TMD DISCUSSION PAPER NO. 112 POVERTY-FOCUSED SOCIAL ACCOUNTING MATRICES FOR TANZANIA James Thurlow International Food Policy Research Institute Peter Wobst International Food Policy Research Institute and Centre for Development Research (ZEF) Trade and Macroeconomics Division International Food Policy Research Institute 2033 K Street, N.W. Washington, D.C. 20006, U.S.A. March 2003 TMD Discussion Papers contain preliminary material and research results, and are circulated prior to a full peer review in order to stimulate discussion and critical comment. It is expected that most Discussion Papers will eventually be published in some other form, and that their content may also be revised. This paper is available at http://www.cgiar.org/ifpri/divs/tmd/dp.htm Abstract The development of effective and sustainable economic policies for Tanzania requires access to appropriate databases. One such database is a social accounting matrix (SAM) that details the structure of the entire economy, taking into account the patterns of production and demand, and various institutional relationships. Prior to this study the most recent SAM for mainland Tanzania was for 1992 and was based on past household budget and labor force surveys. Following the release of newer versions of these two surveys as well as a new input-output table for 1992, it is desirable to construct a new SAM for the country. Furthermore, given that Tanzania is committed to reducing national poverty, it is necessary that this new SAM is able to address questions related to poverty and inequality. This paper outlines the process of developing SAMs for mainland Tanzania. Although this process was formally a collaborative project undertaken by the International Food Policy Research Institute (IFPRI) and the Tanzanian National Bureau of Statistics (NBS), the final project team also contained representatives from the Bank of Tanzania, the Tanzanian Revenue Authority, the University of Dar es Salaam, and the Economic and Social Research Foundation (ESRF). The collaboration combined IFPRI’s technical experience with the collaborators’ knowledge of the country and available statistical data. The SAMs were constructed during a series of workshops held in Tanzania aimed at capacity building, information sharing, and first- hand data validation. Furthermore, the process was made sufficiently flexible to allow the NBS to easily update the SAM to subsequent years. The updating process is also detailed in this paper. Although only the 2000 SAM is described, the process is identical for the other Tanzanian SAMs published by IFPRI for 1998 onwards. Acknowledgements The work presented in this paper forms part of a collaborative project with the Tanzanian National Bureau of Statistics (NBS) that aims at the development of social accounting matrices for economic policy analysis in Tanzania. The Royal Danish and Dutch Embassies jointly funded the project, with the latter funding the International Food Policy Research Institute’s (IFPRI) component. Part of the process of constructing the SAM took place over three two-week workshops held in Tanzania, with representatives from the NBS, the Bank of Tanzania, the Tanzanian Revenue Authority, the University of Dar es Salaam, and the Economic and Social Research Foundation (ESRF). While this paper documents IFPRI’s involvement in the project, the results are product of the efforts of the many people who are involved in the project and have participated in the workshops. Contents 1. Introduction and Background ..................................................................................................... 1 2. The Structure of a Social Accounting Matrix ............................................................................. 2 Activities and Commodities........................................................................................................ 3 Trade and Marketing Margins .................................................................................................... 3 Government Income and Payments ............................................................................................ 4 Domestic Non-government Institutions...................................................................................... 4 Home and Final Household Consumption.................................................................................. 5 3. Constructing the Prior ................................................................................................................. 5 4. The Balancing Process.............................................................................................................. 17 The Cross-entropy Balancing Method ..................................................................................... 17 Setting the Constraints on the Prior .......................................................................................... 20 5. Describing the Final Social Accounting Matrix ....................................................................... 22 Micro SAM Accounts............................................................................................................... 22 The Structure of Production and Trade..................................................................................... 28 Household Income and Expenditure......................................................................................... 36 Government, Savings, Investment and the Rest of the World.................................................. 39 6. Summary ................................................................................................................................... 40 References..................................................................................................................................... 40 Appendix 1: The SAM Construction Process............................................................................... 42 Appendix 2: Final 2000 Micro SAM for Tanzania (Tsh Bil.) ...................................................... 46 1 1. Introduction and Background This paper outlines the process of constructing social accounting matrices (SAMs) for mainland Tanzania.1 A SAM is a consistent data framework that captures not only the information contained in the national income and product accounts and the input-output table, but also takes into consideration the monetary flows between institutions within the economy. The SAM is an ex-post accounting framework since, within its square matrix, total receipts must equal total payments for each account contained within the SAM. As an attempt to reflect the specific structure of the Tanzanian economy, the SAM’s accounts are disaggregated across various activities, commodities, production factors, and households. However the task of constructing a SAM is complicated by this need for greater disaggregation. Since the required information is not contained within a single data source, information from various sources has to be compiled and then made consistent with one another. This process is itself valuable since it reveals inconsistencies between a country’s various statistical sources, and by doing so, highlights areas where data reliability is weakest. Beyond its role as a consistency-check, the objective of a SAM is to provide a multi-sector, economy-wide database that facilitates policy-analysis. The work presented in this paper forms part of a collaborative project between the International Food Policy Research Institute (IFPRI) and the Tanzanian National Bureau of Statistics (NBS) aimed at developing poverty-focused SAMs for Tanzania. Although the SAMs outlined in this paper do not make use of information contained in earlier SAMs, it does represent the culmination of a project that also involved the updating of the 1992 SAM originally developed by Wobst (1998). The most recent available data was used during the compilation of the SAM. This information included the results of the 2000/01 Household Budget Survey (HBS) (NBS, 2002a) and the 2000/01 Labor Force Survey (LFS) (NBS, 2002c). With the exception of the intermediate demand structure, which is taken from the 1992 input-output table (NBS, 1999), the 2000 SAM detailed in this paper is based solely on data for the year 2000. However, the project also 1 Mainland Tanzania is distinct from Zanzibar. Hereafter ‘Tanzania’ is used to describe only mainland Tanzania 2 developed a framework that would allow for the easy updating of SAMs to alternative years. In such cases it was necessary to combine information from different years, albeit using the same method of compilation. Within the joint NBS/IFPRI project a series of SAMs for four consecutive years from 1998 to 2001 were constructed.2 Section 2 briefly reviews the structure of a social accounting matrix. The process of constructing a SAM begins with the compilation of a prior SAM that represents the first attempt to place the available data into a SAM framework. Given the diversity of data sources, this information is almost always inconsistent and results in inequalities between receipts and payments in the SAM accounts. Section 3 details the various data sources used and the assumptions made during the construction of the prior SAM for Tanzania. Once the prior SAM has been compiled the reliability of the various data sources is assessed based on the inequalities between row and column accounts. This is done before attempting to balance the SAM so that the additional information on relative data reliability can be included as weighted constraints in the cross-entropy estimation of the final SAM. The cross-entropy balancing approach is presented in Section 4 together with a description of the constraints imposed on the estimation procedure. Finally Section 5 provides a description of the Tanzanian economy as it is reflected in the final SAM. 2. The Structure of a Social Accounting Matrix3 A Social Accounting Matrix (SAM) is a comprehensive, economy-wide data framework, typically representing the economy of a country.4 More technically, a SAM is a square matrix in which each account is represented by a row and a column. Each cell shows the payment from the account of its column to the account of its row – the incomes of an account appear along its row, its expenditures along its column. The underlying principle of double-entry accounting requires 2 The SAMs for 1998, 1999, 2000, and 2001 are available at www.ifpri.org. A step-by-step guide to updating the Tanzanian SAM is also provided in the appendix. 3 This discussion on SAMs is drawn from Lofgren et al (2002). 4 For general discussions of SAMs see Pyatt and Round (1985), and Reinert and Roland-Holst (1997); for perspectives on SAM-based modeling see Pyatt (1988), and Robinson and Roland-Holst (1988). 3 that, for each account in the SAM, total revenue (row total) equals total expenditure (column total). Table 2.1 shows an aggregated SAM (with verbal explanations in place of numbers). With one exception, it contains all the features of the final SAMs presented in this paper. The exception is that, in the final SAMs, taxes are paid to tax accounts, disaggregated by tax type, each of which forwards its revenues to the core government account. The tax types are divided into direct taxes (on domestic non-government institutions and factors), sales taxes, import taxes, export taxes, and producer/activity taxes. Activities and Commodities The SAM distinguishes between ‘activities’ (the entities that carry out production) and ‘commodities’ (representing markets for goods and non-factor services). SAM flows are valued at producers’ prices in the activity accounts and at market prices (including indirect commodity taxes and transactions costs) in the commodity accounts. The commodities are activity outputs, either exported or sold domestically, and imports. This activity-commodity separation is preferred since it permits activities to produce multiple commodities (for example, a dairy activity may produce the commodities cheese and milk) while any commodity may be produced by multiple activities (for example, activities for small-scale and large-scale maize production may both produce the same maize commodity). In the activity columns, payments are made to commodities (intermediate demand), factors of production (value-added comprising of operating surplus and compensation of employees), as well as producer tax accounts. In the commodity columns, payments are made to domestic activities, the rest of the world, and various tax accounts (for domestic and import taxes). This treatment provides the data needed to model imports as perfect or imperfect substitutes vis-à-vis domestic production. Trade and Marketing Margins Domestic and international trade flows in the SAM are explicitly associated with transactions (trade and transportation) costs, also referred to as marketing margins. For each commodity, the 4 SAM accounts for the costs associated with domestic, import, and export marketing. For domestic marketing of domestic output, the marketing margin represents the cost of moving the commodity from the producer to the domestic consumer. For imports, it represents the cost of moving the commodity from the border to the domestic market, while for exports it shows the cost of moving the commodity from the producer to the border. Government Income and Payments The government is disaggregated into a core government account and different tax collection accounts, one for each tax type. This disaggregation is necessary since otherwise the economic interpretation of some payments is often ambiguous. In the SAM, direct payments between the government and other domestic institutions are reserved for transfers. Finally, payments from the government to factors (for the labor services provided by government employees) are restructured in the SAM. The preferred approach is to reallocate such payments to a commodity for government services (public administration) that pays a government service activity, which in turn, pays the labor account Domestic Non-government Institutions The domestic non-government institutions consist of households and enterprises. The enterprises earn factor incomes (a reflection of ownership of capital and/or land) and may also receive transfers from other institutions. Their incomes are used for corporate taxes, enterprise savings, and transfers to other institutions. As opposed to households, enterprises do not demand commodities. Assuming that the relevant data are available, it is preferable to have one or more accounts for enterprises when these have tax obligations and a savings behavior that are independent of the household sector. It is possible to disaggregate the enterprise sector in a manner that captures differences across enterprises in terms of tax rates, savings rates, and the shares of retained earnings that are received by different household types. For example, in some settings it may be appropriate to disaggregate enterprises into non-agricultural (with earnings from non-agricultural capital), small-scale agricultural enterprises (with earnings from land and 5 capital of small farmers), and large-scale agricultural enterprises (with earnings from land and capital of large farmers).5 Home and Final Household Consumption The SAM distinguishes between home (own) consumption of activities and marketed consumption of commodities by households. Home consumption, which in the SAM appears as payments from household accounts to activity accounts, is valued at producer prices, i.e., without marketing margins and the sales taxes that may be levied on marketed commodities. Final household consumption of marketed commodities appears as payments from household accounts to commodity accounts, valued at consumer prices that include marketing margins and commodity taxes. 3. Constructing the Prior The initial task of building a SAM involves compiling data from various sources into a SAM framework. This information comes from national accounts, household budget and labor force surveys, foreign trade statistics, government budgets, balance of payments, and various other government publications. In other words, this information is drawn from a number of sources that may use (i) different disaggregations of sectors, production factors, and socio-economic household groups, (ii) different years and/or base-year prices, and (iii) different data collection and compilation techniques. Consequently, the initial or prior SAM will inevitably include imbalances between row and column account totals. The development of a disaggregated microeconomic (Micro) SAM first requires the construction of an underlying macroeconomic (Macro) SAM that contains all economic control totals. The construction of a Macro SAM is centered around the national accounts and typically replicates the major national accounts values. 5 Due to data limitations enterprises are not disaggregated further in the Tanzanian SAMs. Instead, capital earnings from the agricultural sector are distributed directly from the agricultural capital account to households, while capital earnings from the non-agricultural sector are channeled through the enterprise account. Hence, the enterprise account in the Tanzania SAM represents non-agricultural enterprises only. 6 Table 2.1: Basic SAM Structure Activities Commodities Factors Households Enterprises Government Investment Rest of the World Total Activities marketed outputs home- consumed outputs Activity income (Gross Output) Commodities intermediate inputs transactions costs private consumption government consumption investment exports Demand Factors value-added factor income from RoW factor income Households factor income to households inter- household transfers surplus to households transfers to households transfers to households from RoW household income Enterprises factor income to enterprises transfers to enterprises transfers to enterprises from RoW enterprise income Government producer taxes, value-added tax sales taxes, tariffs, export taxes factor income to government, factor taxes transfers to government, direct household taxes surplus to government, direct enterprise taxes transfers to government from RoW government income Savings household savings enterprise savings government savings foreign savings savings Rest of the World (RoW) imports factor income to RoW surplus to RoW government transfers to RoW foreign exchange outflow Total activity expenditures supply factor expenditures household expenditures enterprise expenditures government expenditures investment foreign exchange inflow 7 Account by account, this section first outlines the construction of the prior Macro SAM and explains how each Macro SAM entry is disaggregated to arrive at the prior Micro SAM. It should be noted that the prior Macro SAM is already balanced since it is drawn from consistent national accounting data. However the prior Micro SAM is not balanced as it uses information from many data sources that are inevitably inconsistent. 6 Table 3.1 shows the Macro SAM constructed for Tanzania for the year 2000.7 Each of the entries in the SAM is discussed in detail below. The notation for the SAM entries is row then column, and the values are in billions of current (2000) Tanzanian Shillings. i. (Factors, Activities)… 6,698 This is the value of gross domestic product (GDP) at factor cost and is taken directly from national accounts (NBS, 2002d). In constructing the Micro SAM this single value is disaggregated across 43 activities and 12 factor categories.8 The initial process of disaggregation entailed deriving a level of value-added attributable to the home production factor. This was calculated by summing the total value of each household’s home consumption for each activity (home consumption is discussed in detail later in this section). The remaining value-added was then disaggregated across activities using information from both national accounts and the 1992 input-output (IO) table. The 14-sector disaggregation of GDP at factor cost from the national accounts is insufficient to provide values for the 43 activities included in the final Micro SAM. Therefore, whenever further disaggregation was necessary the IO table’s disaggregation of that particular sector was used. 6 The same data sources and methods that were used in the construction of the 2000 SAM were also used for the 1998, 1999, and 2001 SAMs. While these additional SAMs are not discussed in the text, they are available at www.ifpri.org. Since much of the survey data is for the year 2000, these additional SAMs are updates of the 2000 SAM (with new information being used wherever possible). 7 The Macro SAM in this table is not the prior but the final Macro SAM. However the differences between the two SAMs at this aggregate level is at the most very small since the information in the prior Macro SAM is largely maintained during the balancing process. The reason for displaying the final Macro SAM is that the reader will find it more convenient if the cited Macro and Micro SAMs are consistent. 8 The disaggregated activity, commodity, factor, and household categories are defined in Section 5, which outlines the final balanced Micro SAM. 8 Table 3.1: Prior Macro SAM for Tanzania for 2000 (TSh Bil.) Activities Commod- ities Factors House- holds Enter- prises Govern- ment Taxes Invest- ment Rest of World Total Activities 10,744 1,684 12,428 Commodities 5,690 303 4,382 483 1,267 1,065 13,190 Factors 6,698 6,698 Households 4,627 1,860 62 210 6,759 Enterprises 1,997 1,997 Government 1 745 746 Taxes 40 466 18 85 136 745 Savings 608 201 457 1,267 Rest of World 1,676 56 1,733 Total 12,428 13,190 6,698 6,759 1,997 746 745 1,267 1,733 Note: The Macro SAM is balanced in the prior since it is largely constructed from consistent national accounting data. The Macro SAM presented is the final Macro SAM that may have changed slightly during the balancing process. However, these changes are at the most very small. 9 The distribution of sectoral value-added across capital and labor for the non-agricultural sectors was taken from the 1992 IO table. According to the IO table, the share of capital in value-added in the agricultural sectors is above 90 percent. Given that this is an overestimation, extensive discussion was undertaken between project collaborators, country representatives, and agricultural data-specialists. Finally, non-subsistence value-added in the agricultural sectors was shared equally between capital and labor.9 Capital was then further disaggregated into land (30 percent) and capital (70 percent) value-added. These calculations produced a distribution of total GDP at factor cost across sectors, and across broadly defined labor, capital, and land factors. The labor value-added by sector was further disaggregated across labor categories using information from the 2000/01 Labor Force Survey (LFS). An imputed wage had to be calculated for agricultural labor since respondents in the LFS were not asked to state an income if they were engaged in work on their own farms. Total labor value-added by sector was imputed by taking the average wage earned by hired labor in agriculture (taken from the LFS) and then multiplying this value through by the total number of people who reported working in each agricultural sector. ii. (Commodities, Activities)… 5,690 The share of intermediate inputs to value-added was taken from the 1992 IO table. In the current Macro SAM the ratio of intermediates to value-added is 0.84. Each sectors’ demand for each intermediate commodity was determined using the sectoral information in the IO table. iii. (Taxes, Activities)… 40 The total value of activity taxes was taken from the Tanzanian Revenue Authority (2002). This figure represents payments by activities to the government for such items as business licenses, departure charges, motor vehicle taxes, road toll, stamp duty, and other non-tax revenue. The 9 One possible suggestion that might explain this high-capital intensity is that subsistence or home production is sometimes treated in the system of national accounts as gross operating surplus. 10 relative distribution of activity taxes was taken from the IO table’s estimates of other indirect taxes and subsidies. iv. (Activity, Households)… 1,684 The payment from households to activities represents households’ consumption of own production. This production is measured at producer (or farm-gate) prices. The value of total household consumption was taken from national accounts. For both the Macro and Micro SAMs this was distributed across own and final consumption using information from the 2000/2001 HBS. Respondents in the HBS were asked to keep a journal of incomes and expenditures over a period of 30 days. Regarding consumption, households were asked to state whether the commodities that were consumed were purchased, or whether they were produced or gathered by the household. In the SAM the latter was treated as own household consumption of agricultural commodities as well as processed food and beverages. According to the HBS, own consumption constitutes 27.7 percent of total household consumption. This is largely consistent with the estimates from the NBS, which, in 2000, attributed 31.8 percent of total consumption to own household production (BOT, 2001). Since the HBS did not take into account the value of owner-occupied dwellings, the initial step in disaggregating own consumption across activities involved removing the share of non- monetary GDP attributable to owner-occupied dwellings. This share was derived from national accounts and amounted to 27.9 percent of non-monetary GDP. The remaining own consumption was distributed across activities using consumption shares taken from the HBS measure of own consumption.10 v. (Activities, Commodities)… 10,744 Total marketed output is the difference between gross output (12,388) and the value of own household consumption (1,648). Gross output is the sum of intermediate demand, GDP at factor 10 There is some discussion on the consumption patterns of households in the final section of this paper. The complete list of household own and final consumption shares, as well as the SPSS code used to generate these shares, can be obtained from the authors. 11 cost, and activity taxes. While the SAM distinguishes between activities and commodities, and thus would facilitate interactions between single/multiple activities and single/multiple commodities, the IO table does not allow for this distinction. Therefore the disaggregation of this cell in the Micro SAM results in single entries along the main diagonal of the activity- commodity sub-matrix (or a one-to-one mapping between activities and commodities). vi. (Commodities, Commodities)… 303 The payment by commodities to commodities is a condensed version of the treatment of trade margins in the final Micro SAM. In the Micro SAM there are separate margin accounts for the trade costs incurred through the marketing of each commodity. This value of transaction costs is further disaggregated to distinguish between the costs incurred by imports, exports, and domestically produced and sold goods. Unlike most other entries in the Macro SAM, this entry was first calculated on a disaggregated level, and then aggregated to arrive at a final Macro SAM value. Trade costs per unit of marketed output were calculated using the survey data on trade margins by broad economic sector produced during the compilation of the 1992 IO table. These margins are shown in Table 3.2. Table 3.2: Trade Margins from the IO Table by Broad Commodity Categories (1992) Commodity Category Final and Intermediate Demand (1992 TSh Mil.) Trade Margin (1992 TSh Mil.) Trade Margin to Total Demand (Percentage) Agricultural Crops 286,245 84,779 42.1 Livestock and other animals 79,509 15,027 23.3 Hunting 2,529 106 4.4 Forestry 33,890 7,055 26.3 Manufacturing 824,109 104,073 14.5 Total 1,226,282 211,040 20.8 Source: 1992 IO Table The variation in trade margins within these broad sectors was determined by using the variation found in the differences between 1998 consumer and producer prices exclusive of sales taxes. 12 The reason for only using the variation in relative producer/consumer prices is that more confidence is placed on the detailed survey work of the IO table than on the estimates of relative prices (which are typically more difficult to measure). vii. (Taxes, Commodities)… 466 While the Macro SAM in Table 3.1 shows only a single row and column for taxes, this accounts actually consists of a number of distinct tax accounts. These include specific accounts for activity, income, sales, factor, import, and export taxes. The commodity tax entry can therefore be disaggregated to include indirect taxes (366) and import tariffs (100). There are also export taxes in the 2000 SAM, but their combined value is only 6.9 million Tanzanian Shillings. Import tariff and export tax rates by commodity category were derived from trade information from the Customs and Excise Department at the Tanzanian Revenue Authority (TRA, 2002).11 It should be noted that the implied aggregate import tariff rate of 6 percent is based on tariff collections and therefore, due to collection inefficiencies and exemptions, this might not coincide with the tariff book rate. Indirect taxes on commodity sales were taken from the Tanzanian Revenue Authority (TRA, 2002). This value was disaggregated across commodities using excise and value-added tax rates found in the trade data described above. It is assumed that the sales taxes imposed on imported commodities also apply to domestically produced goods sold domestically. viii. (Rest of World, Commodities)… 1,676 Data on merchandise imports was obtained from the Customs and Excise Department at the Tanzanian Revenue Authority (TRA, 2002). Data on imports of services were obtained from the Bank of Tanzania (BOT, 2001). Since the merchandised data was measured inclusive of cost, insurance and freight (CIF), it was necessary to reduce the value of transportation and insurance 11 As is typical of trade data, the information on imports, exports, and taxes was converted from the harmonized system (HS) of product classification into the 43 commodity categories used in the SAM. Although this mapping between HS and SAM categories is not presented in this document, it is available from the authors. 13 services imported into the country in order to avoid double counting the importing of these services. The Bank of Tanzania assumes there is a 13 percent difference between imports measured at CIF and free-on-board (FOB) prices. Therefore, this portion of the value of merchandised imports was subtracted from the value of transport and insurance imports according to the weighted shares of these services’ imports. ix. (Commodities, Households)… 4,382 As already noted in the discussion of own household consumption, the disaggregation of total household consumption from the national accounts used information from the HBS. According to this survey, final household demand constitutes 72.3 percent of total household consumption spending. Final consumption was distributed across commodities using household consumption shares calculated from the HBS raw data. x. (Commodities, Government)… 483 The value of government consumption spending is taken from national accounts. All of government spending is for the purchase of the government services commodity. In this way the government is treated as both a sector producing government services, and a demander of these services. xi. (Commodities, Investment)… 1,267 The aggregate value of investment spending is taken from national accounts. It is distributed across commodities according to the 1992 IO table’s investment shares. xii. (Commodities, Rest of World)… 1,065 The aggregate value of exports to the rest of the world is taken from national accounts (NBS, 2002). This is then disaggregated across commodities using the detailed commodity information from the Customs and Excise Department (TRA, 2002). 14 xiii. (Households, Factors)… 4,627 This value is the sum of all land, labor, and agricultural capital value-added generated during production, less any factor taxes and factor payments abroad. The distribution of labor income across households is determined using household labor-income shares as reported in the HBS. Agricultural capital and land returns are distributed according to households’ stated land ownership in the HBS.12 xiv. (Enterprises, Factors)… 1,997 The sum of all non-agricultural capital value-added (or gross operating surplus) is paid to enterprises. It is therefore assumed that all non-agricultural capital is subject to direct taxation by the government. xv. (Taxes, Factors)… 18 Factor taxes are taken from the Tanzanian Revenue Authority (TRA, 2002) and include pay-roll levies. Based on discussions with the TRA, this tax is distributed across high-educated male and female workers assuming the same tax rate on each factor’s income. xvi. (Rest of World, Factors)… 56 Factor remittances abroad are taken from the national accounts. Based on discussions with the NBS and the Bank of Tanzania, these remittances are made by agricultural (20 percent) and non- agricultural (80 percent) capital. 12 An alternative distribution of agricultural capital income could have been derived from households’ ownership of cattle, sheep and other relevant assets. Land ownership is preferred since (i) it is difficult to compile a weighted index of asset ownership when the assets are measured in incomparable units (e.g., cattle and tractors); and (ii) the distribution of livestock assets closely matches that of land ownership (this is especially true for cattle). 15 xvii. (Households, Enterprises)… 1,860 Enterprise payments to households are treated as the sum of enterprise income less direct taxes and dividend payments to the government. These payments are distributed across households according dividend earnings reported by households in the HBS. xviii. (Government, Enterprises)… 1 Dividend payments from enterprises to the government are taken from national accounts (NBS, 2002). xix. (Taxes, Enterprises)… 136 Enterprise direct taxes are taken from the Income Tax Department at the Tanzanian Revenue Authority (TRA, 2002). Among other things, this includes such items as taxes on limited companies and parastatals, training levies, excess-profit taxes, and estate duties. xx. (Households, Government)… 62 Government transfers to households are taken from national accounts (NBS, 2002). The distribution of this value across households is determined by households’ reported income from government in the HBS. xxi. (Households, Rest of World)… 201 Aggregate household income from the rest of the world is treated as a residual between households’ income and expenditure. By assumption, this is evenly distributed across higher income rural and urban households. 16 xxii. (Taxes, Households)… 85 The value of direct taxes on households is taken from the Income Tax Department at the Tanzanian Revenue Authority (TRA, 2002), and represents total P.A.Y.E. taxes. This value is distributed across households according to their relative income tax payments as reported in the HBS. xxiii. (Savings, Households)… 608 Taking total domestic savings and removing government savings from this value determines total household savings. Both of these values are reported in national accounts (NBS, 2002). Savings across households is distributed according to households’ relative savings as reported in the HBS. xxiv. (Government, Taxes)… 745 Each of the tax accounts mentioned above pay their total tax revenue to the government. xxv. (Savings, Government)… 201 Government savings is taken from national accounts (NBS, 2002). xxvi. (Savings, Rest of World)… 457 Foreign savings or borrowing is treated as a residual in the SAM in order to balance total savings with total investment spending. 17 4. The Balancing Process Although the Macro SAM is already balanced in the prior, the data used to disaggregate the SAM is from a number of possibly inconsistent sources. This results in a number of imbalances between row and column accounts in the prior Micro SAM, which must be removed in order to arrive at a final balanced SAM for Tanzania. A cross-entropy approach to SAM estimation is used for the balancing process leading from the unbalanced prior to the balanced final SAM. Since data availability and data consistency are limited, the cross-entropy approach is an appropriate tool for estimating a balanced and consistent database starting from an unbalanced database that contains all available information. The Cross-entropy Balancing Method 13 A SAM can be defined as a matrix T of monetary flows i, jT representing payments by account j to account i. Following the convention of double-entry bookkeeping, total receipts and total expenditures of a particular agent i have to be equal (i.e., respective row and column sums are balanced). This is shown below. i, j j, ii j j = =y T T ∑ ∑ Dividing every cell entry of the flow matrix T by its respective column total generates a matrix A of column coefficients: i, j i, ji, j ij T= with = 1 iA A y ∀∑ In matrix notation it follows that: y = A y⋅ 13 For a detailed discussion see Robinson et al (2001). 18 Balancing a SAM is an underdetermined estimation problem using information from various sources and for various years. The cross-entropy approach allows the incorporation of errors in variables, inequality constraints, and prior knowledge about any part of the SAM (as opposed to just row and column sums as in the RAS balancing method).14 These features of the cross- entropy estimation technique allow considerable flexibility in incorporating specific information and implementing limits to which the estimation results are restricted. The general cross-entropy approach is described by the following optimization problem.15 min ln * i, j* i, j i j i, j * ** * i, j i, jj i j j A ( )A A s.t. : = and =y yA A 1 i ⋅∑∑ ∀∑ ∑ In this equation A is a coefficient matrix representing any prior SAM. Despite being inconsistent and imbalanced, this prior SAM represents the starting point of the balancing process aimed at determining a new and balanced coefficient matrix *A .16 The described problem is set up to minimize the entropy difference between the two coefficient matrices. This becomes clearer if the optimization problem is rearranged as follows: 14 Following information theory developed by Shannon (1948) and further developed by Theil (1967) the expectation of separate information values can be described as the expected information of data points: lnn i i i-1 i p p-I( p : q ) = - q∑ where q and p are prior and posterior probabilities regarding a set of events iE and -I(p:q) is the Kullback-Leiber (1951) measure of the ‘cross-entropy’ distance between the two probability distributions. The cross-entropy approach minimizes the cross-entropy distance between the probability distributions that are consistent with the information in the data and the prior. 15 As formulated by Golan, Judge, and Robinson (1994) to update an input-output table by solving for a new coefficient matrix A, which minimizes the entropy difference between the underlying prior A and the new matrix A. 16 This means that the prior A does not need to satisfy the model y = A y⋅ , but the sum of its column coefficients adds up to one, i.e. i , j i = 1 jA ∀∑ . 19 min ln ln* * i, j i, j i, j i j ( - )A A A⋅∑∑ Additional equality and inequality constraints can be formulated as linear ‘adding-up’ constraints on various elements of the SAM. For an aggregator matrix G, which has ones for those Micro SAM entries that correspond to a certain Macro SAM aggregate and zeros otherwise, the formulation for k such aggregation constraints is given by (k)(k) i, ji, j i j = G T γ⋅∑∑ where (k)γ is the value of the aggregate and the ijT 's are the Micro SAM flows. Measurement errors in variables can be incorporated into the system through y = x + e where y is a vector of row sums and x the initially known vector of column sums measured with error. The error e is defined as a weighted average of known constants i i, w i, w w = e W v⋅∑ where w is a set of weights W, v are constants, and weights are subject to i, w i, w w = 1 with 0 1W W≤ ≤∑ For the purposes of the Tanzanian Micro SAM, a symmetric distribution around zero given lower and upper bounds is generally chosen using five weights.17 Consequently, the optimization 17 Note that if the error distribution is symmetrically centered on zero and all weights are equal – as are their initial prior values – then the respective error equals zero. 20 problem of minimizing the entropy difference now contains a term for the weights W, as shown below. min ln ln ln* * i, j i, j i, w i, wi, j i j i w ( - ) + W WA A A   ⋅ ⋅    ∑∑ ∑∑ Solving the above optimization problem in conjunction with the constraints imposed on the system derives a balanced SAM that is as close to the prior SAM as possible while still satisfying the constraints. By varying the value of the standard errors on the constraints it is possible to adjust the confidence placed on various areas within the prior. For instance, it is possible to lower the standard errors on the macro control constraints so as to ensure a closer match of the Micro SAM’s aggregates to those found in national accounts. The remainder of this section outlines the constraints used for the Tanzanian SAMs. Setting the Constraints on the Prior Based on discussions with the National Bureau of Statistics, various constraints were imposed on the model according to the perceived reliability of the data. Given the discrepancies between commodity demand and supply, it was decided that the levels of sectoral gross output, total market supply, and total demand are known only with considerable error.18 As such a high standard error was placed on the column and row totals implying that gross output, supply and demand are able to readily shift between sectors. Furthermore, the error distribution on gross output was assumed to be uniform. However, GDP at factor cost is fixed in aggregate so that it matches the national accounts value. Other macroeconomic aggregates (taken from the Macro SAM) that were maintained in the final Micro SAM include: 18 These inconsistencies between demand and supply are likely to be due to the need to reevaluate GDP, since the last rebasing of GDP took place in 1992. GDP is currently being rebased to the year 2001 and this will undoubtedly reduce the imbalances in the commodity accounts. 21 • Imports and exports • Household own and final consumption • Government consumption • Investment demand • Household and enterprise direct taxes • Import tariffs and indirect commodity taxes The HBS contains recent and detailed information on household expenditures. As a result, household consumption shares across activities and commodities were held fixed with error in order to maintain households’ consumption patterns in the final SAM.19 Finally, the ratio between certain cells was held fixed with error. This allowed, for example, the fixing of import tariff rates for individual commodities. In this case a particular commodity’s share of total imports was allowed to vary but the tariff rate imposed on the imported commodity was maintained. Other areas where fixed shares with errors were imposed include: • Import, export, and domestic transactions cost margins • Household and enterprise private savings rates All of these constraints are imposed based on considerations of data reliability. For example, data that is current is given preference over outdated information. Information that is difficult to measure is treated with less certainty than other sources. As a result, information that is seen as more reliable includes: 19 Although it is possible to simply fix certain shares within the SAM it is often preferable to allow some flexibility between the prior and final SAM shares. This would reflect the fact that all data is likely to be measured with some error and it also prevents over-constraining the estimation procedure. As mentioned earlier in this section, confidence in a particular data source, in this instance the HBS’ household consumption shares, is reflected in the size of the standard error placed around the prior estimate. In the case of household consumption shares, a relatively small standard error was chosen so as to maintain the shares in the final SAM. 22 • National accounts macro-aggregates • Import and export shares by commodity • Import tariff rates on commodities • Private savings rates Other information that were given more flexibility to vary around their expected values include: • Investment shares by commodities • Production and supply shares by sector and commodity • Household consumption shares by activity and commodity20 • Import, export, and domestic transactions cost margins Significant uncertainty is placed on intermediate commodity demands given that they are (i) difficult to measure, (ii) were last measured in 1992, and (iii) are likely to be subject to considerable variation over time. 5. Describing the Final Social Accounting Matrix Based on the data sources detailed in Section 3 and the constraints outlined in Section 4, a final SAM for Tanzanian was estimated using the cross-entropy method. The strength of the cross- entropy balancing procedure lies in the information it is given regarding the reliability of the various data sources. This section presents the structure of the final SAM in order to confirm that it accurately represents the Tanzanian economy in 2000. Micro SAM Accounts The Micro SAM is a more detailed representation of the flows shown in the Macro SAM. In total there are 43 productive activities producing 43 commodities.21 Although the SAM distinguishes 20 These shares were largely maintained during the estimation procedure regardless of the size of the standard error placed on their initial estimates. 23 between producers and commodities, the IO table does not account for differences between supply and use. Consequently, there is a direct mapping between activities and commodities. A description of these disaggregated accounts is presented in Table 5.1. Table 5.1: Micro SAM Account Descriptions Group Account Description Agricultural AMAIZE Growing of maize Sectors APADDY Growing of paddy (Activities) ASORGH Growing of sorghum or millets AWHEAT Growing of wheat ABEANS Growing of beans ACASSA Growing of cassava ACEREA Growing of other cereals AOILSE Growing of oil seeds AROOTS Growing of other roots and tubes ACOTTO Growing of cotton ACOFFE Growing of coffee ATOBAC Growing of tobacco ATEAGR Growing of tea ACASHE Growing of cashew nuts ASISAL Growing of sisal fiber ASUGAR Growing of sugar AOFRVE Growing of fruits and vegetables AOCROP Growing of other crops ALIVES Operation of poultry and livestock AFISHI Fishing and fish farms AHUFOR Hunting and forestry Non-agricultural AMININ Mining and quarrying Sectors AMEATD Processing of meat and dairy products (Activities) AGRAIN Grain milling APFOOD Processed food ABEVER Beverages and tobacco products ACLOTH Textile and leather products AWOODP Wood paper printing ACHEMI Manufacture of basic and industrial chemicals AFERTI Manufacture of fertilizers and pesticides APETRO Petroleum refineries ARUPLA Rubber, plastic, and other manufacturing AGLASS Glass and cement AMETAL Iron, steel, and metal products AEQUIP Manufacture all equipment 21 It would be possible to construct a SAM with the 79 sectors contained in IO table. However, based on discussions during the workshops, it was decided that this level of disaggregation is unnecessary, and that the IO table should be aggregated into 43 sectors. 24 Table 5.1 continued: Micro SAM Account Descriptions Group Account Description Non-agricultural AUTILI Utilities Sectors ACONST Construction (Activities) ATRADE Wholesale and retail trade AHOTEL Hotels and restaurants ATRANS Transport and communication AESTAT Real estate AADMIN Public administration, health, and education APRIVS Business and other services Agricultural CMAIZE Maize Commodities CPADDY Paddy CSORGH Sorghum or millets CWHEAT Wheat CBEANS Beans CCASSA Cassava CCEREA Other cereals COILSE Oil seeds CROOTS Other roots and tubes CCOTTO Cotton CCOFFE Coffee CTOBAC Tobacco CTEAGR Tea CCASHE Cashew nuts CSISAL Sisal fiber CSUGAR Sugar COFRVE Fruits and vegetables COCROP Other crops CLIVES Poultry and livestock CFISHI Fish CHUFOR Hunting and forestry Non-agricultural CMININ Mining and quarrying Commodities CMEATD Meat and dairy products CGRAIN Grain milling CPFOOD Processed food CBEVER Beverages and tobacco products CCLOTH Textile and leather products CWOODP Wood paper printing CCHEMI Manufacture of basic and industrial chemicals CFERTI Manufacture of fertilizers and pesticides CPETRO Petroleum refineries CRUPLA Rubber plastic and other manufacturing CGLASS Glass and cement CMETAL Iron steel and metal products CEQUIP Manufacture all equipment CUTILI Utilities CCONST Construction CTRADE Wholesale and retail trade CHOTEL Hotels and restaurants CTRANS Transport and communication 25 Table 5.1 continued: Micro SAM Account Descriptions Group Account Description Non-agricultural CESTAT Real estate Commodities CADMIN Public administration health and education CPRIVS Business and other service activities Marketing CTDTP-E Export transactions costs Margins CTDTP-D Domestic sales transactions costs CTDTP-M Import transactions costs Factors FSUB Subsistence Factor LCHILD Child labor (age 10 to 14) LNONF Female labor (no formal education) LNFPF Female labor (not finished primary school) LNFSF Female labor (not finished secondary school) LSECF Female labor (secondary or higher education) LNONM Male labor (no formal education) LNFPM Male labor (not finished primary school) LNFSM Male labor (not finished secondary school) LSECM Male labor (secondary or higher education) CAPAG Agricultural capital CAPNAG Non-agricultural capital LAND Agricultural land Households HRBFPL Rural (below food poverty line) HRFBPL Rural (between food and basic needs poverty lines) HRNOED Rural (non-poor – head with no education) HRNFPS Rural (non-poor – head not finished primary school) HRNFSS Rural (non-poor – head not finished secondary school) HRSECP Rural (non-poor – head finished secondary school) HUBFPL Urban (below food poverty line) HUFBPL Urban (between food and basic needs poverty lines) HUNOED Urban (non-poor – head with no education) HUNFPS Urban (non-poor – head not finished primary school) HUNFSS Urban (non-poor – head not finished secondary school) HUSECP Urban (non-poor – head finished secondary school) Taxes DIRTAX Direct taxes on domestic institutions IMPTAX Import tariffs EXPTAX Export taxes ACTTAX Value added or activity taxes INDTAX Indirect or sales taxes FACTAX Factor taxes Other GOV Government Institutional ROW Rest of world Accounts S-I Savings and investment Of the 43 production sectors in the SAM, 21 are in agriculture. The remaining sectors are split between mining (1), manufacturing (13), the rest of the secondary sector (2), and the tertiary sector (6). The same disaggregation applies for commodities. In addition, there are three 26 marketing margin commodity accounts for export, import, and domestic sales transaction costs discussed above. Considerable attention was paid to disaggregating factors. The first factor is the composite subsistence land, labor and capital factor used in the production of own household consumption. Since it is not possible to determine the shares of each of the factor types in this factor category, they are combined into a single factor called subsistence factor. Assuming that subsistence production uses the same technology as non-subsistence production, the share of subsistence labor value-added in total subsistence value-added is 36.2 percent. Labor was disaggregated largely according to gender and education as reported in the LFS. The exception to this was the child labor category, which includes all working children between the ages of ten and 14.22 Adult workers were divided into male and female labor categories, and then disaggregated further according to their highest level of education attained. The educational categories chosen included: (i) no formal education (including adult education); (ii) not finished primary school; (iii) not finished secondary school; and (iv) completed secondary school or higher education. Table 5.2 shows the distribution of the labor force across the various labor categories in 2000/01.23 Children are an important source of labor in Tanzania. Non-subsistence child labor accounts for 8.6 percent of the total workforce. However, as will be seen in Table 5.5, they only contribute 0.3 percent to GDP at factor cost or total value-added. Female and male adults appear to be as likely to work with 51.7 and 48.3 of the adult non-subsistence labor comprising women and men respectively. Within both male and female adult labor the largest category are workers who have not yet finished secondary school. Male workers hold a larger number of the higher educated jobs, and women hold a proportionally larger share of uneducated positions. 22 A detailed description of the LFS codes used and the SPSS syntax are available from the authors. 23 The total workforce shown in the table is lower than the official estimate of 16.9 million people. This is because a number of respondents in the LFS declared that they were working but did not specify their job. It is impossible to determine whether these respondents incorrectly answered positively to being employed, or whether there was an omission regarding their job description. As such these people have been dropped from estimates of the labor force. 27 Table 5.2: Labor Force by Labor Category (2000/01) Age and Gender Category Education Category Number of Workers Share of Total Workers Subsistence labor 5,937,131 36.2 Child labor (ages 10 to 14) 1,403,358 8.6 Female No formal education 1,527,131 9.3 Not finished primary school 672,474 4.1 Not finished secondary school 2,344,897 14.3 Secondary or higher education 143,315 0.9 Total adult female 4,687,817 28.6 Male No formal education 788,193 4.8 Not finished primary school 928,912 5.7 Not finished secondary school 2,407,857 14.7 Secondary or higher education 249,685 1.5 Total adult male 4,374,646 26.7 All labor categories 16,402,952 100.0 Source: Authors’ calculations using the Labor Force Survey 2000/01 (NBS, 2002) Capital is separated into agricultural and non-agricultural capital depending on the sector in which it is employed, and land is only used in agricultural production. Households are initially separated into rural and urban. The remaining disaggregation is based on the income level of the household and on the education of the head of the head of the household. In terms of adult-equivalent income levels, the poorest households are those below the food poverty line, followed by households who fall between the food and basic needs poverty lines. The remaining households that do not fall into either of these categories (approximately 60 percent of the population) are divided according to the highest educational attainment of the head of the household. These include: (i) no formal education (including adult education); (ii) not finished primary school; (iii) not finished secondary school; and (iv) completed secondary school or higher education. This disaggregation of households uses the poverty lines developed for the official report on the HBS (NBS, 2002b). The motivation behind using rural/urban and education in further disaggregating households is derived from the finding of the HBS report that these are the two factors which account the most for the incidence of household poverty. According to the same report, there is no greater likelihood of a household falling into poverty if the head of that household is female given that other socio-economic conditions are equal. As such, households 28 were not disaggregated according to the gender of the head of the household. The distribution of household population across the household categories is shown in Table 5.3. Of the total Tanzanian population of 31.3 million people, 80.3 percent live in rural areas. The largest concentration of the population is in those rural households whose household head has not finished secondary school. A similar concentration is also found in the urban areas. The concentration of the population in these household categories coincides with the concentration of adult labor into categories with the same attained education level. The remaining household populations are relatively evenly distributed within the urban and rural classifications. Given this relatively even distribution of the population and the identification of the two poverty lines, this household disaggregation appears to be appropriate for poverty analysis. Table 5.3: Household Population by Household Category (2000/01) Rural/ Urban Education Category Number of People Share of Total Population Rural Below food poverty line 5,080,859 16.2 Between food and basic needs poverty lines 4,605,455 14.7 Non-poor – head with no education 3,512,349 11.2 Non-poor – head not finished primary school 3,499,736 11.2 Non-poor – head not finished secondary school 7,842,113 24.9 Non-poor – head finished secondary school 661,535 2.1 Total rural 25,202,047 80.3 Urban Below food poverty line 674,816 2.2 Between food and basic needs poverty lines 712,486 2.3 Non-poor – head with no education 422,993 1.4 Non-poor – head not finished primary school 689,084 2.2 Non-poor – head not finished secondary school 2,462,953 7.9 Non-poor – head finished secondary school 1,146,635 3.7 Total Urban 6,108,967 19.7 All households (total population) 31,311,014 100.0 Source: Authors’ calculations using the Household Budget Survey 2000/01 (NBS, 2002) The Structure of Production and Trade Tanzania is largely an agricultural economy with 46.3 percent of total GDP at factor cost being generated within the agricultural sectors. Table 5.4 shows the distribution of production and trade across the sectors included in the SAM. Column one shows the importance of each sector in generating total value-added in the economy. The single largest agricultural sector in Tanzania 29 is maize production accounting for almost one tenth of total value-added. Other large sectors within crop agriculture include fruits and vegetables (6.6 percent), and paddy (3.7 percent). The animal-related and forestry sectors together contribute 11 percent towards total value-added and are therefore an important component of both agricultural and national production. Beyond agricultural production, the agro-related sectors account for 5.7 percent of value-added. Combined agricultural and agro-related production dominates GDP at factor cost. None of the remaining individual manufacturing sectors account for more than around one percent of total value-added, with total manufacturing value-added equal to only 11.9 percent of GDP at factor cost. The service and remaining secondary sectors are an important source of value-added and together amount to 40.4 percent. Within these sectors it is wholesale and retail trade that contributes the most (10.5 percent) followed by health, education and public services (6.4 percent). Turning to international trade, column two shows that over a quarter of total import expenditures are for the purchase of equipment. Other important imported commodities in order of magnitude are transport and communication services (20 percent), and petroleum products (12.14 percent). With the exception of sugar, which accounts for 2.5 percent of total imports, there are virtually no agricultural imports. The most important source of export earnings is transport and communication services (40.9 percent), which includes tourism. Two other sectors with high shares of total exports include coffee (7.9 percent) and cashew nuts (7.8 percent). Columns three and four show import to domestic demand and export to domestic output ratios. The former shows that around 90 percent of petroleum products used within Tanzania are imported from abroad. Similarly for exports, the food crops have very low export intensities, while the traditional crops have very high proportions of output sold abroad (with the exception of sugar production). Finally, the non-traditional crops have higher export intensities than the food crops but these are still substantially lower than the traditional crop categories. 30 Table 5.4: The Structure of Production and Foreign Trade in Tanzania (2000) Activity/Commodity Share of Total Value (%) Value- added Imports Exports Import- demand Ratio Export- output Ratio Maize 9.89 0.82 0.09 3.75 0.24 Paddy 3.73 1.09 0.22 6.64 0.73 Sorghum/millet 1.32 0.00 0.01 0.01 0.10 Wheat 0.23 0.96 0.00 43.72 0.19 Beans 2.35 0.00 0.09 0.02 0.64 Cassava 2.01 0.00 0.00 0.00 0.00 Other cereals 0.34 0.00 0.01 0.04 0.39 Oil seeds 1.50 0.01 0.37 0.28 4.27 Other roots/tubes 1.62 0.00 0.00 0.00 0.00 Cotton 0.63 0.00 3.46 0.01 38.41 Coffee 0.76 0.00 7.91 85.09 100.00 Tobacco 0.54 0.01 4.01 0.50 57.27 Tea 0.27 0.01 2.20 1.56 62.16 Cashew nuts 1.04 0.00 7.80 83.59 100.00 Sisal fiber 0.09 0.00 0.00 0.00 0.00 Sugar 1.60 2.53 1.06 28.57 7.50 Fruits/vegetables 6.60 0.40 2.29 2.34 7.03 Other crops 0.80 0.01 0.38 0.37 10.80 Poultry/livestock 3.29 0.14 0.54 1.31 2.49 Fish 4.04 0.01 5.43 0.06 19.09 Hunting/forestry 3.65 0.03 0.46 0.34 3.16 Mining/quarrying 1.41 0.67 1.66 12.28 15.36 Meat/dairy 2.28 0.20 0.05 1.75 0.24 Grain milling 0.66 0.78 0.56 2.43 1.02 Processed food 1.92 3.70 0.58 16.57 1.68 Beverages/tobacco 0.86 0.83 0.10 9.90 0.67 Textile/leather 2.93 3.83 1.41 18.22 4.04 Wood/paper/printing 0.95 3.41 0.47 33.09 3.70 Chemicals 0.21 5.61 0.27 64.48 4.77 Fertilizers/pesticides 0.04 0.62 0.01 51.32 0.50 Petroleum refineries 0.18 12.14 0.01 89.44 0.40 Rubber/plastics 0.23 3.06 0.11 54.84 2.32 Glass/cement 0.39 0.32 0.56 7.97 7.44 Iron/steel/metal 0.56 6.29 0.09 49.18 0.78 All equipment 0.68 28.00 0.65 82.79 6.17 Utilities 1.78 0.00 0.00 0.00 0.00 Construction 4.75 0.09 0.00 0.21 0.00 Wholesale/retail trade 10.51 0.00 0.00 0.00 0.00 Hotels/restaurants 2.60 0.00 0.00 0.00 0.00 Transport/comm. 5.70 19.96 40.86 69.14 74.46 Real estate 5.82 0.00 0.00 0.00 0.00 Health/education/Govern. 6.38 0.88 4.61 1.05 3.42 Business/services nec. 2.87 3.62 11.68 22.17 36.88 All sectors 100.00 100.00 100.00 15.44 9.57 Source: Authors’ calculations from the 2000 SAM for Tanzania 31 Based on these results, the structure of the Tanzanian economy is typical of a sub-Saharan country. There is a high dependence on agricultural production, with exports being dominated by traditional, and to lesser extent non-traditional, crops. The exporting of food crops that does take place is likely to be trade with other countries in the region. The manufacturing base is relatively small and is largely concentrated in the production of agro-related products. Finally, retail trade and public services drive production in the service sector. Table 5.5 shows percentage usage of each factor in each sector’s production. Conversely, Table 5.6 shows each factor’s usage across all sectors. As can be seen from these two tables, the subsistence factor, which represents home consumption, is largely used in the production of food and non-traditional crops. As would be expected, there is some degree of home processing of agricultural products into food and related products. The subsistence factor is also used in the production of real estate services, which represents the value of owner-occupied dwellings. The concentration of subsistence factor usage in food and non-traditional crops is even clearer in Table 5.6, where it can be seen that most of this factor is used in maize, and fruit and vegetable production. By construction, land is only used in the agricultural sectors. However, within agriculture land is somewhat concentrated in the animal-related and forestry sectors. Exceptions to this include maize, paddy, and fruits and vegetables. As noted above, child labor is an important labor category in Tanzania. Almost 70 percent of all child labor is used within the poultry and livestock, fruits and vegetables, and cotton sectors. Comparing male and female labor intensities reveals some division of labor in agricultural and agro-related production. Table 5.5 shows that more female labor is used in the production of food crops, while more male labor is used in the traditional crop and animal-related sectors. Outside of agriculture the labor division is more pronounced. Very little female labor is used in manufacturing and services, with the exception of the hotel and catering, and textile sectors, which use higher skilled female labor. Manufacturing and service production is dominated by male labor. 32 Table 5.5: Factor Employment within Sectors in Tanzania (2000) Female Labor by Education Level Activity Subsist Factor Land Child Labor None NFP NFS Sec+ Maize 67.8 4.6 0.2 1.8 1.2 8.1 0.0 Paddy 23.0 11.2 0.2 1.9 2.1 18.8 0.1 Sorghum/millet 66.3 5.0 0.4 3.9 0.6 5.2 0.0 Wheat 5.4 14.2 0.0 0.0 0.0 0.0 0.0 Beans 30.3 10.3 0.0 3.8 2.5 20.4 0.1 Cassava 81.0 2.8 0.1 1.1 0.6 3.4 0.0 Other cereals 12.2 13.1 0.5 5.5 2.3 19.4 0.1 Oil seeds 27.0 10.9 0.6 3.4 1.6 17.1 0.0 Other roots/tubes 53.1 7.0 0.8 1.6 1.3 12.0 0.0 Cotton 0.0 14.9 5.2 1.1 3.1 12.2 0.0 Coffee 8.4 13.7 0.0 2.1 1.2 15.4 0.0 Tobacco 0.0 14.9 0.0 2.8 2.3 14.9 0.2 Tea 1.6 14.8 0.0 0.0 0.0 0.0 0.0 Cashew nuts 0.0 14.9 1.0 3.1 0.7 10.5 0.0 Sisal fiber 0.0 15.0 0.0 5.4 0.0 0.0 0.0 Sugar 1.5 14.5 0.0 0.0 27.4 0.0 0.0 Fruits/vegetables 40.2 8.6 0.7 1.2 1.0 16.5 0.3 Other crops 44.3 8.3 1.1 1.0 0.0 12.0 0.1 Poultry/livestock 18.4 12.0 3.9 2.4 1.0 17.8 0.8 Fish 6.0 14.0 0.0 2.0 0.0 3.4 0.0 Hunting/forestry 60.2 5.8 0.0 0.0 0.0 11.0 0.0 Mining/quarrying 0.0 0.0 0.1 0.1 0.0 0.0 0.1 Meat/dairy 87.1 0.0 0.0 0.0 0.0 0.1 0.2 Grain milling 0.0 0.0 0.3 0.5 3.5 25.5 0.0 Processed food 16.3 0.0 0.0 0.2 0.1 0.7 4.0 Beverages/tobacco 13.1 0.0 0.0 0.0 0.0 0.1 0.4 Textile/leather 0.0 0.0 0.2 0.8 0.8 14.4 4.8 Wood/paper/printing 0.0 0.0 0.0 0.0 0.0 0.3 0.6 Chemicals 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Fertilizers/pesticides 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Petroleum refineries 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Rubber/plastics 0.0 0.0 0.0 0.0 0.0 4.1 2.2 Glass/cement 0.0 0.0 0.0 0.1 0.0 0.0 0.0 Iron/steel/metal 0.0 0.0 0.0 0.0 0.0 0.0 0.5 All equipment 0.0 0.0 0.0 0.0 0.0 0.5 0.2 Utilities 0.0 0.0 0.0 0.0 0.0 0.5 0.4 Construction 0.0 0.0 0.2 0.1 0.0 0.3 0.7 Wholesale/retail trade 0.0 0.0 0.0 0.1 0.2 1.3 0.4 Hotels/restaurants 0.0 0.0 0.0 1.2 1.7 8.9 0.8 Transport/comm. 0.0 0.0 0.0 0.0 0.0 0.6 1.1 Real estate 76.1 0.0 0.0 0.0 0.0 0.0 0.7 Public services 0.0 0.0 0.0 0.1 0.4 8.8 23.4 Business/services nec. 0.0 0.0 0.1 0.1 0.5 3.8 3.3 All sectors 25.1 4.1 0.3 0.9 1.1 7.0 2.1 Source: Tanzanian SAM 2000. NFP means ‘not finished primary school’; NFS means ‘not finished secondary school’; and Sec+ means ‘finished secondary or tertiary school. 33 Table 5.5 continued: Factor Employment within Sectors in Tanzania (2000) Male Labor by Education Level Capital Activity None NFP NFS Sec+ Agri Non-agri All Factors Maize 0.8 2.2 2.5 0.2 10.6 0.0 100.0 Paddy 1.1 6.3 8.4 0.8 26.2 0.0 100.0 Sorghum/millet 1.8 2.0 2.9 0.1 11.7 0.0 100.0 Wheat 0.0 0.0 47.4 0.0 33.0 0.0 100.0 Beans 1.0 4.0 3.3 0.2 24.0 0.0 100.0 Cassava 0.6 2.0 1.6 0.2 6.6 0.0 100.0 Other cereals 2.7 6.3 6.9 0.4 30.7 0.0 100.0 Oil seeds 2.3 6.7 5.0 0.2 25.4 0.0 100.0 Other roots/tubes 0.0 4.3 3.6 0.2 16.3 0.0 100.0 Cotton 3.0 11.4 14.0 0.2 34.8 0.0 100.0 Coffee 2.3 8.9 15.4 0.6 31.9 0.0 100.0 Tobacco 3.4 10.4 14.0 2.2 34.8 0.0 100.0 Tea 0.0 49.2 0.0 0.0 34.4 0.0 100.0 Cashew nuts 5.5 12.3 17.2 0.0 34.9 0.0 100.0 Sisal fiber 13.7 11.0 16.7 3.2 35.0 0.0 100.0 Sugar 0.0 22.8 0.0 0.0 33.8 0.0 100.0 Fruits/vegetables 1.7 3.8 5.2 0.4 20.2 0.0 100.0 Other crops 3.6 7.0 2.9 0.2 19.4 0.0 100.0 Poultry/livestock 3.5 5.5 5.9 0.7 28.0 0.0 100.0 Fish 6.9 15.6 19.5 0.0 32.6 0.0 100.0 Hunting/forestry 1.1 0.0 7.4 0.8 13.6 0.0 100.0 Mining/quarrying 0.0 0.1 1.3 0.1 0.0 98.3 100.0 Meat/dairy 0.0 0.2 0.5 0.0 0.0 11.9 100.0 Grain milling 1.7 2.3 34.8 7.7 0.0 23.7 100.0 Processed food 0.3 0.5 4.8 3.2 0.0 69.9 100.0 Beverages/tobacco 0.0 0.0 1.1 15.9 0.0 69.3 100.0 Textile/leather 0.4 4.0 21.4 5.3 0.0 48.0 100.0 Wood/paper/printing 1.4 1.0 15.9 4.0 0.0 76.8 100.0 Chemicals 0.0 0.0 81.4 0.0 0.0 18.6 100.0 Fertilizers/pesticides 0.0 0.0 70.6 0.0 0.0 29.4 100.0 Petroleum refineries 0.0 0.0 24.5 3.0 0.0 72.5 100.0 Rubber/plastics 0.0 0.3 11.1 4.0 0.0 78.3 100.0 Glass/cement 0.2 0.4 18.5 3.6 0.0 77.2 100.0 Iron/steel/metal 1.2 3.0 10.4 8.2 0.0 76.8 100.0 All equipment 0.0 2.2 2.7 3.5 0.0 90.9 100.0 Utilities 1.1 0.8 7.9 9.7 0.0 79.8 100.0 Construction 1.3 7.3 45.9 11.4 0.0 32.8 100.0 Wholesale/retail trade 0.2 0.5 2.8 2.2 0.0 92.4 100.0 Hotels/restaurants 0.1 0.3 6.9 3.3 0.0 76.8 100.0 Transport/comm. 0.1 0.3 4.1 6.1 0.0 87.7 100.0 Real estate 0.0 0.0 0.8 5.5 0.0 16.8 100.0 Public services. 0.1 1.3 14.2 47.5 0.0 4.3 100.0 Business/services nec. 0.4 2.4 12.5 8.5 0.0 68.5 100.0 All sectors 1.1 3.5 9.0 5.7 9.6 30.5 100.0 Source: Tanzanian SAM 2000. NFP means ‘not finished primary school’; NFS means ‘not finished secondary school’; Sec+ means ‘finished secondary or tertiary school; and Agri means ‘Agriculture’; 34 Table 5.6: Factor Employment across Sectors in Tanzania (2000) Female Labor by Education Level Activity Subsist Factor Land Child Labor None NFP NFS Sec+ Maize 26.7 10.9 5.3 19.7 11.2 11.4 0.1 Paddy 3.4 10.2 2.1 7.5 7.4 10.0 0.2 Sorghum/millet 3.5 1.6 1.6 5.6 0.7 1.0 0.0 Wheat 0.1 0.8 0.0 0.0 0.0 0.0 0.0 Beans 2.8 5.9 0.0 9.8 5.4 6.8 0.1 Cassava 6.5 1.4 0.9 2.4 1.1 1.0 0.0 Other cereals 0.2 1.1 0.5 2.0 0.7 0.9 0.0 Oil seeds 1.6 4.0 2.8 5.4 2.2 3.6 0.0 Other roots/tubes 3.4 2.7 4.1 2.8 2.0 2.7 0.0 Cotton 0.0 2.3 10.8 0.7 1.8 1.1 0.0 Coffee 0.3 2.5 0.0 1.8 0.9 1.7 0.0 Tobacco 0.0 2.0 0.0 1.6 1.2 1.1 0.1 Tea 0.0 1.0 0.0 0.0 0.0 0.0 0.0 Cashew nuts 0.0 3.8 3.4 3.5 0.6 1.5 0.0 Sisal fiber 0.0 0.3 0.0 0.5 0.0 0.0 0.0 Sugar 0.1 5.6 0.0 0.0 40.9 0.0 0.0 Fruits/vegetables 10.6 13.8 16.1 8.8 6.2 15.5 1.0 Other crops 1.4 1.6 2.8 0.9 0.0 1.4 0.0 Poultry/livestock 2.4 9.6 42.4 8.6 3.0 8.3 1.3 Fish 1.0 13.7 0.0 8.8 0.0 1.9 0.0 Hunting/forestry 8.8 5.2 0.0 0.0 0.0 5.7 0.0 Mining/quarrying 0.0 0.0 0.4 0.1 0.0 0.0 0.1 Meat/dairy 7.9 0.0 0.0 0.0 0.0 0.0 0.2 Grain milling 0.0 0.0 0.7 0.4 2.2 2.4 0.0 Processed food 1.2 0.0 0.0 0.5 0.2 0.2 3.7 Beverages/tobacco 0.4 0.0 0.0 0.0 0.0 0.0 0.2 Textile/leather 0.0 0.0 1.6 2.6 2.1 6.0 6.7 Wood/paper/printing 0.0 0.0 0.1 0.0 0.0 0.0 0.3 Chemicals 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Fertilizers/pesticides 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Petroleum refineries 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Rubber/plastics 0.0 0.0 0.0 0.0 0.0 0.1 0.2 Glass/cement 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Iron/steel/metal 0.0 0.0 0.0 0.0 0.0 0.0 0.1 All equipment 0.0 0.0 0.0 0.0 0.0 0.0 0.1 Utilities 0.0 0.0 0.0 0.0 0.0 0.1 0.3 Construction 0.0 0.0 3.0 0.3 0.0 0.2 1.5 Wholesale/retail trade 0.0 0.0 0.5 1.3 2.3 1.9 1.9 Hotels/restaurants 0.0 0.0 0.2 3.3 4.2 3.3 1.0 Transport/comm. 0.0 0.0 0.0 0.0 0.0 0.5 3.0 Real estate 17.7 0.0 0.0 0.0 0.0 0.0 2.1 Public services 0.0 0.0 0.0 0.7 2.3 7.9 71.1 Business/services nec. 0.0 0.0 0.9 0.4 1.2 1.6 4.6 All sectors 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Source: Tanzanian SAM 2000. NFP means ‘not finished primary school’; NFS means ‘not finished secondary school’; and Sec+ means ‘finished secondary or tertiary school. 35 Table 5.6 continued: Factor Employment across Sectors in Tanzania (2000) Male Labor by Education Level Capital Activity None NFP NFS Sec+ Agri Non-agri All Factors Maize 7.2 6.2 2.7 0.3 10.9 0.0 9.9 Paddy 3.9 6.7 3.5 0.5 10.2 0.0 3.7 Sorghum/millet 2.2 0.8 0.4 0.0 1.6 0.0 1.3 Wheat 0.0 0.0 1.2 0.0 0.8 0.0 0.2 Beans 2.1 2.7 0.9 0.1 5.9 0.0 2.4 Cassava 1.1 1.1 0.4 0.1 1.4 0.0 2.0 Other cereals 0.8 0.6 0.3 0.0 1.1 0.0 0.3 Oil seeds 3.1 2.8 0.8 0.0 3.9 0.0 1.5 Other roots/tubes 0.0 2.0 0.7 0.0 2.7 0.0 1.6 Cotton 1.7 2.0 1.0 0.0 2.3 0.0 0.6 Coffee 1.6 1.9 1.3 0.1 2.5 0.0 0.8 Tobacco 1.6 1.6 0.8 0.2 2.0 0.0 0.5 Tea 0.0 3.7 0.0 0.0 1.0 0.0 0.3 Cashew nuts 5.2 3.6 2.0 0.0 3.8 0.0 1.0 Sisal fiber 1.2 0.3 0.2 0.1 0.3 0.0 0.1 Sugar 0.0 10.3 0.0 0.0 5.6 0.0 1.6 Fruits/vegetables 10.3 7.0 3.8 0.5 13.8 0.0 6.6 Other crops 2.6 1.6 0.3 0.0 1.6 0.0 0.8 Poultry/livestock 10.6 5.1 2.1 0.4 9.6 0.0 3.3 Fish 25.1 17.9 8.8 0.0 13.7 0.0 4.0 Hunting/forestry 3.7 0.0 3.0 0.5 5.2 0.0 3.6 Mining/quarrying 0.0 0.0 0.2 0.0 0.0 4.5 1.4 Meat/dairy 0.0 0.1 0.1 0.0 0.0 0.9 2.3 Grain milling 1.0 0.4 2.6 0.9 0.0 0.5 0.7 Processed food 0.6 0.3 1.0 1.1 0.0 4.4 1.9 Beverages/tobacco 0.0 0.0 0.1 2.4 0.0 1.9 0.9 Textile/leather 1.0 3.3 7.0 2.7 0.0 4.6 2.9 Wood/paper/printing 1.2 0.3 1.7 0.7 0.0 2.4 1.0 Chemicals 0.0 0.0 1.9 0.0 0.0 0.1 0.2 Fertilizers/pesticides 0.0 0.0 0.3 0.0 0.0 0.0 0.0 Petroleum refineries 0.0 0.0 0.5 0.1 0.0 0.4 0.2 Rubber/plastics 0.0 0.0 0.3 0.2 0.0 0.6 0.2 Glass/cement 0.1 0.0 0.8 0.2 0.0 1.0 0.4 Iron/steel/metal 0.6 0.5 0.6 0.8 0.0 1.4 0.6 All equipment 0.0 0.4 0.2 0.4 0.0 2.0 0.7 Utilities 1.7 0.4 1.6 3.0 0.0 4.7 1.8 Construction 5.8 9.8 24.3 9.6 0.0 5.1 4.7 Wholesale/retail trade 1.6 1.4 3.3 4.0 0.0 31.8 10.5 Hotels/restaurants 0.1 0.2 2.0 1.5 0.0 6.6 2.6 Transport/comm. 0.5 0.5 2.6 6.1 0.0 16.4 5.7 Real estate 0.0 0.0 0.5 5.6 0.0 3.2 5.8 Public services. 0.8 2.4 10.1 53.4 0.0 0.9 6.4 Business/services nec. 1.0 1.9 4.0 4.3 0.0 6.5 2.9 All sectors 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Source: Tanzanian SAM 2000. NFP means ‘not finished primary school’; NFS means ‘not finished secondary school’; Sec+ means ‘finished secondary or tertiary school; and Agri means ‘Agriculture’. 36 Consulting Table 5.6 shows that 40.9 percent of female labor that has not finished primary school is employed in the sugar sector. This large share highlights the important distinction that is made the SAM between subsistence and non-subsistence factors. In this case there are similarly educated women working elsewhere in the economy but they are largely employed in subsistence production, and as such are not distinguished in this SAM. Consequently, all categories (except for the first one) should be interpreted as non-subsistence factors, with the subsistence factor containing an amalgamation of the various labor and non-labor factors. Agricultural capital is used most intensively in the traditional crop sectors, where estate production is more prevalent. However, the importance of capital in agricultural production is substantially lower than in non-agricultural production. As expected, the most capital-intensive sector is the mining sector, followed by manufacturing. However, the largest concentration of total non-agricultural capital is not within these sectors. With the exception of real estate, and public and private services, the services sector is also highly capital intensive and contains most of the country’s non-agricultural capital resources. Household Income and Expenditure Starting with household income, Table 5.7 shows how the income earned by each income source is distributed across households. For the sake of convenience those households that lie above the basic needs poverty lines have been aggregated. However in the final SAM in the appendix these households are further divided into categories based on the education of the head of the household. Table 5.8 shows the distribution of household income across each of the income sources. In interpreting these tables it is important to keep in mind each households’ share of the total population as shown in Table 5.3 above. 37 Table 5.7: Distribution of Income Sources across Households (2000) Rural Urban Income Source BFL BFBL AFBL BFL BFBL AFBL All Subsistence factor 8.9 13.1 67.8 0.3 0.4 9.5 100.0 Child labor 5.3 5.4 52.4 0.5 4.9 31.5 100.0 Female (None) 12.0 11.6 57.4 2.6 2.4 14.0 100.0 Female (NFP) 7.0 7.4 56.2 1.6 2.5 25.3 100.0 Female (NFS) 2.1 3.4 49.0 1.2 2.2 42.1 100.0 Female (Sec+) 0.0 0.7 18.3 0.6 1.8 78.6 100.0 Male (None) 13.5 21.3 52.0 1.5 1.2 10.5 100.0 Male (NFP) 8.8 9.2 54.4 1.8 2.4 23.4 100.0 Male (NFS) 2.9 4.2 54.1 1.0 1.6 36.2 100.0 Male (Sec+) 0.6 0.9 31.4 0.3 0.9 65.9 100.0 Agricultural capital 12.4 15.5 65.1 0.9 0.5 5.6 100.0 Land 12.2 15.3 64.9 1.0 0.6 6.0 100.0 Non-agricultural capital 0.6 1.6 48.3 2.4 6.2 40.9 100.0 Government 6.9 7.0 64.8 1.0 0.8 19.5 100.0 Rest of world 0.0 0.0 45.7 0.0 0.0 54.3 100.0 Total household income 5.2 7.3 54.9 1.2 2.4 29.0 100.0 Source: Tanzanian SAM 2000. NFP means ‘not finished primary school’; NFS means ‘not finished secondary school’; Sec+ means ‘finished secondary or tertiary school; BFL means ‘below food poverty line’; BFBL means ‘between food and basic needs poverty line’; and ABL means above basic needs poverty line. Table 5.8: Household Income by Income Source (2000) Rural Urban Income Source BFL BFBL AFBL BFL BFBL AFBL All Subsistence factor 42.9 44.7 31.1 5.7 5.7 8.2 24.8 Child labor 0.3 0.2 0.3 0.1 0.1 0.4 0.3 Female (None) 2.1 1.5 1.2 2 2 1.1 0.9 Female (NFP) 1.4 1.1 1.2 1.5 1.5 1.8 1.1 Female (NFS) 2.8 3.2 4.0 6.9 6.9 7.5 7.0 Female (Sec+) 0.0 0.2 1.5 1.1 1.1 4.8 2.0 Male (None) 2.9 3.2 1.4 1.3 1.3 1.2 1.1 Male (NFP) 5.9 4.4 4.2 5.3 5.3 6.0 3.5 Male (NFS) 4.9 5.2 5.2 7.4 7.4 6.2 8.9 Male (Sec+) 0.7 0.7 7.9 1.5 1.5 10.8 5.4 Agricultural capital 22.3 19.9 11.0 7.3 7.3 2.0 9.4 Land 9.6 8.6 4.7 3.5 3.5 1.0 4.1 Non-agricultural capital 3.0 6.2 21.6 55.6 55.6 42.7 27.5 Government 1.2 0.9 1.1 0.8 0.8 0.7 0.9 Rest of world 0.0 0.0 3.5 0 0 5.8 3.1 Total household income 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Source: Tanzanian SAM 2000. NFP means ‘not finished primary school’; NFS means ‘not finished secondary school’; Sec+ means ‘finished secondary or tertiary school; BFL means ‘below food poverty line’; BFBL means ‘between food and basic needs poverty line’; and ABL means above basic needs poverty line. 38 As expected, subsistence factor income is almost entirely earned by rural households, with the poorer rural households being most dependent on this income source. Most of total land and agricultural capital income is earned by higher income households, with rural households earning more than urban households. Given their lower access to subsistence factor income, poorer urban households are more dependent on agricultural capital and land income. However these households do receive a greater share of the indirect returns to non-agricultural capital (i.e., enterprise income), and are also most dependent on this income source. Rural households receive more of and are more dependent on transfer payments from government. Finally, a greater share of higher educated labor income accrues to urban households, despite their lower share of the population. Table 5.9 shows how total household consumption spending is distributed across broad commodity categories. On average 64.7 percent of all households consumption spending is on agricultural and food products.24 Higher-income rural households spend more on food than similarly classified urban households. Amongst poorer households, those below the food poverty line spend less of their income on food consumption than those households between the two poverty lines. Only high-income urban households spend a substantial amount on services. Table 5.9: Household Consumption Spending by Broad Commodity Category (2000) Rural Urban All Commodity category BFL BFBL AFBL BFL BFBL AFBL Agriculture and food 66.9 68.2 65.2 67.7 72.0 61.4 64.7 Other manufactured goods 15.1 13.0 16.4 22.5 16.0 19.1 16.9 Utilities 0.5 0.5 0.8 1.3 1.0 1.4 0.9 Real estate 13.5 14.5 11.2 2.7 3.3 1.7 8.6 Other services 4.0 3.8 6.4 5.8 7.7 16.4 8.9 All commodities 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Source: Tanzanian SAM 2000. BFL means ‘below food poverty line’; BFBL means ‘between food and basic needs poverty line’; and ABL means above basic needs poverty line. 24 Food products include meat and dairy products, grain milling, food, and beverages. 39 Government, Savings, Investment and the Rest of the World Government income is generated through its tax receipts. The most important of these are indirect and direct taxes, which contributes 49.1 and 29.5 percent to total government revenue. Import taxes are another important revenue source. Government expenditure is dominated by consumption spending, although the government did manage to save 27 percent of its income in 2000. Table 5.10: Government Income and Expenditure (2000) Income Value (Tsh Bil) Share Expenditure Value (Tsh Bil) Share Enterprise dividends 1 0.2 Consumption spending 483 64.7 Direct taxes 220 29.5 Transfers to households 62 8.3 Import taxes 100 13.3 Savings 201 27.0 Export taxes 0.0 0.0 Value-added taxes 40 5.4 Indirect taxes 367 49.1 Factor taxes 18 2.4 Total 746 100.0 Total 746 100.0 Source: Authors’ calculations from the 2000 SAM. Total savings in Tanzania is heavily dependent on the inflow of foreign savings. Total savings received from abroad amount to over 36 percent of total savings available for investment. Remaining savings come from households (48 percent), and the government (15.9 percent). All savings is directed towards gross fixed capital formation, which represents both public and private investment in the economy. Table 5.11: Savings and Investment (2000) Savings Value (Tsh Bil) Share Investment Value (Tsh Bil) Share Households 608 48.0 Fixed capital formation 1,267 100.0 Government 201 15.9 Foreign 457 36.1 Total 1,267 100.0 Total 1,267 100.0 Source: Authors’ calculations from the 2000 SAM. The final account in the SAM is the current account. As can be seen from Table 5.12, the value of imports far exceeds the value of exports. This large trade deficit is made possible by the 40 inflow of foreign capital into the country. While some of this inflow is foreign borrowing, much of it is in the form of foreign aid (grants). As such, while the deficit is a substantial share of total receipts, it should in part be interpreted as a dependence on foreign goodwill. In the case of Tanzania the inflow of foreign aid has been extensive. Table 5.12: Tanzanian Current Account (2000) Receipts Value (Tsh Bil) Share Payments Value (Tsh Bil) Share Exports 1,065 61.5 Imports 1,676 96.7 Household remittances 210 12.1 Factor remittances 56 3.3 Deficit 457 26.4 Total 1,733 100.0 Total 1,733 100.0 Source: Authors’ calculations from the 2000 SAM. 6. Summary The objective of this research has been to not only compile a SAM for Tanzania using the recently released household budget and labor force surveys, but also to construct a framework that can be used to update the SAM as new data becomes available. While only the Tanzanian SAM for the year 2000 has been discussed in detail, the same process of construction applies to the SAMs for 1998, 1999, and 2001. Furthermore, the process of compiling and balancing the SAM is described in the appendix, and the appropriate GAMS code used during this process is available from the International Food Policy Research Institute. References BOT (2001) Economic Bulletin for the Quarter Ended June 30, 20001. Vol. XXXI, No.2. Bank of Tanzania, Dar es Salaam, Tanzania. Brooke, A., Kendrick, D., Meeraus, A., and Raman, R. (1998) GAMS: A User’s Guide. Washington, D.C. 41 Lofgren, H., Harris, R.L. and Robinson, S. (2001) “A Standard Computable General Equilibrium (CGE) Model in GAMS”. Trade and Macroeconomics Division Discussion Paper No. 75, International Food Policy Research Institute, Washington, D.C. NBS (1999) Input-Output Table of Tanzania for 1992. National Bureau of Statistics, Dar es Salaam, Tanzania. NBS (2002a) Household Budget Survey 2000/01, Vol. V. National Bureau of Statistics, Dar es Salaam, Tanzania. NBS (2002b) Household Budget Survey 2000/01: Final Report. National Bureau of Statistics, Dar es Salaam, Tanzania. NBS (2002c) Labor Force Survey 2000/01. National Bureau of Statistics, Dar es Salaam, Tanzania. NBS (2002d) Revised National Accounts for Tanzania, 1988-2001. National Bureau of Statistics, Dar es Salaam, Tanzania. Pyatt, G. (1988). “A SAM Approach to Modeling.” Journal of Policy Modeling, Vol. 10. Pyatt, G. and J. Round (1985). Social Accounting Matrices: A Basis for Planning. World Bank, Washington, D.C. Reinert, K. A., and D. W. Roland-Holst (1997). “Social Accounting Matrices,” in J. F. Francois, and K. A. Reinert (eds.) Applied Methods for Trade Policy Analysis: A Handbook. Cambridge University Press, New York. Robinson, S., Cattaneo, A., and El-Said, M. (2001) “Updating and Estimating a Social Accounting Matrix Using Cross Entropy Methods.” Economic Systems Research, Vol. 13. 42 Robinson S. and D.W. Roland-Holst (1988). “Macroeconomic Structure and Computable General Equilibrium Models.” Journal of Policy Modeling, Vol. 10. TRA (2002). Tax Revenue Report. Tanzanian Revenue Authority, Dar es Salaam, Tanzania. Wobst, P. (1998) “A 1992 Social Accounting Matrix for Tanzania.” Trade and Macroeconomics Division Discussion Paper No. 30, International Food Policy Research Institute, Washington, D.C. Appendix 1: The SAM Construction Process This appendix outlines the process of compiling the database and constructing the final SAM. This is done in three stages: 1. New data from the government publications are collected for the relevant year and entered into the Excel data file. 2. The prior SAM is constructed using information from the data file. This prior SAM is then scrutinized for any obvious inconsistencies between the various data sources. 3. The SAM is balanced using cross-entropy and a final SAM is produced. This final SAM is then scrutinized for any substantial differences between the prior and final SAMs. Since this process is clearly defined, it has largely been automated. While the example below has been developed for Tanzania, with some adjustments this process could be adapted to fit most countries. Compiling the Database The file containing the data for the SAM is ‘MasterDataFile.xls’. As described on the first worksheet, this file contains three types of data: (i) data that needs to be collected annually; (ii) data that is only available from periodic surveys; and (iii) data that is compiled using information from the first two types of data. Each of these required data is discussed in turn. 43 i. Annual Data (Orange sheets) Since this information is available annually, it needs to be updated for each new SAM. The relevant worksheets that require such information include: • ‘GDP (Nat Acc)’ – National Accounts (NBS) • ‘Taxes (Inl. Rev)’ – Income Tax Dept. (TRA) • ‘Trade (C&E)’ – Customs and Excise (TRA) • ‘GDP (BoT)’ – National Accounts (BOT) ii. Periodic Data (Green sheets) Since this information is drawn from a variety of periodic surveys and publications, it is only necessary to enter new data when new surveys and publications become available.25 The relevant worksheets that require such information include: • ‘Lab Inc (LFS)’ – Labor Force Survey (NBS) • ‘HHD Inc-Exp (HBS)’ – Household Budget Survey (NBS) • ‘HHD Cons (HBS)’ – Household Budget Survey (NBS) • ‘IO Table (NBS)’ – Input-output Table (NBS) • ‘Margins (IO)’ – Input-output Table (NBS) iii. Automatic Data (Purple sheets) This information is compiled using information from the annual and survey data. For example the Macro SAM combines information from national accounts, the household budget survey, the tax tables, and the input-output table. The various tables included in this 25 The SPSS syntax used in the compilation of these tables is available from the authors. This includes a detailed outline of the commodity classification used in the HBS, and the classification of households and labor in the HBS and LFS. 44 section are read into the General Algebraic Modeling System (GAMS) and are used during the construction of the prior SAM.26 The relevant worksheets that contain such information include: • ‘MacroSAM’ • ‘Value Added’ • ‘Trade’ • ‘Taxes’ • ‘Sav-Inv’ Constructing the Prior SAM Two GAMS files are needed to construct the prior SAM. The first file (‘LoadData.gms’) loads the necessary data from the data file constructed in stage one. The second file (‘Prior.gms’) uses this data to compile the prior Micro SAM. This is then exported to the SAM file (‘MasterSAMFile.xls’) where the prior Macro SAM and prior Micro Coefficient SAM are constructed.27 The sheets in the SAM file containing information on the prior include: • ‘PriorMicro’ – Direct output from GAMS • ‘PriorMacro’ – Compiled from the prior Micro SAM • ‘PriorCoeff’ – Compiled from the prior Micro SAM At this stage the prior Macro SAM should be checked for differences in the row and column account totals. If differences exist then there is likely to be some error in the entry of the data in the data file. The prior Micro SAM should also be checked for any substantial differences between demand and supply in the commodity accounts, and between income and expenditure in the household accounts. If such differences exist then inconsistencies exist between the various data sources. The prior data should only be adjusted if there exists new information on how to 26 For details on GAMS see Brooke et al (1998). 27 The SAM file must be closed when the prior is constructed in GAMS. 45 correct for these differences. In the absence of additional and confirmed information, it is preferable to allow the balancing procedure to adjust for these differences. Balancing the SAM The final stage of constructing the SAM involves balancing the prior SAM using cross-entropy estimation techniques (as described in Section 4). The SAM is balanced by running the cross- entropy code in GAMS (‘Balance.gms’). This code has been tested on SAMs for 1998 to 2001 and has found solutions for each of those years without requiring any adjustment to the code.28 It therefore can be expected that in the case of Tanzania the SAM construction and updating process for subsequent years will be easily achieved as well. The prior SAM is loaded from the SAM file (‘MasterSAMFile.xls’, Sheet ‘PriorMicro’), and after balancing is exported back to the same file (‘MasterSAMFile.xls’, Sheet ‘FinalMicro’).29 The sheets containing information on the final Macro and Micro SAMs include: • ‘FinalMicro’ – Direct output from GAMS • ‘FinalMacro’ – Compiled from the final Micro SAM • ‘FinalCoeff’ – Compiled from the final Micro SAM • ‘FinalDiff’ – Absolute difference between cells in the prior and final SAMs. • ‘FinalPerc’ – Percentage difference between cells in the prior and final SAMs. 28 It is possible to place more stringent constraints on the cross-entropy model and find final SAMs that are closer to the prior, however the improvements are typically small. 29 The SAM file must be closed since the prior SAM is read into GAMS and then the final SAM is exported back into the SAM file. 46 Appendix 2: Final 2000 Micro SAM for Tanzania (Tsh Bil.) AMAIZE APADDY ASORGH AWHEAT ABEANS ACASSA ACEREA AOILSE AROOTS ACOTTO ACOFFE ATOBAC ATEAGR ACASHE ASISAL ASUGAR AOFRVE AOCROP AMAIZE APADDY ASORGH AWHEAT ABEANS ACASSA ACEREA AOILSE AROOTS ACOTTO ACOFFE ATOBAC ATEAGR ACASHE ASISAL ASUGAR AOFRVE AOCROP ALIVES AFISHI AHUFOR AMININ AMEATD AGRAIN APFOOD ABEVER ACLOTH AWOODP ACHEMI AFERTI APETRO ARUPLA AGLASS AMETAL AEQUIP AUTILI ACONST ATRADE AHOTEL ATRANS AESTAT AADMIN APRIVS CMAIZE 42 CPADDY 82 CSORGH 2 CWHEAT 1 CBEANS 24 CCASSA 3 CCEREA 3 COILSE 7 CROOTS 6 CCOTTO 2 CCOFFE 4 CTOBAC 8 CTEAGR 2 CCASHE 1 CSISAL 3 CSUGAR 27 COFRVE 10 COCROP 1 CLIVES 7 2 0 CFISHI CHUFOR CMININ 47 Appendix 2 continued: Final 2000 Micro SAM for Tanzania (Tsh Bil.) AMAIZE APADDY ASORGH AWHEAT ABEANS ACASSA ACEREA AOILSE AROOTS ACOTTO ACOFFE ATOBAC ATEAGR ACASHE ASISAL ASUGAR AOFRVE AOCROP CMEATD CGRAIN CPFOOD CBEVER CCLOTH 1 3 7 2 1 0 0 0 0 5 1 1 0 0 CWOODP 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 CCHEMI 0 0 0 1 0 0 0 0 0 CFERTI 0 1 0 0 1 0 0 7 2 3 4 3 0 0 CPETRO 0 1 2 3 2 1 1 0 1 CRUPLA 0 0 0 CGLASS CMETAL 2 6 5 0 0 0 0 0 0 3 3 1 0 0 0 2 3 CEQUIP 0 0 0 0 0 0 0 0 1 CUTILI 0 0 0 0 0 0 0 0 0 0 1 0 3 0 3 1 0 0 CCONST 1 0 0 0 0 0 0 0 0 0 0 CTRADE 36 9 5 1 3 0 2 2 1 18 6 6 4 1 1 3 7 1 CHOTEL CTRANS 9 8 4 1 1 0 0 1 0 5 2 7 1 1 0 4 2 0 CESTAT 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 CADMIN CPRIVS 0 0 0 0 0 0 0 0 0 0 0 0 CTDTP-E CTDTP-D CTDTP-M FSUB 449 57 59 1 48 109 3 27 57 4 0 2 178 24 LCHILD 1 0 0 0 0 1 1 2 1 3 1 LNONF 12 5 3 6 1 1 3 2 0 1 1 2 0 5 1 LNFPF 8 5 1 4 1 1 2 1 1 1 1 0 29 4 LNFSF 54 47 5 32 5 4 17 13 5 8 5 7 73 6 LSECF 0 0 0 0 0 0 1 0 LNONM 5 3 2 2 1 1 2 1 1 1 4 1 8 2 LNFPM 15 16 2 6 3 1 7 5 5 5 4 9 9 1 24 17 4 LNFSM 16 21 3 7 5 2 2 5 4 6 8 5 12 1 23 2 LSECM 1 2 0 0 0 0 0 0 0 0 1 0 2 0 CAPAG 70 65 10 5 38 9 7 25 18 15 16 13 6 24 2 36 89 10 CAPNAG LAND 30 28 4 2 16 4 3 11 8 6 7 5 3 10 1 15 38 4 ENTR HRBFPL HRFBPL HRNOED HRNFPS HRNFSS HRSECP HUBFPL HUFBPL HUNOED HUNFPS HUNFSS HUSECP GOV DIRTAX IMPTAX EXPTAX