Policy Discussion Paper

Transcription

1 Policy Discussion Paper No. 97/11 University of Adelaide Adelaide SA 5005 Australia LABOUR STANDARDS AND TRADE FLOWS OF OECD COUNTRIES Cees van Beers October 1997

2 CIES POLICY DISCUSSION PAPER 97/11 LABOUR STANDARDS AND TRADE FLOWS OF OECD COUNTRIES Cees van Beers Department of Economics Leiden University P.O. Box RA Leiden The Netherlands ph: +31 (0) fax: +31 (0) C.vanbeers@Law.Leidenuniv.nl October 1997 The author is grateful to the Netherlands Organization for Scientific Research who provided financial support. ii

3 SUMMARY This study is concerned with the impact of differences in labour standards among OECD countries on intra-oecd trade flows. In theory labour costs in relatively high-labour standards' countries are relatively high and therefore competitiveness of labour-intensive industries will be relatively low. In this study an index representing stringency of labour standards in OECD-countries based on judgements of actual labour regulations has been used in a trade flow equation to investigate the effect of labour standards' strictness on foreign trade flows. It is argued that differences in labour standards among OECD-countries affect both labour- and capital-intensive trade flows if these are produced with relatively more highskilled labour. The reason is the higher demand elasticity of high-skilled labour compared with low-skilled labour. Higher labour standards result in higher labour costs in industries that use relatively more high-skilled labour. Labour-intensive exports are negatively affected if they are produced with high-skilled labour while no effect can be found if produced with low-skilled labour. Capital-intensive exports produced with high-skilled labour are also negatively affected while that part of these exports produced with low-skilled labour is positively affected. The relatively low demand elasticity of low-skilled labour combined with high substitution possibilities of capital for lowskilled labour provide an explanation for this result. No significant effect on imports can be found. iii

4 LABOUR STANDARDS AND TRADE FLOWS OF OECD COUNTRIES Cees van Beers 1. INTRODUCTION The question whether labour standards should be uniform over the world is a matter of intense dispute. Unlike the discussion on environmental standards, the issue of uniform labour standards is not a new one. Already in the nineteenth century concerns were expressed by industries that improving working conditions by means of abolishing child labour and limiting the working week would put them at a competitive disadvantage in comparison with foreign industries (Hoekman and Kostecki, 1995: 263). In the United States the first "Fair Labour Standards Act" proposals appeared in the 1930s (Charnovitz, 1987: 568). During the Uruguay Round the issue of labour standards was unsuccessfully introduced by the United States and France. At the Singapore WTO-conference of December 1996 it was decided that the issue of labour standards remained a matter of the International Labour Organization. In future trade talks the issue of labour standards will be at the international trade agenda as it remains a means for high-standard nations to try to maintain a protectionistic policy against imports from lowstandard countries (Anderson, 1995). The central issue of the discussion on the relationship between labour standards and international trade consists of two parts. First, whether countries with high labour standards experience a loss of competitiveness in the production and exports of labour-intensive products as compared with low labour standards countries. This becomes manifest in either less production of labour-intensive goods or migration of labour-intensive industries to lower standard nations. Both phenomena will be revealed in a relatively low (high) level of labourintensive exports in the high-standard (low-standard) country. Second, whether import barriers 1

5 should be introduced against products imported from countries with low labour standards with the aim to avoid a 'race to the bottom', i.e., a high standard country lowers it's standards to remain or improve it's competitiveness. The present study is concerned with the first issue and consists of an empirical test of the hypothesis whether OECD-countries with high labour standards experience a lower level of labour-intensive exports than OECD-countries with low labour standards. The two issues referred to above are mostly analyzed within the context of North-South trade flows. This focus overlooks the importance of the issue among developed countries, i.e., countries with similar political systems (see also Swinnerton and Schoepfle, 1994: 70). Particularly the increasing regionalization of world trade leads to increasing fears of displacement of production from high-standard to low-standard members of regional integration agreements. For example, the already achieved (Greece, Portugal and Spain) and expected (Czech Republic, Hungary and Poland) increase of members of the European Union leads to a greater number of members that are relatively low-income countries. Inter-industry trade within such an economic integration scheme will rise and therewith the emergence of greater adjustment problems than in the past (Sapir, 1995: 799). The issue of harmonization of labour standards among developed countries will therefore remain a matter of importance in the near future. The second reason for an analysis within the framework of OECD countries is the availability of data on labour standards. Although still limited, the necessary data for constructing indices that represent strictness of a country's labour standards are available for the majority of OECD countries while for non-oecd nations no reliable data are present at all. A relatively low level of labour-intensive exports of countries experiencing relatively high labour standards may be expected. Labour costs are a substantial part of production costs of labour-intensive commodities. Relatively high labour standards increase labour costs and 2

6 therewith production costs which results in a loss of competitiveness. At the import side it can be expected that high labour standards countries experience a relatively high level of imports of labour-intensive commodities. The costs of domestic production is high relative to costs of production abroad and therefore domestic consumers will substitute foreign products for domestic ones. In other words, high labour standards generate relatively less (more) labourintensive exports (imports). The next section presents a short discussion on theoretical and empirical work that has been performed so far. Section 3 discusses a methodology to measure (differences in) labour standards. In section 4 a distinction of trade flows according to factor- and skill-intensity is presented. Section 5 presents a trade flow model extended with labour standards indices. The estimation results of this model are presented in section 6. The last section shows the conclusions. 2. PREVIOUS STUDIES ON THE RELATIONSHIP BETWEEN TRADE AND LABOUR STANDARDS Most of the discussion on the relationship between trade and labour standards is of a policyoriented or theoretical nature (see also Sapir, 1995; Krueger, 1997). Recently theoretical work has been done by Brown, Deardorff and Stern (1996) who attempted to analyze the effects of labour standard differences on foreign trade patterns in several standard trade models. An important finding of their investigations is that harmonization of labour standards is not in the interest of high-labour standard industrialized countries. They assume that labour standards are primarily labour using and thus distract from labour that would otherwise be available. In the countries where they are introduced, labour standards make labour more scarce. In a Heckscher- Ohlin framework the implementation of higher labour standards leads to a decrease of a 3

7 country's labour endowment (Brown, Deardorff and Stern, 1996: 254). Labour costs and therefore the price of the labour-intensive commodity rises and thus in a two-countries, twofactors, two-commodities world an improvement of the terms-of-trade of the country that is relatively well-endowed with labour is the result. This improves (deteriorates) comparative advantage of the labour (capital) rich country and leads to the conclusion that welfare of capital rich countries decreases if higher labour standards are imposed on countries that are relatively well-endowed with labour. The decline of consumers' welfare in capital-rich countries is larger than welfare increase of producers and workers in labour-intensive industries in these countries. In policy discussions, however, it is not national welfare impacts that are central but the interests of certain groups like labour unions or producers (Srinivasan, 1994). This gives rise to a generally wrong belief that capital-rich countries' welfare would favour from harmonization of labour standards because their import-competing industries would meet less competition from industries in lower-standard countries. Empirical work on the relationship between trade and labour standards is very scarce. A possible reason for the little number of empirical studies is the lack of quantitative data on labour standards. Especially when developing countries are involved this is a severe problem. A recent study has been performed by Rodrik (1996) and concentrates on North-South trade relations. He regressed the ratio of textile and clothing exports to other exports as a proxy for labour-intensive commodity trade on a country's labour/land ratio (proxy for labour endowment of a country relative to its land endowment) and average years of schooling in the population aged over 25 (proxy for human capital) and on different combinations of labour standards indicators. The several proxies for labour standards used in Rodrik (1996) are (1) total number of ILO conventions ratified by a country (TOTCONV), (2) number of ILO conventions ratified by a country among six conventions relating to "basic worker rights"(bwrconv), 1 (3) a 1 These contained the conventions 29 (Forced Labour), 87 (Freedom of Association and Protection of the Right to 4

8 synthetic indicator combining indicators of civil liberties and political rights (DEMOC), (4) an indicator for child labour (CHILD), (5) statutory hours of work in a normal working week in manufacturing or construction (HOURS), (6) days of annual leave with pay in manufacturing (LEAVE), (7) percentage of the labour force that is unionized (UNION). Insignificant results for all the labour standard indicators are reported except for the variable "HOURS". A positive effect is reported which means that longer statutory hours go together with a stronger comparative advantage in textiles and clothing (Rodrik, 1996: 20). 3. HOW TO MEASURE LABOUR STANDARDS? The intensity with which labour regulations affect labour costs and therefore competitiveness of industries and countries depends on (1) the relative strictness as revealed by the labour standard indices reported in Table 1, and (2) the importance of labour costs as part of total production costs. In the present section attention is paid to the first point. The second factor will be discussed in the next section. A quantifiable measure of strictness of labour standards is very hard to define. A problem in defining labour standards by means of the seven variables presented above is that they probably produce a lot of noise. An example is the number of ILO-conventions ratified. A country that has ratified quite a number of ILO-conventions intends to put much effort in establishing a relatively high level of labour standards but it does not contain information on how such an intention is translated in actual labour regulations. Therefore information on actual labour Organize), 98 (Right to Organize and Collective Bargaining), 105 (Abolition of Forced Labour), 111 (Discrimination) and 138 (Minimum Age). 5

9 regulations is necessary. 2 Civil liberties and political rights are a condition for but do not necessarily cause higher labour standards. The percentage of the labour force that is unionized depends strongly on what is considered as union membership which can differ strongly among countries for other reasons than differences in labour standards. 2 Van Beers and van den Bergh (1997) have argued that in case of environmental standards a distinction needs to be made in input-oriented and output-oriented standards. Input-oriented standards are connected to input efforts of governments while output-oriented standards use the actual results or a proxy for the results of stringent regulations. 6

10 Working Time TABLE 1. Labour standard indices: 18 OECD countries Fixed- Term Contract Employment Protection Minimum Wages Employees Representation Rights Total Austria Belgium-L Canada Denmark Finland France Germany Greece Ireland Italy Netherlands Norway Portugal Spain Sweden Switzerland United Kingdom United States Notes: 0 = government regulations are either light or non-existent 1 = government regulations are intermediate 2 = government regulations are strict Source: OECD Employment Outlook (1994) The concrete results of labour regulations are a better representation of the actual costs of a relatively high labour standard to be borne by a producer than measures based on input efforts of governments. In case of the seven proxies mentioned in the previous section, HOURS and LEAVE can be considered as actual measurable regulations. In OECD Employment Outlook (1994: 152) an attempt has been made to construct a labour standard indicator based on actual labour regulations in eighteen OECD-countries. The indicator 7

11 is a synthetic index which is calculated by adding together partial indices that measure the relative stringency of government regulations on working time, employment contracts, minimum wages and workers' representation rights. In Table 1 these indices are reported, added and ranked. Although the indices involve a measure of judgement, their advantage as compared with a number of Rodrik's (1996) standards is that they represent the actual results of labour regulations and therefore provide a better indication of strictness of labour standards. The main element behind the judgement is whether labour standards are determined by government regulations or by more decentralised systems such as collective agreements or individual contracts. In the latter case labour standards are less strict as there is room for differences among industries. Working time as reported in Table 1 consists of normal and maximum number of hours per working day or week; overtime provisions; rest periods. The table reveals that Spain, Greece and Ireland are judged strict with regard to working-time arrangements. This cannot be explained by the contractual weekly working hours but by the existence of government regulations by law (and not collective agreements) so that less exceptions based on individual cases are allowed. The low score of Denmark, United Kingdom and the United States can be attributed to the absence of legal limits on working time (OECD, 1994: 143). The existence of fixed-term contracts in contrast to traditional employment protection legislation point at less strict labour standards. Employment protection regulations include legislation and rules with respect to individual and collective dismissal of workers. These are stringent in Greece, Ireland, Italy and Spain. The establishment of a statutory minimum wage, and the extension of minimum wage clauses concluded in collective agreements to enterprises that have not been a signatory party points at relatively stringent labour standards. These can be found in France, Greece, Italy and Spain. Employment representation rights are manifested by the presence of union delegates 8

12 or workers councils and their presence indicates higher labour standards. It is interesting to observe that especially South European countries' standards that are rather stringent on workingtime, employment protection and minimum wages are rather loose with respect to workers representation rights. A glance at the ranking of the total score in Table 1 shows that countries that reveal relatively stringent labour standards are France, Germany, Greece, Italy, Spain and Sweden. The stringency of labour standards in the South European countries can be explained by centralised arranged employment protection and minimum wages while in case of Germany and Sweden employees representation rights are the cause of relatively high labour standards. Both the United Kingdom and the United States show low scores on all indices reported. 4. FACTOR-INTENSITY AND SKILL-INTENSITY OF TRADE FLOWS Labour costs are more (less) important in total production costs if the production process is relatively labour-intensive (capital-intensive). Thus it can be expected that exports of labourintensive commodities will be relatively low in countries with relatively high labour standards while their exports of capital-intensive commodities will be relatively high. The trade flows to be analyzed in the next section are distinguished in trade flows produced by relatively labourintensive and those produced by relatively capital-intensive production techniques. The distinction between labour-intensive and capital-intensive export commodities has been based on the United Nations Standard International Trade Classification (SITC) and has been published by UNIDO (1981). The classification is reported in Appendix A. Besides labour-intensity of the production process also skills of labour are important for the ultimate effect of labour standards on exports especially in a sample of developed countries. Empirical work suggests that the demand for high-skilled labour is less elastic than for lowskilled labour for reasons of employers investing in education and skills of their workers (see 9

13 Filler, Hammermesh and Rees, 1996: 168). Therefore an increase of labour costs as a result of more stringent labour standards reduces demand for high-skilled labour less than for low-skilled labour. As a result total labour costs in industries that make use of relatively much high-skilled labour will rise more than in industries that mainly produce with low-skilled workers. In the latter industries workers will be laid off quicker. The competitive position of high-skilled labour-intensive industries will be affected more than that of low-skilled labour-intensive industries. With the aim to take into account the skill-intensity of export commodities the two categories labour- and capital-intensive commodities are distinguished in a high-skilled and a low-skilled component (see Appendix A). If labour-intensive commodities are produced with relatively much high-skilled labour a negative effect on competitiveness may be expected. If production processes are capital- and high-skills intensive, labour costs still play an important role for reasons of capital-skill complementarity. If commodities are produced with relatively much low-skilled labour, stricter labour standards will increase labour costs less as compared with high-skilled produced goods. Thus the competitive position of industries making use of relatively much low-skilled labour will be affected negatively but relatively less. If the elasticity is rather large, the impact on competitiveness may even be absent. If production processes are characterized by low-skilled capital-intensive production techniques, labour costs play a minor role in total production costs. A rise in labour costs leads to a decrease of the relative price of capital which encourages a substitution of capital-intensive for labour-intensive production. Thus it can be expected that countries with relatively high labour standards will export relatively more low-skilled capital-intensive commodities. The effects of stringent labour standards on exports are summarized in Table 2. 10

14 TABLE 2 Expected signs of impact of high-labour standards on exports of different factor- and skill-intensity Low-Skilled High-Skilled Labour-intensive -/0 - Capital-intensive THE TRADE FLOW EQUATION EXTENDED WITH LABOUR STANDARDS The trade flow equation or gravity model is a powerful econometric tool for analyzing bilateral trade flows (Linnemann, 1966). For the aim of this study it is very useful because it takes into account bilateral trade flows and therefore provides us with a good opportunity to examine how differences in strictness of labour standards between countries influence foreign trade flows. One may wonder why a Heckscher-Ohlin-Vanek framework has not been used. There are two reasons to prefer a bilateral trade flows model. First, a disadvantage of the Heckscher-Ohlin approach is that it is based on multilateral trade flows. Effects of differences in labour regulations on trade flows between countries may cancel out in such an approach as multilateral trade is an aggregate of bilateral trade flows. 3 This drawback is not present in case of a bilateral trade flows model. Second, as has been shown by several authors the gravity model is consistent with factor endowment models (Bergstrand (1989), Deardorff (1995)). The gravity model has the following form: β β β β β β Pij uij ij β0 i j i j ij X = Y Y N N D e e (1) with: Xij = Trade flow of country i to country j in thousands of US $ 3 If, in a three country world, country X has stricter (looser) labour regulations than country Y (Z), X's exports to Y (Z) are assumed to decrease (increase). Aggregation of both trade flows into one multilateral trade flow may cancel out these effects. 11

15 Yi = Gross Domestic Product of country i in billions of US $ Yj = Gross Domestic Product of country j in billions of US $ Ni Nj = Population of country i in millions = Population of country j in millions Dij= Distance between country i and j in nautical miles Pij uij = Dummy variables = Log-normally distributed disturbance term In this equation bilateral trade is considered to be dependent on: (i) The potential demand of the importing country, which is represented by it's national product; it is assumed that imports of a country are positively related to it's domestic product; moreover, often population is included; a country with a large population can much easier specialize in a wide range of commodities and, consequently, will be less dependent on foreign trade. (ii) The potential supply of the exporting country, which is caught by the national product and the population of the exporting country. (iii) The resistance to trade which is introduced in the model by the geographical distance between important trading centres. This is a proxy for transportation costs including time costs, and it also reflects cultural distance, knowledge of markets, etc. These five variables explain what has been called the "normal" (or "typical" or "potential" trade) between country i and j. It is expected that b1 and b2 are positive and that b3, b4 and b5 are negative. In order to take account of specific deviations from the normal trade pattern, dummy variables (Pij) are included as well, reflecting membership of a customs union and the 12

16 circumstance that country i and j are neighbours, or the circumstance that there are important trade obstacles between two countries. We also include the variables LANDi and LANDj - land area of country i and j - which are assumed to exert a negative influence on trade flows. The larger a country's total area, the smaller the fraction of its economic activity that is expected to cross borders. The new element in this study is the extension of the basic model with variables that describe the strictness of domestic labour regulations. These variables, LabSi and LabSj, are incorporated to test the hypothesis whether stringent domestic labour standards exert a depressing influence on competitiveness resulting in lower exports and higher imports. The log-linear version of the extended model is: log X = log β + β log Y + β log Y + β log N + β log N + ij 0 1 i 2 j 3 i 4 j β log D + β A + β EC + β EFTA + β log LAND 5 ij 6 ij 7 ij 8 ij 9 i + β log LAND + β LabS + β LabS 10 j 11 i 12 j (2) with: Aij = dummy with value 1 if both exporter i and importer j are adjacent countries and zero otherwise. ECij = dummy with value 1 if both exporter i and importer j are a member of the European Community and zero otherwise. EFTAij = dummy with value 1 if both exporter i and importer j are a member of the EFTA and zero otherwise. LANDi = land area of country i in millions of ha. LANDj = land area of country j in millions of ha. LabSi = measure for the strictness of labour standards in exporting country i. LabSj = measure for the strictness of labour standards in importing country j. 13

17 The coefficients b6, b7, and b8 are expected to have positive values, as they represent a geographical situation or specific trade arrangements that push the level of bilateral trade flows up. b9 and b10 are assumed to be negative. With respect to the labour standard variables, b11 is assumed to have a negative and b12 a positive value. 6. ESTIMATION RESULTS In Table 3, seven regressions are reported, based on data for the year The dependent variable in the first one is total bilateral trade flows among the sample countries. The variables LabSi and LabSj present the synthetic index of labour standards (see Table 1) for the exporting and importing country respectively. 4 The coefficients b1, b2, b3 and b5 show significant values with signs in conformity with theoretical expectations. Coefficient b4 is not significant which can be explained by multicollinearity between Yj and Nj. The coefficients b6 till b10 reveal correct signs but are not significant except for b6 and b7. 5 Relatively stringent labour standards do not affect total bilateral trade flow as can be concluded from the insignificant estimates for b11 and b12. This is not surprising as the dependent variable Xij consists of a mix of labourintensive, capital-intensive and resource-intensive goods (UNIDO, 1981). 6 On theoretical 4 Additional regulations agreed upon by employers and unions outside the governmental framework are not taken into account in the OECD assessment. However, such regulations are most important in a decentralised system in which government regulations are not very extensive. As argued in section 3, labour standards are relatively less strict in such a system. Therefore it can be expected that nongovernmental regulations will hardly affect the competitiveness of a country. 5 The critical remarks of Polak (1996) on estimating integration effects in the framework of a gravity model are not valid here because the distances between countries participating in economic integration schemes are not very large in the present study's sample. 6 As an alternative, it is possible to regress unit labour costs (ULC) of the sample countries on the labour standards index. This generates an estimate of with a t-value of 1.18 and leads to the conclusion that relatively stringent labour standards do not affect ULC significantly which is consistent with the conclusion coming out of regression 1. As no data for ULCs at a disaggregated level are available, it is not possible to run this exercise to find comparable estimates for regressions 2 to 7. 14

18 grounds it is expected that labour-intensive and capital-intensive trade react to relatively stringent labour standards in an opposite way towards each other while resource-intensive trade is assumed to be not affected at all. 7 Therefore in total bilateral trade flows these effects are cancelled out. Regression 2 and 3 show estimates of a trade flow equation with labour-intensive bilateral trade flows and capital-intensive bilateral trade flows respectively as the dependent variable. The Yi coefficient in regression 2 is low compared with regression 3 and shows that countries at a relatively high level of economic development do export relatively less labour-intensive goods and relatively more capital-intensive goods. Labour-intensive goods do not play an important role in intra-ec trade patterns but they do in intra-efta patterns and in case of adjacency. For capital-intensive commodities it is just the other way around. Interesting is the negative significant estimate for LANDi in both regressions. The explanation for the insignificant result in regression 1 can be attributed to the existence of resource-intensive commodities. These are mostly found and exported in countries with a large size. This positive effect on resourceintensive goods counteracts and even cancels out the negative effects of land size on labour- and capital-intensive commodities' bilateral trade. The slightly negative nonsignificant estimate for the stringency of labour standards variable on the exports of labour-intensive goods (b11 in regression 2) reveals that countries with relatively stringent labour standards do not export relatively less labour-intensive goods as compared with the average pattern. Although the sign of the estimate is correct, the nonsignificance makes it different from theoretical expectations. The main result of regression 3 is the positive impact of labour standard stringency on capitalintensive bilateral trade flows although not significant. This leads to the conclusion that countries with relatively stringent labour standards do not export significantly more capital 7 Resource-intensive industries are not very sensitive to the costs of high labour standards as their main production factor is the resource. 15

19 goods than the average pattern. On the import side no significant impact of relatively stringent labour standards on both bilateral labour-intensive and capital-intensive imports can be found. 16

20 TABLE 3 Estimation results of trade flow equation with labour standards: 1992 Regression Trade Flow Xij Xli Xci Xlils Xlihs Xcils Xcihs Constant 5.09 ** (9.28) 5.04 ** (7.85) (-0.43) 7.13 ** (10.31) 0.40 (0.56) (-1.47) ** (-3.47) Yi (β 1) 1.13 ** (10.13) 0.79 ** (5.55) 2.03 ** (12.08) (-0.50) 2.09 ** (13.01) 1.67 ** (10.85) 2.68 ** (17.24) Yj (β 2) 0.73 ** (6.60) 0.75 ** (6.02) 0.77 ** (4.97) 0.59 ** (4.40) 0.76 ** (5.26) 0.93 ** (6.08) 0.98 ** (6.27) Ni (β 3) ** (-3.11) 0.22 (1.48) ** (-5.50) 1.30 ** (8.10) ** (-6.60) ** (-3.88) ** (-9.71) Nj 0.09 (β 4) (0.76) 0.04 (0.27) 0.13 (0.78) 0.23 (1.50) 0.03 (0.21) (-0.06) (-1.27) Dij (β 5) ** (-10.72) ** (-8.11) ** (-8.50) ** (-6.22) ** (-8.84) ** (-8.43) ** (-6.42) Aij 0.33 * (β 6) (2.54) 0.51 ** (3.53) 0.23 (1.32) 0.80 ** (5.32) 0.21 (1.26) 0.30 (1.47) 0.24 (1.13) ECij 0.21 * (β 7) (2.12) 0.08 (0.71) 0.31 * (2.33) (-0.83) 0.31 * (2.25) 0.01 (0.04) 0.45 ** (2.94) EFTAij 0.11 (β 8) (0.77) 0.40 * (2.39) 0.32 (1.65) 0.57 ** (3.20) 0.31 (1.66) 0.60 ** (2.62) 0.11 (0.46) LANDi (β 9) (-1.50) ** (-6.25) ** (-3.21) ** (-10.79) * (-2.26) ** (-5.21) ** (-5.57) LANDj (β 10) (-1.52) * (-2.13) (-1.03) ** (-3.76) * (-0.29) (-1.34) (-0.43) LabSi (β 11) (-1.33) (-1.49) 0.02 (1.19) (-0.02) ** (-3.51) 0.12 ** (5.27) ** (-4.20) LabSj 0.01 (β 12) (0.57) (0.10) 0.03 (1.65) (-1.00) (0.22) 0.03 (1.29) 0.01 (0.52) Sigma 0.83 ** Log- Likelihood R 2 adjusted * = significant at 5%; ** = significant at 1%. 17 (24.74) 0.84 ** (24.74) Notes: see main text for explanation of variables; for data sources, see Appendix B. Trade flow and GDP data are real. Regression 1-5: t-values based on White (1980) heteroscedasticity corrected standard errors. Regression 6-7: Tobit-estimates to take account of zero bilateral trade flows (only two trade flows are zero). LabS-variables are based on the synthetic index presented in section 3. Xij = all trade; Xli (Xci) = labour-intensive (capital-intensive) trade; Xlils (Xlihs) = labour-intensive lowskilled (high-skilled) trade; Xcils (Xcihs) = capital-intensive low-skilled (high-skilled) trade.

21 Regressions 4 and 5 present a distinction of labour-intensive commodities, i.e. those produced with relatively much low-skilled labour and those produced with relatively much high-skilled labour. Regression 4 learns that b1 is insignificant and b3 is strongly significant and positive. 8 The size of population and thus the labour force in the exporting country is an important determinant of bilateral exports of labour-intensive low-skilled goods. This suggests that these goods are Heckscher-Ohlin goods (see Bergstrand, 1989: 146). Labour-intensive lowskilled commodities do not play an important role in the European Union as is shown by the insignificant estimate of b7. These goods are important in intra-efta trade and in adjacency trade. The significant negative impact of both LANDi (b8) and LANDj (b9) is theoretically expected. The most surprising revelation is that labour standards do not exert a significant impact on both exports and imports of labour-intensive low-skilled commodities. A relatively high elasticity of demand for unskilled labour may be the explanation. Regression 5 shows the effects on high-skilled labour-intensive commodities. All coefficients b1 till b10 correspond with theoretical expectations. Note that now intra-european Union trade appears to be significant positive (b7) which - as expected - is in contrast to the findings of regression 4. The coefficient of b11 is significant negative. Countries with relatively strict labour standards experience relatively less exports of high-skilled labour-intensive commodities than the average. 9 The presence of substantial labour costs as a result of higher skilled labour used in labour-intensive production processes means that more strict labour standards are important additional labour costs which reduce competitiveness. On the import side no significant effect could be found. In regressions 6 and 7 the capital-intensive trade flows are distinguished in a low-skilled and 8 Note that income per capita can be considered as a proxy for labour costs (see Rodrik, 1996: 36). 9 The presence of human capital is an important determinant for high-skilled commodities' exports. Generally the level of income per capita is a good indication of the level of human capital. Higher income per capita is connected to more human capital. 18

22 a high-skilled component. For both regressions the estimates of b1 till b10 are plausible. Note, again, that intra-european trade is important in case of capital-intensive commodities but only for the part that is produced with high-skilled labour while intra-efta trade is significantly higher on average if capital-intensive goods' production is low-skilled. The positive estimate of b11 in regression 6 reveals that countries with relatively strict labour standards export relatively more capital-intensive goods produced with low-skilled labour. As capital and low-skilled labour can be considered as substitutes this result suggests that increasing labour costs as a result of relatively stringent labour standards go together with more capital-intensive production. 10 The slight significant negative estimate of b11 in regression 7 can be explained by the capital-skills complementarity. If an industry gets more capital-intensive, it also requires more high-skilled labour. No effects on the imports could be found. The nonsignificant effect on the imports which is found in all regressions does not correspond with theoretical expectations. This suggests that increasing labour costs as a result of higher labour standards do not affect domestic consumers' import demand. The reasons for this result may be twofold, (1) either increasing labour standards do not increase labour costs substantially to make domestic consumers switch from domestic to foreign products, or (2) labour costs are increased sufficiently to let domestic consumers switch but this effect is counteracted by trade barriers. The relatively high demand elasticity for low-skilled labour suggests that labour costs increase not substantially if labour standards are relatively high which makes the first reason plausible. The second suggestion may prevail with regard to high-skilled commodities as the lower demand elasticity for high-skilled labour suggests that labour costs do increase and induce consumers to switch to demand of foreign goods. However, caution should prevail in drawing conclusions with regard to imports as the present analysis is not aimed at 10 Estimation results of high-skilled and low-skilled trade flows not distinguished to labour-and capital-intensive parts show a significant negative effect of relatively high labour standards on high-skilled exports. A positive effect on low-skilled exports can be reported. This latter result can be attributed to the capital-intensive component. The 19

23 answering these questions explicitly. The model has also been estimated for a selected sample of the European Union countries only. These results are not presented here. 11 The estimates show that no changes appear as compared with the estimates for the total sample of eighteen OECD countries. Again, both labour-intensive and capital-intensive commodities produced relatively high-skilled intensive, are negatively affected by relatively high labour standards. If the two kinds of commodities are produced with a relatively low-skilled intensive production technique, labour-intensive exports are not significantly affected while capital-intensive exports are positively affected but at a lower significance level (10%) than in the OECD-sample. Also, in the smaller European Union sample, imports are not affected by differences in labour standards. 7. CONCLUSIONS This paper examines whether differences in labour standards' strictness among OECD countries affect export flows among these countries, i.e. countries with similar political systems. In theory it can be expected that labour costs in countries with relatively high labour standards are relatively high and therefore these countries experience a loss of competitiveness. On the import side a positive effect of relatively higher labour standards can be expected as domestic consumers will substitute foreign products for domestic ones. A major problem is to construct a reliable measure for differences in stringency of labour standards. In the present study a synthetic index constructed by the OECD has been used to measure the stringency variable. The index has been built up of partial indices which consist of results are not presented but are available upon request from the author. 11 These results are available upon request from the author. 20

24 judgements by the OECD on whether working time, fixed term contracts, employment protection, minimum wages and employees representation rights are light, intermediate or strict. Although far from perfect, the advantage of these indices is that they are constructed by taking into account whether they are determined by government regulations (more tight standards) or by decentralised systems (less tight standards). An important finding is that in Europe relatively stringent labour standards are experienced by France, Germany, Greece, Italy, Spain and Sweden. In Germany and Sweden this is mainly caused by employees representation rights while in the Southern European nations rigid employment protection and minimum wages cause labour standards to be relatively stringent. The synthetic index is used as an independent variable in a trade flow equation to investigate it's impact on bilateral trade flows. In correspondence with another empirical study no significant impact of labour standards stringency on exports of labour-intensive commodities can be found. However, if the bilateral trade flows are also distinguished according to differences in skill-intensities a significant negative impact is found on exports of both labourintensive and capital-intensive commodities that are produced with relatively much high-skilled labour. The reason suggested is the relatively inelastic demand for high-skilled labour which causes labour costs to rise more than in case of low-skilled labour intensive commodities. If labour-intensive exports are produced with relatively much low-skilled labour no significant effect could be found. The relatively high elasticity of low-skilled labour demand provides an explanation for this empirical finding. Furthermore a positive significant effect is found on capital-intensive low-skilled exports. This means that these exports are mainly coming from countries with relatively high labour standards. Substitution of capital for low-skilled labour is an important explanation. Estimates for intra-european trade flows reveal a similar pattern. Imports of the commodities classified are not significantly affected by differences in labour standards. This suggests the existence of possible trade barriers for high-skilled commodities' 21

25 imports and lack of labour costs increases in low-skilled commodities' production as a result of a relatively high elasticity of low-skilled labour demand. 22

26 APPENDIX A Table A.1 Factor-Intensity of Manufactured Commodities, SITC rev. 2 Labour-Intensive Commodities Capital-Intensive Commodities Source: UNIDO, 1981:

27 Table A.2 Skill-Intensity of Manufactured Commodities, SITC rev. 2 High Skilled-Intensive Commodities Low Skilled-Intensive Commodities Source: UNIDO, 1981:

28 APPENDIX B. Data Sources of Variables The data for the bilateral trade flows are import data of 1992 obtained from United Nations Statistical Papers, Series D. The income- and population data are reported in IMF International Statistics, May The distances are those used by Linnemann (1966). The variables "LAND" (millions of ha. in 1990) were obtained from Human Development Report Price indices for exports and imports used to convert from nominal to real variables were reported in World Tables The countries used in the study are reported in Table 1 in the main text. 25

29 REFERENCES Anderson, Kym (1995), 'The Entwining of Trade Policy with Environmental and Labour Standards', Seminar Paper 95-01, (Since published as Ch. 15 in The Uruguay Round and the Developing Countries, edited by W. Martin and L.A. Winters, Cambridge and New York: Cambridge University Press, 1996.) Bergstrand, Jeffrey (1989), 'The Generalized Gravity Equation. Monopolistic Competition, and the Factors-Proportions Theory in International Trade', Review of Economics and Statistics, 71, LXXI: Brown D., Alan Deardorff and Robert Stern (1996), 'International Labour Standards and Trade: A Theoretical Analysis', in J. Bhagwati and R. Hudec (eds.), Fair Trade and Harmonisation; prerequisites for Free Trade? (Cambridge Massachusetts: MIT Press). Charnovitz, Steve (1987), 'The Influence of International Labour Standards on the World Trading Regime. A Historical Overview', International Labour Review, 126, 5, Deardorff, Alan (1995), Determinants of Bilateral Trade: Does Gravity Work in a Neoclassical World? NBER Working Paper No Filler, Randall K., Daniel S. Hammermesh and Albert Rees (1996), The Economics of Work and Pay, (New York: HarperCollins Publishers). Hoekman, Bernard and Michel Kostecki (1995), The Political Economy of the World Trade System. From GATT to WTO, (New York: Oxford University Press). Krueger, Alan B. (1997), International Labour Standards and Trade, in: M. Bruno and B. Pleskovic (eds.), Annual World Bank Conference on Development Economics, (Washington DC). Linnemann, Hans (1966). An Econometric Study of International Trade Flows (Amsterdam: North Holland Publishing Company). 26

30 OECD (1994), Employment Outlook (Paris). Polak, Jacques J. (1996), 'Is APEC a Natural Regional Trading Bloc? A Critique of the 'Gravity Model' of International Trade', The World Economy, 19, Rodrik, Dani (1996), 'Labour Standards in International Trade: Do They Matter and What Do We Do About Them', (Overseas Development Council). Sapir, André (1995), 'The Interaction Between Labour Standards and International Trade Policy', The World Economy, 18, 6, Srinivasan, T.N. (1994), 'International Labour Standards Once Again', in: International Labour Standards and Global Economic Interpretation, Proceedings of a Symposium, (Washington DC: US Department of Labour). Swinnerton, Kenneth and Gregory K. Schoepfle (1994), 'Emerging Themes', in: International Labour Standards and Global Economic Interpretation, Proceedings of a Symposium, (Washington DC: US Department of Labour). UNIDO (1981), World Industry in 1980 (New York: United Nations). Van Beers, Cees and Jeroen J.C.J.M van den Bergh (1997), 'An Empirical Multi-Country Analysis of the Impact of Environmental Regulations on Foreign Trade Flows', Kyklos, 50, White, Halbert (1980). 'A Heteroscedasticity-Covariance Matrix Estimator and a Direct Test for Heteroscedasticity', Econometrica, 48: