| The Pattern of Industrial Growth in Chile (1977) Josino Moraes Latin America Economic Researcher www.josino.net email: josinomoraes@hotmail.com Contents I. Introduction…………………………………………………. II. Theoretical Framework III. Chenery's Development Patterns…………….. IV. Chile's Pattern of Industrial Growth V. Technological Innovations VI. Conclusions Footnotes Appendix A Appendix B Appendix C Bibliography………………………………………………………………………… Introduction Industrialization is often associated with economic development in the sense of high rates of growth of per capita income. This was true in the case of the developed countries. The principal aim of this paper is to verify whether or not the Chilean industrialization can be asso¬ciated with significant economic development. It is useful to define the precise meaning of the terms employed. Industrialization may have completely different meanings and is not always explicitly specified among the authors writing in this field(l). It is seen, in quantitative terms as the increasing share of industrial production in aggregate product and as the increasing share of labour force employed in the industrial sector. Indus-trial sector is considered either as the aggregate for the mining, manufacturing and construction sectors or only as the manufacturing and construction sectors. The meaning used is specified in each case. Industrialization, in qualitative terms, is meant as the spread of technological innovations, forms of organization of enterprises, etc. This word shall be used within the limits of the analysis in the sense of an increasing share of industrial produc¬tion in aggregate production. This is briefly discussed in section IV, while section V explores the spread of tech¬nology within the manufacturing sector. Given this meaning of the word, a study is made of whether the process of industrialization has been accompanied by high rates of growth of per capita product with regard to the past and with regard to the developed countries at the same moment in time. The idea earlier expressed by Chenery is that "industrialization is the main hope of most poor countries" This paper argues that this contention must be rejected by the Chilean experience, when industrialization is given the meaning as defined above. In section II is described the theoretical frame-work of the present discussion on industrialization and growth. It is mainly a discussion of the predictive value of cross-section analysis for infering intertemporal changes in the structure of production as income levels rise, in which Chenery argues for and Kuznets, after 1966,against. Section III is a closer look at Chenery's de-rived patterns of development based on cross-section data, Section IV is a study of the Chilean pattern of industrial growth for the period 1908-1971 compared with Chenery's derived pattern of development in which is included the Chilean case. There is also a discussion on the reasons for Chile's particular industrial growth and its relation with aggregate growth. Finally, in section V, the matter of technological innovations is taken up and a comparison is made between Chile's manufacturing sector and Sweden's industrial sector. II. Theoretical Framework This paper is concerned, in the theoretical framework, with theory of the economic growth of nations. Although econ¬omic theories of growth have been formulated as early as the mid-nineteenth century by the Classical school, it is diffi-cult to speak of a theory which can explain the interaction of all the essential factors involved in the process of economic growth 3). Jungenfeldt points out that the theory consists properly of a great many models each of them con¬structed for a specific and limited problem (4}. The first explicative variable often associated with rate of growth of per capita income is the rate of investment which, as uznets points out, is a minor detail in accounting for the striking differences in levels of performance and pat-terns of growth that are observed between the developed and the less developed countries(5). At any rate, it is an ex¬plaining factor of economic growth. A second line of reasoning about growth is the relation between aggregate growth and pro¬duction structure. It is the transformation of the production structure (the shift from agriculture to industry for example) accompanied by economic growth which is brought into focus. Kuznets's studies on the association between production structure and economic growth were done in two ways(6 . First, arranging the countries by their per capita income for a recent period and observing the differences in production structure, i.e., share of labour force employed in each sector and in national product. This is the cross-section approach. Second, a more direct approach consists of the study of the long term trends in production structure and income per capita for the developed countries. Considering only three major sectors of production: primary (agriculture, forestry and fishing), industry (mining, manufacturing and construc¬tion) and services (all other sectors), some of the results obtained were: as per capita income levels rise, -the share of primary production in national product decreases, -the share of industry production in national product increases, -the share of services in national product shows no consistent movement. Both the cross-section analysis and the long term trends provided silr]ilar results. At first, Kuznets was impres¬sed with the similarities between historical and cross-country patterns. After 1966 he changed his original point of view assessing: "The value of such (cross-section) analysis for generating some preliminary hunches can not be denied. But unless innovational changes can somehow be taken into account in the use of the cross-section data proper, use of its re¬sults may lead to erroneous inferences concern¬ing past changes in structure in the process of growth. And the same applies, pari passu, to application of cross-section analysis to projections into the future." (7) In 1971, Kuznets extended this point of view: "It may well be that the discrepancies (bet¬ween estimations based on cross-section analysis and long term records) are analy¬tically significant, reflecting as they do changes over time in the pattern of rela¬tions between per capita income and shares of sectors -an inference that is suggested by the continuously changing technological and institutional framework within which modern economic growth takes place. The discrepancies are ,therefore, to be studied -not removed. To put it paradoxically, the value of the cross-section may lie not in its capacity to predict correctly the magnitude of changes over time; but rather in its revelation of the discrepancy bet¬ween its implication and the observed his¬torical change -a discrepancy indicating that temporal changes must have occurred in the relations. But the proper iden¬tification of such temporal changes re-quires the evidence of long term records!"( 8 ) However, Chenery in 1968 extended his previous study of 1960, applying more formal econometric methods to reafirm the capacity of cross-section analysis to predict correctly the magnitude of changes over time. Chenery's analysis of 1968 shall be tested for its long run predictive capacity in the Chilean case. Although it will not suffice to deny the. capacity of his cross-section analysis to predict changes over time, which would require studying a group of countries, it provides a good point of departure in the discussion of the Chilean industrialization. Temin has carried out a regression analysis based on time-series data for a set of developed countries which in spite of some objections concerning the problem of agri¬cultural development as income levels rise, found results concerning the industrial sector very similar to those of Chenery (9) Kuznets' work of 1971, based on a 1958 cross section, shows conformity of long term changes with those indicated in the cross-sectionfor the developed countries included in the sample. This is not true for the three less developed countries (Honduras, Philippines and Egypt). The share of the primary sector actually de¬clined while per capita product failed to rise signifi¬cantly whereas the cross-section would indicate no change in the share of this sector. (10) This is more likely the Chilean case, at least after the 1930's depression. III. Chenery' s Development Patterns Chenery and Taylor's paper of 1968 is the basis for the present summary. Some eleCments of the general expla¬nation which are found in Chenery's earlier paper of 1960 and his later work of 1975 will be taken into account in seeking to construct a better picture. The first two parts of the study consist of A. Reestimation of multiple regression describing inter-country growth patterns for major sectors and country groups. B. Comparison of postwar changes in each group of countries to the intercountry regression and to the historical patterns of the developed countries. The third part is an analysis of twelve countries' industrial sectors designed to provide a less aggregated view of production patterns, which is not taken into ac-count for it is out of the limits of this paper. Our focus is on an aggregate level. A. Variation among countries 1) Hypothesis A development pattern may be defined for a given country by the time paths of variables describing production, international trade and resource all¬ocation in each sector. A comparable cross-section pattern may be defined by the variation in the same set of variables among countries at a given moment in time. The two patterns can be compared by ex-pressing both as functions of per capita income and other variables. The intercountry pattern of any year is generated by the historical development patterns of all countries in prior years. If each country's pattern is dominated by a set of univer¬sal factors common to all, the cross-section rela¬tions will reveal some of the characteristics of these underlying factors. The universal factors suggested to explain the inter-country uniformities include: i) similarities in production relations -common production functions, substitution of capital for labour with rising income, etc, ii) similarities in domestic demand -both in pri¬vate consumption and public expenditures, iii) similarities in opportunities for trade and international capital movements. A country's development pattern can be described in terms of three components: a) the normal effect of universal factors which are related to the level of income b) the effect of other general factors such as market size or natural resources over which the government has little or no control, c} the effect of the country's individual history, its political and social objectives and the particular policies the government has followed to achieve them. Cheners primary concern is the identification of the uniform factors a and b which affect all countries. Chenery's explanation for the uniform patterns of change in the structure of production as income levels rise is an effort to take into account not only demand conditions (Engel's law) but also sup-ply conditions. Engel's law explains the rise in the share of industrial output in national pro-duct specifying that the income elasticity of demandfor food is less than for industrial products, implying that the share of food in total consump¬tion will fall as the level of income rises. 2} Econometric procedure The data are from the United Nations national ac-counts, which provide a sample of fifty four countries over the period of 1950-1963. The de-pendent variables in the regression equations are the shares of the three major components in GNP: = share of primary production (agriculture and mining) xm = share of industry (manufacturing and construction} xs _ share of services (all other sectors) The breakdown follows that of Kuznets, except that mining is combined with agriculture because of its similar role in trade. The set of explanatory variables are: y = per capita GNP (in 1960 dollars) N = population in millions k = share of gross fixed capital formation in GNP ep = share of primary exports in GNP em = share of manufactured exports in GNP All the explanatory variables are utilized in a simple multiple regression equation. This method assumes that the effect of each variable is additive in loga¬rithms and independent of the values taken by the others. Regression equations. Three logarithmic equations from cross-section data are estimated: Regression A Iny+ (Iny)2+ 1n N+ + j In k + In ep * In em Regression B + 9 In y t (In Y)2 + In N Regression C In xi = cxY 4 ~,9 In y + In N Regression A provides a better explanation of the varia¬tion in the shares of industry and primary production than the equation used in 1960 (Regression 0). After using the pooled sample (all countries) cross-section patterns, Chenery explores an alternative formulation in which the sign equationsare fitted for sub-groups sample. to determine the merits of same regres¬of the total The first subdivision is made between large and small countries. Large countries, it is said, are expected to industrialize earlier than small ones becausaeconomies of scale shift their comparative advantage toward industry. However, the importance of this effect declines as incomesrise, and it may ultimately be outweighed by greater exports of manafac ture d goods from small countries. The next step is to take up the effects of natural resources and trade patterns in the small country group. The small countries are divided into two groups on the basis of an index of trade orienta¬tion -toward primary or manufactured exports. All these subdivisions are confirmed by regression results. Three development patterns are identified: large countries, small industry-oriented countries and small primary-oriented countries. Chile is included among the small primary-oriented countries. The results of regressions A and 0 which here are interesting for the share of industry in this group are: Regression A 1.6875 - 1.4666 In y + .1474(ln y)2- .0305 In N + (.84) (.29) (.02) (.02) + .1619 in k - .2406 In ep+ .0058 In em R2=.798 (.04) (.03) (.01) Regression C in x = - 3.7951 + .3439 In y + .0569 in N R2=.697 m (.09) (.02) (.02) B. Variation over time 1) Historical evidence Taking data for primary and industrial production from Temin for nine developed countries within the period 1870-1950,Chenery finds that they fit quite well (time -series) with the cross-section patterns for the large countries. 2) Historical vs. intercountry variation in the post-war period. The time series elasticity has been computed for each country by fitting regression C without the population term to the data for 1950-1963. The median values are quite close to the cross-section elasticities for all three groups. Chenery finds 12 that the majority of countries tend to move parallel to the cross-section pattern. Chenery has made predictions for country groups using regression B and the observed change in per capita product from 1950-1952 to 1961-1963. These projections from cross-section regressions underestimate the actual decline in primary output by about 5 % and the rise of industry by about the same amount. The major problem with Chenery analysis is that it deals with a short time span in the time-series for less developed. IV. Chile's pattern of industrial growth The main aim in this section is to constrast Chenery's derived pattern of industrial growth based on cross-section data with the actual Chilean industrial growth for the period 1908-1971. National accounting first appeared in Chile in 1940. However,there are estimations for the variables studied here from 1908.They are from BCLA(Economic Commission of Latin America) and Ballesteros and Davis.A historical se¬ries is constructed_ based on these sources for the evolu¬tion of the share of industry(manuf acturing and construe¬tion,the same definition Chenery uses) and of per capita GDP. The results are presented in table 1. These results for the period 1908-1940 seem quite reasonable in the light of Chile's economic history.In fact,during the period 1908-1929 within the denominated period of development out-wards (1830-1930) there was considerable expansion in the Chilean economy,with regard to the growth of domestic pro-duct. This was a consequence of production in the mining sector, and principally by some foreign-owned saltpetre and copper mines which were oriented to international trade. The impact of the great depression was enormous in Chile and the per capita product level of 1928 was not regained until the early 40's. The first feature that can be noticed from the results presented in table 1 is that it seems faulty reasoning to conclude that there exists just one development pattern for the Chilean economy in the long run,i.e. a uniform chan¬ge of structure of production as income levels rise. Share of GDP GDP per capita M anuf ac . % I Construe. % II Manuf ac . Construe. 111=1+II Indexes 1960100 IV in 1960 US 1960=294 V 1908 9 51 150 9 10 51 150 9 54 159 10 1 10 53 156 9 59 173 2 3 9 59 173 4 10 54 159 5 11 50 147 6 9 58 171 7 8 62 182 8 9 62 182 9 11 48 141 20 10 54 159 1 13 46 135 2 12 49 144 3 12 60 176 4 12 65 191 5 11 65 191 6 10 65 191 7 9 68 200 8 7 84 247 9 8 91 268 30 9 80 235 1 8 66 194 2 12 47 138 3 12 54 159 4 11 64 188 5 11 67 197 5 12 67 197 7 11 78 229 8 12 73 215 9 11 74 218 40 12 3 15 77 226 1 15 3 18 88 259 2 14 4 17 83 244 3 16 3 19 85 250 4 17 3 20 86 253 5 18 3 21 90 265 I II 1II=1+II IV V 1946 16 4 20 82 241 7 17 4 21 87 256 8 17 3 20 95 279 9 18 3 21 89 262 50 17 3 20 91 268 1 16 3 19 93 273 2 15 3 18 97 285 3 17 3 20 96 282 4 18 3 21 95 279 60 25 5 3o 100 294 1 24 5 29 103 303 2 25 5 30 106 312 3 26 6 32 108 318 4 27 6 33 110 324 5 26 6 32 114 335 6 25 6 31 123 362 7 27 5 32 122 359 8 30 5 35 123 362 9 29 4 33 127 373 70 29 4 33 129 379 1971 27 6 33 133 391 It seems more reasonable to distinguish two periods, the period 1908-1929 when there seems to be no association between the share of industry and income levels and the subsequent period after 1929. The year 1908 is chosen as the beginning of the first period because it is the earliest year with available records.Th e ending of this first period is undoubtely the year 1929,when the expansion of the mining sector (mainly saltpetre production) breaks down. As the beginning of the second period the year 1940 is chosen main¬ly for two reasons. First, because it is possible to dis¬tinguish the beginning of a period characterized by a stea¬dy rise of the share of industry in aggregate production. Secondly, because the years immediately after 1929 present a great disturbance caused by the great depression. A closer study of both periods will be undertaken. 1) Period 1908-1929 The available data for this period are only for the manufacturing sector which is the most important one within the industrial sector (table 1). The fitting of a logar--ithlnic line of the form indicated by Chenery' s regression equation C (section III,pg. 11) without the component of population gives the following result: In x' = .481 - ,e543 In y `e •393 m (.77) (.15) x' ., share of manufacturing in GDP m y .. per capita GDP in 1960 US $ The fitting of these observations into a logarithmic line is quite poor,as indicated by the coefficient of deter¬mination.This is done here only to confirm on a more objec¬tive basis the preliminary hunch given by table 1, that there is no positive association between the share of the manufacturing sector and income levels. In fact, the above equation gives a negative association between the two variables as indicated by the negative sign of the co-efficient with respect to per capita GDP (- 0,543). In other words, the income elasticity for the share of indus¬try is negative. 2) Period 1940-1971 In this period there is a clear association between the share of industry in domestic product and per capita product levels. In fact, fitting a logarithmic line of the form of regression C without the population term, gives the following result : In x = -10.186 + 1.543 In y R2 = .835 m (.78) (.14) m ..share of industry y ..per capita GDP in 1960 U5 $ This equation reveals an extremely high income elasticity for the share of industry contrasting with that indicated by Chenery's regression equations. In fact, the normal values for the share of industry given by regressions A and C based on the data for the explanatory variables are calculated and presented in table 2 with the actual values. The data for the explanatory variables are given in appendix B. As can be seen in table 2, the actual values of the share of industry in GDP are much higher than the expected values indicated by both regression equations. The Chilean case presents a high positive deviation in the more recent period suggesting this particular phenomenon of high rates of shift of the production structure with low rates of growth in per capita product. The predictions from regression A are calculated here to see the influence of the other ex-planatory variables (other than per capita product and size). They are not significant as can be observed in table 2. Table 2 Predicted and actual values for the share of industry, 1940-1970 . Predictions on basis of Actual Regression A = Regression C } , % ' r % 1 + % 1940 15 16 15 1945 i'6 17 21 1950 17 17 20 1954 18 18 21 1960 17 18 30 1965 18 18 32 1970 18 20 33 Source : See appendix B Graphically, the differences between the normal values given by Chenery's regressions and the actual values are shown in figure 1. The above regression equation de-rived from the actual values for the share of industry in the period 1940-1971 is plotted against Chenery's regression A. The first conclusion that emerges from the above results is that the effects of the universal factors sug¬gested by Chenery (pg.8) are not significant in explaining the remarkable rising share of the industrial sector, The reasons for this peculiar Chilean development pattern must be sought in its economic history. 250 300 350 400 per capita GDP in 1960 US $ figure 1 3) Reasons for the deviation Some of the explanatory factors of the Chilean development patterns appearing in these two periods have to be sought in government policy. The contrast between eco¬nomic policy put into practice before and after 1929 is noticeable. The impact of the great depression on the Chilean economy originated a series of transformations which are the basis of Chile's development pattern during this period. During the period 1908-1929, the expansion of the Chilean economy was based on the mining sector and the eco¬nomic policies were concentrated upon the same. partner's description of the dominant economic policy between 1906-1920 makes it clear that the industrial sector was not con¬sidered important and that the problems on focus were in¬ternational trade and construction of railways throughout the country. (11) Within this context, it is not surpris¬ing that a positive association is not found between the share of manufacturing sector and per capita GDP level. The rising income provided by the mining sector did not stimulate domestic manufacturing production, but possibly import of manufactured products. The immediate effect of the crisis of 1929 was a drainage of the Central Bank's gold reserves.. Import res¬trictions and tariffs were adopted. The economic policies after 1929 are characterized by a high degree of protec¬tionism (12) . The former demand for import of manufactur¬ed goods created a basis for domestic manufacturing produc tion. A considerable movement of capital to the menufactur ing sector took place. Between 1333 and 1934 it is almost tripled (13). The economic policies after 1929 were designed to favour the expansion of the manufacturing and construction sectors. Some of the issues were: 1) stimulus to building activity; 2) easier credit conditions for those sectors; 3) promotion of industrial production by various develop¬ment banks (Institutes de Fomento) coordinated 'since 1940 by CORFO ( Corporaci6n de Fomento de la Produc cian ); 4) creation of several state-owned indus trial enterprises such as ENDESA ( Empresa Nacional de Eleotricidad ), ENAP ( Empresa Nacional del Petr6-leo ) and CAP ( Companla de Acero del Pacifica ). The high positive deviation of the share of industry shown in Figure 1 is approximately offset by the sharp decline in the share of agriculture in G. L. P. A source of explanation, as suggested by Chenery and Kuznets, is income distribution. (14) Although a systematic study of the long run income distribution in Chile could not be found, several sources suggest a tendency for income concentration in Latin America as a whole and for Chile. (15) If the additions to income are concentrated to the high income groups, which have a high marginal propensity to consume industrial products, a steady rising of demand in the industrial sector results, while demand in the agricultural sector remains constant. However, this does not seem to be a relevant source of explanation in the Chilean case. In fact, Chile has changed from a net exporter to a net importer of agricultural products(16) One element often suggested for explaining the crisis of the Chilean agricultural sector is the structure of property. However, the whole of Latin America has a very similar structure of property and the intensity of the decline verified in Chile is higher than in the rest of the countries. Another explicative factor is the fairly permanent price control on the agricultural products`. This is certainly an important factor explaining the lack of domestic supply to cover domestic demand of agri-cultural products in the more recent period. There have been several factors playing im¬portant roles in the forming of Chile's particular development patterns. Of these, the most signi ficant has been government policy, a crucial fac¬tor in the increment of Chile's industrialization relative to the cross-sectio.al normal pattern after 1929. 4) Industrialization and rates of growth If industrialization is seen as the in-creasing share of industrial production in aggre¬gate production, this process began after the crisis of 1929. The data of Table 1, taking three-year average for the extreme points gives the following results : Compound annual rates of g rowth of per capita GDP 1908-1910 to 1928-1930 2.5 1939-1941 to 1969-1971 1.6 % These figures do not allow one to infer that indus¬trialization in the Chilean case is accompanied by higher rates of growth of per capita GDP neither with regard to the past nor with regard to the developed countries. On the contrary at the same time as industrialization takes place in Chile the gap between Chile and the developed coun¬tries is widening The rates of growth of per capita product in Chile do not mean that all the groups in society have improved in real income. It is more likely that this "growth" has been concentrated to the high income groups. If one applies the method recently suggested by Kuznets of calculating rates of growth, taking into account this distortion in income distribution, it would give considerably lower rates of growth in the Chilean case(i8). Another adjustment sug¬gested by Kuznets, when comparing rates of growth of a less-developed country such as Chile with the developed ones, would also give lower rates of growth. This later adjust¬ment is due to the existence of different price structures in both cases. The ratio of industrial and service prices to agri¬cultural prices are higher in Chile than in the developed countries or than in the international marketa(19 . As the rate of growth of aggregate output can be seen as each single sector's share in aggregate output multiplied by its rate of growth, any adjustment that diminishes the weights of the industrial sector and services will decrease the aggregate rate of growth. This is the case in Chile when the sectoral weight of the industrial sector prd ser¬vices is calculated with regard to the price Structure in the developed countries. Industrialization in Chile has not been accompanied by high rates of growth of per capita product. Nor can it be seen as a process accompanied by the transformation of the country from a poor one to a rich one(20). There is no sign of such a process in the Chilean case. When industrial¬ization is considered as the rising share of labour force employed in the industrial sector within which is only in¬cluded the factory sector (defined as the manufacturing sector having at least 5 employees a plant) this process has been working in Chile(21). However if industrialization is seen as the share of labour force employed in the industrial sector, when the entire manufacturing sector is considered , this process has not taken place in Chile, because the share has been fairly constant(22). Industrialization in qualitat¬ive terms, as the spread of technology in general is an extensive area of research. However, whether or not tech¬nological innovations have influenced the manufacturing sector can be explored. This will be done in the next section for it will allow for a better understanding of the Chilean problem of industrial growth. V . Technological Innovations In studying the evolution of the manufacturing sec-tor there are some signs of a lack of technological inno¬vations in Chile's industrial growth. During the last 100 to 150 years the developed countries have had high rates of growth, which can for the most part be attributed to technological innovations. Innovations raise output per worker and finally output per capita. The raising of indus¬trial production's share of aggregate production in the de¬veloped countries was steadily accompanied by technological innovations. This seems not to be true in the Chilean case. Two proxy variables are used for the identification of technological innovations : the evolution of labour pro¬ductivity and the evolution of average plant size in the manufacturing sector as a whole. A comparative study is made, with Sweden representing developed countries. There is no particular reason for choosing Sweden other than the facility of data, and, as the developed countries have had a quite similar development pattern, there can be no rel¬evant objection to this procedure. 1) Evolution of Labour Productivity There are three sources on the long run evolution of manufacturing output in Chile. The prime source is that of Ballesteros and Davis from 1963 (23). The other two are from UCLA and Munoz (see appendix C). ECLA's series is based on Ballesteros and Davis's for the period 1908-1929 but after 1929 gives strong downward estimations. Munoz's series diverges from that of Ballesteros and Davis for the period during the First World War giving strong upward estimations and showing the same time path as the data of Carmagnani who studies an earlier period (24). The high rates of increase in manufacturing output indicated by Munoz and Carmagnani for the period 1914-1918 seem to agree with the data for employment given by the Anuarios Estadisticos for the period.However this is not relevant for the long run records discussed here. Prom the study of these sources it seems sensible to chose either Ballesteros and Davis's series or Muffoz's as the more optimistic data and on the other hand ECLA's series as the more pessimistic. Munoz's series is imple¬mented here because it covers the period until 1965 while Ballesteros and Davis's stops at 1957. Concerning the evolution of the labour force employed, the data are from Anuarios Estadisticos and Censos Manufactureros. The re-salts are condensed in table 3. Table 3 Indexes of Labour Productivity in Chile and Sweden, 1916 to 1965. Chile (manufacturing) Sweden (industry plus handicraft) ECLA Muno z 1916-1920 100 100 100 1921-1925 113 95 92 1926-1930 114 104 125 1931-1935 96 95 114 1936-1940 100 121 142 1941-1945 88 150 151 1946-1950 81 156 200 1951-1955 102 180 251 1956-1960 109 173 301 1961-1965 124 191 402 Source : See Appendix C The results obtained on the basis of ECLA's estimates of manufacturing output indicate practically no increase in labour productivity and even MuEoz' s, which apparently give the highest output development, show an insignificant in crease in productivity. In fact, an increase of 90 % in 50 years is very low when compared with 300 % in the Swedish case. 2) Evolution of Average Plant Size The development of average plant size in Chile for the period 1916-1972 are shown in table 4. They are calcu¬lated in different ways for two reasons. First, to attempt to eliminate the influence of small establishments on average size. Secondly, due to changes of classification in the Statistical Sources. The discontinuity of the data is due to lack of information. Indexes for the average plant size in the Swedish industrial sector are presented in the last column. They are utilized to show the long run increase in average plant size in a developed country as Sweden. This trend is also found in the American manufacturing sector between 1914-1937(25). As can be seen, the results for Chile show no trace of increases in average plant size, whichever of the first six columns is observed. On the other hand, the Swedish industrial sector shows an increase of 65 % between 1922-1964. The bias observed between 1946 to 1956 in column III can perhaps be explained by the fact that the data for the period 1942-1956 as well as 1916-1926 are from the Anuarios Estadisticos and it is possible that the sample underesti Table 4 Average Plant Size in Chile, Sweden,1916-1972 over 100 workers II Chile over 100 total VI Sweden indexes VII over 10 workers I over 5 total III over 10 over 50 total IV total V 1916 44 237 22 7 46 242 24 8 50 285 25 9 25 20 24 1 24 2 25 100 3 26 103 4 26 115 5 26 114 6 26 105 1928 30 101 1937 34 52 165 277 106 1942 29 3 31 4 33 5 35 6 47 7 48 108 8 40 111 9 38 111 50 38 108 1 45 116 2 45 122 3 45 120 4 48 127 5 49 132 6 47 137 1957 35 55 190 322 141 1960 4g 58 182 148 1 49 58 182 153 2 51 60 176. 154 3 52 61 180 158 1964 54 64 182 165 Table 4 (cont.) Chile over 10 workers I over 100 workers II over 5 total III over 10 total IV over 50 total V over. 100 total VI 1965 54 66 177 6 54 66 178 7 55 240 31 67 177 292 8 247 185 302 9 241 186 304 70 238 191 301 1 256 191 322 1972 255 201 325 total = wage plus salary earners Source: See Appendix C mates the weight of the small establishments. The Canso uanufacturero of 1957, which is a much more reliable source, provides an average near the historicpl one. Why this does not happen in the former period is possibly due to the fact that the Anuarios for this period are with regard to many issues much better. As indexes are used for the average plant size in the Swedish industrial sector, an estimation is made of recent average plant size in the Swedish manufacturing sec-tor to show the differences between both countries. Table 5 Recent Average Plant Size Chile Sweden over 5 (total) 31 68 over 10 (workers} 44 79 over 100(workers} 240 344 Sources : Chile, table 4, 1967 Sweden, Industri 1971. Over 5 (total}, tabell 1, pg. 50; over 10 and over 100(workers), tabell 2, pg. 60. Data from 1971. Average plant size is much higher in the Swedish manufacturing sector than in the Chilean. different conclusion than that reached by Ricardo Lao 26). The author studying the period 1937-1957 finds a trend towards concentration of the labour force to larger plants in the manufacturing sector. The extension of these data backwards (1937-1916) and forwards (1957-1972) suggests no basis for such a conclusion. Since this study only covers the evolution of average plant size, the index of concentration in 1967 is calculated, using the same pro¬cedure as Lagos (27). In fact, an index of 0.69 is obtained which lies a little below the author's estimate of 0.71 for 1957. It is however, a little above the index for 1937. These are Lorenz's indices. Lagos makes other estimations on concentration of value added in 1957 and finds that the value added gener¬ated by the larger plants is higher in Chile than in the United States.These studies are carried out to show levels and trends of oligopolization in the Chilean markets of manufactured products. This method is undoubtely a rough approximation as a study of market structure. However, when it is applied for a long time span,it suggests an interesting new hypothesis for further explorations : that these markets had some degree of oligopolization from the beginning. This is suggested by the presence of large plants in the beginning of this century (28). One interesting feature to point out is the basis for Lagos's interpretation of the results. It is the transferring of the idea corresponding to the actual devel-opment experienced by the developed countries ( from free competition to oligopol ) to the development in Chile. It is the same as seeing industrialization in Chile as being associated with high rates of growth of per capita product, which was true for the developed countries. As Myrdal points out, this is often the source of mistakes(29) Chile's economic history does not necessarily follow the economic history of today's industrial countries. Technological innovations seem not to have been sig nificant in the development of the Chilean manufacturing sector. One of the reasons for this could be the lack of competition among enterprises, a consequence of some degree of oligopolization in the market structures. A second and important reason is the high degree of protectionism which has characterized Chile since 1929, and has principally, protected the domestic production of manufactured goods. VI. Conclusions The Chilean industrialization, in the sense of rising share of industrial product in agregate product, has not been accompanied by high rates of growth of per capita product either with regard to the past or with regard to the developed countries. One elenent that aggravates this performance is that these rates of growth are, because of the higher weight of the industrial pro¬duction and services in aggregate production, estimated to be higher than if they were estimated on the basis of international relative prices. This is due to the higher prices of industrial goods and services relative to agri¬cultural goods. As a consequence, Chile, being a less-developed country, has a structure of production very similar to that of the developed countries, a feature that could not be expected on the basis of a cross-section analysis. The main reasons attributed (in this paper) to this particular development pattern after 1929 is the major role played by government policy in increasing Chile's industrialization relative to the cross- sectional normal pattern studied here. The direction of state affairsseems to have favored the concerns of industry. The crisis of the agricultural sector has been, at least partially, a con-sequence of these economic policies. The Chilean industrialization, in qualitative terms, i.e., the presence of technological innovations, seems to be insignificant when studying the evolution of the manu¬facturing sector. The constant average plant size and low rates of increase in labour productivity in the long run point in this direction. The lack of technological innovations is attributed to some degree of oligopolization in the beginning of the century (a closer study of this hypothesis is required) and the high degree of protectionism for the manufacturing sector. Footnotes (1) (2) (3) (4) (5) (6) (7) (8) (9) A study of this matter is found in Freyssinet, pgs. 149-150 Chenery, 1955, pg. 450 See an interesting digression about this point in Kuznets, 1965, pgs. 4-5 Jungenfeldt, pg. 1 Kuznets, 1959, Pg. 74 xuznets, 1959, pgs. 43-57 Quoted by Sutcliffe, pg. 57 r uznets, 1971, pg. 198 Temin Pg. 181 (10) ruznets, 1971, pgs. 155 and 174 (11) Martner, Tomo TI, pgs. 590-658 (12) Jeanneret, pg. 70, and Ellsworth pg. 49 (13) Ellsworth, pg. 29 (14) Kuznets, 1971, pg. 129 and Chenery, 1960, pg. 647 (15) A study of the tendency for income distribution between 1940-1953 is found in Pinto, pg. 272 for Latin America, see Kuznets, 1973 (16) Pinto, pgs. 231-233 (17) In 1953, when the Chilean industrial sector (manufac¬turing plus construction) accounted for 20 % of GNP, the ratio of Chile's per capita income to the average of developed countries was 1/2.5; in 1974 when the Chilean industrial sector accounted for 39 % the same ratio was 1/8. The difference has more than tripled. The original data for these estimations are from Chenery, 1960, Table 1, pg. 632, and United Nations Statistical Yearbook of 1975, Table 188, pgs. 672-682 and Table 192, pg. 694. Developed countries here con¬sidered: United Kingdom, France, Belgium, Nether-lands, Germany, Austria, Denmark, Norway, Sweden, Italy, Japan, United States, Canada, Australia. (18) Xuznets, 1973 (19) Kuznets, 1971, pg. 137 (20) Chenery, 1975, pg. 8 afirms : "Taken together these processes /of which one is a rising share of industry in GDP / describe the different dimensions of the overall structural transformation of a poor country into a rich one", and, in pg. 135 : "In describing the processes of development we have tried to replace the notion of dichotomy between less-developed coun¬tries and developed countries with the concept of a transition from one state to the other." These ideas are highly doubtful when studying the Chilean case in the long run. (21) Utilizing the Censos de Poblaci6n of 1920 and 1970, the Anuario Estadistico of 1920 and the Censo Manu¬facturero of 1967 in order to isolate the factory sector from the entire manufacturing sector, the fol¬lowing results are obtained : manufacturing (factory) construction mining 1920 do 5': 8 3.9 4.4 1970 1 .3 6.2 3.2 24.7 14.1 (22) Utilizing the Censos de Poblaci6n of the following results are obtained : 1920 and 1970, 1920 1970 manufacturing 17.9 17.4 construction 3.9 6.2 mining 4.4 3.2 26.2 26.8 (23) Ballesteros and Davis, table 1, column 3, pg. 160 (24) Carmagnani, Appendice I, pg. 169 (25) Blair, pg. 125 (26) Lagos, capitulo VI (27) The data are from Censo Manuf acturero of 1967, toms III, pg. 31 (28) The idea expressed by Griffin, pg. 276, that there was a transition from small plants into large plants in Spanish America is not true in the Chilean case. (29) Myrdal, pg. 21 ADpendix A Sources qf table 1. col. 1 1908-1939, ECLA, Col. I and II 1940-1954, 1960-1971, Cuentas Nacionales, Chile, cuadro N°- 5, pg 151, Cuantas Nacionales, Chile, cuadro Ng 2, pg 22 The share of each sector's value added in GDP is approximated by the share of the sector's income in domestic income. This is because the national accounts for the period 1940-1954 only present the latter account. For the period 1960-1971 the two accounts give approximately the same results and for the sake of consistency, the same account is used. Col. IV 1908-1955, Indexes of output from Ballesteros and and Davis, table 1, pg 160. These in¬dexes are divided by population figures from Censos de Poblaci6n. A simple in¬terpolation is used for the years between censuses, 1955-1960, Statistical Yearbook of 1964, table 174, pg 546, 1960-1971, Cuentas Nacionales, table 16, column 4, pg 54. The three series are reduced to a single one in the following way : the data for the period 1960-1971 are transformed into indexes, the index for 1960 is used as the base (= 100) and then the series for the period 1955-1960 through the year 1960 and the extended series through the year 1955 are joined to the base. Col. V The GDP per capita for 1960 is from the Statistical Yearbook of 1975, table 193, pg 701. The rest of the column is constructed on the basis of column IV. Appendix B Sources of table 2. Data for projection from cross-section regression equations A and C. I y in 1960 US$ II N million III k IV e P V em % 1940 226 5.0 10.9 15.9 1945 265 5.4 8.6 12.9 1950 268 5.8 10.0 9.7 1954 279 6.3 8.4 6.7 (insignificant) 1960 294 7.4 15.4 12.4 1965 335 8.1 16.2 12.9 1970 379 8.9 14.9 14.6 Col. I table 1,pg 14, 15 Col. II Censos de Poblaci6n. For the years 1945, 1954, 1965, a simple interpolation is used. Gross fixed capital formation ( inversion interna bruta en capital fijo ) in GDP. 1940-1954 Cuentas Nacionales, cuadro 4, pg 150, current pe¬sos for GDP; cuadro 46, pg 185, current pesos for capital formation. 1960-1970 : Cuentas Nacionales cuadro 3, pg 24, current escudos for GDP ; cuadro 6, pg 30, current escudos for capital formation. Export of primary products in GDP, estimated in the following way. Export in GDP, Cuentas Nacio¬nales, cuadro 5, pg 145, cuadro 4,pg 150, cuadro 9, pg 38 and cuadro 3, pg 24. Share of primary products in export from Boletin del Banco Central NQ 335-6, pg 81, N2 554, pg 424, N2 394, pg 1334. Appendix C, Sources of table 3 Sweden Aberg, tabell B:3 4 pg 113. The original data are transformed into indexes, using the index for 1916-1920 as the base ( = 100 ). Chile Output : Munoz, cuadro II, 2.1, column C, pg 158. For the period 1936-1940, simple interpolation between 1937 and 1941 is used. ECLA, Addendum, table I-1,Industrial Production Indexes. For the years 1964_1965 the indexes from Industrias Manufactureras, 1968, pg 52, are used. Labour Force: Anuarios Estadisticos and Censos Manufactureros de 1957 and 1967. For the periods 1926-1942 and 1957 1965, simple interpolation is used. Sources of table 4 Sweden Aberg, tabell B:9, pg 119 and tabell B:10, pg 120. The indexes for total employment are divided by the indexes of total plants (arbetsstallen). Chile 1916-1926 and 1942-1956 : Anuarios Estadisticos 1927 : Censo Manufacturero 1928 1937 : Censo Manufacturero 1937 1960-1972 : Industrias Manufactureras. Bibliography Anuarios Estadisticos de la Republica de Chile, Industria Manufacturera, Oficina Central de Estadistica, 1916-1925, 1942-1956. Ballesteros, A.T. and Davis, E.T. "The growth of Output and Employment in Basic Sectors of the Chilean Economy, 1908-1957", Economic Development and Cultural Change, XI:2, January 1963.° Blair, J.M. "Does Large Scale Enterprise Result in Lower Costs? Technology and Size" , American Econ¬omic Review, papers and proceedings, N° 2, May 1948. Boletin del Banco Central de Chile, N°- 335-6, enero-febre¬ro 1956; N°- 554, abril 1974; N 394, diciem bre 1960. Carmagnani, M. "Sviluppo Industrialle e Sottosviluppo Economico. 11 case cileno (1860-1920)", Fondazione Luigi Einandi, Torino, 1971. Censos Manufactureros : Censo de la Industria Manufactu¬rers y del Comercio de 1928, Republica de Chile, Direccion General de Estadistica; Cen¬so Industrial y Comercial de 1937, Chile, Di¬reccion general de Estadistica; III Censo Na¬cional de ianufacturas, Chile, 1960, Direccion de Estadistica y Censos (data referred to 1957); IV Censo Nacional de Manuf ac turas, Instituto Nacional de Estadisticas, Chile, Tome 111,1971, (data referred to 1967). de Poblacion : Censo de la Republica de Chile, 28 de noviembre de 1907, Comisi6n Central del Censo, Santiago 1908; Censo de Poblacion de la Republica de Chile, 15 de diciembre de 1920, Direccion General de Estadisticas, Santiago 1925; Resultados del 10°- Censo de Poblacion noviembre de 1930 y estadlsticas comparativas con censos anteriores, Direcci6n General de Estadistica, Santiago 1931-1935; XIII Censo de Poblaci6n, 29 de noviembre 1960, Serie A, resu¬men del pals, Direccion de Estadistica y Censos, Republica de Chile.(Data referred to 1940, 1952 and 1960); XIV Censo Nacional de Poblacion y III de Vivienda, abril 1920, muestra de adelan to de cifras censales, total del pals, Instituto Nacional de Estadisticas, Chile, marzo 1971. Chenery, B.H. (1955) "The Role of Industrialization in Development Programmes" in "The Ecomics of Underdevelopment" ed. by Agarmola, A.N. and Singh, S.P., New York, Oxford University Press, 1963 (First Edition 1955). Chenery, B.H. (1960) "Patterns of Industrial Growth", American Economic Review, 50, September 1960, pgs. 624-654. Chenery, B.H. and Taylor, L. (1968) "Development Patterns: Among Countries and Over Time", The Review 'of Economics and Statistics, Volume ,L. No. 4, November 1968. Chenery, B.H. and Syrguin, M. (1975) "Patterns of Develop_ ment : 1950-1970", published for The World Bank by Oxford University Press, 1975. Cuentas Nacionales de Chile: 1940-1954, Corporation de Fomento de is. Production (CORED), Editorial del Pacifico S.A., Santiago de Chile, 1957; 1960-1971, Oficina de PlanificaciSn Nacional (ODEPLAN) ECLA, Economic Comission for Latin America (Comision Economica pare America Latina), "El Proceso de Industrialization en America Latina", Addendum E/CN 12/716/ Add 2. Ellsworth, P.T."Chile an Economy in Transition ",The Macmillan Company, New York, 1945. Freyssinet,J. "Le Concept de sous-dveloppment", Publica tions de is. Faculte de Droit et des Sciences Economiques de Grenoble, Serie economie du developpment, Vol. I, Mouton Cie. 19'36. Griffin, K. "i nderdevelopment in Spanish America" }George Allen and Urwin Ltd., London, 1969. Industri 1971, Del 1, Sveriges Officiella Statistik, Statistiska centralbyran, Stockholm 1973, (data from 1971) Industrias Manufactureras 1960-1972, Instituto Nacional de Estadisticas, Santiago de Chile. Jeanneret, T. "El vistema de Prctecci6n a la Industria Chilena" in "P roceso a la Industrializaci6n Chilena", Ediciones Nueva Universidad, Vice¬rrectoria de Comunicaciones, Universidad Ca¬t6lica de Chile, 1972. Jungenfeldt, K.G. "Ekonomisk tillvaxt", Sartryck, kap. I, Akademibokhandeln i Stockholm A.B.,, Lund 1974. S. (1959) "Six Lectures on Economic Growth", The Free Press of Glencoe, Illinois, 1959. S. (1965) "Economic Growth and Structure", W.Y. Norton and Co. Inc., New York, 1965. S. (1971) "Economic Growth of Nations. Total Output and Production Structure", The Belknap of Harvard University Press, Cambridge, Massachusetts, 1971. (1973) "Population, Capital and Growth", Selected Essays, W.W. Norton Co. Inc., New York, 1973. See "Problems in Comparing Recent Growth Rates for Developed and Less-Developed Countries". "La Industria en Chile: antecedentes estructu¬rales", Universidad de Chile, Instituto de Economia, Santiago de Chile 1966. Estudio de Politica Comercial Chilena e His¬toria Econdmica Nacional, Tomo I, II, Impren¬ta Universitaria, Santiago de Chile, 1923. "Crecimiento Industrial de Chile 1914-1965", Universidad de Chile, Instituto de Economia y Planificaci6n, 1968. "Vrldsfattigdomen", Utrikespolitiska Institutet 1972. "Chile, un caso de desarrollo frustrado", Edi¬torial Universitaria, Santiago de Chile, 1973. Sutcliffe, R.B. "Industry and Underdevelopment", Develop¬ment Economic Series, ed. by Arthur Hazlewood, Addison-Wesley Publishing Co. Inc., 1971. Temin , P. "A time-series Test of Patterns of Industrial Growth", Economic Development and Cultural Change, 15, January 1967, pgs. 174-182. United NatUnited Nations Statistical Yearbook, 1964 and 1975 Aberg, Y. "Produktion och produktivitet i Sverige 1861-1965", Industries utredningsinstitut Almquist & Wiksell, Stockholm, 1969. 1 |