Offshoring and Outsourcing Potentials: Evidence from German Micro‐Level Data

• Author: Tobias Brändle,Andreas Koch
• Published date: 01 September 2017
• Date: 01 September 2017
• DOI: http://doi.org/10.1111/twec.12439

Offshoring and Outsourcing Potentials:

Evidence from German Micro-Level Data

Tobias Br€

andle and Andreas Koch

Institute for Applied Economic Research (IAW), T€

ubingen, Germany

1. INTRODUCTION

OVER the past few decades, technological progress in information and communication

technologies has reduced economic and cultural distances (global village scenario,

death of distance) and led to substantial growth in the national and international trade of

goods and services, both within and across the boundaries of firms (see, for instance, Hij-

zen, 2006; Merino and Rodr

ıguez, 2007). It has also facilitated the easier fragmentation of

production processes, thus generating a ‘new type’ of trade characterised by international

specialisation across different stages of single-value chains, frequently referred to as ‘tasks’

(cf. Autor, 2013), rather than across whole products.

1

The modern theoretical literature

typically envisions labour inputs as a sequence (or even a continuum) of tasks that are

differently susceptible to offshoring, depending on what is commonly referred to as the

costs of offshoring (e.g. Grossman and Rossi-Hansberg, 2008, 2012). In this literature, off-

shoring relates to the location of task performance, typically the domestic or a foreign

economy that offers some cost advantage for certain tasks. A separate strand of literature,

going back to Antr

as and Helpman (2004), focuses on contract incompleteness, which adds

an organisational dimension to fragmented value chains: the choice between arm’s length

transactions, referred to as ‘outsourcing’, and the vertical integration of input suppliers,

which leads to intra-firm trade.

2

There is a large body of empirical literature analysing the determinants and labour market

effects of offshoring. It is worth pointing out that the economic effects of offshoring, for

example on wages or employment, seem to be more pervasive and stronger than those of

trade in general (Feenstra and Hanson, 1997; Krugman, 2008; Feenstra, 2010; Harrison and

This paper is part of the research project ‘Trade in tasks –potentials for internationalization and their

effects on the wage structure and composition of employment’, funded by the German Research Founda-

tion (DFG) (BO 2793/3-1). We use data from the 1991, 1998, 2006 and 2012 waves of the BIBB/BAuA

Labour Force Survey in Germany, provided by the Research Data Centre of the Federal Institute for

Vocational Education and Training (BIBB). We thank our colleagues Bernhard Boockmann, Carina Den-

gler, Peter Eppinger, Wilhelm Kohler and Katja Neugebauer, as well as the participants of the PhD Sum-

mer Workshop at University Hohenheim, the ETSG Conference, the SEEK workshop at ZEW

Mannheim, the Brown Bag Seminar at University of T€

ubingen, the FIW-Seminar in International Econ-

omics at WIIW Wien, the Annual Conference of the Verein f€

ur Socialpolitik in Du

¨sseldorf, the MIUR-

PRIN Workshop in Milan and two anonymous referees for their useful comments. Jane Dreyer, Patrick

Maier, Elitza Milenkova, Miriam Morlock, Matthias Seckler and Lisa Tarzia are gratefully acknowl-

edged for excellent research assistance. All errors remain ours alone.

1

For an overview of the increased (international) fragmentation of production processes, see Jones and

Kierzkowski (1990) and Jones (2000).

2

See Antr

as (2016) for a comprehensive treatment of contractual incompleteness in input trade.

©2016 John Wiley & Sons Ltd 1775

The World Economy (2017)

doi: 10.1111/twec.12439

The World Economy

McMillan, 2011).

3

Almost the entire empirical literature focuses on observed levels of off-

shoring, typically applying a much broader definition of offshoring than trade in tasks as anal-

ysed in theoretical models. Much of the theoretical literature relies on the notion of different

tasks costing different amounts to offshore, which in turn evokes the notion of the offshoring

potential as a key characteristic of tasks, jobs and occupations. In this sense, theoretical mod-

els typically feature a margin separating tasks and jobs of a certain value chain that are actu-

ally performed offshore from tasks and jobs for which offshoring remains a potential. Almost

by definition, then, the notion of offshoring potential seems empirically elusive, as it does not

show up in an economic activity per se. When referring to offshoring potentials, the empirical

literature therefore typically uses general characteristics of industries (e.g. labour intensities)

or jobs (e.g. skill levels) (Blinder, 2006, 2009). In contrast, the theoretical literature empha-

sises that the costliness of offshoring is conceptually orthogonal to the labour intensity of

industries and the skill intensity of jobs (see in particular Grossman and Rossi-Hansberg,

2008). The same applies to the potential of tasks or jobs to be outsourced.

In this paper, we try to bridge this gap between the theoretical and empirical literature by

proposing and implementing an empirical metric for the potential of jobs to be offshored, and a

separate empirical metric for the potential of jobs to be outsourced. Such metrics should allow

for a better explanation of the observed levels of offshoring across jobs, industries and countries.

In addition, they allow one empirically to address observed gaps between offshoring potentials

and actual levels of offshoring, an issue that has so far not been addressed at all in the empirical

literature. And finally, with a metric at hand, one may examine empirically whether the off-

shoring potential (as opposed to the observed levels of offshoring) has an independent effect on

labour market outcomes. However, in this paper, we are interested in designing the metric as

such, and not in how offshoring potentials affect labour market outcomes.

4

The question of ‘how many jobs can be offshored’ (Blinder, 2006) remains relevant and has

not been finally answered.

5

However, more important from our perspective is the question of

‘which jobs can be offshored’. There is an intensive discussion about particular job characteris-

tics that might influence a firm’s decision to shift a job to another firm and/or country. We argue

that this decision has two dimensions. First, there are considerations determining whether to pro-

vide a certain input in the value chain inside the firm’s boundaries or to purchase it on the market

(‘make-or-buy decision’). We shall henceforth refer to these considerations as determining the

outsourcing potential of a job.

6

Second, a choice has to be made about whether to make or buy a

certain input or task on the home market or abroad. In the following, considerations that favour

3

See Mankiw and Swagel (2006), Crin

o (2009) and Wagner (2011) for recent overviews, and

Geishecker (2006), Crin

o (2010), Baumgarten (2015) and Karpaty and Tingvall (2015) for evidence on

the effects of offshoring. More recently, services offshoring and its consequences have gained more

attention in the literature; see, for example, Winkler (2010) and Eppinger (2014).

4

A recent example analysing the link between offshoring and outsourcing potentials is the working

paper by Hogrefe (2013). He analyses the effect of offshoring on relative labour demand measured as

the share of routine tasks in occupations in German manufacturing industries. The dependent variable

can also be used as a proxy for how easily jobs can be offshored in the future.

5

A recent article by Blinder and Krueger (2013) summarises the attempts and methods to answer this

question for the United States. Additionally, it can be argued that the public debate and the scientific

reality need to be further aligned (Smith, 2006).

6

With regard to the outsourcing margin, things like contractibility and relationship specificity come into

play. The most straightforward idea is that of the transaction cost approach (Whinston, 2003), which argues

that the more relationship-specific the task, the lower its outsourcing potential. The property rights theory by

Antr

as (2003) and Antr

as and Helpman (2004) furtherintroduces headquarter intensity intothis margin.

©2016 John Wiley & Sons Ltd

1776 T. BR€

ANDLE AND A. KOCH

offshore provision will be referred to as the offshoring potential.

7

Both decisions are often based

on often similar –though not identical –job characteristics. The empirical literature on trade in

tasks mainly considers the determinants of offshoring and outsourcing as a single issue (see, e.g.

Becker and Muendler, 2012; Becker et al., 2013).

8

The paper contributes to the literature by dis-

entangling the two dimensions of offshoring and outsourcing.

9

However, potential differences in

the determinants of offshoring and outsourcing have not been comprehensively considered in the

literature so far. A comprehensive empirical classification of the determinants of how easily jobs

can be offshored and outsourced is still missing.

10

The second contribution of the present paper

is to complement the existing (empirical) research by developing and testing more general mea-

sures affecting the offshoring and outsourcing potentials of occupational activities.

Our metrics for the offshoring and the outsourcing potential are based on a host of differ-

ent job characteristics. Using rich, representative individual-level data on German employees,

we can derive condensed individual, occupation, task and industry-specific indicators relying

on principal component analysis (PCA).

The paper proceeds as follows: in Section 2, we discuss the existing literature in search of job

characteristics that seem promising candidates as determining factors of offshoring and out-

sourcing potentials. Section 3 introduces the data and the construction of variables, focusing in

particular on the characteristics, identified in Section 2, that possibly determine the offshoring

and outsourcing potentials of jobs. Section 4 describes the methodology used for the construc-

tion of our two metrics for the offshoring and outsourcing p otentials. Section 5 presents a

descriptive overview of the empirical results obtained, comparing the values of our indicators

for different levels of aggregation and comparing our results with those of other indicators pro-

posed in the previous literature. Moreover, we analyse potential individual and aggregate drivers

of offshoreability and outsourceability. Section 6 concludes the paper with a brief summary.

2. LITERATURE REVIEW

a. Operationalising Offshoring

The literature on offshoring (or trade in tasks) has so far mostly addressed offshoring in terms

of one-dimensional decisions (see, for instance, Lo Turco and Maggioni, 2012; Blinder and Krue-

ger, 2013): firms either perform a job within their company and country or they do so offshore.

However, the determinants of organisationally relocating (i.e. outsourcing) a specific activity

might differ from the factors affecting the decision of whether or not to relocate an activity geo-

graphically –that is, whether to perform it in the home country or abroad. In our approach,

7

On the offshoring margin, the literature mostly stresses cost differences (wage differences) as playing

a major role in the decision by combining the given wage (or more generally, the factor price difference)

with the above-mentioned offshoring cost schedule. Intuitively, the costs of geographical dislocation

seem important here (cf. Grossman and Rossi-Hansberg, 2008).

8

This view is based on examples using single job characteristics, such as communication intensities and

interactivity requirements (Leamer and Storper, 2001; Blinder, 2006), routineness of tasks (Autor et al.,

2003; Spitz-Oener, 2006) or the standardisation of goods (Costinot et al., 2011).

9

The difference between these two dimensions was first highlighted in theoretical work by Antr

as and

Helpman (2004); see also Antr

as (2016). The empirical literature mainly considers only one direction.

Kohler and Smolka (2015) are among the very few who explicitly consider both dimensions separately.

10

This means that we consider more than one determinant, such as routines in Costinot et al. (2011) or

performance requirements in Becker and Muendler (2012) and others (see Section 2b of the paper).

©2016 John Wiley & Sons Ltd

OFFSHORING AND OUTSOURCING POTENTIALS 1777