SciELO - Scientific Electronic Library Online

Home Pagelista alfabética de revistas  

Servicios Personalizados




Links relacionados


Cuadernos de economía

versión On-line ISSN 0717-6821

Cuad. econ. v.40 n.120 Santiago ago. 2003 

Cuadernos de Economía, Año 40, Nº 120, pp. 285-318 (Agosto 2003)


Rainer Thiele*
Daniel Piazolo**


This paper describes the construction of a Social Accounting Matrix (SAM) for Bolivia for the year 1997. Three distinctive features render the SAM a useful starting point for distributional analyses. First, production in the agricultural and services sector is split up into formal and informal activities to account for the fact that poverty is largely confined to the latter. Second, factor and household accounts exhibit a high level of disaggregation, thus permitting the monitoring of the factorial and personal income distribution. Finally, the SAM contains a detailed system of accumulation balances which reveals the distribution of assets among household groups.

Keywords: SAM, Structural Adjustment Programs, Poverty, Income Distribution, Bolivia.


En el artículo se describe la construcción de una matriz de contabilidad social (SAM) Bolivia para el año 1997. Por tres razones, la SAM resulta útil como punto de partida para analizar la distribución de ingresos. Primero, se dividen los sectores agricultura y servicios en actividades formales e informales, para tomar en consideración que la probreza se concentra en actividades informales. Segundo, las cuentas de los factores de producción y de los hogares muestran un alto nivel de desagregación. Por lo tanto, se puede analizar la distribución factorial y personal de los ingresos de manera muy detallada. Finalmente, la SAM contiene un sistema detallado de saldos de activos y pasivos y sus cambios para diferentes grupos de hogares que revelan la distribución de los bienes entre ellos.


After 20 years of structural adjustment programs, there is an ongoing debate about their economic and social impact (for an overview, see Thiele and Wiebelt, 2000). The distributional consequences of adjustment measures are particularly contentious. Employing, among others, econometric methods (e.g. Behrmann et al. 2001) and simulations in Computable General Equilibrium (CGE) models (e.g. Vos 2001; Paes de Barros and Corseuil 2001), a number of studies for Latin America obtain empirical evidence that strongly varies with the specific measures and countries considered. Somewhat surprisingly, given its reputation as an early and profound reformer, Bolivia is among the few Latin American countries for which information on the distributional impact of adjustment is very scarce. As a contribution to filling this gap, this paper presents a data set for Bolivia for the year 1997 which will later be used for CGE model simulations. The data are organized in a Social Accounting Matrix (SAM), which is the most convenient framework for keeping track of the circular flow of income and expenditures in an economy.

Three distinctive features render the present SAM a better starting point for distributional analyses than the two main recent SAM applications for Bolivia by Alarcón et al. (2000) and Jemio (2001b). First, in the agricultural and services sector, production is split up into formal and informal activities. This is important because Bolivia has a large informal economy where poverty is widespread and where the response to structural adjustment measures is likely to differ from the formal economy. Second, compared to the two previous studies, a stronger emphasis is put on monitoring the factorial and personal distribution of income via the distinction of 4 production factors and 6 household groups. Finally, based on Jemio (2001b), but disaggregated by households, detailed accumulation balances reveal the distribution of assets and indicate to which extent different household groups have access to the resources they need to finance investment in excess of their own savings. With these features, the SAM includes the information necessary to analyze the main transmission mechanisms _changes in relative factor and commodity prices, changes in public redistribution, and changes in the returns on assets and the costs of liabilities_ by which structural adjustment affects the well-being of households.

The paper is structured as follows. Chapter 2 shortly outlines the origins of the SAM framework and its possible uses for policy analysis. Chapter 3 provides a general overview of the components that make up the SAM for Bolivia. Chapter 4 describes the construction of the Input-Output Table around which the SAM is built, and discusses some main structural characteristics of the Bolivian economy which can be derived from the Input-Output Table. The distribution of income across factors and households is shown in Chapter 5. Chapter 6 deals with the accumulation of assets and liabilities by households. The paper closes with some concluding remarks.


The use of SAMs to record all the transactions that take place in an economy during a year is preceded by a long history of macroeconomic accounting (Kehoe 1998), which can be traced back at least to the year 1758 when Quesnay published his famous Tableau économique. In the twentieth century, Kuznets' work on national income accounts, Leontief's work on input-output matrices, and the statistical reconciliation of these two strands under the United Nations System of National Accounts (SNA), provided the base for the later development of SAMs. In fact, SAMs can be viewed as a straightforward extension of input-output tables in that they capture the distribution of income in addition to production and demand and thereby close the income circle.

The development of social accounting went hand in hand with the development of planning models that used these data. Such models can basically be divided into two categories. On the one hand, SAM-based multiplier models have been developed, in which certain variables, such as investment or government demand, are set exogenous and then changed in order to analyze how the changes feed through the economic system.1 These multiplier models are demand-driven, since no constraints on supply are specified, and are thus Keynesian in spirit. On the other hand, SAMs have been used to provide data bases for constructing CGE models. A main feature of (most) CGE models is that relative price movements tend to play a central role in determining production and demand decisions, making them more neoclassical in spirit.

While SAMs constitute a powerful tool for policy analysis, their limitations have to be kept in mind. Most notably, a SAM only provides a snapshot for a particular year, which is not necessarily characteristic of the economy at hand if, for example, an exogenous shock hit the economy in that year. Furthermore, due to lags in data collection, SAMs often are not up-to-date. This problem can partly be resolved by means of procedures for updating data bases, such as the maximum entropy approach suggested by Robinson et al. (2001). Finally, a SAM cannot readily be used as an input for policy analysis. To conduct a CGE analysis, most nominal flows given in the SAM have to be split up into their price and quantity component, and certain parameters such as trade elasticities have to be specified outside the SAM framework.


A schematic representation of the SAM structure chosen for Bolivia is given in Table 1.2 The Bolivian SAM consists of four different types of accounts. First, product supply and demand is described by a set of commodity accounts, where the row shows the cost components that add up to overall supply from domestic sources, plus imports, while the column lists the domestic demand components and exports. Second, factor accounts depict how value added is distributed to the domestic factors of production, and how this factor income is transformed into income accruing to the various institutional agents (government, households, companies, rest of the world) identified in the SAM. Third, current accounts show the sources from which institutions receive income, and the uses to which they put that income. Part of it is consumed, part is redistributed among the institutional agents themselves, and the remaining is saved. Among the institutional accounts are two important macroeconomic balances, namely the government budget and the current account of the balance of payments.


All in all, with slight variations in exposition, these first three kinds of accounts can be found in almost every SAM. The main peculiarity here lies in the level of disaggregation among commodities, factors, and institutions, which will be discussed in detail below. In conceptional terms, the capital accounts, which are required to establish the link between savings and investment, constitute the distinctive feature of the Bolivian SAM. In most existing SAMs, it is simply assumed that the savings of the different institutions are collected in a "savings pool" and then made available for investment, without any reference to the process of financial intermediation. Here, by contrast, all major flows of funds occurring to match the positions of surplus units (institutions for which savings exceed investment) and deficit units are recorded. The explicit consideration of financial market transactions carries a number of advantages.3 Viewed from the perspective of structural adjustment programs, the main advantage appears to be that in certain areas such as monetary policy the transmission that runs via the financial system can be analyzed.


1. Construction of the Input-Output Table

The Input-Output Table forms the core of the SAM as it describes the structure of the economy.4 Based on the data for 35 sectors from the Instituto Nacional de Estadistica (INE) (2001a) of Bolivia an aggregated Input-Output table with 13 sectors was constructed.5 The chosen aggregation reflects the favored focus of the analysis and groups together activities with similar demand and supply characteristics. As discussed later, the remaining 13 sectors differ quite significantly, e.g. in their trade shares and their dependency on intermediate products. The assignment of the 35 sectors of the Input-Output Table to the 13 sectors of the SAM is represented in Overview 1.

Most aggregated sectors are derived through the simple addition of the individual sectors. However, the composition of the service sectors and the agricultural sectors is more complex. For these two sectors, a disaggregation into informal and formal activities was considered necessary because most of the poorer people in Bolivia pursue informal activities, and because production characteristics (e.g. trade shares) differ between formal and informal activities. Informal activities are identified according to the value added share of unincorporated capital. A high share of unincorporated capital is assumed to reflect a high percentage of informal activity. The derived shares of informal activities for eight service sectors are given in Table 2 and range from 0 percent in communications to 65 percent in trade. For transport and storage, financial services and company services we assume that the share of formal activity is 100 percent. In domestic services wages account for all value added. Here, a modification is introduced: The complete sector is attributed to informal services.

Source: Calculated on the basis of Instituto Nacional de Estadística (INE) (2001a, 2001b).

Table 2 also presents each sector's percentage share of the total value added of the nine service sectors and the derived weighted share of the informal and formal activities. For all nine service sectors, informal services account on average for 20 percent of the activities.

A similar aggregation procedure using the value added share of unincorporated capital was performed for the division between traditional (informal) and modern (formal) agriculture. Table 3 gives the percentage share of traditional and modern activities in each agricultural sector, the weight of each agricultural sector (according to the value added) and the derived overall share of traditional and modern agricultural activities. 67 percent of the Bolivian agriculture is traditional, 33 percent is modern.


From the 13-sector Input-Output Table, various indicators describing demand and supply in the Bolivian economy can readily be calculated. The structure of the sectoral use of goods is given in Table 4. This table shows for each sector the absolute values of the components that make up domestic absorption: intermediate demand, private consumption, government consumption and investment (which is the sum of fixed capital formation and inventories). Subtracting imports and tariffs from domestic absorption yields the domestic use of domestically produced goods. Adding to this exports results in the overall use of domestically produced goods.6

An examination of the relative shares of the different components reveals an enormous variety among sectors. Intermediate demand accounts for 83 percent of domestic absorption for crude oil and natural gas whereas the share is only about 1 percent in the public sector and in the coca sector. Coca is the sector where final private consumption is with almost 99 percent the largest component of absorption. Consumer goods (63 percent) and electricity, gas and water (56 percent) are the sectors with the next highest share of private consumption. Government consumption is only a component of domestic absorption in the public sector. Investment is a decisive component of absorption for construction (almost 96 percent) and capital goods (59 percent). These structural characteristics are important determinants of the likely impact of structural adjustment. If, for example, investment is cut back as a result of fiscal consolidation, this will lead to a contraction of the construction and capital goods sector.

In foreign trade, overall imports correspond to 16 percent of domestic absorption, a moderate share for a small economy such as Bolivia. There are no imports of coca, crude oil and natural gas, and construction. However, for capital goods imports correspond to 86 percent, for intermediate goods to 38 percent and for consumer goods to almost 15 percent of domestic absorption. Table 4 also includes the import structure of the Bolivian economy. Almost 90 percent of all imports are directed to the three sectors capital goods (44 percent), intermediate goods (27 percent) and consumer goods (17 percent).


Exports account for about 12 percent of the overall use of domestic goods. As in the case of imports, this points to a rather low integration of the Bolivian economy into world markets. Very large shares are reported for coca (more than 80 percent) and mining (65 percent). The export ratio differs substantially between the informal and the formal agricultural sectors and underlines the necessity to distinguish between them: Whereas only less than 9 percent of traditional agriculture is exported, more than 25 percent of modern agriculture goes abroad.7 The export structure shows that 76 percent of all exports stem from four sectors: Consumer goods (25 percent), formal services (22 percent), mining (18 percent) and intermediate goods (11 percent).

The trade pattern that emerges for Bolivia has important implications for the distributional analysis of adjustment programs. All policies that lead to a (real) devaluation of the Boliviano, for example, will benefit outward-oriented sectors with reasonable flexibility to shift between domestic and world markets, such as mining and modern agriculture. Outward-oriented sectors with low flexibility, such as capital goods where domestic production cannot easily substitute for imports, will hardly benefit. And sectors mainly producing nontradables, such as informal services where most of the urban poor earn their living, will be hurt by a depreciation.

The overall demand structure of the Bolivian economy is also represented in Table 4. It shows that the demand for formal services (28 percent) and for consumer goods (20 percent) are quite important as a share of overall demand. The next three sectors (informal services, intermediate goods and public sector) have a share of around 9 percent each of overall demand. Coca accounts for only half of one percent of total demand, and for less than four percent of all exports. However, since it employs many smallholders (Section IV.1) and is regionally concentrated, a separate consideration of coca is justified, e.g. for the analysis of coca eradication programs.

Table 5 presents the structure of the sectoral production costs in Bolivia in 1997. For each of the 13 sectors the costs for intermediates and the factors of production are given as absolute values and as percentage shares.

This representation shows that coca (93 percent), traditional agriculture (75 percent), mining (74 percent) and the public sector (73 percent) are the sectors with the highest share of costs for the factors of production in the sectoral net production. The sectors capital goods (76 percent), consumer goods (71 percent) and intermediate goods (67 percent) are the sectors with the highest share of intermediate costs.


The sectoral cost structure is of major importance for the examination of adjustment programs. The capital goods sector, for instance, with its high share of intermediate costs and a large share of imported intermediates will not be very responsive to cost pressures, given that it can barely substitute imports by domestic production. By contrast, sectors facing a high cost share for primary factors and reasonably flexible factor markets (like coca and traditional agriculture) will be much more able to pass cost pressure on to factors. These effects are, in turn, very important for the distributional aspects of adjustment programs.


The reforms typically required in adjustment programs affect the real income position of households via four basic channels. First, most structural measures alter the use and remuneration of production factors and thus have an impact on factor incomes that varies with factor endowments. Second, fiscal consolidation usually entails a reduction and/or restructuring of transfer payments to households. Third, a liberalization of the financial sector changes the returns on assets as well as the costs of liabilities. Fourth, microeconomic reforms such as cuts in food and energy subsidies lead to changes in relative consumer prices, thereby affecting the real income distribution as long as consumption patterns are not identical across households. The informational base the SAM provides for modeling these four channels will be discussed in the following.

1. Distribution of Factor Income

The net value added created in the economy flows as a reward to the factors used in the production process. In order to obtain a rich enough picture of the resulting factorial income distribution, we extend the conventional breakdown of value added into labor and capital income by distinguishing four different production factors: Skilled labor, unskilled labor, corporate capital, and unincorporated capital. While the two labor categories correspond to the distinction between employees and workers made in Bolivian statistics, the two types of capital are characterized by different ownership structures. Corporate capital is owned by private and public enterprises and all corporate capital income is retained in these enterprises. Unincorporated capital, by contrast, belongs to households who also receive the respective factor income. This factor income, in turn, comprises both profits distributed to owners of enterprises and mixed income earned by self-employed workers, the latter being a conglomerate of a return on investment and other components such as land rents.

Arriving at the distribution of income to these four factors involved several steps. Basic data for breaking down value added into its components were provided by INE (2001b) which calculated for 1998 the shares of wages, operating surpluses and mixed income in total value added. These data were adapted as follows. First, to obtain 1997 figures, it was assumed that the structure of value added did not change between 1997 and 1998. Then, wages were distributed between skilled and unskilled labor according to the information given in the 1997 employment survey (INE, 1997). Finally, operating surpluses were split up into distributed profits and corporate capital income by taking the aggregate value of retained earnings given in the national accounts as total corporate capital income, and by assuming that both types of income have the same sectoral composition.

The factorial income distribution for the 13 production sectors that results from these calculations is shown in Table 6. A very diversified picture emerges. Traditional agriculture and informal services where, by assumption, only unincorporated capital is used as a production factor, account for the bulk of mixed income. Corporate capital income and distributed profits are highest in formal services, a result that is partly owed to the large size of the sector. As a share of sectoral net value added, corporate capital income, for example, is much higher in both the oil and gas and the utilities sector. Skilled laborers earn most of their income either from providing formal services or from serving in the public administration, while unskilled workers receive a high share of their income from modern agriculture, manufacturing, construction and public services. Overall, with such a sectoral diversity of factor incomes, adjustment policies that induce structural change are likely to have a substantial impact on Bolivia's factorial income distribution.

Knowledge of the functional income distribution constitutes only a first step towards assessing a country's distributional situation. More direct insights can be gained by tracing the flow of income from factors to households. In doing so for Bolivia, we identify six different types of households: smallholders, agricultural workers, employees, non-agricultural workers, urban informals, and employers. The disaggregation is basically made along functional lines, i.e. households with similar factor endowments are lumped together. This is justified because factor income is the single-most important income source in Bolivia given the low degree of redistribution (see Section V.2). Furthermore, workers and those involved in informal activities are disaggregated regionally as their consumption patterns tend to vary between regions (see Section V.3).


Appendix Table 1 (columns 14-18; rows 19-27) shows how income is distributed from the 4 production factors to the 6 household groups and to enterprises. 5 household groups obtain factor income from one single source: smallholders and urban informals only earn mixed income, agricultural and non-agricultural workers only unskilled labor income, and employees only skilled labor income. Employers are the sole exception as this group does not only include capital owners but also self-employed people with mixed income, such as providers of financial services, who cannot meaningfully be counted as informals.

What these income flows mean for the well-being of households is revealed by the average income figures presented in Table 7. Not surprisingly, it turns out that smallholders are worst off, followed by urban informals and the two categories of worker households. The rather large gap between smallholders and urban informals on the one hand and agricultural and non-agricultural workers on the other hand is largely due to the fact that unpaid family workers are included among the former. But even if unpaid family workers are excluded from the calculation, the ranking remains unchanged. Smallholders then earn about 30 percent less than agricultural workers, and urban informals almost 10 percent less than non-agricultural workers. Incomes of employees and employers are substantially above the national average. As the by far richest household group, employers earn more than ten times the amount that goes to smallholders. Such a considerable range of income levels suggests that the classification chosen here has succeeded in isolating household groups with different living standards and thus provides a useful starting point for distributional analyses. Nevertheless, substantial within-group disparities still exist at the chosen level of disaggregation. As a consequence, the Gini coefficient for the income distribution resulting from the SAM only amounts to slightly above 0.4, whereas most previous studies based on disaggregated household surveys obtain Gini coefficients of more than 0.5 (Thiele 2003).

2. Redistribution of Income

Various transactions taking place between institutions, such as government transfers to households and remittances from Bolivians living abroad, modify the factorial income distribution described above. In the SAM, the six household groups, the government, private and public enterprises, and the rest of the world are identified as participants in this process of redistribution. Financial institutions are assumed to act as mere intermediaries (see Section VI.2). Their current transactions are allocated to the two kinds of enterprises. As a result, the institution "private enterprises" does not only include private firms, but also commercial banks, insurance companies and pension funds, while the institution "public enterprises" consists of state-owned and capitalized firms, the Central Bank, and other financial institutions such as development banks.

Table 8 shows the part of the inter-institutional transactions that involves the six household groups; a full description of the income flows between institutions is given in Appendix Table 1. It turns out that households as an aggregate have

net receipts of about 2.5 billion Bolivianos. The most significant transactions recorded are public transfers (2.5 billion Bolivianos), government revenues (2 billion Bolivianos), interest receipts on deposits and dividends on equities (3.0 billion Bolivianos), and interest payments on loans (1.1 billion Bolivianos). Pensions and the corresponding contributions account for the bulk of the flows from and to the government. Direct taxes and transfers, the classical means of public redistribution, only play a minor role, with 200 and 400 million Bolivianos, respectively, as do remittances, with less than 200 million Bolivianos.

Since there is no survey information available on the payments and receipts of different household groups, the disaggregation of the overall amounts had to be based on a number of assumptions, of which the most important are:

(i) public pensions are confined to employees;
(ii) only employers and employees hold equities and deposits in the private and public banking system;
(iii) only employers and employees have access to private loans, while public loans are distributed to smallholders and informals;
(iv) smallholders and urban informals do not pay taxes;
(v) all household groups receive remittances;
(vi) all household groups except employers receive transfers from the government.

Based on these assumptions, overall flows were allocated among households in fixed proportions to their income levels. What comes out is that the bulk of redistribution involves employers and employees, while the poorer household groups are barely affected. Although this inevitably is a very stylized pattern, one can at least conclude that the redistribution process in Bolivia does not do much to correct for the disparities occurring in the primary income distribution. Poor households may not be totally excluded from formal social security and financial markets as assumed here, but their access definitely is very limited. Adding to this the low level of transfers, redistribution cannot be markedly progressive and thus is unlikely to constitute a major link between structural adjustment and the well-being of poor households.

3. Structure of Household Demand

At given nominal factor and transfer incomes, the sectoral composition of private demand determines how the real income position of households might be affected by structural adjustment policies via changes in relative consumer prices. Since sectoral expenditures cannot be observed directly, the calculation of the demand vector for each household group had to proceed in two steps. First, consumption data from the national accounts were grouped into six commodity aggregates (see Table 9), and then disaggregated by household group combining information from the 1999 MECOVI survey (INE, 2001c) and from Jemio (1993). Although the latter only identifies three types of households (rural, lower-income urban, upper-income urban), it had to be used here because the survey does not provide the necessary data for all relevant commodities. Due to lacking evidence, the shares spent on clothing and footwear and on consumer durables were set equal for the two rural households, for non-agricultural workers and informals, and for employees and employers, while expenditures on all services apart from transport were calculated residually. Second, the resulting commodity demand schedule was translated into a sectoral demand schedule employing a transformation matrix based on Jemio (1993).

In the commodity demand schedule shown in Table 9 two structural characteristics stand out. Food, beverages and tobacco are on much higher demand in poorer than in richer households and, among poorer households, rural spending exceeds urban spending. The mirror image can be observed with respect to services: wealthier people spend more than poorer people, and the urban poor more than the rural poor, on services such as health and education; only the budget shares allocated to transport are more or less equal across household groups, although the aggregate figures are likely to mask differences such as a move from public to private means of transportation with rising living standards.

This structure of demand carries over to the sectoral consumption pattern reported in Table 10. The share of the consumption basket devoted to the output of traditional agriculture, for instance, ranges from about 5 percent for employees and employers to more than 10 percent for smallholders, and services contribute between 30 percent (for smallholders) and 50 percent (for employers) to overall expenditures. In the case of services, the composition of demand also differs among households: the relative importance of informal services is highest in poor urban households, while rural households have to rely for the most part on formal services as they lack access to many of the services only provided by the urban informal economy. Overall, the structural variation in consumption seems to be large enough to constitute a potential mechanism through which adjustment measures can affect the distribution of income.


This chapter deals with the part of income institutions retain for investment in physical and financial capital. It first discusses the savings and investment recorded for the year 1997 (Section VI.1), and then links these flows to the corresponding stocks which indicate the wealth of the different institutions (Section VI.2).

1. Saving and Investment

At given factor incomes, inter-institutional transfers, and consumption expenditures, the amount saved by each domestic institution can be calculated residually from the accounting identity.

(1) factor income + net transfers = consumption + savings.

External savings are determined by the current account balance. As shown in Appendix Table 1 (columns 30-43; rows 19-28), employers and employees dominate household savings. All other household groups exhibit saving rates of around zero. Households and enterprises together account for about half of overall savings. The other half is contributed by the government and the rest of the world, reflecting that Bolivia in 1997 ran a budget surplus and a current account deficit.

Total savings (8.176 billion Bolivianos) determine the resources available for investment in physical capital. The institutional composition of this investment is documented in the national accounts, except for the different household groups. The breakdown of household investment had to rely on two assumptions: first, gross fixed capital formation, which includes the establishment of residential buildings, was assumed to have the same distribution as the respective capital stocks (see below); second, aggregate inventories were distributed among smallholders, urban informals, and employers _ i.e. among those household groups that are also producing units _ taking shares in total mixed income as weights.

Compared to the government and enterprises, whose investment amounts to 2.086 and 5.666 billion Bolivianos, respectively, households do not invest much (Table 11). For informals and smallholders, investment is even slightly negative as the reduction in inventories overcompensates positive fixed capital formation. Furthermore, the investment projects households undertake are largely confined to construction; only employers utilize a non-negligible amount of capital goods.

The low investment levels realized by smallholders and informals point towards a possible persistence of poverty because for these household groups, who derive their income from self-employment, capital formation arguably is the most important means to raise living standards in the medium to long run. As a complement to private investment, capital expenditures by the government on public goods such as infrastructure may also have a considerable impact on the well-being of smallholders and informals. Hence, if public investment is cut back during adjustment, this constitutes another mechanism through which macroeconomic reforms may be transmitted to the household level.

While overall savings in the economy have to be equal to overall investment, this is not true for individual institutions. Here, saving-investment surpluses by households and the rest of the world correspond with deficits by enterprises and the government. These surpluses and deficits are balanced via the financial system. For each institution, the identity.

(2) saving -investment = D assets-D liabilities = D net assets.

must hold, i.e. institutions with a saving-investment surplus accumulate (net) financial assets and thereby finance the deficits of the other institutions. Some of the financial transactions, for example the purchase of shares in enterprises, take place directly between institutions, while others involve one of the following financial intermediaries: the Central Bank, commercial banks, other financial institutions, and pension funds.

The matrix describing the flow of funds between all these economic agents replicates the structure of the matrix for 1998 given in Jemio (2001b). Since neither the existing case studies nor the Bolivian household surveys contain appropriate information about the financial transactions conducted by individual household groups, the disaggregation again produces a stylized picture. For shares, deposits and loans, the same assumptions as in Section V.2 were applied. In addition, it was assumed that only employees accumulate private pension rights, and that only employers hold foreign assets. Currency holdings were calculated residually so as to balance the flow-of-funds system.

The financial flows between all institutions identified in the SAM can be found in Appendix Table 2 (columns 31_44; rows 1_14). For the six household groups, savings allocated to pension funds, deposits in and loans from the commercial banking system, and foreign assets turn out to be the most significant items, which are all confined to the two richer household groups (see columns 1-6 and rows 37_44 for assets; columns 7_14 and rows 31_36 for liabilities). The participation of poorer households in the financial system is low, both as creditors and debtors. From a distributional viewpoint, the most relevant feature emerging from the flow of funds is that smallholders and informals have very limited access to credit. This result may be somewhat overstated because in reality the two groups are likely to obtain at least some loans from commercial banks. Nevertheless, it seems reasonable to conclude that credit constraints _ probably combined with a low saving capacity and other institutional factors not captured in the SAM _ act as a break on investment by smallholders and informals.

2. Net Wealth

The flows just described link the stocks at the beginning to those at the end of 1997. Stocks can also be subject to revaluations over the year. In the highly dollarized Bolivian economy, such revaluations are assumed to be driven by exchange rate changes, with a depreciation of the Boliviano implying an appreciation of existing assets and liabilities and vice versa. For financial stocks, the accumulation process is described by

(3) assetst = assetst -1 + D assets + revaluation of assets,


(4) liabilitiest = liabilitiest -1 + D liabilities + revaluation of liabilities.

A similar relationship holds for physical capital:

(5) capital stockt = capital stockt -1 + investment-depreciation + capital gains.

From equations (3) to (5), net wealth can be derived as

(6) net wealtht = assetst -liabilitiest + capital stockt .

A full account of these stock-flow relationships for all institutions is given in Appendix Table 2. Here we focus on the net wealth position of the different household groups, which is of particular interest from a distributional point of view. To arrive at the net wealth figures, end-of-period stocks at the aggregate household level were first set equal to the beginning-of-period stocks for 1998 given in Jemio (2001b), and the beginning-of-period stocks were then derived by calculating backwards. The disaggregation of financial stocks was achieved by making the same assumptions as in Section VI.1, except for currency holdings which were fixed at the same proportion of income for each household group. Physical capital was distributed among households using the information the MECOVI survey provides about fixed assets such as buildings and vehicles.

Table 12 shows the end-of-period stocks and the resulting net wealth. It turns out that physical capital is by far the dominating asset in the portfolio of Bolivians. This is even true for those two household groups (employers and employees) with the strongest links to the financial system. Just as the income distribution, the distribution of net wealth exhibits a clear distinction between the 4 poorer and the 2 richer household groups. The overall dispersion of net wealth is even higher than the dispersion of income, with an average employer's net assets exceeding those of an average smallholder by a factor of 15.8 Among the poorer households, smallholders and urban informals appear to be in a somewhat better position in terms of net wealth than in terms of income as they own relatively large physical capital stocks.


This paper has described the construction of a Social Accounting Matrix (SAM) for Bolivia for the year 1997. The SAM displays a number of distributional features of the Bolivian economy, of which the most important are:

(i) Smallholders and urban informals together account for about two thirds of the total labor force, but for less than a quarter of total income;

(ii) the richest household group (employers) receives slightly more than ten times the income of the poorest household group (smallholders);

(iii) disparities in the distribution of wealth are even somewhat wider, with employer's net assets exceeding those of smallholders by a factor of 15;

(iv) the poor household groups are characterized by low savings, low investment, and a low participation in the financial system.

To arrive at a SAM that captures these distributional features in a consistent way, data from different sources _ primarily the 1997 Input-Output table, the 1997 national accounts, and two household surveys for 1997 and 1999 _ had to be reconciled. Moreover, various assumptions had to be made because at the given level of disaggregation not all the required information was available. The resulting data base is therefore in parts somewhat stylized, but it is still likely to provide a reasonable approximation of the structural characteristics prevailing in Bolivia, rendering it a useful starting point for further analyses.

While international comparisons are inherently limited by country-specific features, at least two generalizations from the Bolivian SAM seem to be justified. First, since dual production patterns can be found almost everywhere in the developing world, the distinction between formal and informal activities made in this paper might serve as a blueprint for future SAM applications. Second, the household disaggregation according to factor endowments for Bolivia might also be the best choice for most other low and lower-middle income countries, given that redistribution tends to play a major role only at higher per capita income. Finally, it would in general be desirable to integrate financial accounts into SAMs but data requirements may often be too demanding.


Alarcón, J., J. van Heemst, and N. de Jong (2000), Extending the SAM with Social and Environmental Indicators: An Application to Bolivia. Economic Systems Research 12 (4): 473_496.         [ Links ]

Behrman, J., N. Birdsall, and M. Székely (2001), Pobreza, Desigualidad y Liberalización Comercial y Financiera. Working Paper 449 (Washington, D.C., Interamerican Development Bank).         [ Links ]

Bulmer-Thomas, V. (1982), Input_Output Analysis in Developing Countries: Sources, Methods and Applications (Chichester, John Wiley & Sons).         [ Links ]

INE (Instituto Nacional de Estadística) (1997), Encuesta Nacional de Empleo III: Noviembre 1997, La Paz.         [ Links ]

INE (Instituto Nacional de Estadística) (2001a), Cuentas Nacionales 1988_1997, La Paz.         [ Links ]

INE (Instituto Nacional de Estadística) (2001b), Bolivia: Valor Agregado Bruto A Sus Componentes Segun Actividad Economica 1998, La Paz (unpublished manuscript).         [ Links ]

INE (Instituto Nacional de Estadística) (2001c), Encuesta Mecovi: Noviembre 1999, La Paz.         [ Links ]

Jemio, L.C. (1993), Micro- and Macroeconomic Adjustment in Bolivia (1970_89): A Neostructuralist Analysis of External Shocks, Adjustment and Stabilization Policies (The Hague, PhD Dissertation, Institute of Social Studies).         [ Links ]

Jemio, L.C. (2001a), Macroeconomic Adjustment in Bolivia since the 1970s: Adjustment to What, By Whom, and How? Analytical Insights from a SAM Model, Kiel Working Paper 1031 (Kiel, Institute of World Economics).         [ Links ]

Jemio, L.C. (2001b), Debt, Crisis and Reform in Bolivia: Biting the Bullet (Hampshire, Basingstoke).         [ Links ]

Kehoe, T. (1998), Social Accounting Matrices and Applied General Equilibrium Models, in: I. Begg (ed.), Applied Economics and Public Policy (Cambridge University Press).         [ Links ]

Lysy, F. (1977), Creating a Consistent Data Set for the Research Model of Malaysia (unpublished manuscript).         [ Links ]

Paes de Barros, R. and C.H. Corseuil (2001), Apertura Económica y Distribución del Ingreso en Brasil, in : E. Ganuza, R. Paes de Barros, L. Taylor, and R. Vos (eds.), Liberalización, Desigualidad y Pobreza: America Latina y el Caribe en los 90 (Eudeba, Buenos Aires).         [ Links ]

Pyatt, G. and J. Round (1985), Social Accounting Matrices: A Basis for Planning (World Bank, Washington, D.C.).         [ Links ]

Robinson, S., A. Cataneo and M. El-Said (2001), Updating and Estimating a Social Accounting Matrix Using Cross Entropy Methods. Economic Systems Research 13 (1): 47_67.         [ Links ]

Thiele, R. (2003), The Social Impact of Structural Adjustment in Bolivia. Journal of International Development 15 (1): 1-21.         [ Links ]

Thiele, R. and M. Wiebelt (2000), Sind die Anpassungsprogramme von IWF und Weltbank gescheitert? Eine Bestandsaufnahme der Erfahrungen aus zwei Jahrzehnten. Discussion Paper 357 (Kiel, Institute of World Economics).         [ Links ]

Vos, R. (1991), Social Accounting, Flow of Funds and Capital Accumulation, in: J.V. Alarcón, I. van Heemst, S. Keuning, W. de Ruijter, and R. Vos (eds), The Social Accounting Framework: Concepts, Construction and Applications (Aldershot).         [ Links ]

Vos, R. (2001), Liberalización Económica, Ajuste, Distribución y Pobreza en Ecuador, 1988-1999, in: E. Ganuza, R. Paes de Barros, L. Taylor, and R. Vos (eds.), Liberalización, Desigualidad y Pobreza: America Latina y el Caribe en los 90 (Eudeba, Buenos Aires).         [ Links ]

* Research Fellow, Kiel Institute for World Economics, Duesternbrooker Weg 120. D-24105 Kiel / Germany. E-mail:r.
This paper is part of a research project on "Poverty Impacts of Macroeconomic Reforms: Stabilization and Structural Adjustment Programs in Bolivia." Financial support by the Kreditanstalt für Wiederaufbau (German Development Bank) is gratefully acknowledged. The authors thank Lykke Andersen, Ernesto Cupé, and Osvaldo Nina for the constructive suggestions made during a workshop at the Kiel Institute for World Economics, Manfred Wiebelt for helpful comments on an earlier version of this paper, and Jann Lay for effective research assistance.

** Researcher, Feri Research GmbH, Rathausplatz 8-10. D-61348 Bad Homburg / Germany. E-mail:

JEL Classification: D57, O54.

1 For a number of illustrative examples of this sort of modelling, see the volume edited by Pyatt and Round (1985).

2 For an algebraic representation of the SAM, see Appendix 1.

3 Vos (1991) provides a detailed account of these advantages.

4 The Input-Output Table is derived through a calculation process using the technology matrix and the make matrix of an economy. For a detailed description of the matrices used in the SAM construction see Bulmer-Thomas (1982). For Bolivia, these matrices are provided by the Instituto Nacional de Estadistica (INE) (2001a).

5 The Bolivian statistics include an additional dummy sector 36, called imputed bank service charge, which was eliminated in order to construct a square (35x35) Input-Output table following a similar procedure to the one suggested in Lysy (1977, p. 9). Whereas Lysy allocates the values of the imputed bank service charge to the other sectors in proportion to each sector's value added, we allot this in proportion to the value added share of incorporated capital (see also Chapter IV), assuming that only sectors with a high share of incorporated capital demand a high share of the imputed bank services.

6 It should be noted that all macroeconomic totals in the Input-Output Table are equal to the national account figures calculated by INE (2001a).

7 Since the export share of informal services is negligibly small, the exports of informal services are assigned as exports of formal services. The export share of formal services then amounts to 9 percent of the overall use of domestic goods.

8 This result does not necessarily imply lower returns on the wealth of richer households because some returns (most notably the imputed rents for owner-occupied housing) do not show up as income in the SAM.


This Appendix specifies in algebraic form the main identities and constraints that can be derived from the aggregate SAM given in Table 1. The commodity account can be written as

Equation (1) expresses GDP at market prices from the expenditure side as

and from the value-added side as

Rewriting equation (1a) gives domestic absorption:

By combining the information given in the current and capital accounts, the budget constraints of the institutions identified in the SAM can be derived. For the government, the current balance is

The government's capital balance is

Substituting equation (3) into equation (4) and rearranging the resulting expression yields the government budget constraint:

Equation (5) states that the overall fiscal deficit (left-hand side) has to be financed by borrowing from other institutions.

The budget constraint for households can be derived in a similar fashion, using the identity between the sources and uses of income

and the balance of capital transactions

Substituting equation (7) into equation (6) yields the household budget constraint

which implies that income in excess of consumption and tax payments is used for physical and financial investment.

For companies, the current balance

and the capital balance

lead to the company budget constraint

In the external sector, the current account equation is given as

the capital account equation as

and therefore the external sector budget constraint as

Equation (14) simply states that Bolivia's current account deficit is financed by net foreign borrowing.

Finally, adding up all the budget constraints yields the savings-investment balance

The savings-investment states that aggregate domestic investment is financed by domestic saving (S) and foreign saving (DAf).


hh, g, c, f, fi superscripts denoting the institutions (households, government, companies, rest of the world, financial institutions); in case of two superscripts the first indicates the institution that receives a payment


C consumption, MI mixed income,
I investment, A absorption,
X exports, NT net transfers,
M imports, INT interest payments,
VA value added, S savings,
TI indirect taxes, T taxes,
SUB subsidies, DL change in net liabilities,
W wages, DA change in net assets,
P profits, CA current account.





Creative Commons License Todo el contenido de esta revista, excepto dónde está identificado, está bajo una Licencia Creative Commons