The Implications of Globalization for Firms’ Demand for Skilled and Unskilled Labor

Scheuer, Christian; Sørensen, Anders; Rosholm, Michael Working Paper The Implications of Globalization for Firms’ Demand for Skilled and Unskilled Labor Working paper, No. 8-2007 Provided in Cooperation with: Department of Economics, Copenhagen Business School (CBS) Suggested Citation: Scheuer, Christian; Sørensen, Anders; Rosholm, Michael (2007) : The Implications of Globalization for Firms’ Demand for Skilled and Unskilled Labor, Working paper, No. 8-2007, Copenhagen Business School (CBS), Department of Economics, Frederiksberg, https://hdl.handle.net/10398/7625 This Version is available at: https://hdl.handle.net/10419/208540 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-nc-nd/3.0/ Department of Economics Copenhagen Business School Working paper 8-2007 THE IMPLICATIONS OF GLOBALIZATION FOR FIRMS`DEMAND FOR SKILLED AND UNSKILLED LABOR Michael Rosholm Christian Scheuer Anders Sørensen ____________________________________________________ Department of Economics -Porcelænshaven 16A, 1.fl. - DK-2000 Frederiksberg The Implications of Globalization for Firms’ Demand for Skilled and Unskilled Labor1 Michael Rosholm Department of Economics,Aarhus School of Business, and Centre for Economic and Business Research (CEBR), Christian Scheuer Department of Economics, Copenhagen Business School, and Centre for Economic and Business Research (CEBR) and Anders Sørensen Department of Economics, Copenhagen Business School, and Centre for Economic and Business Research (CEBR). November 2007 1Acknowledgements: The authors gratefully acknowledges financial support from the Tuborg Foundation. The usual disclaimer applies. Address: Department of Economics, Copenhagen Business School and Centre for Economic and Business Research, Porcelaenshaven 16A, 2000 Frederiksberg, Denmark. Abstract This paper investigates the impact of globalization, in the sense of increasing international trade, on the demand for skills in Danish manufacturing companies. The study is based on a unique data set that enables us to develop rich measures of international outsourcing and import penetration. Moreover, the data also allows several strategies to strengthen the causal interpretation of our results. The main finding of the analysis is that it is of crucial importance to distinguish imports - both in the form of outsourcing and overall imports - by country-of-origin. We find that international trade with low-wage countries leads to skill-upgrading.Thisis especially pronounced for import penetration with a ceteris paribus contribution of around fifty percent to skill-upgrading. Moreover, we find that import penetration in goods originating from high-wage countries lead to skill-downgrading.This latter result suggests that Danish manufacturing has comparative advantage in skillintensive production when compared to low-wage countries, but in unskill-intensive production when compared to high-wage countries. Keywords: Skill-upgrading, Low-wage country outsourcing, Low-wage country import penetration, Comparative advantage JEL:F14,J24;L60 1 Introduction This paper investigates the implications of globalization - in the form of increasing international trade - for the demand for skilled and unskilled labor in Danish manufacturing companies, where skills are measured by the educational composition of the firms’ labor input. The study is based on a unique combination of data sets yielding a matched employer-employee sample containing information on output, labor inputs, and international trade, all measured at the firm level, and covering the period 1999-2002. This data set enables us to develop measures of international trade at a very detailed level, both in terms of the type of trade and in terms of the origins of the goods traded. Specifically, our focus is international outsourcing and import penetration. We consider two testable hypotheses that are readily derived from standard trade models of comparative advantages: •Hypothesis 1: The relative demand for skilled labor increases with the extent of international outsourcing to low-wage countries; both at the firm and industry level. •Hypothesis 2: The relative demand for skilled labor increases with industry exposure to imports from low-wage countries. Moreover, we study the effect on the relative demand for skilled labor of exposure to international trade from high-wage countries; both international outsourcing and import penetration. According to the standard comparative advantage trade models it is unclear whether the relative demand for skilled labor should increase, decrease, or remain unchanged in the case of increasing trade with high-wage countries, as we have no priors as to whether Danish manufacturing has comparative advantages in skilled or unskilled labor when compared to other high-wage countries. International outsourcing is a measure of imported inputs to the production process - including the final stages of assembly - measured at the firm level. Import penetration is total imports at the (detailed) industry level and is a measure of foreign competition in product markets. For both types of international trade, we distinguish between trade with low-wage versus high-wage countries, in order to investigate if the impacts of trade are similar irrespective of origin. 1 Our main findings are that it is of crucial importance to distinguish international trade - whether in the form of outsourcing or import penetration - by country-of- origin. We find that international trade from low-wage countries - both international outsourcing and import penetration - leads to skill upgrading in Danish manufacturing firms. We also find that import penetration from high-wage countries leads to skill downgrading. The results established in relation to import-penetration suggests that Danish manufacturing has comparative advantages in processes that use skilled labor intensively when compared to low-wage countries and comparative advantages in processes that use unskilled labor intensively when compared to high-wage countries. To evaluate the importance of internationalization for skill-upgrading, we calculate the (ceteris paribus) contributions of each explanatory variable to skill-upgrading. More precisely, we multiply the mean change in explanatory variables by the matching regression coefficients and divide it by the mean change in dependent variable. Doing this we find that import penetration from low-wage countries accounts around fifty percent of the shift towards skilled labor. This is by far the largest contributor in the analysis and emphasizes the importance of import penetration from low-wage countries for skill upgrading. An important feature of globalization, and a driving mechanism for increasing skill upgrading in developed countries - in the sense of firms employing increasing fractions of workers with formal education - has been attributed to international outsourcing. Since the seminal work by Feenstra and Hanson (1996),theideathat firms offshore productive activities that use unskilled labor intensively to low-wage countries has been used intensively. Empirically, international outsourcing is often found to be of significant importance for skill upgrading, see Feenstra and Hanson 1996,1999), and Feenstra (2004). The relationship between international outsourcing and skill upgrading is usually investigated using aggregated measures of international outsourcing based on data from input-output tables, and hence, they do not distinguish between country-of- origin, i.e. they do not allow a decomposition of outsourcing to low- and high-wage countries; exceptions are Ekholm and Hakkala (2006), and Hijzen, Görg and Hine (2005). Moreover, such aggregated data exclude international outsourcing of the final stage of production, e.g., assembly; an activity where many low-wage countries have their comparative advantage, see Ng and Yeats (1999). In this paper, measures of international outsourcing are based on imports at the firm level. We develop different measures of international outsourcing that differ with respect to the type of good, its origin, and the level of aggregation. 2 More precisely, the type of goods considered includes all imports into the firm, imports from foreign manufacturing industries only, and imports from the same foreign industry as that in which the firm is located. With respect to country of origin, we divide goods according to the level of GDP per capita in the originating country. The aggregation level varies from the single firm to 316 manufacturing industries and further to 55 manufacturing industries. The established results are robust to the use of different measures of international outsourcing. Another important aspect of globalization is that increasing international competition in product markets that force domestic firms out of business or makes them change their product mix. However, import penetration is often argued to be without empirical relevance for skill upgrading. This view is supported by shift-share analyses, see for example Berman, Bound and Griliches (1994) and Autor, Katz and Krueger (1998). These studies find that within-industry effects dominate skillupgrading, whereas between-industry effects explaining skill-upgrading are modest. This evidence is interpreted as indicating that increasing competition in final goods markets is without importance for skill upgrading, because trade is argued to affect the composition of skills through a changing industry structures. A recent study by Bernard, Jensen and Schott (2006),however,finds that import penetration plays an important role in the sense that firms adjust their product mix in response to international trade pressures, especially, when exposure to competition from low-wage countries is high. This suggests that firms may shift towards more skill intensive activities as a consequence of increasing international competition in product markets and that this shift may take place within industries; the authors also find it crucial to distinguish import penetration by country-of-origin. Inspired by the findings in Bernard, Jensen, and Schott (2006), we develop measures of import penetration broken down by country-of-origin. These measures are defined as the overall imports from specific foreign manufacturing industries, and indicate the extent of product market competition from abroad. Hence, we both introduce measures of international outsourcing and import penetration after country-of-origin. According to our knowledge, this is the first paper to include both aspects of internationalization for skill-upgrading. Methodologically, our study is based on estimation of the translog cost function, hence, it allows for direct structural interpretation and inference. In order to strengthen the causal interpretation of the parameters of interest, we reformulate the translog equations in deviations from firm means, i.e. we specify a fixed effects type model, thus exploiting only within-firm variation for identification. Moreover, the different levels of aggregation used for the measures international outsourcing 3 (e.g. aggregation at different manufacturing industry levels) is essentially an instrumental variables strategy, and thus it further strengthens the causal interpretation given to these parameters, while import penetration measures are by construction aggregate measures. The next section describes the basic equations in the translog cost model and discusses the econometric strategy. Section 3 describes the data set applied in the analysis, as well as the constructed measures of international outsourcing, import penetration, and a proxy for a potentially confounding variable; skill-biased technological progress. Section 4 presents empirical results. Section 5 concludes. 2EconometricFramework Our empirical specification is based on the so-called translog cost function.1We log-linearize the cost function and follow the framework in Brown and Christensen (1981) assuming the different types of labor to be variable inputs and capital to be aquasifixed input. What distinguish the present study are two important things, firstly the specification here applies to the firm level rather than the industry level, and second, as something new we add international trade measures assumed also to be quasi fixed inputs 2. Below the model for two labor types is described, whereas the general model for more labor types - as well as a more detailed derivation of the translog cost function - is found in the appendix. The translog cost function generates a wage cost share equation of the following form: Sskilled,i,j,t =αskilled,i +β1ln µwskilled,i,t wunskilled,i,t ¶+β2ln (Yi,t)+β3ln (Ki,t) +β4ln (LWOUTSi,t)+β5ln (HWOUTSi,t) +β6ln (LWPENj,t)+β7ln (HWPENj,t) +β8ln (TECHi,t)+εskilled,i,t 1The transcendental logarithmic function was developed in Kmenta (1967) to approximate the CES-function, later applications more closely related to this study is Christensen and Greene (1976) and Berndt (1991) 2A variable input is assume to be optimally used in the short run, where as a quasi fixed input is assumed not to be subject of short run optimization, but by second order effects they affect the decision of how to optimally use the variable inputs. It is standard in this literature to treat capital as quasi-fixed, see Brown and Christensen (1981). We further argue that the opening of new markets and technological progress possess the characters of quasi fixed inputs. 4 where Sskilled,i,j,t is the cost share of skilled labor to total labor costs in firm i,located in industry j, observed at time t. The cost share of skilled labor measures the firm’s demand for skilled labor, which is derived using Shephard’s lemma. wskilled/wunskilled is the relative wage of skilled labor to unskilled labor, Yis total production, Kis the capital stock, OUTS is international outsourcing, PEN is import penetration, and TECH is a variable measuring technology level. LW and HW denote low-wage and high-wage countries, respectively. αskilled,i is a firm fixed effect, and εskilled,i,t is an i.i.d. error term. The measures of import penetration is industry measures per seandassuchtheyarelabelledj. In general estimating a translog cost function generates a series of wage cost share equations, one for each type of variable (labor) input. Imposing the standard restrictions of symmetry and homogeneity of 1 degree in prices allows us to reach the specification above where one of the equations are now redundant. In the case of two labor inputs, the estimates of the similar equation for the wage cost share of unskilled labor can be recovered from the symmetry restrictions without actually estimating the equation. The translog cost function is also estimated for three skill levels, and in this case, the restrictions have to be incorporated directly into the estimation procedure. We refer the interested reader to the appendix where the general model for htypes of labor is developed. In what follows all regressions are performed using Fixed Effects estimation techniques, implying that the model parameters are identified using only variation in the explanatory variables within firms over time. In the three labor types classification, this implies that the cost share equations are specified in deviations from individual firm averages, and this equation is then estimated using restricted maximum likelihood, incorporating all the structural parameter restrictions. Hence, our hypothesis is not that the firms with most outsourcing or import penetration has the highest share of skilled labor, but rather that within a given firm an increase in outsourcing or import penetration will cause a change towards a larger share of skilled or educated workers. Outsourcing measures at the firm level can be criticized due to endogeneity in the sense that they are determined simultaneously with the firm’s decision regarding its skill composition. To alleviate these problems, we use an approach based on the aggregation of the firm specific variables to the most detailed manufacturing industry level. This implies that the endogeneity problem remains only to the extent that the endogeneity is industry specific. This is similar to an instrumental variables approach, where the instrumental variable is substituted for the endogenous 5 The difference between import penetration and international outsourcing is that the former measures total imports from individual foreign manufacturing industries, whereas the latter measures import from all (or some) foreign manufacturing industries to the individual Danish manufacturing industry or firm. The measure of import penetration in industry k∈DB111 is defined as PENk= Total X j Mj,k where PENkrefers to import penetration, and Mj,k denotes the amount of import from foreign industry kto domestic industry j. The measures of import penetration are also calculated separately for LW- and HW-countries: PENk=HWPENk+LWPENk 3.3 Skill-Biased Technological Change It is difficult to procure measures for skill-biased technological changes. However, given the role of skill-biased technological change as a potential confounding variable, it is important to try to include a measure of it. In the literature, different variables have been used to capture the increasing efficiency of skilled and educated labor. Autor, Katz, and Krueger (1998) show that the diffusion of computers and related technologies is an important source of changes in the relative demand of skills and thereby in the relative efficiency of skilled labor. We follow Machin and Van Reenen (1998), who use R&D intensities to explain skill upgrading. Hence, R&D intensities are used as measures of technological stage of development using the OECD ANBERD database. R&D data that are compatible with the applied databases, i.e., industry-structure ISIC Revision 3 within manufacturing, only exist for the DB53 level covering 13 such industries within manufacturing. R&D intensities are defined as R&D expenditures in an industry divided by industry production. relevant product from the external trade statistics. After having calculated external trade by country-of-origin for all product accounts, the Input-Output table with external trade divided by country-of-origin is constructed using the standard method for constructing Input-Output tables, see Statistics Denmark (1986). 12 Table 3 contains a description of the main explanatory variables used in the empirical model specified below. <Table 3 about here > Note that all variables are measured in changes over time. Hence, the skilled workers’ wage share increase by over two percentage points, while at the same time, their relative wages decline by two log-points. For both the measures of outsourcing and the measure of import penetration, we observe quite large increases over time, especially for outsourcing to low-wage countries. 4 Regression Results Table 4 presents results using different measures of internationalization. Models 1-4 include the measure of international outsourcing based on all imports to individual manufacturing industries, i.e., OUTSall DB03,LWOUTSall DB03,andHWOUTSall DB03, where DB03 implies that we use the instrumental variable described in Section 2.2.1 above. It is evident that skill upgrading is more pronounced in firms located in industries with increasing international outsourcing to low-wage countries. Increasing outsourcing to high-wage countries does not appear to affect the relative demand for skilled labor; independent of whether we use instrument or firm specific outsourcing.6 <Table 4 about here> Next we turn to import penetration in Models 5 and 6. In Model 5, it is seen that an overall measure of import penetration does not have a statistically significant impact on the relative demand for skilled labor. However, when calculating the measure separately for HW and LW countries in Model 6, we see that this overall picture hides important information. Namely, import penetration has very 6To allow our study more comparability to the existing literature on the effects of international trade on skill upgrading, we make a short deroute from our micro foundation and perform all regressions using trade intensities instead of our log specification. I.e., instead of e.g. ln (LW OU T Si,t) we include ³LW OUT Si,t Yi´. The results are reported in an appendix (Tables A3-A6). The results are similar to the results presented below, hence, we shall not discuss them further. 13 important effects on skill upgrading, but it is crucial to distinguish between imports from HW and LW countries; firms located in industries exposed to extensive import penetration from low-wage countries increase the relative demand for skilled labor more than other firms, while firms located in industries with high import levels from HW countries lower their relative demand for skilled labor, that is, import penetration from high-wage countries leads to skill downgrading. Overall, this picture suggests that Danish manufacturing firms have comparative advantages in skill-intensive production when compared to low-wage countries and comparative advantages in production processes intensive in unskilled labor when compared to other high-wage countries. Finally, both the measures of international outsourcing and import penetration are included in Models 7 to 10. The main results are again that skill upgrading is affected positively by international outsourcing to low-wage countries and positively by import penetration from low-wage countries, whereas it is affected negatively by import penetration from high-wage countries. Other explanatory variables also affect the relative demand for skilled relative to unskilled labor. It decreases in the relative price of skilled labor, and is more or less unaffected by the size of the firm in terms of its output. The measure of the capital stock does not appear to affect the relative demand for skilled versus unskilled labor. A potential explanation is that time series for physical capital are smooth and therefore it is difficult statistically to distinguish between a time trend and the development in physical capital, see for example Islam (1995).Inallmodels presented in Table 4 the TECH variable capturing technological progress have a significant positive effect on the relative demand for skilled to unskilled labor. The measures of international outsourcing applied in Table 4 may be criticized on the grounds that they include all imports, i.e., imports of e.g. energy and agricultural products are included. To overcome this critique, we re-estimate the models presented in Table 5 using OUTSbroad DB03,DB111 and OUTSnarrow DB03,DB111.Itisevidentthat theconclusionsofTable4arerobusttothechangeinthemeasuresofinternational outsourcing. <Table 5 about here> Next, we turn to the case with three labor types; unskilled, vocational, and academically educated workers. This division is of interest because it is interesting 14 to investigate if skill upgrading takes place between unskilled labor and skilled labor or between unskilled labor and academically educated labor. <Tables 6 about here> The main results are presented in Table 6. For both groups of skilled labor, vocational and academically educated, the effects of international trade are more or less the same as in the two skill groups case. The relative demand for both vocational and academically educated labor is positively affected by import penetration from low-wage countries, whereas import penetration from high-wage countries leads to skill downgrading for both types of skilled labor. Moreover, import penetration from high-wage countries reduces the demand for both types of labor. However, when it comes to international outsourcing, the relative demand for vocationally educated labor is positively affected by outsourcing to low-wage countries, while in general, academically educated labor is not. Finally technological progress is seen to increase the demand for long educated labor in relations to unskilled labor only, since vocational labor is seen to be unaffected. It is evident from Table 6 that higher wages of vocational workers in relation to wages of unskilled workers lead to lower demand for vocational skills. Surprisingly, such a conclusion can not be made for academic workers, since a wage increase for this group in relation to unskilled workers lead to higher relative demand for academic workers. This result can arise because the relative wage for academic workersentersinthewageshareofthegroup. Whentherelativewageratefor academic workers increases, this can result in a higher wage share even when the employment share decreases and can be due to low substitutability between academic and unskilled workers and possibly by a larger degree of inertia in layoffs within this group (due to contractional bindings and that they most likely are harder to replace). The likelihood for this result increases when the time period under investigation is short as in the present study. We have estimated the regressions of Table 6 using employment shares as dependent variable and find that higher relative wages of both vocational and academic education lead to lower demand for the two education lengths, respectively. It is also seen that the effect on skill-upgrading from changing relative cross wages are negative. Taken at face value this suggests that workers with vocational education and workers with academic educations are complements, i.e., when the relative wage for vocational skills increases, the demand for both skill types decreases. It could, however, also be due to correlation between the two relative wages, which 15 could potentially generate the negative correlation between changing relative cross wages and changing wage shares. Our regressions based on the employment share as dependent variable, suggest that this is the case. Demand for vocational skilled workers seems to be mostly unaffected by firm size and capital stock, where as the demand for long educated is decreasing in firm size and as expected increasing in total amount of capital. <Table 7 about here> Again we re-estimate model (5) and (6) from table 6 using OUTSbroad DB03,DB111 and OUTSnarrow DB03,DB111. Table 7 contains the results. Again we conclude that the results of Table 6 are fairly robust to the change in the measures of international outsourcing, even though outsourcing from low wage countries does loose its significance in its explanatory power on the demand for vocational educated workers. 4.1 Empirical Importance of Internationalization In order to assess the empirical importance for skill upgrading of the various explanatory variables, we calculate average contributions to skill upgrading using β1∆ln µwskilled,mun wunskilled,mun ¶/∆Sskilled +β2∆ln (Yi,t)/∆Sskilled +β3∆ln (Ki,t)/∆Sskilled +β4∆ln ¡LWOUTSb h,t¢/∆Sskilled +β5∆ln ¡HWOUTSb h,t¢/∆Sskilled +β6∆ln (LWPENk,t)/∆Sskilled +β7∆ln (HWPENk,t)/∆Sskilled +β8∆ln (TECHDB53,t)/∆Sskilled =1 where the ∆referstothechangeinthemeanofthevariableinquestionfrom1999 to 2002. It is important to emphasize that the contributions determined using this method are ceteris paribus contributions. This method enables us to investigate the importance of the increasing international trade patterns that may contribute to explaining the observed development in the relative demand for skilled labor. <Tables 8 about here> 16 The results are presented in Table 8. It is found that increasing import penetration from low-wage countries on average contributes as much as 40 percent of skill-upgrading. The result suggests that import penetration from low-wage countries may be a more considerable driver for skill-upgrading than any other driver included in the regression. Turning to the same exercise for 3 types of labor, it is found that the ceteris paribus contribution to skill-upgrading from import penetration from low-wage countries is an astonishing 60 percent for academic education, whereas it equals 20 percent for vocational education. The results are presented in Table 9. <Tables 9 about here> 5Conclusion This paper studies the relationship between skill-upgrading and internationalization. The study is based on a unique data set for Danish Manufacturing that enables us to develop measures of international outsourcing and import penetration that focus on where imports originate rather than on their overall level. The main finding suggests that it is of crucial importance to distinguish import by country-of-origin. It is found that international trade from low-wage countries - both international outsourcing and import penetration - lead to skill-upgrading.Moreover,wefind that import penetration from high-wage countries lead to skill-downgrading. To evaluate the importance of internationalization for skill-upgrading, we determine the (ceteris paribus) contributions of each explanatory variable to skillupgrading. Doing this we find that import penetration from low-wage countries accounts around fifty percent of the shift towards skilled labor. This is by far the largest contributor in the analysis and emphasizes the importance of import penetration from low-wage countries for skill upgrading. The main result that import penetration from low-wage countries is of great importance for skill-upgrading; especially, for academic education, are really interesting and important. The importance of import penetration for skill-upgrading has to a large degree been disregarded in the literature for a long time and the focus has mainly been on international outsourcing. In the Journal of Economic Perspective summer 1995 there was a symposium on income inequality and trade. For example, 17 Wood (1995) "..argue[s] for what is still a minority view among economists: that the main cause of the deteriorating situation of unskilled workers in developed countries has been expansion of trade with developing countries." This is not a view that has had any power of penetration. Our result is somewhat related to the view by Wood since we find that the fall in demand for unskilled labor relative to skilled labor is greatly influenced by increasing import penetration from developing countries. 18 References [1] Anderton, B., and P. Bretton (1999): “Outsourcing and Low-Skilled Workers in the UK”, Bulletin of Economic Research,51(4),267-285 [2] Autor, D.H., L.F. Katz, and A.B. Krueger (1998): “Computing Inequality: Have Computers Changed the Labor Market?” Quarterly Journal of Economics, 113 (4), 1169-1213. [3] Berman, E., J. Bound and Z. Griliches (1994): “Changes in the Demand for Skilled Labor within U.S. Manufacturing: Evidence from the Annual Survey of Manufactures,” Quarterly Journal of Economics,109(2), 367-398. [4] Berman, E., J. Bound, and S. Machin (1998): “Implications of Skill-Biased Technical Changes: International Evidence,” Quarterly Journal of Economics, 113(4), 1245-1280. [5] Bernard, A.B., J.B. Jensen, and P.K. Schoot (2006): “Survival of the Best Fit: Exposure to Low-Wage Countries,” Journal of International Economics [6] Berndt,E.R.(1991):The Practice of Econometrics Classic and Contemporary, Addison-Wesley Publishing Company [7] Brown, R. S., and L. R. Christensen, (1981): Estimates of elasticities of substitution in a model of partial static equilibrium: An application to US agriculture, 1947-1974,inE.R.BerndtandB.C.Field(eds.):Modellingandmeasuring natural resource substitution, 209-229, MIT Press. [8] Christensen, L. R. and W. H. Greene (1976): “Economics of scale in U.S. Electric Power Generation, The Journal of Political Economy, 84, 655-676. [9] Edin, P-A, P Fredriksson, and O Åslund (2003): “Ethnic Enclaves and the Economic Success of Immigrants: Evidence from a Natural Experiment”, Quarterly Journal of Economics, 118(1),329-357 [10] Ekholm, K., and K. Hakkala (2006),“TheEffect of Offshoring on Labor Demand: Evidence from Sweden”, CEPR working paper #5648 [11] Feenstra, R.C. (2004),Advanced International Trade: Theory and Evidence, Princeton University Press [12] Feenstra, R.C. and G.D. Hanson (1996), “Globalization, Outsourcing, and Wage Inequality”, American Economic Review, 86(2), Papers and Proceedings, 240-245 19 [13] Feenstra, R.C. and G.D. 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 [14] Feenstra, R.C. and G.D. Hanson (2003),“Global Production Sharing and Rising Inequality: A Survey of Trade and Wages,” in Kwan Choi and James Harrigan (eds.): “Handbook of International Trade,” Basil Blackwell. [15] Geishecker, I. (2006), “Does Outsourcing to Central and Eastern Europe Really Threaten Manual Workers’ Jobs in Germany?”, The World Economy,29(5), 559-583 [16] Görg, H., A. Hanley, and E. Strobl (2007), “Productivity effects of international outsourcing: Evidence from plant level data” , memeo [17] Hijzen, A., H. Görg and R.C. Hine (2005), “International Outsourcing and the Skill Structure of Labor Demand in the United Kingdom”, Economic Journal, 115, 506, 860-878 [18] Islam, N. (1995) growth empirics: a panel data approach, Quarterly Journal of Economics, 443, 1127—1170. [19] Katz, L.F., and K.M. Murphy (1992): “Changes in Relative Wages, 1963-1987: Supply and Demand Factors,” Quarterly Journal of Economics,107(1),35-78. [20] Kmenta, J (1967): “On Estimation of the CES Production Function”, International Economic Review 8, 180-189. [21] Machin, S., and J. Van Reenen (1998): “Technology and Changes in Skill Structure: Evidence from Seven OECD Countries,” Quarterly Journal of Economics, 113(4), 1215-1244. [22] Ng, F., and A.J. Yeats (1999), “Production Sharing in East Asia — Who Does What for Whom and Why?”, Policy Research Working Paper 2197, World Bank [23] Statistics Denmark (1986), “Commodity Flow Systems and Construction of Input-Output Tables in Denmark”, Arbejdspapir nr. 15 Statistics Denmark (2000): “IDA Database,” Copenhagen. [24] Statistics Denmark (2003): “FIDA Database,” Copenhagen. [25] Wood, A. (1995), “How Trade Hurt Unskilled Workers", Journal of Economic Perspectives, 9(3),57-80 20 A Appendix 1: The Model in the general case of H-types of workers We analyze the relationship between globalization and the firm’s relative demand for different types of labor within a factor demand framework. The estimated system of equations is derived from a simple quasi-fixed translog cost function (Christensen and Greene 1976,BrownandChristensen1981,Berndt1991). The starting point is the cost function C(Y,p)denoting the minimum cost of producing Yat given prices pevaluated at the relevant cost-minimizing choice of inputs. By expanding the log of the cost function ln C(Yit,p iht)in a second order Taylor series about the point piht =0,Y it =0,wherepiht denote input prices for type hlabor, and Yit the output in firm iat time t,weobtain ln C(Yit,p iht)=β0+ H X h=1 ∂ln Cit ∂ln piht ln piht +∂ln Cit ∂ln Yit ln Yit +1 2 H X h=1 H X m=1 ∂2ln Cit ∂ln piht∂ln pimt ln piht ln pimt +1 2 ∂2ln Cit ∂ln Yit∂ln Yit (ln Yit)2+1 2 H X h=1 ∂2ln Cit ∂ln Yit∂ln piht ln Yit ln piht This cost function may contain both variable and quasi-fixed inputs in the vector piht. In our case the variable inputs are different types of workers (by education groups), with cost given by the wages PH h=1 wh. Quasi-fixed inputs can be seen as inputs that are not used optimally (in a long run sense) in the short run, but are important for the production process. First, we assume the amount of physical capital Kto be quasi-fixed, implying that firms cannot adjust capital in the short run. Second, our hypothesis is that the production process in terms of the labor composition might be influenced by international trade, especially, from low wage countries. Therefore, international outsourcing OUTS and import penetration PEN are included as quasi-fixed inputs along with measures of technological progress TECH. Further identifying the derivatives as our parameters of interest the cost function can be expressed as: ln Cit =β0+PH h=1 αw hln wiht +αYln Yit +αKln Kit +αoln OUTSit +αoln PENit +αoln TECHit +1 2PH h=1 PH m=1 βw hm ln wiht ln wimt +PH h=1 βY hln wiht ln Yit +PH h=1 βK hln wiht ln Kit +PH h=1 βo hln wiht ln oit +PH h=1 βo hln wiht ln PENit +PH h=1 βo hln wiht ln TECHit (1) 21 Table 5: Change in Skilled Wage Share as Dependent Variable, 1999-2002, Fixed Effects Estimation Two Education Groups, Outsourcing Measures narrow DBDB OUTS 111,03 and broad DBDB OUTS 111,03 broad DBDB OUTS 111,03 =Ω broad DBDB OUTS 111,03 =Ω narrow DBDB OUTS 111,03 =Ω narrow DBDB OUTS 111,03 =Ω (11) (12) (13) (14) Relative wagemun, skilled -0,0464 -0,0204 -0,0465 -0,0234 (3.23)** (1,40) (3.23)** (1,61) ln(Yi) -0,0029 -0,0048 -0,0029 -0,0049 (1,48) (2.45)* (1,47) (2.51)* ln(Ki) -0,0011 -0,0014 -0,0011 -0,0013 (0,68) (0,89) (0,69) (0,82) ln(Ω) 0,0017 0,0015 (2.56)* (2.44)* ln(LWΩ) 0,0009 0,0006 (1,89) (1,58) ln(HWΩ) 0,0000 0,0003 (0,03) (0,56) ln(PENDB111) -0,0017 -0,0017 (1,79) (1,77) ln(LWPENDB111) 0,0168 0,0171 (10.12)** (10.36)** ln(HWPENDB111) -0,0181 -0,0184 (10.05)** (10.26)** ln(TECH DB53) 0,0054 0,0030 0,0053 0,0030 (5.15)** (2.83)** (5.04)** (2.82)** Observations 20184 20184 20184 20294 Number of Firms 6612 6612 6612 6649 Note: * significant at 5 % level; ** significant at 1 % level Table 6: Change in Skilled and Educated Wage Share as Dependent Variable, 1999-2002, Fixed Effects Estimation, Three Education Groups, Outsourcing Measures all OUTS and all DB OUTS 03 (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) ------------------- Wage Share, Vocational Education ------------------- Relative wagemun, Vocational skills -0,0161 -0,0167 -0,0166 -0,0169 -0,0178 -0,0086 -0,0159 -0,0163 -0,0077 -0,0082 (1,15) (1,19) (1,18) (1,20) (1,27) (0,61) (1,13) (1,17) (0,55) (0,58) Relative wagemun, Long education -0,0265 -0,0264 -0,0265 -0,0262 -0,0263 -0,0234 -0,0264 -0,0262 -0,0235 -0,0233 (4.40)** (4.37)** (4.39)** (4.35)** (4.36)** (3.89)** (4.38)** (4.35)** (3.92)** (3.87)** ln(Yi) 0,0037 0,0034 0,0038 0,0033 0,0043 0,0029 0,0037 0,0034 0,0025 0,0021 (1,76) (1,61) (1,84) (1,55) (2.07)* (1,41) (1,75) (1,59) (1,20) (0,98) ln(Ki) -0,0014 -0,0012 -0,0013 -0,0013 -0,0012 -0,0018 -0,0014 -0,0012 -0,0018 -0,0019 (0,68) (0,58) (0,63) (0,65) (0,62) (0,92) (0,69) (0,59) (0,89) (0,93) ln(OUTSDB03) 0,0023 0,0023 (3.80)** (3.66)** ln(OUTS) -0,0002 -0,0262 (0,57) (4.35)** ln(LWOUTSDB03) 0,0012 0,0008 (2.68)** (1,68) ln(HWOUTSDB03) 0,0006 0,0008 (0,81) (1,19) ln(LWOUTS) -0,0004 -0,0004 (1,40) (1,31) ln(HWOUTS) 0,0013 0,0015 (0,33) (0,40) ln(PENDB111) 0,0014 0,0010 0,0014 (1,51) (1,16) (1,51) ln(LWPENDB111) 0,0102 0,0099 0,0097 (6.76)** (6.47)** (6.38)** ln(HWPENDB111) -0,0090 -0,0089 -0,0086 (5.30)** (5.22)** (4.99)** ln(TECH DB53) 0,0016 0,0014 0,0017 0,0013 0,0015 0,0002 0,0015 0,0013 0,0004 0,0000 (1,69) (1,44) (1,82) (1,36) (1,58) (0,22) (1,59) (1,30) (0,37) (0,03) Table 6: Change in Skilled and Educated Wage Share as Dependent Variable, 1999-2002 - (continued) ------------------- Wage Share, Academic Education ------------------- Relative wagemun, Vocational skills -0,0265 -0,0264 -0,0265 -0,0262 -0,0263 -0,0234 -0,0264 -0,0262 -0,0235 -0,0233 (4.40)** (4.37)** (4.39)** (4.35)** (4.36)** (3.89)** (4.38)** (4.35)** (3.92)** (3.87)** Relative wagemun, Long education 0,0308 0,0308 0,0306 0,0307 0,0309 0,0317 0,0309 0,0309 0,0316 0,0315 (5.76)** (5.75)** (5.73)** (5.73)** (5.78)** (5.96)** (5.78)** (5.77)** (5.95)** (5.93)** ln(Yi) -0,0173 -0,0172 -0,0174 -0,0173 -0,0174 -0,0191 -0,0174 -0,0173 -0,0191 -0,0188 (10.30)** (10.09)** (10.38)** (10.14)** (10.40)** (11.43)** (10.31)** (10.11)** (11.36)** (11.07)** ln(Ki) 0,0105 0,0104 0,0101 0,0103 0,0104 0,0097 0,0105 0,0104 0,0095 0,0096 (6.49)** (6.44)** (6.25)** (6.37)** (6.42)** (6.03)** (6.48)** (6.43)** (5.92)** (5.98)** ln(OUTSDB03) -0,0003 -0,0004 (0,61) (0,77) ln(OUTS) -0,0002 -0,0002 (0,57) (0,61) ln(LWOUTSDB03) 0,0012 0,0006 (3.17)** (1,76) ln(HWOUTSDB03) -0,0009 -0,0006 (1,59) (1,09) ln(LWOUTS) 0,0003 0,0003 (1,36) (1,14) ln(HWOUTS) -0,0004 -0,0004 (1,40) (1,31) ln(PENDB111) 0,0012 0,0011 0,0012 (1,61) (1,56) (1,63) ln(LWPENDB111) 0,0116 0,0112 0,0115 (9.62)** (9.21)** (9.41)** ln(HWPENDB111) -0,0104 -0,0101 -0,0102 (7.60)** (7.37)** (7.44)** ln(TECH DB53) 0,0050 0,0051 0,0050 0,0049 0,0050 0,0034 0,0049 0,0050 0,0034 0,0034 (6.42)** (6.61)** (6.44)** (6.37)** (6.43)** (4.35)** (6.26)** (6.43)** (4.25)** (4.31)** Observations 5651 5651 5651 5651 5651 5651 5651 5651 5651 5651 Note: * significant at 5 % level; ** significant at 1 % level Table 7: Change in Skilled and Educated Wage Share as Dependent Variable, 1999-2002, Fixed Effects Estimation, Three Education Groups, Outsourcing Measures broad DBDB OUTS 111,03 and narrow DBDB OUTS 111,03 br o DB OUTS=Ω broad DBDB OUTS 111,03 =Ω narrow DBDB OUTS 111,03 =Ω narrow DBDB OUTS 111,03 =Ω (11) (12) (13) (14) ------------------- Wage Share, Vocational Education ------------------ - Relative wagemun, Vocational skills -0,0161 -0,0076 -0,0181 -0,0078 (1,15) (0,54) (1,29) (0,55) Relative wagemun, Long education -0,0264 -0,0234 -0,0260 -0,0234 (4.38)** (3.88)** (4.31)** (3.88)** ln(Yi,) 0,0038 0,0025 0,0043 0,0029 (1,80) (1,20) (2.07)* (1,40) ln(Ki,) -0,0014 -0,0019 -0,0012 -0,0019 (0,68) (0,96) (0,62) (0,97) ln(Ω) 0,0020 0,0000 (3.38)** (0,01) ln(LWΩ) 0,0004 0,0005 (0,95) (1,27) ln(HWΩ) 0,0013 -0,0006 (2.05)* (0,97) ln(PENDB111) 0,0011 0,0014 (1,24) (1,50) ln(LWPENDB111) 0,0099 0,0103 (6.46)** (6.73)** ln(HWPENDB111) -0,0089 -0,0091 (5.20)** (5.33)** ln(TECH DB53) 0,0015 0,0003 0,0015 0,0003 (1,59) (0,30) (1,56) (0,34) ------------------- Wage Share, Academic Education ------------------- Relative wagemun, Vocational skills -0,0264 -0,0234 -0,0260 -0,0234 (4.38)** (3.88)** (4.31)** (3.88)** Relative wagemun, Long education 0,0310 0,0316 0,0311 0,0317 (5.79)** (5.95)** (5.81)** (5.97)** ln(Yi,) -0,0173 -0,0190 -0,0178 -0,0195 (10.30)** (11.33)** (10.59)** (11.63)** ln(Ki,) 0,0105 0,0096 0,0103 0,0096 (6.47)** (6.00)** (6.40)** (5.98)** ln(Ω) -0,0005 0,0013 (1,03) (2.71)** ln(LWΩ) 0,0007 0,0008 (2.10)* (2.55)* ln(HWΩ) -0,0010 0,0004 (1.97)* (0,92) ln(PENDB111) 0,0012 0,0010 (1,58) (1,34) ln(LWPENDB111) 0,0112 0,0110 (9.11)** (8.99)** ln(HWPENDB111) -0,0100 -0,0100 (7.29)** (7.31)** ln(TECH DB53) 0,0049 0,0035 0,0048 0,0035 (6.30)** (4.39)** (6.23)** (4.45)** Observations 5651 5651 5651 5651 Note: * significant at 5 % level; ** significant at 1 % level Table 8: (Ceteris Paribus) Contributions of Explanatory Variables to Skill-Upgrading – Two Category Classification of Skills (1) (3) (5) (6) (7) (9) (11) (12) (13) (14) Relative wagemun, skilled 4% 4% 4% 2% 4% 2% 4% 2% 4% 2% ln(Yi,) -2% -3% -2% -4% -3% -4% -3% -4% -2% -4% ln(Ki,) -1% -1% -1% -1% -1% -1% -1% -1% -1% -1% ln(OUTSDB03) 2% 2% 2% 2% ln(LWOUTSDB03) 6% 3% 3% 1% ln(HWOUTSDB03) 0% 0% 0% 0% ln(PENDB111) -1% -1% -1% -1% ln(LWPENDB111) 40% 39% 39% 39% ln(HWPENDB111) -6% -6% -6% -6% ln(TECH DB53) 10% 10% 11% 6% 11% 6% 11% 6% 11% 6% Table 9: (Ceteris Paribus) Contributions of Explanatory Variables to Skill-Upgrading – Three Category Classification of Skills (1) (3) (5) (6) (7) (9) (11) (12) (13) (14) ------------------- Wage Share, Vocational Education ------------------- Relative wagemun, Vocational skills 1,4% 1% 2% 1% 1% 1% 1% 1% 2% 1% Relative wagemun, Long education 1,5% 2% 2% 1% 2% 1% 2% 1% 1% 1% ln(Yi,) 3% 4% 4% 3% 3% 2% 4% 2% 4% 3% ln(Ki,) -1% -1% -1% -2% -1% -2% -1% -2% -1% -2% ln(OUTSDB03) 2% 2% 2% 0% ln(LWOUTSDB03) 3% 2% 1% 1% ln(HWOUTSDB03) 0% 1% 1% -1% ln(PENDB111) 1% 0% 0% 1% ln(LWPENDB111) 21% 21% 21% 22% ln(HWPENDB111) -1% -1% -1% -1% ln(TECH DB53) 3% 4% 3% 0% 3% 1% 3% 1% 3% 1% ------------------- Wage Share, Academic Education ------------------- Relative wagemun, Vocational skills 5% 5% 5% 5% 5% 5% 5% 5% 5% 5% Relative wagemun, Long education -4% -4% -4% -4% -4% -4% -4% -4% -4% -4% ln(Yi,) -39% -40% -40% -43% -39% -43% -39% -43% -40% -44% ln(Ki,) 26% 25% 25% 24% 26% 23% 26% 24% 25% 23% ln(OUTSDB03) -1% -1% -1% 4% ln(LWOUTSDB03) 6% 3% 3% 2% ln(HWOUTSDB03) -2% -1% -2% 1% ln(PENDB111) 1% 1% 1% 1% ln(LWPENDB111) 58% 56% 56% 55% ln(HWPENDB111) -4% -4% -4% -4% ln(TECH DB53) 24% 24% 24% 16% 23% 16% 23% 17% 23% 17% Table A1: Share of Firms with at Least one Employee of a Given Educational Length 1999 2000 2001 2002 -------------------- All Manufacturing Firms -------------------- No Education 84.8% 83.8% 83.1% 82.5% Vocational Education 86.2% 86.3% 86.5% 86.5% Short Academic Education 30.6% 31.2% 31.9% 32.3% Medium Academic Education 30.1% 30.8% 31.2% 31.4% Long Academic Education 18.3% 18.9% 19.4% 20.2% ----- Manufacturing Firms With 10 or More Employees ----- No Education 99.5% 99.4% 99.3% 99.1% All Educations 99.9% 99.9% 99.9% 99.9% ----- Manufacturing Firms With 50 or More Employees ----- No Education 100.0% 100.0% 100.0% 100.0% Vocational or Short Academic Education 100.0% 100.0% 100.0% 100.0% Medium or Long Academic Education 95.8% 96.0% 96.3% 96.6% Table A2: Number of Firms, Aggregate Turnover and Capital Stock in Danish Manufacturing 1999 2000 2001 2002 -------------------- All Manufacturing Firms -------------------- Number of Firms 13,341 13,132 12,788 12,379 Turnover in Billions D.Kr. 465 518 552 550 Capital in Billions D.Kr. 434 478 488 496 ----- Manufacturing Firms with 10 or More employees ----- Number of Firms 5,205 5,146 5,019 4,814 Turnover in Billions D.Kr. 436 488 521 521 Capital in Billions D.Kr. 414 457 468 475 ----- Manufacturing Firms with 50 or More employees ----- Number of Firms 1,420 1,460 1,416 1,355 Turnover in Billions D.Kr. 362 413 444 444 Capital in Billions D.Kr. 363 405 416 422 Table A3: Using Intensities: Change in Skilled Wage Share as Dependent Variable, 1999-2002, Fixed Effects Estimation Two Education Groups, Outsourcing Measures all OUTS and all DB OUTS 03 (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Relative wagemun, skilled -0,0495 -0,0500 -0,0483 -0,0505 -0,0506 -0,0484 -0,0500 -0,0506 -0,0475 -0,0489 (3.45)** (3.49)** (3.37)** (3.52)** (3.53)** (3.38)** (3.49)** (3.53)** (3.31)** (3.41)** ln(Yi,) -0,0024 -0,0023 -0,0024 -0,0025 -0,0024 -0,0024 -0,0025 -0,0024 -0,0025 -0,0026 (1,23) (1,18) (1,23) (1,29) (1,23) (1,24) (1,27) (1,20) (1,28) (1,35) ln(Ki) -0,0012 -0,0012 -0,0012 -0,0010 -0,0011 -0,0012 -0,0012 -0,0012 -0,0013 -0,0011 (0,74) (0,72) (0,76) (0,64) (0,71) (0,75) (0,74) (0,71) (0,79) (0,67) OUTSDB03/ YDB03 0,0119 0,0164 (1,24) (1,69) OUTS/ Y 0,0002 0,0003 (0,06) (0,09) LWOUTSDB03/ YDB03 0,0887 0,0758 (3.20)** (2.58)** HWOUTSDB03/ YDB03 -0,0037 0,0000 (0,34) 0,00 LWOUTS/ Y 0,0140 0,0134 (2.39)* (2.30)* HWOUTS/ Y -0,0092 -0,0094 (1,93) (1.97)* PENDB111/ YDB111 -0,0780 -0,0891 -0,0780 (2.23)* (2.50)* (2.23)* LWPENDB111/ YDB111 0,5231 0,3803 0,5142 (3.24)** (2.23)* (3.19)** HWPENDB111/ YDB111 -0,2164 -0,2048 -0,2155 (4.30)** (4.02)** (4.28)** TECHDB53/ YDB53 0,2835 0,2884 0,2775 0,2867 0,2863 0,2723 0,2791 0,2861 0,2641 0,2708 (5.07)** (5.17)** (4.96)** (5.14)** (5.13)** (4.87)** (4.99)** (5.13)** (4.72)** (4.85)** Observations 20184 20184 20184 20184 20184 20184 20184 20184 20184 20184 Number of Firms 6612 6612 6612 6612 6612 6612 6612 6612 6612 6612 Note: * significant at 5 % level; ** significant at 1 % level Table A4: Using Intensities: Change in Skilled Wage Share as Dependent Variable, 1999-2002, Fixed Effects Estimation Two Education Groups, Outsourcing Measures broad DBDB OUTS 111,03 and narrow DBDB OUTS 111,03 broad DBDB OUTS 111,03 =Ω broad DBDB OUTS 111,03 =Ω narrow DBDB OUTS 111,03 =Ω narrow DBDB OUTS 111,03 =Ω (11) (12) (13) (14) Relative wagemun, skilled -0,0502 -0,0474 -0,0505 -0,0472 (3.50)** (3.31)** (3.53)** (3.29)** ln(Yi,) -0,0025 -0,0025 -0,0024 -0,0025 (1,27) (1,28) (1,25) (1,30) ln(Ki) -0,0012 -0,0013 -0,0012 -0,0012 (0,73) (0,79) (0,71) (0,77) Ω/ YDB03 0,0162 0,0084 (1,60) (0,77) LWΩ/ YDB03 0,0978 0,0920 (3.07)** (2.92)** HWΩ/ YDB03 -0,0014 -0,0052 (0,13) (0,42) PENDB111/ YDB111 -0,0889 -0,0792 (2.49)* (2.26)* LWPENDB111/ YDB111 0,3515 0,4883 (2.06)* (3.01)** HWPENDB111/ YDB111 -0,2035 -0,2093 (4.00)** (4.13)** TECHDB53/ YDB53 0,2798 0,2649 0,2843 0,2646 (5.00)** (4.73)** (5.09)** (4.73)** Observations 20184 20184 20184 20184 Number of Firms 6612 6612 6612 6612 Note: * significant at 5 % level; ** significant at 1 % level