Britain's pattern of specialization in manufactured goods with developing countries and trade protection

Britain's Pattern of Specialization in Manufactured Goods with Developing Countries and Trade Protection - SWP425 Vlorld Bank Staff Working Paper No. 425 October 1980 Prepared by: Vincent Cable and Ivonia Rebelo (Consultants) - conomic Analysis and Projections Department 1980 k N.W. C. 20433,. U.S.A. d interpretations in this document are those of the authors > ancd shouTld not be attributed to the World Bank, to its affiliated 1 n n X X ( F 'W/ organizations, or to any individual acting in their behalf. 1 > j (nl ) (   The views and interpretations in this document are those of the authors and should not be attributed to the World Bank, to its affiliated organizations, or to any individual acting in their behalf. WORLD BANK Staff Working Paper No.425 BRITAIN'S PATTERN OF SPECIALIZATION IN MANUFACTURED GOODS WITH DEVELOPING COUNTRIES AND TRADE PROTECTION This work in progress report is part of an inquiry being undertaken by the World Bank in conjunction with scholars from twelve industrial countries into the penetration of the markets of industrial countries by exports of manufactures from developing countries. The project seeks to establish the shares of industrial country markets held by the developing countries, changes in such shares in the 1970s, and why they vary among industry groups and countries. The aim is to assist developing and industrial countries to improve their policies through a better understanding of trade patterns and protectionist pressures. This paper studies the influences determining the emergence of protectionist pressures in the industrial countries in response to market penetration by products from other countries, and analyses the trigger points for taking protectionist measures. The problem is approached in two stages. Firstly, an explanatory outline for the pattern of manufacturing trade is mapped. Secondly, an attempt is made to explain the determinants of protectionist reaction. It is found that there is a strong association between the pattern of trade with developing countries and variables relating to aspects of "human capital", a much weaker relationship between trade patterns with other groups of countries, and virtually none for changes in trade patterns over time. The measurement of what has been called the "new protection", the selective use of non-traditional forms of trade barriers, directed mainly at non-industrial country competitors, presented considerable technical difficulties. However, use of a dummy variable permitted an analysis which showed a protectionist response to trade patterns apparently designed to negate or frustrate trade changes based upon the UK's comparative advantage in trade with developing countries. We would like to thank the participants in the workshop on market penetration held in Kiel in November 1979 for their comments. Prepared by: Vincent Cable and Ivonia Rebelo (Consultants) Economic Analysis and Projections Department Copyright c 1980 The World Bank 'NTERNATIONAL bMONETAAY FUND 1818 H St. N.W. JOIT LIBRARY Washington D.C. 20433, U.S.A. MAY 1 19P4 INTERlATIONAL BANg FOR RECONSTRUCTION AND DEVELOPpfNT WAS}INGTON, D.C. 20431   Introduction In line with the objectives of the World Bank project as a whole, our aim is "to study the influences determining the emergence of protectionist pressures in the developed countries in response to market penetration by products from other countries and to analyse the "trigger points" for taking protectionist measures". The problem is approached in two stages. The first is to try to map out an explanatory outline for the pattern of manufac- turing trade with developing countries (and the world as a whole). In effect, we seek to establish the current determinants of comparative advantage and changes in it. This work follows to some degree in well-trodden footsteps and as far as the UK is concerned takes advantage of recent advances in the presentation of UK trade data by industry and insights obtained in analytical work already done by Department of Industry economists. The second stage is to try to explain why a protectionist reaction is triggered in some industries to a greater extent than in others. The degree of protection afforded by the proliferation of non-tariff barriers is extremely difficult to measure in terms of tariff equivalents or any other continuous variable, and cruder measures have to be employed. An attempt was made to use 0-1 variables allowing a greater range of alternatives to reflect different degrees of severity of protection, in particular the use of quotas rather than tariffs, and semi-sensitive' rather than sensitive' tariff treatment under the GSP scheme. Section 1 sets out the variables employed and data used. Section 2 deals with the methodology and results of a study of patterns of manufacturing specialisation with developing countries and other groups of countries. Section 3 deals with the methodology and results of a study of the inter- industry variation in UK protection as it applies to trade in manufactures in developing countries.  -2- I. DATA USED The analysis of both trade patterns and of the 'triggering' of trade protection is carried out using 87 3 digit non-food manufacturing (MLH) categories. The empirical foundations of the study rested upon the avail- ability of trade data expressed in terms of manufacturing census categories so that hypotheses_,can be tested which relate trade patterns and their changes. to industry and labour characteristics derived from the census. It is now possible to conduct such analysis for the UK as a result of the publication by the government for the years from 1970 of imports and exports expressed in terms of 3 digit MLH categories(l), of the estimation by Wells and Imber of import penetration and export sales ratios from 1968 to 1976(2) and latterly of the publications of more disaggregated and updated estimation of these ratios(3). These published sources provide imports and exports only in aggregate -terms and the authors were fortunate to have unpublished data giving a geographical breakdown. The sample excludes 17 items from the three digit list in addition to food items. The main criterion for exclusion is that the category of goods is predominately untraded, applying the criterion that exports and imports of a traded good together account for 10% of UK home consumption. This is however only a partial remedy; in trade with particular groups of countries for particular years the sample of items used to explain patterns of trade specialisation and trade protection contains a number of items which are not involved in trade either as imports or exports. A substantial number of variables is used as dependent or independent variables in both exercises. The studies are distinct though the variables  - 3 - are for the most part common to both. They are set out below in the context of the relevant hypotheses and formally defined in Appendix 1. II. STUDY I - THE PATTERN OF SPECIALISATION Refinement and testing of existing trade theory has given us a reasonably solid base from which to examine the pattern of UK manufacturing trade with developing countries. Much of this work points to the usefulness of a modified version of the Heckscher-Ohlin factor proportions theory in explaining patterns of trade: "a significant proportion of the variance in the commodity composition of trade in manufactured goods can be explained statistically by variations in three factor inputs, physical capital, human capital and labour. The proportion of variance explained goes as high as 45x----"(4). The crude factor endowment hypothesis (based on capital and labour alone) has been rendered inadequate by demonstration, and then repeated confirmation, of the Leontief paradox. Internationalisation of Leontief's work has been established that not only for the US, but for Canada, West Germany, Japan and in other "capital rich" economies, exports have a lower ratio of capital to labour than imports(5). Data limitations have made tests difficult to apply to the UK but recent work, including that below, gives strong support for all trade in manufactured goods, and even for trade with non-oil ldcs, to the existence of this "paradox". There is a rich literature which this paper cannot pursue in any detail explaining and interpreting this paradox and which leads to doubt over its significance, if the factor endowments of supplying industries and other than two factors are considered. Suffice it to say that trade patterns have been more satisfactorily explained if 'human capital' is intro- duced. Human capital can in turn be split into two separate components, that  of the skill of the labour force and that of innovativeness. Wolter found for Germany that though both were a strong positive influence on German trade performance, innovativeness (measured by R & D expenditure) was particularly so(6). Subsequent testing of hypotheses about trade patterns has encompassed a wider range of explanatory variables including the degree of. product differentiation, raw material use, and various aspects of indus- trial structure(7). The predicted pattern of British trade is less likely to conform to the stereotype of an 'advanced' industrial country than some others. The UK has been through a long period of relative decline in manufacturing trade. The level of per capita income has sunk - relatively - to a point where this is now about half of some of its main trade rivals (US, Germany, Scandinavia). The capital stock-is relatively old. There is much ancedote and!some evidence to the effect that labour markets work more slowly than elsewhere, and in particular that the differential between skilled and unskilled work is inadequate to produce a rapid movement of resources towards activities utilising relatively large amounts of human capital. In such an economy present and future line of specialisation in trade is, if not indeterminate, certainly unclear. The effects of this decline in relative performance can be subdivided into two(8). One is that British industrial trade performance increasingly fails - largely because of market malfunctioning - to reflect a comparative advantage in products utilising a high human capital content. The other is that Britain's comparative advantage is itself changing in the direction of goods which require less skill and inventiveness. The policy implications of these two interpretations of decline are radically different; the latter,  for example, implies that in the absence of bold policy initiatives to revise the quality of the labour force and to influence the direction of resources towards more technologically advanced activities, Britain's economically efficient trade pattern in manufactures will increasingly resemble that of the more advanced Southern European or Far Eastern newly industrializing countries (NICs). An analysis of changing trade patterns may help to illustrate whether the UK displays some of the characteristics of an advanced industrial economy (with developing countries). In other contexts, in trade with more advanced industrial economies, the U.K. is beginning to show some of the characteristics of a developing country. Empirical testing of such hypotheses have been inhibited until very recently by the lack of a suitable UK data base, of trade flows expressed in terms of disaggregated industry groups (or vice versa) and also by the absence of suitably disaggregated capital stock estimates. Some tests have however now been carried out by Department of Industry economists, notably by Owen and White(9). They have tried to explain the pattern of UK manufactured trade for 1970, 1975 and 1976 (94 categories, non-food) with the world as a whole, the OECD, the EEC and developing countries. They used multiple regression analysis to explain variations in an index of trade performance by industry, which was measured by (trade balance . UK apparent consumption). Their main conclusions, relevant to this paper, were: (i) a model based on factor proportions and industrial structure characteristics achieved a good fit (R2 = .39 to .41, for different years) for trade with developing countries. (ii) there was strong confirmation of the Leontief paradox in UK trade with the world as a whole and with developing countries in  particular, ie, British revealed compararative advantage in trade with developing countries was in relatively labour intensive goods (in relation to capital). (iii) British revealed comparative advantage in manufacturing trade with developing countries lay with industries which were high :wage (the most significant factor) but labour intensive, used an above-average proportion of skilled and white collar personnel, were more highly concentrated in terms of ownership, more likely to be foreign owned and to have good industrial relations. The same factors appeared to explain trade patterns with the world as a whole, though less convincingly than for developing countries, while trade with the,EC could not be explained at all in terms of the set of variables used. One factor which appeared to be unimportant in all contexts was the use of R & D. As part of the present-exercise these tests were conducted again but in a somewhat modified form, with a wider range of dependent and explanatory variables, more recent data and a more specific focus on UK-developing coun- try trade. The Hypothesis The specific hypothesis is that the pattern of UK manufacturing trade can be explained by combinations of a) factor endowments: human and physical capital resources; b) characteristics of industrial structure - the degree of foreign ownership, or firm size - which, other things being equal, influence the relative trade performance of particular sectors; c) the regio- nal concentrationfof manufacturing - as-between UK regions with different factor endowments -- influencing the extent to which inter-regional trade flows  serve as a substitute for international trade; d) the incidence of trade protection. We would expect that a) and c) would be a dominant factors in trade with countries with the most radically different factor endowments, that is with developing countries, and b) in trade with comparable industrial economies. Over time we would expect relationships to be less clear-cut, reflecting the tendency of the UK to slide down the league of industrial countries, and also a large amount of government intervention with no obvious logic in terms of inter- national specialisation. Dependent Variables The Owen White study used a measure of 'revealed comparative advantage' as an index of inter-industry trade performance. More alternative measures of trade performance are considered, in particular separating out exports and imports. The reasoning behind this was that a possibly misleading picture is created by the use of a measure of performance which includes trade balance as the numerator when there are many industries with substantial but roughly balanced intra-industry trade. Four alternative measures of trade performance - export-sales ratio (ES), import penetration (IP), and two measures of revealed comparative advantage (RCAI and RCA2) - were used. (For a definition of the variables see Appendix 1). Several problems arise in defining these variables. Firstly, domestic and foreign products in the same category may be different and non-competing. For example, imported carpets from developing countries and East European centrally planned economies are usually handwoven, catering for a different market to domestic production than imports in the same category  - 8 - from industrial countries (which use quite different production techniques). But there is no clear cut dividing line between competing and non-competing products, and in the absence of an objective criterion - and suspecting this was not a major problem - we disregarded the problem except in the one obvious case of carpets. Secondly, where trade and production data are combined in a common index (eg, IP), the production (but not the trade) figure may be inflated by double-counting for industries where intra-industry sales are a high proportion of the total. For example, for cars, aerospace, shipbuilding and electronic components, domestic sales of intermediate products are often counted more than once, as components and as part of the final product. In order to minimise this distortion a more conventional measure of revealed comparative advantage which uses only trade data was used (RCA 1). But for contrast another measure which gave some weighting to domestic sales was also tried (RCA 2). This other measure was also used since the former measure "does not distinguish between significant trade balances and those without..... Moreover because they focus on traded products it could be that in some industries the independent variables which refer to the entire industry concerned are highly unrepresentative of the independent variable which apply to these particular products" (10). There were practical difficulties in obtaining a very disaggregated picture of trade patterns in terms of geographical regions for any year other than 1978. The following units were considered for 1970 and 1978: the world as a whole, developing countries (defined to exclude all OECD countries, all East European countries, Israel, Turkey, and including Communist Asian countries), subdivided into OPEC and non-OPEC; Japan and COMECON. Trade with the US and the EEC was considered for 1978 alone.  -9 Independent Variables (i) the choice of independent variables to represent factor endowments centered on proxies for human and physical capital. There are inherent difficulties in representing simply a multifaceted concept such as human capital and severe problems in measuring physical capital in the absence of capital stock data at the level of disaggregation with which we are working. The human capital endowment was represented by two of its principle compo- nents: the innovativeness of firms resulting from a high R & D effort, and skill in the workforce as a whole resulting from substantial training and experience of complex operations. The first (R & D) was represented by a measure of expenditure on R & D as a proportion of industry sales but the figures were aggregated to 32 categories and we had to assume that the same coefficients applied at a disaggregated level. The second was initially represented by a series of indirect measures taken from 1975 manufacturing census data. (The 1977 figures are now available but do not significantly affect the inter-industry differences). Within the considerable limitations of UK census data, these give necessarily rather unsatisfactory proxy measures of human capital for the sample of industries. The data used were the share of 'operatives' in the labour force (OPSH) (in effect the share of manual workers, skilled and unskilled, rather than clerical, managerial and scienti- fic staff); the share of women employees (WMSH); and the average wage AVWAGE. For the last of these we elected to use the average annual earnings for the labour force as a whole rather than the average weekly manual wage employed by Owen and White since it seemed likely to incorporate, in a more comprehensive way, the human capital element in .the industry work force (but any measure would be distorted by labour market imperfections). After experimenting with  - 10 - these variables we were aware of the availability of a direct skill measure (SKILL) derivable from a 1970 population census classification of the work force of industries in terms of (five) social classes. Although this was superior to our proxy variables in that it measured skill directly, the differentiation is based on an arbitrary and somewhat questionable definition of skill. The boundary between semi-skilled and skilled is very important for our definition and the qualifications for admission to the "skill" category in the census did not appear particularly demanding. The most "skill"-intensive manufacturing industry on the population census scale was footwear and several textile trades, notably knitwear, were not far behind. There were also difficulties in the calculations of physical capital intensity (K/L) since capital stock data do not exist in the form required. One possibility is to use a measure of capital stock which is inferred from accumulated depreciation over a specified period of years. This is the approach used by Owen and White drawing upon NEDO estimates. Another approach, which also uses flows over time as a substitute for stock, is to measure cumulative net capital investment over a fixed period, assuming that this correlates with capital stock. This is the method used here. These measures have obvious deficiencies but when tested there was a high correlation between them. A measure of value added per man was incorporated (VAPM). This should reflect the combined influence of human and physical capital, (but it has deficiencies - notably the use of a flow measure, affected strongly by inter- annual variations in profits). A final variation was to measure labour intensity in relation to output rather than capital stock, that is, the inverse of productivity (LAB).  - 11 - (ii) We incorporated several variables intended to capture the influence of industrial structure. We would expect this to be important in explaining the pattern of trade with other industrial countries. We incorporated one measure of economies of scale, the average firm size (AFSZ). There has been a long history of official intervention in UK industry designed to promote amalgamations, the underlying assumption being that economies of scale are a significant influence on trade performance. Clearly this could be refined in various ways to allow for differences between plant and firm size, and the difference between average firm size and the degree of industrial concentra- tion all of which represent different aspects of scale economies. Another factor of possible importance is the degree of concentration of ownership (CONC), the hypothesis being that import penetration is less where the domestic producers are more organized along oligopolistic lines. Finally we consider foreign investment in UK industry (FINV). It would be a plausible hypothesis that industries highly penetrated by foreign investment have a higher degree of involvement in international trade, both in terms of export performance and import penetration, because of the facility of trade between subsidiaries. Unfortunately the measure is not entirely satisfactory since information is only available at an aggregated level (in 16, two-digit categories) and the assumption is made that the industrial subdivisions share the same coefficient as the larger categories. (iii) in addition we incorporated a series of measures designed to reflect the regional distribution of employment in the UK. We would expect, a priori, that the pattern of comparative advantage in trade with developing countries would overlap with the UK regional structure of employment to the extent that  - 12 - international trade is a partial substitute for inter-regional trade. The regions of the UK characterized by high unemployment should, if there is a functioning labour market, also have relatively low wages, and attract indus- tries which use relatively more intensively labour which is more available and cheaper than elsewhere. Regional policy has been active, if erratic, and one of its aspects has been an attemp.t to subsidise labour in high unemployment regions by such devices as the Regional Employment Premium to attract or retain relatively labour intensive footloose industries. These are also the industries in which developing countries would be expected to have a compara- tive advantage. Unfortunately there data is not available on the detailed regional distribution of employment. The best available is a breakdown into 11 planning regions. We calculated three indices. One, RBSI, measured the share of employment by industry in those planning regions which are of above average unemployment and are major beneficiaries of regional assistance (N. Ireland, Wales, Scotland, the North of England). Another RBS2 also included 'intermediate' regions (Yorkshire and the North West). A third RBS3 mea- sured crudely the degree of concentration of employment by region. These three variables are correlated with each other, but only RBS1 and RBS2 sig- nificantly. (iv) interindustry differences in the extent of protection will clearly bias trade performance on the import side. The measurement of protection raises, however, complex issues which will be discussed below, and the use of a protection measure as an independent variable (PRMSA to PRMSE) is treated as part of the second study.  - 13 - The Model By assuming a linear relationship between our dependent and independent variables we can estimate multiple regression of the following form: IP )) ES ) RCA1) = a + b (K) + c (HK) + d (IS) + e (RBS) + f (PRM) +L RCA2) K is physical capital, HK is human capital, IS is industrial structure, RBS is regional bias in employment, PRM is protection. Each of these is repre- sented by several variables in combination or independent of each other: a is a constant, b to i are regression coefficients and i is the error term. The significance and sign of coefficients b to f should give us some indication of the force of a modified factor proportions model of international trade. But it should be stressed that these regressions are, at best, half of a test of trade theory. They also serve as a descriptive data analysis designed, by identifying factors which correlate significantly with trade patterns, to indicate likely sources of trade adjustment difficulty, permitting us to formulate hypotheses that can then be used for an analysis of protection.  Table 1: ESTIMATION OF REGRESSION COEFFICIENTS - DEPENDENT VARIABLE IS REVELAED COMPARATIVE ADVANTAGE (n = 87) Dependent Variable AFSZ R+D K/L VAPM OPSH WMSH AVWAGE RBSI Const R2 F RCA1 1978 .144 .0002 -2.02 -.244 -65.03 -.241 .719 -.168 53.80 .46 9.61 LDC+OPEC (.721) (.552) (-1.53) (-1.76)* (-3.10)*** (-1.64) (1.065) (-1.60) (1.92) RCA1 1978 -.163 .1553 -.288 -.101 -36.87 -.267 .404 -.181 30.32 .52 12.47 LDC (-.114) (.611) (-.304) (-1.02) (-2.45)** (-2.53)** (.835) (-2.40)** (1.51) RCA1 1978 -.592 -.571 .427 .747 -3.23 .261 .633 .402 -2.92 .03 .39 Japan (-.710) (-.385) (.772) (1.295) (-.369) (.426) (.228) (.915) (-.250) RCA1 1978 .610 .960 -.181 .iil -19.79 .182 -.354 .108 17.55 .07 .89 EEC (.481) (.426) (-.216) (-127) (-1.48) (.203) (-.825) (1.63) (.987) RCA1 1978 -.347 -.184 .261 .220 21.39 .338 .834 -.391 -18.67 .19 2.68 USA (-.559) (-1.67)* (.634) (.497) (3.27)*** (.738) (.396) (-.119) (-2.14)** RCA1 1978 -.449 -.119 -.879 -.386 -5.33 .261 .256 .123 -2.39 .20 2.82 COMECON (-1.14) (-.169) (-.336) (-1.416) (-1.29) (.901) (1.919)* (.592) (-.432) RCA1 1978 .158 -.118 -2.31 -.150 -103.99 .111 .154 -.167 48.68 .20 2.75 World (.375) (-.158) (-.825) (-.508) (-2.35)** (.359) (1.076) (-.752) (.822) RCA1 1970 .148 .055 -2.42 -.385 -37.03 -.209 .161 -.351 8.83 .46 9.49 LDC+OPEC (.749) (.157) (-1.84)** (-2.86)*** (-1.78)* (-.144) (2.411)** (-3.37)*** (.318) RCA1 1970 -.006 .016 -.735) .327 -3.44 .0035 .0005 .018 .678 .04 .48 Japan (-.148) (.236) (-.029) (1.25) (-.868) (.126) (.413) (.898) (.128) RCA1 1970 .547 .449 -.620 -.109 -8.286 .087 .450 .197 -4.982 .12 1.61 COMECON (.866) (.040) (-1.482) (-2.33)** (-1.251) (1.662k (2.111)** (.592) (-.563) RCA1 1970 .533 -.414 -6.01 -.726 -42.38 .251 .0274 -.666 1.232 .23 3.34 World (1.139) (-.498) (-1.939)* (-2.24)* (-.863 (.730) (1.730)* (-2.704)*** (.018) RCA1 1978-70 -.0069 -.182 .380 -.391 .545 .163 -.208 -.023 .156 .05 0.58 LDC+OPEC (-.298) (-.445) (.249) (-.024) (.226) (.096) (-.027) (-1.879)* (.048) RCA1 1978-70 -.175 -.754 2.003 .002 -46.89 .234 -.122 .029 53.88 .06 0.78 Japan (-.593) (-.014) (1.023) (1.152) (-1.511) (.108) (-1.218) (.190) (1.300) RCA1 1978-70 -.846 -.174 .614 .386 .642 .014 -.579 -.123 -.271 .03 0.38 World (-1.363) (-.016) (1.496) (.886) (.099) (.316) (-.276) (-.377) (-.031) * Significant at 10% confidence level. ** Significant at 5% confidence level. I *** Significant at 1% confidence level. Note: T - statistic in brackets.  - 15 - Table 2: ESTIMATION OF REGRESSION COEFFICIENTS - DEPENDENT VARIABLES ARE IMPORT PENETRATION AND EXPORT/SALES RATIOS Variable AFSZ R+D K/L VAPM OPSH WMSH AVWAGE RBSI Const R2 F 1978 .252 .953 -.256 .0001 14.360 .1303 .381 .088 -13.19 .50 11.18 LDC (.442) (.941) (-.679) (.228) (2.401)** (3.110)*** (.198) (2.942)*** (-1.650)** ES 1978 .363 .244 -.506 -.0002 -5.479 -.735 .705 .018 11.09 .24 3.50 LDC (.669) (2.538)** (-1.409) (-.490) (-.963) (-l.843)* (.038) (.618) (1.458) 1978 .607 1.510 -1.554 -.00002 -2.987 -.031 .577 .105 20.87 .13 1.68 World (.202) (2.827)*** (.780) (-.011) (-.094) (-.140) (.568) (.666) (.495) 1978 .115 1.499 -2.143 -.0008 -48.43 -.429 .811 .087 52.95 .27 4.22 World (.431) (3.193)*** (-1.224) (-.664) (-1.746)* (-.221) (.908) (.063) (1.424) 1978 .046 .099 -.343 -.0009 -2.098 -.048 -.002 -.040 10.392 .03 .39 Japan (.499) (.606) (-.561) (-1.436) (-.216) (-.713) (-.542) (-.825) (.801) ES 1978 .020 .582 ..172 -.0002 -4.688 .0060 .0003 -.0095 4.763 .15 1.68 Japan (.018) (2.147)** (1.83)* (-2.235)** (-1.698)* (.618) (.432) (-1.35) (2.006)** 1978 .927 .607 -.927 -.0002 -20.312 .272 .003 -.246 12.48 .48 10.34 USA (1.535)* (5.663)*** (-2.318)** (-.431) (-3.207)*** (.613) (1.297) (-.077) (1.475) ES 1978 .0365 .339 -.393 -.003 -.176 -.178 .418 .005 2.688 .24 3.58 USA (.858) (4.430)*** (-1.387) (-1.06) (-.039) (-.564) (.287) (.205) (.488) 1978 .119 -.523 -.170 .00001 .848 -.178 -.660 .556 2.023 .09 1.13 COMECON (.904) (-.253) (-.196) (.145) (.613) (-1.845)* (-1.48) (.803) (1.080) ES 1978 -.108 -.155 -.964 -.0002 -2.12 -.571 .777 .539 .958 .21 3.09 COMECON (-.613) (-.496) (-.082) (-1.63) (1.14) (-.440) (1.303) (.580) (.387) IF 1978 -.698 .487 -.483 .0005 -15.452 .231 .586 -.454 11.943 .17 2.33 EEC (-.457) (1.794)* (-.048) (.565) (-.964) (.211) (1.134) (-.564) (.551) 1978 -.283 .564 -.171 .0008 -19.862 .108 .569 -.013 9.817 .13 2.94 ES EEC (-.231) (2.595)** (-.211) (.901) (-.545) (1.206) (1.377) (-.208) (.572) 1978-70 .079 .138 -.905 -.0010 -2.315 .359 .229 .135 -3.136 .16 2.18 11'LDC+OPEC (1.054) (1.037) (-1.819)* (-1.905)* (-.940) (.651) (.904) (3.413)*** (-.298) 1978-70 .104 .292 -.140 -.000 -.218 .121 .278 .144 -.140 .11 1.44 ESLDC+oPEC (.688) (1.092) (-1.405) (-.195) (-.138) (1.096) (.597) (1.816)* (-.067) 1978-70 .599 .107 -.677 -.0009 6.257 .166 .339 -.124 -9.411 .12 1.57 Japan (1.157) (1.167) (-1.978)**(-2.488)** (1.152) (.436) (1.940)* (-.457) (-1.297) 1978-70 -.321 -.339 .222 -.0003 2.320 .115 -.251 -.184 -.520 .08 .98 Japan (-1.301) (-.071) (1.360) (-1.469) (.896) (.636) (-.300) (-1.421) (-.150) 1978-70 .306 -.898 -.108 -.001 11.567 -.125 -.863 .176 -4.32 .24 3.57 COMECON (1.739)* (-.029) (-.093) (-.877) (.627) (-.967) (-.614) (1.902)* (-.175) 1978-70 .117 -.271 -.180 -.0003 -5.595 -.501 -.143 .588 -9.521 .11 1.42 COMECON (.270) (-.351) (-.628) (-1.025) (-1.229) (-1.572) (2.576)** (2.576)** (-1.565) 1978-70 .008 .013 -.133 -.119 2.575 .0188 .001 -.009 -3.338 .21 3.10 lWorld (.870) (.757) (-2.091)**(-1.755)* (2.555)** (2.662)** (3.022)*** (-1.906)* (-2.478)** 1978-70 -.007 .190 .004 .710 .385 .002 .003 .050 -.970 .11 1.36 ESWorld (-.868) (1.399) (.071) (1.330) (.479) (1.805)* (1.363) (1.235) (-.903) * Significant at 10% confidence level * Significant at 5% confidence level *** Significant at 1% confidence level Note: T-statistic in brackets  - 16 - Table 3: ESTIMATION OF REGRESSION COEFFICIENTS; USE OF ALTERNATIVE MEASURES INCLUDING SKILL AS INDEPENDENT VARIABLES (n = 87) Dependent variables SKILL K/L R+D LAB RBSI Const. R2 F (share of unskilled) IP 1978 .120*** -.375 .09 8.577 LDC (.044) (1.555) .121*** .0410 -.482 .07 4.29 (.0413) (.129) (1.599) s .109*** .001 59.304*** -4.772** .19 7.59 (.039) (.121) (16.464) (1.912) ES 1978 -.0605* 6.743*** .03 3.448 LDC (.0325) (1.235) -.055* .261*** 6.069*** .09 5.337 (.0316) (.099) (1.221) -0.056* .254** 9.876 5.354*** .09 3.715 (.0318) (.097) (13.473) (1.569) R 2 IP 1978 -.241 1.459*** 120.980* 28.479*** .15 4.956 World (.161) (.506) (68.229) (7.972) es 1978 -1.665* 1.702 46.478 35.747*** .17 5.687 World (.841) (3.538) (65.233) (7.578) IP 1978 -.102*** .688*** -1.845 6.605*** .36 15.771 USA (.036) (.114) (15.416) (1.790) ES 1978 -.0422* .318*** 31.338*** 1.279 .38 12.864 USA (.022) (.069) (9.433) (1.075) IP 1978 -.123 1.260 .582** .0683 .11 2.123 EEC (.089) (.871) (.271) (.0804) ID 1978 -.015** -.024 6.554** .680 .10 2.982 COMECON (.007) (.022) (3.033) (1.980) * Significant at 10% confidence level ** Significant at 5% confidence level *** Significant at 1% confidence level  Table 4: ESTIMATION OF REGRESSION COEFFICIENTS, USE OF ALTERNATIVE MEASURES INCLUDING FOREIGN INVESTMENT AS INDEPENDENT VARIABLES (n=87) 2F Dependent Variables FINV CONC; SKILL R+D K/L RBSI Const R F RCA 1 1978 .645*** .227*** -.391*** -.267 -1.144* -.377 10.44* .44 10.49 LDC+OPEC (.196) (.079) (.118) (.367) (.616) (.110) (5.54) RCA 2 1978 1.162*** .241 -.032 -.442 -1.470 -.229 -14.083 .16 2.59 World (.419) (.170) (.252) (.787) (2.531) (.23.6) (11.878) RCA 2 1978 -.116 -.068** .022 -.172 .162 .009 -.111 .11 1.631 USA (.064) (.026) (.038) (.119) (.201) (.036) (1.807) IP 1978 -.136* -.020 -.073* .302** -.268 .105 2.656 .246 4.305 LDC+OPEC (.073) (.02,9) (.044) (.136) (.229) (.136) (2.055) IP 1978 .293 .216* -.281* 1.179 -1.560* .129 26.503** .187 3.028 World (.282) (.115) (.170) (.530) (.892) (.159) (7.995) IP 1978 .145** .462** -.112*** -.563*** -.436** .002 0.762 . .462 11.333 USA (.059) (.188) (.036) (.112) (.188) (.034) (1.693) IP 1978 .318** .104* -.156* .326 .505 -.005 11.267*** .202 3.330 EEC (.146) (.059) (.088) (.276) (.40) (.082) (4.136) IP 1978-70 .128* .037 -.052 .059 -.635*** .165*** .753 .234 4.039 LDC+OPEC (.060) (.028) (.042) (.137) (.221) (.039) (1.985) IP 1978-70 -.5017** -.0016 .006 -.013 .005 -.011** .995*** .119 1.776 World (.0098) (.006) (.000) (.018) (.031) (.005) (.272) * Significant at 10% confidence level. ** Significant at 5% confidence level. *** Significant at 1% confidence level. F Note: Standard error in. brackets. -^  Table 5: CORRELATICN BETWEEN DEPENDENT VARIABLES .-IN REGRESSION ANALYSIS 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 1. IP 1978 1.00 -.08 .32** .13 .26** -.01 .00 .05 -.05 .16 -.08 .04 .11 .24* .03 .01 LDC 2. ES 1978 1.00 .13 .04 .33** .73** .03 -.05 .44** .61** ,57** .35** -.09 .35** .15 .26** LDC 3. IP 1978-70 1.00 .22* .12 LDC+OPE C 4. ES 1978-70 1.00 72** LDC+OPEC 5. IP 1978 1.00 .84** .00 World 6. ES 1978 1.00 .03 World 7. IP 1978-70 1.00 .32** World 8. ES 1978-70 1.00 World 9. IP 1978 1.00 .57** USA 10. ES 1978 1.00 USA 11. IP 1978 1.00 .85** EEC 12. ES 1978 1.00 EEC 13. IP 1978 1.00 -.02 COMECON 14. ES 1978 1.00 COMECON 15. IP 1978 1.00 .26** Japan 16. ES 1978 1.00 Japan Indicates significant at 5, confidence level. * Indicates significant at 1% confidence level.  Table 6: Correlation Coefficients From Matrix of Protection Measures and Explanatory Variables c o o