This action might not be possible to undo. Are you sure you want to continue?
FACTOR ENDOWMENTS AND THE DISTRIBUTION OF INDUSTRIAL PRODUCTION ACROSS THE WORLD
Thiam Hee Ng
Working Paper No 6 February 2002
Statistics and Information Networks Branch of UNIDO
This paper has not been edited. The author is a staff member of the United Nations Industrial Development Organization (UNIDO), Investment Promotion and Institutional Capacity Building Division, Statistics and Information Networks Branch (correspondence to email@example.com). The author thanks Helmut Forstner, Anders Isaksson and Ghislain Robyn for helpful comments and suggestions. The views expressed in the paper are those of the author and not necessarily those of UNIDO or its member states.
Abstract: The trend towards greater integration of the world economy suggests that we would expect manufacturing production to become more concentrated. In this paper, we will use industrial production data from a cross-section of developed and developing countries to analyse the distribution of industrial production. Using standard deviation and Gini index to measure concentration, we find evidence of higher concentration in several industries. However, for most industries, the concentration is relatively unchanged or lower. Traditional theory of location predicts the location of industry based on factor endowments. We conduct regression analysis to determine if factor endowments are significant in explaining the location of industrial production. Our results show that factor endowments are significant in explaining the location for about one-third of the industries.
Keywords: Manufacturing Sector, Distribution of Industrial Production, Factor Endowments. JEL Classification: F15, L60, R30.
1. Introduction In recent years we have witnessed a trend towards growing integration of the world economy. This trend is likely to affect how industrial production is distributed across the world. In an integrated world economy, one would expect companies to shift their production to the country with the lowest cost. The extent to which this takes place will depend partly on the economies of scale of a particular industry. Industries with substantial economies of scale might be expected to be concentrated in a few countries, while industries with less economies of scale will more likely be distributed across several countries. Further, the growing integration of the world economy may also lead to clustering of industrial sectors that share synergies. For example, in the Silicon Valley, the computer industry has clustered together in a single location where access to new technologies is easily accessible and disseminated.
There are both positive and negative implications from such a trend. The growing integration of the world economy is likely to lead to production being concentrated in countries with the lowest cost. As a result, these countries benefit through exploiting economies of scale. Furthermore, the clustering together of firms can lead to greater synergy and collaboration among themselves that would lead to higher innovation and productivity. However, against these benefits, there are adjustment costs that will have to be borne. For example, workers in the sectors that have relocated will have to be retrained. Another drawback is that when one country specialises in a limited number of industries, it makes itself more susceptible to external economic shocks. This point has been highlighted recently by the slump in the technology sector causing sharp reduction in industrial production in Singapore and Taiwan whose industries rely heavily on the technology sector. Hence, the trends towards greater concentration of industrial production could lead to larger fluctuations in the business cycles.
In recent years, new theories of the location of production based on spatial distance and increasing returns to scale have been developed. These theories, known as the
The presence of increasing returns to scale is another factor that pushes production to be concentrated near areas with large markets. One is the availability of data and the other is that manufacturing activity is the only one that can easily be relocated to other countries4. we aim to collect and analyse the data on industrial production across a cross-section of developing and developed countries. which are based on the endowments of resources2. We will do this at two different levels of aggregation. we wish to see how the distribution of industrial production across the world has changed. 3 We will use sector to refer to broad aggregate sector in the economy such as manufacturing. First. see Leamer (1984). we will examine how the distribution of the broad aggregate sectors3 in the economy – agriculture. 4 With the improvement of communications technology. The second objective of this paper is to analyse whether the distribution of factor endowments is significant in explaining the location of industrial production. manufacturing. We have two main objectives for this paper.“new economic geography” 1. In the first part. We will use industry to refer to the ISIC 3-digit level disaggregated manufacturing sector such as food industries etc. These results are in contrast with the traditional theories on the location of industry. In particular. Both these factors suggest that firms tend to agglomerate in regions or locations with large domestic markets. Our decision to examine the manufacturing sector is driven by two factors. larger distances from the market make it more expensive to transport goods and acquire information about the market. 2 . First. Previous studies on the distribution of industry have concentrated almost solely on the developed countries and we feel that this gap should be filled. In the second part. mining and services – are affected by the factor endowments in the economy. mining. services and agriculture. we will examine how factor endowments influence the location of production of each manufacturing industry. larger distances from the suppliers of intermediate goods also make for more costly to obtain inputs for the production process. argue that distance affects production in two ways. 2 For example. We will use detailed industrial sector data for a broad cross-section of countries to examine the trend in world industrial production. certain types of services such as customer support can also be relocated to other countries. Second. 1 See Krugman (1991) and Fujita and Thisse (1996) for a survey of the theories of “new economic geography.
Section 3 will describe the dataset and also provide a summary of the data. There is no change in the number of industries for which factor endowments are significant between 1980 and 1990. Literature Review The trend towards integration in the world economy has prompted the emergence of several studies that examine the distribution of industrial production. Most of these studies were conducted using European data. for most of the industries. manufacturing and services. 2. The next section will present the results of the distribution analysis and regression. factor endowments have low explanatory power for the mining sector. Concluding remarks will be given in Section 5. For the manufacturing sector as a whole.Our results on the distribution of industrial production show that there are certain industries that have become more concentrated between 1980 and 1990. By examining the manufacturing industry shares for each country compared with the EU average. the industries began to diverge since the early 1980s. there is little change in concentration or even a decline. In the next section. they find that although industries in Europe showed signs of convergence in the 1970s. Our regression results for the broad aggregate sectors in the economy show that factor endowments are significant in determining the location of the production of agriculture. Hence. However. we will review some of the literature and empirical results of related studies. However. our results suggest that there has been little change in the distribution of industrial production between 1980 and 1990 despite the globalisation trend. The regression results on individual manufacturing industries show that factor endowments are significant in explaining the location of production in about one-third of the industries. Midelfart-Knarvik et al (2000) examine the trend in the location of industrial production in Europe. Each country’s industrial structure was becoming more dissimilar from that of their European partners. They point out 3 . It will also present the empirical methodology used in our estimation. there is a very slight increase in concentration.
They find that slow-growing. suppose that China is relatively abundant in labour. As a result. Harrigan (1995) uses data from a group of OECD countries to examine the importance of factor endowments to manufacturing. He finds that manufacturing employment has become more concentrated in a few countries. He finds that endowments of capital are a source of comparative advantage for most industries while the effects of endowments of skilled and unskilled labour on comparative advantage are ambiguous. there have been a number of papers that explored the implications of the Heckscher-Ohlin model from the production side. Amiti (1999) using both production and employment data also finds that there has been a trend towards greater concentration in Europe. he does not find manufacturing exports to be more concentrated. For example.the rapid expansion of high-technology industries in Ireland and Finland as one of the dramatic changes in European industry. An increase in labour endowment will then have the effect of increasing the production of labour-intensive goods such as textiles. also using European data. Part of the reason for the ambiguity of the results of skilled labour may be due to the fact that the services sector is more skill-intensive than manufacturing. labour-intensive industries are becoming more concentrated in the peripheral. Traditionally. The Rybczynski theorem shows that as endowment of a given factor increases. empirical work on the Heckscher-Ohlin theory has been carried out using trade data. low-wage economies. He also shows that lowtechnology sectors are the ones that are the most geographically concentrated. This theory can be seen as a combination of the Rybczynski theorem with the assumption of identical preferences. Much of the traditional empirical studies on the location of production are based on the Heckscher-Ohlin theory. However. Combining the results of the Rybcyznski theorem with identical preferences across countries gives us the conclusion that countries will export goods whose production are intensive in the factors that are relatively abundant in that countries. an increase in the endowment of skilled labour will lead to larger services sector and relative contraction 4 . shows that industrial specialization has increased in Europe over the period 1972-1996. recently. Brülhart (2001). the country will increase production of the good that uses that factor intensively. However.
We follow this strand of research in our paper and use data on industrial production in our empirical tests. which can be written as: sik(t) = xik(t) / GDPi(t) 5 . Our sample includes both developed and developing countries. This dataset provides us with data on industrial production for 44 countries for the years 1980 and 1990.1 Data Source and Definition The source of our data for industrial production is the United Nations Industrial Development Organization (UNIDO) Industrial Statistics Database. Harrigan (1997) expands on this result and shows that technological differences are important in explaining the variation of output in a panel of OECD countries. In our empirical analysis.of manufacturing. Harrigan and Zakrajsek (2001) also find an important role for factor endowments in determining the location of production. The list of countries in our sample is given in Appendix 1. we will be using the share of value added of the industry in the country’s GDP. We choose to measure industrial production using value added. The number of industries available for each country differs slightly from country to country since some of the developing countries do not have the full 28 manufacturing industries. We feel that this reflects the true contribution of the country to the manufacturing process. Using a dataset containing a mixture of both OECD and nonOECD countries. The difference in the number of industries available for each country does not affect our analysis since we are interested in the distribution of industries across countries. 3. Our basic unit of analysis for industrial production will be the value added of industry k in country i at time t which we will denote xik(t). Data 3. Using gross output as the measure of industrial production will likely overstate the industrial production of countries that do only minimal value added on imported components.
Finally. In addition to data on industrial production. manufacturing. Data on capital stock is from UNIDO’s own estimates using the perpetual inventory method and a 15% depreciation rate. Unskilled labour is defined as workers without any education. The proportion of unskilled. defined as: rik (t ) = sik (t ) 1 å sik (t ) n i × 100 where sik(t) is the share of sector k value added in country i’s GDP at time t. we use data on the aggregate shares of agriculture. semi-skilled and high-skilled. 3. semi-skilled labour are workers with primary and secondary education. mining and services in the overall economy. semi-skilled and high-skilled labour is then multiplied by the total labour force to obtain the number of unskilled. Using the share of value added allows us to make cross-country comparisons without the difficulties of converting the data to a common currency. A value greater than 100 means that the sector share is larger than the world average while a value less than 100 means 6 . Data for arable land area are obtained from the World Bank World Development Indicators.2 Summary of Data Table 1 presents each country’s value added share relative to the world’s average. Data for the breakdown by sectors in the overall economy are obtained from the World Bank World Development Indicators. The data on total labour force for the country is obtained from the World Bank’s World Development Indicators as well. We present the countries with the highest five and lowest fives values of the relative share for the years 1990 and 1980. we classify labour endowments into three different categories. we also require data on factor endowments. unskilled. using data on educational attainment from Barro and Lee (1993). semi-skilled and high-skilled labour in the economy. A value of 100 means that the sector share in overall GDP is exactly the same as the world’s average. and high-skilled labour are workers with tertiary education.where GDPi(t) is the GDP of country i at time t. For our regression analysis.
Developing countries also tend to specialize in the production of resource-based products such as leather products. Both developed countries and developing countries have substantial production of the industrial chemical and other chemical industries. It is surprising to note that Jamaica 7 . the importance of Hong Kong and Korea is diminishing over time as production of these industries are being moved to lower cost countries such as Mauritius and Tunisia. we find that countries with large endowments of softwood forests dominate the paper products industry. Indonesia. Tunisia has a very high concentration in leather products while Zimbabwe has a strong presence in footwear. Another successful industrialiser. The wood products industry tends to be dominated by countries with substantial endowment of forests such as Malaysia and Indonesia. two Asian “tigers” – Singapore and Korea – hold the top two positions. Looking at total manufacturing. Both of these countries have manufacturing sector shares that are roughly double that of the world average. industrialised so rapidly over the ten-year period that it managed to move out of the bottom five list by 1990. In 1990. beverages and tobacco are mostly concentrated in the developing countries. For a lot of these countries. the growth of the Mauritian wearing apparel industry is spectacular. Glancing through the figures. We also observe the increase of manufacturing share in developing countries such as Chile and Tunisia. In the textiles and wearing apparel industries. we can see that there are substantial differences between the top and bottom countries. these industries may be the only substantial manufacturing entities. the relative shares in individual industries are much more dispersed than the total manufacturing shares. The production of low-technology manufacturing goods such as food products. the furniture industry is dominated by a different group of countries. In fact.that the sector share is smaller than the world average. High values of the relative share indicate that the economy is relatively specialized in that sector. However. the Mauritian wearing apparel industry had a share of GDP ten times as large as the world average. The relative share can also be viewed as an indicator of specialization. Not surprisingly. Similarly. footwear and wood products.
is relatively specialized in the production of industrial chemicals and the Philippines is relatively specialized in other chemicals. several African economies are relatively capital-scarce. We present the relative share of factor endowments for 1980 and 1990 for the top five and bottom five countries in Table 2. We will now turn our attention to the distribution of factor endowments. 8 . Denmark and Sweden all have substantial amounts of capital per worker. Malaysia figures prominently in the production of rubber products probably due to the fact that it is a large rubber producer. Several papers have shown that there is a trend towards greater specialization in industrial production in European countries. The developed countries dominate the top rankings for relative endowments of high-skilled labour. 4. Norway. with a relative share ten times the world average. We are interested in seeing whether this trend is also visible in a broader cross-section of countries. The poor African economies have substantial endowments of unskilled labour while in the developed world the endowments of unskilled labour is negligible. The distribution of endowments was relatively stable over the ten-year period. respectively.1 Distribution of Industrial Production One of the aims of the paper is to examine the change in the distribution between countries of industrial production over the period 1980 to 1990. The Philippines is the only developing country among the top five in that list. both Korea and Malaysia are becoming increasingly specialized in this industry. This is most likely due to the large copper industry. Chile is highly specialized in the production of non-ferrous metals. Empirical Results 4. Since education takes a long time. There is a mix of developing countries and developed countries with substantial endowments of semi-skilled labour. The Nordic countries are highly capital intensive. Finland. Regarding the electrical machinery industry. Singapore is becoming much more specialized in the non-electrical machinery industry. we see little change in the relative endowments of the skilled-labour force over the period of ten years. Unsurprisingly. China and India have the two largest labour endowments amounting to 26 times and 14 times the world average.
We also calculate the Gini index. Several manufacturing industries show an increase in the standard deviation while others show a decline. for each of the manufacturing industries. a low value of the standard deviations suggests that the distribution of that industry is more dispersed with production spread out over many countries. beverages and tobacco have lower concentration than most other industries. Manufacturing industries whose concentration declined include professional equipment. Chart 1 and Chart 2 present the standard deviation of the relative share for each of the 28 manufacturing industries and for the manufacturing sector as a whole. These are relatively basic industries that usually cater to local tastes.We will use standard deviation and the Gini index as our measure of concentration. industrial chemicals. On the other hand. petroleum refining. so it is not surprising to see that their production is quite evenly distributed. This is probably due to a few countries. and paper products. and non-electrical machinery. A quick look at the charts shows us that there is no uniform trend. plastic products. that specialize heavily in producing wearing apparel for exports. the standard deviation remains roughly the same. While textiles production seems to be quite evenly distributed. A high value of the standard deviations implies that the production of that industry is concentrated in only a few countries. For quite a number of industries. The distribution of total manufacturing is quite even with a relatively low level of standard deviation. notably Mauritius. fabricated metal products. petroleum and coal products. Hence. Similarly. iron and steel. pottery and china. another measure of concentration. Low-technology industries such as food. Other industries that have shown substantial increases in concentration include footwear. The Gini index has a minimum value of zero when the 9 . the wearing apparel industry is quite concentrated. there is a high level of concentration in leather products. There is little change in the value of standard deviation over the period. an increase in the standard deviation implies that there is a trend towards greater concentration or specialization in that industry. which also underwent a large increase in concentration over the period.
which will be called “Concentrated to Concentrated”. The second group. the value of the Gini index increases up to a maximum value of one. we observe no clear trend towards greater concentration. This is consistent with the earlier results using standard deviations. We perform a similar classification exercise for the nine least concentrated industries in 1980 where we divide them again into a group made up of industries that remain as one of the bottom nine least concentrated industries in 1990 (“Dispersed to Dispersed”) and another group consisting of industries that did not remain in the bottom nine (“Dispersed to 5 The ranking of industries by concentration is based on the Gini index. First we take the nine industries that were most concentrated5 in 1980 as and divide them into two groups. As the distribution of industrial production becomes more concentrated. Chart 3 and Chart 4 present the value of the Gini indices for the various manufacturing industries for the years 1980 and 1990. called “Concentrated to Dispersed”. consists of the industries that remained as one of the top nine most concentrated industries in 1990. An increase in the Gini index implies an increase in concentration while a decrease in the Gini index implies a decrease in concentration. Hence. The first group. We find that manufacturing as a whole is more evenly distributed than are the individual industries. Several industries have become much more concentrated but most others remained roughly the same or have become less concentrated. there are several industries whose concentration increased over the period while others decreased. 10 .distribution is perfectly even. is made up of industries that were no longer in the top nine most concentrated industries in 1990. This result more or less matches what we see using the standard deviation measure of distribution. This means that all countries have the same relative share in their manufacturing sector. In order to facilitate analysis we will divide the 28 manufacturing industries into five groups following the typology of MidelfartKnarvik et al (2000). Overall. As before. In this extreme case. the Gini index can be interpreted in the same way as the standard deviation measure. We will also attempt to analyse and identify the characteristics of industries whose concentration changed over the period. one country has all the production of the industry in question.
which classifies industries based on their technology intensity. Finally. Then we calculate the unweighted average growth rate for each manufacturing industry and rank them. This is not surprising. the middle 12 are called “Medium” and the bottom 12 called “Low”. We will attempt to characterise the industries by using three indicators – the degree of economies of scale. The 12 industries with the highest degree of economies of scale are classified as “High”. The measure for the technology level is from OECD (1997). Similarly. we would also expect industries with a high technology level to have higher barriers to entry and hence likely to be concentrated in only a few countries.Concentrated”). we calculate the growth rate in terms of value added in US dollars for each manufacturing industry in each country over the period 1980-1990. we divide the industries into three categories. 6 First. It is also interesting to note that these industries are relatively fastgrowing with four of them classified as fast-growers and two as medium-growers. The indicator for the degree of economies of scale is obtained from Pratten (1988). 11 . We classify as “High” the OECD categories of ‘high technology’ and ‘medium-high technology’. We first examine the group of industries called “Concentrated to Concentrated”. the level of technology and the growth of the industry. we will classify the nine manufacturing industries with the highest growth6 as “High”. The final group known as the “Residual” consists of industries that do not meet the above criteria. the next ten as “Medium” and the last nine as “Low”. with only one exception. since we would expect the production of industries with a high degree of economies of scale to be concentrated in a few locations in order to minimize production costs. where the minimum efficient scale for 36 industries is given. High. as “Medium” the OECD category of ‘medium-low technology’ and as “Low” for the OECD category of “low technology”. Medium and Low. We then map the 36 industries into the 28 industrial sectors used in this paper. For each indicator. What we are probably witnessing in this case is that wearing apparel production is being concentrated in only a few countries with low labour cost. The wearing apparel industry is the only exception to the high degree of economies of scale or high technology level. The first observation is that this group of industries has either a high degree of economies of scale or a high technology level. We divide those industries into three groups.
They are also relatively slow-growing industries. technology levels and growth rates. and petroleum and coal products industries. The “Dispersed to Concentrated” group is also characterized by low or medium-level economies of scale and technology. technology level and growth rate. which has a high degree of economies of scale and medium-level technology. 4. while the geographically more dispersed industries have lower economies of scale. we will perform regression analyses on the aggregate share of sectors in the economy and on individual manufacturing industries. Summarizing the results. We also find that there are substantial variations in concentration across the various industries. In order to do this. we find that – by and large – the geographically concentrated industries have higher degrees of economies of scale. beverages. However. 12 . The growth rate for this group ranges from high to low. There are several industries for which production has become less concentrated over the period. The industries that are in the “Dispersed to Dispersed” group are characterised by low or medium levels of economies of scale and technology with one exception. The group called the “Residual” is a diverse group without any dominant characteristics. This group consists of the wood products. In terms of industry characteristics. we can say that both the standard deviation and the Gini coefficient results show that certain industries have become more concentrated. The exception is petroleum and coal products. Two out of the three industries are low-technology and have a low degree of economies of scale. the “other chemical” industries.2 Regression Results In this section we aim to find out about the importance of factor endowments in explaining the location of manufacturing production.We next turn to the group of industries that went from “Concentrated to Dispersed”. there is little evidence that the manufacturing sector as a whole is becoming much more concentrated.
Therefore.Our empirical framework is based on that developed by Harrigan and Zakrajsek (2001). Taking into account that the matrices B3 and B4 are symmetric and that the revenue function is homogeneous of degree one. data on relative prices across industries for a cross-section of countries are not available. v ) = b 0 + b1 ln( p) + b 2 ln( v ) + 1 1 ln( p)B 3 ln( p) + ln( v )B 4 ln( v ) + ln( p)B 5 ln( v ) 2 2 where r is the revenue function. The factor endowments we use for this regression are the share of semi-skilled labour in total labour force. p is a vector of final goods prices and v is a vector of endowments. We feel that manufacturing goods fit this assumption relatively well. However. Assume that the production side of the economy can be described by a translog revenue function: ln r (p. First we regress aggregate sector shares of the economy– agriculture. This assumption is plausible if we believe that all goods are tradable and each country is small and faces an exogenous relative goods price. pik(t) is the relative price of industry k in country i at time t and vim(t) is the endowment of factor m in country i at time t. the share of skilled labour in total labour 13 . manufacturing. we assume that relative prices are uncorrelated with factor endowments. mining and services – on factor endowments for 1990 and 1980. we can differentiate the above revenue function with respect to prices to obtain: sik (t ) = β 0 + β1 å ln( pik (t )) +β 2 å ln(vim (t )) k m where sik(t) is the sector share of industry k in country i’s GDP at time t. With the above assumption we can proceed to estimate the following equation using OLS: sik (t ) = β 0 + β1 å ln(vim (t )) + ε ik (t ) m We will perform the above regression both at the broad economic sector level and at the manufacturing industry level.
we find that both capital and high-skilled labour are a strong positive factor for the share of services in GDP in 1990. we find that increases in semi-skilled labour raise the share of manufacturing in GDP in 1990. Table 6 and Table 7 present the results of the regressions for individual manufacturing industries and total manufacturing. for 1980. Semi-skilled labour. Factor endowment is less important for the manufacturing sector and has become less important over the years. The changing nature of the services industry where more technology is being used now probably leads to the growing importance of high-skilled workers. the importance of factor endowments has also become stronger as evidenced by the higher R-square. 14 . has a negative coefficient in 1980 while the positive coefficient for high-skilled labour is no longer significant. Another way to view it is that a positive coefficient for the factor means that the factor is a source of comparative advantage while a negative coefficient for the factor means that the factor is a source of comparative disadvantage. we find that higher high-skilled labour leads to a lower mining share while higher capital leads to a higher mining share. This is in contrast to 1980 when an increase in high-skilled labour contributed to a higher share of manufacturing. In short. we find that factor endowments are a strong determinant for the agriculture sector. For the manufacturing sector. Table 4 and Table 5 present the results from this regression. Due to the large number of coefficients. only capital leads to a lower share of agriculture. For 1990. Over time. A positive coefficient for a given factor means that an increase in that factor will lead to a higher share of the sector in question while a negative coefficient means that it will lead to a lower share. of skilled labour. We find that increases in semi-skilled labour. The coefficients presented in the tables are related to the Rybczynski derivatives. high-skilled labour and capital all lead to a lower share of agriculture in 1990. Interestingly.force7 and capital per worker. however. However. none of the endowments are significant in explaining the share of mining. In 1980. we have 7 We cannot include the share of unskilled labour in the regression since it is linearly dependent on the shares of semi-skilled and. The mining sector does not seem to be affected much by factor endowment while factor endowments are becoming more important for the services sector.
However. electrical machinery and transport equipment. Capital is significant for the printing and publishing. beverages. electrical machinery and non-electrical machinery. we find that out of the 29 industries. These are all relatively capital-intensive. We also find that endowment of semi-skilled labour is becoming more important in explaining the location of low-technology manufacturing industries.decided to present just the summary of the results. High-skilled labour has limited importance in explaining the location of manufacturing production. Capital is significant for several capitalintensive industries such as fabricated metal products. non-electrical machinery. we find that 10 out of the 29 industries have significant coefficients. fabricated metal products and professional equipment industries. 11 have significant coefficients. For rubber products and tobacco industries. we did not find significant coefficients for other capital-intensive industries such as petroleum refining. high-skilled labour is not significant even for high-technology industry such as electrical machinery. This means that as a country accumulates capital. the share of these manufacturing industries will decline. There is also substantial change in the importance of factor endowments between 1980 and 1990. the results of the regressions show that factor endowments are important only for about one-third of the manufacturing industries. This could be due to a trend of moving production of low-technology goods from rich countries to developing countries. increases in semi-skilled labour increases the industry’s share. the importance of capital disappears over time. Several low-technology industries such as food products. Overall. The endowment of semi-skilled labour is also significant in several of the low-technology industries. This allows us to see the factors that are significant in explaining the share of a given industry in GDP. 15 . Turning to the results of the regression for 1980. tobacco and footwear have significantly negative coefficients for capital. We find that for some of the capital-intensive industries such as transport equipment. Interestingly. The tables show the signs of the coefficients that are significant at the 5% level. Looking at the result for 1990.
Conclusions We find that the distribution of industrial production in our cross-section of countries has shown limited change over time. We also find that factor endowments are significant in explaining the share of agriculture. However. We also find that the set of industries for which factor endowments are significant changes between the two periods. A few industries have become more concentrated but for most of them concentration remained the same or even declined. Results for the individual industries show that factor endowments are significant in only one-third of the industries in both 1980 and 1990. we find that factor endowment is not significant in explaining the share of mining in the economy. manufacturing and services in the overall economy. 16 .5.
“Evolving Geographical Concentration of European Manufacturing Industries”. 137. (1991). and International Specialization: Estimating the Neoclassical Model. M.2: Studies on the economics of integration. “The Location of European Industry. OECD (1997). Geography and Trade. 1970-86. Sources of International Comparative Advantage. (1988). E. Redding and A. Science. (1997). (2001). “Factor Endowments and the International Location of Production: Econometric Evidence for the OECD.” American Economic Review. and E. Barro. 4349. Harrigan. M. Weltwirtschaftliches Archiv. Leamer.References Amiti. Zakrajsek (1999). S. and J-F Thisse (1996). (1999). Brülhart. mimeo. (1984).H. and W.” NBER Working Paper. “International Comparisons of Educational Attainment. J.” Centre for Economic Policy Research Discussion Paper No. (1995). “ Factor Supplies and Specialization in the World Economy. Cambridge. No. P..J. 573-593.J.J. Paris. Technology and Industry: Scoreboard of Indicators 1997. J. Massachusetts. “Economics of Agglomeration.” Journal of International Economics. 39. J.G. 7848. “A Survey of the Economies of Scale. No. 87. European Commission. MIT Press. “Technology. 1344. 123-41. M. 17 .” Weltwirtschaftliches Archiv. OECD. 215-243. Pratten. Factor Supplies. Lee (1993). Fujita. K. R. Midelfart-Knarvik. Harrigan.” NBER Working Paper. Venables (2000). C. 475-94. “Specialization Patterns in Europe.” in Commission of the European Communities: Research on the “cost of non-Europe” vol.” Report prepared for the Directorate General for Economic and Financial Affairs. Massachusetts. H. Overman. Harrigan. 135. Krugman. Cambridge. MIT Press.
Appendix 1 List of Countries in the Sample Algeria Australia Austria Bangladesh Belgium Canada Chile Colombia Costa Cyprus Denmark Ecuador Egypt Gambia Greece Hong Kong India Indonesia Iran Jamaica Jordan Kenya Korea Malawi Malaysia Mauritius Netherlands New Zealand Pakistan Peru Philippines Senegal Singapore South Africa Sri Lanka Sweden Tunisia Turkey United Kingdom Cameroon United States Venezuela Zambia Zimbabwe 18 .
Chart 1: Standard Deviation of Relative Share by Manufacturing Sector Petroleum refineries Other chemicals Industrial chemicals Printing and publishing Paper products Furniture Wood products Footwear Leather products Wearing apparel Textiles Tobacco Beverages Food products Total manufacturing 0 50 100 150 200 1990 1980 Chart 2: Standard Deviation of Relative Share by Manufacturing Sector Other manufactured products Professional equipment Transport equipment Electrical machinery Non-electrical Machinery Fabricated metal products Non-ferrous metals Iron and steel Non-metallic mineral products Glass products Pottery and china Plastic products Rubber products Petroleum and coal products 0 50 100 1990 1980 150 200 Note: The charts show the standard deviation of the relative value added share for each industry. 19 .
7 Chart 4: Gini Index for Distribution of Industries Other manufactured products Professional equipment Transport equipment Electrical machinery Non-electrical Machinery Fabricated metal products Non-ferrous metals Iron and steel Non-metallic mineral products Glass products Pottery and china Plastic products Rubber products 0 0.2 0.6 0.2 0.1 0.3 1990 0.3 1990 0.5 0. 20 .7 Note: The charts show the Gini coefficient for the relative value added share for each industry.6 0.4 1980 0.4 1980 0.Chart 3: Gini index for Distribution of Industries Petroleum and coal Petroleum refineries Other chemicals Industrial chemicals Printing and Paper products Furniture Wood products Footwear Leather products Wearing apparel Textiles Tobacco Beverages Food products Total manufacturing 0 0.1 0.5 0.
4 166.3 187.6 313.5 43.1 8.1 Indonesia Bangladesh Gambia India Sri Lanka Singapore Korea United Kingdom Peru United States 1980 25.1 30.1 9.9 182.7 247.1 182.3 1990 1.6 362. 21 .8 329.Table 1: Distribution of Industrial Production Across Countries Total Manufacturing Lowest Gambia Bangladesh Malawi Cameroon Senegal Highest Korea Singapore Chile Peru Tunisia Food Products Lowest Hong Kong Malawi India Iran Islamic Republic Mexico Highest Philippines Chile Jamaica Costa Rica Zambia Beverages Lowest Bangladesh India Egypt Indonesia Iran Islamic Republic Highest Zimbabwe Zambia Jamaica Mauritius Cameroon 1990 30.3 24.9 27.5 20.0 215.7 165. A value of more than 100 means that the industry share in GDP is higher than the cross-country average and a value of less than 100 means that the industry share in GDP is less than the average.2 234.2 Bangladesh India Indonesia Egypt Iran Islamic Republic Zambia Colombia Cameroon Jamaica Peru 1980 4.0 36.1 273.3 6.8 22.3 16.0 212.8 203.2 31.6 185.8 199.7 37.9 165.4 47.8 159.7 Note: This table shows the industry value added share relative to GDP for the countries divided by the crosscountry mean and multiplied by 100.4 213.0 203.2 45.0 Bangladesh Indonesia India Mexico Hong Kong New Zealand Malawi Costa Rica Peru Senegal 1980 18.7 244.1 34.2 7.0 27.9 1990 22.4 177.4 267.0 34.9 204. The highest five and the lowest five values for each industry is shown in the table.4 230.6 165.9 365.3 8.0 16.4 175.3 194.2 34.6 10.6 23.6 561.
6 4.5 260.9 284.1 29.6 41.1 269.1 24.7 238.1 368.9 9.6 277.6 Jordan Mexico Netherlands Sweden Denmark Korea Hong Kong Egypt Peru Colombia 1980 22.3 16.0 23.9 292.9 386.2 Sweden Egypt India New Zealand South Africa Jamaica Korea Sri Lanka Cyprus Jordan 1980 13.1 27.0 28.7 235.4 8.5 11. The highest five and the lowest five values for each industry is shown in the table.5 31.0 2.1 15.5 Note: This table shows the industry value added share relative to GDP for the countries divided by the crosscountry mean and multiplied by 100.6 22.2 417.5 219.4 347.2 13.1 221.5 5.3 19. 22 .4 Bangladesh Indonesia Egypt Pakistan India Hong Kong Mauritius Cyprus Korea Tunisia 1980 0.9 21.0 950.4 235.5 207.4 4. A value of more than 100 means that the industry share in GDP is higher than the cross-country average and a value of less than 100 means that the industry share in GDP is less than the average.8 394.8 329.2 1990 0.3 280.Table 1: Distribution of Industrial Production Across Countries (continued) Tobacco Lowest Ecuador South Africa Iran Islamic Republic New Zealand Sweden Highest Jordan Sri Lanka Tunisia Indonesia Jamaica Textiles Lowest Singapore Senegal Sweden Mexico Netherlands Highest Zimbabwe Pakistan Korea Peru Hong Kong Wearing Apparel Lowest Senegal Netherlands Sweden Iran Islamic Republic Ecuador Highest Mauritius Tunisia Hong Kong Cyprus Sri Lanka 1990 2.3 344.7 247.5 392.3 162.4 219.1 221.7 17.3 305.0 1036.6 31.5 346.1 12.0 1990 18.7 190.5 19.5 274.
6 10.9 Note: This table shows the industry value added share relative to GDP for the countries divided by the crosscountry mean and multiplied by 100.4 1990 2.8 16.6 13.2 12.9 185.2 166.7 5.6 357.4 9.7 18.3 15.7 327.8 218.3 20.5 575.8 11.9 9.1 1990 6.5 183.8 329.9 19.8 205.Table 1: Distribution of Industrial Production Across Countries (continued) Leather Products Lowest Egypt Malaysia Denmark Sweden Netherlands Highest Tunisia Korea Algeria Mauritius Cyprus Footwear Lowest Malaysia Sweden Netherlands Singapore India Highest Zimbabwe Cyprus Chile South Africa Sri Lanka Wood Products Lowest Senegal Mexico India Egypt Pakistan Highest Tunisia Malaysia Indonesia Sweden Chile 1990 13.8 392.2 468.5 211.2 329.0 803.6 22.8 12.2 7.2 14. The highest five and the lowest five values for each industry is shown in the table.1 339.7 181.7 357.4 274.7 4.3 257.6 205.7 Bangladesh Pakistan India Mexico Egypt Malaysia Sweden New Zealand Canada Cyprus 1980 3. 23 .0 363.6 477.2 29.7 8.2 Pakistan Bangladesh India Indonesia Philippines Cyprus Zimbabwe Chile Jamaica Venezuela 1980 11.8 270.6 288.8 302.8 566.8 25.7 17.4 223.8 197.6 Indonesia Malaysia Egypt Philippines Sri Lanka Peru Cyprus Korea New Zealand Pakistan 1980 7.3 240.5 260. A value of more than 100 means that the industry share in GDP is higher than the cross-country average and a value of less than 100 means that the industry share in GDP is less than the average.0 26.5 24.2 12.
3 217.4 278.3 254.0 25.3 11.2 27. The highest five and the lowest five values for each industry is shown in the table.4 345.5 1990 7.3 10.2 15.6 197.0 323. A value of more than 100 means that the industry share in GDP is higher than the cross-country average and a value of less than 100 means that the industry share in GDP is less than the average.6 200.4 271.9 India Bangladesh Indonesia Pakistan Sri Lanka Belgium Cyprus Austria Jamaica Costa Rica 1980 1.8 259.7 366.9 189.3 25.5 7.5 2.8 345.3 Indonesia Mauritius Bangladesh Pakistan Iran Islamic Republic Canada Sweden Chile New Zealand United States 1980 10.5 18.4 288.5 178.8 4.4 189.7 19.1 30.9 241.5 22.2 203.4 252. 24 .2 218.2 14.1 1.8 2.9 196.8 188.3 13.7 211.6 16.4 24.2 17.3 Note: This table shows the industry value added share relative to GDP for the countries divided by the crosscountry mean and multiplied by 100.8 5.3 314.8 243.7 4.0 26.7 210.2 394.2 1990 14.4 213.8 280.2 251.9 8.Table 1: Distribution of Industrial Production Across Countries (continued) Furniture Lowest India Senegal Sri Lanka Pakistan Cameroon Highest Belgium Cyprus Austria Jamaica Denmark Paper Products Lowest Senegal Cameroon Iran Islamic Republic Mauritius Pakistan Highest Chile Sweden Canada New Zealand South Africa Printing And Publishing Lowest Cameroon Bangladesh India Iran Islamic Republic Indonesia Highest United Kingdom United States Singapore Canada Sweden 1990 1.1 Bangladesh Indonesia Iran Islamic Republic Sri Lanka Pakistan United Kingdom United States Sweden Netherlands New Zealand 1980 5.9 215.
8 211.8 Note: This table shows the industry value added share relative to GDP for the countries divided by the crosscountry mean and multiplied by 100.4 6.1 1990 17.7 180.5 21.5 20.3 12.3 220.9 46.3 6.4 576.Table 1: Distribution of Industrial Production Across Countries (continued) Industrial Chemicals Lowest Cyprus Hong Kong Sri Lanka Cameroon Ecuador Highest Jamaica Belgium Netherlands Korea Singapore Other Chemicals Lowest Cameroon Hong Kong Sri Lanka Senegal Mauritius Highest Korea Chile Philippines Singapore Peru Petroleum Refining Lowest Iran Islamic Republic Bangladesh Denmark Kenya Cyprus Highest Venezuela Tunisia Peru Jamaica Ecuador 1990 6.4 Bangladesh Denmark Austria New Zealand India Singapore Venezuela Colombia Jamaica Turkey 1980 1.6 21.7 39. The highest five and the lowest five values for each industry is shown in the table.5 12.6 218.0 183.6 230.6 206.5 1990 1.0 298.5 270.7 218. 25 .0 15.7 251.2 8.8 Algeria Iran Islamic Republic Cyprus Malawi Hong Kong Belgium Korea Jamaica United Kingdom United States 1980 6.4 188.3 174.3 659.3 2.8 228.9 8.2 203.1 418.2 271.8 282.7 22.8 294.3 Senegal Cameroon Hong Kong Iran Islamic Republic Indonesia Korea Peru Chile United Kingdom Singapore 1980 24.7 16.3 193.9 275.0 35.1 225.1 10.7 220.2 163.2 46.7 10.8 713.2 37.9 666.4 15. A value of more than 100 means that the industry share in GDP is higher than the cross-country average and a value of less than 100 means that the industry share in GDP is less than the average.0 264.9 25.0 238.5 48.3 20.
0 238.6 321.4 11.7 8.1 3.6 Iran Islamic Republic Costa Rica Philippines Bangladesh Malaysia Korea Egypt Turkey South Africa United Kingdom 1980 2. The highest five and the lowest five values for each industry is shown in the table.8 411.6 488.1 192.3 363.0 8.7 506.6 275.9 171.8 15.5 Jordan Bangladesh Egypt Netherlands Malawi Malaysia Korea Zambia Zimbabwe United Kingdom 1980 2.7 208.7 191.7 174.8 179. A value of more than 100 means that the industry share in GDP is higher than the cross-country average and a value of less than 100 means that the industry share in GDP is less than the average.7 500.1 31.0 13.7 27.1 347.0 485.5 542.1 15.3 Note: This table shows the industry value added share relative to GDP for the countries divided by the crosscountry mean and multiplied by 100.0 206.8 187.5 19.8 16.2 233.2 30.1 23.2 185.1 7.0 28.8 15.9 157.5 560. 26 .8 12.3 219.3 1990 6.5 1990 10.8 174.2 35.5 7.8 164.1 Bangladesh Indonesia India Pakistan Sri Lanka Hong Kong Singapore United Kingdom Korea Venezuela 1980 0.4 151.4 10.4 8.1 512.1 10.0 24.3 443.Table 1: Distribution of Industrial Production Across Countries (continued) Petroleum And Coal Products Lowest Costa Rica Bangladesh Philippines Australia Hong Kong Highest Turkey Chile Korea South Africa Denmark Rubber Products Lowest Egypt Bangladesh Hong Kong Jordan Cyprus Highest Korea Malaysia Sri Lanka Zambia Zimbabwe Plastic Products Lowest Bangladesh Pakistan India Sri Lanka Egypt Highest Korea Hong Kong Singapore United Kingdom United States 1990 2.2 28.0 146.
5 527.6 199.9 280.7 22.8 5.6 271.6 17.5 226.3 222.7 31.0 Note: This table shows the industry value added share relative to GDP for the countries divided by the crosscountry mean and multiplied by 100.0 7.5 5.7 14.7 266.7 27.7 25.7 783.2 18.9 198.6 19.1 284.8 Bangladesh Cameroon Hong Kong Philippines India Jordan Tunisia Cyprus Korea United Kingdom 1980 10.5 29.6 331.6 214.2 Kenya Bangladesh Indonesia Cyprus Canada Jamaica United Kingdom Korea Tunisia Turkey 1980 11.9 207.2 43.1 26.9 209.1 348.7 138.3 666.8 330.6 213.3 643.7 12. The highest five and the lowest five values for each industry is shown in the table.9 232.8 322.0 10.4 17.5 252.6 365. A value of more than 100 means that the industry share in GDP is higher than the cross-country average and a value of less than 100 means that the industry share in GDP is less than the average.Table 1: Distribution of Industrial Production Across Countries (continued) Pottery And China Lowest Mauritius Hong Kong Kenya Canada India Highest Algeria Jamaica Belgium Tunisia New Zealand Glass Products Lowest Mauritius Bangladesh Cyprus Hong Kong India Highest Korea Turkey Austria Colombia South Africa Non-Metallic Mineral Products Lowest Malawi Bangladesh Hong Kong Cameroon Sri Lanka Highest Jordan Tunisia Korea Cyprus Zambia 1990 3.9 21.4 169.7 15. 27 .8 140.8 222.0 32.5 211.4 1990 6.8 27.1 7.1 Cyprus Mauritius Bangladesh Hong Kong India Korea Austria Peru Mexico United Kingdom 1980 2.1 19.3 233.1 18.8 1990 13.3 359.
3 199.4 22.1 Sri Lanka Hong Kong Indonesia Kenya Bangladesh Zimbabwe South Africa Korea Jamaica Belgium 1980 10.9 15.9 18.0 1062.6 680.4 165.3 Bangladesh Indonesia Sri Lanka Pakistan Cameroon Hong Kong South Africa Zimbabwe Sweden United States 1980 4.6 18.4 21.6 11.5 161.2 377.4 349.0 0.2 Pakistan Costa Rica Ecuador Zambia India Chile Peru Australia Canada South Africa 1980 1.4 17.2 20.9 506.3 12.8 181.1 235.0 660. The highest five and the lowest five values for each industry is shown in the table.4 5.8 10.0 21.5 6.5 281.0 20.6 23.5 13.6 278.6 205.0 1990 7. A value of more than 100 means that the industry share in GDP is higher than the cross-country average and a value of less than 100 means that the industry share in GDP is less than the average.8 176.2 27.4 771.6 180.Table 1: Distribution of Industrial Production Across Countries (continued) Iron And Steel Lowest Hong Kong Sri Lanka Bangladesh Kenya Mauritius Highest Algeria Korea Jamaica Zimbabwe South Africa Non-Ferrous Metal Lowest Bangladesh Pakistan Costa Rica Ecuador Zambia Highest Chile Peru Venezuela Australia Tunisia Farbricated Metal Products Lowest Bangladesh Pakistan Sri Lanka India Senegal Highest Belgium Korea Singapore Sweden Zimbabwe 1990 7.1 318.6 14.2 10.4 Note: This table shows the industry value added share relative to GDP for the countries divided by the crosscountry mean and multiplied by 100.6 3.0 216. 28 .7 206.8 210.1 262.0 221.9 1990 0.7 235.5 283.0 4.0 292.6 10.7 17.3 191.7 993.9 334.7 195.
1 441.6 3.2 5.6 257.5 240.4 8.4 228. A value of more than 100 means that the industry share in GDP is higher than the cross-country average and a value of less than 100 means that the industry share in GDP is less than the average.1 273.8 411.4 5.9 6.0 7.3 740.0 9.6 179.8 10.4 7.8 336.2 Senegal Cameroon Jordan Bangladesh Cyprus Singapore Hong Kong Korea United Kingdom United States 1980 3.5 2.0 6.1 216.6 297.5 453.2 15.5 285.0 644.9 Bangladesh Tunisia Ecuador Indonesia Kenya United Kingdom United States Sweden Singapore Denmark 1980 2.4 536.9 262.2 325.3 317.7 243.5 18. The highest five and the lowest five values for each industry is shown in the table.0 Note: This table shows the industry value added share relative to GDP for the countries divided by the crosscountry mean and multiplied by 100.1 307.9 269.2 1990 4.9 1990 2.8 3.8 7.0 351.7 804.0 19.1 265.8 16.0 8.2 247.6 17.Table 1: Distribution of Industrial Production Across Countries (continued) Non-Electrical Machinery Lowest Bangladesh Ecuador Senegal Kenya Malawi Highest Singapore Korea United Kingdom Sweden United States Electrical Machinery Lowest Senegal Cameroon Sri Lanka Cyprus Bangladesh Highest Singapore Korea Malaysia Austria United Kingdom Transport Equipment Lowest Jordan Cameroon Cyprus Mauritius Bangladesh Highest Korea United Kingdom Sweden United States Singapore 1990 2.5 263.9 19.3 3.0 2.8 4.8 Jordan Cameroon Bangladesh Sri Lanka Senegal Singapore United Kingdom United States Sweden Canada 1980 0.8 192. 29 .2 447.7 288.4 309.8 17.
3 617.Table 1: Distribution of Industrial Production Across Countries (continued) Professional Equipment Lowest Bangladesh Cyprus Sri Lanka Indonesia Greece Highest United States Hong Kong Singapore Mauritius Korea Other Manufactured Products Lowest Senegal Egypt India Iran Islamic Republic Ecuador Highest Belgium Korea Mauritius Hong Kong South Africa 1990 0.0 20.1 439.7 315.5 729.4 14.3 299.9 8.2 245.6 Note: This table shows the industry value added share relative to GDP for the countries divided by the crosscountry mean and multiplied by 100.0 325. 30 .8 9.9 201.7 259.2 14.8 274.6 5. The highest five and the lowest five values for each industry is shown in the table.1 Egypt Iran Islamic Republic Indonesia Sri Lanka Turkey Highest Hong Kong Korea Singapore South Africa Belgium 1980 3.1 10. A value of more than 100 means that the industry share in GDP is higher than the cross-country average and a value of less than 100 means that the industry share in GDP is less than the average.4 409.0 452.3 567.4 8.2 Bangladesh Indonesia Turkey Sri Lanka Philippines Hong Kong United States Singapore United Kingdom Denmark 1980 0.7 361.5 328.9 16.0 8.5 320.4 3.3 13.6 1.1 1.8 280.0 210.0 5.3 820.3 252.8 7.9 1990 1.
8 2599.2 300.3 Iceland Guyana Cyprus Gambia Mauritius China India United States Indonesia Japan 1980 0.5 253.3 1.3 294.2 245.2 1.6 2.0 0.6 3.8 348.7 333.0 2.0 0.1 289.5 376.2 1.8 508.1 Uganda Bangladesh Rwanda China Gambia Norway Finland Denmark Netherlands Japan 1980 1.4 3.7 1.0 3.3 1990 1.7 2568.3 1384.0 239.5 404.7 3.4 2.5 292.2 351.6 1.6 31 .0 348.0 0.8 3.4 2.5 440.0 1439.Table 2: Distribution of Factor Endowments Across Countries Total Labour Force Lowest Iceland Guyana Cyprus Mauritius Gambia Highest China India United States Indonesia Japan Capital Per Worker Lowest Uganda Malawi Sierra Leone Bangladesh Mali Highest Norway Finland Japan Denmark Sweden Share Of Unskilled Labour Lowest Denmark New Zealand Finland Japan Canada Highest Mali Gambia Benin Sierra Leone Bangladesh 1990 0.0 519.6 1.5 2.5 522.6 280.4 2.4 2.0 1.2 274.6 1.8 484.4 1.8 334.0 275.7 Austria Japan Australia United States France Gambia Mali Benin Sierra Leone Pakistan 1980 0.4 249.0 304.2 272.0 484.7 1990 0.
0 16.8 511.4 224.9 44.6 140.4 613.1 Gambia Malawi Rwanda Uganda Mali Canada New Zealand United States Australia Philippines 1980 1. The highest five and the lowest five values for each factor is shown in the table.6 57.4 3.8 25. A value of more than 100 means that the factor endowment is higher than the cross-country average and a value of less than 100 means that the factor endowment is less than the average.4 3.6 136.5 151.9 152.3 36.7 143.0 134.9 503.8 4.5 269.0 133.8 551.2 1990 1.6 351.6 484.2 3.1 145.3 35.4 Note: This table shows the value of factor endowment for the countries divided by the cross-country mean and multiplied by 100.7 3.5 38.4 3.8 279.5 150.8 4.9 Gambia Mali Benin Sierra Leone Pakistan Austria Trinidad And Tobago Jamaica Guyana France 1980 16.3 431.Table 2: Distribution of Factor Endowments Across Countries (continued) Share Of Semi-Skilled Labour Lowest Mali Gambia Benin Sierra Leone Bangladesh Highest Jamaica Trinidad And Tobago Guyana Austria France Share Of High-Skilled Labour Lowest Gambia Rwanda Uganda Malawi Mali Highest Canada United States New Zealand Australia Philippines 1990 18.0 38. 32 .3 139.6 4.
1980-1990 Economies of Scale Concentrated to Concentrated Electrical machinery Non-electrical machinery Petroleum refining Non-ferrous metal Wearing apparel Professional equipment Concentrated to Dispersed Wood Products Beverages Petroleum and coal products Dispersed to Dispersed Other chemicals Food products Non-metallic mineral products Textiles Glass products Printing and publishing Plastic products Dispersed to Concentrated Fabricated metal products Leather products Residual Rubber products Paper products Industrial chemicals Furniture Iron and steel Pottery and china Footwear Transport equipment Tobacco Other manufactured products Medium Medium High High Low Medium Technology Growth High High Medium Medium Low High Medium High High Medium High High Low Low High Low Low Medium Low Medium High High Low Medium Low Medium Medium Low High Low Medium Low Medium Low Medium Medium Low Low Low Medium Low High Medium Low Medium Low Low Medium Low Medium High Low Medium Medium Low High Low Low Medium Low High Low Medium Medium Low High Low Medium Low Medium High Low Low High Medium Medium Medium High Note: Please refer to pages 10-12 of the text for the description of the classifications and indicators. 33 .Table 3: Changes in Concentration.
248** (0. The independent variables are the log of factor endowments as a percentage of total labour.521) 72 0.668) 4.180 (2.553 (0.562* (13.529) -5.813) 2.492** (0.225) 1.411** (14.282 (16.827) -0.31 Mining 3.426 (0.08 Services -5.095) 72 0.099* (2.81 Standard errors in parentheses * significant at 5%.115* (2. The dependent variable is the sector value added as a percentage of GDP. ** significant at 1% Table 5: Regression of Sector Shares on Economy-wide Endowments.135* (1.736) 2.12 Services -5.475 (1.017) -0.666) -2.481 (0.076** (0.506* (0. 1990 Semi-skilled Labour High-skilled Labour Land Capital Constant Agriculture -5.897) 69.692) -6.979** (14.353 (0.591) 1. ** significant at 1% Note: Table 4 and Table 5 present the results of the OLS regressions.388 (0.430 (0.571) Manufacturing 5.994) 5.729 (0.651 (0.812 (1.819) -5.63 Observations 72 R-squared 0.017) 0.405) 1.414* (1.397** (1.032 (2.734) 66.451 (2.594) 72 0.662 (0.738) -1.647) -23.828** (2.607) 4.480) -0.273** (0.053) 0.Table 4: Regression of Sector Shares on Economy-wide Endowments.555* (10.127) 0.482) -0.983* (0.666) -31.157) 72 0.734) -0.842) 69.865** (17.740) 2.901) Manufacturing 6.911) 0. 34 .054** (14.52 Observations 72 R-squared 0.634) 72 0.364** (0.44 Mining 2.74 Standard errors in parentheses * significant at 5%.045) -0.452) -1. 1980 Semi-skilled Labour High-skilled Labour Land Capital Constant Agriculture -2.857* (0.067 (0.850 (1.090 (15.491) -1.071) 72 0.929) 49.572 (0.313 (2.
15 0.28 0. 1990 Semi-skilled Labour + + + High-skilled Labour + Capital R 2 Total manufacturing Food products Beverages Tobacco Textiles Wearing apparel Leather products Footwear Wood products Furniture Paper products Printing and publishing Industrial chemicals Other chemicals Petroleum refineries Petroleum and coal products Rubber products Plastic products Pottery and china Glass products Non-metallic mineral products Iron and steel Non-ferrous metals Fabricated metal products Non-electrical Machinery Electrical machinery Transport equipment Professional equipment Other manufactured products - + + - + + - + + + 0.13 0.08 0.03 0.21 0. – denotes that the coefficient is significantly negative at the 5% level.41 0.34 0.37 0.11 0.05 0.03 0. + denotes that the coefficient is significantly positive at the 5% level.16 0.13 0.51 0.11 0.45 0.02 0. 35 .21 0.05 0.Table 6: Regression of Manufacturing Industries on Factor Endowments.25 0.28 0. The dependent variables are the industry value added as a share of GDP.02 0.19 Note: Table 6 presents the summary of the result from OLS regression. The independent variables are the factor endowments as a percentage of total labour force.02 0.21 0.32 0.39 0.31 0.
53 0.44 0.Table 7: Regression of Manufacturing Industries on Factor Endowments.18 0.21 Note: Table 7 presents the summary of the result from OLS regression.35 0.09 0.11 0.41 0.03 0.36 0. 36 .23 0. 1980 Semi-skilled Labour Total manufacturing Food products Beverages Tobacco Textiles Wearing apparel Leather products Footwear Wood products Furniture Paper products Printing and publishing Industrial chemicals Other chemicals Petroleum refineries Petroleum and coal products Rubber products Plastic products Pottery and china Glass products Non-metallic mineral products Iron and steel Non-ferrous metals Fabricated metal products Non-electrical Machinery Electrical machinery Transport equipment Professional equipment Other manufactured products High-skilled Labour Capital + R 2 + - + + + + + + + + 0.05 0.01 0.19 0.21 0.16 0.20 0.17 0.60 0.09 0.10 0. – denotes that the coefficient is significantly negative at the 5% level.12 0.09 0. The dependent variables are the industry value added as a share of GDP. + denotes that the coefficient is significantly positive at the 5% level.43 0.07 0.09 0.28 0.13 0.19 0. The independent variables are the factor endowments as a percentage of total labour force.