Ranking Sectors Changes of the Malaysian Economy: Input-Output Approach

This study attempts to re-investigate the production structure change for Malaysia economy through the ranking sectors changes over the period 1983-2000. We used four input-output tables had published so far by Department Statistics of Malaysia (DSOM) for the period under study. The study employed the Leontief model for demand side (Input inverse (I-A)) for forward linkages indices, while supply side (Output Inverse, (I-O)) for backward linkages indices to examine the ranking sectors structure changes. New evidence is found in this study: first, the integration degree between demand and supply side for the Malaysian economy still remain weak. Second, the rank correlation coefficients between forward and backward indices are not significant and very weak. Third, the linkages between the commodities sectors and the rest of the economy still remain weak. Fourth, there is still a high dependency on the primary sectors, such us Oil palm, Rubber primary products and Wood sectors. Finally, fifth, the main results of the development policies were to transform Malaysia from an exporter to an importer foodstuff and other agriculture products.


Introduction
One of the objectives all less developed countries have set themselves is rapid growth in income per head.Rising incomes are associated in both time-series and cross-section studies with a rising share of industry in gross domestic product (GDP) [BULMER-THOMAS, 1982].
The development process can be carried out in a number of ways, but each new industrial investment will offer opportunities for other suppliers (backward linkages) and provide input for utilisation by other users (forward linkages).Furthermore, these backward and forward linkages are not reflected in market prices and therefore represent externalities, which could cause the social benefits of investment to diverge from the private benefits [BULMER-THOMAS, 1982;p.190].It might appear, therefore, that by concentrating on those sectors with high backward or forward linkages, the development process could be speeded up.If, furthermore, we were prepared to assume [HIRSCHMAN, 1958;p.102]common techniques across countries for each sector, a common set of relative prices and a distribution of income consistent with the eventual emergence in each less developed country of the structure of demand to be found in developed countries, then we could select our key industrial sectors for promotion by reference to the backward and forward linkage found in developed countries.
The average backward and forward linkage indices are greater in the developed countries than those in the less developed countries, and the indices of the coefficients of variation are lower in the developed countries than in the less developed countries, presumably revealing a lower level of integration of these economy's industries [BOUCHER, 1976;P.318].
We shall argue below that these assumptions are too strong and that the ranking of sectors or investments in terms of linkages in this way is not a very satisfactory guide to development planning.However, first we must show how we might measure such linkages using input-output tables for the Malaysian Economy, for it should be clear that such tables offer an excellent opportunity to quantify a concept which would otherwise remain empirically intractable [BULMER-THOMAS, 1982;P.192].The structural linkage of sectors can be described by two types of linkage effects, which can be measured in the framework of technology matrices.These linkage effects are the backward linkage effect and the forward linkage effect.
Before reviewing the theoretical basis of the linkage argument, two important points should be noted.
First, measures of linkage should not be confused with sectoral (income or employment) multipliers.Sectoral multipliers are designed to measure the impact of an increase in final demand on income or employment [BEKHEET, 2009].Measures of linkage are designed to assess the impact of an increase in final demands on gross outputs.A high value for backward or forward linkages does not imply a correspondingly high value for the income or employment multipliers, a point overlooked by some writers who seem to assume that high linkages mean a high domestic value-added content.For the above reason I will study the multipliers in a separate paper [BEKHET, 2010].
Secondly, it is important to distinguish between measures of linkage based on the existing technology of a economy's structure of production, and measures of linkage based on the existing interdependence of domestic sectors of production.In the latter case, backward and forward linkages measure the impact of a unit increase in final demand on domestically supplied inputs and outputs, and the appropriate matrix for calculating linkages is (I-A d ) -1 , where A d is the matrix of domestic transaction coefficients.In the former case, measures of linkage are based on the technology matrix (I-A) -1 , where A is the matrix of total (domestic plus imported) transaction coefficients.Hence, in this case, backward and forward linkages measure the impact of a one unit increase in final demand on total supply, rather than gross output, In this study, I will be using the technology matrix (I-A d ) -1 , because the imports coefficients are not available for all the Malaysian input-output tables [Department Statistics Of Malaysia, 2009].
The subsequent nine sections of this paper are structured as follows.Section 2 deals with the problem and objectives of the paper.Section 3 deals with the definitions of the input coefficients matrix, A, and output coefficients matrix, O. Section 4 deals with the interpretation of the input Leontief inverse, (I-A) -1 and the output Leontief inverse, (I-O) -1 .Section 5 examines with backward linkages and forward linkages.Section 6 gives empirical results of the linkage indices for the Malaysian economy.Section 7 offers some policy implications.In Section 8 some concluding remarks are made.

The problem
In Malaysia, as in most resources-rich developing countries, the availability of foreign exchange generated by the rapidly growing export of resources has been of great importance to the process of economic development.The aim of Malaysia development policy has been, primarily, to invest in the commodities sectors.The rationale behind this policy was to build a solid base for the Malaysian economy; by using the resources revenues (such as Crude Oil, Mining & Quarrying, Palm Oil, and Rubber products) to support the establishment of large scales enterprises, which could produce intermediate products at competitive prices for the other industries in the economy.This would thus aid the integration of the national economy.Secondary aims were to assist in income redistribution, import substitution, export growth and agricultural modernization.
Unfortunately, such a policy of inter-sectoral imbalance between economic sectors has lead to a poorly integrated economy in the short-run, causing a heavy dependence on imports.The presently existing weak forward and backward linkages between sectors are cited among the problems existing in the Malaysian economy.
In addition, the planners' policy towards the industrial sector regarding the adoption of advanced technology has resulted in production below its potential maximum in the short-run.This is because a number of structural "bottlenecks" developed, such as an insufficiently trained labour force and a lack of managerial and technical skills, as well as a heavy bureaucratic and hierarchical structure of organisation.

Objectives of the Analysis:
This study aims to assess the success or failure of Malaysian economic policy with input-output analysis.A static input-output model is used.Unfortunately, dynamic input-output models must be ignored, as the necessary capital matrix is not available for the Malaysian economy.The period of study is 1983-2000, during which time four input-output tables were established.The year 2000 is chosen as the closing year of study because this is the last year for which an input-output table is available.
It would be expected that in resources-rich developing economy, such as that of Malaysia, substantial structural change will take place over time.In particular, one might expect marked changes in the technologies employed, especially the nature of inter-industry trading.Also, change in the level and mix of final demand for produced goods would be expected to occur.One would anticipate that the role of state economic planning would be to facilitate and direct such developments.
Input-output analysis is well suited to the analysis of the nature of economic development through changing demand and changing technology.There are so many input-output techniques we can use to explore the ranking change of the sectors, such as linkages, multipliers, matrix triangularsation.Thus this study uses one of these techniques of input-output methods to explore the structural change of the Malaysian economy which is linkages analysis.It leads towards the conclusion that economic integration has occurred in Malaysia during the period of study.Also, there is evidence of increasing efficiency in the Malaysian economy.However, there remain substantial benefits from further integration which economic plans thus far have not exploited.

The Input Coefficients Matrix and the Output Coefficients Matrix
The input-output model describes two aspects of the relationships among participants in the production process.It can answer both:' Who receives from whom? ' and 'Who gives to whom?'.Accordingly, the structure of the relationships may also be approached in two ways.We may examine how much is needed of the output of preceding stages, or of the primary inputs, for some purpose (either for final use or for a unit output of some industry); this is the input approach.But we may examine what will come out of something, either of primary inputs or of the unit output of some industry, in successive stages or in final use; this the output approach.
These approaches describe the transactions of products and values in two opposite directions.One of them asks: 'Where do they come from?', the other: 'Where do they go?'.
For the purpose of the analysis of these linkages, I will use the following definitions.
First, the input coefficients matrix, A, can be used to analyse backward linkages, (i.e., intermediate inputs as a share of total inputs, including value added).Mathematically we can represent the input coefficients matrix, in element form; as: This is as earlier defined in the literature (see LEONTIEF, 1963 and1966).But we can rewrite it in matrix form: Here, z ij is intermediate demand, x j is total demand, and x ˆ is the diagonalised vector x, as a matrix; Z is the matrix of intermediate transaction.
Second, several authors [BULMER-THOMAS, 1982;NUGENT, 1973;et. all] have suggested that an alternative point of view can be taken with the basic input-output model.This alternative relates sectoral gross production to the primary inputs (that is, to a unit of value entering the interindustry system at the beginning of the process).This approach is made operational by essentially transposing our vertical (column) view of the model to a horizontal (row) one.Instead of dividing each column of Z by the gross input of the sector associated with that column, divide each row of Z by the gross output of the sector associated with that row.We used O to denote the direct output coefficients matrix that results.The output coefficients matrix, O, can be used to analyse forward linkages.(i.e., intermediate sales as a share of total sales including final demand).We can define the output coefficients matrix, O, by: Or in matrix notation: The input and output coefficients matrices for Malaysian Economy are available with author.

The Input Leontief Inverse and the Output Leontief Inverse
In this section, I should explain the meaning of the input Leontief inverse, (I-A) -1 , and the output Leontief inverse, (I-O) -1 .Briefly, the former is based on a matrix of technical input coefficients A. The latter uses technical output coefficients O.
First, the input Leontief inverse, (I-A) -1 , elements may be interpreted [JONES, 1970;p.325]as follows: (1) The elements of (I-A) -1 represent the increase in output of the ith industry to supply the inputs required for a unit of final demand in the jth industry.
(2) The ith row sum of (I-A) -1 is the increase in total output of the system required to utilize the increase in output from an initial unit of primary input into industry i.
(3) The column sums of (I-A) -1 represent the increase in total output of the system required to supply inputs for initial unit increase in final demand from each industry j.Second, the output Leontief inverse, (I-O) -1 element may be interpreted as follows: (1) The elements of (I-O) -1 represent the increase in output of the jth industry required to utilize the increase in output brought about by a unit of primary input into the ith industry.
(2) The ith row sum of (I-O) -1 is the increase in total output of the system required to utilize the increase in output from an initial unit of primary input into industry i.
(3) The column sum of (I-O) -1 , like the row sum of (I-A) -1 , has to do with the effect of a unit expansion of primary inputs into (or for final demand, from) all industries.The input and output inverse matrices for Malaysia are available with author.For the purpose of this paper; I will define the elements of the input Leontief inverse matrix to be c ij and the elements of the output Leontief inverse matrix to be v ij .

The Data and Methodology
Basically, the present study uses secondary data based on the four input-output tables compiled for the Malaysian economy so far.These tables were produced by the Department of Statistics.For analytical and comparable purposes, the original input-output tables consisting of different number of sectors are aggregated into 39 sectors based on International Standard Industrial Classification (ISIC).These sectors are shown in Table 1.

Backward linkages
The backward linkage effect allows one to find the dependence of one industry on other industries in respect of the supply of inputs.It measures the extent to which one industry utilizes the outputs of industries.This implies that for a sector with a high backward linkage effect, by increasing the output of the specific industry a powerful stimulus is set into operation in other industries, to increase the outputs of those industries.The aim of this section is to measure the potential for other activities resulting from investment in any sector.One possible measure of direct backward linkage from the input coefficient matrix, A, is the sum of the column elements [CHENERY, 1958;p.492].i.e. s j = A ' i (5) s j will measure the ratio of purchased inputs to the value of total production x j , and i is the unit (summation) vector.I show the results of this measure for the Malaysian economy for 1983-2000 period in the Table 2. Table 2 These show the direct backward linkages, derived from Equation ( 5).The value of direct backward linkages determines the values of input percentage of the value of production in these sectors.The remaining input value is attributable to factors used in other establishments.
The key points to note from Table 2 are the significant change in ranking of the most sectors except the Oils and Fats product.This sector is kept its ranking with high ranking for all tables under study.Agriculture, it will be noted, does not exhibit a significant long-term change in its ranking.Crude oil & Mining and Quarrying is mostly ranked last in all years.The ranking of this sector decreases markedly in the post 1983 period.But this only measures direct backward linkages and takes no account of the indirect stimuli given to the economy if investment takes place.This measure has three deficiencies [JONES, 1970;p. 324]: double counting of causal linkage, neglect of indirect impacts, and failure to distinguish the domestic effect from those operating on foreign economies.The first problem is that in an input-output framework, sales of industry A to industry B are recorded as A's forward linkages and B's backward linkages, but only one of these can be effective in a causal sense.Causality is at the root of the HIRSCHMAN hypothesis using input-output interdependence as a proxy for linkages [JONES, 1970;p.325].
To measure both the direct and indirect effects, we need the LEONTIEF inverse matrix (I-A) -1 (the input Inverse).We can get direct and indirect backward linkage for any sector j by the sum of the column elements of input inverse [YOTOPOULOS and NUGENT, 1976;p.335],as: We see that l j is the sum of the elements in column j of the LEONTIEF inverse.Now each element in column j measures the direct and indirect impact of the inverse of one unit in the final demand for industry j on each of the n industries.It must be noted that Equation 6 would be used also as a multiplier (see BEKHET, 2009).The results for the Malaysian economy are shown in Table 3. Table 3 The comments made above about the significant changes in ranking shown in Table 2 are equally applicable to Table 3.Typically, these elements are defined in terms of gross output values, and l j is then the aggregate or economy-wide gross output generated by an increase of one unit in final demand in industry j.However, a normalization procedure is often carried out, by comparing the average stimulus created by sector j with the overall average [RASMUSSEN, 1957;pp.133-140].The direct and indirect backward linkage index then becomes: The numerator denotes the average stimulus imparted to other sectors by a unit's worth of demand for sector j.The denominator denotes the average stimulus for the whole economy when all final demands increase by unity.Equation ( 7) has been applied to the input-output tables for the Malaysian economy.The results are shown in Table 4. Table 4 As noted with Table 3, the comments on the changes in ranking applied to Tables 2 still apply when Table 4 is examined.The difference between l j , as defined in Equation ( 6), and q j , as defined above, is the normalization in the latter by the number of sectors and by the double sum of columns and rows.Since the number of sectors and the double sum are obviously the same for any one country, q j is simply perfectly correlated with itself after normalization by a constant [YOTOPOULOS and NUGENT, 1976;p.340].It follows that q j >1 implies a jth sector where investment yields above average backward linkages, while the opposite is true for q j <1.
When q j >1, it means that an industry would need a comparatively large production increase to cope with one unit increase in the final demand for the product of industry j.The economic interpretation of q j >1 would be that the industry j would draw heavily on the rest of the industries, compared with other industries.On the other hand, q j <1 means that the industry j does not draw heavily on the rest of the industries.This measure was first devised by RASMUSSEN [1957], as the index of the power of dispersion (corresponding to the index of backward linkage).It is worth noting that this measure pre-dated ideas about the role of linkage in industrial development strategy, and was simply regarded as useful summary measure of the structural interdependence of an economy [McGILVRAY, 1977;p.50].Reference to the ranking of q j alone would not be sufficient to assist industrial planning, for a number of reasons.A high index could have been achieved, although only one or two sectors stand to gain from the backward linkages created by the investment.This can be taken into account by considering the dispersion of the stimuli according to the formula for the coefficient of variation: This equation has been applied to the input-output tables for the Malaysian economy, for the period under study.These results are shown in Tables 5. Table 5 The changes in ranking shown in Table 5 show some variation on those revealed in Tables 2-4.The ranking for Crude oil, Mining & Quarrying sector is moving up for all years, rather than at the last, whilst Oils & Fats product kept its rank at the first.There is significant change of the ranking for the most sectors for all tables.Agriculture, however, shows remarkable change over time.A low J j means that the investment in sector j would stimulate other sectors in an even manner, while a high J j means that the benefits of the stimuli provided by backward linkage would be unevenly shared [BULMER-THOMAS, 1982;p,191].On the other hand, in that case a relatively high value of J j can be interpreted as showing to what extent a particular industry draws heavily on one or a few industries.Thus, a low value of J j can be interpreted as that a particular industry draws evenly on other industries.

Forward linkages
The basic idea of forward linkage is to trace the output increase which occurs, or might occur, in using industries when there is a change in the sector supplying inputs.The forward linkage effect measures the dependence of one specific industry on other industries, in respect of the supply of its output as inputs to these industries.For an industry with a high forward linkage effect, it implies that by expanding the output of a specific industry a powerful stimulus is generated in other industries, by way of absorbing the output of the specific industry as inputs to other industries.The meaning of direct forward linkage may be derived from the output coefficient matrix O.The direct forward linkage is the sum of the row elements of O matrix [YOTOPOULOS and NUGENT, 1973;p.161]:s i = O i (9) Here, s i denotes the ratio of intermediate demand to total demand, x i , for a given product.These ratios for the Malaysian economy are shown in Table 6.Table 6 The key points to note from Table 6 are the changes in ranking of the most sectors were fluctuating during the period under study.But there is some sector still keeping their ranking, these are Animal feeds product; Oil Palm Primary product; Electricity & Gas; Health; Education; Real estate & Ownership dwelling; Hotel & Restaurant; Building & Construction; and Furniture & Fixtures sectors.Agriculture, it will be noted, does have a decreasing significant long-term change in ranking.Once again, Animals Feeds product is ranked first in all years.While the ranking of the Health and education sectors are last in all tables.However, this only measures direct forward linkage, and takes no account of the indirect stimuli given to the economy if the investment goes ahead.The measurement of direct and indirect forward linkage effects may be derived from the output inverse (I-O) -1 , using the technical output coefficients matrix O (intermediate sales as a share of total sales including final demand), [CARTER and BRODY, 1970;pp.252-253].We can get direct and indirect forward linkages from the sum of the row of the output inverse indicate forward linkage.l i = (I-O) -1 i (10) The (I-O) -1 indicates the increase in the output of the sector i needed in order to cope with a unit increase in the final demand for the product of each industry [BOUCHER, 1976;p.314].The results for the Malaysian economy are shown in Table 7. Table 7 The comments made above about the significant changes in ranking shown in Table 6 are equally applicable to Table 7.
High forward linkages occur when a sector's output is, or could be, used by many other sectors as an input.By expanding capacity in such a sector, inducements are provided to using industries which now have an incentive to expand output, to take advantage of the increased availability of inputs.Given our interpretation of the ijth element of the output inverse, a suitable measure of forward linkages might therefore be the row sum of this inverse, which becomes: This equation has been applied to the input-output tables for the Malaysian economy.These results are shown in the Tables 8. Table 8 The comments on the changes in ranking also apply to Tables 6 and 7.The only change we can note is that the Oil Palm sector became second in ranking for 1983, 1987, 1991 and 2000 tables.It is apparent that q i >1 implies a sector with high forward linkage.It would mean that the industry i, in general, would have to increase its output more than the rest of the industries for a given increase in final demand on the system of industries, while the opposite is true where q i <1.The index q i (i = 1, 2, …, n) is thus termed the index of sensitivity of dispersion of the industries under consideration.The numerator in Equation ( 11) refers to the ith row sum of the Leontief inverse, which in turn measures the total impact on sector i when the final demand for all sectors increases by unity.If this impact is large, it suggests that increased investment in sector i would induce output increases in all using sectors, as users take advantage of the increased availability of inputs.It might seem, therefore, that q i is a good measure of forward linkages.This measure was first devised by RASMUSSEN, as the 'Index of Sensitivity of Dispersion' (forward linkage).But this measure, according to the ranking of q i alone would not be sufficient to determine industrial planning.Another possibility also suggested by Rasmussen [RASMUSSEN, 1957;pp.138-139] it to look at the variance associated with each industry as: I have applied this equation to the input-output tables for the Malaysian economy tables for 1983-2000 period.The results are shown in Tables 9. Table 9 The changes in ranking shown in Table 9 show some variation on those revealed in Tables 6-8.Oil Palm Primary is fluctuating in the ranking for all years rather than second, whilst Oils & Fats product sector moves up to first in all years.The ranking of the Agriculture and Industry sectors move down in most years.If we compare the ranking of health, Education and other serves sectors with previous tables, we can see a significant change for these sectors.A high value of J i can be interpreted as showing to what extent a particular industry draws heavily on one or a few industries.A low variance shows that the system of industries draws relatively evenly on industry i and it might be concluded that in this case the row sum might be a reliable indicator of forward linkages.This is not the case, for the problem is not the dispersion of sales across industries, but the existence of sales that are a large share of a small industry.Thus a unity J i , indicating sales to all industries, could still give distorted row sums if those sales represented a large share of inputs into small industries [JONES, 1970;P.326].This measure of forward linkage is quite different from the backward linkage, because it measures the forward linkage as the increase in output of all using industries, rather than as the increase in output of the (one) supplying industry.

Results Analysis for the Malaysian Economy
To measure the linkage effects of the industrial sectors, the empirical results of the linkage indices are constructed in the framework of inter-industrial production relations.The data used for the construction of the indices are the input-output transaction coefficients matrices for the Malaysian economy.In order to identify the high backward and forward linkage effects of sectors, the industrial sectors with q j > 1 and low J j , and with q i > 1 and low J i are selected and shown in Tables 10 to 13. Tables 10-13 The input-output table for 1983 shows that there were sixteen sectors with high backward linkage effects.Of these sectors, one was Oils & Fats products and Foods production other Industries and the remaining fourteen were non-agricultural sectors.Next, there were nineteen sectors with high forward linkage effects, of which the highest ranking was the Animal Feeds product sector.The second ranking was the primary producing sectors such as Oil Palm primary products.The remaining seventeen sectors were the non-agricultural sectors.The two primary producing sectors and the sector of Non-Electrical Machinery and Equipment did not show a backward linkage effect.Agriculture appeared to be very weakly linked to the national economy, giving rise to the suspicion that it was an enclave sector.It appears that the Agriculture sector, with its potential importance for import substituting and export promoting industries, had few links with the national economy.
However, the other three tables for 1987, 1991 and 2000 show the impact of a planning policy that paid greater attention to the structural sector change.From these three tables it will be noted that structural change becomes in most sectors linked to the national economy in the post 1983 era.But I think this change still far away from planners' targets.As can also be seen from Tables 10-13 Manufacturing Industries, and these sectors seem to have a fluctuating position during the period under study.Sectors with a high forward linkage effect and a high backward linkage effect could be regarded as key sectors of the Malaysian economy in the period under study.In addition, these sectors should be given high priority by planners in investment planning.These sectors for 1983, 1987, 1991 and 2000 tables are shown in Table 14.Table 14 In tables 15-18, I have presented the matrices of Rank Correlation Coefficients among eight alternative linkage indices, including all the indices defined above.These results are based on the four input-output tables for the Malaysian economy, for which all eight indices, have been calculated from the original input-output tables.An examination of these matrices of Rank Correlation Coefficients shows that some of the indices (Coefficients of Variation) are quite unrelated, for all Indices.Also, the relations between s j and s i ; l j and l i ; q j and q i are uncorrelated.Note, in particular, that the backward and forward indices, q j , s j , l j ; and q i , s i , l i are correlated, respectively with the indices which I have used above.Therefore, the main result of this analysis is that the integration degree between demand and supply side for the Malaysian economy still remain weak.Tables 15-18 In fact, given the nature of the key sectors ( and the emphasis on their spread effects), it may well be that the faster growth rates may be found in other sectors not identified as key sectors in Tables 10-13.BLUMENFELD [1955] noted the same problem whilst discussing the economic base model.Such sectors may be those with the greatest potential for achieving import substitution [HEWINGS, 1982].It is expected that such sectors will be reflected as key sectors in the statistical data of coming years.

Policy Implications
The theoretical basis and aims of Malaysian planning policy since 1980 have been discussed in Section (2) and the details in [CHING, 2005].To briefly summarise, the main aim of the planners was to develop the commodities sectors and integrate them with the rest of the economy.It would therefore be expected that the indirect linkages for these sectors would have a high ranking in terms of backward and forward linkages.The results shown in Tables 10-13, and discussion in the previous sections (5 and 6), show how far this policy has been successfully achieved.The tables show that although some progress has been made, it falls far short of what the planners desired.The linkages between the commodities sectors and the rest of the economy still remain weak.There is still a high dependency on the primary sectors, such us Oil Palm, Rubber Primary products, and Crude Oil, Gas, Mining & Quarrying, and Wooden Sectors.Unfortunately, however, the primary sectors remain a classic example of an enclave export-oriented industry, superimposed on an entirely different type of economy, without any significant economic linkages between it and the rest of the economy.Agriculture, however, has been one area where planning policy has had some success in establishing linkages.But it had low backward linkages because its cost (input) structure is dominated by non-wage costs paid to factors of production.Also, it had low forward linkages since most of its output goes to private consumption.The main results of the policy were to transform Malaysia from an exporter to an importer of foodstuffs and other agricultural products [www.upe.gor].In addition, the declining rate of growth in the Agriculture sector was the most profound factor in widening economic inequalities between urban and rural areas during the period of the plans.The failure in agriculture resulted in rural income remaining low, especially jungles areas.Policy emphasised the domestic substitution of some of the growing volume of imported products, and in particular those of oil and rubber derived products.This implies that the planners should have undertaken the construction of a number of industrial projects linking commodities sectors to the consumer, either in the form of final consumable products, or in the form of intermediate goods utilised by other sectors.This, however, did not take place to any great extent.
The fact that not all potential linkages can be translated into actual linkages suggests the need for a modified form of linkages analysis, in which technological coefficients are adjusted for those growth stimuli which are not feasible for Malaysia; such non-feasibility will be determined by considerations of market-size, efficiency, comparative costs, natural resources, etc.The inducement which remains, as measured by the backward and forward linkages analysis, would then be a better guide to the selection of 'key' sectors.Furthermore, sectors with high backward linkages have a high dependence on intermediate goods, which are typically capital-intensive.In the context of DCs, particularly Malaysia, we are therefore asking planners to give priority to sectors which directly or indirectly are capital-intensive; although the argument over choice of techniques is complex; this is not a position to which most LDCs would want to be committed.However, one needs to consider a more fundamental set of objections to linkage analysis based on economic theory.Industrialization is not usually considered as an objective in itself, but as proxy for the rise in real income which is supposed to accompany it.If, however, we consider real income growth per head as our objective, then each investment needs to be evaluated in terms of its direct and indirect income (not output) impact, which can be done by reference to the income multipliers introduced in BEKHET [2009] .

Conclusions
In this paper I have made an effort empirically to identify key sectors and structural changes in the Malaysian economy during the period 1983-2000, using the input-output tables for this period.In closing it may be appropriate to allude to a possible limitation of this study.The input-output relations used above assume that a given output requires inputs in fixed proportions, so that the production structure in various industries or groups of industries is fairly stable.This may be true of the modern sector industries in an underdeveloped economy, but it is well-known that the primitive sector in such economies is marked with variable coefficients with a high degree of substitution among inputs [BOUCHER, 1976;p.318].A high degree of aggregation may fail to reveal the true pattern of linkages in such an economy [see BULMER-THOMAS, 1982, Ch.12].To the extent that the modern sector dominates the primitive sector in such economies, the linkage value calculated above represents mostly the situation in the modern sector.In view of this, one is not quite sure if the values of linkages calculated above encompass both the primitive and the modern sector in these years, or the modern sector alone.Needless to say, it would be most desirable to utilize a large number of tables and to extend the time series.Yet the concept of linkages is a powerful tool in the economies of development.These results although admittedly tentative, indicate that the linkage indices merit further attention and empirical research.In this paper, I have explored only one of the input-output techniques to measure the success of development in Malaysia.The next paper uses multipliers technique to measure the success of development policy.Sector Names: as shown in Table 1.
, some sectors did change a great deal as a result of post 1983 changes in planning policy.i.e., Paper & Printing products, Basic Metal; other Transportation Equipment; Wholesale & Retail Trade; Real Estate & Ownership dwelling.In fact the ranking of the Crude Oil/Gas/Mining & Quarrying sector in the Malaysian economy during 1983-2000 has changed.The main differences are for commodity sectors, i.e.

Table 4 .
Index of Power Dispersion of Backward Linkages, q j .

Table 5 .
Coefficient of Variation of Backward Linkages, J j .

Table 6 .
Direct Forward Linkages, s i .

Table 7 .
Indirect Forward Linkages, l i .

Table 8 .
Index of Sensitivity of Dispersion of Forward Linkages, q i .

Table 9 .
Coefficient Variation of Forward Linkages, J i .

Table 10 .
Backward and Forward Linkages for 1983.

Table 11 .
Backward and Forward Linkages for 1987.

Table 12 .
Backward and Forward Linkages for 1991.

Table 13 .
Backward and Forward Linkages for 2000.