A structural analysis of a regional economy using Social Accounting Matrices: 1990-1999

© Investigaciones Regionales. 5 – Páginas 113 a 138 Sección ARTÍCULOS A structural analysis of a regional economy using Social Accounting Matrices: 1990-1999* M. Carmen Lima1, M. Alejandro Cardenete1, G.J.D. Hewings2y José Vallés Ferrer3 ABSTRACT: Social accounting matrices (SAM) are an instrument that enlarges the information provided by the input-output analysis. These matrices study the intersectoral relationships of an economy, the behaviour of the consumers, the public sector or the foreign sector, as long as they complete the income flow of rent. In this work, we use the SAM for Andalusia (region southern Spain) 1990, 1995 and 1999, to conduct a structural analysis of the Andalusian economy by means of the «path analysis» methodology and a multiplier decomposition. With these techniques, we obtain the changes in productive structure and we quantify the influence of sectoral shocks on this regional economy. Finally, we also identify which sectors have most strongly contributed to the regional economic activity in the last decade. JEL classification: C67, D57, R15. Key words: social accounting matrix, regional accounting, structural analysis. Un análisis estructural de una economía regional a través de matrices de contabilidad social: 1990-1999 RESUMEN: Las matrices de contabilidad social (MCS) son un instrumento que permite ampliar la información proporcionada por el análisis input-output al recoger además de las relaciones intersectoriales de una economía, el comportamiento de los consumidores, el sector público o el sector exterior, logrando así completar el flujo circular de la renta. En este trabajo utilizamos las matrices de contabilidad social para Andalucía correspondientes a los años 1990, 1995 y 1999, elaboradas en trabajos previos. Con dicha información realizamos un análisis estructural de la economía an- * The first two authors wish to thank the support from XT 2002-00037 and centrA (Fundación Centro de Estudios Andaluces). The second author also thanks to SEC 2003-05112/ECO and FEDER. Finally, we thank to A. Manresa, F. Sancho and two anonymous referees, for their helpful comments. All errors are our responsibility. 1Universidad Pablo de Olavide. 2REAL-Universidad de Illinois (Urbana-Champaign). 3Universidad de Sevilla. Dirección para correspondencia: Department of Economy and Business;Universidad Pablo de Olavide at Seville,Ctra. Utrera, km 1. 41013-Sevilla, Spain, e-mail: [email protected] Recibido: 3 de marzo de 2004 / Aceptado: 25 de noviembre de 2004. 113 05 Lima 7/6/05 08:37 Página 113 daluza mediante la metodología denominada paths analysis y la descomposición de multiplicadores. Con estas técnicas obtendremos los cambios experimentados en la estructura productiva y cuantificaremos la influencia que han ejercido los shocks experimentados por cada sector sobre sí mismos y sobre el resto de sectores de esta economía regional. Finalmente, nos preguntaremos qué sectores han contribuido en mayor medida a la activación económica regional. Clasificación JEL: C67, D57, R15. Palabras clave: matrices de contabilidad social, contabilidad regional, análisis estructural. 1. Introduction Social accounting matrices (SAM) are databases comprising economic transactions which enable us to extract information on the different economic agents such as the producers, the consumers, the government and the foreign sector; as well as on the behaviour of the productive factors. They complete the information provided by the input-output analysis, whose limitations have been deeply discussed in the literature1, with the regional or national accounting and the surveys of family constraints, among other databases. The interest on SAMs is based on the fact that not only do they study the production relationships among the economic sectors but also the transactions that take place among the different institutions of an economic system in terms of revenues or consumption. Besides their statistical content, the SAMs have became a useful tool for the evaluation of interventions from the political economy in national or regional frameworks. If a SAM is available for more than one year, it is feasible to carry out a complete analysis of the productive structure of the economy and also to obtain a perspective of the changes that have occurred. Several methodologies are able to outline such analysis in a particular economy. In section two, we present a methodology based on a three-dimensional landscape2called «structural path analysis». Through this methodology, we can extract the main tendencies in the behaviour of an economy and we can also develop its corresponding structural view. For this purpose, we derive a hierarchy of the economy by way of the calculation of two types of indexes: the «absorption effects» or forward linkages and the «diffusion effects» or backward linkages. In section three we analyse interdependences and decompose the backward and forward linkages in own, open and circular effects following a classical multiplier decomposition. In section four we include an employment multiplier because we consider that this real variable will provide valuable information in terms of elasticity bet114 Lima, M.C., Cardenete, M. A., Hewings, G. J. D. y Vallés, J. 1See in this respect Roland-Holst, D.W. (1990). 2For more details, see Hewings, G.J.D., Sonis, M. et al. (1997), or Sonis, M. et al. (1997), about the economies of Chicago and Indonesia respectively. 3See Cardenete, M.A. (1998), and Cardenete, M.A. and Moniche, L. (2001), respectively. 05 Lima 7/6/05 08:37 Página 114 ween economic activity and capacity of employment creation. In section five, we come up with an empirical application of both methodologies on the SAMs for Andalusia in the years 1990 and 1995, elaborated in previous works3. We will also present a first approach for 19994. This exercise will point out the key sectors of the regional economy, the type of interrelationships and the nature of linkages inside it. Having these multipliers for three different databases, we can extract conclusions for each year, and get a perspective of the evolution along the whole period. Finally, we outline the main conclusions. 2. Methodology: Structural Path Analysis and Multiplier Product Matrix The SAM accounts are divided in two blocks: exogenous and endogenous. The classification in one group or another will depend on the aspects that are to be studied. In this type of linear general equilibrium models, it is possible from a mathematical point of view, to consider all the variables except one as endogenous variables (those whose rent level or production we want to explain). Nevertheless, it is not very realistic to build a model without recognizing as exogenous those variables that are determined outside of the productive system, or those that are used as instruments of the political economy (such as taxes, subsidies, transfers, public expenses,...) since in fact, the changes in these ones will determine the behaviour of the endogenous variables. To carry out the structural analysis of an economy, and to know what type of linkages work inside it, we should observe the changes in the intermediate flow levels among sectors. Following Hewings and Sonis (1997), we use an instrument to study the interrelationships of an economy by means of the calculation of a «Multiplier Product Matrix» (MPM), which we get from the SAM multipliers matrix. If we reorder the sectoral relationships according to their importance, we can analyse how a change in the final demand of a sector, affects the final demand of the economy (diffusion effect or backward linkage). We can also interpret how a change in the rest of sectors influences one in particular (absorption effect or forward linkage). These effects provide a clear orientation about the key sectors in the growth of an economy. They are useful to design performances about political economy as well, as they are supported by their high multiplier effect and the important influence of such interventions. To analyse the sectoral interdependences in an economy, we calculate the Multiplier Product Matrix, MPM, starting from the average tendency matrix of the SAM. We identify these matrices by a subindex, t, according to the base year (A90, A95 and A99 in A structural analysis of a regional economy using Social Accounting Matrices: 1990-1999 115 4This first version has been calculated by the application of an updating technique called CEM (Cross Entropy Method) on the SAM for Andalusia 1995, carried out by Cardenete, M.A. and Sancho, F. (2004). Using this methodology, we can introduce known information inside the cells of the estimated SAM (prior information), letting us to use it for structural analysis because there are changes in the technical coefficients (see Robinson et al. (2001)) 05 Lima 7/6/05 08:37 Página 115 this case). These matrices have been calculated by dividing every SAM column vector by the corresponding sum of that column, being nthe number of endogenous variables (the productive sectors, the production factors and the consumers).We calculate the associate inverse matrix Bt= (I– At)–1, being Ian n ×nidentity matrix. The sub-indexes i, jmake reference respectively to the rows and columns of the corresponding matrices. Following the path analysis methodology, we derive two vectors of multipliers, where each element corresponds to the sum of a column or a row respectively: B.j = 冱 n i=1 bij j= 1...n[1] Bi. = 冱 n j=1 bij i= 1...n[2] being bij components of the associated inverse matrix Bt. Next we define the Multiplier Product Matrix as the product of the row and the column multipliers corrected by a factor that we call the «global intensity» (V), which corresponds to the sum of all the elements of the associate inverse matrix: MPM = ᎏ V 1 ᎏ | Bi.B.j | i, j= 1...n[3] where V= 冱 n i=1 冱 n j=1 bij [4] This new matrix will identify those sectors whose structural connections generate a higher impact than the average upon the rest of the economy, whether they experience a change in their own sector or as an answer to changes detected in the rest of the system. Rasmussen (1956) and Hirschman (1958) classify these sectors as «key sectors». In short they include two indexes: • Diffusion effect or backward linkage, BLj: BLj=j= 1...n[5] • Absorption effect or forward linkage, FLi: BLi=i= 1...n[6] The interpretation of these coefficients is as follows: if the backward linkage is greater than 1 (BLj greater than 100% in percentage terms), a unit change in the final demand of sector jwill generate an increase above the average in the global activity of the economy. If the forward linkage is greater than 1 (FLi greater than 100% in percentage terms), a unit change in all the sectors of the final demand will generate an increment above the average in sector i. A key sector is the one with both indexes greater than one. Bi. ᎏ ᎏ 1 n ᎏ V B.j ᎏ ᎏ 1 n ᎏ V 116 Lima, M.C., Cardenete, M. A., Hewings, G. J. D. y Vallés, J. 05 Lima 7/6/05 08:37 Página 116 3. Classical multiplier decomposition in Social Accounting Matrices The present paper is located into the multisectoral linear models, in which we assume the exogeneity of prices. We work with three databases corresponding to three SAMs for Andalusia. We consider as endogenous those accounts that are part of the economic interrelations determined outside of the economic system (the production factors, the productive sectors and y the private sector); while the exogenous ones are tools for the political economy (as the public sector, the foreign sector and the capital)5. The multiplier decomposition was initially proposed by Stone (1978) and Pyatt and Round (1979). Later on, Defourney and Thorbecke (1984) and again Pyatt and Round (1985) have been working on it. We also have spanish references as Polo, Roland-Holst and Sancho (1991), among others. It is also interesting to highlight the works from the regional point of view developed by Cardenete and Sancho (2003) for Andalusia, de Miguel, Manresa and Ramajo (1999) for Extremadura or Llop and Manresa (2003) for Catalonia. The formulation of these linear models of general equilibrium is as follows: Let ynbe: yn= (I – An)–1 ⋅ x= Ma ⋅ x[7] where ynis the column vector of the total rent of the endogenous accounts, Iis an identity matrix of order n ×n, An is the average tendency matrix of expenditure between the different endogenous accounts and x is the vector that collects the flows of rent that the endogenous accounts receive from the exogenous ones. A generic element of An as aij is interpreted as the expense carried out in ifor each unit of expense of the sector j. Ma is the so called Accounting Multipliers Matrix and an element maij indicates the effect that an exogenous unit of rent on an endogenous account j, generates on the rent of the endogenous account i. In other words, the interpretation would be how many monetary units of rent are generated in sector ibecause of the circular flow of rent when sector jreceives a unitary shock. If we sum up these values of Ma by columns, we get the total effect of an exogenous shock received by one account on the rest of the economic activity. This way, the account with the greatest multiplier value points out one sector with an important influence on the rest of the economy when it is involved in an economic development policy. The multiplier decomposition can be carried out in two ways: the additive or the multiplicative one. Both of them enable us to split the process of generation of rents in an economy. In this work we use the multiplicative procedure, which distinguishes among the own effects, the open effects and lastly, the circular effects. To start with, we outline the structure of the SAM that we are using. In our endogenous accounts we find the two productive factors (capital and labour —accounts (11) and (12), resA structural analysis of a regional economy using Social Accounting Matrices: 1990-1999 117 5Revising the literature, there are alternative classifications, for example the ones proposed by Polo, C., Roland-Holst, D. and Sancho, F. (1991) that endogenizes the capital account, or Llop, M. and Manresa, A. (2003) with a foreign sector endogenization. 05 Lima 7/6/05 08:37 Página 117 pectively—), the private sector represented by the consumers (13) and finally ten activity sectors [accounts (1) to (10)]. Our exogenous accounts, following the most common approaches in the literature are three: the public sector (14), the savings and investment (15), and the foreign sector (16). Following Pyatt and Round (1979), we have decomposed the matrix of accounting multipliers in other three matrices by means of a multiplicative expression. The first matrix is called matrix of circular effects (Ma3) and reflects the effect that an exogenous injection of rent generates on the very account due to the circular flow of the rent. The second matrix is known as the matrix of open or crossed effects (Ma2), and the elements of its main diagonal are identity submatrices. It shows the effects on the rest of accounts of a shock received by one particular account. Finally, we have the matrix of own or internal effects (Ma1), also known as matrix of transfers because the first element of the main diagonal is an identity submatrix (there are no transfers among the productive factors), the second shows the transactions among institutions and the later includes the interindustrial transactions, and is in fact the inverse of Leontief. To interpret the multiplicative decomposition in terms of relative importance of each element on the total effect, we can express it in an additive form as: Ma = I+ (Ma1–I) + (Ma2–I) Ma1+ (Ma3–I) Ma2Ma1[8] In the previous expression, the identity matrix enable us to discount the initial injection of rent of each of the effects, so that we work with a net multiplicative decomposition. 4. Employment Multipliers Moreover it is possible to calculate one more multiplier to extract the accounts that generate more employment when receiving a unitary exogenous injection of rent. The employment multipliers are the result, in the first place, of a new diagonal matrix that we call E. This matrix includes the quotients between the volume of employment and the total resources for each productive sector. In the second place, we multiply this matrix with the part of Ma that incorporates the rows and columns corresponding to the productive sectors (in our case the order of this matrix is 10 ×13). When increasing the rent of an endogenous account, we will obtain the effects of this change in the corresponding column of the partition of Ma and, by means of the diagonal matrix E, we convert this impact into number of jobs. This way the expression of the employment multiplier,Me, is the following: Me = E * Ma [9] An element meij, is the increment in the volume of employment of the sector i when the sector jreceives a unitary exogenous injection6. If we analyse the sum of 118 Lima, M.C., Cardenete, M. A., Hewings, G. J. D. y Vallés, J. 6Additional information about the employment multiplier and a comparison with other type of multipliers, is provided in Arango (1979). 05 Lima 7/6/05 08:37 Página 118 columns, we have the effect on the employment at a global level, which entails the reception of an exogenous monetary unit on a particular sector. As far as rows is concerned, they show the increment that the activity sector in question experiences in its employment if the rest of sectors receive the exogenous monetary unit. As we are dealing with very small figures in absolute terms, we proceed to the normalization of the multipliers based on the average values by row and column and total average value. We get the new results by following these steps: • We calculate the columns and row average values. • We derive the total average value by means of the sum of all the values of Me divided by the number of elements of Me. • We divide the average values by rows and columns by the total average value. If the result is greater than 1, the normalized figure indicates an employment multiplier over the average. This process enables us to carry out comparisons that can be easily interpreted. Accordingly, they can be used as a reference to contrast if a value is greater than what we consider an average reaction or not. Thereby, we get a classification of sectors that are able to transform their activity increments into new employment. 5. Empirical application 5.1. Structural Path analysis: backward and forward linkages After the MPM calculations for the three databases, we reach a classification following the definition of key sectors, by means of the analysis of the backward and forward linkages. We can select those cases in which an over-average reaction is expected for the whole economy due to a modification in a sector demand, or as a consequence of a demand change in the rest of the economy. The greatest forward linkage value in percentage terms is 312.50% for 1990 and it corresponds to «Consumers (13)». The one for backward linkages is 125.50% for «Commercial services (9)». Applying the MPM matrix, the greatest coefficient is precisely located in (13,9) position7. We can reorder the MPM so that the highest multipliers are located in the main diagonal in order to obtain a graphical representation of the MPM with this new sorting. In table 1, the backward and forward linkages have been calculated for 1990, 1995 and 1999 from the greater to the smaller values. In order to analyse the information from an aggregate point of view, we present one three-dimensional graphic —landscape—, for each period. These landscapes are drawn with the previously mentioned reordering for an easier comparison between one period and another. In the three landscapes in the Appendix at the end of the work, we observe an activity reduction in 1995 and a recovery that slightly exceeds the initial situation of 1990. These results show the better behaviour of the Andalusian economy for the last A structural analysis of a regional economy using Social Accounting Matrices: 1990-1999 119 7MPM calculation has not been included in the paper in order to avoid a wider Appendix, any consultation will be attended. 05 Lima 7/6/05 08:37 Página 119 SAM corresponding to 1999, on account of the recovery from the crisis of the first years of the nineties. From the values of the multipliers, the policy maker can derive the intensity of interactions between sectors and, also, the local demand interactions and moreover to develop strategic plans to increase the activity in the keysectors identified. 120 Lima, M.C., Cardenete, M. A., Hewings, G. J. D. y Vallés, J. Table 1. Backward and forward linkages 1990, 1995 y 1999 (in percentage terms) Andalusia 1990 Andalusia 1995 Andalusia 1999 Backward linkages Forward linkages Backward linkages Forward linkages Backward linkages Forward linkages BLj ranking Fli ranking BLj ranking Fli ranking BLj ranking Fli ranking 1st 9 125.50% 1st 13 312.50% 1st 9 129.02% 1st 13 318.90% 1st 9 136.53% 1st 13 367.74% 2nd 8 121.86% 2nd 4 177.23% 2nd 6 121.17% 2nd 12 186.13% 2nd 10 126.49% 2nd 12 215.08% 3rd 11 113.45% 3rd 12 152.34% 3rd 10 120.10% 3rd 4 151.09% 3rd 6 123.83% 3rd 11 126.81% 4th 12 113.45% 4th 11 139.55% 4th 8 114.63% 4th 11 148.80% 4th 11 121.14% 4th 6 108.21% 5th 10 109.59% 5th 6 117.87% 5th 5 113.63% 5th 6 119.61% 5th 12 121.14% 5th 8 96.61% 6th 1 108.00% 6th 8 81.00% 6th 12 111.46% 6th 8 73.00% 6th 5 108.21% 6th 9 81.73% 7th 5 107.40% 7th 7 65.03% 7th 7 102.14% 7th 9 59.80% 7th 3 103.43% 7th 4 71.70% 8th 7 101.33% 8th 2 58.28% 8th 3 100.74% 8th 7 55.16% 8th 8 101.36% 8th 3 51.50% 9th 6 95.44% 9th 1 51.17% 9th 1 100.45% 9th 3 50.21% 9th 13 95.29% 9th 7 51.41% 10th 13 92.84% 10th 9 49.85% 10th 13 88.86% 10th 1 44.39% 10th 7 94.09% 10th 1 35.78% 11th 3 84.91% 11th 3 42.79% 11th 11 88.39% 11th 2 35.83% 11th 1 92.77% 11th 5 31.79% 12th 4 72.78% 12th 5 31.57% 12th 4 70.78% 12th 5 33.38% 12th 2 40.67% 12th 2 31.63% 13th 2 53.45% 13th 10 20.82% 13th 2 38.63% 13th 10 23.70% 13th 4 35.06% 13th 10 30.02% Source: Own elaboration through SAMs for Andalusia 1990, 1995 and 1999. Note the meaning of key sector using the case of «Capital (12)» in 1990 as an example. By consulting Table 1, we can see that one change in the final demand of this sector generates an increase in the activity of the economy, that is, the rest of the sectors get a 13% above the expected average reaction. This fact means that when capital increases in the Andalusian economy, it generates a pulling effect in the rest of sectors even above its own experienced shock. This is called «diffusion effect» or backward linkage. As for the «absorption effect» or forward linkage, a change of one unit in the final demand of all the sectors produces an increase of «Capital (12)» activity of more than 52%; again above the average. We could conclude that capital strongly reacts in moments of economic good-behaviour and, it is also pushed by the rest of sectors to a larger extend than the average reaction. As the two previous behaviours are greater than 100%, «Capital (12)» account is classified as a key sector for the Andalusian economy in 1990. Other key sectors for this year are «Labour (11)» and «Consumers (13)» (although its «diffusion effect» did not exactly reach 100%, we consider that 92.84% is a high enough percentage, specially when the sector presents the highest «absorption effect» with a triple reaction over the expected average when reacting to an increase in the rest of the activity sectors). Finally, «Commerce (6)» also registers a similar behaviour to the consumers. In 1995 the figures for «Capital (12)» and «Commerce (6)» remain among the relevant sectors in terms of generation of economic activity. «Consumers (13)» are ta05 Lima 7/6/05 08:37 Página 120 ken out as a key sector because, although they show an even higher absorption effect than in 1990 (close to 318.90%), they continue the decreasing tendency of the diffusion effect of the previous period. A similar process appears in «Labour (11)» sector. In 1999, we highlight the growth of the «diffusion effect» and «absorption effect» in «Capital (12)» account that strongly behaves as a key sector for the Andalusian economy. «Labour (11)» recaptures its position of 1990 as a key sector, and «Commerce (6)» consolidates its place as a key sector. «Consumers (13)» recover positions ending up improving their capacity to influence the rest of sectors through increases in the demand. They also improve their capacity to take advantage of the expansion moments reflected in increments in the final demand of the rest of sectors. Finally, «Other services (8)» joins the group of growth accelerators in this region. We will now study those sectors that, although they do not behave as key sectors because they register a low forward linkage value, they certainly do have a great capacity to accelerate the economic activity when they experience a change in their own final demand, that is to say, they have a high «diffusion effect» or backward linkage. Such is the case of «Commercial services (9)», with the highest value in this category, and «Other Services (8)» in 1990; once again «Commercial services (9)» and «Non Commercial services (10)» in 1995; and, finally for 1999, «Commercial services (9)» which repeats again, conforming its position as a sector of high «diffusion effect» for the whole decadeWe get the same behaviour for «Non Commercial services (10)» in 1995. These data show the high relevance of services in the Andalusian economy, once we have confirmed the important influence of a demand increment both in private and public services, on the rest of activity sectors. If we focus on the sectors that exert the least impulse on activity when they experience an increase on their final demand, that is, those that are not able to transmit their growth to the rest because of their low «diffusion effect». We can highlight «Extractives (2)» and «Manufacturing industry (4)» for the three years. We would like to point out that the first sector keeps a specially marked downward tendency in 1995 which is still present in 1999. «Manufacturing industry (4)» which registered a 27% below average, also experiencing a drastic fall in the early nineties, concluding with an «diffusion effect» of only 35.06%, the smallest value among those registered in 1999. With this result we see the reduced capacity of the secondary sector to reactivate the Andalusian economy. Regarding the evolution along time of the sectors that generate important backward linkages, the decade shows that «Commercial services (9)»(where hiring house services and machinery renting are included) is in the first position throughout the whole time. This behaviour confirms the huge capacity to impulse the rest of the andalusian activity sectors along the nineties. «Other Services (8)» (financial intermediation services, insurance services and pensions), move from second position at the beginning of the nineties, to fourth place in 1995 and finishes the decade in eighth place, being an example of continuous descent. A similar behaviour is observed as for «Agriculture, Cattle & Forestry and Fishing (1)». The opposite case is true for «Commerce (6)», which begins in ninth position, getting second place by the middle of the nineties and reaching the top positions in 1999. We highlight the volatility of «Labour (11)» sector that moves from third place in 1990 to eleventh position in 1995, returA structural analysis of a regional economy using Social Accounting Matrices: 1990-1999 121 05 Lima 7/6/05 08:37 Página 121 (6)» seems to be more receptive in moments of good economic activity, which will entail the creation of new employment. Table A.2, presents the employment multipliers corresponding to 1995. Reading the figures of the columns, we find again the accounts of the precedent year with the exception of «Construction (5)» which registers a soft slope that locates it below the average reaction. Revising the data by rows, we find the same accounts of 1990. The employment generation capacity of «Commercial services (9)» with 2.7 times the average reaction, is only surpassed by «Commerce (6)». If in 1990 «Manufacturing industry (4)» registered values below our reference, during this year the situation becomes even worse. Finally, Table A.3 comprises the employment multipliers for 1999. To begin with the columns, the sectors that, when receiving an increment in their final demand of a monetary unit, are able to impulse the employment in the rest of activity sectors; are the same ones than in 1995. However, there is a new incorporation that can suppose an important change for Andalusian economic activity. For the first time in the decade of the nineties, the account of «Manufacturing industry (4)» experiences a change that makes it join the group of activity sectors that produce new figures of employment. As for rows, there are three accounts that are greater than the established standard and register multipliers greater than one: on the one hand «Commerce (6)» together with «Commercial services (9)» which is strengthened as a key sector in terms of employment, and on the other hand «Manufacturing industry (4)» that doubles the average. This way, the sector (4) is a key account for the employment creation in 1999 and we also consider it as a key sector for the regional economic planning. Following the evolution of the employment multipliers in the decade, we can summarize such information in table 2 and table 3: 128 Lima, M.C., Cardenete, M. A., Hewings, G. J. D. y Vallés, J. Table 2. Employment multipliers for Andalusia in the nineties. Column analysis. 12345678910 1990 1.542 0.222 0.468 0.553 1.057 1.202 0.769 0.842 1.278 2.936 1995 1.275 0.294 0.521 0.485 0.813 1.278 0.921 0.860 2.026 2.662 1999 1.317 0.575 0.573 1.021 0.891 1.496 0.834 0.753 1.612 1.566 ∆ 1995/90 –17% 33% 11% –12% –23% 6% 20% 2% 59% –9% ∆ 1999/90 –15% 159% 23% 85% –16% 24% 9% –11% 26% –47% Source: Own elaboration starting from calculation of the multipliers of employment 1990-95-99. Table 3. Employment multipliers for Andalusia in the decade of the nineties. Row analysis. 12345678910 1990 1.699 0.049 0.109 0.805 0.517 2.907 0.555 0.456 1.102 1.800 1995 1.101 0.213 0.092 0.539 0.219 2.323 0.665 0.548 2.711 1.587 1999 0.823 0.413 0.060 1.999 0.191 2.552 0.460 0.644 2.107 0.750 ∆ 1995/90 –35% 334% –15% –33% –58% –20% 20% 20% 146% –12% ∆ 1999/90 –52% 740% –45% 148% –63% –12% –17% –41% 91% –58% Source: Own elaboration starting from calculation of the multipliers of employment 1990-95-99. 05 Lima 7/6/05 08:37 Página 128 A structural analysis of a regional economy using Social Accounting Matrices: 1990-1999 129 We have calculated some variation rates from 1990 to 1995 and from 1990 to 1999, so we establish a comparison of the situation from the beginning of the period until the end. We found very heterogeneous behaviours, for example the important growth experienced by «Extractives (2)» which is followed by «Manufacturing industry (4)» although if we observe the data in detail, the take off of this multiplier remains stable from 1995 onwards. The rest of accounts register a moderate growth between 10% and 25%. Such is the case of (3), (6), (7) and (9). The most significant fall is a 47% of «Commercial services»(10). In relation to the evolution by rows, again we highlight «Extractives (2)», followed by «Manufacturing industry (4)» and «Commercial services (9)». Similar reductions take place in the multipliers of the primary sector, the energy production, the construction or the non commercial services. To summarize with, we have found relevant information from these multipliers. Firstly, there is one sector that behaves as a key sector both in terms of backward and forward linkages and employment. That is the case of «Commerce (6)». Any political decision focused on increasing the final demand of this sector —i.e. fiscal incentives for promotion of new firms able to widen the actual supply, together with subsidies for these initiatives or public investment on physical infrastructure for their establishment—, is expected to induce a very good reaction in terms of activity and employment of the regional economy. There is no doubt that the regional government has taken advantage of this opportunity, probably setting aside some structural reforms that could have improved the poor reaction of industrial sectors in terms of generation of value added. We have found a very stable group of sectors responsible for employment generation in the rest of the economy: the private and public services. Once again our results describe the Andalusian economy as a region where service activities monopolize the higher percentage of generation of rent and employment. We have also detected that the nice figures in terms of employment multipliers have became better all along the period, probably induced by a more flexible labour market framework that could be partially responsible for a more elastic behaviour of accounts like «Manufacturing industry (4)». 6. Concluding remarks The goal of combining fields of industrial concentration with a development strategy which takes advantage of the endogenous character of each region and its dynamics9, compels us to study those sectors that are able to generate growth and distribute the added value in a national or regional economy. In this work we have outlined a structural analysis of the Andalusian economy using Social Accounting Matrices. The temporal scenario considered was the decade of the nineties, and we have used the SAMs for the years 1990, 1995 and a first version for 1999. From «structural path analysis» methodology, we extract a graphical representation of a «three-dimensional landscape» that captures the structure of relationships 9For more information see Curbelo, J.M. (1988). 05 Lima 7/6/05 08:37 Página 129 among the productive sectors of the Andalusian economy. These linkages provide information to analyse the effect of a change in the final demand of a sector on the whole Andalusian economy or to measure the influence of the expansion of one sector on the rest of them. All the necessary information has been collected in the backward linkages or «diffusion effects» and the forward linkages or «absorption effects». Moreover, the results obtained for the Andalusian economy show that the productive factors, the consumers account and some commercial sectors, generate important multiplier effects on economic activity all through the decade, with the very small exceptions. From 1995 on, growth-employment elasticity decreased considerably (the labour factor was displaced to third place at the end of the ranking as regards generation of the «diffusion effects» in this year). It is also important to remark that «Construction (5)» stayed between seventh and fifth positions as regards the «diffusion effect», showing its capacity as an stimulator of the economic activity. It is important to outline that «Manufacturing industry (4)» is unable to work as a developer of economic activity on the Andalusian economy, ending up the decade with a very limited capacity of influence on the rest of sectors, even in moments when manufacturing demand increased. This weakness is even more remarkable if we keep in mind that its reaction in moments of optimal behaviour of the rest of the sectors becomes worser as time goes by. Such behaviour restricts the effectiveness of certain investment policies, due to the apparent rigidity of the secondary sector. As for services, they show a high «absorption effect» in the whole period. This result was expected in this research due to the weight of this sector in the Andalusian economy. We must point out the good behaviour of «Commercial services» (9) as well as «Non-Commercial services»(10) or public services. The sector with an exemplary capacity to generate added value is the one of «Commerce (6)», which includes tourist activities, since it is able to register huge linkages in both senses. In this work we have also introduced a methodology of classic multiplier decomposition from SAMs, by means of a multiplicative disaggregation that separates the net effects of an initial shock in the own, open and circular effects. The use of the SAM enables us to complete the information derived from Leontief technology, and quantify the importance of the feedback effects generated by the own circular flow of rent. In relation to the multiplier decomposition, we have come across with a fall of the total effects of 25% in the whole decade (the own net effects and the circular effects ended up registering a reduction of 60%). We have also noticed that in spite of this falling evolution, the circular effects keep the greater values, which proves the importance of the feedback that is taking place in the Andalusian economy. Furthermore, we have completed the analysis of multipliers with a fourth multiplier in terms of creation of new jobs. Later we have carried out an empiric application obtaining the corresponding multipliers for the three representative years of the decade of the nineties: 1990, 1995 and 1999. The results show the reaction of the private sector, the government or the savings and investments when facing a change in the final demand. We have also found interesting answers in relation to the elasticity of each of the productive sectors as for their capacity to generate employment. The employment key sectors are those that react generating employment above the average value when they receive an exogenous injection or when the rest of the economy 130 Lima, M.C., Cardenete, M. A., Hewings, G. J. D. y Vallés, J. 05 Lima 7/6/05 08:37 Página 130 experiences the shock. Such accounts remain invariable during the decade and they are the primary sector and services in general. It is important to underline that in 1999 we detect a new dynamic account, the one of «Manufacturing industry (4)» which for the first time becomes an employment key sector. Such a behaviour means a new stage for the secondary sector in Andalusia. It is important to remark that, employment keysectors, do not coincide with classical multiplier decomposition keysectors and the policy maker must balance whether to push one activity sector or another. The previous result makes us think if it is reasonable to finance the secondary sectors through regional policy funds like European Regional Development Fund (ERDF) or European Social Fund (ESF), specially once we have pointed out their rigidity. In fact, if we follow our previous reasoning, the region of Andalusia, —classified as Objective 1 for the european regional policy—, should concentrate on activities able to generate a high value added, as those we have highlighted, in order to get the best results from the european support. But in this case, the regional growth model would be very dependent on a few service sectors. In this sense, we think that, with the aim of reducing future obstacles to regional development, policy makers should direct their efforts in combining actions able to capture the higher multiplier effects, with others that improved the low reaction of inelastic sectors. Such decision would probably mean a redefinition of the priorities in the current regional policy. This long run bet, would derive in changes in intervention axes of Regional Development Plans (RDP) and Community Support Framework (CSF) for 2000-06 and in the one in current negotiation for 2007-2013. In conclusion, we consider that it is necessary to work with the results of this paper in a double sense: trying to catch up the advantages of multipliers and at the same time, designing an strategy to improve the behaviour of less dynamic accounts. Furthermore, the advantage of this work is that we can extract strengths and weaknesses of a regional economy. Obviously, we must keep in mind that we are working with the limitations of a linear model derived from a SAM. Of course, we must interpret our results under the caution of the limitation of the statistical databases. Moreover, the path analysis technique that we have developed, can be complemented with a prices model for a better understanding of the results. To deepen in other aspects as to asses the impact of an specific political decision —i.e. a change of the amount of European Funds received in Andalusia— in terms as consumer’s welfare, income, GDP or again price levels; we could develop a Computable General Equilibrium (CGE) model. In these models, we set a group of functional relations that describe the behaviour of different agents and then search for the corresponding solution. This more sophisticated tool would be helpful to polish the final results. CGE models advance information on the results that can be expected after an intervention, and they point out the prospective reaction of the most important regional economic linkages. To finish with, in this work we have outlined those keysectors of an economy which can be used to analyse problems of regional planning by means of lineal general equilibrium models from SAMs. The main objective has been to study the internal arrangements within the Andalusian activity sectors from an aggregate point of view, in order to determine their potentialities and weaknesses. This type of exercise can A structural analysis of a regional economy using Social Accounting Matrices: 1990-1999 131 05 Lima 7/6/05 08:37 Página 131 provide with ex-ante and ex-post exercises to the object of assessing the effects of choosing certain investment projects instead of others. Decisions of this type can condition regional growth in the long term, generating strangulations in the productive activity if an adapted development strategy is not properly designed. References Arango, J. (1979): «Multiplicadores derivados de un modelo input-output regional», Investigaciones Económicas, 5-26. Bosch, J. et al. (1997): Evaluación del Impacto Económico de la Construcción de la Red de Cable de Banda Ancha en Cataluña. Institut D’Estudis Territorials, Barcelona. Cardenete, M.A. (1998): «Una matriz de contabilidad social para la economía andaluza: 1990». Revista de Estudios Regionales, 52:137-153. Cardenete, M.A. y Moniche, L. (2001): «El nuevo marco input-output y la SAM de Andalucía para 1995». Cuadernos Ciencias Económicas y Empresariales, 41:13-32. Cardenete, M.A. y Sancho, F. (2004): «Sensitivity of simulation results to competing SAM updated». The Review of Regional Studies, 34 (1):37-56. Cardenete, M.A. y Sancho, F. (2003): «Evaluación de multiplicadores contables en el marco de una matriz de contabilidad social regional». Investigaciones Regionales, 2:121-139. Curbelo Ranero, J.M. (1988): «Crecimiento y equidad en una economía regional estancada: el caso de Andalucía (un análisis en el marco de las matrices de contabilidad social)». Investigaciones Económicas, XII (3):501-518. Defourney, J. y Thorbeke, E. (1984): «Structural Path Analysis and Multiplier Decomposition within a Social Accounting Matrix framework». The Economic Journal, 94:111-134. De Miguel, F.J.; Manresa, A. y Ramajo, J. (1999): «Matriz de Contabilidad Social y multiplicadores contables: una aplicación para Extremadura». Estadística Española, 40 (143):195-232. Hewings, G.J.D. y Sonis, M. et al. (1997): «The Hollowing-Out process in the Chicago economy, 19752011». Geographical Analysis, 30:217-233. Hirschman, A. (1958): The strategy of economic development. New Haven: Yale University Press. Kehoe, T.J.; Manresa, A.; Polo, C. y Sancho, F. (1988): «Una matriz de contabilidad social de la economía española». Estadística Española, 30 (117):5-33. Leontief, W. (1941): The Structure of American Economy, 1919-1924: An Empirical Application of Equilibrium Analysis. Harvard University Press, Cambridge. Llop, M. y Manresa, A. (2003): «Análisis de multiplicadores lineales en una economía abierta». Working Paper Serie de Economía E/2002/21. Fundación Centro de Estudios Andaluces (centrA). Pyatt, G. y Round, J. (1979): «Accounting and fixed price multipliers in a Social Accounting Matrix framework». The Economic Journal, 89:53-69. Pyatt, G. y Round, J. (1985): Social Accounting Matrices: a basis for Planning. The World Bank, Washington. Polo, C.; Roland-Holst, D.W. y Sancho, F. (1991): «Descomposición de multiplicadores en un modelo multisectorial: Una aplicación al caso español». Investigaciones Económicas, XV (1):53-69. Rasmussen, P. (1956): Studies in Inter-Sectorial relations. Einar Harks, Copenhagen. Robinson, S.; Cattaneo, A. y El-Said, M. (2001): «Updating and Estimating a Social Accounting Matrix Using Cross Entropy Methods». Economic Systems Research, 13(1):47-64. Roland-Holst, D.W. (1990): «Interindustry analysis with social accounting methods». Economic Systems Research, 2 (2):125-145. Sonis, M.; Hewings, G.J.D. y Sulistyowati, S.(1997): «Block structural path analysis: applications to structural changes in the Indonesian Economy». Economic Systems Research, 9:265-278. Sonis, M.; Hewings, G.J.D. y Guo, J. (2000): «A new image of Classical Key Sector Analysis: Minimum information decomposition of the Leontief inverse». Economic Systems Research, 12(3):401-423. Stone, R. (1978): The Disagreggation of the Household Sector in the National Accounts, World Bank Conference on Social Accounting Methods in Development Planning, Cambridge. 132 Lima, M.C., Cardenete, M. A., Hewings, G. J. D. y Vallés, J. 05 Lima 7/6/05 08:37 Página 132 A structural analysis of a regional economy using Social Accounting Matrices: 1990-1999 133 Table A.1. Employment multipliers for Andalusia in 1990 Accounts 12345678910111213 Sum Average Total Normaliz. of row value average value 1 0.950 0.023 0.049 0.116 0.095 0.092 0.080 0.094 0.099 0.087 0.108 0.108 0.108 2.008 0.154 0.091 1.699 2 0.002 0.028 0.007 0.002 0.002 0.002 0.003 0.002 0.002 0.002 0.002 0.002 0.002 0.058 0.004 0.049 3 0.004 0.002 0.079 0.003 0.004 0.004 0.004 0.005 0.005 0.006 0.004 0.004 0.004 0.129 0.010 0.109 4 0.074 0.019 0.036 0.173 0.089 0.059 0.069 0.068 0.071 0.070 0.075 0.075 0.075 0.951 0.073 0.805 5 0.016 0.007 0.015 0.009 0.413 0.015 0.015 0.020 0.022 0.021 0.020 0.020 0.020 0.611 0.047 0.517 6 0.233 0.068 0.153 0.131 0.233 0.795 0.200 0.243 0.272 0.234 0.291 0.291 0.291 3.437 0.264 2.907 7 0.036 0.017 0.028 0.023 0.039 0.033 0.245 0.039 0.041 0.040 0.038 0.038 0.038 0.657 0.051 0.555 8 0.024 0.010 0.018 0.016 0.028 0.026 0.026 0.221 0.039 0.042 0.030 0.030 0.030 0.539 0.041 0.456 9 0.060 0.026 0.038 0.031 0.056 0.065 0.055 0.074 0.609 0.061 0.076 0.076 0.076 1.303 0.100 1.102 10 0.003 0.002 0.001 0.001 0.001 0.001 0.001 0.002 0.002 2.108 0.002 0.002 0.002 2.128 0.164 1.800 Sum of columns 1.402 0.202 0.426 0.503 0.961 1.093 0.699 0.766 1.162 2.670 0.646 0.646 0.646 Average value 0.140 0.020 0.043 0.050 0.096 0.109 0.070 0.077 0.116 0.267 0.065 0.065 0.065 Total average 0.091 Normalized value 1.542 0.222 0.468 0.553 1.057 1.202 0.769 0.842 1.278 2.936 0.710 0.710 0.710 Source: Own elaboration through SAM for Andalusia 1990. Appendix 05 Lima 7/6/05 08:37 Página 133 134 Lima, M.C., Cardenete, M. A., Hewings, G. J. D. y Vallés, J. Table A.2. Employment multipliers for Andalusia in 1995 Accounts 12345678910111213 Sum Average Total Normaliz. of row value average value 1 0.598 0.006 0.023 0.072 0.046 0.042 0.031 0.033 0.037 0.038 0.028 0.038 0.038 1.033 0.079 0.072 1.101 2 0.004 0.130 0.023 0.009 0.007 0.004 0.003 0.004 0.004 0.004 0.003 0.004 0.004 0.200 0.015 0.213 3 0.003 0.001 0.060 0.002 0.002 0.003 0.002 0.003 0.003 0.003 0.002 0.002 0.002 0.086 0.007 0.092 4 0.036 0.007 0.024 0.106 0.058 0.041 0.033 0.034 0.036 0.033 0.026 0.036 0.036 0.506 0.039 0.539 5 0.004 0.001 0.002 0.001 0.171 0.004 0.002 0.003 0.004 0.005 0.002 0.003 0.003 0.206 0.016 0.219 6 0.131 0.024 0.105 0.068 0.131 0.584 0.137 0.138 0.179 0.169 0.141 0.191 0.191 2.179 0.168 2.323 7 0.023 0.010 0.023 0.019 0.033 0.042 0.313 0.028 0.030 0.028 0.020 0.027 0.027 0.624 0.048 0.665 8 0.019 0.005 0.021 0.014 0.024 0.032 0.025 0.226 0.031 0.041 0.021 0.028 0.028 0.514 0.040 0.548 9 0.099 0.027 0.092 0.057 0.110 0.166 0.114 0.147 0.135 0.166 0.116 0.157 0.157 2.543 0.196 2.711 10 0.004 0.001 0.003 0.002 0.003 0.004 0.003 0.006 0.005 1.444 0.004 0.005 0.005 1.489 0.115 1.587 Sum of columns 0.920 0.212 0.376 0.350 0.587 0.922 0.664 0.621 1.462 1.921 0.363 0.491 0.491 Average value 0.092 0.021 0.038 0.035 0.059 0.092 0.066 0.062 0.146 0.192 0.036 0.049 0.049 Total average 0.072 Normalized value 1.275 0.294 0.521 0.485 0.813 1.278 0.921 0.860 2.026 2.662 0.504 0.680 0.680 Source: Own elaboration through SAM for Andalusia 1995. 05 Lima 7/6/05 08:37 Página 134 A structural analysis of a regional economy using Social Accounting Matrices: 1990-1999 135 Table A.3. Employment multipliers for Andalusia in 1999 Accounts 1 2 345678910111213 Sum Average Total Normaliz. of row value average value 1 1.157 0.004 0.020 0.013 0.026 0.035 0.021 0.024 0.037 0.039 0.040 0.040 0.040 1.499 0.115 0.140 0.823 2 0.003 0.628 0.082 0.003 0.007 0.004 0.003 0.003 0.004 0.004 0.004 0.004 0.004 0.752 0.058 0.413 3 0.002 0.001 0.081 0.000 0.002 0.003 0.002 0.002 0.003 0.003 0.003 0.003 0.003 0.110 0.008 0.060 4 0.209 0.057 0.155 1.352 0.341 0.209 0.186 0.146 0.195 0.191 0.200 0.200 0.200 3.640 0.280 1.999 5 0.004 0.001 0.003 0.000 0.313 0.003 0.002 0.002 0.004 0.006 0.003 0.003 0.003 0.347 0.027 0.191 6 0.232 0.047 0.213 0.029 0.264 1.449 0.241 0.244 0.371 0.341 0.406 0.406 0.406 4.647 0.357 2.552 7 0.024 0.014 0.029 0.004 0.033 0.042 0.481 0.026 0.037 0.035 0.038 0.038 0.038 0.838 0.064 0.460 8 0.048 0.011 0.054 0.007 0.061 0.079 0.055 0.403 0.087 0.096 0.091 0.091 0.091 1.173 0.090 0.644 9 0.153 0.040 0.157 0.021 0.188 0.254 0.166 0.191 1.501 0.272 0.298 0.298 0.298 3.836 0.295 2.107 10 0.011 0.002 0.011 0.001 0.013 0.016 0.011 0.013 0.019 1.205 0.021 0.021 0.021 1.365 0.105 0.750 Sum of columns 1.844 0.805 0.803 1.430 1.248 2.096 1.169 1.054 2.257 2.193 1.102 1.102 1.102 Average value 0.184 0.081 0.080 0.143 0.125 0.210 0.117 0.105 0.226 0.219 0.110 0.110 0.110 Total average 0.140 Normalized value 1.317 0.575 0.573 1.021 0.891 1.496 0.834 0.753 1.612 1.566 0.787 0.787 0.787 Source: Own elaboration through SAM for Andalusia 1999. 05 Lima 7/6/05 08:37 Página 135 136 Lima, M.C., Cardenete, M. A., Hewings, G. J. D. y Vallés, J. Figure A.1. Landscape Andalusia 1990 (numeraire 1990) 9 Commercial Services 8 Other services 11 Labour 12 Capital 10 Non-commercial services 1 Agriculture, cattle & forestry 5 Construction 7 Transport and Comunications 6 Commerce 13 Consummers 3 Electricity and natural gas 9 Commercial Services 4 Manufacturing industry 2 Extractives 10 Non-commercial services 5 Construction 3 Electricity and natural gas 1 Agriculture, cattle & forestry 2 Extractives 7 Transport and Comunications 8 Other services 6 Commerce 11 Labour 12 Capital 4 Manufacturing industry 13 Consumers 0 0,2 0,4 0,6 0,8 1 1,2 1,4 1,6 1,8 05 Lima 7/6/05 08:37 Página 136 A structural analysis of a regional economy using Social Accounting Matrices: 1990-1999 137 Figure A.2. Landscape Andalusia 1995 (numeraire 1990) 9 Commercial Services 8 Other services 11 Labour 12 Capital 10 Non-commercial services 1 Agriculture, cattle & forestry 5 Construction 7 Transport and Comunications 6 Commerce 13 Consummers 3 Electricity and natural gas 9 Commercial Services 4 Manufacturing industry 2 Extractives 10 Non-commercial services 5 Construction 3 Electricity and natural gas 1 Agriculture, cattle & forestry 2 Extractives 7 Transport and Comunications 8 Other services 6 Commerce 11 Labour 12 Capital 4 Manufacturing industry 13 Consumers 0 0,2 0,4 0,6 0,8 1 1,2 1,4 1,6 1,8 05 Lima 7/6/05 08:37 Página 137