Offshoring and occupational wages: Some empirical evidence

This article was downloaded by: [Heriot-Watt University]On: 08 October 2014, At: 16:47Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH,UK

The Journal of InternationalTrade & Economic Development:An International andComparative ReviewPublication details, including instructions for authorsand subscription information:http://www.tandfonline.com/loi/rjte20

Offshoring and occupationalwages: Some empirical evidenceArne Bigsten a b , Dick Durevall a b & Farzana Munshi ca Department of Economics , University ofGothenburg , Gothenburg , Swedenb Gothenburg Centre for Globalisation andDevelopment , Gothenburg , Swedenc Department of Economics and Social Sciences , BracUniversity , Dhaka , BangladeshPublished online: 20 Apr 2011.

To cite this article: Arne Bigsten , Dick Durevall & Farzana Munshi (2012) Offshoringand occupational wages: Some empirical evidence, The Journal of International Trade &Economic Development: An International and Comparative Review, 21:2, 253-269, DOI:10.1080/09638191003615612

To link to this article: http://dx.doi.org/10.1080/09638191003615612

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all theinformation (the “Content”) contained in the publications on our platform.However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, orsuitability for any purpose of the Content. Any opinions and views expressedin this publication are the opinions and views of the authors, and are not theviews of or endorsed by Taylor & Francis. The accuracy of the Content shouldnot be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions,

http://www.tandfonline.com/loi/rjte20

http://www.tandfonline.com/action/showCitFormats?doi=10.1080/09638191003615612

http://dx.doi.org/10.1080/09638191003615612

claims, proceedings, demands, costs, expenses, damages, and other liabilitieswhatsoever or howsoever caused arising directly or indirectly in connectionwith, in relation to or arising out of the use of the Content.

This article may be used for research, teaching, and private study purposes.Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expresslyforbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

Dow

nloa

ded

by [

Her

iot-

Wat

t Uni

vers

ity]

at 1

6:47

08

Oct

ober

201

4

http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/page/terms-and-conditions

Offshoring and occupational wages: Some empirical evidence

Arne Bigstena,b, Dick Durevalla,b* and Farzana Munshic

aDepartment of Economics, University of Gothenburg, Gothenburg, Sweden;bGothenburg Centre for Globalisation and Development, Gothenburg, Sweden;

cDepartment of Economics and Social Sciences, Brac University, Dhaka, Bangladesh

(Received 6 February 2009; final version received 12 January 2010)

Offshoring has changed the pattern of international competition; laborin specific occupations rather than whole firms and sectors are nowfacing competition. Accordingly, wages in offshorable occupations areaffected in new ways. In this article, we investigate the effects ofoffshoring on relative occupational wages in 13 countries for 1990–2003.Our findings show that offshoring competiveness is associated withhigher relative wages in offshorable occupations, and that export growthof IT-related services leads to higher relative wages in offshorableoccupations, whereas import growth of such services reduces them.

Keywords: offshoring; globalization; service trade; wages; outsourcing

JEL Classifications: F1; F15

1. Introduction

Offshoring1 of electronically traded services from high-wage to low-wagecountries began in earnest in the early 1990s. Increased trade and capitalmobility together with improvements in information and telecommunicationtechnology have accelerated the process, and more and more IT-relatedservices are being offshored (Amiti and Wei 2005; Blinder 2005; van Welsumand Vickery 2005). This new wave of globalization brings new challenges:both high-skilled workers (e.g. software engineers, researchers, and analysts)and low-skilled workers (e.g. call center operators and data entry clerks) inhigh-wage developed countries now face competition from their counter-parts in low-wage developing countries. Competition and specializationincrease productivity in general, and therefore, the prosperity of theparticipating countries. But does offshoring benefit everyone? In this article,we undertake an exploratory analysis of available cross-country data to seeif we can identify any systematic effects of the offshoring of IT-relatedservices on relative wages.

*Corresponding author. Email: [email protected]

The Journal of International Trade & Economic DevelopmentVol. 21, No. 2, April 2012, 253–269

ISSN 0963-8199 print/ISSN 1469-9559 online

� 2012 Taylor & Francis

http://dx.doi.org/10.1080/09638191003615612

http://www.tandfonline.com

Dow

nloa

ded

by [

Her

iot-

Wat

t Uni

vers

ity]

at 1

6:47

08

Oct

ober

201

4

One of the reasons behind the apprehension in the public debate and inthe media over offshoring of electronically traded services in developedcountries is the fear of job loss and downward pressure on real wages incertain high-skilled occupations, where these countries have had acomparative advantage for a long time (Amiti and Wei 2005; Coe 2008).Many of these jobs have been offshored to developing countries where thejob (task) can be done at much lower cost and delivered electronically atnegligible cost.

By offshoring routine tasks to low-wage developing countries, producers inhigh-wage developed countries can specialize in complex tasks in which theyhave comparative advantages, and hence expand and create more such jobs.Overall, the real incomes of the origin (developed) country are likely toincrease as a result of cheaper imports from the destination (developing)country. The idea is similar to the standard gains from trade story: trade(offshoring) reduces cost of labor if the various segments in the productionchain are allocated according to comparative advantages. Although someworkers will of course suffer from temporary dislocation due to the trade, fullemployment can be attained in the long run through market adjustments.Therefore, also this trade can bring benefits to all participating countries.

Ireland is a good example in this context. In the early 1990s, it attracted alot of US offshored services, since wages were competitive. However, as wagesrose in Ireland, the USA and other offshoring countries started to look for newdestinations. For example, China, the Philippines, Malaysia, and India, inparticular, have become attractive offshoring locations. However, by nowwages in some offshored occupations are increasing rapidly in India as well.Interestingly, the search-engine company Like.com decided to close its Indiancenter, because it was not cost-effective anymore; wage levels of softwareengineers in Bangalore and California had become too similar.2

Research on the impact of globalization or market integration on labormarkets, factor prices, production patterns, and welfare has mainly beenbased on models with complete goods produced in one location. Withincreased offshoring, trade in specific tasks or intermediate goods has cometo play a very significant role, which also affects factor price movements. Todate, little research has been done on the impact of offshoring on incomedistribution in the participating countries, although some studies haveanalyzed the effects of offshoring on labor demand (Ekholm and Hakkala2006), on employment (Scholler 2006; Hijzen et al. 2007b), and on relativewages of skilled to unskilled workers (Feenstra and Hanson 1996, 1999;Grossman and Rossi-Hansberg 2006).

We analyze a panel of 13 countries in Europe and Asia for the period1990–2003. To our knowledge, this is the first study that exploits varia-tions across countries to try to understand the potential effects of off-shoring on relative wages in offshorable and non-offshorable occupations.We find that offshoring competiveness, measured with A.T. Kearney’s

254 A. Bigsten et al.

Dow

nloa

ded

by [

Her

iot-

Wat

t Uni

vers

ity]

at 1

6:47

08

Oct

ober

201

4

(2005) Global Services Location Index (GSLI)TM, is associated with higherrelative wages in offshorable occupations. Moreover, increases in exportsof IT-related services are associated with higher relative wages, while thereis some, but weaker, evidence of a reverse effect for increases in imports.This suggests that increased relative demand for these occupations booststheir relative wages. We also find that the differences in relative wages arethe largest in developing countries, and that they decline as GDP (GrossDomestic Product) per worker increases across countries. Since we do notanalyze a full model and only study 13 countries over 14 years, thesefindings should be treated with some caution but we still believe that theyare interesting and may inform the debate.

The remainder of the article is organized as follows: Section 2 describesoffshorable jobs and the countries involved in offshoring, Section 3 discussessome relevant literature on offshoring and our analytical approach, Section 4describes the data and the variables used in the empirical analysis, Section 5presents the results, and Section 6 concludes the article.

2. Offshoring

As a background to the analysis, we begin by briefly describing thecharacteristics of offshorable jobs and the countries to which these jobs areoffshored.

2.1. Offshorable jobs

This article analyzes the offshoring of what is defined as Mode 1 servicesunder the General Agreement on Trade in Services (GATS). These servicesare traded electronically and often over long distances. van Welsum andVickery (2005) identify a set of occupations which potentially would beaffected by offshoring. According to their classification, offshorable jobs areoccupations that embody high explicit information but do not require face-to-face contact (codified knowledge), and most importantly, information andcommunication technology is intensively used in the process of production aswell as in trade of the service. More generally, ‘routine’ tasks (Levy andMurname 2004) and ‘codifiable’ tasks (Leamer and Storper 2001) are mostsuitable for offshoring. van Welsum and Vickery (2005) predict that around20% of total employment in Organization for Economic Co-Operation andDevelopment (OECD) countries is potentially offshorable, while Bardhan andKroll (2003) predict that about 11% of total US employment is.

Offshoring can involve both high-skill and low-skill jobs. Examplesof workers include call center operators (Friedman 2004), radiologists(Pollak 2003), software programmers (Thurm 2004), and people preparingtax forms (Robertson et al., 2005). Non-offshorable occupations are thosethat require face-to-face personal contact and where the services cannot

The Journal of International Trade & Economic Development 255

Dow

nloa

ded

by [

Her

iot-

Wat

t Uni

vers

ity]

at 1

6:47

08

Oct

ober

201

4

be delivered electronically. Most personal services are non-offshorable;e.g. taxi drivers, child-care workers, nurses, barbers, automobile mecha-nics, cooks, as well as jobs in construction and mining. Hardly anyphysician jobs, except in radiology, are offshorable, since they requireface-to-face personal contact. Some government services may not requireface-to-face contact, but are still generally not offshorable for politicalreasons.

With technological progress, more and more jobs are becomingoffshorable. Blinder (2006) and Jensen and Kletzer (2005) predict that thenumber of offshorable jobs in the USA will soon exceed the number ofmanufacturing jobs. For example, services in wholesale and retail trade aregood candidates for offshoring due to the steady increase in internetretailing. In the health sector, laboratory tests are suitable for offshoring.Jobs in financial services as well as attorneys writing routine contracts canalso be offshored. However, most jobs in the tourist sector cannot beoffshored, except for reservation clerks. Furthermore, research-relatedand innovative jobs use substantial information technology but are notoffshorable. Leamer and Storper (2001) use the term ‘double-edgedgeography of the internet age’ to explain how the internet is spreading outroutine tasks while at the same time concentrating (agglomerating) researchand innovative activities.

2.2. Inshoring countries

The GSLI, developed by A.T. Kearney (a leading international managementconsulting firm), provides a ranking of offshore locations according totheir attractiveness to investors, a measure we use in the analysis. It iscreated by evaluating 40 countries based on 40 individual metrics groupedinto three criteria: financial attractiveness, skills and availability of people,and business environment.3 Financial attractiveness includes compensationlevels, infrastructure, and tax and regulatory costs. People skills andavailability is based on remote services sector experiences and qualityratings, labor force availability, education and language, and attrition risk.Business environment includes infrastructure, cultural exposure, andsecurity of intellectual property. The GSLI is thus a broad measure ofcompetiveness in providing offshoring services.

In the GSLI list, the top six countries are Asian: India, China, Malaysia,the Philippines, Singapore, and Thailand. Central and Eastern Europe areattractive to many non-English speaking high-wage European countries dueto language skills and geographical proximity; skills in European languagesother than English are uncommon in low-wage Asian countries. Amitiand Wei (2005) show that four out of the five top outsourcers of businessservices in 2002 were non-English speaking countries: Germany, Japan, theNetherlands, and Italy.


Dow

nloa

ded

by [

Her

iot-

Wat

t Uni

vers

ity]

at 1

6:47

08

Oct

ober

201

4

3. Literature review

The classical models of trade focussed on the exchange of finished goods.Since communication costs were high, it was reasonable to assume thatsplitting-up of production processes was only done in close proximity. Overtime, technological improvements and the decline of communication costshave made more and more goods and services tradable, which have led tochanges in the pattern of comparative advantages.

Rapid improvements in telecommunication technology since the 1990shave made the cost of trading services small, which has increased thetradability of services. Blinder (2006) calls this the ‘Third IndustrialRevolution’. The concurrent opening-up of large economies like China,India, and Russia has speeded up of this process.

The splitting-up of production processes has made high-wage (devel-oped) countries offshore various labor-intensive tasks of production to low-wage countries. Several empirical studies have found that offshoring oflabor-intensive jobs has increased wage inequality in developed countries(Feenstra and Hanson 1999; Hijzen 2007a), and it has been argued that itleads to a tendency towards factor price equalization (Deardorff 2001).

To our knowledge, there are no studies investigating the impact ofoffshoring of IT-related services on relative wages of offshored occupations.However, some studies (Feenstra and Hanson 1999; Ekholm and Ulltveit-Moe 2007) analyze the impact of offshoring on relative wages of skilledworkers using the definition of offshoring as import of all intermediateinputs, both material and services. Feenstra and Hanson (1999) findthat over 1979–1990 offshoring was responsible for 17.5%–40% of theincrease in the relative wages of non-production US workers. Ekholm andUlltveit-Moe (2007) develop a general equilibrium model of imperfectcompetition, where the impact of offshoring on relative wages of skilledworkers depends on two opposite forces: vertical specialization andcompetition. Greater vertical specialization may increase the skill premiumand therefore wage inequality in industrialized countries. On the other hand,increased competition may reduce the wage premium and therefore wageinequality in the same countries. The recent fall in the gap between non-production and production workers in US manufacturing may, according toEkholm and Ulltveit-Moe (2007), be due to the dominance of the secondforce.

Also, some papers study the impact of offshoring on employment.Scholler (2006) finds that offshoring had a negative impact on manufactur-ing employment in Germany. However, by analyzing data for several OECDcountries for the period 1995–2003, van Welsum and Reif (2005) concludethat there was no absolute decline in employment in most countries,although the occupations in question did experience slower employmentgrowth during the period.


Dow

nloa

ded

by [

Her

iot-

Wat

t Uni

vers

ity]

at 1

6:47

08

Oct

ober

201

4

The emergence of offshoring has been a challenge to the standardtheoretical predictions of the impact of globalization on skilled andunskilled employment and wages.4 First, offshoring involves both high-skilland low-skill occupations. It is therefore difficult for the traditionalcomparative advantage theory to identify clear ‘‘winners’’ (skilled workersin developed countries) and ‘‘losers’’ (unskilled workers in developedcountries) of globalization in terms of skill-level. Second, in addition totrade in complete goods as in standard trade theory, trade in specific tasks orintermediate goods has come to play an increasingly significant role ininternational factor price movements.

The changing pattern of global trade caused by offshoring demands newtheories, and some have begun to emerge regarding both trade in tasks andtrade in final goods.5 Several authors have developed models to investigatethe impact of offshoring on factor incomes (Jones and Kierzkowski 1990,2001; Arndt 1997; Kohler 2004). Generally, these models conclude thatalmost anything can happen to the structure of earnings depending on theconfiguration of sectoral factor-intensities, the relative factor intensities inthe task reallocated abroad, and relative factor endowments.

However, recently Grossman and Rossi-Hansberg (2008) presented amodel yielding somewhat more clear-cut results by placing certainrestrictions on the offshoring technology. Instead of assuming that certainactivities in certain industries can be offshored (as was done in previouswork), they assume that that the offshoring technology only varies acrosstasks and not across industries. This means that the productivity effect isconcentrated to the sector that makes intensive use of offshorable tasks.Grossman and Rossi-Hansberg decompose the effect into three compo-nents: a productivity effect, a relative-price effect, and a labor-supplyeffect. The productivity effect refers to the reduction in production costsderived from increased offshoring. This means that a firm can obtaininputs more cheaply than before and hence become more profitable,which in turn leads to an increase in the demand for labor in offshorableoccupations. The relative-price effect refers to the effect of changes inrelative goods prices due to offshoring. Reduced prices of goods thatuse offshorable labor intensively will tend to reduce wages of labor inoffshorable occupations in the origin country (the Stolper–Samuelssonprediction). The labor-supply effect is similar to the impact of increasedlabor supply in an economy, which will generally decrease wages. Theeffect arises because workers in offshorable occupations are released fromtheir jobs due to offshoring. If the productivity effect dominates the othertwo, then wages of the workers in offshorable jobs may actually increase.When putting their model to use, they find evidence that the combinedeffect of productivity and labor-supply effects led to increased real wagesof unskilled US workers over 1997–2004 (Grossman and Rossi-Hansberg2006).6


Dow

nloa

ded

by [

Her

iot-

Wat

t Uni

vers

ity]

at 1

6:47

08

Oct

ober

201

4

Thus, the literature does not provide much guidance in terms of clear-cutpredictions that can be tested empirically. Given the ambiguities in theliterature, we use a partial equilibrium approach to investigate the question-at-hand empirically. We can think of a simple model, with relative demandand relative supply curves for offshorable versus non-offshorable labor.Increased net-export of offshorable services, ceteris paribus, raises therelative demand for workers in the offshorable occupations. If at the sametime the relative supply curve shifts less, there will be an upward pressure onwages in the offshorable occupations. The less elastic the supply of workersin offshorable occupations is, the stronger is the effect of an increase inthe relative demand for this type of workers on wages in offshorableoccupations. Hence, the faster export in the relevant sector grows, the morethe relative offshorable occupation wages increase.

It seems reasonable to hypothesize that the shape of the demand curvefor this type of labor is relatively similar across countries, or at least that themajor inter-country differences pertain to relative labor supply. In advancedeconomies, there is a relatively high elasticity of supply of labor becauselabor is flexible with regard to the types of job that it can do and labormarkets are transparent and well integrated. In poor economies with lessintegrated labor markets, the elasticity of the relative supply of labor curvewill tend to be lower. Furthermore, poor economies have a more limitedability to use the educational system to produce the skills required in theexpanding outsourcing sectors. We therefore predict that a certain increasein demand for labor increases relative wages more in poor countries than inwealthy ones.

4. Data and variables

This section explains the choice of variables and describes the data used inthe empirical analysis. Table 1 reports the descriptive statistics of the mainvariables used in the analysis.

van Welsum and Vickery (2005) identify some occupations that arepotentially affected by offshoring in Europe and other countries using theInternational Standard Classification of Occupation-1988 (ISCO-88).7 Theoccupational wages around the world (OWW) database,8 which we use inthe analysis, also follows the ISCO-88 classification. With this link we endup with wages of 23 offshorable occupations that match the classification ofvan Welsum and Vickery (2005). To evaluate the effects of offshoring onoccupational wages, we compare the wages of each of these occupationswith an average of 14 non-offshorable occupations over 1990–2003 forseveral countries. This differs from Feenstra and Hanson (1999), Ekholmand Ulltveit-Moe (2007), and Grossman and Rossi-Hansberg (2006), whererelative wages of skilled to unskilled workers are analyzed using data fromthe US manufacturing sector. A list of offshorable and non-offshorable


Dow

nloa

ded

by [

Her

iot-

Wat

t Uni

vers

ity]

at 1

6:47

08

Oct

ober

201

4

occupations with the corresponding ISCO code is reported in Tables A1 andA2 in the Appendix.

In our analysis, we focus on countries ranked by A.T. Kearney’s (2005)GSLI. Twenty-one countries, developed as well as developing, are classifiedas the most attractive offshore locations in Europe and Asia. However, dueto missing data in the OWW database, only 13 can be used. These (in orderof rank) are India, China, the Philippines, Singapore, Thailand, the CzechRepublic, Slovakia, Poland, Hungary, Romania, the UK, Germany, andPortugal.

Our variable of interest is the relative wage in offshorable occupations.We construct this variable as the ratio of each offshorable occupationalwage to the average wage level of all non-offshorable occupations. Anincrease in the ratio implies that the offshorable occupational wages increaserelative to those in non-offshorable occupations.

It is difficult to measure the extent of offshoring; only parts of theproduction processes are offshored, and they are regarded as inputs toproduce the final goods. Hence, we have to use indirect measures. Thecompetitiveness of a country as an offshoring destination, GSLI, providesinformation about the attractiveness, and more attractive locations shouldexport more business services and therefore have higher relative wages thanless attractive locations. We thus expect this variable to have a positiveimpact on relative wages in offshorable occupations. Although GSLI ispartly subjective, it is unlikely to be a biased by construction in favor of ourhypothesis, since cost advantage is an important driver of offshoring andhigh compensation costs have a negative effect on the GSLI (A.T. Kearney2005). The main drawback of GSLI is that we only have data for 2004.

There is some information on exports and imports of services;particularly ‘computer and information services’ and ‘other business

Table 1. Descriptive statistics 1990–2003.

Variables Mean Standard deviation N

Log of relative wage (the ratioof wages of offshorable tonon-offshorable occupations)

0.234 0.517 1526

Index of competiveness 5.377 0.636 1570Log of export of business servicesas a fraction of GDP

720.152 3.214 1555

Log of import of business servicesas a fraction of GDP

719.637 3.200 1555

Log of GDP per workera 10.073 0.815 1570Log of openness (trade as afraction of GDP)

4.179 0.689 1570

Note: aReal GDP per worker is in thousands of 2000 PPP dollars.


Dow

nloa

ded

by [

Her

iot-

Wat

t Uni

vers

ity]

at 1

6:47

08

Oct

ober

201

4

services’ from the IMF balance of payments database have been used asproxies for offshoring (Amiti and Wei 2005; Scholler 2006). Data on‘computer and information services’ are unfortunately not available formany of our countries for the study period. We therefore use information on‘other business services’ (business services for short) only, measured inUS dollars. The business services are traded electronically and include,among others, accounting jobs, management consulting, legal services,research and development, architectural and engineering services, and otherback-office operations (United Nations 2002).9 The countries in our sampleare both exporters and importers of such services, and hence we use bothexport and import of business services as measures of offshoring. These canbe viewed as indicators of the strength of offshoring. Business service exportand import are divided by GDP to control for the absolute size of theeconomy.

The impact of offshoring on wages is likely to vary across occupationsand levels of development, and there are large differences in the level ofdevelopment among the 13 countries in our sample. In the panels, we thususe country, occupation and time dummies to capture fixed effects, whenadequate. We also control for productivity (and indirectly for the levelof development) by including real GDP per worker, measured in thousandsof 2000 constant dollars, and openness, measured by total internationaltrade divided by GDP. To test for differences depending on the level ofdevelopment, interaction terms constructed with offshoring and productivityare included in some specifications.10

5. Empirical analysis

To investigate the impact of service offshoring on relative occupationalwages, various versions of the following model are estimated,

Wcit ¼ a1 þ a2OUTþ a3PRODct þ a4OPENct þDc þDi þDt þ ucit ð1Þ

The subscripts c, i, and t are indexes for country, occupation and time.The dependent variable Wcit is the log of the ratio of wages for eachoffshorable occupation relative to the country average of non-offshorableoccupational wages. OUT stands for the index of competiveness, GSLI, orthe log of export and import of business services as a share of GDP. PRODct

and OPENct denote the log of real GDP per worker and log of openness(trade as share of GDP). The fixed effects for country, occupation, and timeare represented by Dc Di, and Dt They capture unobserved effects that areconstant across countries, occupations and time. Finally, a1 is the interceptand ucit is an error term. Standard errors are corrected for clustering usingtwo-way clustering at the country and year levels, using the approachdeveloped by Cameron, Gelbach, and Miller (2006).11


Dow

nloa

ded

by [

Her

iot-

Wat

t Uni

vers

ity]

at 1

6:47

08

Oct

ober

201

4

The results from the estimated models are presented in Tables 2 and 3.We begin by estimating a simple model using the competitive measure GSLIand occupation and time fixed effects (column 1, Table 2); we do not includecountry fixed effects since GSLI is constant over time for each country. Thecoefficient has the predicted sign and is highly significant: an increase incompetiveness from 5.03 (Romania), a low level, to 6.87 (India), the highestlevel, would raise the difference in relative wages by about 60%. Thissuggests that the most competitive offshoring locations attract the mostoffshoring jobs, which results in higher relative wages in offshorableoccupations. We next add productivity and openness individually (columns2–3) and jointly (column 4) but they are not significant. The competivenessindex is still clearly significant in column 2 and 3, but drops to just below the10% significance level when both productivity and openness are included(column 4). However, the estimated coefficient does not change substantiallyin any of the specifications.

Since we expect that the impact on relative wages is smaller the moredeveloped the country is, we next add an interaction term, productivitytimes GSLI (column 5). The size of the coefficient on competiveness is nowsubstantially higher and significant at the 5% level, and the interactionterm is negative, as expected, and significant the 10% level. The coefficientson productivity and openness are both positive and significant. It isobvious from column 5 that a marginal change in competitiveness has astronger effect in the poorest countries than in the developed onessince development is closely related to high productivity. This is consistentwith our hypothesis that the impact on relative wages is likely to be higherin poor countries with lower elasticity of labor supply than in moredeveloped ones.

Table 3 reports estimations of models with business services exportsand imports as explanatory variables instead of competitiveness. Since theyvary over time, we also include country dummies with Romania as theomitted country. However, to allow comparison with the model with thecompetitiveness index, i.e. column 2 in Table 2, the first specification(column 1) does not have country dummies. The coefficients on exports andimports have the opposite signs, as expected, but only the one on exports issignificant, and productivity has a negative but insignificant coefficient.Hence, the two specifications are consistent.

Country dummies are added in column 2. This makes exportsinsignificant. Since several countries in the sample had rapidly rising exportand import shares during the study period, this could be due to slowresponses in relative wages. Hence, we add lagged exports and imports incolumn 3. The coefficients on contemporaneous and lagged exports aresignificant and of similar size but with opposite signs; the ones on importsare also similar with opposite signs, but they are not significant. The modelthus suggests that at least increased export growth of business services


Dow

nloa

ded

by [

Her

iot-

Wat

t Uni

vers

ity]

at 1

6:47

08

Oct

ober

201

4

Table

2.

Theeff

ects

ofoffshoringonrelativeoccupationalwages:1990–2003.

(1)

(2)

(3)

(4)

(5)

Competitivenessindex

0.338***(0.112)

0.257**(0.129)

0.336***(0.112)

0.236(0.145)

3.455**(1.590)

Productivity

70.095(0.081)

70.123(0.106)

1.637*(0.875)

Openness

70.023(0.078)

0.044(0.094)

0.313*(0.167)

Competitiveness6

productivity

70.351*(0.181)

Occupationdummies

Yes

Yes

Yes

Yes

Yes

Tim

edummies

Yes

Yes

Yes

Yes

Yes

Constant

72.052***(0.616)

70.651(1.394)

71.957***(0.737)

70.416(1.590)

717.68**(8.244)

Observations

1526

1526

1526

1526

1526

R-squared

0.527

0.538

0.528

0.540

0.580

Notes:Significance

levelsare

indicate

as*10%,**5%,***1%.Standard

errors,in

parentheses,are

clustered

attheoccupation/countrylevel.


Dow

nloa

ded

by [

Her

iot-

Wat

t Uni

vers

ity]

at 1

6:47

08

Oct

ober

201

4

Table

3.

Theeff

ects

onrelativeoccupationalwages

ofserviceexportsandim

ports.

(1)

(2)

(3)

(4)

(5)

(6)

Export

share

0.075***(0.026)

0.036(0.050)

0.074*(0.045)

Import

share

70.046(0.031)

0.005(0.060)

70.075(0.057)

Export

share

lagged

70.071***(0.023)

Import

share

lagged

0.067(0.050)

Growth

rate

ofexports

0.072***(0.022)

0.063***(0.022)

0.110**(0.052)

Growth

rate

ofim

ports

70.069*(0.038)

70.063(0.038)

70.126(0.197)

Productivity

70.176(0.113)

0.370*(0.194)

0.389**(0.195)

0.392**(0.186)

0.399**(0.174)

70.280***(0.093)

Openness

0.175(0.153)

0.124(0.080)

Countrydummies

Germany

70.512(0.439)

70.832**(0.394)

70.803***(0.272)

70.856***(0.250)

United

Kingdom

70.0667(0.421)

70.367(0.382)

70.333(0.267)

70.364(0.245)

Singapore

70.583*(0.334)

70.507(0.360)

70.513*(0.261)

70.898**(0.360)

Portugal

0.152(0.187)

0.142(0.194)

0.142(0.184)

0.0831(0.177)

Czech

70.188(0.217)

70.124(0.233)

70.129(0.169)

70.268(0.186)

India

1.055***(0.157)

1.084***(0.169)

1.089***(0.153)

1.152***(0.148)

China

0.398**(0.195)

0.432**(0.200)

0.439***(0.164)

0.479***(0.152)

Philippines

0.164(0.135)

0.276*(0.143)

0.279***(0.0546)

0.113(0.163)

Thailand

0.672***(0.098)

0.759***(0.104)

0.757***(0.0318)

0.557***(0.172)

Slovakia

70.175(0.192)

70.110(0.215)

70.116(0.144)

70.287(0.190)

Poland

70.056(0.172)

70.202(0.187)

70.212**(0.0944)

70.256***(0.0905)

Hungary

70.309(0.197)

70.340(0.212)

70.352**(0.150)

70.485***(0.169)

Tim

edummies

Yes

Yes

Yes

Yes

Yes

Yes

Occupationdummies

Yes

Yes

Yes

Yes

Yes

Yes

Constant

2.212**(0.955)

72.949(1.833)

73.974**(1.646)

73.912**(1.689)

74.579***(1.587)

2.221**(1.030)

Observations

1511

1511

1366

1366

1366

1366

R-squared

0.531

0.713

0.717

0.717

0.718

0.514

Notes:Significance

levels;*10%,**5%,***1%.Romania

isthebase

forthecountrydummies.Thelikelihood-ratiotest

thatmodel

(4)isnestedin

(3)isw2(2)¼

0.03,p-value

0.98.Standard

errors,in

parentheses.Thestandard

errors

are

clustered

atcountryandyearlevels(see

Cameronet

al.2006).


Dow

nloa

ded

by [

Her

iot-

Wat

t Uni

vers

ity]

at 1

6:47

08

Oct

ober

201

4

increases relative wages of offshorable occupations, whereas increasedimport growth of business services might reduce them. To test this, we re-estimate the model with growth rates of export and import shares (column4): both export growth and import growth are significant, and a likelihood-ratio test comparing the models in column 3 and 4 shows that therestrictions imposed are valid (w2(2) ¼ 0.03, p-value 0.98). The models showthat 1% increase in the growth of exports as a share of GDP contributes to0.072% increase in relative wages.

Next, openness is added, but it is not significant and does not alter theearlier findings (column 5). We also included interaction dummies betweenproductivity and exports and imports, measured both as levels and growthrates, but they were insignificant (not reported). This is probably because thenon-linearity, reported in Table 2, mainly is due to the cross-countryinformation in the data; the within-country effects, which are estimatedwhen country dummies are included, are weak because of the short-timeperiod analyzed. If productivity is interpreted as a measure of development,re-estimation of the model in column 5 without country dummies (column 6)is revealing. The coefficient on productivity is negative, and significant,indicating that less-developed countries have higher relative wages, while thecoefficient on productivity in column 5, the within-country model, ispositive, indicating that increasing productivity reduces differences inrelative wages over time.

Overall, the main results make sense and appear to be fairly consistent inall specifications; the competiveness index has a strong impact on relativewages, and export growth of business services raises relative wages of theoffshorable occupations, while import growth tends to reduce wagedifferences. The finding in Table 2 that the impact is smaller the moredeveloped the country is, does not hold when we exploit all the panel datainformation. The level of development (productivity) has a negative effectacross countries, while it tends to raise relative wages for offshorableoccupations within countries when controlling for unobserved country fixedeffects. The cross-country effect is probably due to higher wages indeveloped countries for non-offshorable jobs, since most of them are inthe non-tradable sector (the Balassa–Samuelson model), whereas the within-country effect could be due to a more rapid impact of increases inproductivity on offshorable occupations, which are in the tradable sector, incombination with our relatively short sample.

6. Concluding remarks

Rapid improvements in information and communication technology duringthe last decade have increased the tradability of services. Such trade hasreceived a lot of media and political attention recently, especially indeveloped countries. Although service offshoring is still at a relatively low


Dow

nloa

ded

by [

Her

iot-

Wat

t Uni

vers

ity]

at 1

6:47

08

Oct

ober

201

4

level, it is growing with technological improvements. It is therefore of greatinterest to understand its impact on wages and income distribution.

We have analyzed whether offshoring of service jobs changes the wagesin offshorable occupations relative to wages in other occupations, using dataon relative occupational wages in 13 countries for 1990–2003. Our findingsshow that offshoring competiveness is associated with higher relative wagesin offshorable occupations, and that export growth of IT-related servicesleads to higher relative wages in offshorable occupations, whereas importgrowth tends to reduce them. There is also some evidence that the impact ofoffshoring on relative wages across countries is larger the lower the level ofdevelopment, although our finding is probably due to the Balassa–Samuelson effect where relative wages in the non-tradable sector are higherthe higher the GDP per capita. To detect this effect within countries, muchlonger time series are required than currently available.

Instead of debating whether offshoring is good or bad, the focus in themore developed countries should primarily be on how much of the increasednational income caused by offshoring should be redistributed for re-trainingand unemployment insurance for the workers whose jobs have beenoffshored. Interventions to deal with structural change and incomedistribution problems are probably also important for the maintenance ofpopular support for a policy of openness (Coe 2008). In the poorercountries, wage differentials are increasing more strongly, while the capacityto intervene to deal with inequities is more limited. Still, one can hardlydeplore the fact that skilled jobs are moving there and increasing theiroverall income level, even if it takes a longer time for the effects to spread tobroader segments of the population.

The empirical results in this article are quite strong statistically, but ourfindings are admittedly tentative – addressing the issues at hand in a moresatisfactory fashion requires better data. We should also admit that there is as yetno firm theoretical basis with clear-cut predictions, and that we were restricted todo a partial equilibrium analysis using incomplete measures of offshoring and alimited number of occupational wage data observations. Nonetheless, we dobelieve that the difference between developing and developed countries withregard to the impact is interesting and significant, and that it should survive amore comprehensive econometric analysis with better data.

Notes

1. Offshoring in this article covers both intra-firm and arm’s length offshoring,because we cannot differentiate between the two forms in our data.

2. This information is accessed online at http://influencepeddler.blogspot.com3. We use the index published in 2005 (A.T. Kearney 2005). In the most recent

indexes, 50 countries are ranked with 43 metrics.4. Following the two-by-two-by-two Hecksher–Ohlin intuition, integration in

commodity markets induces factor price convergence between developed anddeveloping countries through Stolper–Samuelson effects (Stolper and


Dow

nloa

ded

by [

Her

iot-

Wat

t Uni

vers

ity]

at 1

6:47

08

Oct

ober

201

4

http://influencepeddler.blogspot.com

Samuelson 1941; Samuelson 1953). An increased wage inequality is conse-quently predicted in developed countries.

5. See Baldwin (2006) for a recent review.6. Rodriguez-Clare (2007) uses a similar approach, where the impact of offshoring

on average wages is analyzed with three effects: productivity effects, terms oftrade effects, and world-efficiency effects. Amiti and Wei (2009) find that serviceoffshoring has a positive effect on labor productivity in US manufacturing.

7. See van Welsum and Vickery (2005) for a detailed methodology on how toselect occupations.

8. Freeman and Oostendorp (2000) created this database by standardizing theILO October Inquiry data, which is essentially a large country-occupation-timematrix reporting occupational wages for 164 occupations for more than 150countries in the respective national currencies.

9. It would have been preferable to use data on all producer services, but we donot have access to comprehensive data on this variable. Our chosen variableshould be taken to be an indicator that is proxying for the ideal variable.

10. GDP per worker and openness were obtained from the Penn WorldTable 6.2.

11. The program used for clustering was written by M. Petersen and modified by D.Taylor. It is available on http://www.stanford.edu/*djtaylor/research. SeeCameron et al. (2006), Petersen (2008) and Gow, Ormazabal, and Taylor (2009)on two-way clustering.

References

Amiti, M., and S. Wei. 2005. Fear of service outsourcing: Is it justified? EconomicPolicy 20: 308–48.

Amiti, M., and S. Wei. 2009. Service offshoring and productivity: Evidence from theUS. World Economy 32: 203–20.

Arndt, S.W. 1997. Globalization and the open economy. North American Journal ofAmerican Economics and Finance 8: 71–9.

A.T. Kearney. 2005. Building the optimal global footprint, http://www.atkearney.com/ (last accessed 9 October 2008).

Baldwin, R. 2006. Globalization: The great unbundling(s). Report to PrimeMinister’s Office, Economic Council of Finland.

Bardhan, A.D., and C. Kroll. 2003. The new wave of outsourcing. Berkeley, CA:Fisher Center Research Reports, University of California.

Blinder, A.S. 2005. Fear of offshoring. CEPS Working Paper no. 119.Blinder, A.S. 2006. Offshoring: The next industrial revolution? Foreign Affairs 85: 113–28.Cameron, A.C., J.B. Gelbach, and D.L. Miller. 2006. Robust inference with multi-

way clustering (September 2006). NBER Working Paper no. T0327.Coe, D.T. 2008. Jobs on another shore. Finance and Development 45: 1.Deardorff, A.V. 2001. Fragmentation in simple trade models. North American

Journal of Economics and Finance 12: 121–37.Ekholm, K., and K. Hakkala. 2006. The effect of offshoring on labour demand:

Evidence from Sweden. CEPR Discussion Paper 5648.Ekholm, K., and K.H. Ulltveit-Moe. 2007. A new look at offshoring and inequality:

Specialization versus competition. CEPR Discussion Paper 6402.Feenstra, R.C., and G.H. Hanson. 1996. Globalization, outsourcing, and wage

inequality. American Economic Review 86: 240–5.Feenstra, R.C., and G.H. Hanson. 1999. The impact of outsourcing and high-

technology capital on wages: Estimates for the United States, 1979–1990.Quarterly Journal of Economics 114: 907–41.


Dow

nloa

ded

by [

Her

iot-

Wat

t Uni

vers

ity]

at 1

6:47

08

Oct

ober

201

4

http://www.stanford.edu/~djtaylor/research

http://www.atkearney.com/

http://www.atkearney.com/

Freeman, R.B., and R. Oostendorp. 2000. Wages around the world: Pay acrossoccupations and countries. NBER Working Paper Series 8058.

Friedman, T. 2004. What goes around . . . . The New York Times, Online edition,Feb. 26.

Gow, I., G. Ormazabal, and D. Taylor. 2009. Correcting for cross-sectional and time-series dependence in accounting research. Memo. Palo Alto, CA: Graduate Schoolof Business, Stanford University.

Grossman, G.M., and E. Rossi-Hansberg. 2006. The rise of offshoring: It’s notwine for cloth anymore. Paper prepared for the symposium sponsored bythe Federal Reserve Bank of Kansas City on ‘The New EconomicGeography: Effects and Policy Implications’, August 24–26, in JacksonHole, Wyoming.

Grossman, G.M., and E. Rossi-Hansberg. 2008. Trading tasks: A simple theory ofoffshoring. American Economic Review 98: 1978–97.

Hijzen, A. 2007a. International outsourcing, technological change, and wageinequality. Review of International Economics 15: 188–205.

Hijzen, A., M. Pisu, R. Upward, and P. Wright. 2007b. Employment, job turnoverand the trade in producer services: Firm level evidence. Research Paper 2007/37,GEP, University of Nottingham.

Jensen, J.B., and L.G. Kletzer. 2005. Tradable services: Understanding the scope andimpact of service offshoring. Institute for International Economics WorkingPaper 05-9.

Jones, R.W., and H. Kierzkowski. 1990. The role of services in production andinternational trade: A theoretical framework. In Political economy of interna-tional trade, ed. R.W. Jones and A. Krweger. Oxford: Blackwell.

Jones, R.W., and H. Kierzkowski. 2001. A framework for fragmentation. InFragmentation: New production patterns in the world economy, ed. S.W. Arndtand H. Kierzkowski, 17–34. Oxford and New York: Oxford University Press.

Kohler, W. 2004. International outsourcing and factor prices with multistageproduction. Economic Journal 58: 547–68.

Leamer, E.E., and M. Michael Storper. 2001. The economic geography of theinternet age. Journal of International Business Studies 32: 641–65.

Levy, F., and R. Murname. 2004. The new division of labor. Princeton, NJ: PrincetonUniversity Press.

Petersen, M.A. 2008. Estimating standard errors in finance panel data sets:Comparing approaches. Review of Financial Studies 22: 435–80.

Pollak, A. 2003. Who’s reading your x-ray? The New York Times, Online edition,November 16.

Robertson, J., D. Stone, L. Niedewanger, M. Grocki, E. Martin, and Ed. Smith.2005. Offshore outsourcing of tax-return preparation. The CPA Journal, OnlinePublication of New York State Society of CPA, June. http://www.nysscpa.org/cpajournal/2005/605/essentials/p45.htm

Rodriguez-Clare, A. 2007. Offshoring in a Ricardian world. NBER Working Paper13203.

Samuelson, P.A. 1953. Prices of factors and goods in general equilibrium. Review ofEconomic Studies 21: 1–20.

Scholler, D. 2006. Service offshoring: A challenge for employment? Evidence fromGermany, University of Hohenheim no. 23/2006.

Stolper, W., and P.A. Samuelson. 1941. Protection and real wages. Review ofEconomic Studies 9: 51–68.

Thurm, S. 2004. Tough shift-lesson in India: Not every job translates overseas. TheNew York Times, Online edition, March 3.


Dow

nloa

ded

by [

Her

iot-

Wat

t Uni

vers

ity]

at 1

6:47

08

Oct

ober

201

4

United Nations. 2002. Manual of statistics of international trade in services.Statistical Papers, Series M no.86.

van Welsum, D., and X. Reif. 2005. Potential offshoring: Evidence from selectedOECD countries. Paris: OECD.

vanWelsum, D., and G. Vickery. 2005. Potential offshoring of ICT-intensive occupations.In Enhancing the performance of the services sector, 179–204. Paris: OECD.

Appendix

Table A1. Offshorable occupations.

OWW code ISCO-88 code Occupations

11 2147 Coalmining engineer14 2147 Petroleum and natural gas engineer46 412/3/4/9 Office clerk (subgroup 1)52 2146 Chemical engineer76 2143 Power distribution and transmission engineer77 412/3/4/9 Office clerk (subgroup 2)45/91/130/134/140 4111 Stenographer–typist in different sectors94 3433 Book-keeper in retail trade97 4222 Hotel receptionist120 4221 Airline ground receptionist128 4223 Telephone switchboard operator129 2411 Accountant132 4114 Book-keeping machine operator133/138 2132 Computer programer in different sectors135/141 4113 Card- and tape-punching machine operator

in different sectors136 3412 Insurance agent142 412/3/4/9 Office clerk (subgroup 3)

Table A2. Non-offshorable occupations.

OWW code ISCO-88 code Occupation

13 9311 Underground helper in coalmining15 3117 Petroleum and natural gas extraction technician56 9322 Laborer in manufacturing of industrial chemical61 2230 Occupational health nurse81 7137 Building electrician82 7136 Plumber84 7141 Building painter85 7122 Bricklayer98 5122 Cook99 5123 Waiter100 9132 Room attendant or chambermaid111 8323 Motor bus driver114 3141 Ship’s chief engineer118 3143 Air transport pilot


Dow

nloa

ded

by [

Her

iot-

Wat

t Uni

vers

ity]

at 1

6:47

08

Oct

ober

201

4

Documents

Offshoring and occupational wages: Some empirical evidence