Labor market effects of international outsourcing: How measurement matters

https://doi.org/10.1016/j.iref.2008.10.006Get rights and content

Abstract

As regards labor market effects of International Outsourcing, empirical results differ strongly. This is not only due to different data, the use of different indices adds to the puzzle. This paper investigates the importance of measurement differences for analyzing labor market effects of International Outsourcing. To this end, several indices are compared with respect to their design, their descriptive properties, their quality in proxying International Outsourcing activities, and their econometrical performance. As the results show, International Outsourcing effects depend strongly on measurement differences and the level of industry aggregation. Considering these results, different empirical findings can be reconciled.

Introduction

Since International Outsourcing moved into the focus of political and social discussion, it has been blamed to reduce relative demand for low skilled labor. Thus, beside skill-biased technological change, International Outsourcing is seen as one main culprit for labor market disruptions in industrialized countries. While outsourcing is defined as the procurement of inputs from an external supplier, it is the international component, namely the use of a production fragment produced abroad, that achieves most attention in public discussion as well as in economic research. Even if International Outsourcing is already seen as a world-wide phenomenon, Kierzkowski (2005, p. 235) correctly mentions that “it is only the beginning of what seems an inexorable process”.

In order to investigate labor market effects of International Outsourcing, a wide area of empirical research emerged.1 Since it is not possible to directly observe International Outsourcing on an aggregated macro level, there is a need to proxy it. Thus, several indices were developed and a few of them are very common in use. Within a descriptive analysis, Campa and Goldberg (1997) e.g. measure International Outsourcing using an index called vertical specialization. They show for the period 1974–1993, that International Outsourcing increased strongly in the US, Canada and the UK, but not in Japan. Hummels, Ishii, and Yi (2001) measure International Outsourcing as imported inputs used to produce products that are afterward exported. Based on OECD input-output tables they document several key aspects of International Outsourcing for various countries. Yeats (2001) uses the measure of imported inputs in total imports and mentions for a variety of countries that International Outsourcing is already at a quite high level.

Even though it is difficult to classify empirical contributions, they can be divided into two broad groups: some contributions show insignificant effects of International Outsourcing while others support the importance of International Outsourcing for changes on the labor market. Berman, Bound, and Griliches (1993) first estimate labor market effects of International Outsourcing using a narrow measure the parts and components purchased from abroad. Regressing the share of high skilled wages in total wages on the components of a quasi-fixed cost function, including their International Outsourcing proxy, they show for the US manufacturing sector that International Outsourcing has only small effects while it is the labor saving technical change that turns out to be the main driving force. Amiti and Wei (2004) investigate the role of service outsourcing for the US and the UK showing that it is on a much lower level but increases at a faster pace than material outsourcing in both countries. Using imported inputs in total inputs, they estimate labor market adjustment effects with a standard labor demand equation. As a result they also present only insignificant effects of service outsourcing on job growth in the UK.2 Thus, they summarize that service outsourcing does not induce a fall in aggregate employment but could lead to overall positive effects since it increases the productivity within industries.

By contrast, Feenstra and Hanson, 1996a, Feenstra and Hanson, 1996b, Feenstra and Hanson, 1999 first present a positive, statistically significant effect of International Outsourcing on the change in the non production wage share of the US manufacturing industry. Using the index imported inputs in total inputs they highlight the importance of International Outsourcing for understanding changes in labor demand and first note that measurement differences can be one crucial point for achieving different results. Egger and Egger (2002) examine the effects of International Outsourcing within the involved low-wage countries. As proxy they use i.a. imported inputs in total imports and find a significant positive (negative) effect of imports (exports) on wages in the manufacturing industry. Focusing on the manufacturing sector in France, Strauss-Kahn (2003) shows that International Outsourcing contributes significantly to the decline of the share of unskilled workers in employment. She bases her calculations on an index called vertical specialization and, like Berman et al. (1993) and Feenstra and Hanson, 1996b, Feenstra and Hanson, 1997, estimates labor market effects using a cost share equation of a translog function. Hijzen, Görg, and Hine (2004) estimate the effects of International Outsourcing on labor demand in the manufacturing sector in the UK using a very narrow measure inputs in an industry imported from the same industry. As a result they note that International Outsourcing nevertheless has a strong negative effect on the demand of low skilled workers and thus, is an important component in explaining the changing skill structure. Geishecker and Görg, 2005, Geishecker and Görg, 2008 show for the German economy that International Outsourcing may have different adjustment effects for different levels of industry aggregation. As index they use imported inputs in total output. While for the manufacturing sector as a whole, effects of International Outsourcing are not significant, results differ when considering a more disaggregated industry level. Estimating a microeconomic log wage equation they show that, while low skilled workers in the low skill intensive industries experience significant reductions in their real wage, there is no such effect for low skilled workers in the high skill intensive industries. On the other hand, high skilled workers significantly gain from fragmentation only in the high skill intensive industries while the effect on their real wage in the low skill intensive industries is not significant. Hijzen (2007) investigates the effects of International Outsourcing and skill biased technical change on factor prices in the UK for the period 1993–1998. Using two indices, a more general and a narrow one, he shows that International Outsourcing effects are significant, however, technological change is the predominant force behind the change in relative wages. As Egger and Egger (2005) mention, most of the empirical contributions analyzing International Outsourcing do not control for possible spill over effects between industries. When considering these important features, labor market implications of International Outsourcing are expected to get even magnified. Table 1 highlights the indices used and the effects achieved by summarizing the existing literature.

As this literature review shows, empirical results differ strongly. Some of them support the importance of International Outsourcing for changes in factor prices, others refuse these effects and show that it is the skill biased technical progress that matters most. The use of different indices to proxy International Outsourcing activities adds to the puzzle. Basic presumptions made on the origin of different empirical results are that measurement differences may play a crucial role (Feenstra & Hanson, 1996b) and that the level of industry aggregation matters (Geishecker and Görg, 2005, Amiti and Wei, 2005). This contribution answers both of them. Investigating if measurement differences may be one reason for achieving different empirical results it turns out that fundamental differences exist between different International Outsourcing indices. While some of them achieve significant results on more aggregated industry levels, others significantly affect labor markets on more disaggregated industry levels. Shedding more light on measurement differences in combination with the aggregation bias, the comparative analysis in this paper reconciles different empirical findings.

The remainder of the paper is structured as follows. Section 2 investigates the design of commonly used International Outsourcing indices. They are formally defined and compared with respect to their theoretical compositions. In order to assess the descriptive properties of the indices, International Outsourcing is calculated for Germany within the period 1991–2000 in Section 3. Section 4 investigates the quality of the indices. Therefore, several shift share analyses are applied examining whether the indices really capture International Outsourcing activities. Section 5 analyzes the performance of the different indices when they are used to estimate labor market adjustment effects. Based on data from the German Socio Economic Panel (GSOEP) several panel data estimations are applied regressing the within industries' wage differential on the different indices and control variables. Section 6 concludes by summarizing the major findings.

Section snippets

Design: first theoretical differences

In order to examine measurement differences, the paper considers four International Outsourcing indices very common in use: Imported Inputs in Total Imports (IITM), Imported Inputs in Total Inputs (IITI), Imported Inputs in Gross Output (IIGO), and Vertical Specialization (VS).3

Measurement: analyzing descriptive properties

As the second step, this section investigates descriptive properties of the indices. Within a German case study, the level as well as the development of International Outsourcing are calculated for several levels of industry aggregation.9

Quality: extracting the driving forces

This section investigates the quality of the different indices. With respect to the design of the indices it is possible that they indeed capture International Outsourcing activities but that there are other forces at work that drive the main part of their variance. If e.g. a highly vertically specialized industry increases its share of production relative to economy wide production, an index is forced to increase even though the outsourcing activity of the industry has not changed or even

Performance: estimating adjustment effects

In order to investigate the importance of measurement differences for estimating labor market adjustment effects, several panel data analysis are applied in this section. However, before entering the econometrics, the expected effects of International Outsourcing on the wage differential are discussed from a theoretical point of view. Among the various theoretical contributions on how International Outsourcing affects domestic labor markets, few of them are well known and regarded as milestones

Conclusions

The results of empirical contributions investigating labor market effects of International Outsourcing differ strongly. While some of them significantly support the importance of International Outsourcing to explain changes on the labor market, others yield only insignificant effects. The different results depend on different data situations, however, the use of different indices to proxy International Outsourcing activities adds to the puzzle. This paper examines the importance of measurement

Acknowledgements

Thanks are due to Klaus Beckmann, Michael Berlemann, Daniel Bernhofen, Barbara Dluhosch, Mario Holzner, Ronald Jones, Michael Landesmann, Thierry Mayer, Daniel Mirza, Nikolas Mueller-Plantenberg, Ines Pelger, Robert Stehrer, Marcel Thum, Joachim Wagner and Julia Woerz.

References (39)

  • KierzkowskiH.

    Outsourcing or fragmentation: Blessing or threat

    International Review of Economics and Finance

    (2005)
  • KimuraF. et al.

    Two-dimensional fragmentation in East Asia: Conceptual framework and empirics

    International Review of Economics and Finance

    (2005)
  • KohlerW.

    A specific-factors view on outsourcing

    North American Journal of Economics and Finance

    (2001)
  • AmitiM. et al.

    Fear of service outsourcing: Is it justified?

  • AmitiM. et al.

    Service off shoring, productivity, and employment: Evidence from the United States

  • ArndtS.W.

    Globalization and the gains from trade

  • ArndtS.W.

    Super-specialization and the gains from trade

    Contemporary Economic Policy

    (1998)
  • BardhanA.D. et al.

    The new wave of outsourcing, research report, Fischer Center for Real Estate and Urban Economics

    (2003)
  • BermanE. et al.

    Changes in the demand for skilled labor within U.S. manufacturing: Evidence from the annual survey of manufacturing

  • Cited by (26)

    • Offshoring, industry heterogeneity and employment

      2021, Structural Change and Economic Dynamics
    • The effects of service offshoring on employment

      2019, Structural Change and Economic Dynamics
      Citation Excerpt :

      The results of the above research show some differences in terms of the sign and extent of the impact of offshoring on employment levels. As pointed out by Horgos (2009), apart from the data used in the analysis, this is due to the different offshoring indices used, as well as the level of disaggregation of the sectoral data. In general, a higher degree of disaggregation leads to negative effects, while the effects obtained take a positive sign when the analyses are based on more aggregated data.

    • Foreign capital, public infrastructure, and wage inequality in developing countries

      2014, International Review of Economics and Finance
      Citation Excerpt :

      However, an enormous number of empirical studies contend that the contemporary developing countries have suffered from the widening skilled–unskilled wage gaps during the last decades. The related empirical literature can be referred to Feenstra and Hanson (1996), Alarcon and Mckinley (1997), Wood (1997), Beyer, Rojas, and Vergara (1999), Feenstra and Hanson (2003), Kijima (2006), Horgos (2009), Lam and Liu (2011), Mehta and Hasan (2012), and Kamal, Lovely, and Ouyang (2012), which strongly shows that such phenomenon has widely prevailed in some Asian and Latin American countries. The increasing skilled–unskilled wage inequality in developing countries arouses the great interest of many economic theorists.

    • Can we really trust offshoring indices?

      2013, Structural Change and Economic Dynamics
    • Multi-sourcing as an entry deterrence strategy

      2013, International Review of Economics and Finance
      Citation Excerpt :

      Thus, multi-sourcing acts as an entry deterrence strategy. In light of the fact that existing literature did not pay much attention to vertical technology transfer and multiple sourcing, which are the focuses in the present study, we can relate, therefore, this paper to the literature on international outsourcing (Antràs & Helpman, 2004; Deardorff, 2005; Glass & Saggi, 2001; Grossman & Helpman, 2002, 2003; Horgos, 2009; Jones, 2005; Long, 2005; Marjit & Mukherjee, 2008; Matsuoka & Fukushima, 2010; Shy & Stenbacka, 2003; Tomiura, 2009). The remainder of the paper is organized as follows.

    • Public infrastructure provision and skilled-unskilled wage inequality in developing countries

      2012, Labour Economics
      Citation Excerpt :

      However, enormous empirical researches and data suggest that developing countries are instead suffering from increases in the skilled–unskilled wage gap. Representative empirical literature includes Feenstra and Hanson (1996), Alarcon and Mckinley (1997), Wood (1997), Khan (1998), and Horgos (2009), which strongly shows that the skilled–unskilled wage inequality has risen in Southeast Asian and Latin American countries. Theoretical economists have long tried to explain the reason behind the growing wage inequality in developing countries.

    View all citing articles on Scopus
    View full text