifts,
which, together with changes in the supply of skills, under-
lie the bulk of the changes in the earnings distribution over
the last several decades. At a basic level, the distinction be-
tween high, middle and low skills adds an important degree
of freedom to the model, allowing for non-monotone move-
ments in wage levels and wage inequality as seen in the data.

A second distinction is that, while factor-augmenting
technological improvements always increase the real earn-
ings of both skilled and unskilled workers in the canonical
model, this may not be so in the presence of endogenous
allocation of workers to tasks. Acemoglu and Autor (2011)
show that a factor-augmenting technological improvement
(e.g., an increase in Aj) can reduce the wages of middle
skill workers. In particular, this happens when new technolo-
gies, by increasing the productivity of high skill workers, en-
courage some of the tasks previously performed by middle
skill workers to be shifted to high skill workers, but a cor-
responding shift of low skill tasks to middle skill workers is
not profitable.®

Arguably the most important innovation offered by this
task-based framework is that it can be used to investigate the

8Loosely, this happens when Jy shifts down considerably while I
does not change by much.
The “task approach” to labor markets: an overview

189

implications of capital (embodied in machines) directly dis-
placing workers from tasks that they previously performed.
In general, we would expect that tasks performed by all three
skill groups are subject to machine displacement. Based
on the patterns documented in the data above, as well as
the general characterization of machine-task substitution of-
fered by Autor et al. (2003), I believe that the set of tasks
most subject to machine displacement in the current era
are those that are routine or codifiable.? Such tasks are pri-
marily, though not exclusively, performed by medium skill
workers (e.g., high school graduates and those with less than
a four-year college degree). For this reason, let us suppose
that there now exists a range of tasks [/’, 7] C [Jz Iq] for
which ox (i) increases sufficiently (with fixed cost of capi-
tal r) so that they are now more economically performed by
machines than by middle skill workers. For all the remain-
ing tasks, i.e., for alli ¢ [/’, 7], we continue to assume that
ax(i)=0.

What are the implications of this type of technological
change for the supply of different types of tasks and for
wages? Acemoglu and Autor (2011) answer this question
formally. Here, I provide the intuition. A key observation is
that this form of technological change has the potential to
generate the patterns of wage changes and polarization that
have been widely discussed in the recent literature (Autor
and Dorn forthcoming; Goos et al. 2011; Firpo et al. 2011;
Autor et al. 2006, 2008; Goos and Manning 2007); that is,
a task-replacing technological change can directly reduce
wages of a skill group even as it raises total output. The rea-
son, intuitively, is that a task-replacing technological change
squeezes out the type of worker previously performing these
tasks, thereby creating “excess supply.” These workers are
therefore reallocated to tasks for which they have lower
comparative advantage, which pushes their wages down. !°
Simultaneously, by reducing the cost, and hence increasing
the intensity of use, of the newly automated tasks, the task-
replacing technological change complements each of the re-
maining tasks performed by labor. These countervailing ef-
fects imply that the real wage of the group that is directly
displaced by technology in a subset of its original tasks

°Tasks with these attributes may also be particularly well suited to
offshoring since their codifiability makes them readily tradable as the
price of communications falls (also a technological change).

!0Notice, however, that this reallocation process does not imply that
these displaced tasks are obsolete—in fact, just the opposite. As the
cost of performing routine tasks has declined by orders of magnitude,
their use in production has grown explosively—think, for example, of
the amount of processing power that goes into a single Google query.
However, because these tasks are now performed by capital rather than
labor, the consequences for the earnings power of workers who previ-
ously held comparative advantage in these tasks are at best ambiguous.

does not have to fall in real terms. But such an outcome is
possible—perhaps even likely—in realistic cases.!!

The polarization of occupational employment also fol-
lows from this mechanism. As has been documented in the
U.S. and many E.U. countries, the occupations that have
contracted most rapidly as a share of total employment over
the last three decades—in particular, clerical, administrative
support, sales, production and operative positions—are rea-
sonably well characterized as routine task-intensive: many
of the core tasks of these occupations follow precise, care-
fully codified procedures. Because of exponential declines
in the cost of computing power, these tasks are increas-
ingly ripe for automation and hence are reassigned from
labor to capital. As workers lose comparative advantage in
routine-intensive activities, a greater mass of skills is real-
located towards the tails of the occupational distribution—
both towards high skill analytic, reasoning and problem
solving tasks and, ironically, towards traditionally low skill,
in-person service tasks—thus leading to employment polar-
ization.

3 Challenges in putting tasks to the test

The task model offers a potentially powerful framework for
studying how changes in skill supplies, technologies, and
trade and offshoring opportunities jointly shape the demand
for labor, the assignment of skills to tasks and the real wages
commanded by workers of different skill groups. There are
substantial challenges, however, that confront a researcher
in bringing these ideas to the data.

3.1 Measuring tasks

A first challenge is measurement. Conventional skill prox-
ies, such as education and experience, are broadly collected
by survey data sources and are widely accepted by applied
economists as valid proxies for workers’ stocks of human
capital. In the canonical setting where the “law of one price”

1 As detailed in Acemoglu and Autor (2011), there are also interest-
ing general equilibrium effects on other skill groups. As the middle
skill group is displaced, the task boundaries will change, encroaching
on the other two skill groups. The relative degree of encroachment on
L versus H workers depends on the shape of the comparative advan-
tage schedules in the neighborhood of the initial set boundaries. If H
workers have strong comparative advantage relative to M in their ini-
tial tasks, then the upper boundary will move little. If L workers have
relatively weak comparative advantage in their initial tasks, then the
predominant direction of task reassignment will be that tasks previ-
ously performed by L workers are reallocated to M workers. Although
total output necessarily rises with the advent of task-replacing tech-
nological change, the net effect need not be positive even for groups
indirectly affected. For example, if M and L workers are sufficiently
close substitutes at the margin, the displacement of M workers into L
tasks can reduce the wages of both groups.

a Springer
190 D.H. Autor

for skill prevails, a worker’s human capital is also a suffi- ing to automate. Unfortunately, these overlaps among occu-
cient statistic for her productivity and hence her potential pations in “task space” are in no way visible from standard
earnings. Thus, while one could ask what tasks a worker occupational classification schemes that group occupations
performs using her stock of human capital in the canoni- roughly according to the services that they provide (health
cal setting, the question is not particularly interesting since services, production, analysis, etc.) rather than the tasks that
job tasks—and more broadly, occupations—are not well- they encompass. Thus, a researcher who wants to identify
defined economic constructs in this framework.12 task commonalities that cross occupational boundaries is
By contrast, a key premise of the task approach is that al- forced to make additional subjective judgments. Such judg-
though skills are used to perform tasks, skills are not tasks. ment calls are generally undesirable in empirical work since
To understand how and why shifts in technology and trade they leave an uncomfortable amount of discretion for re-
affect skill demands and earnings, we therefore need to be searchers to shape the data to fit their priors.13
able to measure the mapping between skills and tasks and To reduce the role of subjectivity in the task categoriza-
observe the changes in this mapping over time. This presents tion step, an alternative approach to grouping occupations
a substantial measurement challenge since consistent infor- manually is to append a set of standardized job descriptors
mation on job tasks is not commonly collected by represen- to each occupation. These descriptors can in turn be used to
tative data sources. develop task measures. In the U.S., two major data sources,
Researchers have taken at least three approaches to re- the Dictionary of Occupational Titles (DOT) and its succes-
solving this measurement problem. A first is to use occu- sor, the Occupational Information Network (O*NET), offer
pations as proxies for job tasks. This has a natural appeal sources for job descriptors, and both have been used fre-
since occupations can readily be conceptualized as bundles quently in empirical work on job tasks.
of tasks that workers are required to perform. Because oc- The primary strength of this method of classifying tasks
cupational classification schemes employed by national sta- is that its main inputs, externally supplied job content mea-
tistical agencies typically contain hundreds of distinct occu- sures, can be carefully validated by the statistical agencies
pations, however, occupational codes are far too unwieldy that supply them. The approach of assigning tasks to occu-
in their raw form to serve as task measures. To make this pations using job task descriptors has two major limitations,
problem manageable, it is necessary to boil down a large however, one intrinsic and the other reflecting shortcomings
set of occupation codes to a lower dimensional object. One in existing data collection efforts. Fundamentally, assigning
method is to aggregate many detailed occupations into a few task measures to occupations overlooks all heterogeneity in
broad categories, e.g., professional, technical, managerial, job tasks among individuals within an occupation. It’s self-
clerical, production, service, etc. This is typically straight- evident that individual worker skills and actual job tasks dif-
forward because most occupational schemes are hierarchi- fer among workers within an occupation, and it seems likely
cal. Hence, detailed codes are readily collapsed upward into that these within-occupation skill-task assignments are an
their parent categories. important component of the overall equilibrium relationship
A limitation of this approach, however, is that it obscures between skills and tasks. Thus, at best, occupation level task
any similarities in task content that cross broad occupational measures provide a rough approximation to the microeco-
boundaries. For example, office clerical workers and assem- nomic assignment process.
bly line machine operators have much in common from the Perhaps even more importantly, task measures assigned
perspective of the task framework: both make extensive use at the occupational level offer a static view of the tasks
of routine tasks that have high potential for automation. Sim- an occupation comprises. If, as the task framework sug-
ilarly, both truck drivers and food service workers engage in- gests, job tasks assignments are subject to continual re-
tensively in non-routine manual tasks requiring detailed vi- optimization, this static perspective is problematic. Under
sual recognition and flexible adaptation to a changing physi- the reasonable assumption that task changes within occupa-
cal environment, tasks that have proven extremely challeng- tions (the ‘intensive margin’) tend to move in the same direc-
tion as task changes made visible by changes in the relative
sizes of occupations (the ‘extensive margin’), then measur-
12 Indeed, one simple sense in which the task approach has been pro-
ing job tasks using static measures of occupational content
ductive is by spurring economists to focus attention on shifts in oc-
cupational composition—most notably, in the case of employment will systematically understate the extent of the task reallo-
polarization—rather than maintaining an exclusive focus on the wage cations taking place. Fortunately, this problem can be over-
returns to skills. It seems self-evident that the changing distribution come if occupational job descriptor databases are regularly
of occupational assignments—that is, shifts in what people do at their
jobs—contains information about the evolving nature of production in
the economy. But perhaps because occupations do not have a theoret- 13 On the other hand, workers’ occupational classifications are gener-

ical analogue in the canonical production framework, they have until ally hard-coded in survey and administrative data sets. Hence, using
recently received limited empirical attention from economists (though these codes directly is empirically transparent despite its other disad-
see Goldin 2002 for a notable exception). vantages.
The “task approach” to labor markets: an overview 191

refreshed. In the case of the U.S. Dictionary of Occupational from survey respondents alongside other demographic, em-
Titles, updates were infrequent and appeared to suffer from ployment and wage data. Adding task measures directly to
status quo bias (Miller et al. 1980). In the case of O*NET, worker surveys places no restrictions on the variability of
the sheer number of distinct occupations—approximately tasks within as well as across occupations, does not inad-
1,000 in O*NET release 14.0—and the vast quantity of vertently impose the assumption that occupational tasks are
unique scales make it extremely costly to refresh the entire static, and allows task measures to be designed for test-
database at high frequency. ing specific hypotheses. There have been several success-
Alongside these intrinsic limitations of occupation-level ful efforts in this direction to date. Most significantly, the
task measures, there are practical difficulties stemming from IAB/BIBB labor force data, collected in 1979, 1984/85,
the design of the job content descriptors provided by the 1991/92, 1998/99 and 2005/06, offer detailed self-reported
major U.S. data sources (DOT and O*NET). In both data data on workers’ primary activities at their jobs.15 To my
sources, job content measures are often vague, repetitive, knowledge, the task measures in the IAB/BIBB were first
and constructed using ambiguous and value-laden scales brought to prominence among economists worldwide by a
that are likely to confuse respondents.14 The DOT, which sardonic article by DiNardo and Pischke (1997), who pur-
was first issued by the U.S. Department of Labor in 1939 and ported to assess whether pencil use had changed the U.S.
last revised in 1991, contains 44 objective and subjective job wage structure during the 1980s. Subsequent work by Spitz-
content scales. When the DOT was replaced by the O*NET Oener (2006), Dustmann et al. (2009), Antonczyk et al.
in 1998, the complexity of the database increased by an or- (2009), Black and Spitz-Oener (2010), and Gathmann and
der of magnitude. Version 14.0 of the O*NET database, re- Schönberg (2010), along with other recent studies that are
leased in June of 2009, contained 400 separate rating scales, too numerous to list here, used the job activity questions in
which is almost half as many scales as the number of occu- the IAB/BIBB data to measure job tasks and explore links
pations coded by O*NET. These scales were not of course between technological change, changes in task inputs, and
developed with “task measurement” in mind since the de- shifts in wage structure. One striking finding from this body
sign of DOT and O*NET preceded recent research interest of work is that changes in the task structure of employment
in job tasks. In practice, this means that researchers who in Germany between 1979 and 1999 are primarily accounted
wish to use these databases as sources for task measures are for by changes in task input within detailed occupations
essentially required to pick and choose among the plethora rather than shifts in employment across occupational cate-
of scales available, a problem that is much more severe for gories.
O*NET than for DOT. Researcher discretion again becomes The IAB/BIBB data were not, to my knowledge, purpose-
paramount in this data construction process, and some trans- built to measure economy-wide levels or over-time changes
parency is inevitably lost. While I have found that task mea- in work activities (i.e., job tasks). Several more recent data
sures distilled from DOT and O*NET can serve as powerful collection efforts explicitly tackle these goals. The British
proxies for occupational tasks, I am at best only moderately Skills Survey, fielded in 1986, 1992, 1997, 2001 and 2006
comfortable with these tools because their complexity and by Francis Green and collaborators, has sought to provide
opacity places little discipline on how they are applied and consistent measures of skills used in the workplace by sur-
interpreted. veying workers about their work activities, training require-
A third approach to task measurement that avoids some ments, use of technologies, and freedom to exercise dis-
of the pitfalls above is to collect job task information directly cretion in performing their jobs (see Felstead et al. 2007
for a summary of recent findings from this body of work).
The survey of Skills, Technology, and Management Prac-
14 For example, Item 30 of the Skills questionnaire of O*NET release
tices (STAMP) fielded by Michael Handel provides a de-
14.0 asks respondents to rate the importance of Systems Evaluation, tailed cross-sectional view of work activities in the U.S.,
defined as “Identifying measures or indicators of system performance
and the actions needed to improve or correct performance, relative and has the potential to provide a valuable time-series view
to the goals of the system” of Labor Employment and Administra- of the evolution of these activities provided that the survey
tion (2009). The O*NET survey anchors for this question include: is repeated (Handel forthcoming). In addition, a survey con-
Level 2, determining why a coworker was unable to complete a task ducted by the Princeton Data Improvement Initiative (PDII)
on time; Level 4, understanding why a client is unhappy with a prod-
uct; and Level 6, evaluating the performance of a computer system in 2008 collected a series of experimental workplace task
of Labor Employment and Administration (2009). As a 2010 National measures for a representative sample of U.S. households
Research Council (Tippins and Hilton, 2010, p. 178) report noted, “it is
difficult. . . to conceive of a well-defined metric that would unambigu-
ously place the level of systems evaluation required for “evaluating the 15 Unfortunately, the set of job activity questions used varies substan-

performance of a computer system” either above or below the level of tially across the different survey years. This almost certainly reduces
systems evaluation required for “understanding why a client is unhappy the reliability of the IAB/BIBB data as a source for tracking the evolu-
with a product.” tion of job task inputs in aggregate.
192 D.H. Autor

(see Autor and Handel forthcoming). While this survey will 3.2 The counterfactual problem
not offer a time-series component, it serves as a useful
testbed for exploring the value-added of person level task Alongside the practical challenge of measurement, the task
measures alongside occupation level measures. Most re- framework poses two central conceptual challenges that the
cently, researchers at IAB have fielded a new survey instru- literature is now beginning to tackle. A first concerns the in-
ment that seeks to rigorously operationalize five major task trinsic difficulty of identifying credible counterfactuals in a
constructs: analytic tasks, interactive tasks, manual tasks, setting where the allocation of labor to sectors (occupations,
routine tasks, and autonomy-demanding tasks (Matthes et al. tasks) is determined by comparative advantage, as in a Roy
2012). and Donald (1951) model.17 To see the problem, consider
One lesson these recent initiatives offer is that designing the question of what is the “market wage” of a worker of a
survey questions that successfully measure routine, codifi- given skill level in the task model above. At one level, the
able tasks is itself a challenging task. The heart of the dif- answer is self-evident. In the equilibrium of the model, a
ficulty is that what is routine from the perspective of hu- “law of one price” for skill prevails: there is one wage WH ,
man labor is generally not routine from the perspective of WM and WL for each of the three skill levels.18 However,
machine automation and vice versa. For example, mopping these equilibrium wages are contingent on the assignment
floors is a mundane repetitive task from the perspective of of skills to tasks. In reality, each worker in the task economy
janitors. The fact that the visual recognition and environ- could perform any of the tasks available, and would earn
mental adaptability required to perform this task pose daunt- a different wage in each of the three segments of the task
ing technical challenges for computer science is a charac- space. Moreover, if a non-trivial subset of workers were to
teristic of current computer technology, not a characteristic reallocate their labor across task categories, this would al-
of the work from a human labor perspective. Consequently, ter the aggregate wage offered to each skill group, since it
asking workers to assess whether their job tasks are codi- would shift the effective supply of skill.
fiable from the perspective of machine execution—roughly These notional reallocations are out-of-equilibrium
akin to asking truck drivers to compare their skills to the thought experiments and, superficially, have little practical
current capabilities of Google’s driverless car—is unlikely relevance.19 They become relevant, however, when shocks
to provide reliable information.16 A more promising ap- to the economy shift the allocation of skills to tasks—for
proach may be to ask workers concretely about which of example, the case discussed above where a task-replacing
many commonplace tasks they regularly perform, and then technological change displaces middle skill workers from
to apply outside expertise to classify whether these tasks are a set of tasks they’d previously performed. In a canonical
routine or non-routine from a machine execution perspec- production model with fixed roles for each input, this tech-
tive. Of course, we should expect that the range of tasks that nological advance would be equivalent to an increase in the
can be automated will continue to expand outward from the
most “routine” core—performing basic mathematical oper- 17 See Heckman and Honore (1990), Heckman and Sedlacek (1985),

ations, storing and retrieving data—towards tasks requiring and Heckman and Scheinkman (1987).
18 In an extension to the model, Acemoglu and Autor (2011) con-
ever greater adaptability, judgment and, perhaps eventually,
creativity. sider a setting in which each worker i has an endowment of each of
the three skills, li , mi and hi , only one of which can be used at a
time and where the distribution of endowments varies over the pop-
16 For example, among the task questions that Michael Handel and I ulation. Each worker therefore chooses which skill to apply to which
contributed to the PDII survey, we included the following item in- task to maximize earnings, and hence the wage of worker i is equal to
tended to measure the frequency of routine tasks, “How much of your wi = max[WL li , WM mi , WH hi ], where WL , WM and WH are in turn
workday (involves/involved) carrying out short, repetitive tasks? [al- determined by the supply of skills to tasks (that is, aggregate supply to
most all of the time, more than half the time, less than half the time, each skill group is jointly determined with the wage level). This setting
almost none of the time, don’t know]” Consistent with expectations, gives rise to both greater wage heterogeneity and richer wage dynam-
managers and professionals report performing relatively few routine ics than the baseline case, and also captures, in my view, an important
tasks. To our surprise, however, workers in service occupations (e.g., additional element of realism. When a skilled production worker loses
food service, cleaning, security) report spending about as much time his job due to plant shutdown and takes employment instead as a food
performing routine tasks as clerical workers, and transportation work- service worker, it is plausible to think that he has not only changed job
ers report substantially more. In retrospect, we should have anticipated tasks but also changed the skill set used to perform these tasks—for
exactly this pattern of responses. To the typical delivery driver, security example, switching from using m skills for production tasks to l skills
guard or food service worker, it is almost surely the case that their pri- for service tasks. This extension to the model is also consistent with the
mary job tasks appear quite repetitive—though these tasks don’t appear findings in Neal (1999) and Gathmann and Schönberg (2010) that job
as such to computer scientists. Thus, the notion of “routineness” that changers tend to switch into occupations that are similar in task content
we had in mind in writing the survey question is simply alien to how to their prior jobs, and when they move further afield by changing in-
most survey respondents think about it. (In our partial defense, Handel dustries or by substantially changing job tasks, they tend to experience
and I were invited to join the PDII effort shortly before the final survey a wage penalty.
went into the field, and hence, we did not have the opportunity to pilot 19 Unless, of course, one posits a set of plausible frictions in the econ-

the task questions. I nevertheless view our initial effort as naive.) omy that lead to some degree of misallocation at any given moment.
The “task approach” to labor markets: an overview 193

relative supply of middle skill labor, which would reduce These features add to the realism of the model, but they
the middle skill wage and, in most cases, raise the wage pose challenges for empirical implementation because the
of other factors of production—low and high skill labor, productivity of factors in performing tasks to which they
in particular—through q-complementarity. In a setting with are not currently assigned is inherently hard to estimate.
comparative advantage, however, the effects are more nu- For example, an ambitious study by Costinot and Donald-
anced and contingent. The substitution of capital for labor in son (2012) studies how reductions over the course of 160
performing middle skill tasks reduces the total set of tasks years in the cost of moving agricultural goods across the
performed by M labor and causes the threshold tasks, IL∗ U.S. have affected consumer welfare through three distinct
and IH∗ , to shift outward (falling in the case of IL∗ and ris- mechanisms: direct reductions in transport costs; gains from
ing in the case of IH∗ ).20 These shifts impact WM through price convergence across markets; and, most relevant here,
three channels. One is the standard diminishing marginal gains from reallocation of land parcels to crops that are more
productivity effect seen in the canonical model. A second valuable when distant markets become accessible (e.g., avo-
is a q-complementarity effect, here operating directly on M cados are now grown in California for consumption in Man-
labor: reflecting their lower cost, the newly automated tasks hattan). The innovation that makes this last crucial step fea-
are used more intensively in production, which in turn raises sible is that the authors exploit agronomic data to estimate
the marginal product of all other tasks, including those per- the potential productivity of each parcel of land for produc-
formed by M labor. The third operative channel is realloca- ing all food crops, not simply those with which it is currently
tion: as IL∗ and IH∗ are shifted outward, M labor is reassigned planted. This ingenious approach is not necessarily broadly
to a set of tasks in which it was initially less productive, thus feasible in other settings, unfortunately. In a recent paper,
reducing its marginal product. In short, how a task-replacing for example, Hsieh et al. (2012) seek to assess how much
technological change (or an increase in own labor supply) the reductions in discrimination against women and minori-
affects the earnings of M workers depends importantly on ties over six decades have increased aggregate productivity
how productive M labor is in alternative uses—that is, in the by increasing allocative efficiency in the labor market (in
our terminology, the assignment of skills to tasks). Since no
tasks to which it is not initially assigned.
labor market data equivalent to agronomic data are available
These same forces also determine how the task-replacing
for estimating counterfactual task productivities by gender
technological change affects the relative earnings of H and
and race, the Hsieh et al. estimation is necessarily more de-
L workers. We noted above that an M-labor replacing tech-
pendent on distributional assumptions than is the Costinot-
nological change causes IL∗ and IH∗ to shift outward, thus
Donaldon study.21
encroaching on the tasks previously performed by H and L
labor. What determines the degree of encroachment is the 3.3 On the “wage return” to tasks22
degree of substitutability between M labor and H or L la-
bor at the respective task thresholds, IL∗ and IH∗ . If H la- A related conceptual and empirical challenge concerns the
bor has strong comparative advantage relative to M labor in appropriate use of task measures in wage regressions. A key
performing the tasks just above IH∗ and L labor has com- implication of the theoretical framework above is that hold-
paratively weak comparative advantage relative to M labor ing the task efficiencies (that is, the α(·) s) constant, changes
in performing tasks just below the IL∗ threshold, then M la- in the market value of tasks should affect the evolution of
bor will primarily be allocated downward as the M-labor wages by skill group. In particular, the model predicts that if
displacing technology is adopted, meaning that WL falls rel- the relative market price of the tasks in which a skill group
ative to WH and wage inequality between high and low skill holds comparative advantage declines, the relative wage of
workers rises. If the pattern of comparative advantage is re- that skill group should also decline—even if the group re-
versed, then WL rises relative to WH and wage inequality allocates its labor to a different set of tasks (i.e., due to the
falls. Thus, comparative advantage shapes the impact of a change in its comparative advantage).
shift in technology or skill supplies anywhere in the task Critical to this prediction is the distinction made between
distribution on the task assignment, productivity and wage the wages paid to a skill group and the wages paid to a given
of all other factors.
21 Over the short and medium term, changes in supplies and technolo-
20 Recall that in equilibrium, the cost of performing the I ∗ threshold gies will generally lead to evolutionary rather than revolutionary real-
L
task with either M or L labor must be equated, and similarly for M locations of factors. If so, it may be sufficient for many leading empir-
and H labor at the IH∗ threshold task. By squeezing M workers into ical questions to be able to estimate the distribution of potential factor
a smaller set of tasks, a middle skill task-replacing technical change productivities in a neighborhood nearby to the current equilibrium al-
initially reduces their marginal productivity, hence WM . Because it is location rather than across all possible scenarios.
now cheaper to perform the threshold tasks with M labor than with 22 This discussion of the use of task measures in wage regressions

either L or H labor, IL∗ must shift downward and IH∗ upward until the draws from Acemoglu and Autor (2011), and some text is repeated
no arbitrage conditions are again satisfied. from that source.
194 D.H. Autor

task—a distinction that is meaningful because the assign- may be more informative about the wage structure than the
ment of skills to tasks is endogenous. To see the implica- alternative of regressing wages on tasks.
tions of this distinction, consider a technological change that A growing literature is now grappling with these chal-
raises the productivity of high skill workers in all tasks (e.g., lenges. A recent working paper by Firpo et al. (2011), for
an increase in AH ). The model implies that this would ex- example, develops an innovative method for measuring the
pand the set of tasks performed by high skill workers (i.e., impact of changing task prices on wage structure. Using a
lowering IH ), so that some tasks formerly performed by simple statistical model of occupational wage setting, their
medium skill workers would now be performed by high skill work predicts that occupations that are specialized in tasks
workers instead. Thus, relative wages paid to workers per- that have declining market value should see a reduction in
forming these (formerly) middle skill tasks would actually both mean occupational wages and the variance of occupa-
increase since they are now performed by the more produc- tional wages, and vice versa for tasks with rising prices. This
tive high skill workers.23 But crucially, the analysis in Ace- latter (variance) effect stems from the interaction between a
moglu and Autor (2011) also proves that the relative wage of falling task price and a fixed distribution of task efficiencies
medium skill workers, who were formerly performing these within an occupation; as the market value of a given task
tasks, would fall. falls, the range of wages paid to workers with differing pro-
One important implication of this set of observations is ductivities in that task compresses along with it.
that the intuitively appealing approach of regressing wages One issue that needs further study in the Firpo, Fortin and
on current tasks performed by workers to estimate the “re- Lemieux approach is that changes in task prices will pre-
turns to tasks” may generate potentially misleading results. sumably lead to changes in the self-selection of workers into
The appeal of estimating task-wage regressions is almost re- occupations—as is implied by the comparative advantage
flexive for labor economists steeped in the Mincerian tradi- model and, more generally, by any model in which work-
tion of regressing log wages on years of schooling and po- ers make maximizing choices. These endogenous changes in
tential experience. At the heart of the Mincerian approach occupational entry and exit should also affect occupational
is the observation that a worker’s human capital, proxied by wage means and variances. At present, the approach taken
schooling, is a quasi-fixed stock variable, determined prior by Firpo, Fortin and Lemieux is similar to the Mincer regres-
to labor market entry. In this setting, a log earnings regres- sion in that it implicitly treats the skill allocations (here, the
sion can recover the market rental rate of human capital (i.e., assignment of skills to tasks) as predetermined. This simpli-
the return to education)—at least under certain stringent as- fies the predictions of the statistical model but imposes an
sumptions. unknown cost in terms of economic realism.
By contrast, the set of tasks that a worker performs on the
job is an endogenous state variable that is simultaneously
determined by the worker’s stock of human capital and the 4 Some concluding concerns
contemporaneous productivity of the tasks that human cap-
ital could accomplish. This implies that task assignments Revealed preference demonstrates that I am optimistic about
are themselves a function of the current wage distribution, the potential of the “task approach” to labor markets to offer
and hence a regression of wages on job tasks (rather than on insights into the interactions among skill supplies, techno-
worker skills) will be inherently difficult to interpret. As an logical capabilities, and potentially trade and offshoring op-
extreme example of this pitfall, note as above that the rela- portunities, in shaping the demand for skills, the assignment
tive wage paid to a task can move in the opposite direction of skills to tasks, and the evolution of wages. Rather than
from the relative wage paid to the skill group that initially recapitulate the sources of my enthusiasm, let me conclude
performed that task. By contrast, the relative wage paid to by offering three recommendations to researchers working
a given skill group always moves in the same direction as in this literature.
its comparative advantage—that is, a technological change A first concerns task definitions. Following Autor et al.
that increases the productivity of a skill group necessarily (2003), researchers have often used a rough taxonomy of
raises its relative wage. This suggests that even in a task set- three broad task groups: routine cognitive and manual tasks,
ting, the Mincerian approach of regressing wages on skills which are readily codifiable for purposes of machine per-
formance; abstract analytical and managerial tasks, which
may require creativity, hypothesis formation, problem solv-
23 Nor is this notion far-fetched. Skill levels in production and cleri-
ing or persuasion; and non-routine manual tasks (or, simply,
cal occupations, as measured by the college employment or wage-bill manual tasks), which may require physical flexibility and
share, have risen as employment in these occupations has declined (Au-
tor et al. 2003). A plausible interpretation of this pattern is that ed- adaptability, visual recognition, or non-scripted communi-
ucated workers have comparative advantage in the set of non-routine cations. Commencing with the influential work of Blinder
tasks in these occupations that remain. (2009), economists have also focused on another dimension
The “task approach” to labor markets: an overview 195

of task content: “offshorability,” describing tasks that may they are not dependent on face-to-face interactions, not that
be performed in a remote location without substantial qual- they are routine per se.27
ity degradation.24 In a similar vein, researchers sometimes classify tasks
While these four task attributes—routine, abstract, man- as non-offshorable because they require workers to exercise
ual, offshorable—are broadly distinct, there are important judgment, think creatively, or engage in problem-solving.
overlaps among them. For example, many routine codifi- Yet, it cannot be the case that overseas workers are un-
able tasks (e.g., performing calculations, checking a doc- able to exercise judgment or perform higher cognitive func-
ument for spelling errors) can potentially be automated or tions! Clearly, what makes these particular non-routine tasks
difficult to offshore is that they are typically performed
offshored. On the other hand, many tasks that are eminently
by workers in management, leadership or decision-making
offshorable are clearly non-routine—for example, providing
roles—and these roles in turn demand face-to-face interac-
technical support to frustrated computer users, interpreting
tions. Here again, the offshorability question turns on qual-
medical x-rays, or taking drive-through orders at roadside
ity degradation due to distance, not on a task’s routineness
McDonalds.25 or non-routineness.
How should these overlaps affect task classification? A These observations raise the concern that the emerging
rudimentary rule of thumb is that task classification schemes task literature may be significantly inside the frontier of
should provide separate scales for distinct task attributes. what is feasible in terms of precise terminology and con-
In practice, however, researchers frequently employ broad sistent measurement. Addressing these shortcomings should
classification schemes that collapse distinctions among at- therefore be a high priority on the research agenda since it is
tributes. For example, in constructing measures of task “off- difficult for a field to advance when researchers lack shared
shorability,” it is commonplace for researchers to classify definitions of core constructs or consistent measures of these
tasks as offshorable if they are highly codifiable, and hence constructs.
can be performed without direct supervision, or if they do To advance this agenda, my second recommendation to
not require face-to-face contact and hence suffer little qual- researchers is to use, re-use, recycle, replicate, repeatedly
ity degradation when performed remotely. This broad defi- apply and attempt to converge upon a shared and standard-
nition of offshorability is problematic. As Blinder and coau- ized set of task measures.28 Although empiricists may have a
thors have emphasized in their discussions of offshorabil- preference for their own purpose-built task measures, social
ity (cf. Blinder 2009; Blinder and Krueger forthcoming), science will be better served if, at a minimum, researchers
use “off the shelf” measures alongside these alternatives.
the key attribute that makes a task potentially “offshorable”
This practice is critical for illuminating when the potentially
is that it does not depend on face-to-face worker contact
divergent conclusions reached by competing studies reflect
or close proximity between worker and customer.26 This
disagreements about substance or instead discrepancies in
requirement does not exclude many non-routine tasks— measurement. While there is surely room for improvement
such as interpreting x-rays, or providing technical support— on both fronts, a lack of clarity on either hinders intellectual
nor does it include all routine tasks, such as generating progress.
web pages dynamically in response to user queries. While As one small step towards speeding the process of con-
many routine task-intensive activities appear to be poten- vergence, I have placed all task measures used in my pub-
tially suited for offshoring, what makes them suitable is that lished papers—along with associated crosswalks and sup-
porting documents—on my MIT homepage for download.29
If other researchers are similarly interested in contributing
24 One may ask what distinguishes offshorability from traditional trade
to this public good, I am pleased to host and maintain a web
in goods. Perhaps the best answer is that offshoring refers primarily page that provides access to data sets and documentation
to service tasks rather than physical goods. Many of these services are
non-storable and hence need to be produced in real time at the point at
developed by various task projects for use by the broader
which they are used (e.g., teaching, technical support, person-to-person research community.
sales). Thus, a valid measure of “offshorability” describes which of Finally, let me remark on the difficulties on extrapolating
these tasks must be performed in the location where they are consumed from theory to policy. One of the ironies of the computer era
and which can be delivered from a distance.
25 Surprisingly, the voice heard through the loudspeaker when you
27 Having made these distinctions, researchers may draw different con-
place your drive-through order may not belong to a worker in the
restaurant. clusions about whether the widely observed phenomenon of employ-
26 How could a task not require face-to-face contact but still need to ment polarization is due to automation, offshoring or some third set of
factors. But this is a question of cause-and-effect, not measurement.
be performed nearby to a customer site? Package delivery is one such
28 It is regrettably the case that there are almost as many distinct task
example. United Parcel Service (UPS) does not require face-to-face
contact with package recipients, but it does need to drive up to their classifications as there are papers in the task literature.
houses to deliver packages. 29 http://web.mit.edu/dautor/www.
196 D.H. Autor

is that information technology has increased labor’s compar- monplace frustration of calling a software firm for tech-
ative advantage in traditionally low skill service tasks, such nical support only to discover that the support technician
as food preparation and personal care, relative to its value knows nothing more than what is on his or her computer
in traditional middle skill tasks such as computation, infor- screen—that is, the technician is a mouthpiece, not a prob-
mation processing, and repetitive production activities. One lem solver. This example captures one feasible division of
inference that economists and policymakers are tempted to labor: machines performing routine technical tasks, such as
draw from this trend is that advanced countries should give looking up known issues in a support database, and workers
up on “middle skill” education because there is no future for performing the manual task of making polite conversation
middle skill jobs. This conclusion is mistaken for multiple while reading aloud from a script. But this is not generally
reasons. A first is that education is cumulative; students can- a productive form of work organization because it fails to
not attain high skills (e.g., proving theorems) without first harness the complementarities between technical and inter-
mastering middle skills (e.g., arithmetic). Since it is diffi- personal skills. Stated in positive terms, routine and non-
cult to forecast which students might ultimately succeed in routine tasks will generally coexist within an occupation to
the higher echelons of education, there is substantial option the degree that they are complements—that is, the quality of
value in investing universally in students’ middle skills. The the service improves when the worker combines technical
efficiency case for these investments is complemented by an expertise and human flexibility.31
equity case. Choosing to not universally invest in students’ This reasoning suggests that many of the middle skill
“middle skills” would imply foreclosing the economic hori- jobs that persist in the future will combine routine tech-
zons of many citizens at an early age—an idea that few nical tasks with the set of non-routine tasks in which
democratic societies would want to embrace. workers hold comparative advantage—interpersonal inter-
Second, my reading of the evidence is that middle skill action, flexibility, adaptability and problem-solving.32 Med-
jobs are not slated to disappear. While many middle skill ical para-professions are one leading example of this virtu-
tasks are susceptible to automation, many middle skill jobs ous combination, but this example is not a singularity. This
demand a mixture of tasks from across the skill spec- broad description also fits numerous skilled trade and re-
trum. To take one prominent example, medical parapro- pair occupations—plumbers, builders, electricians, HVAC
fessional positions—radiology technicians, phlebotomists, installers, automotive technicians—marketing occupations,
nurse technicians, etc.—are a numerically significant and and even modern clerical occupations that provide coor-
rapidly growing category of relatively well-remunerated, dination and decision-making functions rather than sim-
middle skill occupations. While these para-professions do ply typing and filing. Indeed, even as some formerly mid-
not require a college degree, they do demand one to two dle skill occupations are stripped of their routine technical
years of post-secondary vocational training. Significantly, tasks and arguably deskilled—for example the stockbroking
mastery of “middle skill” mathematics, life sciences, and an- occupation—other formerly high-end technical occupations
alytical reasoning is indispensable for success in this train- are made accessible to workers with less esoteric techni-
ing.30 cal mastery, for example, the nurse practitioner occupation
Why are these middle skill jobs likely to persist and, that increasingly performs diagnosing and prescribing tasks
in lieu of physicians. I expect that a significant stratum of
potentially, to grow? My conjecture is that many of the
middle skill, non-college jobs combining specific vocational
tasks currently bundled into these jobs cannot readily be
skills with foundational middle skills—literacy, numeracy,
unbundled—with machines performing the middle skill
adaptability, problem-solving and common sense—will per-
tasks and workers performing the residual—without a sub-
sist in coming decades.
stantial drop in quality. Consider, for example, the com-
The economics profession is very far from a full under-
standing of the interactions among rising worker skills, ad-
30 Simultaneously, it becomes more critical than ever to improve work-
vancing technology, improvements in offshoring and trade
ers’ capabilities to add value in manual and service tasks. While one opportunities, and shifting consumer demands in determin-
may legitimately be skeptical that education can do much to im-
prove productivity in labor-intensive, technologically-lagging tasks, it
ing the division of labor, the growth of aggregate produc-
is worth noting that workers with post-secondary education appear to tivity, and the level and inequality of earnings within and
earn more in essentially all walks of life—including mundane ser-
vice occupations—than do those with a high school or lesser educa- 31 Lawrence Katz memorably titles workers who virtuously combine
tion (Carnevale et al. 2011). Barring the unlikely case that these wage
differences entirely reflect self-selection on unobserved ability into technical and interpersonal tasks as “the new artisans” (see Friedman
post-secondary education rather than post-secondary education’s value 2010).
added, these earnings differentials suggest that increasing the skills and 32 In general, these same demands for interaction frequently privilege

capabilities of workers that will perform the lower skill tasks in the face-to-face interactions over remote performance, meaning that these
economy may be productive in aggregate and may have the secondary same middle skill occupations may have relatively low susceptibility
benefit of containing the growth of inequality. to offshoring.
The “task approach” to labor markets: an overview 197

between skill groups. The “task approach” to labor markets equilibrium mapping between production tasks and produc-
does not come close to offering a solution to this vast in- tive factors responds dynamically to shifting economic and
tellectual puzzle. It may however serve to put a few puz- technological forces. The “task approach” to labor markets
zle pieces into their rightful locations. If so, this will mark offers one promising, though far from complete, set of con-
progress towards a worthy goal. ceptual tools for studying these forces.
In this brief essay, I first sketch a formal model, devel-
oped by Acemoglu and Autor (2011), that offers a flexible
Executive summaries and tractable approach for analyzing the interactions among
skill supplies, technological capabilities, and potential trade
The canonical production function found in economic mod- and offshoring opportunities in shaping the demand for fac-
els is anthropomorphic—it’s built to resemble humans. It tors, the assignment of factors to tasks, and the productivity
features a role for labor and a role for machinery (capi- and hence the wage of each factor. This model removes what
tal), and, in general, these roles are distinct. This is not we view as an artificial set of distinctions typically made be-
illogical—after all, these are separate inputs. But what pre- tween labor, capital, and trade (or offshoring) in the canoni-
cisely is distinctive about the role of each input is left cal production model, where different factors of production
opaque, and the nature of the interactions among them is play distinct and often incommensurate roles. In this alter-
highly constrained. In particular, capital is either a comple-
native model, the fundamental units of production are job
ment or a substitute for labor, different types of labor are
tasks, which are combined to produce output. Tasks can be
either complements or substitutes for one another, and these
supplied by domestic labor, foreign labor, or capital, the ca-
roles are essentially fixed. Changes in the supply of each
pabilities of which may change over time. Consequently, the
input can of course affect marginal products, but each fac-
assignment of factors to tasks in this model is determined by
tor’s “purpose” in the production function is both distinct
comparative advantage; in equilibrium, only the least-cost
and static.
factor is assigned to any given task.
These restrictions stem from the fact that the canonical
The second part of the essay briefly discusses some of
production function implicitly equates two distinct aspects
the key conceptual and practical challenges that researchers
of production. One aspect is which factors are used as in-
face in bringing the “task approach” to the data. The litera-
puts, e.g., capital, high skill labor, low skill labor. The other
ture is already grappling with these issues, as I note below,
is what services these factors provide. In the canonical setup,
and I’m cautiously optimistic that progress will be rapid.
a factor’s identity and its role in the production function are
I end by highlighting some of the pitfalls that researchers
synonymous. In reality, however, the boundary between “la-
bor tasks” and “capital tasks” in production is permeable may encounter in applying the task approach, most promi-
and shifting. We observe numerous instances in which tech- nently, loose and inconsistent task definitions that tend to
nological advances allow machinery to substitute for labor sow confusion, and the beguiling temptation for researchers
in performing a specific set of tasks—for example, dispens- to confidently forecast the “future of work” based only on
ing boarding passes and assigning seats at airport check-in the very modest understanding that we have developed to
counters in lieu of gate agents—while simultaneously com- date.
plementing labor in other tasks, for example, allowing gate
agents to rapidly identify alternate flights and issue new tick-
ets. This evolving division of labor has a clear economic Kurzfassungen
logic: novel tasks—those demanded by new products, tech-
niques, or services—are often assigned first to workers be- Die gängige Produktionsfunktion, die in ökonomischen
cause workers are flexible and adaptive. As these tasks are Modellen angewandt wird, ist anthropomorph – sie ist so
formalized and codified, they become fallow for automation konstruiert, dass sie menschliches Verhalten abbildet. Sie
since machinery typically has a cost advantage over human weist sowohl der Arbeit als auch dem Kapital eine Rolle
labor in rote execution of repetitive tasks. zu – und im Allgemeinen sind diese Rollen verschieden.
A growing body of literature argues that the shifting al- Das ist nicht unlogisch, da es sich um verschiedene In-
location of tasks between capital and labor—and between putfaktoren handelt. Jedoch bleibt unklar, was genau die
domestic and foreign labor—has played a key role in reshap- jeweilige Rolle der beiden Inputfaktoren ausmacht. Zu-
ing the structure of labor demand in industrialized countries dem sind die Interaktionen zwischen beiden in diesem
in recent decades. Analyzing this phenomenon is difficult Modell hochgradig eingeschränkt. Insbesondere ist Kapi-
within the canonical production function framework, how- tal entweder ein Komplement oder ein Substitut für Ar-
ever, because the assignment of tasks to labor and capital is, beit, und unterschiedliche Arbeitstypen sind entweder Kom-
for most conceptual purposes, static. To overcome this con- plemente oder Substitute füreinander, wobei diese Rollen
ceptual hurdle, it is valuable to consider a setting where the grundsätzlich festgelegt sind. Natürlich können Änderungen
198 D.H. Autor

des Angebots jedes Inputfaktors die Grenzprodukte beein- bietet, um die Interaktionen zwischen Qualifikationsange-
flussen, aber der „Zweck“ jedes Inputfaktors in der Produk- bot, technologischen Fähigkeiten, etwaigen Handels- und
tionsfunktion ist sowohl distinkt als auch statisch. „Offshoring“-Möglichkeiten bei der Faktornachfrage, der
Diese Restriktionen leiten sich aus der Tatsache ab, dass Faktorzuweisung auf die Aufgaben, sowie der Produktiv-
die gängige Produktionsfunktion implizit zwei verschiedene ität und damit der Entlohnung jedes Faktors zu analysieren.
Aspekte der Produktion gleichsetzt. Ein Aspekt ist, welche Die künstlichen Unterscheidungen, welche typischerweise
Faktoren als Input eingesetzt werden, z.B. Kapital, hoch- zwischen Arbeit, Kapital und Handel (oder „Offshoring“)
qualifizierte Arbeit, gering-qualifizierte Arbeit. Der andere im gängigen Produktionsmodell gemacht werden, in dem
ist, welchen Nutzen diese Faktoren stiften. Nach dem gängi- verschiedene Produktionsfaktoren unterschiedliche und oft-
gen Ansatz sind die Faktor-Identität und seine Rolle in mals inkommensurable Funktionen übernehmen, werden
der Produktionsfunktion synonym. In der Realität jedoch durch dieses alternative Modell beseitigt. In diesem alterna-
ist die Grenze zwischen „Arbeitsfunktionen“ (labor tasks) tiven Modell sind die grundlegenden Einheiten der Produk-
und „Kapitalfunktionen“ (capital tasks) in der Produktion tion Arbeitsanforderungen (job tasks), die kombiniert wer-
durchlässig und veränderlich. Wir beobachten unzählige den, um einen Output zu produzieren. Diese Aufgaben kön-
Beispiele, wo bei bestimmten Tätigkeiten Arbeit aufgrund nen durch heimische Arbeit, ausländische Arbeit oder Kap-
des technischen Fortschritts durch Kapital ersetzt wird – ital erledigt werden, wobei sich die jeweiligen Fähigkeiten
zum Beispiel bei der Ausgabe von Flugtickets und von Sitz- über die Zeit verändern können. Daraus folgt, dass in diesem
plätzen an Check-in-Schaltern in Flughäfen – während gle- Modell die Faktorverteilung auf die Aufgaben vom kom-
ichzeitig bei anderen Aufgaben und Tätigkeiten ein komple- parativen Vorteil bestimmt wird; im Gleichgewicht ist einer
mentäres Verhältnis zwischen Arbeit und Kapital entsteht, gegebenen Aufgabe nur der günstigste Faktor zugeteilt.
etwa indem das Flughafenpersonal durch die Technik in die Der zweite Teil des Essays diskutiert kurz einige der
Lage versetzt wird, rasch Alternativflüge zu finden und neue wichtigsten konzeptionellen und praktischen Herausforderun-
Tickets auszustellen. Diese fortschreitende Zweiteilung der gen, die Forscher dabei haben, den TASKS-Ansatz em-
Arbeit hat eine klare ökonomische Logik: Neuartige Auf- pirisch zu überprüfen. Die Literatur setzt sich wie bereits
gaben – wie sie neue Produkte, Technologien oder Di- erwähnt mit diesen Fragen auseinander und ich bin verhal-
enstleistungen erfordern – werden oftmals zuerst Arbeit- ten optimistisch, dass sie schnelle Fortschritte machen wird.
nehmern zugeteilt, da Menschen flexibel und anpassungs- Abschließend hebe ich einige der Fallstricke hervor, de-
fähig sind. Sobald sie stärker formalisiert und kodifiziert nen sich Forscher bei der Anwendung des TASKS-Ansatzes
sind, sind sie reif für die Automatisierung, da Maschi- gegenüber sehen. Dies gilt insbesondere für die vagen
nen bei routinemäßiger Ausführung sich wiederholender und inkonsistenten Aufgabendefinitionen, die tendenziell
Tätigkeiten typischerweise Kostenvorteile gegenüber men- Verwirrung stiften sowie die trügerische Versuchung für
schlicher Arbeit haben. Forscher, die „Zukunft der Arbeit“ lediglich aufgrund der
Ein wachsender Teil der Literatur argumentiert, dass die sehr bescheidenen Erkenntnisse, die wir bis heute entwick-
veränderliche Allokation der Aufgaben zwischen Kapital elt haben, vorhersagen zu wollen.
und Arbeit – sowie zwischen heimischer und ausländis-
Acknowledgements Autor acknowledges funding from the National
cher Arbeit – eine Schlüsselrolle in der Neuordnung der Science Foundation (grant SES-1227334). I thank Joachim Möller and
Struktur der Arbeitsnachfrage der industrialisierten Länder Brendan Price for valuable comments that improved this essay.
in den letzten Jahrzehnten gespielt hat. Jedoch ist die Anal-
yse dieses Phänomens im Rahmen der gängigen Produk-
tionsfunktion schwierig, weil dort die Aufgabenverteilung References
auf Arbeit und Kapital, aus überwiegend konzeptionellen
Gründen, statisch ist. Um dieses konzeptionelle Hinder- Acemoglu, D., Autor, D.H.: Skills, tasks and technologies: implica-
nis zu überwinden ist es wichtig, eine Situation zu betra- tions for employment and earnings. In: Handbook of Labor Eco-
nomics, vol. 4B, pp. 1043–1171 (2011)
chten, wo die Gleichgewichtsfunktion zwischen Produktion- Acemoglu, D., Zilbotti, F.: Productivity differences. Q. J. Econ. 116(2),
saufgaben und Produktionsfaktoren dynamisch auf verän- 563–606 (2001)
derliche ökonomische und technologische Kräfte reagiert. Antonczyk, D., DeLeire, T., Fitzenberger, B.: Polarization and rising
Der „TASK-Ansatz“ für Arbeitsmärkte bietet ein vielver- wage inequality: comparing the U.S. and Germany. IZA Discuss.
Pap. 4842 (2010)
sprechendes (wenn auch keineswegs vollständiges) Set Autor, D.H., Dorn, D.: The growth of low skill service jobs and the po-
von konzeptionellen Werkzeugen, um diese Phänomene zu larization of the U.S. labor market. Am. Econ. Rev., forthcoming
studieren. Autor, D.H., Handel, M.J.: Putting tasks to the test: human capital, job
tasks and wages. J. Labor Econ., forthcoming
In diesem knappen Essay beschreibe ich zuerst ein
Autor, D.H., Levy, F., Murnane, R.J.: The skill content of recent tech-
formales Modell, entwickelt von Acemoglu and Autor nological change: an empirical exploration. Q. J. Econ. 118(4),
(2011), welches einen flexiblen und handhabbaren Ansatz 1279–1333 (2003)
The “task approach” to labor markets: an overview 199

Autor, D.H., Katz, L.F., Kearney, M.S.: The polarization of the U.S. Katz, L.F., Goldin, C.: The Race Between Education and Technol-
labor market. Am. Econ. Rev. 96(2), 189–194 (2006) ogy. Harvard University Press, Cambridge (2008). Number March
Autor, D.H., Katz, L.F., Kearney, M.S.: Trends in U.S. wage inequality: 2007
revising the revisionists. Rev. Econ. Stat. 90(2), 300–323 (2008) Katz, L.F., Margo, R.A.: Technical Change and the Relative Demand
Bessen, J.E.: Was mechanization de-skilling? The origins of task- for Skilled Labor: the United States in Historical Perspective
biased technical change. Boston University School of Law Work- (2012)
ing Paper, pp. 11–13 (2011) Levy, F., Murnane, R.J.: The New Division of Labor: How Computers
Black, S.E., Spitz-Oener, A.: Explaining women’s success: technolog- Are Creating the Next Job Market. Princeton University Press,
ical change and the skill content of women’s work. Rev. Econ. Princeton (2004)
Stat. 92(1), 187–194 (2010) Matthes, B., Christoph, B., Janik, F., Ruland, M.: Collecting informa-
Blinder, A.S.: How many US jobs might be offshorable? World Econ. tion on job tasks an instrument to measure tasks required at the
10(2), 41–78 (2009) workplace in a multi-topic survey. Technical report, November,
Blinder, A.S., Krueger, A.B.: Alternative measures of offshorability: IAB 2012
a survey approach. J. Labor Econ. (forthcoming) Miller, A.R., Treiman, D.J., Cain, P.S., Roose, P.A. (eds.): Work, Jobs
Carnevale, A.P., Rose, S.J., Cheah, B.: The college payoff: educa- and Occupations: A Critical Review of the Dictionary of Occupa-
tion, occupations, lifetime earnings. Technical report, George- tional Titles. National Academy Press, Washington (1980)
town University Center on Education and the Workforce (2011) Neal, D.: The complexity of job mobility among young men. J. Labor
Costinot, A., Vogel, J.: Matching and inequality in the World Economy. Econ. 17(2), 237–261 (1999)
J. Polit. Econ. 118(4), 747–786 (2010) U.S. Department and Training Administration of Labor Employment
Costinot, A., Donaldson, D.: How Large Are the Gains from Economic Data Dictionary: O*NET 14.0 data dictionary. Technical report,
Integration? Theory and Evidence from U.S. Agriculture, 1880– National Center for *NET Development (2009)
2002 (2012) Roy, A.D.: Some thoughts on the distribution of earnings. Oxf. Econ.
DiNardo, J.E., Pischke, J.-S.: The returns to computer use revisited: Pap. 3(2), 135–146 (1951)
have pencils changed the wage structure too? Q. J. Econ. 112(1), Spitz-Oener, A.: Technical change, job tasks, and rising educational
291–303 (1997) demands: looking outside the wage structure. J. Labor Econ.
Dustmann, C., Ludsteck, J., Schönberg, U.: Revisiting the German 24(2), 235–270 (2006)
wage structure. Q. J. Econ. 124(2), 843–881 (2009) Tippins, N.T., Hilton, M.L. (eds.): A Database for a Changing
Fackler, M.: Nissan and others add factories in emerging markets. New Economy: Review of the Occupational Information Network
York Times, 8 June (2008) (O*NET), committee ed. The National Academies Press, Wash-
Felstead, A., Gallie, D., Green, F., Zhou, Y.: Skills at Work, 1986 to ington (2010)
2006. ESRC Centre on Skills, Knowledge and Organisational Per-
formance, Cambridge (2007)
Firpo, S., Fortin, N.M., Lemieux, T.: Occupational tasks and changes David H. Autor is a Professor and Associate Department Head of the
in the wage structure. IZA Discuss. Pap. 5542 (2011) Massachusetts Institute of Technology Department of Economics. He
Friedman, T.L.: The election that wasn’t. New York Times, 23 Oct. is also a Faculty Research Associate of the National Bureau of Eco-
(2010) nomic Research and Editor in Chief of the Journal of Economic Per-
Gathmann, C., Schönberg, U.: How general is human capital? A task- spectives (published by the American Economic Association), and has
based approach. J. Labor Econ. 28(1), 1–49 (2010) served on the Board of Editors at the American Economic Journal: Ap-
Goldin, C.: A pollution theory of discrimination: male and female dif- plied Economics and the Journal of Labor Economics. Autor received a
ferences in occupations and earnings. NBER Work. Pap. 8985 B.A. in Psychology from Tufts University in 1989 and a Ph.D. in Pub-
(2002) lic Policy at Harvard University’s Kennedy School of Government in
Goldin, C., Katz, L.F.: The origins of technology-skill complementar- 1999. His current fields of specialization include labor market impacts
ity. Q. J. Econ. 113(3), 693–732 (1998) of technological change and globalization, earnings inequality, disabil-
Goos, M., Manning, A.: Lousy and lovely jobs: the rising polarization ity insurance and labor supply, and temporary help and other interme-
of work in Britain. Rev. Econ. Stat. 89(1), 118–133 (2007) diated work arrangements. Autor is the recipient of an NSF CAREER
Goos, M., Manning, A., Salomons, A.: Explaining job polarization in award for his research on labor market intermediation, the Alfred P.
Europe: the roles of technology and globalization. CEP Discuss. Sloan Foundation Fellowship, and the Sherwin Rosen Prize in 2008
Pap. dp1026 (2011) for outstanding contributions in the field of Labor Economics. In 2012,
Handel, M.J.: What do people do at work: a profile of U.S. jobs from he was elected to the American Academy of Arts and Science, and he is
the survey of workplace skills, technology, and management prac- also a Fellow of the Society of Labor Economists and was a recipient
tices (STAMP). Technical report, OECD, forthcoming of both the John T. Dunlop Outstanding Scholar Award given by the
Heckman, J.J., Honore, B.E.: The empirical content of the Roy model. Labor and Employment Relations Association (2006) and MIT Under-
Econometrica 58(5), 1121–1149 (1990) graduate Economics Association Teaching Award (2005). Autor is also
Heckman, J.J., Sedlacek, G.: Heterogeneity, aggregation, and market Co-director of the MIT School Effectiveness and Inequality Initiative
wage functions: an empirical model of self-selection in the labor (SEII), and Associate Director of the NBER Disability Research Cen-
market. J. Polit. Econ. 93(6), 1077–1125 (1985) ter, which is funded by the Social Security Administration. He is also
Heckman, J.J., Scheinkman, J.: The importance of bundling in a a Research Associate of the IZA in Bonn, the IAB in Nuremberg, and
Gorman-Lancaster model of earnings. Rev. Econ. Stud. 54(2), the Institute for Research on Poverty at the University of Wisconsin,
243–255 (1987) Madison. Prior to obtaining his Ph.D., Autor spent three years direct-
Hsieh, C.-T., Hurst, E., Jones, C.I., Klenow, P.J.: The Allocation of ing efforts in San Francisco and South Africa to teach computer skills
Talent and U.S. Economic Growth (2012) to economically disadvantaged children and adults.

File and source

File: the-task-approach.pdf
Size: 535,306 bytes
SHA-256: 602be604d3a6288604a6424f599e4a087fb5ca85d0bc9fb66516591d77b27783
Our copy: the-task-approach.pdf
Original: link.springer.com