Tanzania: Detailed Assessments Using the Data Quality Assessment Framework (DQAF)
Author: International Monetary Fund
Publication Date:
23 Mar 2004
eISBN:
9781451838350
Language:
English
Keywords:
ISCR; CR; balance of payments; consumer price index; information; debt; commercial bank; business plan; base year; producer price index; Balance of payments statistics; Trade balance; Monetary statistics
Download PDF (822.5 KB)

The data module of this Report on the Observance of Standards and Codes (ROSC) of Tanzania provides a summary assessment of practices on the coverage, periodicity, and timeliness of the data categories against the IMF’s General Data Dissemination System (GDDS). It is complemented by an assessment of the quality of national accounts, consumer and producer price indices, and government finance, monetary, and balance-of-payments statistics, using the Data Quality Assessment Framework (DQAF) developed by the IMF’s Statistics Department (STA).

Abstract

The data module of this Report on the Observance of Standards and Codes (ROSC) of Tanzania provides a summary assessment of practices on the coverage, periodicity, and timeliness of the data categories against the IMF’s General Data Dissemination System (GDDS). It is complemented by an assessment of the quality of national accounts, consumer and producer price indices, and government finance, monetary, and balance-of-payments statistics, using the Data Quality Assessment Framework (DQAF) developed by the IMF’s Statistics Department (STA).

Detailed Assessment Using the Data Quality Assessment Framework (DQAF)

The following detailed information on indicators of statistical practices in the areas of the national accounts, consumer and producer price indices, government finance, monetary, and balance of payments statistics was gathered from publicly available documents and information provided by the Tanzanian officials. This information, which is organized along the lines of the generic DQAF (see Appendix I), was used to prepare the summary assessment of data quality elements, based on a four-part scale of observance, shown in Tanzania’s Report on the Observance on Standards and Codes (ROSC)—Data Module.

I. National Accounts

0. Prerequisites of quality

0.1 Legal and institutional environment

0.1.1 The responsibility for collecting, processing, and disseminating statistics is clearly specified

According to the Statistics Act 2002, second schedule, the National Bureau of Statistics (NBS) may collect statistics on national accounts. Although the schedule does not indicate these statistics as such. However, various items in the schedule such, as “Primary and secondary production,” “Expenditure and consumption,” and “Income, earnings, profits, and interest” taken together would represent the national accounts. In principle, the Statistics Act would not exclude an agency other than the NBS from compiling and publishing the national accounts. However, in practice it is accepted that the NBS is the sole official producer of these statistics. Even if an agency other than the NBS began to compile national accounts, article 15 of the Statistics Act gives the NBS the right to establish statistical standards for those statistics, and article 25 of the Act requires NBS approval prior to publication of these data.

0.1.2 Data sharing and coordination among data producing agencies are adequate

Arrangements and procedures are in place to secure the efficient and timely flow of source data for the national accounts from data producing agencies such as the Ministry of Agriculture and Food Security, to the NBS. There are regular meetings with major data compilers, such as examples include the Bank of Tanzania (BoT), the Tanzania Revenue Authority (TRA), and the Tanzanian Investment Centre (TIC). The contacts with some data compilers, such as the Ministry of Mining, are less regular.

0.1.3 Respondents’ data are to be kept confidential and used for statistical purposes only

Article 18 of the Statistics Act stipulates that no individual information may be disclosed or so arranged that the particulars of any person, business or undertaking are identifiable. There are certain exceptions though, namely when (1) previous consent in writing has been obtained from the respondent, (2) the respondent itself has made the information available for general access, (3) the respondent is the only undertaking or business within a particular sphere of activities,1 or (4) the information is available to the public under any enactment or public document. In addition the following information is exempt: (i) the names of individual undertakings or businesses together with their addresses, industrial classification, and the number of persons engaged, (ii) details of external trade, movement of ships and aircraft, and cargo handled at ports, and (iii) information relating to a local authority or other statutory body. Also, statistical records may be disclosed for bona fide research or statistical purposes provided that all identifying characteristics have been removed and the information is not likely to enable identification of the particular person, undertaking, or business.

Respondents are usually informed of their obligations with regard to the provision of information. For example, the survey form of the Annual Survey of Industrial Production mentions regulations under which the surveyed establishment is required to complete the form, and advises the respondent that returns will be treated as strictly confidential and used for statistical purposes only. Similar information is included in many other questionnaires or in letters accompanying the NBS questionnaires. However, there are exceptions, such as the questionnaire of the Quarterly Survey of Industrial Production.

Article 27 of the Statistics Act specifies that staff or other persons employed by the NBS are not allowed to communicate unpublished information, or to use it for personal gain. Infringement is an offence liable to a fine not exceeding T Sh 1 million (about US$1,000), imprisonment for a term not exceeding 12 months, or both. The Civil Service Department Standing Orders 1994, which are applicable to all civil servants, include similar restrictions on the communication of internal information. Furthermore, the NBS has drafted Staff Regulations that include a section on Discipline and a Code of Conduct that prohibits staff from unauthorized disclosure of nonpublished data.

No strict rules have been developed to ensure that residual disclosure does not occur when aggregations of survey information are disseminated. However, the NBS ensures that survey forms are only made accessible to staff who require the information in the performance of their duties. Such forms are stored in locked cupboards, offices are secured with keys when staff is not present, and the NBS premises are guarded at nights. Computers are password-protected. As well, the destruction of survey forms occurs under strict supervision as required by Article 22 of the Statistics Act.

0.1.4 Statistical reporting is ensured through legal mandate and/or measures to encourage response

Truthful and timely reporting to all NBS questionnaires is obligatory under the terms of Article 27 of the Statistics Act. Noncompliance is liable on conviction to a fine not exceeding T Sh 600,000 (about US$600) or to imprisonment for a term not exceeding six months, or both. However, the NBS does not make use of its right to prosecute and rather seeks to convince nonresponding persons or businesses of the importance of statistics. Outright refusal to respond to questionnaires is exceptional in Tanzania.

Conflicts between the NBS’ legal authority to produce statistics and other provisions have not been noted recently. For instance, obtaining detailed information from the TRA or the National Social Security Fund is not a significant problem for the NBS.

The NBS considers the burden placed on respondents. The reduction of the burden on respondents for providing statistical data is also one of the NBS’ coordinating activities according to Article 5 of the Statistics Act 2002. Assistance given to respondents in completing and submitting forms often take the form of personal contacts by staff of the regional offices. These contacts also serve to seek the cooperation of respondents. Other measures developed to create goodwill include the organization of meetings with enterprises. Such activities and the preparation of information leaflets, contacts with business associations, etc., are planned to be further developed with the creation of a marketing department. Usually, the questionnaires provide a point of contact for respondents.

0.2 Resources

0.2.1 Staff, financial, and computing resources are commensurate with statistical programs

Overall, staff resources fall somewhat short of what would be needed to allow the NBS to undertake extensive analysis or to expand the statistical system, for instance, by developing quarterly accounts and implementing the recommendations of the System of National Accounts 1993 (1993 SNA). For certain demanding operations, staff are borrowed from other departments or temporary workers are hired. The NBS considers the qualifications of its staff to be adequate. Training is largely done on-the-job; there only limited opportunities for receiving more formal training in international guidelines and procedures to compile data. The NBS is successful in retaining a core contingent of trained staff; the staff turnover is very small. Since it has acquired the status of a Government Executive Agency in 1999, the NBS has introduced a salary structure that compares favorably with overall government, and which has helped improve staff morale.

With the help of donors, computing resources are overall at a sufficient level and the software utilized is modern. Most staff have a computer for their use.

Until recently, the financial resources available to the NBS were insufficient in certain areas, resulting in publication delays and gaps in the compilation of essential statistics such as the Survey of Industrial Production that had to be discontinued for several years. Meanwhile, good progress has been made in securing a more solid financial basis for the NBS’ operations, although more resources are still needed in view of the NBS’ development program. At the proposal of the President’s Office—Planning and Privatization (POPP), the government has now guaranteed the resources to compile an extensive set of core statistics. Financing for other statistics has to come from “selling” statistical services to other institutions, among which most notably are foreign donors. Financing outside government has also to be found for several overhead costs, such as the upkeep of buildings and other basic facilities.

0.2.2 Measures to ensure efficient use of resources are implemented

Measures are in place to ensure that resources are used efficiently. The NBS places importance on the development of managerial skills for its staff. Both the progress of the statistical work as well as the financial achievement of the planned activities is monitored. The Treasury also insists on regular reporting on pain of not releasing funds for the next quarter.

Efficiency in data collection and processing are sought by editing data as soon as possible after their collection, preferably already in the field. Extensive use is further made from computer assisted checking on unlikely values. Post enumeration surveys are sometimes used as an additional check.

0.3 Quality awareness

0.3.1 Processes are in place to focus on quality

Managers are sensitive to various dimensions of data quality. In 1999, the NBS conducted an extensive self-assessment of its performance on nine criteria according to the model developed by the European Foundations for Quality Management Excellence Model. Currently, it uses among other things the output based Open Performance Review Appraisal (OPRAS) to appraise staff performance.

The NBS formulates its mission as follows: “To facilitate planning and decision making within the government and the business community, to stimulate research and inform public debate through the provision of relevant, reliable, and timely statistics, and a quality statistical service in general.”

0.3.2 Processes are in place to monitor the quality of the collection, processing, and dissemination of statistics

The NBS may institute, on an ad hoc basis, technical commissions to guide major statistical operations such as new censuses. These commissions, in which donors, users, and providers of data are represented, discuss the structure of the questionnaires, the timing of the data collection, the time at which statistical results should be available, and so forth. The NBS’s business plan is presented in the ministerial Advisory Board. This Board councils the minister responsible for statistics on the adequacy of the Bureau’s plan and performance on functions regarding setting targets and carrying out policy priorities, including:

  • the development and maintenance of a strategic framework,

  • setting the objectives and priorities of the Bureau,

  • the acceptability of the annual reports and accounts, and

  • the evaluation of the Bureau’s performance.

Feedback from users is encouraged in the introductory part of several publications, and by a suggestion box placed at the head office. However, periodic users’ surveys are not conducted.

0.3.3 Processes are in place to deal with quality considerations, including tradeoffs within quality, and to guide planning for existing and emerging needs

The NBS makes decisions taking into account tradeoffs within overall quality. The wish to have certain data available at the time parliament is opened plays an important role, for instance. The reluctance to revise the first estimates of the national accounts obviously has important tradeoffs in the reliability of the estimates.

The introduction of commercial principles in the NBS’s operations has led to a higher awareness of new and emerging data requirements. At the request of the BoT, for instance, the NBS is now considering to improve the timeliness of certain statistical series.

1. Integrity

1.1 Professionalism

1.1.1 Statistics are compiled on an impartial basis

There is no formal provision explicitly addressing the professional independence of the NBS or prohibiting others from interfering in the compilation and/or dissemination of statistical information. However, the status of government executive agency implies distance from the tutelage of other government bodies. The Statistics Act also gives the Director General rights to engage such officers and employees as he may consider necessary or desirable. The NBS does not think its professionalism is being challenged by any outside agency.

Professionalism is among other things promoted by hiring new staff on the basis of transparency of procedures and a competitive interview. Similar principles are followed concerning the promotion of staff. Outside assignments and, to some extent, the presentation of technical papers and seminars given by those returning from overseas courses and conferences are further incentives for developing professional competency.

1.1.2 Choices of sources and statistical techniques are informed solely by statistical considerations

The NBS management considers itself completely free in the choice of the source data and statistical techniques. The reasoning for methodological choices it has made is being collected for future internal reference under the title Methods and Standards.

1.1.3 The appropriate statistical entity is entitled to comment on erroneous interpretation and misuse of statistics

The NBS management believes it is its responsibility to clarify misinterpretation of their data. However, in practice, corrective action is infrequent.

1.2 Transparency

1.2.1 The terms and conditions under which statistics are collected, processed, and disseminated are available to the public

The Statistics Act is available in the NBS library and its website includes a profile of the NBS summarizing its vision and mission, as well as the legal foundations.

1.2.2 Internal governmental access to statistics prior to their release is publicly identified

There is extensive internal government access to statistics prior to their release, which reflects, among other things, the desire to receive comments that could lead to improved estimates. For instance, the NBS will provide the national accounts data to the POPP in March, and they will be made public by the Minister in his mid-June Budget speech. At the occasion of the Budget speech, the POPP also distributes its annual Economic Survey, which contains many data compiled by NBS. The extensive internal government access to unreleased data is identified, but probably not to its full extent, in the IMF Dissemination Standards Bulletin Board (DSBB) website; national publications do not make reference to internal government access.

1.2.3 Products of statistical agencies/units are clearly identified as such

The products of the NBS are clearly identified as such. In the main statistical publications that disseminate NBS data, examples being the Economic Survey of the POPP and the publications (including the website) of the BoT, appropriate notes identify the NBS as source. Although the Bureau does not actively request attribution, nonattribution does not seem an issue in practice.

1.2.4 Advance notice is given of major changes in methodology, source data, and statistical techniques

The NBS has not yet developed the standard procedure of giving advance notice of major changes in methodology, etc., in its statistical series. The main users of the statistics are likely to be aware of such future changes, though, from informal briefings in interagency meetings.

1.3 Ethical standards

1.3.1 Guidelines for staff behavior are in place and are well known to the staff

The Civil Service Department Standing Orders includes a Code of Conduct applicable to all government employees. In addition, the NBS has developed a Code of Ethics that will be attached to the NBS Staff Regulations. Staff is informed about ethical standards during the first week of employment. Issues of proper ethical conduct may further be addressed in the one-day meeting the NBS annually organizes for its entire staff.

2. Methodological soundness

2.1 Concepts and definitions

2.1.1 The overall structure in terms of concepts and definitions follows internationally accepted standards, guidelines, or good practices

The general framework for compiling the Tanzanian national accounts is largely based on 1968 System of National Accounts (SNA), but it also includes certain elements for the 1993 SNA such as the valuation of imports of goods at free-on-board (f.o.b.) prices. The NBS is working toward further incorporation of the 1993 SNA, and according to plans, the new revised time series, which are scheduled to be finalized by 2003/2004 in addition to the introduction of a new base year and benchmarking, will be based on the 1993 SNA.

2.2 Scope

2.2.1 The scope is broadly consistent with internationally accepted standards, guidelines, or good practices

The national accounts statistics for Tanzania cover the tables and accounts that the Intersecretariat Working Group on National Accounts (ISWGNA) has defined as a minimum requirement for the implementation of 1993 SNA, but not yet those whose compilation the ISWGNA has recommended.

The minimum set refers to the following:

  • annual value added and gross domestic product (GDP) at current and constant prices by activity,

  • annual expenditures of GDP at current and constant prices,

  • annual value added components at current prices by activity,

  • sequence of accounts for the total economy (up to financial accounts) with an annual frequency, and

  • annual rest of world accounts (until net lending).

In addition to these core tables and accounts, Tanzania also publishes GDP estimates by region. Sector accounts, except for general government, are not being compiled.

The 1993 SNA tables and accounts whose regular publications the ISWGNA recommends are the following:

  • quarterly value added and GDP at current and constant prices by activity,

  • quarterly expenditures of GDP at current and constant prices,

  • annual supply and use tables.

Tanzania does not yet compile quarterly national accounts, but according to plans they will be introduced in a few years; the target year is not yet set. Annual supply and use tables are not compiled. In Tanzania, input-output tables have been compiled for 1954, 1961, 1969, 1976, and 1992. However, the annual accounts have not been revised to incorporate the results of the 1992 input-output table. Presently, there is a project to develop the capacity of NBS to produce Social Accounting Matrices (SAMs). The plan is to produce a SAM for the new benchmark year 2000. In principle the SAM, while particularly focusing on the household sector, also incorporates supply side data.

The Tanzanian national accounts define residency in accordance with the 1993 SNA and the fifth edition of Balance of Payments Manual. Included are the following in particular:

  • territorial enclaves such as Tanzanian embassies in the rest of the world,

  • bonded warehouses and free trade zone activities, and

  • workers who work only part of the year in another country.

Nonetheless, in practice it is very difficult to distinguish between workers who are only temporarily away from Tanzania, and thus should be treated as Tanzanian residents, and those that remain abroad for more than a year.

Only mainland Tanzania is included in the national accounts.

The production boundaries of the Tanzanian national accounts are not fully in accordance with the 1993 SNA recommendations. The scope of output measurement includes among other things the following items:

  • goods for own final consumption, and

  • output of goods for own account fixed capital formation

In variance with the 1993 SNA, not included items are:

  • research and development on own account,

  • mineral exploration,

  • production of entertainment, literary or artistic originals,

  • production of computer software, and

  • illegal output sold to willing buyers, with an exception for mineral smuggling.

The asset boundary has yet to include newly introduced recommendations of the 1993 SNA. Among the tangible assets the following should be included:

  • Defense-related assets that could be used for civilian purposes. This is in the Tanzanian accounts limited to buildings and road construction.

  • Valuables and historical monuments. These are not included in the Tanzanian accounts.

  • Agricultural work-in-progress. Only the annual increase in livestock is estimated in the accounts.

Among intangible assets the following should be included:

  • Mineral exploration, whether successful or not. This is not included in the Tanzanian accounts.

  • Systems and standard application computer software and databases (purchased or built in-house). In Tanzania there are problems with capturing computer software and databases, either purchased or not. Thus, these are not included as intangible assets.

  • Entertainment, literary or artistic originals. Such items are not included in the Tanzanian accounts.

  • Patented entities. These are not included.

  • Leases and other transferable contracts. These are not included.

There are no routines to keep the deviations from the 1993 SNA under review. A major revision of the national accounts, which among other things will serve to implement the 1993 SNA, is according to the NBS’ plans to be finalized by 2003/2004.

2.3 Classification/sectorization

2.3.1 Classification/sectorization systems used are broadly consistent with internationally accepted standards, guidelines, or good practices

The Tanzanian national accounts do not include the compilation of institutional sector accounts. However, there are plans to introduce institutional sector accounts within a few years.

Economic activities of producer units are classified according to the second revision of the International Standard Industrial Classification of all Economic Activities (ISIC Rev. 2) with national adaptations in the agricultural breakdown. The nine major activities distinguished are: Agriculture, Forestry, fishing, and hunting; Mining and quarrying; Electricity and water; Construction; Wholesale and retail trade, hotels and restaurants; Transport and communications; Finance, insurance, real estate and business services; and Public administration and other services. Imputed bank service charges are treated as intermediate consumption by a nominal industry.

For national accounts purposes, the Harmonized System used in the foreign trade statistics to classify imports and exports of goods is converted to the Standard Industrial Trade Classification (SITC).

Current expenditures by general government are classified by 14 categories of the Classification of Functions of Government (COFOG).

No functional classification, such as the Classification of Individual Consumption by Purpose (COICOP) is used to break down the final consumption expenditures of households. The benchmark estimates from 1992 are, however, estimated at a detailed level using a national Input Output Trade Classification (IOTC).

2.4 Basis for recording

2.4.1 Market prices are used to value flows and stocks.

The valuation rules used for recording flows are partially in accordance with the 1993 SNA. Market output is valued at factor costs instead of at basic prices or producer’s prices. Output for own use is valued at the same prices as market output (factor costs). The valuation of intermediate consumption is purchaser’s prices, including sales tax, excise taxes, and non-deductible value added tax (VAT). The deductible part of VAT, which was introduced in 1998, is in principle excluded from the valuation of final use.

Corrections are made when transfer prices are detected, but this is not very common in actual practice. Information on insurance and freight for merchandise imports is available. BoT total imports and exports are valued on an f.o.b., basis. Transactions in foreign currency are converted using the average exchange rate for a year, which is published by the BoT.

2.4.2 Recording is done on an accrual basis

Most transactions and flows are recorded on an accrual basis, but all government-related and some external transactions are recorded on a cash basis. Except for the annual increase in livestock, no work-in-progress is recorded.

2.4.3 Grossing/netting procedures are broadly consistent with internationally accepted standards, guidelines, or good practices

Transactions between establishments within the same enterprise are recorded on a gross basis to the extent that the data sources allow.

3. Accuracy and reliability

3.1 Source data

3.1.1 Source data are collected from comprehensive data collection programs that take into account country-specific conditions

The NBS maintains a Central Register of Establishments (CRE) that has been constructed on the basis of the census, administrative records on licenses from district councils and municipalities, the Tanzania Revenue Authority (TRA), the National Social Security Fund (NSSF), the register of companies, and the vocational education training center. Agricultural, fishing, and forestry establishments are included in the CRE. The CRE was established in 1992. The present register suffers from not being regularly updated. The problems relate to the register including units that have not been operating for some time, and not including new units established after 1992. The business census that will be conducted in 2003 will be important for the updating of the register. This business census will cover both Tanzania mainland and Zanzibar.

The table below summarizes for each ISIC section the main basic sources for the annual accounts estimates and their coverage. In constructing this table, the results of all current exhaustive and sample surveys that directly measure the activity have been considered as positive contributions to coverage.

Main Sources for the Annual Accounts

article image
article image
article image
article image

The above sources are supplemented with a complete range of agricultural prices, industrial producer price indices, the consumer price index, indirectly measured prices for foreign merchandise trade based on values and quantities, and the Cost of Construction Index.

In the table’s section on sources for the production approach, the lack of sufficient basic data for manufacturing, trade, transport, and real estate stands out. For manufacturing this is due to the large informal sector. It can also be noted that GDP estimates for the mentioned sectors to a large extent are based on extrapolation. A range of indicators, with varying degrees of appropriateness for this purpose, is the basis for these extrapolations.

Source data that are based on sample surveys generally use scientific random sampling techniques, use proper imputation methods to handle nonresponse, and determine grossing-up factors scientifically. However, owing to the inadequate register, the sampling frames may be of suspect quality. Especially in the case of manufacturing there are indications that this leads to a severe underestimation of value added. The Input-Output Table Construction Project (1999) indicated an underestimation of total GDP in 1992 of about 20 percent, nearly half of which can be attributed to manufacturing.

When an annual survey is carried out, normally information is collected on both type of income and expenditure. However, for most of the industries, like agriculture, mining and quarrying, manufacturing, construction, trade, transport and private services, there is no information available on intermediate consumption. In all these cases, fixed value added/output ratios based on the benchmark year are used.

Household budget surveys (HBSs) are conducted on a regular basis, the latest one in 2000/2001, for which a representative sample of 22,178 households was interviewed. The sample design and estimation procedures ensured that the estimates are representative of Both the mainland total and the regional populations. The HBS generates detailed data on purchases of consumption goods, durable goods, and own-account production for consumption, as well as purchases of valuables and own-account fixed capital formation. In the current national accounts, most production and expenditure benchmarks are based on the HBS 1992. For final household consumption expenditure, the benchmarking was conducted at a detailed product level.

Comprehensive government finance data are available; they partially originate from the report by the Accountant General for central government, and partially from the Report of Income and Expenditure of Local Government. The data are provided on a detailed level, and allow the estimation of output, intermediate consumption, fixed capital formation by assets and final consumption expenditure. The central government data include extrabudgetary funds. All defense related expenditures are included in the data for the public sector. Estimates on the capital stock owned by the government are not made.

When annual services do not exist, routine or ad hoc surveys/censuses are conducted. These will be used for benchmarking purposes. Examples are the HBS, which according to plans should be carried out approximately every fifth year, and the Annual Survey of Industrial Production, which is now taking place for 2002. The previous annual Survey of Industrial Production was conducted for the year 1990.

3.1.2 Source data reasonably approximate the definitions, scope, classifications, valuation, and time of recording required

Source data generally are consistent with the definitions, scope, and classifications needed for national accounts purposes. The data normally are consistent with the time of recording, reference period, and the valuation of national accounts estimates. Exceptions are the agricultural statistics and central government finance statistics. For agriculture the data provided relate to the agricultural year, e.g., October to September, while the central government financial statistics follow the fiscal year from July to June. However, in these cases steps are taken to make a better allocation in time. For agriculture production the data are related to the year when the harvesting normally takes place. The central government accounts data for the calendar year are estimated by averaging two subsequent fiscal years. A preferred method would be to compile the central government accounts on the basis of the monthly or quarterly revenue and expenditure data.

3.1.3 Source data are timely

Source data are generally available on a timely basis. The exception is the Producer Price Index (PPI), which used to be timely. It is now rather delayed and the national accounts have to use proxies, which are later on revised. If source data are late, actions are taken, usually by way of informal telephone contact. If necessary, higher-level staff gets involved.

3.2 Statistical techniques

3.2.1 Data compilation employs sound statistical techniques

The approach for measuring GDP is from data on production. Expenditures on GDP are entirely or largely estimated using the commodity flow approach.

Production Approach Procedures

In total, Tanzania’s national accounts include estimates for 28 industries. The output and intermediate consumption estimates for most of the service activities are compiled at a relatively aggregated level, using extrapolating techniques. Agricultural output is compiled at a detailed product level, while formal manufacturing output is extrapolated using information from the Quarterly Survey of Industrial Production at three-digit ISIC level. Intermediate consumption for agriculture is estimated as a fixed proportion of crop output and livestock output respectively. For manufacturing, intermediate consumption is also estimated as a fixed ratio derived from the base year.

Almost a decade has passed since the benchmark year and the economy will have changed substantially. In 2000 the National Accounts Department, under the guidance of the Food and Agricultural Organization (FAO), carried out a revision exercise of the agricultural GDP estimates.2 This exercise indicates that, for instance, for livestock products intermediate consumption as a percentage of output was 24 percent in 1992, and that this had increased to 35 percent in 1999 at current prices. The current national accounts still presume a fixed percentage of 10 percent. The study also indicated that the prices of inputs had increased more than the prices of livestock products. To improve the quality of the national accounts GDP estimate one should, therefore, try to incorporate more updated information on intermediate inputs, and seek to have information on intermediate inputs on a more regular basis than at present.

Tanzania had a broadly comprehensive range of economic statistics available around 1992, the year upon which the present estimates of GDP are based. Since that baseline, GDP estimates are to a large extent compiled by extrapolation. A range of indicators is used to make these extrapolations, the indicators having various degrees of robustness for the purpose. However, the reliance on fixed ratios that are more than five years old when estimating output at current prices from the production side is problematic. The major activities that use these fixed ratios are transport, real estate and the informal manufacturing sector.

Concerning the techniques used to address specific issues of GDP compilation the following is observed:

  • The output of owner-occupied dwellings is based on HBS 1992 data, which included a special question on the value of owner-occupied dwellings. Separate estimates are made for rural and urban dwellings, and both are assumed to move at current prices with the appropriate levels of population growth. The deflator is the rent sub-index of the CPI.

  • The concept of work in progress is limited to recording the increase in livestock reared for purposes of food, as made available by the MWLD. Regarding large construction work, the rule is to record partial output as fixed capital formation in proportions to the payments made.

  • In the national accounts only changes in livestock are included in the inventories.

  • In contrast to the recommendations in the 1993 SNA, the perpetual inventory method is not used as basis for estimating consumption of fixed capital. Consumption of fixed capital is estimated directly based on information about depreciation on capital equipment, either from the accounts or from the surveys conducted for national accounts compilation.

  • No effort is undertaken to convert government receipts and expenditures (such as taxes on products or compensation of employees), which are recorded on a cash basis in the government finance statistics, to an accrual basis.

Regarding procedures for compiling volume measures of GDP the following is noted:

  • The general approach is to extrapolate estimates for a benchmark year by output volume and price indices. The prices used are generally as specific as possible with regard to the variables being deflated/inflated in terms of concept, coverage and reference period. However, in a few cases the correspondence is less close. Examples are government value added, which is deflated by the overall CPI.

  • Single indicator methods are used to compile value added estimates at constant prices. The 1993 SNA recommends double deflation (or double inflation). This is something countries should strive at because the assumption that the prices of intermediate consumption change in the same measure as output may not be valid.

  • In constant prices, customs and taxes on imported products are extrapolated with the constant price value of imports, while sale taxes on local products are extrapolated by the appropriate production index, and the residual taxes are deflated by the CPI. Subsidies at constant prices are inappropriately estimated by deflating current price values by the CPI.

  • Output volume of trade (and value added) is estimated by applying benchmark trade margins on the constant price production levels of monetary agriculture, mining, manufacturing, and relevant components of imports.

  • No chain indices are used to measure the volume change of GDP. Presently, the base year is 1992, but by 2003 or 2004 this will be changed to 2000. The policy is to change base years as soon as new surveys or censuses that give good benchmark estimates are conducted. An updated benchmarking will be important, as the Tanzanian economy has changed substantially since 1992.

Expenditure Approach Procedures

GDP cannot be calculated from the expenditure approach, among other things because there are no independent sources available for estimating household final consumption expenditure.

Gross fixed capital formation has two components: construction, and machinery and equipment. The construction estimates are derived directly from the output of the construction industry, with appropriate downward adjustments for repairs and maintenance. Machinery and equipment totals are derived directly from imports data with appropriate adjustments for taxes and installation costs. Government estimates for machinery and equipment are derived from accounts analysis. Gross fixed capital formation is compiled both by activity and by type of assets.

As indicated earlier, changes in inventories consist only of changes in livestock made available by the MWLD. At present no attempts are being made to investigate whether it would be possible to extend the coverage of changes in inventories with data from agencies likely to hold large inventories (wholesalers, oil companies, etc.).

The main source for the benchmarking of household final consumption expenditure is the HBS 1992. The benchmarking was compiled at detailed product level (Input-Output Trade Classification), and the HBS data were adjusted for intermediate consumption items such as purchases of fertilizer and seeds, as well as for underreporting of beer and cigarettes. After the benchmark year, household final consumption expenditure has been extrapolated as a fixed ratio of GDP at factor costs, both at constant and current prices.

Government final consumption expenditure is derived directly from central and local government accounts. All information on aid-funded expenditure outside the government budget is included. Government final consumption expenditure is classified by 14 categories of the Classification of Functions of Government (COFOG).

Concerning the techniques used to address a number of specific compilation issues of the expenditure approach, the following should be mentioned:

  • Government final expenditure is exclusive of incidental sales.

  • Expenditures of residents abroad are not captured in household final consumption expenditures but are included in imports.

  • Expenditure of nonresidents in the economy are in theory not included in the household final consumption expenditure but are included in exports.

  • Expenditures on items that are considered stores of wealth (such as jewelry, works of art) are not included in the estimates of valuables.

As concerns the volume measures of expenditures on GDP the following is noted:

  • Household final consumption expenditure at constant and current prices is estimated as a fixed ratio of GDP at factor costs, while the consumption at current prices is estimate as a fixed ratio of output at current prices. This means that the implicit deflator of household final consumption expenditure is not consistent with the CPI.

  • Government final consumption expenditure is deflated by the overall CPI, and not as recommended by the cost components of output for final use.

  • The calculation of gross fixed consumption capital at constant prices varies by type of investment goods. Imported goods are extrapolated by import volumes.

  • For constant price estimates, both imported and exported goods are extrapolated by import/export volumes at a rather detailed commodity level.

  • Changes in inventories include only changes in livestock. Correct methods are used to derive volume measurements.

3.2.2 Other statistical procedures (e.g., data adjustments and transformations, and statistical analysis) employ sound statistical techniques

Procedures are to a certain extent developed to adjust data sources to improve the coverage, definitions, classifications, and valuation conforming to international guidelines.

There are some procedures to incorporate unrecorded activities, as e.g., the informal manufacturing sector. The informal sector is rather important in Tanzania and a small-scale annual statistics survey to measure changes in this informal sector activity would be recommendable. Besides this, the national accounts also have procedures for making adjustments related to hunting activities, e.g., for poaching, and to mining and quarrying for mineral smuggling. However, as is the case in most countries, no estimates are made for clearly illegal transactions such as drug trafficking.

For industries where there traditionally are few or no surveys conducted, e.g., services, a special questionnaire used specifically for the compilation of the GDP is used.

3.3 Assessment and validation of source data

3.3.1 Source data—including censuses, sample surveys and administrative records—are routinely assessed, e.g., for coverage, sample error, response error, and nonsampling error; the results of the assessments are monitored and made available to guide planning

For most of the large surveys used in the national accounts compilation, e.g., the HBS 2000/2001, information about sampling errors is available, as well as on over/under coverage, measurement problems, and nonresponse. However, for instance no sampling errors are published with the Annual Survey of Industrial Production or with the Quarterly Manufacturing Survey. When available, the information on both sampling and nonsampling errors is reviewed.

To the mission’s knowledge, the various imputation techniques to deal with imputation methods for nonresponse in surveys are known. It is, however, underlined that nonresponse is a small problem for annual surveys in Tanzania, while it seems to be a growing problem for more frequent surveys.

All data received by the NBS are routinely checked for outliers and other improbable results. The routines will normally detect such problems at an early stage of the processing and data will be verified with the respondents. If the National Account Department finds outliers or other impossible results, these values will be discussed with the data providers. Surveys and censuses are routinely audited to verify the accuracy of the data collection.

Accuracy of administrative budgetary data, international trade, and other secondary sources used to compile the national accounts statistics is routinely assessed by the national accounts specialists.

Before use for national accounts purposes the source data are analyzed for temporal consistency and consistency with related data sources. For instance, data are checked with previous reported data, and reported data on investments in buildings and other constructions are checked with data reported for the construction activity.

Consistency checks on definitions, valuation, reference periods and classifications are, by and large, conducted. Source data are also be analyzed in the context of revisions, as is previously done in a project regarding the revision of the estimates for agriculture. The aim of the project was to assemble, review, analyze, and document causes for discrepancies in agricultural data originating from different sources and to propose how best these could be reconciled.

3.4 Assessment and validation of intermediate data and statistical outputs

3.4.1 Main intermediate data are validated against other information where applicable

The data compiled from main sources used to compile national accounts statistics are normally validated against other primary/secondary sources. The National Accounts Department strives to remove any inconsistencies between different data sources.

3.4.2 Statistical discrepancies in intermediate data are assessed and investigated

Systematic procedures exist to assess routinely the potential discrepancy in intermediate data. One example is the transfers from central to local government, which are compared with transfers received by the local government. If an inconsistency is detected, this will be reported back to the data providers and adjustments will be made to remove the discrepancies.

3.4.3 Statistical discrepancies and other potential indicators of problems in statistical outputs are investigated

No supply and use framework is currently used to investigate discrepancies and make the estimates consistent. However, a project is ongoing to develop the capacity of NBS to produce Social Accounting Matrices (SAMs). The plan is to produce a SAM for the new benchmark year 2000. In principle the SAM, which particularly focuses on the household sector, also incorporates supply side data. Therefore, if the compilation of SAMs continues after the end of the project period, this can be a valuable tool to investigate discrepancies between the supply of goods and services and their use.

The NBS validates its estimates against estimates made by others, for example, estimates for exports and imports from the balance of payments are validated against similar data from the customs.

3.5 Revision studies

3.5.1 Studies and analyses of revisions are carried out routinely and used to inform statistical processes

As revisions from the preliminary to the final results are generally very minor, revision studies are not carried out on a regular basis.

4. Serviceability

4.1 Relevance

4.1.1 The relevance and practical utility of existing statistics in meeting users’ needs are monitored

There are several channels through which the NBS consults users about their concerns. Among others, a formal meeting takes place between different government agencies before the release of agricultural production. In addition the NBS has instituted several technical commissions, or user/producer workshops, in which donors, government agencies, other users, and providers of data are represented.

NBS staff regularly participate in international statistical meetings and seminars organized by international and regional organizations such as Southern African Development Community (SADC).

4.2 Timeliness and periodicity

4.2.1 Timeliness follows dissemination standards

In accordance with the recommendations of the GDDS, the annual accounts are published within 6–9 months after the end of the reference year.

4.2.2 Periodicity follows dissemination standards

The national accounts estimates are disseminated with annual periodicity, as recommended under the GDDS.

4.3 Consistency

4.3.1 Statistics are consistent within the dataset

The annual accounts are internally consistent in the sense that a statistical discrepancy is shown that bridges the gap between GDP by the production approach and GDP by the expenditure approach. The discrepancy has been small; on average about 1.5 percent of GDP during the last six years, with a minimum value of 0.6 percent and a maximum of 2.6 percent for the individual years. However, it should be noted that some of the components of expenditure are estimated on a rather weak basis, in particular by extrapolating with fixed rates from the base year. For example, this is the case for household final consumption expenditure.

The estimates at current prices are consistent with the matching series at constant prices and the deflators.

4.3.2 Statistics are consistent or reconcilable over a reasonable period of time

The national accounts have been compiled since the 1950s, and during the mid-seventies the 1968 SNA was implemented and series of new benchmark estimates were compiled. The latest major revision of the GDP estimates, with the adoption of the base year 1992, has been worked back to 1987. The revised time series were published in 1997, together with a discussion of concepts and definitions, sources, and methods used.

4.3.3 Statistics are consistent or reconcilable with those obtained through other data sources and/or statistical frameworks

The national accounts statistics are reportedly to be consistent with the balance of payments and government finance statistics. However, it is worth noting that the table “International transactions” in the national accounts publication is not completely in line with Tanzania’s Balance of Payments published in the BoT’s Economic bulletin.

4.4 Revision policy and practice

4.4.1 Revisions follow a regular, well-established, and transparent schedule

The revision of provisional estimates is predetermined and stable from year to year, but could be more clearly announced to the general public.

The major revisions of national accounts do not follow a predetermined cycle. The policy is that the major revisions should take place when new large surveys and censuses are available for making new benchmarking estimates. The last major revision was conducted during 19951997, with 1992 as the benchmark/base year. The next revision has started and will, according to the plan, be finalized in 2003/2004. The NBS has not developed the procedure of giving advance notice of major changes in methodology, in its statistical series. However, from informal briefings and interagency meetings the main users of national accounts are likely to be aware of the prospective revision.

To avoid breaks in the time series, major new source data are only incorporated when there is a change of base year.

4.4.2 Preliminary data are clearly identified

Since 1997 and the release of the revised time series, the national accounts publication has given no indication, either in the text nor in the tables, that some data are preliminary and can be subject to revisions. Prior to that, the text of the publication explicitly stated that the last two or three years were subject to revision.

Revised data are disseminated with the same level of detail as the preliminary data.

4.4.3 Studies and analyses of revisions are made public

All major issues that gave rise to revisions of the current GDP estimates are explained in the report Revised National Accounts of Tanzania 1986–1996, with an annex giving the descriptions of revised sources and methods. However, the publication contains no analysis of the revised sources such as was done in earlier publications, for instance the Revised National Accounts of Tanzania 1976–1990.

5. Accessibility

5.1 Data accessibility

5.1.1 Statistics are presented in a way that facilitates proper interpretation and meaningful comparisons (layout and clarity of text, tables, and charts)

The national accounts data are published in the NBS’s National Accounts of Tanzania 1989–2001 in a clear manner, but this publication suffers from not including any analysis of current-period developments, and an improved layout could have facilitated interpretation of the data. Datasets are published with various level of detail, and regional GDP is also published. Several tables contain ratios such as shares in GDP, regional GDP per capita, and average annual growth rates of GDP at constant prices.

The national accounts data are made public by the Minister of Finance in his mid-June Budget speech, when also the annual Economic Survey produced by the POPP is distributed. In this publication the national accounts data are accompanied by an analysis of the most recent developments, and several charts are presented to facilitate the analysis.

5.1.2 Dissemination media and formats are adequate

As noted above, national accounts data are first released in the Minister’s Budget speech and in the Economic Survey following the Budget speech. The NBS publication National Accounts of Tanzania will normally be issued in August. Access to data via the Internet is not yet possible, but the preparations for establishing the website have started.

5.1.3 Statistics are released on a pre-announced schedule

Formally, the schedule for the release of the national accounts data is not pre-announced. However, since the data are made public via the Minister’s Budget speech every year, the public will be aware of the release date.

5.1.4 Statistics are made available to all users at the same time

Simultaneous release is achieved through the issuance of the Budget documents. The press is not briefed in advance.

5.1.5 Nonpublished (but nonconfidential) subaggregates are made available upon request

Unpublished, but nonconfidential, national accounts data can be provided on request for a fee. The availability of additional information is not widely publicized by the NBS.

5.2 Metadata accessibility

5.2.1 Documentation on concepts, scope, classifications, basis of recording, data sources, and statistical techniques is available, and differences from internationally accepted standards, guidelines, or good practices are annotated

Extensive documentation on concepts and methods is published when the base year is changed. The publications are available on request, although they are difficult to find after some years. Other metadata are rather scarce. Information about response rates to main surveys is not given. Neither is any linkage to other major data systems such as the balance of payments or government finance statistics established. For example, the BoT is not referred to as the source for the table “International Transactions” in the publication National accounts of Tanzania.

The GDDS summary methodologies and other related metadata are updated with several months’ delay.

5.2.2 Levels of detail are adapted to the needs of the intended audience

The annual brochure Tanzania in Figures informs the general users about national accounts and other statistical series. A rather comprehensive publication on sources and methods is produced on occasion of the change of base year.

5.3 Assistance to users

5.3.1 Contact person for each subject field is publicized

All official publications of the NBS clearly indicate how the Bureau can be contacted. In the national accounts publication, only the NBS and the National Accounts Department are mentioned. No specifics such as telephone numbers, postal/facsimile/e-mail addresses, contact person for particular types of questions, etc., are given.

Documentation has not been developed to educate users of related datasets. Assistance to users is not monitored through surveys.

5.3.2 Catalogs of publications, documents, and other services, including information on any charges, are widely available

The NBS is setting up a section on the marketing of statistical products. However, currently no Catalogs of publications are available.

Table 1.

Data Quality Assessment Framework: Summary of Results for National Accounts

(Compiling Agency: National Bureau of Statistics)

article image
article image

II. Price Statistics (Consumer Price Index)

0. Prerequisites of quality

0.1 Legal and institutional environment

0.1.1 The responsibility for collecting, processing, and disseminating statistics is clearly specified

According to the Statistics Act 2002, second schedule, the National Bureau of Statistics (NBS) may collect statistics on “Wholesale and retail prices of commodities, rents, and cost of living.” In principle, the Statistics Act would not exclude an agency other than the NBS from compiling and publishing statistics, including a consumer price index (CPI).3 However, in practice it is accepted that the NBS is the sole official producer of the CPI. Even if an agency other than the NBS began to compile such statistics, article 15 of the Statistics Act gives the NBS the right to establish statistical standards for them and article 25 of the Act requires NBS approval prior to publication.

0.1.2 Data sharing and coordination among data producing agencies are adequate

The NBS is the sole responsible agency for the collection of both the weights and the prices that are used to produce the CPI. No procedures have to be established to ensure data sharing and coordination.

0.1.3 Respondents’ data are to be kept confidential and used for statistical purposes only

Article 18 of the Statistics Act stipulates that no individual information may be disclosed or so arranged that the particulars of any person, business or undertaking are identifiable. There are certain exceptions though, namely, when (1) previous consent in writing has been obtained from the respondent, (2) the respondent itself has made the information available for general access, (3) the respondent is the only undertaking or business within a particular sphere of activities,4 or (4) the information is available to the public under any enactment or public document. In addition, the following information is exempt: (i) the names of individual undertakings or businesses together with their addresses, industrial classification, and the number of persons engaged, (ii) details of external trade, movement of ships and aircraft, and cargo handled at ports, and (iii) information relating to a local authority or other statutory body. Also, individual statistical records may be disclosed for bona fide research or statistical purposes provided that all identifying characteristics have been removed and the information is not likely to enable identification of the particular person, undertaking, or business.

Respondents are usually informed of their obligations with regard to the provision of information. For example, the survey form of the Annual Survey of Industrial Production mentions the regulations under which the surveyed establishment is required to complete the form, and advises the respondent that returns will be treated as strictly confidential and used for statistical purposes only. Similar information is included in many other questionnaires or in introductory letters sent with NBS questionnaires. However, there are exceptions, such as the questionnaire of the Quarterly Survey of Industrial Production.

Article 27 of the Statistics Act specifies that staff or other persons employed by the NBS are not allowed to communicate unpublished information, to use it for personal gain. Infringement is an offense liable to a fine not exceeding T Sh 1 million (about US$1,000), imprisonment for a term not exceeding 12 months, or both. The Civil Service Department Standing Orders, 1994 which are applicable to all civil servants, include similar restrictions on the communication of internal information. Furthermore, the NBS has drafted Staff Regulations that include a section Discipline and Code of Conduct that prohibits staff from unauthorized disclose of nonpublished data.

No strict rules have been developed to ensure that residual disclosure does not occur when aggregations of survey information are disseminated. However, the NBS ensures that survey forms are only made accessible to staff who require the information in the performance of their duties. Such forms are stored in locked cupboards, offices are secured with keys when staff is not present, and the NBS premises are guarded at nights. Computers are password-protected. There is permanently a staff person at concierge’s desk at the entrance of the Headquarter building, but entrance formalities are minimal. The destruction of survey forms occurs under strict supervision as required by article 22 of the Statistics Act.

0.1.4 Statistical reporting is ensured through legal mandate and/or measures to encourage response

Truthful and timely reporting to all NBS questionnaires is obligatory by force of Article 27 of the Statistics Act. Noncompliance is liable on conviction to a fine not exceeding T Sh 600,000 (about US$600) or to imprisonment for a term not exceeding six months, or both. However, the NBS does not make use of its right to prosecute and rather seeks to convince nonresponding persons or businesses of the importance of statistics. Outright refusal to respond to questionnaires is exceptional in Tanzania.

Conflicts between the NBS’ legal authority to produce statistics and other provisions have not been noted in recently. For instance, the NBS has no big difficulty in obtaining detailed information from the TRA or the National Social Security Fund.

The NBS does consider the burden that statistical reporting places on the public. The reduction of the burden on respondents for providing statistical data is also one of the NBS’ coordinating activities according to Article 5 of the Statistics Act. Assistance that the NBS gives to respondents often takes the form of help in person provided by staff of the regional offices in completing and submitting forms. These personal contacts also serve to seek cooperation of the respondents. Other measures developed to create goodwill include the organization of meetings with enterprises. Such activities and the preparation of information leaflets, contacts with business associations, etc., are planned to be further developed with the creation of a marketing department within the NBS. Usually, the questionnaires provide a point of contact for respondents.

0.2 Resources

0.2.1 Staff, financial, and computing resources are commensurate with statistical programs

Overall, staff resources fall somewhat short of what would be needed to allow the NBS to undertake extensive analysis or to expand the statistical system. For the current CPI program sufficient staff seems to be available, but it may be used for other projects, which for instance during the last year negatively impacted on the validation of prices forms received from the regional offices. This is despite the fact that for certain big operations, arrangements are in place to borrow staff from other departments or to hire temporary workers. The NBS considers the qualification of its staff to be adequate. Training is largely done on-the-job; there are only limited opportunities for receiving more formal training on international guidelines and best practices for compiling data. The NBS is successful in retaining a core contingent of trained staff—the staff turnover is very small. Since it has acquired the status of a Government Executive Agency in 1999, the NBS has introduced a salary structure that compares favorably with overall government, and which has helped to improve staff morale.

With the help of donors, computing resources are overall at a sufficient level and the software utilized is modern. Most staff have a computer at their disposal.

Until recently, the financial resources available to the NBS were not sufficient in certain areas, resulting in publication delays and gaps in the compilation of essential statistics such as the Survey of Industrial Production that had to be discontinued for several years. Meanwhile, good progress has been made in securing a more solid financial basis for the NBS’ operations, although improvements are still possible. At the proposal of the President’s Office—Planning and Privatization (POPP), the government has now guaranteed the resources to compile an extensive set of core statistics. Financing for other statistics have to come from “selling” statistical services to other institutions, among which most notably are foreign donors. Financing outside government has also to be found for several overhead costs such as the upkeep of buildings and other basic facilities.

0.2.2 Measures to ensure efficient use of resources are implemented

Measures are in place to ensure that resources are used efficiently. The NBS performs its functions in accordance with commercial principles and has to ensure that its revenues are sufficient to meet its outgoings. In that framework, it develops strategic plans for periods of three years ahead. The NBS places importance on the development of managerial skills for its staff. Both the progress of the statistical work as well as the achievements of the planned activities are monitored. The Treasury also insists on regular reporting, on the condition of not releasing funds for the next quarter of the year.

Efficiency in data collection and processing are sought by editing data as soon as possible after their collection, preferably already in the field. Extensive use is further made from computer assisted checking on unlikely values. Post-enumeration surveys are sometimes used as an additional check.

0.3 Quality awareness

0.3.1 Processes are in place to focus on quality

Managers are sensitive to various dimensions of data quality. The NBS formulates its mission as follows: “To facilitate planning and decision making within the government and the business community, to stimulate research and inform public debate through the provision of relevant, reliable, and timely statistics, and a quality statistical service in general.”

In 1999, the NBS conducted an extensive self-assessment of its performance on nine criteria according to a model developed by the European Foundations for Quality Management Excellence Model. Currently, it uses among other things the output-based Open Performance Review Appraisal (OPRAS) to appraise staff performance.

0.3.2 Processes are in place to monitor the quality of the collection, processing, and dissemination of statistics

The NBS may institute, on an ad hoc basis, technical commissions to guide major statistical operations such as new censuses. These commissions, in which donors, users, and providers of data are represented, discuss the structure of the questionnaires, the timing of the data collection, the time at which statistical results should be available, and so forth. The NBS’s business plan is presented in the ministerial Advisory Board. This Board councils the minister responsible for statistics on the adequacy of the Bureau’s plan and performance on functions regarding setting targets and carrying out policy priorities, including the following:

  • the development and maintenance of a strategic framework,

  • setting the objectives and priorities of the Bureau,

  • the acceptability of the annual reports and accounts, and

  • the evaluation of the Bureau’s performance.

Feedback from users is encouraged in the introductory part of several publications and by a suggestion box placed at the head office. However, periodic users’ surveys are not conducted.

0.3.3 Processes are in place to deal with quality considerations, including tradeoffs within quality, and to guide planning for existing and emerging needs

The NBS makes decisions taking into account tradeoffs within overall quality. The wish to have certain data available at the time parliament is opened plays an important role, for instance. The introduction of commercial principles in the NBS’s operations has led to larger awareness of new and emerging data requirements. At the request of the BoT, for instance, the NBS is now considering to improve the timeliness of certain statistical series.

1. Integrity

1.1 Professionalism

1.1.1 Statistics are compiled on an impartial basis

There is no formal provision explicitly addressing the professional independence of the NBS or prohibiting others from interfering in the compilation and/or dissemination of statistical information. However, the status of government executive agency implies distance from the tutelage of other government bodies. The Statistics Act also gives the Director General the power to engage such officers and employees as he may consider necessary or desirable. The NBS does not feel that its professionalism is unduly being challenged by any outside agency.

Professionalism is among other things promoted by hiring new staff on the principles of transparency of procedures and competitiveness. Similar principles are followed concerning the promotion of staff. Outside assignments and, to some extent, the presentation of technical papers and seminars given by those returning from overseas courses and conferences are further incentives for improving professional competence.

1.1.2 Choices of sources and statistical techniques are informed solely by statistical considerations

The NBS management considers itself completely free in the choice of the source data and statistical techniques. The reasoning for methodological choices it has made is currently being collected for future reference under the title Methods and Standards.

1.1.3 The appropriate statistical entity is entitled to comment on erroneous interpretation and misuse of statistics

The NBS management sees it as a task to clarify misinterpretation of their data. However, in actual practice corrective action is infrequent.

1.2 Transparency

1.2.1 The terms and conditions under which statistics are collected, processed, and disseminated are available to the public

The Statistics Act is available in the NBS library, and its website includes a profile of the NBS summarizing its vision and mission, as well as the legal foundations.

1.2.2 Internal governmental access to statistics prior to their release is publicly identified

There is extensive internal government access to statistics prior to their release which reflects the desire to receive comments that could lead to improved estimates, as well as the need to accommodate urgent political needs. This access is identified, but probably not to its full extent, in the IMF DSBB website; national publications do not make reference to internal government access.

1.2.3 Products of statistical agencies/units are clearly identified as such

The own products of the NBS are clearly identified as such. In the main outside statistical publications that disseminate NBS data, that is, the Economic Survey of the POPP and the publications (including the website) of the BoT, appropriate notes identify the NBS as source. Although the Bureau does not actively request attribution of the data it compiles, nonattribution does not seem an issue in practice.

1.2.4 Advance notice is given of major changes in methodology, source data, and statistical techniques

The NBS has not yet developed the standard procedure of giving advance notice of major changes in methodology, etc., in its statistical series. The main users of the statistics are likely to be aware of such future changes, though, from informal briefings on occasion of interagency meetings.

1.3 Ethical standards

1.3.1 Guidelines for staff behavior are in place and are well known to the staff

The Civil Service Standing Orders, 1994 includes a Code of Conduct applicable to all government employees. In addition, the NBS has developed a Code of Ethics that will be attached to the NBS Staff Regulations. Staff is informed about ethical standards during the first week of employment.

Issues of proper ethical conduct may further be addressed in the one-day meeting the NBS annually organizes for its entire staff.

2. Methodological soundness

2.1 Concepts and definitions

2.1.1 The overall structure in terms of concepts and definitions follows internationally accepted standards, guidelines, or good practices

The Tanzanian CPI measures the changing cost of a fixed basket of goods and services acquired by the private consumer. The concepts and definitions of the Tanzanian CPI are in broad accordance with the System of National Accounts 1993 (1993 SNA) regarding the weights and valuation of household final consumption expenditures.

2.2 Scope

2.2.1 The scope is broadly consistent with internationally accepted standards, guidelines, or good practices

The CPI covers final consumption expenditure by individual households in twenty urban centers on the Tanzania mainland. Households of all size classes, income levels and occupations are represented. There are two deviations from the concept of households as it is defined in the 1993 SNA. First, excluded are people living in institutions, such as prisoners serving long sentences and elderly living permanently in retirement homes. This exception is in accordance with accepted good practice for CPI compilation. Second, the CPI does not cover households that live outside the urban centers nor are residents of Zanzibar. The partial geographical coverage is not in accordance with the international standards.

The CPI basket of goods and services refers to all final consumption expenditure by households as defined in the 1993 SNA. It includes own-account production of agricultural products, but excludes the consumption of other goods produced on own account. In accordance with international guidelines, expenditure of households for business purposes is excluded. In contrast to the recommendations of the 1993 SNA, but in line with the recommendations of the International Labor Organization (ILO) at the time the CPI basket was chosen, furthermore the following types of expenditure are excluded:

  • services of owner-occupied dwellings,5

  • illegal goods and services, gambling,

  • life insurance, pension funds, and financial intermediation services indirectly measured,

  • second-hand purchases,

  • expenditures of residents abroad.

2.3 Classification/sectorization

2.3.1 Classification/sectorization systems used are broadly consistent with internationally accepted standards, guidelines, or good practices

The grouping of goods and services in the CPI is consistent with the classification of Individual Consumption by Purpose (COICOP) as given in the 1993 SNA. No specific classification, such as the CPC6 or the CPA,7 is applied to categorize goods and services by type.

2.4 Basis for recording

2.4.1 Market prices are used to value flows and stocks.

In accordance with the recommendations of the 1993 SNA, the CPI measures the prices households actually have to pay in order to acquire consumption goods and services, including VAT and other net taxes on the products. In line with ILO recommendations, sales prices are not taken into account.

2.4.2 Recording is done on an accrual basis

In the base year, own-account consumption was recorded as of the times the household used the products up rather than the times that the acquisition was made. This principle is not in accordance with the 1993 SNA because it diverges from the transaction principle.8 However, this divergence has no negative consequences on the robustness of the current CPI.

As regards the recurrent price observation, the NBS always records prices of goods and services in the reference month that purchases are made. However, the price quotes refer to the first half of the reference month only, as by the 15th of the month the price collection forms have to be sent off to headquarters. Moreover, the quotes do not necessarily always refer to the same day of the month: the price collectors have the freedom to collect the prices at their own convenience. This introduces a “noise” element in the basic data that negatively impacts on the accuracy of the index.

For some goods, mainly fresh food items, the prices tend to fluctuate considerably from day to day (or even from hour to hour during the day). It is good practice to collect several quotes for such goods, spread out over the recording period. In Tanzania, quotes for these items are indeed collected during three different days. However, these days are consecutive rather than distributed evenly over the month, which takes much away from the advantage of having price observations at different points of time.

The concentration of all price observations in the first half of the month has its cause in the slow mail transmittal of the prices forms from the regional offices to headquarters. If the CPI is to be ready two weeks after the reference month, the price collection has to stop early. A considerable gain in relevance could be made by electronic data exchange between the NBS offices.

2.4.3 Grossing/netting procedures are broadly consistent with internationally accepted standards, guidelines, or good practices

Issues related grossing/netting regarding the CPI mostly refer to sales of existing goods such as used cars that, according to the 1993 SNA, should be netted out against purchases. The Tanzanian CPI does not take second-hand items into account, and acts in that respect in conformity with the recommendations of the ILO.

3. Accuracy and reliability

3.1 Source data

3.1.1 Source data are collected from comprehensive data collection programs that take into account country-specific conditions

The main source for the CPI weights are data collected through the comprehensive Household Budget Survey (HBS) that was conducted during December 1991 through November 1992. The HBS made use of the National Master Sample that the NBS had developed for all household survey sampling. The National Master Sample frame was initially based on the 1978 Population Census and later updated with information from the 1988 Population Census.

No business register has been used to select the outlet sample for CPI price collection. Here, the NBS relied on the judgment of regional office staff. The outlets are fixed, but frequent replacement is necessary as outlets often change the types goods sold and/or their type of business. Replacement does not give rise to investigations of possible quality differences.

Up till now, the NBS has conducted HBSs every ten years. Good practices in CPI collection indicate that the time between two consecutive household expenditure statistics should not be more than five years. The current CPI basket, which is derived from the 1991/1992 HBS, should therefore be considered too old. The NBS intends to conduct future HBSs on a quinquennial basis.

The 1991/92 HBS covered the whole mainland of Tanzania and, in accordance with good practice, collected household expenditure data over the period of a full year. The sample survey was geographically stratified at the most restricted level distinguished in the National Master Sample, that is, for 292 enumeration areas. In the selected areas, a list was first made of all households. Subsequently a selection of households was made stratified by three income levels (estimated by the assets they possess). The final sample size was 4,823 households, with an approximately equal split between urban areas and rural areas. To ensure the use of correct scientific sampling techniques, the NBS sought advice from the University of Dar es Salaam. Nonresponse and other losses were handled by replacement by households with similar characteristics from a reserve list. The collected data were sufficiently detailed and included goods and services purchased on markets as well as food items produced on own-account.

In the past, the results of the 1991/92 HBS have met with critique because, even after correction, the food component in the CPI basket (71.2 percent) was regarded as very high, and for instance, the weights for the clothing and footwear component (3.7 percent) or rented dwellings (3.9 percent) very low. For comparison, in the 1969 HBS, food represented 47.0 percent of the total; in the 1976/77 HBS this was 64.2 percent. The results of the 2001/2002 HBS, which had a large sample size of 22,178 households, indicate that food expenditures make up 65.4 percent of total consumption. The most recent data therefore seem to broadly support the weights chosen for the current CPI.

The price collection is conducted by the staff of NBS regional offices in twenty urban centers. The periodicity is monthly, which is in line with the frequency of the CPI. Market prices for fruits and vegetables that are sold in heaps are collected from central markets by actual purchases during three consecutive days. The purchased items are weighted in the offices in order to determine the unit price. For other items pricing quotes are collected through personal visits from up to five outlets, depending on the number of outlets available in the regional centre. Prices for rented dwellings are recorded for three categories of dwellings. However, no particular dwellings have been selected and NBS staff record rental prices according to general knowledge. The rates for electricity, water, postage, and telephone are taken from information provided by the producing entities.

The ILO recommends that all goods and services that have a weight of more than 0.01 percent of total expenditures are represented in the CPI. In 1995, the NBS selected 212 items for which the prices would be collected. This number should be considered sufficient. Mostly, items were retained that already existed in the previous 1977-based index, but also a number of new items were added. The item with the largest weight in the total refers to rice (4.525 percent); the item with the smallest weight is women’s blouses (0.007 percent).

To standardize the prices among regional offices, units of measurement by item were defined (1 kilogram, 1 meter, one package of 20, one room, and so forth). However, in most cases further specifications were not given and the price collection forms do not include a column where the collectors themselves can write the detailed characteristics of the items they price every month. As, for instance, the item “biscuits” is not further defined, undue price variations may be recorded when different types of biscuits are priced from one period to the other. This problem has been known for a long time, but there are no indications that measures have been taken to address it.

No ad hoc surveys supplement the regular price survey program. New products are only incorporated in the CPI on occasion of a change of base year. The NBS does not monitor the press or research papers for information on consumer prices.

3.1.2 Source data reasonably approximate the definitions, scope, classifications, valuation, and time of recording required

The notes that were made above in connection with the CPI are also true for the main source data. The main problems are that rural areas and Zanzibar are not covered and that the timing of the price observation is not precise enough. Otherwise, the Tanzania HBSs and regular price surveys reasonably approximate the definitions, scope, classifications, and valuation that are needed to compile the CPI.

The authorities indicate that no special procedures have been developed to adjust the main source data to conform to CPI requirements. For example, no corrections on the 1991/92 HBS were made for probably underreported expenditure categories such as tobacco and alcoholic beverages (beer).

3.1.3 Source data are timely

The expenditure estimates derived from the HBS are available with a lag of one year, which is adequate for CPI purposes. The price collection program usually is timely enough for the compilation of the CPI. However, the observations relate to only the first half of the month over which it is reported. After the 15th of the reporting month, the data collected by each of the regional offices have to be sent to the NBS’s head office for editing and data entry. Usually not more than one urban centre out of twenty is so late with its reporting to the NBS headquarters that the data cannot be incorporated in the index calculation.

3.2 Statistical techniques

3.2.1 Data compilation employs sound statistical techniques

The commodity detail in the base statistics is sufficient for the analysis of price movements. The CPI is published broken down by 12 main COICOP groups. CPIs for the various regions are not compiled (except for Dar es Salaam).

The weights of the CPI closely approximate household consumption expenditure as defined in the 1993 SNA with the exceptions mentioned under section 2.2.

The CPI is computed according to the modified Laspeyres’ index. The lowest short level indices for the CPI are determined by comparing the simple average of the prices observed for each basket item in each urban center with those of the previous month. The level of prices varies considerably across the country. The long-term price relatives form a link to the reference period of December 1994=100. The weights of the index were normalized to December 1994 values to assure that the expenditure weights and the base price period properly correspond. In the aggregation process, the data from the various urban centers are weighted by the average population in these centers.

All calculations are done in the NBS headquarter office. The average urban center prices are calculated by hand as part of the editing process. Imputations that are needed with respect to out of season items are also made by hand. The remainder of the calculations is done by a LOTUS software program provided by the IMF.

Even though the weights of the CPI are derived from the 1991/92 HBS, many decisions concerning the index were only made in 1995. This made it impossible to start the series in 1992. On the basis of available price information, the new series with December 1994=100 have been calculated back until December 1992. The present CPI has not been linked to the former index using splicing methods to form historical series.

Replacement items are used for items that have temporarily disappeared from the market. Such replacement does not give rise to adjustment for probable quality differences. If no replacement item can be found, the weight of the permanently disappeared item is distributed over the remaining items. If prices for a seasonal item are not available in one urban center, its price is imputed by the price changes of that item in other regions. If a seasonal item is not available in any urban center, its price is carried forward until it reappears on the market.

Several compilation practices described above are in accordance with sound practices. This is not true, however, for using regional population as weights in calculating national average prices. SNA. In addition, the current weight and price reference periods of the CPI have become obsolete. This is made worse by not introducing new products, such as cell phones, in the CPI as they gain market share. At the introduction of series with new weights, it is furthermore desirable that new and old series are spliced to form longer historical series. Finally, the unchanged carrying forward of prices in the case where no observation can be made is not a best practice, although it may be acceptable for relatively short periods in countries with low inflation. Preferably, the missing prices are imputed by comparison with the observed prices for similar items.

3.2.2 Other statistical procedures (e.g., data adjustments and transformations, and statistical analysis) employ sound statistical techniques

The proportion that is not covered in regular CPI compilation is reasonably small. With respect to the missing data concerning Zanzibar, separate indices are available compiled by the Zanzibar Department of Statistics. For another large category not included in the CPI—the imputed services of owner-occupied dwellings—very crude estimates are made in the framework of the national accounts compilation.

The section responsible for the CPI does not keep adequate documentation of the methods used to compile the index. For instance, the available descriptions do not contain details on the choices made regarding the basket items. The lack of sufficient institutional memory makes the section vulnerable when experienced staff leaves to work elsewhere.

3.3 Assessment and validation of source data

3.3.1 Source data—including censuses, sample surveys and administrative records—are routinely assessed, e.g., for coverage, sample error, response error, and nonsampling error; the results of the assessments are monitored and made available to guide planning

For the 1991/92 HBS, nonresponse is not an issue as substitute households are used. Sampling errors for the survey were acceptable and made available. No post-enumeration surveys were held to help eliminate reporting errors. For the price survey, sampling errors are not available due to the judgmental nature of the selection of outlets and representative product items.

Although the heads of the regional NBS centers should check the price collectors’ forms before these are sent to Dar es Salaam, in practice no verification takes place. Headquarter staff validates all forms received by checking for obvious mistakes and the price change in comparison with the data for the previous month. In case of unlikely quotes, telephone contact is taken up with the regional offices. Headquarter staff is also charged with making imputations for missing prices, where appropriate.

The persistent problem caused by the fabrication of price quotes by the regional collectors has been noted in the past. The authorities indicated that this problem remains. To address the issue, it was decided that headquarter staff should travel four times a year to the regional centers to verify in person the reported prices (for 19 urban centers outside Dar es Salaam, this would amount to 76 visits). In practice however, also given the work pressure on headquarter staff related to the extensive census and survey program, hardly any such verifications have taken place. The same exceptional situation has caused form validation at Headquarters to be brought down to a minimum during the last year.

3.4 Assessment and validation of intermediate data and statistical outputs

3.4.1 Main intermediate data are validated against other information where applicable

The CPI is not compared with other major price indices such as the PPI; the differences in definition and scope is considered too large to make such an exercise worthwhile. Price indices for foreign trade are not available.

3.4.2 Statistical discrepancies in intermediate data are assessed and investigated

After data entry, the basic CPI data are checked for possible mistakes.

3.4.3 Statistical discrepancies and other potential indicators of problems in statistical outputs are investigated

Before the data are released, again verification takes place to ensure that no mistakes entered the index results because of failures in the automatic data processing. Statistical discrepancies are not part of the compilation process.

3.5 Revision studies

3.5.1 Studies and analyses of revisions are carried out routinely and used to inform statistical processes

An extensive analysis is made whenever a new HBS is prepared. The results of this analysis are published.

Revisions to the monthly price data are extremely rare. Prices reported by the collectors may be revised during the editing process, and this may lead to identification of certain types of mistakes. The imputations made for regions that send in their data too late for incorporation in the published index might generate information to improve the imputation process. However, in practice, standard fixed procedures are used to make such imputations.

4. Serviceability

4.1 Relevance

4.1.1 The relevance and practical utility of existing statistics in meeting users’ needs are monitored

The NBS has not instituted a standing committee for advice on the quality and further development of the CPI, and to review whether the program meets the needs of users of statistics. However, ad hoc technical committees are formed on occasion of major statistical undertakings, such as the organization of a new HBS. Members include representatives of the stake holding ministries, the BoT, academic and research institutes, and international organizations and other donors as appropriate. Users’ surveys are not part of the NBS consultation program.

From time to time, the NBS participates in meetings organized by international and regional organizations where issues related to the CPI are discussed. In the last year, it was represented in three such meetings.

4.2 Timeliness and periodicity

4.2.1 Timeliness follows dissemination standards

The CPI is released to a small group of users on the 15th day of the month after the reference period.9 In October 2002, the NBS resumed its official publication of the CPI with a timeliness of 25 days after the reference month, thus meeting SDDS requirements. Since the Statistical Abstract over the year 1994, the NBS had not published the CPI itself and for a long period the BoT Monthly Economic Review, quarterly Economic Bulletin, and website were the official channels for dissemination of the CPI, the timeliness varying between about one to five months.

4.2.2 Periodicity follows dissemination standards

Since January 1995, the CPI has been compiled on a monthly basis.

4.3 Consistency

4.3.1 Statistics are consistent within the dataset

The CPI is internally consistent. The current computer program ensures that the aggregation of regional indices, if they were compiled, gives the same total as the aggregation over items.

4.3.2 Statistics are consistent or reconcilable over a reasonable period of time

Consistent monthly CPI data are available for an adequate period of time. Methodological notes that explain the main breaks in the series are available on request.

4.3.3 Statistics are consistent or reconcilable with those obtained through other data sources and/or statistical frameworks

The CPI is not used for deflating household final consumption expenditure in the national accounts; any correspondence between the CPI and the implicit deflator for household consumption in the national accounts is coincidental.

4.4 Revision policy and practice

4.4.1 Revisions follow a regular, well-established, and transparent schedule

The HBS follows a regular ten-year schedule; this is well known among the main users. The basis for the CPI is likely to change after the results of a new HBS have become available, but there is no set schedule. The CPI itself is not revised.

4.4.2 Preliminary data are clearly identified

Not applicable, as there are no preliminary data.

4.4.3 Studies and analyses of revisions are made public

New HBS data are extensively analyzed and the results of these studies are published. A fairly extensive note dated January 2001 describes the sources and methods for the CPI including the revisions made. This note is available to the public. On the occasion of the revision of the base year, a workshop for users was organized.

5. Accessibility

5.1 Data accessibility

5.1.1 Statistics are presented in a way that facilitates proper interpretation and meaningful comparisons (layout and clarity of text, tables, and charts)

The main dissemination channel for the CPI to main users is a two-page leaflet closely resembling a press release. This release contains a short discussion of the main changes, a table with the indices for two subsequent months, as well as the month in the previous year that corresponds with the most recent data for the current year. Percentage changes are provided month-to-month and month-to-corresponding month in the previous year periods. The table also includes the weights of the 12 main item groups. Charts facilitate the analysis and a short definition of the CPI is given. On the whole, the presentation facilitates proper interpretation of the CPI. Time series information, however, is not contained in the leaflet. Unfortunately, the leaflet is not made available to the general public and it is even not available for perusal in the NBS library.

The tables of the CPI in the BoT publications present the data in a clear way, but do not include any texts or charts to facilitate their interpretation. On the other hand, the BoT also publishes seasonally adjusted data, that it calculates on the basis of NBS data.

5.1.2 Dissemination media and formats are adequate

The dissemination media for the CPI are not fully adequate. The first information release is not distributed to the press. The data are neither available through an electronic data base managed by the NBS nor on diskettes, and the last comprehensive publication dedicated to the CPI (which contained many details) dates back to June 1995. However, the BoT publishes the CPI both in hard copy and on its website. CPI series are also published in the annual Economic Survey of the POPP.

5.1.3 Statistics are released on a pre-announced schedule

The statistics are always released to select users on the 15th of the month, or the first working day thereafter if the 15th falls in the weekend. This schedule is well-known to the main users. The general public has to rely on publication by the BoT, which does not adhere to a pre-announced schedule.

5.1.4 Statistics are made available to all users at the same time

The CPI is not made available to all users at the same time: the important users receive data that the general public only is able access at a later point in time, and then only indirectly via the BoT releases.

5.1.5 Nonpublished (but nonconfidential) subaggregates are made available upon request

The authorities indicated that nonpublished subaggregates of the CPI are available on request. However, this option is not publicized.

5.2 Metadata accessibility

5.2.1 Documentation on concepts, scope, classifications, basis of recording, data sources, and statistical techniques is available, and differences from internationally accepted standards, guidelines, or good practices are annotated

The methodological description of the CPI that is widely accessible to the general public is the GDDS metadata posted on the DSBB. The authorities update this description, albeit with several months delay. The foreseen hyperlink from the NBS website to the IMF site is not yet established.

The note Sources and Methods for Computation of the National Consumer Price Index in Tanzania is available on request, but this is not publicized.

5.2.2 Levels of detail are adapted to the needs of the intended audience

There are no publications on the CPI specifically directed at general users only. It should be noted that the annual brochure Tanzania in Figures, which is likely to reach the largest non-specialist public, does not contain CPI data.

5.3 Assistance to users

5.3.1 Contact person for each subject field is publicized

All official publications of the NBS clearly indicate how the Bureau can be contacted. Many users with queries prefer to visit the NBS in person. Prompt and knowledgeable support is available to users, but the releases do not indicate specific individuals that can help users on questions regarding the CPI.

The assistance to users is not monitored through surveys.

5.3.2 Catalogs of publications, documents, and other services, including information on any charges, are widely available

The NBS is setting up a section on the marketing of statistical products. However, currently no atalogs of publications are available.

Table 2.

Data Quality Assessment Framework—Summary of Results for Price Statistics (Consumer Price Index)

(Compiling Agency: National Bureau of Statistics)

article image
article image

III. Price Statistics (Producer Price Index)

0. Prerequisites of quality

0.1 Legal and institutional environment

0.1.1 The responsibility for collecting, processing, and disseminating statistics is clearly specified

According to the Statistics Act 2002 (Statistics Act), second schedule, the National Bureau of Statistics (NBS) may collect “Wholesale and retail prices of commodities, rents, and cost of living.” The NBS does not compile a wholesale price index but rather a producer price index (PPI). It is generally accepted that, for most analytical purposes, a PPI is more useful than a wholesale price index.

In principle, the Statistics Act would not exclude that an agency other than the NBS would compile and publish such statistics. However, in practice it is accepted that the NBS is the sole official producer of these statistics. Even if an agency other than the NBS began to compile such statistics, Article 15 of the Statistics Act gives the NBS the right to establish statistical standards for them and Article 25 of the Act requires NBS approval prior to publication.

0.1.2 Data sharing and coordination among data producing agencies are adequate

The NBS is the sole responsible agency for the collection of both the weights and the prices that are used to produce the PPI. No procedures have to be established to ensure data sharing and coordination.

0.1.3 Respondents’ data are to be kept confidential and used for statistical purposes only

Article 18 of the Statistics Act stipulates that no individual information may be disclosed or so arranged that the particulars of any person, business or undertaking are identifiable. There are certain exceptions though, namely, when (1) previous consent in writing was obtained from the respondent, (2) the respondent itself has made the information made available for general access, (3) the respondent is the only undertaking or business within a particular sphere of activities,10 or (4) the information is available to the public under any enactment or public document. In addition, the following information is exempt: (i) the names of individual undertakings or businesses together with their addresses, industrial classification, and the number of persons engaged, (ii) details of external trade, movement of ships and aircraft, and cargo handled at ports, (iii) information relating to a local authority or other statutory body. In addition, individual statistical records may be disclosed for bona fide research or statistical purposes provided that all identifying characteristics have been removed and the information is not likely to enable identification of the particular person, undertaking, or business.

Respondents are usually informed of their obligations with regard to the provision of information. For example, the survey form of the Annual Survey of Industrial Production mentions the regulations under which the surveyed establishment is required to complete the form, and advises the respondent that returns will be treated as strictly confidential and used for statistical purposes only. Similar information is included in many other questionnaires or in introductory letters sent with NBS questionnaires. However, there are exceptions, such as the questionnaire of the Quarterly Survey of Industrial Production.

Under Article 27 of the Statistics Act, staff or other persons employed by the NBS are not allowed to communicate unpublished information or to use it for personal gain. Infringement is an offence liable to a fine not exceeding T Sh 1 million (about US$1,000), imprisonment for a term not exceeding 12 months, or both. Civil Service Department Standing Orders, 1994 which are applicable to all civil servants, include similar restrictions on the communication of internal information. Furthermore, the NBS has drafted Staff Regulations that include a section on Discipline and a Code of Conduct prohibiting staff from unauthorized disclosure of nonpublished data.

No strict rules have been developed to ensure that residual disclosure does not occur when aggregations of survey information are disseminated. However, the NBS ensures that survey forms are only made accessible to staff who require the information in the performance of their duties. Such forms are stored in locked cupboards, offices are secured with keys when staff is not present, and the NBS premises are guarded at nights. Computers are password-protected. There is permanently a staff person at concierge’s desk at the entrance of the Headquarter building, but entrance formalities are minimal. The destruction of survey forms occurs under strict supervision as required by Article 22 of the Statistics Act.

0.1.4 Statistical reporting is ensured through legal mandate and/or measures to encourage response

Truthful and timely reporting to all NBS questionnaires is made obligatory by force of Article 27 of the Statistics Act. Noncompliance is liable on conviction to a fine not exceeding T Sh 600,000 (about US$600) or to imprisonment for a term not exceeding six months, or both. However, the NBS does not make use of its right to prosecute and rather seeks to convince nonresponding persons or businesses of the importance of statistics. Outright refusal to respond to questionnaires is exceptional in Tanzania.

Conflicts between the NBS’ legal authority to produce statistics and other provisions have been not been noted in recently. For instance, the NBS has no big difficulty in obtaining detailed information from the TRA or the National Social Security Fund.

The NBS does consider the burden that statistical reporting places on the public. The reduction of the burden on respondents for providing statistical data is also one of the NBS’ coordinating activities according to Article 5 of the Statistics Act. Assistance that the NBS gives to respondents often takes the form of help in person provided by staff of the regional offices in completing and submitting forms. These personal contacts also serve to seek cooperation of the respondents. Other measures developed to create goodwill include the organization of meetings with enterprises. Such activities and the preparation of information leaflets, contacts with business associations, etc., are planned to be further developed with the creation of a marketing department within the NBS. Usually, the questionnaires provide a point of contact for respondents.

0.2 Resources

0.2.1 Staff, financial, and computing resources are commensurate with statistical programs

Overall, staff resources fall somewhat short of what would be needed to allow the NBS to undertake extensive analysis or to expand the statistical system. For the current PPI program sufficient staff is available. For certain big operations, arrangements are in place to borrow staff from other departments or to hire temporary workers. The NBS considers the qualification of its staff to be adequate. Training is largely done on-the-job; there are only limited opportunities for receiving more formal training on international guidelines and best practices for compiling data. The NBS is successful in retaining a core contingent of trained staff—the staff turnover is very small. Since it has acquired the status of a Government Executive Agency in 1999, the NBS has introduced a salary structure that compares favorably with overall government, and which has helped to improve staff morale.

With the help of donors, computing resources are overall at a sufficient level and the software utilized is modern. Most staff have a computer at their disposal.

Until recently, the financial resources available to the NBS were not sufficient in certain areas, resulting in publication delays and gaps in the compilation of essential statistics such as the Survey of Industrial Production that had to be discontinued for several years. Meanwhile, good progress has been made in securing a more solid financial basis for the NBS’ operations, although improvements are still possible. At the proposal of the President’s Office—Planning and Privatization (POPP), the government has now guaranteed the resources to compile an extensive set of core statistics. Financing for other statistics have to come from “selling” statistical services to other institutions, among which most notably are foreign donors. Financing outside government has also to be found for several overhead costs such as the upkeep of buildings and other basic facilities.

0.2.2 Measures to ensure efficient use of resources are implemented

Measures are in place to ensure that resources are used efficiently. The NBS performs its functions in accordance with commercial principles and has to ensure that its revenues are sufficient to meet its outgoings. In that framework, it develops strategic plans for periods of three years ahead. The NBS places importance on the development of managerial skills for its staff. Both the progress of the statistical work as well as the financial achievements of the planned activities are monitored. The Treasury also insists on regular reporting, on the condition of not releasing funds for the next quarter of the year.

Efficiency in data collection and processing are sought by editing data as soon as possible after their collection, preferably already in the field. Extensive use is further made from computer assisted checking on unlikely values. Post-enumeration surveys are sometimes used as an additional check.

0.3 Quality awareness

0.3.1 Processes are in place to focus on quality

Managers are sensitive to various dimensions of data quality. The NBS formulates its mission as follows: “To facilitate planning and decision making within the government and the business community, to stimulate research and inform public debate through the provision of relevant, reliable, and timely statistics, and a quality statistical service in general.”

In 1999, the NBS conducted an extensive self-assessment of its performance on nine criteria according to a model developed by the European Foundations for Quality Management Excellence Model. Currently, it uses among other things the output-based Open Performance Review Appraisal (OPRAS) to appraise staff performance.

0.3.2 Processes are in place to monitor the quality of the collection, processing, and dissemination of statistics

The NBS may institute, on an ad hoc basis, technical commissions to guide major statistical operations such as new censuses. These commissions, in which donors, users, and providers of data are represented, discuss the structure of the questionnaires, the timing of the data collection, the time at which statistical results should be available, and so forth. The NBS’s business plan is presented in the ministerial Advisory Board. This Board councils the minister responsible for statistics on the adequacy of the Bureau’s plan and performance on functions regarding setting targets and carrying out policy priorities, including the following:

  • the development and maintenance of a strategic framework,

  • setting the objectives and priorities of the Bureau,

  • the acceptability of the annual reports and accounts, and

  • the evaluation of the Bureau’s performance.

Feedback from users is encouraged in the introductory part of several publications and by a suggestion box placed at the head office. However, periodic users’ surveys are not conducted.

0.3.3 Processes are in place to deal with quality considerations, including tradeoffs within quality, and to guide planning for existing and emerging needs

The NBS makes decisions taking into account tradeoffs within overall quality. The wish to have certain data available at the time parliament is opened plays an important role, for instance. The introduction of commercial principles in the NBS’s operations has led to larger awareness of new and emerging data requirements. At the request of the BoT, for instance, the NBS is now considering to improve the timeliness of certain statistical series.

1. Integrity

1.1 Professionalism

1.1.1 Statistics are compiled on an impartial basis

There is no formal provision explicitly addressing the professional independence of the NBS or prohibiting others from interfering in the compilation and/or dissemination of statistical information. However, the status of government executive agency implies distance from the tutelage of other government bodies. The Statistics Act also gives the Director General the power to engage such officers and employees as he may consider necessary or desirable. The NBS does not feel that its professionalism is unduly being challenged by any outside agency.

Professionalism is among other things promoted by hiring new staff on the principles of transparency of procedures and competitiveness. Similar principles are followed concerning the promotion of staff. Outside assignments and, to some extent, the presentation of technical papers and seminars given by those returning from overseas courses and conferences are further incentives for improving professional competence.

1.1.2 Choices of sources and statistical techniques are informed solely by statistical considerations

The NBS management considers itself completely free in the choice of the source data and statistical techniques. The reasoning for methodological choices it has made is currently being collected for future reference under the title Methods and Standards.

1.1.3 The appropriate statistical entity is entitled to comment on erroneous interpretation and misuse of statistics

The NBS management sees it as a task to clarify misinterpretation of their data. However, in actual practice corrective action is infrequent.

1.2 Transparency

1.2.1 The terms and conditions under which statistics are collected, processed, and disseminated are available to the public

The Statistics Act is available in the NBS library, and its website includes a profile of the NBS summarizing its vision and mission, as well as the legal foundations.

1.2.2 Internal governmental access to statistics prior to their release is publicly identified

There is extensive internal government access to statistics prior to their release which reflects the desire to receive comments that could lead to improved estimates, as well as the need to accommodate urgent political needs. This access is identified, but probably not to its full extent, in the IMF’s DSBB website; national publications do not make reference to internal government access.

1.2.3 Products of statistical agencies/units are clearly identified as such

The own products of the NBS are clearly identified as such. In the main outside statistical publications that disseminate NBS data, that is, the Economic Survey of the POPP and the publications (including the website) of the BoT, appropriate notes identify the NBS as source. Although the Bureau does not actively request attribution of the data it compiles, nonattribution does not seem an issue in practice.

1.2.4 Advance notice is given of major changes in methodology, source data, and statistical techniques

The NBS has not yet developed the standard procedure of giving advance notice of major changes in methodology, etc., in its statistical series. The main users of the statistics are likely to be aware of such future changes, though, from informal briefings on occasion of interagency meetings.

1.3 Ethical standards

1.3.1 Guidelines for staff behavior are in place and are well known to the staff

The Civil Service Standing Orders, 1994 includes a Code of Conduct applicable to all government employees. In addition, the NBS has developed a Code of Ethics that will be attached to the NBS Staff Regulations. Staff is informed about ethical standards during the first week of employment. Issues of proper ethical conduct may further be addressed in the one-day meeting the NBS annually organizes for its entire staff.

2. Methodological soundness

2.1 Concepts and definitions

2.1.1 The overall structure in terms of concepts and definitions follows internationally accepted standards, guidelines, or good practices

The PPI measures the average change in the selling prices received by manufacturing enterprises that on average have a labor force of 50 persons or more. The averaging of price changes for individual establishments and commodities makes use of output values as weights. The definition of output for use in the PPI is consistent with the overall concepts of the 1993 SNA. It refers to industrial output without distinction by type of use. Thus, for example, the production by establishments for own account is included.

More sophisticated types of PPIs such as input price and stage of processing indices are not compiled.

2.2 Scope

2.2.1 The scope is broadly consistent with internationally accepted standards, guidelines, or good practices

Regarding the scope, the PPI falls short of internationally accepted good practices in three respects. First, the coverage of index, being limited to the Tanzania mainland, excludes Zanzibar, with the result that the index does not measure changes in producer prices for the whole economy. Second, it covers exclusively manufacturing, whereas good practice would at least also include the activities of mining and the public utilities. It would even be desirable that the index covered other market activities such as transport and business services. Third, the PPI measures exclusively the prices that large (more than 50 workers) establishments receive for their main commodities. A large proportion of manufacturing output in Tanzania is produced by smaller establishments.

2.3 Classification/sectorization

2.3.1 Classification/sectorization systems used are broadly consistent with internationally accepted standards, guidelines, or good practices

The Tanzanian PPI by activity is classified by the International Standard Industrial Classification of All Economic Activities (ISIC) Rev. 2. According to current standards, this breakdown should be by the ISIC Rev. 3. The units distinguished are of the “establishment” type, which conforms with international recommendations. The PPI broken down by product does not follow any specific commodity classification.

2.4 Basis for recording

2.4.1 Market prices are used to value flows and stocks.

The Tanzanian PPI measures changes in market prices. The Annual Survey of Industrial Production, from which the weights are taken, asks for the sales price excluding indirect taxes, rebates, discounts, and drawbacks. The Quarterly Survey of Industrial Production, which generates the recurrent price information, asks for selling prices excluding sales tax and VAT. These different formulations may lead to confusion with the data providers. The methodological documents on the PPI state that ex-factory prices are used, but the questionnaires do not ask for such valuation. It may be noted that none of these prices are in conformity with either the basic price or the producer’s price as defined in the 1993 SNA.11 In practice, though, the influence of such deviations on the PPI may be limited.

2.4.2 Recording is done on an accrual basis

The PPI correctly purports to measure prices of sales effected in the middle of the month.

2.4.3 Grossing/netting procedures are broadly consistent with internationally accepted standards, guidelines, or good practices

Grossing/netting issues mostly arise in the compilation of stage of processing indices. As indicated earlier, no such indices are produced in Tanzania. In principle, transactions between different establishments belonging to the same enterprise are recorded gross. However, such transactions are extremely rare in the Tanzanian circumstances.

3. Accuracy and reliability

3.1 Source data

3.1.1 Source data are collected from comprehensive data collection programs that take into account country-specific conditions

A distinction should be made between source data for the weights of the PPI and the sources for the recurrent collection of prices.

The weights for the PPI have been derived on the basis of a list of industrial establishments generated by the 1989 Census of Industrial Production and updated by the NBS regional offices. Of some 330 establishments with 50 employed persons or more, the 1992 output values by main commodities were collected by way of the Annual Survey of Industrial Production.12 A few establishments with less than 50 workers were included in the selection to assure that all four-digit ISIC groups would be represented in the index. Together, the enterprises in the index accounted for about 85 percent of manufacturing output in the formal sector according to the 1989 Census. Output values for 1992 formed the weights of the PPI during the period 1993–1995, when the fixed-based Laspeyres formula was used.

Although 1992 remained the reference year for the PPI, since 1996 the index has been computed as a chained Laspeyres index. The data needed for the annual reweighing can be derived from the Quarterly Survey of Industrial Production, because this survey asks both for output quantities and selling prices.

The source data for the recurrent prices again is the Quarterly Survey of Industrial Production. Although this survey is conducted quarterly, it generates monthly price quotes.

No ad hoc surveys are conducted to supplement the indicated data sources.

3.1.2 Source data reasonably approximate the definitions, scope, classifications, valuation, and time of recording required

The definitions of the weights and the prices are consistent with the PPI estimates. The scope of their sources has the same limitations as the PPI itself, namely exclusion of Zanzibar, no collection of prices for market activities beyond manufacturing, and restriction to large formal sector establishments. It should be noted that the NBS is aware of the limited activity coverage and plans to expand the PPI to include mining, quarrying, electricity, and water.

The activity classification used is the outdated ISIC Rev. 2. The time of recording sale prices is the 15th of the month, which is consistent with the requirements for PPI compilation.

In variation with accepted standards, no measures are taken to eliminate quality changes from price changes. It is likely that enterprises produce commodities of different qualities, for instance those directed at the foreign market may be of better quality than those for domestic use. A distinction between domestic and foreign customers is not made in the weighting scheme. Although apparently instructions are that prices for domestic use should be reported, such is not indicated on the questionnaire forms. Equally, the influence that sales conditions may have on prices is disregarded. The quality of products is, of course, also likely to change over time. The absence of narrow definitions in respect of the selected commodities gives the PPI characteristics of a unit value index, which makes the index prone to incorrect price measurement and biases over time.

3.1.3 Source data are timely

The source data are not timely enough to allow compliance with GDDS recommendations. Price data for the first month of a quarter necessarily have a delay of two months, to which should be added another month for communication from the NBS regional offices to the head office in Dar es Salaam and data processing. Furthermore, although the authorities indicate that the response rate is good (90–98 percent), late responses may require the imputation of prices for long periods, sometimes even as long as one year. About three to five percent of the prices for the most recent quarter would have to be imputed.

3.2 Statistical techniques

3.2.1 Data compilation employs sound statistical techniques

The index is compiled in sufficient detail, namely for 37 activity groups and 196 commodities.

The elementary indices are defined for individual establishments and commodities. The lowest-level weights are thus the base period output values for distinct commodities in each selected establishment, and no averaging of current prices takes place to determine the elementary price relatives. The formula to aggregate elementary indices to activity and commodity groups currently is the chained Laspeyres formula, which is one of the alternatives generally accepted as good practice. The chained index implies that the weights in any year are formed by the output values recorded for the previous year. Such annual rebasing is generally preferred over an index with fixed weights, because the price index better reflects current circumstances and for instance takes account of substitution effects between products when their prices change differently. Annual chaining also provides an opportunity to review the selected enterprises and commodities on which the index is computed. For instance with the introduction of chaining, fish fillet, malt, and pharmaceutical products were incorporated in the index, while canned meat, baby food, and razor blades were eliminated either because they were not produced anymore, or had insignificant output values in 1995. According to the authorities, all removals and additions were made under the redistribution of the index weights of the group to which the removed or added items belong. The base period prices of the newly added commodities were imputed under the assumption that they showed the same historic price changes as the other items in their group.

At the inception of the index, no adjustments were made to align the original weight reference period (the full year of 1992) and the price reference period (December 1992). According to good practices, such alignment should have been made.

The base year for the PPI has not been changed since 1992. However, with the introduction of chained indices, the exact year in which the index is taken to be equal to 100 has become of less importance.

The method to deal with temporarily missing prices is to carry the last quoted price forward. Although acceptable for short periods, this is not the most adequate methodology if prices are missing for longer periods. The preferred method for dealing with missing prices is to impute a price change on the basis of the observed change in prices for similar products.

No specific methodology has been developed to handle missing prices for seasonal products as seasonal production is not a feature of Tanzanian manufacturing production.

In the years 1992 through 1995, products that became permanently unavailable usually were dropped from the computation of the index, accompanied with a redistribution of the weights these products had in the base year basket. As indicated in section 3.1.1., at the introduction of chained indices in 1996, the selection of enterprises and commodities were reviewed. After 1996, no further replacements have been made for permanently missing prices and no new products have been introduced that come within the scope of the PPI.

3.2.2 Other statistical procedures (e.g., data adjustments and transformations, and statistical analysis) employ sound statistical techniques

No specific adjustments are made on the basic statistics to bring their coverage, definitions, classifications, and valuations closer to international guidelines for the PPI compilation.

As indicated earlier, the Tanzanian PPI covers only a minor section of the total market economy. For specific economic activities, there are alternative statistics available, such as for agriculture, mining, and the public utilities. Generally, no reliable information of producer prices are available for the informal sector, manufacturing establishments employing between ten to 50 persons, and most producers of market services.

The section responsible for the PPI does not keep adequate documentation of the methods used to compile the index. For instance, no detailed description is available on the introduction of the chained Laspeyres formula. The lack of institutional memory makes the section very vulnerable when experience staff leaves to work elsewhere.

3.3 Assessment and validation of source data

3.3.1 Source data—including censuses, sample surveys and administrative records—are routinely assessed, e.g., for coverage, sample error, response error, and nonsampling error; the results of the assessments are monitored and made available to guide planning

The source data received from NBS’s regional offices are edited at headquarters. The checking is mostly done by comparing the last reported three months with the results of the previous quarterly return. In case unlikely prices are noted, clarification is sought from the producers. These contacts are nearly exclusively done by telephone, but they have been rare during the past year. The reported prices tend to remain stable during long periods of time. The influence of outliers is very limited and refers usually to obvious mistakes. The responses on the Quarterly Survey of Industrial Production are also compared with the responses on the annual survey. Occasionally, headquarter staff visit establishments in the field to discuss reporting problems.

Most corrections that are applied concern the imputation of missing prices. The computer program allows headquarter staff to add comments with the stored basic data.

3.4 Assessment and validation of intermediate data and statistical outputs

3.4.1 Main intermediate data are validated against other information where applicable

Before publication, infrequently a comparison is made with the results of the consumer price index. No validation takes place against export prices. The definition differences are considered too large to make such validation a useful exercise.

3.4.2 Statistical discrepancies in intermediate data are assessed and investigated

Unusual fluctuations in the index may lead to further investigations in the accuracy of the data and in possible explanations for the observed trends. An example is the investigations that were made concerning the fall in prices recorded for textiles and soap. The results of these studies were not publicized, although external agencies that had expressed concerns were informed of the findings. A handicap in the dissemination of such investigations is that no official publication of the Producer Price Index (Manufacturing Sector) has taken place since the third quarter of 1995.

3.4.3 Statistical discrepancies and other potential indicators of problems in statistical outputs are investigated

All higher level activity and product-group indexes are derived from the lower level indexes. Hence there are no statistical discrepancies.

3.5 Revision studies

3.5.1 Studies and analyses of revisions are carried out routinely and used to inform statistical processes

The data for the three months received with the most recent quarterly survey are regarded as provisional. Reporting enterprises revise the reported value and volume data very rarely. Revision studies therefore nearly exclusively relate to investigating the prices imputed by the NBS for the months over which no reports were received with the later reported prices. Even when the validity of the imputations is routinely assessed, this does not feed into the imputation process for subsequent periods. The results of the investigations are not publicized.

4. Serviceability

4.1 Relevance

4.1.1 The relevance and practical utility of existing statistics in meeting users’ needs are monitored

No established process of consultation takes periodically place with policy departments, the private sector, universities, and other external agencies to assess whether the program meets the needs of the PPI users.

4.2 Timeliness and periodicity

4.2.1 Timeliness follows dissemination standards

The dedicated NBS publication Producer Price Index has not been disseminated after January 1996, when the data for the third quarter of 1995 were released. However, draft tables containing data for the twelve months of the year are being prepared annually and stored in the NBS library where they are available to users on request. The set of draft tables for January through December 2000 was ready in February 2001. In addition to these annual drafts, the PPI series are made available on a quarterly basis to three or four main users. For instance, data for the first quarter of 2002 were provided to these users in July 2002 (the late distribution was caused by exceptional circumstances). These annual and quarterly drafts cannot count as publications in the sense of the GDDS.

The, in principle, annual Statistical Abstract includes a table with PPI information. The first draft of the Statistical Abstract for 2001 was available in September 2002, but the NBS does not use it to inform the general public. The official hardcopy publication is only expected by January 2003. Also taking into account that the Statistical Abstract has not been published since the issue for 1994, one must conclude that the Tanzania PPI is not yet disseminated to the general public.

4.2.2 Periodicity follows dissemination standards

The Statistical Abstract for 2001 will include a table showing the PPI, broken down by three-digit ISIC major activity groups, for only four months, namely, March, June, September, and December 2001. This falls short of the GDDS recommendation to disseminate the PPI with a monthly periodicity.

4.3 Consistency

4.3.1 Statistics are consistent within the dataset

The PPI series are consistent internally, so that the all items aggregate for the indices by activity is equal to the all-items aggregate by type of commodity.

4.3.2 Statistics are consistent or reconcilable over a reasonable period of time

Consistent time series calculated using the chain-linked method are available for an adequate period of time, that is, 1996 through 2001.

4.3.3 Statistics are consistent or reconcilable with those obtained through other data sources and/or statistical frameworks

The PPI data are in line with national accounts deflators of production as they directly feed in the process of compiling the national accounts.

4.4 Revision policy and practice

4.4.1 Revisions follow a regular, well-established, and transparent schedule

Provisional monthly data may be revised over the same calendar year, but revisions are not applied to earlier periods. This revision policy is known to major users but not made explicit in text accompanying the data sent to them.

The NBS intends to change the base year every ten years, and the base year of the current series (1992) will be changed to 2002. It should be noted that with the introduction of chained indices, and thus annual adaptations of the index weights in 1996, new weights and, in principle, new products can be introduced gradually in the PPI. This diminishes significantly the importance of a change of base year.

4.4.2 Preliminary data are clearly identified

Preliminary data are not identified as such.

4.4.3 Studies and analyses of revisions are made public

Studies of revisions in the monthly data, which pertain in practice to the replacement of estimated by reported figures, are not made. The change of the base year in 1992 gave rise to the publication of a short analysis of the changes. The introduction of the chained indices has not led to publication of analyses.

5. Accessibility

5.1 Data accessibility

5.1.1 Statistics are presented in a way that facilitates proper interpretation and meaningful comparisons (layout and clarity of text, tables, and charts)

The PPI tables in the draft Statistical Abstract are easy to understand. Otherwise than other available tables, they also clearly indicate the geographical restriction to the Tanzania mainland and the fact that December 1992 equals 100 in the index (rather than 1992 as a whole).

The former dedicated publication Producer Price Index (Manufacturing Sector), contained a section “What is new in the report.” This section provided an explanation of the index and it also compared the data by major activity group for the last month of the most recent quarter with those for the last month of the previous quarter and with the base period. The section included a bar chart with perceptual changes to facilitate the interpretation of the data. Similar explanatory comments are not attached to the currently available tables.

5.1.2 Dissemination media and formats are adequate

Currently, the dissemination of the compiled data is exclusively informal. The few major users of the PPI are adequately served, but the general public not at all. The revived Statistical Abstract, being an annual directed at a wide public, will not be an adequate medium to release monthly series.

5.1.3 Statistics are released on a pre-announced schedule

The statistical series are not released according to a pre-announced schedule.

5.1.4 Statistics are made available to all users at the same time

The PPI is not made available to all users at the same time: the important users receive quarterly data that others see only later, if at all. The important users also receive the draft tables containing data for all months of the year, whereas outsiders are not likely to know if and when this information is available.

5.1.5 Nonpublished (but nonconfidential) subaggregates are made available upon request

All subaggregates are made available in the detailed index tables, even to the point of revealing some data (for instance prices for particular brands of cigarettes) that should be treated as confidential.

5.2 Metadata accessibility

5.2.1 Documentation on concepts, scope, classifications, basis of recording, data sources, and statistical techniques is available, and differences from internationally accepted standards, guidelines, or good practices are annotated

The most extensive methodological description of the PPI is provided in the GDDS metadata posted on the DSBB that the authorities update, albeit with several months’ delay. The foreseen hyperlink from the NBS website to the IMF site is not yet established. On request, a three-page methodological description can be obtained from the NBS that explains, among other things, the index formula used and the principle of chaining. However, many details that could interest a sophisticated user are not addressed in this description.

5.2.2 Levels of detail are adapted to the needs of the intended audience

Apart from the methodological description mentioned above, there are no brochures to inform the general user about the PPI.

5.3 Assistance to users

5.3.1 Contact person for each subject field is publicized

All official publications of the NBS clearly indicate how the Bureau can be contacted. Many users with queries prefer to visit the NBS in person. Prompt and knowledgeable support is available to users, but the releases do not indicate specific individuals that can help users on questions regarding the PPI.

The assistance to users is not monitored through surveys.

5.3.2 Catalogs of publications, documents, and other services, including information on any charges, are widely available

The NBS is setting up a section on the marketing of statistical products. However, no Catalogs of publications are available.

Table 3.

Data Quality Assessment Framework—Summary of Results for Price Statistics (Producer Price Index)

(Compiling Agency: National Bureau of Statistics)

article image
article image

IV. Government Finance Statistics13

0. Prerequisites of quality

0.1 Legal and institutional environment

0.1.1 The responsibility for collecting, processing, and disseminating statistics is clearly specified

Although no specific legal responsibility has been assigned to a single agency to compile government finance statistics (GFS), the Public Finance Act, 2001, (Article 5.1 (b)) assigns the responsibility to supervise and monitor the finances of the United Republic to the Minister of Finance. The act also stipulates that the minister presents a full set of government accounts to the National Assembly from time to time, but not less than annually. Article 25 of the act prescribes a list of accounts that the Accountant-General must submit to the Controller and Auditor-General within six months after the end of each fiscal year.14 The Public Finance Regulations, 2001 also mandate the Accountant General to prescribe the basis of accounting to be adopted in Tanzania and the classification system to be used. However, neither the act nor the regulations specifically assigns the responsibility to compile and disseminate government finance statistics in line with international standards such as those set out in the Government Finance Statistics Manual 1986 (GFSM 1986).

0.1.2 Data sharing and coordination among data producing agencies are adequate

Although data sharing and coordination among the agencies producing government data are adequate, the sharing and coordination with other users of data could be improved.

A Budget Ceiling Committee, comprising representatives of all agencies involved in the compilation of government data, has been established. This formal committee meets regularly to discuss progress with the execution of the budget and the stance of the government finances. The Budget Execution Report, compiled by the Policy Analysis Department (PAD) of the Ministry of Finance (MOF), is dependant on the Accountant General’s Department (AGD), the Bank of Tanzania (BoT) and the Tanzanian Revenue Authority (TRA) for data that serves as an input for the report. The formal meeting between these agencies is also supplemented with regular informal contact and feedback. The Commissioner of the PAD also serves as a member on the Monetary Policy Committee of the BoT. Government debt issues and financing of government are also regularly discussed among the agencies.

The relationship between compilers and other macroeconomic data producing agencies, has not been formalized. Although informal contact exists, there remain inconsistencies in the datasets between the various agencies. The lack of data sharing through an electronic network hampers the work of compiling fiscal data between the MOF and the National Bureau of Statistics. The compilation of statistics was identified as a priority sector spending and received an increased budget allocation in the June 2002 budget.

0.1.3 Respondents’ data are to be kept confidential and used for statistical purposes only

Data on government finances are not considered to be confidential. Data collected on the finances of government are used in policy decision-making. As such, the fiscal authorities are accountable to the National Assembly and the public in general. Official publications containing details of government finances (budget document, audit reports, parliamentary submissions, etc.) can be obtained from the government’s publications agency.

0.1.4 Statistical reporting is ensured through legal mandate and/or measures to encourage response

A general legal mandate given to the Permanent Secretary of the MOF encourages compliance with the requirement to provide information on government finances. The Public Finance Act, 2001, (Article 6.4 and 6.5) grants the Permanent Secretary of the MOF the right to request information from all government agencies as he may deem necessary to fulfill his duties and responsibilities. The act also provides the Permanent Secretary with the authority to specify the date by which the information must be submitted and the format in which it should be presented.

0.2 Resources

0.2.1 Staff, financial, and computing resources are commensurate with statistical programs

There are no resources specifically dedicated to compiling GFS. In general, resources to produce government finance data are limited. The Minister of Finance noted in his budget speech for 2002/03: “The shortage of technical and professional staff in government, including accountants, threatens to jeopardize the sustenance of the many reforms the government has implemented including in financial management.” Although the government provides employees with opportunities to attend workshops, seminar and training courses, these events are few due to budgetary constraints. Employees are allowed to further their studies by entering higher training institutions while in employment. A number of government agencies were set up with the aim to provide training opportunities for government employees. Expert consultants often initiate reforms in government and provide on-the-job training to local personnel. However, by and large, employees involved in the compilation of government statistics have had no formal GFS training.

The Integrated Financial Management System (IFMS) was recently implemented as the accounting system for all budgetary accounts. It is envisaged that the accounting systems of all extrabudgetary units be replaced with the IFMS. However, limited financial and computing resources are a serious constraint on this development. A lack of technical knowledge of computer systems also has severe effects on the implementation of reforms.

0.2.2 Measures to ensure efficient use of resources are implemented

The responsibility to compile the GFS is not specifically assigned and the cost associated with compiling GFS is not yet measured. Scarce consultant resources not being utilized to ensure sustainability of compilation procedures.

0.3 Quality awareness

0.3.1 Processes are in place to focus on quality

Participants in the process of compiling government finance data are focused on the quality of information available and strive to improve quality. Accountability to the public and internal financial management improvements have led to improved systems, for compiling data on government finances. Data are produced from actual records in the accounting system of government for which clear rules and a code of conduct exists.

0.3.2 Processes are in place to monitor the quality of the collection, processing, and dissemination of statistics

The collection and processing of government finance data is vested in the accounting system of government. Various processes are in place to monitor the quality of the collection processing and dissemination of the data. In accordance with the Public Finance Act, 2001 (Article 8), the Paymaster-General appoints an Accounting officer for each expenditure vote. This officer is responsible and accountable for all revenue and expenditure of the vote, and the recording of it. Furthermore, the Public Finance Regulations, 2001 also assigned the Accounting officer the responsibility to establish an effective internal audit service in each ministry, department or agency as well as an audit committee. These committees must review all internal and external audit reports and provide advice on action to be taken.

The Controller and Auditor-General (CAG) has the responsibility to perform annual audits and examine financial reports and report on the outcome. The annual audit reports is submitted to parliament and discussed extensively by the Public Accounts Committee. The audited reports are available to the public at the government’s publications agency.

0.3.3 Processes are in place to deal with quality considerations, including tradeoffs within quality, and to guide planning for existing and emerging needs

Although not communicated explicitly, a trade-off between quality and availability of data is made. Comprehensive information on the detailed classification of the expenditure of the extrabudgetary units and development funds is not available from the systems in use. In order to have information available, central government data are presented without a detailed classification of these amounts. Comprehensive data on the local government are not available. Government finance statistics is therefore limited to the central government. Plans to expand the coverage of government and comprehensive statistics on the nature of the expenditure of extrabudgetary units are being developed.

1. Integrity

1.1 Professionalism

1.1.1 Statistics are compiled on an impartial basis

Annual audited data are compiled on an impartial basis as called for the Public Finance Act, 2001. The act stipulates that all officers and employees in the national audit office shall be able to: “perform their functions without fear, favor or prejudice and that no person or authority shall interfere with or exert undue influence on them in the exercise of their powers or in the performance of their professional functions.” The data contained in the quarterly Budget Execution Report, are compiled by the PAD of the MOF from source data obtained from various other departments and agencies. The PAD is also involved in the process of compiling the budget, which may compromise impartiality.

1.1.2 Choices of sources and statistical techniques are informed solely by statistical considerations

No statistical techniques are used in the compilation of the government data since all the data currently disseminated are obtained from actual accounting records.

1.1.3 The appropriate statistical entity is entitled to comment on erroneous interpretation and misuse of statistics

The dissemination of government finances data is accompanied by an extensive analysis of the outcome of government finances that minimizes the possibility of misuse or erroneous interpretation. The official in PAD responsible for disseminating data on government finances provides technical advice and explanations in the publication and on request. However, comments on fiscal data are only addressed through formal channels within the MOF.

1.2 Transparency

1.2.1 The terms and conditions under which statistics are collected, processed, and disseminated are available to the public

The Public Finance Act, 2001 and the Public Finance Regulations, 2001 based on the act are widely available to the public. These documents contain the general terms and conditions under which government finance data are collected. These terms are also disseminated on the DSBB. However, no other documentation describing the process of collection, processing and dissemination are available to the public.

1.2.2 Internal governmental access to statistics prior to their release is publicly identified

Various monthly reports on government finances are compiled for use within the government sector for financial management purposes. These reports are not widely disseminated to the public, although they are made available on request. Neither the availability of the reports, nor the internal access to the statistics, is publicly identified.

All the reports on government finances disseminated to the public are subject to a process of formal approval. The approval processes are not made public.

1.2.3 Products of statistical agencies/units are clearly identified as such

Government finance data disseminated by the PAD are widely used and released in various publications. The statistical product is usually accompanied by a footnote that discloses the origin of the statistics.

1.2.4 Advance notice is given of major changes in methodology, source data, and statistical techniques

Although explanations of major changes in methodology are disseminated with the data, no advance notice is given of major changes in methodology, source data and statistical techniques. Changes in methodology, source data and statistical techniques are disclosed by inserting a break in the time series, which is explained by a footnote to the statistical tables.

1.3 Ethical standards

1.3.1 Guidelines for staff behavior are in place and are well known to the staff

Although compilers of government finance data are not subject to a specific code of conduct guiding their behavior, all government employees are subject to the Civil Service Department Standing Orders, 1994. The document describes the conditions of employment of civil servants, the code of conduct, and disciplinary procedures that will be followed in cases of misconduct. In addition to these general ethical standards set for staff, the Public Finance Regulations, 2001 also governs the code of conduct of accounting officers. These officials, being the custodians of government, are subject to specific regulations on their responsibilities, authority and accountability. Misconduct is clearly defined, and in line with the Public Finance Act, 2000 (Article 44), any person convicted of an offense can be prosecuted.

2. Methodological soundness15

2.1 Concepts and definitions

2.1.1 The overall structure in terms of concepts and definitions follows internationally accepted standards, guidelines, or good practices

The overall structure in terms of concepts and definitions in use in Tanzania only partly adheres to the internationally accepted standard in GFSM 1986. Compilation and dissemination of data on the central government operations are prescribed in the Guidelines for the preparation of medium term plan and budget framework compiled by the MOF and issued to all government agencies by the President’s Office—Planning and Privatization. This framework is used in all budget documentation as well as in the quarterly Budget Execution Report compiled by the PAD of the MOF. The framework for revenue makes a distinction between domestic revenue and grants (all grants are received from foreign donors). The framework for expenditure makes a distinction between recurrent expenditure and development expenditure. Interest payments are identified as a separate item in recurrent expenditure. Although the classification of domestic revenue and recurrent expenditure broadly follows the analytical framework of the GFSM 1986, concepts and definitions of revenue, expenditure, lending minus repayment and financing differ from the international standard in significant ways. (See Section 2.3.1 below.)

Data on central government operations are disseminated in a number of publications using the same source data. However, transactions are presented in different analytical frameworks resulting in different definitions for the deficit of central government.16

Although the AGD of the MOF and the BoT compile various reports on debt, a comprehensive report on the total outstanding debt of the government, compliant with international guidelines, is not disseminated to the public. Data on government debt currently disseminated to the public in the quarterly Bank of Tanzania Economic Bulletin include:

  • Treasury Bills Auctions – Actual Sales by Transactors reporting flows denoted in Tanzania shilling (Table 2.4).

  • Central Government Outstanding Stocks, Bonds and Notes by holders reporting stocks at face value denoted in Tanzania shilling (Table 2.5).

  • External debt developments are denoted in U.S. dollars with government external borrowing separately identified (Table 5.1).

The AGD of the MOF also distributes a Monthly Domestic Debt Report, which summarizes the debt position of the Union government, to other government agencies. This report presents domestic debt according to instrument, type of debt holder and origin of the debt (purchase of buildings, nationalization etc.). However, this report is not disseminated to the public.

The Debt Management Department (DMD) of the BoT compiles a report on the external debt of government comprising mainly bilateral and multilateral loans negotiated by the Union government. Although the coverage is considered to be relatively comprehensive, a few bilateral loans contracted directly by the government of Zanzibar are not included in this database.

The authorities are engaged in a number of initiatives to improve the existing framework and classifications.17 However, no plans have been made to migrate to the GFS 2001 framework.

2.2 Scope

2.2.1 The scope is broadly consistent with internationally accepted standards, guidelines, or good practices

The disseminated summary data on central government operations cover the revenue and expenditure of all budgetary units and extrabudgetary accounts. Revenue collected by extra-budgetary fund is deposited in the central government account on a gross basis. Detailed data on the nature of the expenditure of extrabudgetary accounts are therefore not available. Expenditure also includes all transfers made to Zanzibar, local governments and parastatals. This scope does not include own revenue raised by local governments and matching expenditure financed by these revenues. Approximately thirty per cent of local government total revenue is deemed to be own revenue raised. The government units covered in central government operations of Tanzania broadly meet the guideline of the GDDS. Separate annual reports on the financial position of local governments are available and presented to parliament. However, the local government financial year differs from that of the central government, which limits the possibility of expanding the scope of the data on government finance operations to cover the general government. A project to develop local government quarterly reporting to the Office of the President, Regional Administration and Local Governments, has been initiated and will eventually facilitate consolidation of local government data with central government data in order to derive data on general government operations.

Data on all cash transactions of central government are available, but data also include an amount of grants received in kind. The data on central government operations are presented according to a national analytical framework and do not cover the full scope of data as recommended by the tables recommended in the GFSM 1986.18

Data on central government gross debt, classified by residency, are available. The domestic debt data cover all outstanding debt of the Union government. Under the Government Loans, Guarantees, and Grants Act No. 30 of 1974, only the Minister of Finance is allowed to contract debt on behalf of the government.

In accordance with the Union agreement, only the Union government will incur foreign borrowing. Therefore, foreign debt data only cover the Union government’s external debt. The data on foreign liabilities also include obligations to the IMF, which is the liability of the BoT, but was guaranteed by the government.

2.3 Classification/sectorization

2.3.1 Classification/sectorization systems used are broadly consistent with internationally accepted standards, guidelines, or good practices

The data disseminated on central government operations and debt are not fully consistent with the classification/sectorization recommended in the international standards.

Although the classification of revenue and expenditure broadly follows the analytical framework of the GFSM 1986, concepts and definitions of revenue, expenditure and lending minus repayments differ from this international standard in significant ways:

  • data on revenue distinguish between tax revenue and nontax revenue. Although tax revenue is classified according to the GFSM 1986, nontax revenue includes capital revenue, repayments of loans extended for policy purposes, and revenue generated by government business enterprises, on a gross basis. The GFSM 1986 recommends that:

  • data on current revenue and capital revenue be separated;

  • repayments of loans extended for policy purposes be netted against loans extended for this purpose; and

  • only the net result of revenue and expenditure of government business enterprises be recorded in the accounts of the owning level of government.

Recurrent expenditure does not separately identify current expenditure and capital expenditure. Wages and salaries in the central government operations table also include salaries paid to local government employees without identifying this category separately. The wages and salaries of parastatals are included as a separate line item in government payments for goods and services and transfers. Loans extended for policy purposes are included in expenditure. The GFSM 1986 recommends that:

  • data on current expenditure and capital expenditure be separated;

  • wages and salaries recorded in the account of central government comprise only the remuneration of employees of that level of government. Wages and salaries for local government employees will have to be recorded as a transfer to another level of government in the data of central government. The local government will record the amount received as a grant received from another level of government and will record a matching expenditure on wages and salaries;

  • all unrequited payments to other levels of government be identified separately at the level of government where the transaction occurs;

  • loans extended for policy purposes be excluded from expenditure, netted with repayments and are reported as a separate item;

  • lending minus repayment is not identified separately due to the inclusion of the latter in domestic revenue and the former in recurrent expenditure; Grants include all receipts in cash and in kind, as well as the cash value of debt relief. The GFSM 1986 recommends that:

    • grants received in kind be recorded as a memorandum item because they do not represent a cash flow; and

    • debt relief be recorded as decrease in outstanding liabilities with a memorandum item explaining that the relief resulted in a discrepancy between the net change in debt and financing.

  • A detailed economic classification of development expenditure (also referred to as the project budget) is not available whereas the international practice is to include the expenditure in the economic classification of total expenditure.

  • Domestic financing transactions include the net proceeds from the privatization fund. The GFSM 1986 recommends that all government receipts reducing equity be recorded and a repayment of lending for policy purposes.

  • Domestic financing transactions include the net change in arrears as financing activities whereas that should be recorded as expenditure in a cash system of recording transactions.

  • Government external debt includes guarantees issued by the central government for third party liabilities. However, these amounts are relatively small compared with the total outstanding external debt of government. According to the international standards, outstanding government debt excludes debt guaranteed by government which should however, be noted as a memorandum item in the analytical presentation of debt.19

Transactions between units within the subsectors and all transactions with government units outside the central government (local governments and public corporations) are not clearly identified separately. Data on the economic classification of revenue and expenditure of the sub-sectors of central government are not disseminated, and administrative arrangements have established no clear dividing lines between the transactions of different spheres of government.20

2.4 Basis for recording

2.4.1 Prices used to value flows and stocks reflect actual or expected cash payments.

Central government operations are recorded in accordance with the GFSM 1986 on a cash basis. Tax revenue is recorded when payments are received, and nontax revenue is recorded when deposits are made. Expenditure is recorded when checks are written or payments are made. The difference between checks recorded and cleared with the bank is recorded in a revenue and expenditure “float.” The overall deficit is adjusted with these amounts in order to calculate the cash deficit that equals the net financing transactions.

In accordance with the GFSM 1986, debt liabilities are recorded at face value (i.e., the amount to be repaid at the end of the contract), rather than market value. Transactions in foreign currencies are converted to Tanzania shilling using the official exchange rate prevailing at the time of the transactions. However, the stock of foreign debt are converted using the exchange rate prevailing at the end of the period to which it refers, which is in line with the international standard.

2.4.2 Recording is done on a cash basis

All transactions are recorded when payments are made or received. Recording of transactions is therefore done on a cash basis in accordance with GFSM 1986.

2.4.3 Grossing/netting procedures are broadly consistent with internationally accepted standards, guidelines, or good practices

Grossing/netting procedures are broadly adhered to because all nonindustrial payments and receipts of central government are correctly treated on a gross basis. Corrective transactions on tax refunds and expenditure recovered are correctly netted against the appropriate class of tax or expenditure. However, contrary to the international standard, lending and repayments transactions are treated on a gross basis and included in expenditure and revenue respectively. Activities of departmental enterprises such as the printing agency are also recorded on a gross basis contrary to the international standard which recommends netting their transactions before recording the net result in the accounts of the unit of government they report to.

3. Accuracy and reliability

3.1 Source data

3.1.1 Source data are collected from comprehensive data collection programs that take into account country-specific conditions

Data collection programs in use to compile source data are comprehensive for data on recurrent revenue and expenditure of budgetary accounts, but data collection programs covering extrabudgetary expenditure and development expenditure are not available. Comprehensive data covering the recurrent revenue and expenditure of central government operations are obtained from administrative records generated by the administrative systems of government. Data on tax revenue are compiled by the TRA from their administrative records. Data on the recurrent expenditure of government are compiled by the AGD from records obtained from the IFMS.

Data on grants and donor funds received are compiled from reports by ministries and from the records of donors. A detailed breakdown of development expenditure is not available, but a project has been initiated to improve reports on development expenditure in accordance with the international standards. However, the Minister of Finance noted in his Budget Speech for fiscal 2002/03 that: “There is a possibility that a substantial amount of donor assistance for development projects was disbursed outside the government budget system, and has not yet been reported to government.” Data on grants and donor funds can therefore not be considered to be comprehensive.

Net domestic financing data are compiled by the BoT from the administrative records of the BoT. The Commonwealth Secretariat Debt Recording and Management System (CS-DRMS) shared by the BoT and AGD serve as the source for compiling the external debt data.

3.1.2 Source data reasonably approximate the definitions, scope, classifications, valuation, and time of recording required

The source data for domestic revenue contain enough detail to enable compilers of central government operations reports to classify revenue broadly in accordance with the classifications used in the GFSM 1986, with the exception of repayments of loans previously extended for policy purposes. The source data for recurrent expenditure are compiled in line with the requirements of the analytical framework in use in the country, and are not in compliance with the GFSM 1986. Detailed information on the economic and functional nature of development expenditure and the expenditure of extrabudgetary units is not available from the present system. However, a project in progress in the office of the AGD aims at improving the availability of data on development expenditure in accordance with GFSM 1986. Similarly, a project was initiated to derive data on revenue and expenditure of local government, in accordance with the comprehensive analytical framework of the GFSM 1986, from local government accounting records.

The timing of recording and valuation of the source data are broadly consistent with GFSM 1986 concepts.

3.1.3 Source data are timely

The source data on central government operations and central government debt are timely. Preliminary monthly data on central government operations are available, for internal use by the MOF, within three weeks after the end of the reference period. Although, the monthly data are not disseminated, the quarterly data disseminated to the public have a timeliness of four weeks after the end of the reference period.

All data are considered to be preliminary until the completion of the audit of the annual financial statements of government. Audits are currently completed within eight months after the end of the fiscal year. Data are then revised at which time that are considered to be final.

Comprehensive data for local governments are published annually with a two-year lag after the reference period. The current presentation does not allow for compiling consolidated data on general government.

3.2 Statistical techniques

3.2.1 Data compilation employs sound statistical techniques

Data compilation is based on actual administrative records with no need to employ any statistical techniques in the compilation process. Source data contain sufficient details to enable reclassification into the budget framework and the GFSM 1986 framework.

No process of consolidation of central government operations is followed because the data include all the transactions of the budgetary and extrabudgetary accounts on a gross basis. Wages and salaries of local governments are paid from the budgetary accounts but are not identified separately. Other transfers to local governments and public corporations are also not identified as transfer payments made. Although privatization proceeds are recorded as a separate financing item, other lending and repayments transactions with public corporations is not identified.

There is no need to employ estimations in the calculation of central government operations because actual records cover all transactions.

Data on the general government are not available and no estimations are made.

3.2.2 Other statistical procedures (e.g., data adjustments and transformations, and statistical analysis) employ sound statistical techniques

Actual outcomes for the current year are compared with the outcomes in previous years to derive a medium-term trend. Data adjustments and reclassifications follow sound statistical techniques.

3.3 Assessment and validation of source data

3.3.1 Source data—including censuses, sample surveys and administrative records—are routinely assessed, e.g., for coverage, sample error, response error, and nonsampling error; the results of the assessments are monitored and made available to guide planning

The results contained in the quarterly Budget Execution Report and the monthly debt reports are used for policy decision-making. These reports based on the administrative records of the AGD, BoT and PAD form the basis on which the ceilings and allocation of funds for the consecutive periods are determined. As such, the reports are routinely assessed and monitored by different units of government. The reports are based on actual administrative records, which minimizes the likelihood of inconsistencies. Should deviations from actual outcomes become apparent, their causes are investigated, data are revised in the next publication and corrective measures are initiated to avoid a recurrence of the error or omission. The adjustment line between the overall deficit and net financing also serves as a monthly indicator of the accuracy of source data21. Annual reconciliation of the checks issued and received and the checks cleared through the banking system serves as a further built-in check on the source data. Debt data are compiled from data compiled by the AGD and the BoT.

Net financing data are annually reconciled with data on central government operations. The revenue and expenditure float is also reconciled with outstanding checks and deposits on an annual base. However, the differences in the coverage and valuation of central government operations and central government debt do not facilitate an accurate reconciliation of the stock of debt with the financing of central government.22

3.4 Assessment and validation of intermediate data and statistical outputs

3.4.1 Main intermediate data are validated against other information where applicable

Timing differences in various publications, differences between the analytical frameworks in use, as well as timing differences in the incorporation of revisions have led to inconsistencies in the deficit of central government as published in various publications. Intermediate data on grants received and development projects are unreliable and are subject to frequent revisions. The envisaged improved database for these funds should lead to the availability of better data on these funds.

3.4.2 Statistical discrepancies in intermediate data are assessed and investigated

The authorities involved in the compilation of government statistics are well aware of the discrepancies in data on government finances and efforts are made to improve the data. Efforts are being made to improve systems, train staff in the recording, validating and analyzing data and to expand the coverage of data systems. These efforts however, are limited by a lack of resources and are heavily dependent on donor support.

3.4.3 Statistical discrepancies and other potential indicators of problems in statistical outputs are investigated

Statistical discrepancies and other potential indicators of problems in statistical outputs are investigated when they become apparent. There is no fixed revision schedule apart from the annual finalization of data after the completion of the annual audits. Internal audits are done on a periodic basis in all government agencies and cover scrutiny of output information and system. The audit committees of departments have the responsibility to investigate the discrepancies and to suggest possible action plans.

3.5 Revision studies

3.5.1 Studies and analyses of revisions are carried out routinely and used to inform statistical processes

Revisions of data are incorporated whenever improved information becomes available. Data are finally revised after the completion of the annual audits of government accounts by the CAG. Memorandums for files are prepared on revisions.

4. Serviceability

4.1 Relevance

4.1.1 The relevance and practical utility of existing statistics in meeting users’ needs are monitored

The usefulness of data for users within government and donors is monitored, but the relevance of data and needs of other users are not monitored.

Data on central government operations and budget documentation are prepared according to the same analytical framework as prescribed by the Guidelines for the preparation of medium-term plan and budget framework (see Section 2.1.1 above). Data on central government operations are therefore used in the process of determining and evaluating fiscal policy developments. Progress with the execution of fiscal policy decisions can be measured against stated policy objectives. Members of parliament and selected government committees are considered to be the main users of fiscal data. Donor agencies also have a special interest in the progress made with fiscal policies in general and in the utilization of donor funds. Regular meetings with donors provide an opportunity for feedback from them. However, no regular feedback on the relevance of fiscal data from the public at large is sought.

4.2 Timeliness and periodicity

4.2.1 Timeliness follows dissemination standards

The timeliness of quarterly disseminated central government operations data is four weeks after the end of the reference period. This complies with the GDDS recommendation to publish quarterly central government operations data within one quarter after the end of the reference period. These data are considered to be preliminary. Detailed annual audited data on central government operations become available within eight months after the end of the fiscal year. Annual data on the consolidated general government are not compiled. The GDDS encourages subscribers to publish annual consolidated general government data within six to nine months after the end of the reference period.

A monthly report on central government operations, which is only distributed within government, is available within three weeks after the end of the reference month. The timeliness of central government debt data published in the Bank of Tanzania Economic Bulletin is within 1–2 quarters after the reference period. This complies with the GDDS recommended practice to publish annual central government debt data within 1–2 quarters after the end of the reference period. The PAD of the MOF is engaged in a project that will enable them to publish quarterly reports on the comprehensive debt position of the central government within a month after the end of the reference period. It is expected that these reports will be published on the official website of the Tanzanian government, by the end of 2002.

4.2.2 Periodicity follows dissemination standards

The periodicity of the data on central government operations disseminated to the public is quarterly. This complies with the GDDS requirement to disseminate quarterly data on central government operations. Monthly reports on central government operations are only distributed for use within government.

The GDDS encourage the dissemination of annual general government data. Tanzania does not meet this encouraged extension due to a lack of data on local government operations. However, a project to compile quarterly data on local government operations is underway, and plans call for providing local government data by the end of 2002.

The periodicity of the data on central government debt is quarterly. The GDDS recommended practice is to publish annual debt data, but an extension to quarterly data is encouraged.

4.3 Consistency

4.3.1 Statistics are consistent within the dataset

Data on central government operations is not always consistent within the dataset. Major aggregates on central government operations such as total revenue, grants, recurrent expenditure and development expenditure equals the sum of their respective components. The net result of revenue and grants minus recurrent and development expenditure equals the deficit/surplus of checks issued. However, an adjustment line is added to the deficit in order to match the financing. This adjustment reflects in part checks recorded as revenue or expenditure in the accounting system of government, but not yet cleared by the banking system. This revenue and expenditure “float” is only reconciled with the bank statements on an annual basis.

Total net domestic and foreign financing is calculated from the flows recorded by the BoT in the accounts of central government. Domestic bank financing is derived from the monetary survey compiled by the BoT. Nonbank financing is calculated as a residual item after taking into account amounts received in respect of privatization, amortization payments on debt of parastatals assumed by the central government, and the change in arrears.

Data on financing and government outstanding debt stock differ in coverage and valuation and are therefore not reconcilable. Debt data include all debt incurred by the Union government and are recorded at face value. Data on central government financing cover only the data on Tanzania mainland and are recorded on a cash basis.

The datasets on central government operations disseminated in various publications are also inconsistent with regards to the analytical framework in which data are presented, resulting in different deficit/surplus definitions in use. (see Section 2.1.1). Differences in data used in various publications have also occurred and might be ascribed to the definitions, coverages and the revision policy in use (see Section 4.4.1).

4.3.2 Statistics are consistent or reconcilable over a reasonable period of time

All the parties involved in the compilation of statistics regard consistency in the methodology used in the compilation of government statistics as important. Although previously released data are not adjusted to account for changes due to methodological developments, breaks in the time series are indicated where necessary. Recent reforms in the compilation of source data enhanced the periodicity and timeliness of data without adversely affecting the consistency of datasets over a reasonable period of time.23

The quarterly Budget Execution Report and the Annual Budget Speech, presented to parliament by the Minister of Finance, outlined recent trends in the fiscal data in comparison with the budgeted estimates and historic trends. Explanations for significant deviations between estimates and actual outcomes are provided and any policy changes or economic developments that may have had an impact on the outcome are disclosed in these publications

4.3.3 Statistics are consistent or reconcilable with those obtained through other data sources and/or statistical frameworks

The dataset on central government operations serves as input into the national accounts. However, statistical techniques are employed to derive calendar year data from the fiscal year data. Data are also supplemented with information on local government finances obtained from various other sources. Statistics in the national accounts are therefore not reconcilable with those presented in the central government operations table. Quarterly data on central government operations are available since the start of fiscal 2001/02. These quarterly data will facilitate the compilation of central government operations data for calendar years and will eliminate the need to use statistical conversions. The envisaged expansion of the coverage fiscal data to include local government data will facilitate the compilation of data on general government operations.

The dataset for central government operations and the monetary survey cannot be reconciled due to differences in the coverage of government. The monetary survey includes all the central government accounts of the Union.

An unidentified amount of grants received from foreign donors is not recorded in the government accounts and the nature of grants received is also not identifiable in data on government operations. Government data on grants are, therefore, not consistent with the data contained in the balance of payments statistics.

4.4 Revision policy and practice

4.4.1 Revisions follow a regular, well-established, and transparent schedule

Revisions during the year are performed routinely and do not follow a specific revision schedule. Should any discrepancies from the actual outturn become apparent, the origin of the discrepancy is investigated and the data are revised in the relevant period. The revised data will be contained in the cumulative year-to-date totals of the next publication. Corrective measures are initiated to avoid a recurrence of the error or omission. Annual data are revised once the audited results for all government accounts are available. These audit reports are available within nine months after the end of the fiscal year. There is no specific release or formal announcement of the revisions, and revised data are not clearly identified as such. Furthermore, revisions of government finance data in various other publications are not effected simultaneously, resulting in discrepancies in data contained in these publications. There is also no formal process in place for disseminating all revisions to users of data and other institutions reproducing the data.

4.4.2 Preliminary data are clearly identified

Preliminary data are not clearly identified. All data are considered to be preliminary until such time as audited reports become available. Budget documentation makes a clear distinction between budget estimates and actual outcomes. However, the status of actual outcomes is not identified as being audited or preliminary. Various other publications containing data on government operations also lack this information.

4.4.3 Studies and analyses of revisions are made public

Studies and analysis of revisions are not formally undertaken and disseminated to the public. Reasons for revisions are investigated but are only made available to the public when individual users inquire about the reasons for the revisions. The public is not notified of revisions or the causes. However, the annual audited financial reports of government, which is the final revision of government data, is presented to parliament, discussed by the Public Accounts Committee and can be obtained from the government publishing office.

5. Accessibility

5.1 Data accessibility

5.1.1 Statistics are presented in a way that facilitates proper interpretation and meaningful comparisons (layout and clarity of text, tables, and charts)

Data on central government operations are disseminated in a way that facilitates proper interpretation and some meaningful comparisons. The layout and classification of the data follows budget guidelines rather than the international accepted standard (see Section 3.2.1) resulting in difficulties with international comparability. However, information presented in the Budget Execution Report can be compared with information contained in the annual budget of government. Statistics are thus presented in a way suitable for use in monitoring progress made with government finances and policy execution. However, the unavailability of comprehensive data on the general government can lead to an underestimation of government’s influence on the economy. Furthermore, different presentations of the central government operations table by different institutions also result in a lack of clarity on the financial operations of government.

Data on central government debt are presented in publications, using different currencies (domestic debt data are presented in Tanzania shilling whereas, the external debt of government is presented in U.S.dollars). The framework in which these reports are presented facilitates comparisons of specific debt categories, but does not present comprehensive and meaningful information on the total outstanding debt of the central government. Comprehensive reports on the domestic and foreign debt of government are only available for use within the government, which limits the usefulness of these data.

5.1.2 Dissemination media and formats are adequate

Dissemination media and format for central government operations are adequate. Data on central government operations are disseminated electronically in the Budget Execution Report posted on the official website of the Tanzanian government (http://www.tz.gov). The data are also published in the quarterly Bank of Tanzania Economic Bulletin its website (http://www.bot-tz.org). The Annual Economic Survey produced by the President’s Office Planning and Privatization, also includes tables containing data on central government operations. Data on central government operations are also submitted to the IMF for inclusion in the IFS.

The BoT disseminates data on central government debt in the quarterly Bank of Tanzania Economic Bulletin that is available in hard copy and electronically on its website (http://www.bot-tz.org). The format of debt data is not comprehensive and does not portray the total liabilities of government. Dissemination of a comprehensive debt report on the official website of the Tanzanian government is under consideration (http://www.tz.gov).

5.1.3 Statistics are released on a preannounced schedule

Annual data on government transactions are released at the time of the Minister of Finance’s Budget Speech to parliament. The date of the budget speech is pre-announced and would normally be presented to parliament in June of each year.

Although there is no pre-announced schedule for the release of the quarterly data or an official list of forthcoming publications available. The quarterly Budget Execution Report is released regularly with a timeliness of four weeks after the end of a reference period. The release date is dependent on the completion of:

  • compilation of monthly data to derive the quarterly statistical tables;

  • analyzing and reviewing the results in comparison with the expected outcomes; and

  • writing the review on the progress with budget execution.

5.1.4 Statistics are made available to all users at the same time

Data on government operations and debt are not released to all users at the same time. Quarterly data are made available to all users simultaneously and without preferential treatment for selected users. However, monthly data are distributed to interested users upon request. The fact that these reports are available upon request is not disseminated to the public.

5.1.5 Nonpublished (but nonconfidential) subaggregates are made available upon request

Nonconfidential data are made available upon request. However, the terms and conditions under which data could be provided are not disseminated to the public.

5.2 Metadata accessibility

5.2.1 Documentation on concepts, scope, classifications, basis of recording, data sources, and statistical techniques is available, and differences from internationally accepted standards, guidelines, or good practices are annotated

Metadata describing the central government operations and the central government debt are only disseminated on the DSBB of the IMF. These metadata contain brief descriptions of concepts and classifications in use, and broadly describe differences with international practice. Metadata revisions, in principle, follow the annual cycle mandated by the GDDS, but there have been delays.

5.2.2 Levels of detail are adapted to the needs of the intended audience

No detailed technical descriptions of the methodology and compilation practices other than that those disseminated on the DSSB are available to the public.

5.3 Assistance to users

5.3.1 Contact person for each subject field is publicized

The contact persons for central government operations and debt data are disseminated on the metadata pages of the DSBB. The official website of the Tanzanian government provide a general link for inquiries. All inquiries are then routed to the relevant departments. The Bank of Tanzania Economic Bulletin also provides information of a general contact at the BoT. However, direct contact persons are not identified with the specific data disseminated.

5.3.2 Catalogs of publications, documents, and other services, including information on any charges, are widely available

No catalogs of MOF publications, documents and other services, including information on any charges, are available.

Table 4.

Data Quality Assessment Framework—Summary of Results for Government Finance Statistics

(Compiling Agency: Ministry of Finance)

article image
article image
article image

V. Monetary Statistics

0. Prerequisites of quality

0.1 Legal and institutional environment

0.1.1 The responsibility for collecting, processing, and disseminating statistics is clearly specified

The Bank of Tanzania Act of 1995 (BoT Act) specifies that the Bank of Tanzania (BoT) has the responsibility for collecting statistics on banks and other financial institutions. Thus, according to Section 47 (1) of the BoT Act, “Every bank and financial institution shall furnish to the Bank, at such time and in such manner as may be prescribed by the Bank, such information as the Bank may require for the proper discharge of its functions and responsibilities.” The BoT’s Banks Supervision Department collects data (in duplicate) from all commercial banks, and its Domestic Accounts Department prepares central bank accounts. The BoT Act also specifies that the BoT has the responsibility for compiling and publishing commercial banks’ statistics. Thus, according to Section 47 (2), “The Bank may publish in whole or in part, at such times as it may decide, the information furnished pursuant to subsection (1) of this section, …” Based on these data, the Money and Banking Statistics Division situated within the Monetary and Financial Affairs Department compiles monetary statistics. No other agency than the BoT is involved in producing monetary statistics.

0.1.2 Data sharing and coordination among data producing agencies are adequate

Arrangements are in place to ensure the efficient and timely flow of source data within those BoT departments that are involved with the collection, compilation and dissemination of monetary statistics. Thus, once the Banks Supervision Department has collected data from banks, one copy of the data is transferred to the Monetary and Financial Affairs Department for the compilation and dissemination of monetary statistics. Regular meetings are held between these departments to ensure the use of proper concepts on a consistent basis.

0.1.3 Respondents’ data are to be kept confidential and used for statistical purposes only

Monetary statistics do not disclose data on individual banks. Thus, according to Section 47 (2) of the BoT Act, “The Bank may publish in whole or in part, at such times as it may decide, the information furnished pursuant to subsection (1) of this section, but no information shall be published which would disclose the financial affairs of any customer of a bank or financial institution unless his written consent has been first obtained by the Bank.” This information, however, can be published or disclosed in the public interest in a consolidated form. The BoT Act, which guarantees confidentiality of the respondents’ information, is available to the public at large. All BoT staff are required to abide by the appropriate rules and regulations to prevent disclosure of information of confidential nature, and the staff who are found to have violated these rules and regulations face summary dismissals. Moreover, access to individual data are restricted to the staff who require the information in the performance of their duties, and passwords have been built into the software programs to prevent an unauthorized access to individual data.

0.1.4 Statistical reporting is ensured through legal mandate and/or measures to encourage response

The BoT Act requires all banks to report data to the BoT, and supplementary regulations require that data be reported on a monthly basis and within a 15-day period from the reference date. Should a bank fail to report within the time-limit, the BoT has a right to impose penalties on the bank. Thus, according to Section 47 (4) of the BoT Act, “The Bank may impose on any bank or financial institution which fails to furnish information required under this section, a penalty charge of not less than T Sh 500,000 per day for every day in which the failure continues and the penalty charge may be recovered by deduction from any balance of or moneys owing to the bank or financial institution concerned or as a civil debt.” In actual practice, the penalty imposed is T Sh 1,000,000 per day, although the delays in data reporting are rare. Moreover, banks that expect to report late sometimes obtain prior approval for late reporting from the Banks Supervision Department. For these reasons, the imposition of fines on banks has been rare. Other mechanisms that BoT has put in place to ensure adequate reporting of data include providing a point of contact to respond to questions from data reporters in banks, as well as organizing seminars to provide training to them.

0.2 Resources

0.2.1 Staff, financial, and computing resources are commensurate with statistical programs

The current number of staff (seven) is adequate to perform the required duties related to the compilation of monetary statistics. At the same time, some of these staff are new recruits, who have not yet begun to contribute to work in a manner commensurate with their potential. The Monetary and Financial Affairs Department recruits staff on a highly competitive basis. Only candidates with master’s degrees in economics or related subjects, who also have Bachelor’s degrees at a minimum with an upper second-class, are regarded as eligible for the professional positions. The newly recruited staff are given a two-week exposure to work in the six departments of the Directorate of Economic Policy. All staff are encouraged to attend courses taught at the Bank of Tanzania Training Institute (Mwanza) which, however, offers courses on a rather irregular basis. The staff, selected on a competitive basis, are also allowed to take a leave of absence to pursue a higher-level degree, and are required to sign a bond that commits them to remain in employment with the BoT for a minimum period of five years. Moreover, the staff participates in seminars organized by regional and international organizations. Although no special effort is made, otherwise, to retain the trained staff, BoT’s overall incentives (house loan, training facilities) are regarded as adequate for minimizing the staff turnover. Computing resources, coupled with support services, as well as financial resources, provided for compiling monetary statistics are adequate to perform required tasks.

0.2.2 Measures to ensure efficient use of resources are implemented

Processes and procedures are in place at the BoT to ensure that resources are used efficiently. All departments of the BoT are required to prepare an annual budget, which begins with the process of discussion and agreement between management and staff on the work objectives, and the identification of specific development projects or goals for the year. The draft plans for each department are discussed and reviewed in a meeting of all Deputy Directors of the Directorate of Economic Policy, in a process designed to ensure that overlaps in activities and use of resources are eliminated. The revised plans are then reviewed by senior management and approved by the Board of Directors of the BoT. Progress against plans is monitored by the Director of Directorate of Economic Policy, and all Deputy Directors have access to the summary reports for all BoT directorates, which are prepared by the Domestic Accounts Department of the Directorate of Finance and posted on the BoT’s internal website.

Data procedures in place to minimize errors include the shared database network, Economic Statistics Information System (ESIS), on which each department of the bank posts the latest version of its data tables for use by other departments in their research, data models, etc.

0.3 Quality awareness

0.3.1 Processes are in place to focus on quality

Recognizing that its statistics must have the confidence of users, the BoT seeks to exercise quality control at various stages of data production and dissemination. Thus, in case of monetary statistics, the Banks Supervision Department verifies that data reporting practices followed by the banks are consistent with the regulations.

0.3.2 Processes are in place to monitor the quality of the collection, processing, and dissemination of statistics

The BoT undertakes data reviews to identify problems at the various stages of collection, processing and dissemination of data. For example, the Monetary and Financial Affairs Department recently implemented the ESIS, in order to educate data suppliers of the importance of high-quality data, which culminated in the establishment of the Data Processing Division within the Department. No user surveys, however, are conducted.

0.3.3 Processes are in place to deal with quality considerations, including tradeoffs within quality, and to guide planning for existing and emerging needs

The BoT recognizes that there are tradeoffs among the dimensions of quality (for example between timeliness and accuracy/reliability). Meetings are periodically convened within the BoT to review existing monetary statistics, as well as to identify any emerging data requirements. The Financial Programming Unit located in the Macroeconomic and Financial Programs Department checks intersectoral consistency of data.

1. Integrity

1.1 Professionalism

1.1.1 Statistics are compiled on an impartial basis

The statutory provisions under which the BoT compiles monetary statistics are adequate to support its independence in conducting the function in this area. Professionalism of the staff involved with monetary statistics is ensured through the recruitment process, as well as by the assignment structure. Thus, the recruitment process focuses on those who have graduate and post-graduate qualifications in economics or related subjects, and the assignment structure encourages the staff to prepare research papers and attend training courses and seminars on monetary policy and statistics (although, by and large, the staff find little time to engage in research since much of their time is devoted to statistical compilation). To safeguard professionalism, the BoT staff are required to refrain from participating in political activities, as well as from publishing views that might be considered detrimental to the interest of the BoT or the government of Tanzania.

1.1.2 Choices of sources and statistical techniques are informed solely by statistical considerations

Monetary statistics are compiled from the BoT’s and commercial banks’ accounting data. There is no other source from which data are needed. No use of statistical techniques is involved in monetary statistics compilation.

1.1.3 The appropriate statistical entity is entitled to comment on erroneous interpretation and misuse of statistics

The BoT staff seeks to prevent misinterpretation or misuse of statistics in the press and other media. If misinterpretation of monetary data is found in the newspapers, the BoT may prepares an appropriate reply, which is sent to the press through the Communications Department. However, such instances are rare.

1.2 Transparency

1.2.1 The terms and conditions under which statistics are collected, processed, and disseminated are available to the public

The terms and conditions under which monetary statistics are collected, compiled and disseminated, including the obligation to compile and disseminate the statistics and to safeguard the confidentiality of individual responses, are described in the BoT Act. This document is posted on the BoT website. Moreover, copies of the Act are distributed to all banks and other financial institutions, and are also made available to the public, free of charge, from the BoT headquarters.

1.2.2 Internal governmental access to statistics prior to their release is publicly identified

There is no formal obligation to allow internal government access to monetary data prior to their release. At the same time, monetary data are occasionally provided to the Minister of Finance, upon request, prior to release. In addition, as a member of the Monetary Policy Committee, the Commissioner of the Policy Analysis Department of the Ministry of Finance receives monetary data prior to release. Since there is no formal requirement to this effect, the fact of internal government access to data prior to their release is not publicly identified.

1.2.3 Products of statistical agencies/units are clearly identified as such

The BoT is not identified on a consistent basis as the source of monetary statistics that are published in its publications.

1.2.4 Advance notice is given of major changes in methodology, source data, and statistical techniques

No advance notices on major changes in methodology, source data and statistical changes are provided.

1.3 Ethical standards

1.3.1 Guidelines for staff behavior are in place and are well known to the staff

All BoT staff are required to abide by the appropriate regulations, which are published in Staff Regulations 2002 prepared, updated and circulated by the Human Resource Department. Copies of Staff regulations 2002 are available from the BoT library to all staff.

2. Methodological Soundness

2.1 Concepts and definitions

2.1.1 The overall structure in terms of concepts and definitions follows internationally accepted standards, guidelines, or good practices

The analytical framework used by the BoT in compiling the central bank survey and the depository corporations survey is currently being adapted to the guidelines outlined in the Monetary and Financial Statistics Manual (MFSM). The monetary survey is derived by consolidating the accounts of the central bank and commercial banks. Key aggregates identified in the monetary survey data are net bank credit to the central government; credit to the rest of the public sector (comprising state government and public nonfinancial corporations); credit to the private sector (comprising private enterprises, households and nonprofit organizations, as well as nonbank financial institutions); net foreign assets; currency with the public; demand deposits, time deposits and foreign currency deposits of residents. In the monetary survey data, the division between foreign and domestic positions is based on the residency criterion. However, since the migration to MFSM has not yet been completed, there are certain areas where further work is needed to ensure adherence to the internationally accepted standards. A project has been undertaken, in particular, to introduce the detailed sectorization scheme prescribed in MFSM.

2.2 Scope

2.2.1 The scope is broadly consistent with internationally accepted standards, guidelines, or good practices

The monetary survey currently covers the BoT and 20 commercial banks. This means that it includes (1) the BoT headquarters, (2) BoT’s three domestic branches in the mainland Tanzania, (3) the BoT Zanzibar, (4) headquarters of 20 commercial banks, and (5) resident branches of 20 commercial banks—including those in Zanzibar. It does not include rural banks and nonbank depository institutions.

2.3 Classification/sectorization

2.3.1 Classification/sectorization systems used are broadly consistent with internationally accepted standards, guidelines, or good practices

The BoT’s monetary statistics are in the process of migrating to the guidelines outlined in the MFSM. At present, the sectorization scheme adopted for monetary statistics appropriately distinguishes between nonresident and resident sectors, but the subdivision of the resident sector data does not provide sufficient information on the sectoral distribution of domestic credit. Thus, under the present sectorization scheme, the resident sector data are subdivided by (1) central government; (2) the rest of the public sector (comprising local government and public nonfinancial corporations); and (3) private sector (comprising private enterprises, households and nonprofit organizations, as well as other financial corporations). This means that a separate identification of local government, nonfinancial public enterprises, and other financial corporations is lacking. A project has been undertaken, therefore, to introduce the detailed sectorization scheme prescribed in the MFSM.

On accrued interest on financial assets and liabilities, the recommendation of the MFSM is that it should be added to the financial instrument and sectorized by the debtor sector (in case of assets) or by the creditor sector (in case of liabilities). The BoT, however, includes it in unclassified assets and liabilities.

The classification of financial instruments used for monetary statistics is in broad conformity with the guidelines outlined in the MFSM. Thus, the data separately identifies (1) Monetary gold and Special Drawing Rights; (2) Currency and deposits; (3) Securities other than shares; (4) Loans; but not (5) Shares and other equity; (6) Insurance technical reserves; (7) Financial derivatives; and (8) Other accounts receivable/payable.

In the BoT’s monetary statistics, the securities repurchase agreements (repos) are recorded as collateralized loans. These are not separately identified since these are treated as elements of unclassified assets and liabilities. Data on securities repurchase agreements (repos), however, are available upon request.

2.4 Basis for recording

2.4.1 Market prices are used to value flows and stocks.

The general recommendation of the MFSM is that the valuation of financial assets and liabilities should be carried out on the basis of market prices or market price equivalents. The only exception to this rule is that loans should be valued at their book values without adjustment for expected losses. These principles are broadly followed in the BoT’s monetary statistics.

Thus, (1) monetary gold is valued at market prices (on a daily basis and at the London closing price); (2) loans are valued at current book value without adjustment for expected loan losses arising from anticipated defaults by borrowers; and (3) valuation adjustments (including holding gains and losses on instruments) are presented in the revaluation account in the appropriate sectoral balance sheets or other aggregated presentations for the banking sector. Foreign securities, however, are recorded at face values despite fluctuations in their market prices.

The conversion of foreign-currency denominated assets and liabilities to domestic currency is carried out at prevailing market exchange rates. For this purpose, the BoT as well as the banks use the end-period average rate. While the BoT carries out conversion of its foreign-currency denominated positions on a daily basis, banks do it on a monthly basis.

2.4.2 Recording is done on an accrual basis

As recommended under the guidelines outlined in the MFSM, financial transactions are recorded on an accrual basis in the BoT’s monetary statistics, and are recorded simultaneously between parties.

2.4.3 Grossing/netting procedures are broadly consistent with internationally accepted standards, guidelines, or good practices

Assets and liabilities are presented on a gross or net basis in conformity with guidelines outlined in the MFSM. In particular, (1) claims on a particular transactor or group of transactions are not netted against the liabilities to that transactor or group of transactions; and (2) provisions against loans are shown on a gross basis on the liability side, and loans are included in domestic credit on a gross basis on the asset side. For the central bank subsector and commercial banks subsector, the data on financial assets and liabilities are aggregated into major categories—for example, loans classified by debtor sector and deposits classified by credit sector. In conformity with guidelines outlined in the MFSM, the monetary survey is obtained by canceling out all outstanding claims and liabilities between all commercial banks, while preserving the presentation of data on all stocks that are claims on (and liabilities to): financial corporations subsectors other than commercial banks subsector; other domestic sectors; and nonresidents.

3. Accuracy and reliability

3.1 Source data

3.1.1 Source data are collected from comprehensive data collection programs that take into account country-specific conditions

The source data for compiling the central bank survey are the balance sheet data prepared by the BoT’s Domestic Accounts Department on a daily basis. These balance sheet data are generated by a computerized ledger system based on complete recording. Source transactions are recorded and balance sheet data are generated from a predetermined classification of accounts.

The source data for commercial banks are the individual returns collected on a monthly basis. The incoming returns come in the form of a report form which is filled out by every bank and submitted in duplicate to the BoT’s Banks Supervision Department. This submission takes the form of either a hard copy or an electronic diskette (no on-line submission is available). The reporting format provides, by and large, sufficient detail to compile monetary statistics in conformity with guidelines outlined in the MFSM. The only exception is the area of sectorization where the scheme that is being used does not provide for all sectoral distinctions.

No use of supplementary data are involved. All source data needed for compiling monetary statistics are kept under periodic reviews to ensure that data collection remains comprehensive. The bank reporting system, for example, covers the entire resident banking sector.

3.1.2 Source data reasonably approximate the definitions, scope, classifications, valuation, and time of recording required

The monetary survey is based on complete data reporting by commercial banks. These reported data broadly approximate the required definitions, scope, classifications, valuation, and time of recording, although the sectorization scheme used in the report form is inadequate. No supplementary data sources such as financial market data and data reported from counter parties of commercial banks are used.

3.1.3 Source data are timely

The accounting records of the BoT and commercial banks, and the subsequent preparation of balance sheet data, are sufficiently timely to allow for compilation of monetary statistics. The BoT balance sheet data are compiled on a daily basis with a lag of only one working day. Commercial banks report data on a monthly basis with a lag of 15 days. This enables the monetary survey to be compiled within three weeks.

3.2 Statistical techniques

3.2.1 Data compilation employs sound statistical techniques

Data compilation procedures minimize processing errors such as coding, editing, and tabulation errors. The BoT’s report form is designed to encourage computer processing. The computer program processes source data promptly.

3.2.2 Other statistical procedures (e.g., data adjustments and transformations, and statistical analysis) employ sound statistical techniques

No data adjustments and transformations are involved.

3.3 Assessment and validation of source data

3.3.1 Source data—including censuses, sample surveys and administrative records—are routinely assessed, e.g., for coverage, sample error, response error, and nonsampling error; the results of the assessments are monitored and made available to guide planning

The BoT Governor’s Office receives questions on monetary data from several sources, for example, the government of Tanzania, Members of Parliament, financial reporters and the press at large, as well as the research community. These questions are assigned to the Monetary and Financial Affairs Department, which prepares responses. In the process, the Department uses the opportunity to monitor the accuracy of the BoT’s monetary statistics. In addition, the BoT cross-checks monetary statistics against annual balance sheets of banks after they have been audited. Appropriate measures are undertaken when discrepancies are found.

3.4 Assessment and validation of intermediate data and statistical outputs

3.4.1 Main intermediate data are validated against other information where applicable

No secondary data sources, such as data on financial market operations and sample surveys, are used to check the accuracy of the commercial banks’ balance sheets.

3.4.2 Statistical discrepancies in intermediate data are assessed and investigated

The Monetary and Financial Affairs Department identifies statistical discrepancies and, in collaboration with the Banking Department, investigates their sources.

3.4.3 Statistical discrepancies and other potential indicators of problems in statistical outputs are investigated

Procedures have been established for collaboration between the Monetary and Financial Affairs Department and several other departments—namely, Banking Department, Banks Supervision Department, Domestic Accounts Department and Domestic Markets Department—for resolving problems posed by large or unexplained fluctuations in monetary or credit aggregates.

3.5 Revision studies

3.5.1 Studies and analyses of revisions are carried out routinely and used to inform statistical processes

Monetary survey data are revised. These revisions, however, rarely affect major statistical aggregates since they usually focus on unclassified items, and tend to be of small magnitude. In light of this, no revision studies are undertaken.

4. Serviceability

4.1 Relevance

4.1.1 The relevance and practical utility of existing statistics in meeting users’ needs are monitored

The BoT undertakes certain actions to ensure that monetary statistics address issues of concern to users. A process of consultation takes place periodically with other departments within the BoT. However, neither working groups involving users are established, nor are user surveys organized to ascertain users’ needs.

4.2 Timeliness and periodicity

4.2.1 Timeliness follows dissemination standards

The BoT uses the Monthly Economic Review as its primary vehicle for monetary data dissemination. Although the normal timeliness for the publication is one month, occasional slippages do occur, causing long and variable delays in the dissemination. Thus, the central bank data, as well as the commercial bank data, are disseminated within two to five months after the end-of-the-reference month. This degree of timeliness is not fully consistent with the GDDS, which requires that (1) the monthly data on the central bank should be disseminated within one to two months after the end of the reference month; and (2) the monthly data on commercial banks should be disseminated within two to three months after the end of the reference month.

4.2.2 Periodicity follows dissemination standards

The periodicity of the central bank data is daily, while that of the commercial bank data is monthly, compared with the monthly periodicity required under the GDDS for both the central bank data and the commercial bank data.

4.3 Consistency

4.3.1 Statistics are consistent within the dataset

The BoT’s monetary statistics are internally broadly consistent. Only small discrepancies exist between the central banks’ and commercial banks’ data on interbank positions as recorded in monetary statistics. The BoT has no claims on commercial banks, and the commercial banks have no liabilities to the BoT, while on the banks’ deposits with the BoT only a small discrepancy exists.

4.3.2 Statistics are consistent or reconcilable over a reasonable period of time

Broadly consistent time series data are available from December 1992. Although—whenever changes in methodology are introduced—historical series are reconstructed for only two years, revisions have tended to be small, with no significant impact on the intertemporal consistency of the data series. Methodological notes identify and explain the main breaks and discontinuities in the data series, their causes, as well as adjustments made to maintain consistency over time. For more recent dates, any unusual changes in economic trends are explained in the analytical text included in the publication and in the database accessible to users.

4.3.3 Statistics are consistent or reconcilable with those obtained through other data sources and/or statistical frameworks

The BoT’s monetary statistics are broadly consistent with other statistical systems, such as balance of payments statistics, the international investment position, and government finance statistics. The net foreign assets in monetary statistics are consistent with the corresponding measure derivable from the international investment position. Also, transactions in net foreign assets in monetary statistics are broadly consistent with the corresponding measure derivable from balance of payments statistics. Some differences exist between the two areas of statistics, but they are attributable to differences in valuation. Finally, in government finance statistics, BoT’s records on the government position with the banking sector are used.

4.4 Revision policy and practice

4.4.1 Revisions follow a regular, well-established, and transparent schedule

The BoT’s monetary statistics that are disseminated initially are provisional. The provisional monetary survey data are revised. These revisions, however, rarely affect major statistical aggregates since they usually focus on unclassified items, and tend to be of small magnitude. The revision policy follows a predictable pattern although, in view of the relative insignificance of revisions, no attempt is made to inform the users of statistics of the revision cycle.

4.4.2 Preliminary data are clearly identified

Preliminary data are not consistently identified in all BoT publications, and the users are not always alerted that the initially published data are preliminary and subject to revision. The revised data are disseminated with the same level of detail as the previously published data.

4.4.3 Studies and analyses of revisions are made public

Although monetary survey data are revised, revisions rarely affect major statistical aggregates. They usually focus on unclassified items, and tend to be of small magnitude. In light of this, no revision studies are undertaken.

5. Accessibility

5.1 Data accessibility

5.1.1 Statistics are presented in a way that facilitates proper interpretation and meaningful comparisons (layout and clarity of text, tables, and charts)

The BoT publishes monetary statistics in a clear manner, and charts and tables are disseminated with the data to facilitate the analysis. It publishes datasets with various levels of detail (disaggregation), and includes analysis of current-period developments in each publication. Long-term monetary data series, starting December 1992, are maintained and, upon request, are made available to the users.

5.1.2 Dissemination media and formats are adequate

The BoT releases data through a variety of publications. These are Monthly Economic Review, Quarterly Economic Bulletin, and Annual Economic and Operations Report. All the publications are available on the BoT’s Internet website (http://www.bot-tz.org).

5.1.3 Statistics are released on a pre-announced schedule

The BoT does not disseminate an advance release calendar to provide an advance notice of the release dates for monetary data.

5.1.4 Statistics are made available to all users at the same time

Once the Governor has approved the Monthly Economic Review, which acts as the primary vehicle for the dissemination of monetary statistics, it is sent to the publishers. In the interval between the Governor’s approval and the actual publication, copies are given to all major newspapers in Tanzania, as well as to selected embassies.

5.1.5 Nonpublished (but nonconfidential) subaggregates are made available upon request

Subaggregates, which are not published, are made available upon request approved by the Governor. At the same time, requests for unpublished subaggregates have been rare.

5.2 Metadata accessibility

5.2.1 Documentation on concepts, scope, classifications, basis of recording, data sources, and statistical techniques is available, and differences from internationally accepted standards, guidelines, or good practices are annotated

All BoT publications in which monetary statistics are published contain some notes and/or footnotes providing users with information about data and methodology used to compile data. The only metadata on monetary statistics, however, are those available from the Dissemination Standards Bulletin Board (DSBB).

5.2.2 Levels of detail are adapted to the needs of the intended audience

Metadata on monetary statistics are available from the DSBB, but not from national publications. No attempt is made to develop metadata of the various levels of detail adapted to the needs of the different categories of audience.

5.3 Assistance to users

5.3.1 Contact person for each subject field is publicized

A contact person is identified in the GDDS metadata, although not in the BoT publications. When users request assistance in the form of clarification and explanation of data, the request is channeled to the Monetary and Financial Affairs Department, which prepares an appropriate response.

5.3.2 Catalogs of publications, documents, and other services, including information on any charges, are widely available

A Catalog of publications is available on the BoT’s Internet website (http://www.bot-tz.org). All BoT publications are free. The BoT provides assistance needed to place orders for its publications.

Table 5.

Data Quality Assessment Framework—Summary of Results for Monetary Statistics

(Compiling Agency: )

article image
article image

VI. Balance of Payments Statistics

0. Prerequisites of quality

0.1 Legal and institutional environment

0.1.1 The responsibility for collecting, processing, and disseminating statistics is clearly specified

The Bank of Tanzania (BoT) is the agency responsible for compiling the balance of payments statistics for the United Republic of Tanzania. Although Section 47 (1) of the Bank of Tanzania Act (BoT Act) of 1995 specifies that the BoT has the responsibility for collecting statistics on banks and other financial institutions, the Act does not specify that BoT is also responsible for collecting, processing, and disseminating data on balance of payments.

0.1.2 Data sharing and coordination among data producing agencies are adequate

The system of data sharing among the compiling agencies is generally adequate. However, there is no formal system of coordination between the compiling agencies, or even between the relevant departments of the BoT.

In those instances when the government agencies have coordinated on a specific project, such as the joint travel surveys and surveys of private sector capital flows and stocks, the results have been positive.

0.1.3 Respondents’ data are to be kept confidential and used for statistical purposes only

Respondents’ data are kept confidential and are used for statistical purposes only. The confidentiality of individual respondents’ data is guaranteed, and that guarantee is widely known. According to Section 47 (2) of the BoT Act, “… no information shall be published which would disclose the financial affairs of any customer of a bank or financial institution unless his written consent has been first obtained by the Bank.” The BoT Act, which guarantees confidentiality of the respondents’ information, is available to the general public, and the forms and letters requesting information from entities inform respondents that the information they provide will be kept confidential and used only for the purpose of producing statistics.

Procedures are in place to prevent disclosure of individual data. All BoT staff are required to abide by the appropriate rules and regulations to prevent disclosure of confidential information, and staff who violate those rules and regulations face summary dismissal. Moreover, access to the individual data is restricted to those staff who require the information to perform their duties, and passwords have been built into the software programs to prevent unauthorized access to the data of individual entities.

0.1.4 Statistical reporting is ensured through legal mandate and/or measures to encourage response

The BoT Act requires all banks to report data to the BoT, and supplementary regulations require that the data be reported twice a month and within a 15-day period after the end of the reference period. The regulations also require that the bureau de change report data to the BoT once a month with a timeliness of within 15 days after the end of the reference month. Section 47 (4) of the BoT Act gives the BoT the right to impose penalties should a bank or other financial institution fail to report within the time-limit: “The Bank may impose on any bank or financial institution which fails to furnish information required under this section, a penalty charge of not less than five hundred thousand shillings per day for every day in which the failure continues and the penalty charge may be recovered by deduction from any balance of or moneys owing to the bank or financial institution concerned or as a civil debt.” Other mechanisms that the BoT has put in place to ensure adequate reporting of data include providing a point of contact to respond to questions from data reporters in banks, as well as organizing seminars to provide training for those data reporters.

However, the BoT Act does not give the BoT clear authority to collect data from entities other than banks and other financial institutions. This lack of a clear legal mandate hinders the ability of the division of the BoT responsible for compiling the balance of payments data24 to collect data from entities that are not financial institutions. (See Section 3.1.1 below.) The two successful multi-agency surveys referred to in Section 0.1.2 above have been able to use the authority of the Tanzania Investment Centre (TIC) and that of the National Bureau of Statistics (NBS) to ensure responses from the nonfinancial entities.25

0.2 Resources

0.2.1 Staff, financial, and computing resources are commensurate with statistical programs

The current number of staff of the balance of payments division of the International Economic Department (IED) is considered by the mission to be adequate to perform the required duties related to the compilation of balance of payments statistics, and the number has been expanded in recent years from four to eight.

The IED recruits staff on a highly competitive basis. Only candidates with Master’s degrees in economics and related subjects are regarded as eligible for the professional positions. As with all staff hired by the Directorate of Economic Policy (DEP), newly recruited staff are given a two-week orientation during which they work in the other five departments of the DEP, that is, the Monetary and Financial Affairs Department (MFAD), the Debt Management Department (DMD), the Macroeconomic and Financial Programs Department (MFPD), the Trade, Finance and Investment Policies Department, and the Real Sector Department.

Two members of the staff have attended IMF balance of payments courses in Washington. Staff of the Balance of Payments Division have also attended regional workshops, such as the 2001 workshop on Regional Cooperation and Reconciliation in Balance of Payments Statistics sponsored by the Centre for Research into Economics and Finance in Southern Africa (CREFSA) of the London School of Economics. (See also Section 3.2.1 below.) Although all staff are encouraged to attend courses taught at the Bank of Tanzania Training Institute (Mwanza), the institute offers courses only on an irregular basis, and does not have a course specifically covering balance of payments statistics. Selected staff are also permitted to take paid leave of absence to pursue a higher level degree, for which the BoT pays the tuition fees, and are required to sign a bond that commits them to remain in employment with the BoT for a minimum period of five years following the completion of their degree. Retention of staff is not considered by the mission to be a problem.

The computer resources are considered by the mission to be adequate and computerized systems for the data compilation are in place. The financial resources are allocated as part of the annual budget review process and are considered by the mission to be adequate.

0.2.2 Measures to ensure efficient use of resources are implemented

Processes and procedures are in place at the BoT to ensure that resources are used efficiently. All departments of the BoT are required to prepare an annual budget, which begins with the process of discussion and agreement between management and staff on the work objectives, and the identification of specific development projects or goals for the year. The draft plans for each department are discussed and reviewed in a meeting of all Deputy Directors of the DEP, in a process designed to ensure that overlaps in activities and use of resources are eliminated. The revised plans are then reviewed by senior management and approved by the Board of Directors of the BoT. Progress against plans is monitored by the Director of DEP, and all Deputy Directors have access to the summary reports for all directorates of the bank, which are prepared by the Domestic Accounts Department of the Directorate of Finance and posted on the BoT’s internal website.

Data procedures in place to minimize errors include the shared database network, Economic Statistics Information System (ESIS), on which each department of the bank posts the latest version of its data tables for use by other departments in their research, data models, etc.

0.3 Quality awareness

0.3.1 Processes are in place to focus on quality

There is recognition throughout the BoT that its statistics must have the confidence of its users, and one of the specific objectives of the DEP is improvement in the quality of their data. The ESIS shared database facilitates the cross-checking of related data.

0.3.2 Processes are in place to monitor the quality of the collection, processing, and dissemination of statistics

Quality control processes are in place at every stage of data production and dissemination. The IED monitors the data reporting practices followed by the banks to determine whether they are consistent with the guidelines, and is aware that misclassification continues to be a problem. All levels of the IED and DEP participate in the review of data prior to publication. Despite this, incorrect data have been published on a number of occasions, and inconsistencies within the tables and between the various datasets have not been addressed. (See Sections 2.1.1, and 3.4.2 below.)

No processes are in place to obtain feedback from users of the balance of payments statistics on data quality issues.

0.3.3 Processes are in place to deal with quality considerations, including tradeoffs within quality, and to guide planning for existing and emerging needs

The budget and work objective plans prepared each year by all departments of the BoT address issues affecting the quality of the data, such as, in the case of the IED, plans to improve data sources, or the need for training seminars for the data reporters at financial institutions in order to improve the classification of the data obtained from the ITRS records. In addition, the Financial Programming Unit within the MFPD has been established with the specific responsibility of checking the cross-sectoral consistency of macroeconomic datasets.

1. Integrity

1.1 Professionalism

1.1.1 Statistics are compiled on an impartial basis

The terms and conditions under which the balance of payments statistics are produced promote professional independence. Although the BoT Act does not specifically guarantee the professional independence of the organization, the BoT is not required to submit the balance of payments statistics to any civil servants or ministers for approval prior to their dissemination, and the decision to release the data is made entirely within the BoT.

Professionalism of the staff is assisted by the recruitment process, which focuses on recruiting staff who have graduate and post-graduate qualifications in economics or related subjects. To safeguard professionalism, the BoT staff are required to refrain from participating in political activities, as well as from publishing views that could be considered detrimental to the interests of the BoT or the government of Tanzania.

1.1.2 Choices of sources and statistical techniques are informed solely by statistical considerations

Decisions about the choice of data sources and statistical techniques are made solely on the basis of statistical techniques.

1.1.3 The appropriate statistical entity is entitled to comment on erroneous interpretation and misuse of statistics

The IED staff seek to prevent misinterpretation or misuse of balance of payments statistics in the press and other media by monitoring the financial press on a daily basis. If a misinterpretation of balance of payments data were to be found in the newspapers, the IED would prepare an appropriate reply, which would be sent to the press through the Communications Department of the Directorate of the Governor’s Office. However, the situation has reportedly not arisen to date.

1.2 Transparency

1.2.1 The terms and conditions under which statistics are collected, processed, and disseminated are available to the public

The BoT Act is posted on the BoT website. Moreover, copies of the Act are distributed to all banks and other financial institutions, and are also made available to the public, free of charge, from the BoT headquarters. The metadata posted on the IMF’s DSBB also describe the terms and conditions under which the statistics are collected, and disseminated to the public.

1.2.2 Internal governmental access to statistics prior to their release is publicly identified

There is no formal obligation to allow internal government access to the balance of payments statistics prior to their release. However, although the GDDS metadata on the IMF’s DSBB state that there is no internal government access to the data prior to their release to the public, and that the data are released simultaneously to the ministries and the public, the data are in fact provided to government ministries and agencies, such as the NBS, the MOF, and the POPP, about two weeks prior to their release to the general public. The public is not informed of this internal government access to the data prior to their release to the public.

1.2.3 Products of statistical agencies/units are clearly identified as such

The IED is clearly identified in the footnotes to the balance of payments data tables as the source of balance of payments statistics that are published in the BoT publications. However, while the BoT is listed as being the source of the balance of payments statistics disseminated in the annual Economic Survey published by the POPP, the IED is not identified.

1.2.4 Advance notice is given of major changes in methodology, source data, and statistical techniques

No advance notice is given of major changes in methodology, source data and statistical changes. However, notice of the major changes to the methodology, such as the move to the Balance of Payments Manual (BPM5) methodology in 1996, and of major changes to the data sources, such as the introduction of the new international transactions reporting system (ITRS) in 1998, were given in the BoT publications at the time the changes were introduced.

1.3 Ethical standards

1.3.1 Guidelines for staff behavior are in place and are well known to the staff

A clear set of ethical standards has been prepared, and regulations reflecting those standards are published in Staff Regulations 2002, which is prepared, updated and circulated by the Personnel Management Department of the Directorate of Personnel and Administration. Staff Regulations 2002 is posted on the BoT internal website, and copies are also available to all staff from the BoT library. All BoT staff are required to abide by these regulations, and new staff are made aware of the regulations when they join the BoT.

2. Methodological soundness

2.1 Concepts and definitions

2.1.1 The overall structure in terms of concepts and definitions follows internationally accepted standards, guidelines, or good practices

The overall structure of the balance of payments statistics is broadly in line with the fifth edition of the BPM5. However, the concepts and definitions set out in the BPM5 are not presently followed in several areas:

  • Treatment of interest and amortization arrears. Contrary to the international standards, the debit data on Other Investment income (interest) and loan liabilities that are included in the published aggregates are not equal to the sum of (i) actual interest and loan repayments (amortization payments) made, and (ii) arrears on interest and loan repayments. Instead the amounts included are the interest and loan repayments scheduled to have been paid during the period, based on projections for an extended period, such as a year, provided by the DMD. Although the DMD regularly provides updated information to the IED on the scheduled interest and amortization payments, and the actual payments, until 2002 the updated information on scheduled payments has not been entered into the balance of payments compilation system, reportedly because of a misunderstanding about the nature of the long-term projections. As a result, the debit amounts for Income and Other Investment have been understated in the published aggregate data. The balance of payments statistics for 2002 will reportedly use the correct methodology.

  • Reserve assets. Contrary to the international standards, the reserve assets of Tanzania include the BoT’s holdings of nonconvertible currencies. However, these holdings are reportedly insignificant.

2.2 Scope

2.2.1 The scope is broadly consistent with internationally accepted standards, guidelines, or good practices

In principle, the balance of payments statistics cover the cross-border transactions of the whole territory of the United Republic of Tanzania, including both the mainland and Zanzibar, and its embassies abroad. However, the coverage of the data is incomplete in the following areas:

  • Goods. The authorities are aware that unrecorded trade (primarily smuggling), involving, in particular, exports of coffee to Uganda, and cloves and tanzanite to Kenya, and imports of sugar from Malawi and Zambia, are not included in the balance of payments data.

  • Services. Transactions arising from the construction of the new United States embassy (a nonresident entity) by resident construction enterprises are not included. These are likely to be relatively significant in the next few years. Other transactions of foreign embassies and international organizations (which are nonresidents even though they are located in Tanzania) with Tanzanian residents are also not included.

  • Income. Data on foreign direct investment (FDI) reinvested earnings and portfolio investment income are not covered. The results of a 2000 survey of private sector foreign capital flows and stocks indicated that FDI dividends and interest payments are being significantly under-recorded in the balance of payments statistics.

  • Current transfers. The data do not include workers’ remittances. However, it is possible that those remittances that are made through the banking system are included in the data, but misclassified either under services, or as compensation of employees under income.

  • Capital transfers. The counter-entries for migrant transfers in kind included in the data on exports and imports are not included in the data on capital transfers. No data on migrant transfers in cash are included under capital transfers; it is possible that those transfers in cash that are transacted through the banking sector are included, but misclassified.

  • Acquisition/disposal of nonproduced, nonfinancial assets. No data on these transactions have been reported; it is possible that those transactions made through the banking system are included, but misclassified.

  • Portfolio investment. The balance of payments statistics do not include data on portfolio investment, although the results of the 2000 survey of private sector foreign capital flows and stocks indicate that inward portfolio investment transactions do exist.

On the other hand, there are elements of “over-coverage” in the data, such as instances when a nonresident donor sends capital transfers directly to the nonresident parent company of a resident construction enterprise instead of to the government of Tanzania, and the parent company then forwards those funds to the resident construction company. These amounts are recorded both in the ITRS records as construction services credits, and in the data provided by the donors on capital transfers—in other words, the transactions are being double-counted in the balance of payments statistics. It is expected that this double-counting will be eliminated from the 2002 data, and that the historical data will be revised by the end-of-2002.

2.3 Classification/sectorization

2.3.1 Classification/sectorization systems used are broadly consistent with internationally accepted standards, guidelines, or good practices

The classification of the balance of payments for Tanzania is broadly consistent with the main aggregates of the analytical presentation of BPM5, and consists of the Current Account, the Capital Account, the Financial Account, net errors and omissions, the overall balance, reserve assets, and exceptional financing. However, only the debit and credit totals for the main aggregates are disseminated (goods, services, income, current transfers, capital transfers, direct investment, and other investment), and single lines are disseminated for portfolio investment and reserve assets.

Although not disseminated in the main balance of payments statistics, cumulative monthly debit and credit data for the current year and the same period in the previous year have recently begun to be disseminated in the BoT’s Monthly Economic Review. Those data cover all 11 categories of services, as well as compensation of employees, and investment income, with an “of which” line for interest payments in the case of the latter.

Data for the following items are misclassified:

  • Technical assistance. The counter-part entry for the technical assistance element of current transfers is classified under services as government services n.i.e., rather than under income as compensation of employees.

  • Other investment income (interest) and loan assets and liabilities. The data for the general government sector include (i) the direct debt liabilities and (ii) the contingent debt liabilities (government-guaranteed debt), of parastatals (public enterprises). Both these items should be classified under other sectors, rather than general government.

  • Current transfer debits. The amounts shown in the ITRS records for current transfer debits (all of which are other sector debits) are extremely high (figures of US$123.3 million for 1999, US$76.8 million for 2000, and US$73.8 million for 2001 were published in the first quarter 2002 edition of the quarterly Bank of Tanzania Economic Bulletin (Economic Bulletin)—amounts that represent 76 percent of the current transfer credits for other sectors in 1999, 52 percent in 2000, and 69 percent in 2001). The IED staff are aware that the ITRS data contain numerous misclassifications arising from the reporting banks’ confusion regarding the difference between the balance of payments concept of “transfers” and a normal “bank transfer” that may be made for many different purposes. However, no assessment has been made of the likely extent of the misclassifications of current account debits in the ITRS records, and no adjustments are made to the source data.

In addition, transactions involving the import of electricity from Uganda and Zambia may possibly be misclassified under services, rather than as goods. However, beginning with the data for 2002, it is expected that these imports will be classified under goods, and that the data back to 2000 will be revised to reflect this change in classification.

2.4 Basis for recording

2.4.1 Market prices are used to value flows and stocks

Goods are valued by the Customs Department of the Tanzania Revenue Authority (TRA) at the price actually paid, or payable. The exports data are recorded by Customs on an f.o.b., basis and imports data are recorded on a c.i.f., basis, and the latter are subsequently converted by the IED to the f.o.b., basis required in the balance of payments methodology. The imports data are converted daily by the TRA to the unit of account using the exchange rate prevailing on the day of the transactions. Other transactions are reported in aggregate to the IED by currency and are converted by the IED to the unit of account, using the average exchange rates prevailing during the reporting periods, which vary from two weeks to one year, depending on the data source.

2.4.2 Recording is done on an accrual basis

The balance of payments statistics are recorded on a cash basis, and no adjustments are made to convert the data to an accrual basis.

2.4.3 Grossing/netting procedures are broadly consistent with internationally accepted standards, guidelines, or good practices

The Current Account and Capital Account transactions are generally recorded on a gross basis and those of the Financial Account on a net basis, in accordance with BPM5. However, the following Current Account transactions are recorded on a net, rather than a gross, basis: (i) those income transactions that are subject to withholding taxes; (ii) those intercompany transactions that are settled on a net basis; and (iii) data on communications services that are settled on a net basis, such as the settlements of the Tanzania Postal Corporation and the Tanzanian Telecommunications Company Limited with their international counterparties.

3. Accuracy and reliability

3.1 Source data

3.1.1 Source data are collected from comprehensive data collection programs that take into account country-specific conditions

The primary data sources used to compile the balance of payments statistics for Tanzania are broadly sufficient to compile the balance of payments statistics. The data sources are:

  • Goods. The primary data source is monthly foreign trade data provided by the TRA, which are based on the Customs records. Beginning from 2002, the TRA data are supplemented by information provided to the IED by the Tanzanian Electric Supply Company on imports of electricity from Uganda and Zambia.

Services

  • Transportation. The data source is the ITRS records. The data on freight are estimated as 90 percent of 13 percent of the total c.i.f., value of imports. However, this ratio will be changed by the end-of-2002. (See Section 3.3.1 below.) Although information on cross-border transactions is also obtained from transport enterprises, this is not used as a data source, as the coverage is incomplete—only the Tanzania-Zambia Railways Authority (TAZARA) and the Tanzania Harbours Authority (THA) respond to requests for information, and to date road and air transport enterprises have not provided the requested information, despite reminders.

  • Travel. The data source is the ITRS records. However, these are considered to possibly overstate the level of travel debits and credits, largely because of misclassification of foreign exchange receipts and payments by bureau de change. (For example, in 2001 travel represented 60 percent of total services credits and 42 percent of total services debits.) The data source is expected to be changed to survey data during 2003. (See Section 3.3.1 below.)

  • Communications services. The data source is the ITRS records. Although monthly information on international transactions is also requested from communications enterprises, this information is not used as a data source as only the state-owned Tanzanian Telecommunications Company Limited and Tanzanian Postal Corporation respond, and the five privately-owned mobile telephone companies do not provide the requested information. The data are, however, used as a cross-check for the ITRS transactions of the Tanzanian Telecommunications Company Limited and Tanzanian Postal Corporation.

  • Insurance services. Insurance services for imports are estimated as being 10 percent of 13 percent of the total c.i.f., value of imports. However, this ratio will be changed by the end-of-2002. (See Section 3.3.1 below.) The data source for other insurance service debits and credits is quarterly information provided by the Insurance Commissioner.

  • Government services. The primary data source is the ITRS records. However, for the credit data, only 30 percent of the total amount reported in the ITRS records as being government services is included in the balance of payments, as the total amount is very high, and is known to include receipts that should have been classified by the reporting banks as current transfers, rather than government services. The decision to exclude 70 percent of the amount shown in the ITRS records for government services credits was based on discussions with the commercial banks on the extent of the misclassifications. The sources for the debit data are (i) data provided by the Foreign Accounts Department of the Directorate of Finance of the BoT, which handles all payments for government agencies, (ii) the ITRS records for the transactions of embassies, and (iii) an amount equal to 30 percent of current transfer credits for services, which is treated as a payment by government for technical assistance. (As indicated in Section 2.3 above, this counter-entry should be classified as a payment for compensation of employees, rather than government services).

  • Other services. The data source for all other categories of services is the ITRS records.

Income

  • Compensation of employees. The data source is the ITRS records.

  • Investment income. The data source for credits and all debits except loan interest payments is the ITRS records. The data source for the loan interest payments, both of general government and other sectors, is a monthly report provided by the DMD showing the actual interest payments for the previous month and the scheduled interest payments for the current month. Although the DMD keeps records of interest arrears (stock data), the DMD arrears figure is not used for cross-checking purposes. The amounts included in the balance of payments statistics for arrears are calculated by IED as being the difference between scheduled payments and actual payments for a given month. (See Section 2.1.1 above.) Beginning in 2002, the results of the survey of private sector foreign capital flows and stocks will be used as a data source for revising the balance of payments statistics on FDI and portfolio investment income for the applicable years—for example, the 2000 survey provided data for 1999, and the survey planned for late 2002 will provide data for 2000 and 2001.

  • Current transfers. Prior to 2001, the data source for credits was aggregate information provided by the United Nations Development Program (UNDP) based on data supplied by the individual donor agencies, both for transfers to general government, and transfers to other sectors. Beginning in 2001 the responsibility for compiling the aggregate information moved from the UNDP to the MOF. Because of problems with the MOF data, the 2001 data are estimates. (See Section 3.1.2 below.) The totals are based partly on the MOF data, and partly on UNDP projections for the second half of the 2001 calendar year. The amounts apportioned to current and capital transfers, and the amounts of current transfers apportioned to goods and services are based on the proportions seen in the data for earlier years. The amount of current transfers apportioned to budget support is based on information compiled by the Aid Coordination Division of the IED. The data source for debits is the ITRS records.

  • Capital transfers. The data sources are the same as for the current transfers with the exception of the data on debt forgiveness, which are provided by the DMD.

  • Direct investment financial transactions. The data source is the investment approvals of the Tanzania Investment Centre (TIC), adjusted for the implementation period and the degree of realization. Beginning in 2002, the results of the survey of private sector foreign capital flows and stocks will be used as a data source for revising the balance of payments statistics for the applicable years.

  • Portfolio investment financial transactions. The data source is the ITRS records. Beginning in 2002, the results of the survey of private sector foreign capital flows will be used as a data source for revising the balance of payments statistics for the applicable years.

  • Other investment financial transactions. The data source for trade credits and currency and deposits is the monetary survey data for the banking sector and the ITRS records for other sectors. The data source for loan repayments (amortization payments), both for general government, and for other sectors, is a monthly report provided by the DMD that shows the actual loan repayments for the previous month and the scheduled loan repayments for the current month. Although the DMD keeps records of amortization arrears stocks, the DMD arrears figures are not used for cross-checking purposes. The amounts included in the balance of payments statistics for arrears are calculated by IED as being the difference between scheduled payments and actual payments for a given month. (See Section 2.1.1 above.)

  • Reserve assets. The data source is the monthly assets and liabilities statement of the BoT compiled by the MFAD and posted on the internal BoT network, the ESIS, to which all relevant departments have access.

The ITRS covers all 20 commercial banks and all bureaux de change, and includes the transactions both of the banks’ customers and those made on the banks on their own behalf. There is no value threshold for the ITRS, and the system does not permit the netting of balance of payments transactions, as separate reports on payments and receipts are required. Nevertheless, there are gaps in the data sources, in particular, data for: (i) unrecorded trade, (i) investment income—FDI income (reinvested earnings, dividends and interest) and portfolio investment income, (iii) workers’ remittances, (iv) migrants’ transfers, (v) direct investment financial transactions—equity capital and other capital transactions, and (vi) portfolio investment asset and liability transactions.

External sector developments in the press and other media are monitored on a daily basis, and are usually included in the BoT’s weekly internal newsletter. Staff are expected to be aware of all external developments that could affect the balance of payments, such as changes in commodity prices.

3.1.2 Source data reasonably approximate the definitions, scope, classifications, valuation, and time of recording required

Many of the problems regarding the scope, classifications, and basis of recording identified in Sections 2.2, 2.3, and 2.4 above directly reflect the limitations of the source data. As indicated in Section 3.1.1 above, adjustments are made to convert the data on imports of goods reported by the TRA on a c.i.f. basis, to an f.o.b. basis. However, no adjustments are made for unrecorded trade, or communications services and income transactions reported on a net basis.

Misclassification by banks of ITRS transactions is considered by the IED staff to be a continuing problem, despite efforts in the recent past to provide training for the staff of the banks. The classification problem is particularly acute in the case of credit transactions, many of which are received by telegraphic transfer with little accompanying information on the nature of the transaction.

In addition, deficiencies in the following data sources affect the ability to compile accurate balance of payments statistics:

  • Current and Capital Transfers Credits. The data now provided by the MOF are not suitable for compiling the balance of payments statistics, as they: (i) are annual data compiled on a fiscal year (June 30) basis, rather than a calendar year basis; (ii) do not have complete coverage of the donor agencies; (iii) do not provide a breakdown into current and capital transfers; (iv) do not provide a breakdown into transfers in cash and transfers in kind; and (v) do not provide a breakdown of current transfers into those for budget support, those for goods, and those for services—information required for accurate compilation of the balance of payments statistics. The IED has requested that the MOF provide quarterly data covering all donor agencies, with the breakdowns required for balance of payments compilation purposes, and the MOF is reportedly working to improve the data.

  • Insurance services. The data provided by the Insurance Commissioner cover only total receipts (including premiums) and total payments (including claims) for insurance services. Premiums and claims are not separately identified, and no distinction is made between life and nonlife insurance transactions—additional information that is required to enable insurance services to be calculated more closely in accordance with the BPM5 methodology.

3.1.3 Source data are timely

With the exception of the data on transfers provided by the MOF, all the source data are timely. The trade data have a timeliness of two weeks after the end of the reference period and the data received from the ITRS are very timely—the 20 commercial banks report to the IED twice a month, with a timeliness of one week after the end of the reference period, and the bureau de change report to the IED once a month with a timeliness of 15 days after the period. Penalties for noncompliance ensure that the timeliness deadlines are met. The data on loan transactions and debt forgiveness that are compiled monthly by the DMD and posted on the ESIS network are also very timely—usually within one week following the end of the reference month. The monthly data on reserve assets compiled by the MFAD and the monthly data on grants for budget support compiled by the Aid Coordination Division of the IED have a timeliness of one month after the end of the reference month. The quarterly data provided by the insurance commissioner have a timeliness of one quarter. The timeliness of the annual MOF data on transfers is variable at present, but are more timely than those compiled by the UNDP, which have a timeliness of approximately two years.

3.2 Statistical techniques

3.2.1 Data compilation employs sound statistical techniques

The data compilation procedures used by the IED to compile the balance of payments statistics are reasonably sound, although the possibility of data entry errors exists.

The trade data provided by the TRA are received on diskette. The data from all other sources outside the BoT are received in hardcopy, and in many instances also in electronic formats. The data provided by other departments of the BoT are transmitted via the electronic network, ESIS. All components of the balance of payments are updated monthly, using the monthly data obtained from the ITRS records, the TRA, or from other departments of the BoT. In the case of the transfers data that are provided on an annual basis, the total amount is divided into 12 equal amounts for use in the monthly balance of payments compilation.

The compilation process is as follows. The ITRS data are reported twice a month by each bank on separate forms for receipts and payments, with a page for each of the codes that correspond to the balance of payments components, and show separate aggregate data for each of the currencies in which the transactions were executed. The IED staff key the data into an electronic spreadsheet26 by code and by currency, and print out a copy of the input data. The staff then manually edit the data to verify the accuracy of each individual entry. Once that check has been completed, the electronic compilation program automatically converts the data for each currency to the unit of account, U.S. dollars, using the average exchange rate for the reference period (two weeks), which is obtained electronically through a link to the relevant exchange rates in the ESIS network. The program then produces the various component tables both in U.S. dollars and Tanzania shillings .

In the next step, the IED economists are required to prepare a table indicating the percentage changes from the previous month by code and by reporting bank, as well as to draft text explaining the movements and any significant variations from trend. (The report showing changes and variations from trend is not automatically produced by the compilation program.) The principal economist then reviews the data tables prepared by each section, together with the verification tables and text that analyze the changes, and assesses the movements in light of aspects such as the seasonal pattern of exports, or changes in commodity prices. Economists are asked to investigate any unusual changes, by going back to the reporting bank or other data supplier to determine the reason for the unusual movement. The reporting banks have ready access to the details of the individual transactions and, in consultation with the IED staff, can usually determine whether the change is the result of a misclassification, such as the classification of a payment for capital equipment as services. If insufficient information has been provided to the reporting bank, the IED economists ask the bank to go back to its customer to obtain additional information. However, banks have reportedly become reluctant to do this.

Once the significant variations have been investigated and explained, the principal economist issues instructions to electronically “merge” the separate worksheets into the balance of payments statement, and prepares an analysis of the data. A hard copy of both the text and the balance of payments data is sent to the Deputy Director, the person in charge of the IED, for review. Once approved, the text and the data are sent electronically to the Coordination Division of MFAD, and the data table is archived by IED for the record. The text and tables are reviewed by both the Deputy Director in change of MFAD and the Deputy Director of MFPD, and the IED responds to any queries or clarifications that may arise from those reviews. Once finalized, the Coordination Division incorporates the balance of payments data and text into the relevant publication (the Monthly Economic Review, the quarterly Economic Bulletin, or the annual Bank of Tanzania Economic and Operations Report) which is then sent to the Director of DEP for review and approval. In the case of the Monthly Economic Bulletin and the annual Bank of Tanzania Economic and Operations Report, the draft publication is then submitted to the Governor of the BoT for approval to publish. In the case of the quarterly Economic Bulletin, the authority to approve the document for publication has been delegated to the Director of DEP, with effect from the 2002 editions.

The compilation process as described to the mission appears to have adequate reviews and safeguards against errors. In practice, however, the process does not always prevent errors. For example, the main balance of payments table in U.S. dollars (Table 4.1) published in the fourth quarter 2001 edition of the quarterly Economic Bulletin, the edition that disseminated the first release of the 2001 balance of payments data and was therefore subject to close review, contains errors in the exceptional financing data for 1996 and 1997, in that the sums of the components do not add to the totals.

In addition, it appears that the review process is not always followed, and that economists can amend the data and submit tables for inclusion in publications directly to the Coordination Division of MFAD, without review by their supervisors. This is apparently the explanation for the fact that the Tanzania shillings version of the balance of payments table (Table 4.2) published in the first quarter 2002 edition of the quarterly Economic Bulletin is not consistent with the U.S. dollar version of the table (Table 4.1), and contains many large and obvious errors in the figures both for the 2001 data and for the historical data—such as (i) the reversal of the sign for the overall balances for 2001 and 2000 from negatives to positives, (ii) a figure for the 2001 reserve assets in Tanzania shillings that is smaller than the amount shown in Table 4.1 for 2001 reserve assets in U.S. dollar, (iii) the reversal of the sign for the 1999 reserve assets, (iv) the inclusion of a very large amount of debt forgiveness in 2001 compared with a zero figure in the U.S. dollar version of the table, and (v) the reversal of the sign for the 2000 debt forgiveness element of exceptional financing.

The fact that economists can amend the data and submit amended tables directly to the Coordination Division for publication without review may also explain the numerous and significant inconsistencies between the data in the various detailed tables on exports and imports published in the first quarter 2002 edition of the quarterly Economic Bulletin. (See Section 3.4.2 below.)

It also appears that the IED staff, and the staff of MFAD and MFPD, were unaware of these errors in the published data until they were pointed out by the mission.

The detailed data compilation procedures of the IED staff for the balance of payments statistics are not documented. However, the IED has begun to prepare such documentation with the aim of completing a first draft for internal use by the end of 2002.27

3.2.2 Other statistical procedures (e.g., data adjustments and transformations, and statistical analysis) employ sound statistical techniques

Classification adjustments are made to the data to convert imports reported on a c.i.f. basis to the f.o.b. basis required for the balance of payments statistics. The adjustment is made using an assumption that 13 percent of the total c.i.f. value of imports is freight and insurance, with 90 percent of that amount being freight, and 10 percent being insurance. This adjustment is considered by the mission to statistically sound, provided the assumptions are regularly reviewed. (See Section 3.3.1 for details and recent improvements to the adjustment.)

No coverage adjustments are made for unrecorded trade, or for converting transactions recorded on a net basis to the gross basis required by the balance of payments methodology, or for removing the double-counting of construction services already included in the capital transfer credits.

3.3 Assessment and validation of source data

3.3.1 Source data—including censuses, sample surveys and administrative records—are routinely assessed, e.g., for coverage, sample error, response error, and nonsampling error; the results of the assessments are monitored and made available to guide planning

High value transactions and significant changes in the aggregate data are confirmed by the IED staff with the reporting banks, and can lead to reclassifications. However, as indicated in Section 3.1.2 above, the level of misclassification is still considered to be high, and the IED has identified the need for further training seminars for the staff of reporting banks and bureau de change on the nature of the transactions that should be included under each of the ITRS report form codes.

Source data are assessed, usually as follow-up to a technical assistance mission, to ensure adequate coverage and to identify ways in which the data sources could be improved to address problems. Examples of improvements to the data sources in recent years resulting from this process are:

  • Reserve assets. Prior to 2002, reserve assets were based on the change in the assets shown in the balance sheets of the BoT, converted from Tanzania shillings to U.S. dollars using the average exchange rate for the reference month. Following the recommendation of a recent IMF technical assistance mission, the reserve assets are now converted from Tanzania shillings to U.S. dollars using the end-of-month exchange rates, and calculating the change in reserve assets during a month as being the difference in the U.S. dollar amounts of the stocks of reserve assets for the two relevant months.

  • System for converting import data from c.i.f. to f.o.b. The practice of deducting 13 percent of the c.i.f. value of total imports to arrive at the f.o.b. value, and the assumption that 90 percent is freight and 10 percent is insurance, has been used since 1995. The IED is in the process of introducing a new system, based on information provided by the Pre-Shipment Inspection Company for the years 1999 through 2001 on the actual freight and insurance costs shown on import documents for those goods undergoing pre-shipment inspection. The pilot study showed that freight and insurance were only 10 percent of the total c.i.f. value of imports in 1999, 9.5 percent in 2000, and 9 percent in 2001. The study also indicated that the freight portion was consistently higher than the 90 percent assumed in the past. Under the new system the Pre-Shipment Inspection Company will continue to provide monthly data to the IED on the actual freight and insurance costs shown on the import documents, and the IED will use the percentage indicated by those documents as the basis for the adjustment—with the adjustment percentage being changed each year if applicable. The proportion of freight costs will also be revised each year to reflect the actual data provided by the Pre-Shipment Inspection Company.

  • Travel surveys. Given concerns about the high of levels of travel debits and credits reported through the ITRS records, an alternative survey-based data source for travel has been developed. In March 2001, a committee consisting of the Ministry of Natural Resources and Tourism, the BoT, the NBS, the Immigration Department, and the Zanzibar Commission for Tourism was established and approved two tourism surveys, one of which was a visitor exit survey designed for the balance of payments statistics. A pilot survey was conducted in May–June 2001, the “Low Season”, and two further surveys were conducted, one in July–October 2001 for the “Peak Season,” and one during December 2001–January 2002 for the “Mini Peak Season.” An expenditure model has been developed for the “Low Season” survey, designed to estimate the expenditure of visitors on the basis of the average visitor expenditure by country of residence, by purpose of visit, and by length of stay. Similar expenditure models will be developed for the “Peak Season” and “Mini Peak Season” once the results of those benchmark surveys have been processed. The immigration statistics are being computerized, which will enable the expenditure model to use actual numbers of arrivals. The new data source is expected to replace the ITRS data on travel credits during 2003.

  • Capital flows and stocks survey. The data source for direct investment used to be based on adjusted data on investment approvals made by the TIC. However, a survey of private capital flows and stocks in 1999 conducted in 2000 by the BoT, the TIC and the NBS, with assistance from Development Finance International of the United Kingdom, indicated significantly higher inflows of direct investment equity capital and intercompany loans than the TIC data, which cover only equity capital investment approvals over US$300,000, and do not include the mining companies. The IED is in the process of incorporating the results of the 2000 survey into the balance of payments statistics for 1999 and has compiled revised aggregate data on direct investment in Tanzania for 2000 and 2001 (yet to be disseminated) that are based on the TIC approvals actually implemented, supplemented by information provided by the mining companies. Preparations are almost complete for a new survey of private sector capital flows and stocks to be conducted in late 2002, which will cover data for 2000 and 2001, and will also cover Zanzibar, which was not covered in the 2000 survey.

  • Road transportation credits. Given the nonresponse of road transport enterprises to IED requests for data, and concern that the ITRS records do not fully reflect the cross-border transactions of those road transport enterprises resident in Tanzania that transport goods to neighboring countries, the IED is developing an alternative data source. This will be a data model based on the tonnage transported by road to destinations outside Tanzania, derived from tonnage data provided by the THA, and estimates of the average cost of transport by road to various destinations.

3.4 Assessment and validation of intermediate data and statistical outputs

3.4.1 Main intermediate data are validated against other information where applicable

Significant changes in the exports data are assessed against information on the harvest of various crops (such as cashew nuts, coffee, cotton, cloves, sisal, tea, and tobacco), and the unit price for these crops. The IED compiles and publishes tables in the quarterly Economic Bulletin showing quarterly and annual data on the value, volume, and unit prices of these traditional export items, which assist in this validation process. The exports data for minerals are also assessed against other information, such as production data.

3.4.2 Statistical discrepancies in intermediate data are assessed and investigated

The behavior of data series is not routinely cross-checked with related series, even in the case of data tables compiled by the same sections of the IED. As a result the tables compiled by the IED and disseminated in the BoT’s publications contain numerous inconsistencies.28

The consistency between the data on loan interest and amortization arrears published in the balance of payments statistics and the data on loan interest and amortization arrears in the DMD database is also not routinely checked. As a result, the fact that the balance of payments data were being based on extended-period projections was not noticed for a number of years, and the errors in the balance of payments statistics have yet to be corrected in the disseminated data. The need for such a consistency check was particularly important during recent years when Tanzania’s external debt schedules have been subject to frequent change as part of the Paris Club rescheduling arrangements that formed part of the HIPC29 review process.

In addition, the balance of payments statistics disseminated in the BoT publications do not appear to be checked with the balance of payments data that are disseminated in the annual Economic Survey prepared by the POPP—a high-visibility document that is presented to parliament and used for policy debate—even though the data are apparently compiled by the IED. For example, the balance of payments statistics published in the recently released 2001/2002 Economic Survey have a number of inconsistencies with the data published in the first quarter 2002 edition of the quarterly Economic Bulletin, not only in the 2001 data, which are shown in both publications as being provisional, but also in the historical data, which are stated in the Economic Survey data as being actual data. 30

3.4.3 Statistical discrepancies and other potential indicators of problems in statistical outputs are investigated

The size of net errors and omissions is reportedly closely monitored and the reasons for changes investigated. However, the scrutiny of the size of net errors and omissions does not prevent significant discrepancies in the data approved for publication. For example, until very recently, the IED staff were apparently unaware of a significant increase in the net errors and omissions for 1998 that had occurred following the release of the balance of payments statistics in the first quarter 2002 quarterly Economic Bulletin.31 A subsequent investigation showed that the increase primarily reflects an error in the computer program being used to “migrate” the historical data from an outdated compilation database into a new, more detailed, compilation database, with the result that the 1998 figure for direct investment in Tanzania changed from positive US$172.2 million in the first quarter 2002 edition of the Economic Bulletin to positive US$876.9 million, and the figure for other investment changed from negative US$157.8 million to positive US$75.5 million. The size of net errors and omissions is a key figure in the balance of payments statistics, and had the reasons for the increase been investigated promptly after the “migration” of the data for the affected year, the IED staff would have been immediately alerted to the error in the migration program.

3.5 Revision studies

3.5.1 Studies and analyses of revisions are carried out routinely and used to inform statistical processes

Revisions are made and, although these tend to be major, no formal revision studies are undertaken.

4. Serviceability

4.1 Relevance

4.1.1 The relevance and practical utility of existing statistics in meeting users’ needs are monitored

There is no formal user advisory group and reportedly no queries have been received from users other than government agencies, and international organizations.

There is no established process of consultation with policy departments or ministries, although the NBS, MOF and POPP staff do seek clarification from the IED staff if they have a query about an aspect of the balance of payments statistics. There also does not appear to be a formal process of consultation with those NBS staff who use the balance of payments statistics as a major data source for the external sector of the national accounts, in order to ensure that the balance of payments data meet their needs.

However, the IED has made improvements in the amount of detail disseminated on related data. For example, in 1999 the IED began to publish in the Monthly Economic Review cumulative monthly data for the current year on services and income (Table 5) and monthly data on commodity prices (Table 6). A more detailed version of the aggregate balance of payments data is also expected to be published in the second quarter 2002 issue of the quarterly Economic Bulletin.