1. Structure of government in FSU countries

As defined in GFS methodology, there are two levels of government in the FSU countries-central government and local government. ^2/ For analytical purposes, these two levels of government can be combined to compile data on general government. There is no state/regional level of government. ^3/

Three elements usually exist within the individual levels of government. In GFS terminology these three elements are known as budgetary, extrabudgetary and the social security schemes. In FSU countries, budgetary units consist of central government units covered by the republican budgets and local government units covered by local budgets. ^1/ Operations of budgetary general government are recorded in state budgets, which cover, net of any intergovernmental transfers, the consolidated operations of the central government budget and the local government budgets.

Extrabudgetary funds are defined as units outside the budget that raise monies through, for example, the imposition of taxes or compulsory levies, and use those receipts to provide nonmarket goods and services. A typical extrabudgetary fund found in a number of FSU countries is the Roads Fund, which usually raises revenue (through levies on owners of vehicles, the transport industry, or industry in general, or through ear-marked taxes on fuel) and spends that revenue on the maintenance or construction of roads. In many FSU countries sizeable portions of government operations are carried out through such extrabudgetary funds, although some FSU countries (including Armenia and Kazakhstan) have recently begun to abolish such units, or to incorporate them into the budget.

Social security schemes, the third element found within individual levels of government, are a special category of extrabudgetary units that: (1) carry out the social programs of government (such as payment of pensions and benefits); (2) are funded, at least in part, by compulsory contributions from employees and/or employers; and (3) cover the whole community or particular sections of the community. With the exceptions of Latvia, which integrated these schemes into the budget in 1993, and Kazakhstan and Ukraine, which are currently working towards the same goal, all FSU countries have sizeable social security schemes outside the budget. ^{2/ 3/} These schemes typically comprise the Pension Fund, the Social Insurance Fund and the Employment Fund.^4/

Details of the structure of government in selected FSU countries are set out in the extracts of the Institutional Tables shown in Appendix V.

2. Accounting system underlying the fiscal data

A number of FSU countries are making significant efforts, with the assistance of the Fiscal Affairs Department (FAD) of the Fund, to establish treasury units within the finance ministries responsible for recording the various stages of the payments and receipts of government units. However, at present most FSU countries lack such units. In those FSU countries that have not established treasury functions, the banking system (primarily the central bank) acts as the fiscal agent of government and is the source of the accounting information that underlies the fiscal data.

Accounting systems in FSU countries generally operate in a similar manner. After approval of the budget, staff of the finance ministries prepare a financial plan allocating all expenditure by quarter or month. On the basis of these allocations, the finance ministries authorize the opening, at appropriate state banks, of credit accounts to be used by each spending ministry. The spending ministries then allocate the credit to local branches of their agencies and organizations. The agencies and organizations of the spending ministries carry out the detailed accounting for commitments, for the issuance of purchase orders and for issuance of payment orders to the banks that transact the orders.

The agencies and organizations of the spending ministries are responsible for maintaining detailed accounts of these transactions, and the accounting codes used can be “translated” to the classification codes used in the fiscal reports. Fiscal reports are derived either from transactions shown in the banking system accounts or from data reported by spending ministries, rather than from an integrated set of ledgers maintained by the finance ministries. Transactions recorded in the accounts of spending units are audited concurrently either by the ministries themselves or by staff of the finance ministries, but the year-end “accounts” prepared by the finance ministries summarizing the information provided by the spending ministries are not audited.

3. Source documents for compilation of GFS data in FSU countries

Fiscal reports produced by the finance ministries are the primary source documents used to compile GFS data in FSU countries. These fiscal reports, which summarize transactions recorded during specified periods, are of three types: monthly reports quarterly and annual reports.

1. Coverage of the data

Coverage of GFS data is determined by the definition of government according to GFS methodology. ^1/

In FSU countries, the primary source documents used for compiling GFS data cover only the budgetary elements of central and local governments. Supplementary sources of information on the following activities are often necessary to ensure full coverage: operations of extrabudgetary funds and social security schemes; foreign financing, including loans-in kind; revenue from the sale of grants-in-kind; and lending activities of government.

Sizeable portions of government operations in FSU countries are transacted through extrabudgetary funds and social security schemes. STA technical assistance missions to FSU countries have recommended establishment of a system requiring regular reporting of data on the operations of these funds to the finance ministries, and incorporation of those data into the monthly, quarterly, and annual fiscal reports. ^2/ At the time of writing, no FSU country included adequate data on these operations in its fiscal reports. Furthermore, although data on selected extrabudgetary funds and social security schemes may be available to finance ministries on an ad hoc basis, no FSU country, with the exception of Uzbekistan, had introduced a comprehensive and timely formal reporting system for these units. ^1/

Appendix V lists the extrabudgetary funds and social security schemes in selected FSU countries. Issues affecting two extrabudgetary units, special means accounts, and the hard currency funds, warrant further discussion. Specific problems affecting social security schemes will also be discussed in more detail.

All FSU countries appear to have special means accounts. ^2/ Budgetary units apparently maintain these accounts for recording receipts from activities such as hiring out school halls, and for recording payments on purchases of extra equipment, staff bonuses staff, or repair of buildings, etc. These receipts and payments are not recorded in expenditure and revenue data in the fiscal reports, although any surpluses above a specified level are apparently transferred to the budget at year end. ^3/

The small size of these extrabudgetary funds and the onerous task of collecting data from so many different sources make it impractical to include the transactions of these accounts in the fiscal reporting system and the GFS data of FSU countries. However, any substantial increases in these operations (as appears to have occurred recently in at least one FSU country) will raise two issues affecting the recommended treatment in GFS data. The first issue involves the nature of the activities. If the increased level of receipts is from activities that are still incidental to the main functions of each government unit, or is from fees for schools and hospitals or from regulatory fees, then they continue to be noncommercial in nature, and gross data on receipts and expenditure financed by those receipts should be included in the GFS data. However, if receipts from renting buildings and other activities are deemed to be commercial in nature, then the current situation where gross receipts and payments are excluded from the fiscal data, and only the surpluses are included as revenue, is correct. If these accounts are, in fact, considered to be part of government, a second issue arises concerning measurement of their operations. It appears that the statement on the source of funding for additional expenditure in the annual reports provides this information, but it is not known whether the coverage of the data in this statement is complete. ^{1/ 2/}

Hard currency funds are the second type of extrabudgetary unit that presents special problems. Many FSU countries still operate extrabudgetary funds for their foreign currency transactions, and the transactions of these funds are of particular importance in those countries where receipts from foreign exchange surrenders are thought to be very large. ^3/

Data on the transactions of these accounts is difficult to obtain in a number of FSU countries. A few, such as the Russian Federation, Azerbaijan, and Tajikistan, do include some information in their sub-annual fiscal reports, but this information is insufficient for the compilation of adequate GFS data. ^4/ In rare cases, such as Uzbekistan, information on the operations of the extrabudgetary Hard Currency Fund is now available in sufficient detail to permit the transactions to be incorporated into GFS data. ^5/ In a number of FSU countries, however, inadequate information on the operations of hard currency funds constitutes a serious gap in the coverage of the fiscal data.

As a side issue, special problems related to social security schemes in FSU countries also affect the compilation of GFS data. Most FSU countries operate social insurance funds, pension funds and employment funds that are not Integrated into their fiscal reporting systems. A somewhat unusual feature of these schemes is the extent of the involvement of non-government entities in delivery of the services. Such involvement affects both the timeliness and the nature of the data available, though not the coverage.

The social security schemes are funded primarily through compulsory levies on employers and employees. These levies are collected by the employing organizations and used to pay out pensions, sickness benefits, etc, to employees. In effect, therefore, each enterprise or employing organization is operating a separate fund.

Numerous enterprises and employing organizations are involved in this activity. For example, a STA technical assistance mission estimated that, in 1992, there were between 12,000 and 14,000 such units in Latvia. ^1/ Obtaining timely data from these myriad sources is difficult. Enterprises sometimes send detailed reports on gross receipts and payments only once a year to the central administrative units of the social security schemes. Throughout the year, the only source of data available at central levels may be the net balances. ^2/

Another area where the coverage of the data in monthly, quarterly, and annual reports is inadequate is the recording of foreign borrowing, including loans-in-kind. Many FSU countries have made sizeable drawings under foreign loans in recent years and data on these drawings do not appear to have been included in the fiscal reports. For example, recent Fund staff reports and aide-mémoires identified foreign loans that had not been included in the fiscal reports of Belarus, Estonia, Georgia. Lithuania, Moldova, Tajikistan, and Ukraine. (These examples are discussed in more detail in Section V). In the Russian Federation, a recent STA technical assistance mission was unable to obtain adequate information on 1992 foreign borrowing estimated to total US$ 11 billion. ^3/

Few FSU countries appear to have adequate records on the outstanding foreign debt of central and local governments. ^4/ STA technical assistance missions have emphasized the necessity for developing centralized debt registers covering domestic and foreign debt and reconciling information on debt transactions (drawings, amortizations and interest payments) with data in the fiscal reporting systems. Debt liabilities arising from loans-in-kind present a particular problem in some countries, especially those in which hard currency funds are administered outside the finance ministries, because the finance ministries frequently learn of the existence of liabilities only when the first interest or amortization payments fall due.

In certain FSU countries (such as Armenia, the Baltic states, Kazakhstan, Tajikistan, and Ukraine), some progress has been made with establishing debt registers for the central government covering both domestic and foreign debt. Finance ministries in some countries have also made some progress in canvassing other ministries for details of outstanding debt liabilities arising from loans-in-kind. For example, in Moldova, an attempt is apparently made to record, through an extrabudgetary fund (the Fund for Foreign Credit and Humanitarian Assistance), data on both debt incurred and some of the receipts from sale of goods received under loans-in-kind. However, it appears that these records are only fragmentary and have not been incorporated into the fiscal data. ^1/ In general, however, STA technical assistance missions to FSU countries have been unable to ascertain that adequate arrangements exist for recording in government accounts either the liabilities incurred through loans-in-kind, or the revenue from any sale of the goods received under such loans-in-kind.

Lack of registers to record contingent government liabilities arising from central and local government guarantees of enterprise debt also remains a problem in most FSU countries, although the Baltic states and Moldova appear to maintain some records of these liabilities. Contingent liabilities are not included in GFS data. However, the importance of recording this information is illustrated by the recent example of an FSU country in which enterprise borrowers defaulted on sizeable amounts of loans made under government guarantee, and the debt had to be assumed by the government.

Grants-in-kind should not be included (other than as a memorandum item) in the cash-based GFS data, unless the goods received have been sold. Except in Estonia, it has not been possible to establish how receipts from any such sales are treated or what procedures may be in place to record expenditure financed by those sales. It seems likely, however, that such receipts and payments are not being recorded in the fiscal data at present and should be added to ensure adequate coverage for GFS data. ^2/ In Estonia, a recent STA technical mission found that receipts from sales of goods received under grants-in-kind are placed in counterpart funds held at various banks. The Bank of Estonia apparently has year-end records of the outstanding balances of these funds, but no details about the flow of receipts during the year or about the expenditure from these funds. These receipts and payments are not included in the data in the budget or annual report. ^3/

In addition, coverage of government lending activities (whether financed by foreign borrowing or from domestic resources) in the source documents is currently inadequate for compilation of GFS data. The problem is both widespread and of sizeable proportions. Significant levels of lending activities, financed by foreign borrowing or by borrowing from the central bank and apparently not included in the fiscal data, have been identified in nine FSU countries. ^1/

With the exceptions of Belarus, the Russian Federation and Uzbekistan, the exact nature of these transactions has yet to be examined by STA technical assistance missions. It is possible that some of the transactions may be transfers rather than lending and that some of the lending activities would not, on the basis of GFS methodology, be included in the measurement of the deficit or surplus. However, GFS methodology does require that any lending that is financed by borrowing and that results in a government liability be recorded as a transaction of government and, hence, included in the measurement of the deficit. ^{2/ 3/}

Full details, including sufficient information on the borrower to permit classification of that entity by function, by type (nonfinancial public enterprise, nonprofit organization, household, etc.) and by residency (domestic or foreign), of these transactions should be recorded. The recorded information should also distinguish between loans to an organization and transfers. It is also important that the nature of any transfers (capital or current) be recorded.

Finally in this section on the coverage of the fiscal reports, it has been noted that in Estonia there is a possibility that the coverage of local government data may be adversely affected by the change to the fiscal year of local governments introduced in 1994. As only statistics on general government “can show the overall magnitude of government operations in a country, the allocation of resources through government for various purposes, the aggregate weight of taxes, and the structure of the tax system,” ^1/ compilation of fiscal data for general government of FSU countries is highly desirable. Every effort to compile local government data in a regular and timely manner should therefore be strongly supported.

2. Existing classification codes

Shortcomings in the classification systems of FSU countries have a negative effect on the compilation of accurate and timely GFS data for these countries.

With the exception of Lithuania, the FSU countries continue to use classification systems very similar to that of the former USSR. ^{2/ 3/} Countries that have modified their classification systems have done little more than delete obsolete codes referring to the USSR, add new codes for new national institutions or new taxes, and translate the code book into their national languages. ^{4/ 5/ 6/}

A major problam with the existing classification system in the inability to accurately measure the deficit of surplus of level of government. Even with the disaggregated data available in annual reports it is not possible to measure the deficit or surplus accurately because classification codes often fail to distinguish between amortization payments (which are negative financing transactions) and interest payments (which are items of expenditure). Examples of codes that fail to make this distinction are Razdel 150.4 (Payments for Borrowing from Foreign Governments), Razdel 222.1 (Payments for Loans), and most of the sub-codes of Razdel 239 (Expenditure on Servicing the State Domestic Debt). ^1/ Additional information must be sought from debt register transactions, or from units of government responsible for administering the domestic and foreign debt, to try to reconcile that information with the accounting data, but this is not always possible.

Also affecting accurate measurement of the deficit or surplus is the fact that the classification codes do not distinguish subscriptions paid to the International Monetary Fund from those paid to other international organizations. ^2/ In addition, the lack of razdel codes for receipts from foreign financing and for receipts from the sale of goods obtained through grants-in-kind or loans-in-kind prevents the accurate measurement of the deficit or surplus.^3/

Another major shortcoming of the classification systems used in FSU countries is the inadequacy of statia codes for compiling data on expenditure by economic type (GFS Table C). The deficiencies are numerous. (1) The codes do not identify interest payments (classified under Category 2 in GFS Table C), or distinguish between interest payments and amortization payments. (2) No provision is made for a separate category for employer contributions to social security schemes (classified under Category 1.2 in GFS Table C). ^{1/ 2/} (3) Items that belong in different categories of GFS Table C are combined into single statia codes. [For example, Statia 18 includes items of expenditure on wages and salaries (classified under Category 1.1), and subsidies (classified under Category 3 in GFS Table C), as well as items that are other purchases of goods and services (classified under Category 1.3).]. ^3/ (4) Adequate classification of recipients of subsidies and current transfers is not possible because the codes do not distinguish among subsidies and transfers to nonfinancial public enterprises, financial institutions, and other enterprises; nor is it possible to identify transfers to other levels of government or transfers abroad. (5) No distinctions are made among various types of capital spending (purchases of fixed capital assets, purchases of land and intangibles, capital transfers, etc.), and capital transfers cannot be classified by type of recipient, or residency of recipient (domestic transfers or transfers abroad). (6) Definitions determining the classification of certain items are based on the item value-as a result, the contents of certain codes will change significantly in periods of high inflation. ^4/ (7) Statia codes do not distinguish between lending and expenditure transactions.

Perhaps most significantly, statia classification codes do not cover all items of expenditure. Many FSU countries have therefore introduced a code known simply as “expenditure not covered by statia” in order to balance expenditure classified by statia with total expenditure classified by razdel, When the statia codes have been used by STA technical assistance missions to compile data on expenditure by economic type, the unclassified category has proved to be unacceptably large. (In 1992, the percentages of total expenditure of budgetary central government not able to be classified by statia were 30 percent in Armenia and Turkmenistan, and 50 percent in Azerbaijan.) Furthermore, it appears that the proportion of total expenditure placed in the unclassified category may be growing. (In Azerbaijan, the proportion grew from 37 percent in 1991 to 50 percent a year later.)

Other shortcomings of classification systems used in FSU countries affect compilation of accurate data in various GFS tables. For example, the lack of razdel codes to identify lending by function, or the subsequent repayment of such loans, hinders the compilation of data for GFS Table B1 (Lending minus Repayments by Function.) ^1/ Distinctions used in razdel codes dealing with domestic financing transactions are not always sufficiently specific to permit adequate compilation of GFS tables on financing by debt holder and by debt instrument. ^2/ As well, the codes do not permit capital revenue from the sale of fixed assets or land to be distinguished from receipts from the privatization of enterprises. ^3/

Finally, classification code systems used in FSU countries do not facilitate identification of intragovernmental transactions that should be eliminated in consolidation of central government data. ^4/ For example, there are no separate codes for the subsidies transferred from the budget to the pension funds or the social insurance funds, or for the amounts transferred from the budget to social security schemes as employer contributions. ^5/ Total transfers and subsidies can usually only be determined from the receipt records of the social security schemes. However, these records are often not detailed enough to permit transactions to be classified by economic type because the records do not distinguish between receipts of social security schemes that are employer contributions from government (Category 1.2 in GFS Table C) and those that are subsidies from government (Category 3 in GFS Table C).

3. Level of data aggregation in the fiscal reports

The level of data aggregation in the monthly and quarterly reports places serious constraints on compilation of timely and accurate GFS data in FSU countries. Shortcomings in this area prevent precise measurement of the governments’ deficit or surplus on a monthly or quarterly basis, and accurate compilation of data on expenditure classified by economic type and by function.

The level of data aggregation in monthly and quarterly fiscal reports negatively affects the measurement of the deficit or surplus. Data in monthly and quarterly fiscal reports of FSU countries do not provide sufficient detail to permit financing items to be separated from revenue and expenditure items. For example, Razdel 8 (Receipts from Foreign Economic Activity) includes receipts from foreign loans (which are financing items) as well as receipts from customs and export duties and is usually reported in monthly and quarterly reports as a single line item. (In Moldova, the razdel is also reported as a single line item in the annual report.) Similarly, Razdel 150 (Expenditure on Foreign Economic Activity), which includes both amortization payments on foreign loans and foreign currency expenditure on medicine, food, etc., is reported as a single line item in many monthly and quarterly reports.

In some FSU countries, financing items can be identified from transactions in the debt register. However, this is not always possible and the problem with measuring the level of the deficit or surplus on a monthly or quarterly basis is particularly acute in FSU countries that operate extrabudgetary hard currency funds and do not provide detailed information on the transactions of these funds.

The level of aggregation in the monthly and quarterly fiscal reports, specifically, the difficulty in identifying intragovernmental transfers, mutual settlements and other intergovernmental transactions, also has an adverse effect on the ability to compile consolidated data. ^1/

Another significant problem with data in the monthly and quarterly fiscal reports is lack of information on expenditure classified by statias, the codes that permit an approximate classification of expenditure by economic type. Monthly fiscal reports do not include this information. FSU countries that have separate quarterly reporting systems usually collect data on spending by statia for the social/cultural sector (the Ministry of Education, Ministry of Health, Ministry of Social Security, Ministry of Culture, and Mass Media, etc.), although only Kazakhstan incorporates this data in the quarterly fiscal report. ^2/

Even in Kazakhstan, a recent STA technical assistance mission found that the statia data reported in fiscal reports for the second and third quarters of 1993 covered only 10 to 17 percent of the total expenditure for the period. ^3/ There were two main reasons for this low coverage. The first was the fact that ministries responsible for the National Economy sector (Razdel 100), which accounts for approximately one third of total expenditure in Kazakhstan, are not required to report details of their expenditure classified by statia on a quarterly basis. The second was that the expenditure of razdels controlled by the Ministry of Finance, is not classified by statia. Furthermore, even ministries that are required to report expenditure by statia did not provide full coverage, and reported, on average, only about one quarter of their total expenditure.

Until these problems with coverage of the statia data can be resolved, quarterly fiscal reports of FSU countries cannot be used to compile data on expenditure classified by economic type.

The level of aggregation in monthly, quarterly and, in some cases, annual fiscal reports also presents difficulties for compiling data on expenditure classified by function. The most significant problem exists with the National Economy sector (Razdel 100), which is usually reported as a single line item in monthly and quarterly reports. Expenditure on this sector constitutes a significant proportion of total expenditure in all FSU countries. ^1/ However, insufficient information is available in monthly and quarterly fiscal reports to permit the sector to be classified according to the appropriate functional categories, which include housing, agriculture, mining, transport, and manufacturing. The only exceptions are the quarterly fiscal reports for Kazakhstan and Turkmenistan, which now include information on the spending organization (glava) in sufficient detail to permit an accurate classification. ^2/

Other razdels reported as single lines items in monthly and quarterly fiscal reports of FSU countries also cover a number of different functional categories. (For example, Razdel 222, Other Expenditure, covers 8 of the 14 functions in the GFS classification system. These are general public services, defense, public order and safety, health, social welfare, housing, transport and communications, and interest payments.) These razdels can be adequately classified only on an annual basis in most FSU countries.

Belarus represents a special case in that significant elements of the data in the annual report are recorded at an aggregate level only. Data at the paragraf level are not recorded in the annual report, a circumstance that seriously limits accurate classification of annual data for expenditure by function. (Data on expenditure on the National Economy, however, are reported by glava; thus it is possible to classify by function that important area of expenditure.)

The level of aggregation of part of the data in the annual report for Armenia also hinders accurate classification of expenditure by function in that country. Data in the annual report are not provided at the paragraf level for Razdel 208 (Defense) so it is not possible to separate expenditure on military pensions (which are classified under the social security and welfare category of GFS Table B) from the rest of the expenditure on defense.

Finally, the level of aggregation in monthly and quarterly fiscal reports does not facilitate the separation of tax and nontax current revenue, capital revenue, and repayments of lending. For example, Razdel 12 (Fees and Various Nontax Revenue) is usually reported as a single line item in the monthly and quarterly fiscal reports. This razdel includes paragrafs that cover capital revenue from the sale of fixed assets, receipts from the sale of enterprises, and items that are taxes in GFS methodology, as well as the nontax revenue implied by the razdel title.

4. Exclusion of commercial and industrial activities

GFS methodology specifies that all data should be recorded on a gross basis, except for any commercial or industrial activities that might be included in the source document. In such cases, only the net overall deficit or surplus (in the case of any nonfinancial public enterprises [NFPE’s] included in the document) or the net operating deficit or surplus (in the case of any departmental enterprises included in the document) are recorded in GFS data. ^1/

Source documents used to compile GFS data in FSU countries are largely free of any commercial or industrial activities. The only NFPE commonly found in the source documents is the state lottery. However, both receipts and expenditure of this activity are clearly identified, so that it is possible to remove these from the fiscal data, and add back either the overall surplus as revenue, or the overall deficit as expenditure. ^{2/ 3/}

Lack of detailed information on the defense sector in many FSU countries has precluded a determination of whether there are any defense-related commercial or industrial activities included in the source documents. Classification code books used by FSU countries typically include a razdel for the receipts of the Ministry of Defense (Razdel 12.20, Receipts from Departments and Organizations of the Ministry of Defense), which has been treated as noncommercial in nature in the generic GFS classification set out in Appendix IX and placed under Category 9 of GFS Table A (Administrative fees and charges and incidental sales). However, some uncertainty remains about the true nature of the receipts shown under this code.

It is important to note that the list of spending organizations (glava) in the classification code books of FSU countries is not a list of organizations included in the budget, but rather a list of those organizations that regularly receive transfers or loans from the budget, and for which the finance ministries are authorized to open allocation accounts without special approval from parliament. The lists cannot, therefore, be used as indications that commercial activities have been included in the budgetary accounts. For example, the classification code book for Turkmenistan lists numerous commercial entities, including Turkmenistan Railways, Turkmen Airlines, Turkmen Shipping Company, etc., in this section, but the operations of these organizations are not included in the budgetary accounts of Turkmenistan.

5. Basis of recording the data

As stated in Section III, in most FSU countries data are recorded on a cash basis-a practice which is consistent with GFS methodology. ^1/ The exceptions of Belarus and the Russian Federation were also discussed in that earlier section. However, it is interesting to note that the basis of recording for part of the expenditure data in fiscal reports of the remaining FSU countries extends a step beyond cash.

As in many countries, the fiscal systems in FSU countries distinguish between budgetary organizations and self-supporting organizations. However, FSU countries maintain their accounts on the basis of resources that have been actually spent both by budgetary organizations and by the self-supporting organizations. In GFS methodology, amounts transferred from the government to those self-supporting organizations that are considered to be outside the definition of government (organizations involved in commercial, industrial and financial activities, for instance) would be treated as expenditure of the government at the time of transfer to the organization, rather than at the time of actual spending by the recipient organization.

It has not proved practical to separate out these transactions from the rest of the data recorded in the source documents. Nonetheless, a strict interpretation of GFS methodology would require adjustments for instances of delays between allocations to, and spending of, funds by nongovernment organizations.

Compilation of GFS data is also affected by whether or not the source document accounts include complementary period transactions. In GFS methodology, a complementary period is defined as being any period after the end of the fiscal year during which accounts are held open to permit transactions to be assigned to the previous fiscal year. As the basis of recording in GFS is cash, any transactions between government and the rest of the economy that take place after the end of the fiscal year must be assigned to the fiscal year in which the transactions occur, rather than to the fiscal year in which budget provision was made.

In the FSU countries other than the Russian Federation, the accounts appear to be left open only for mutual settlements and other intergovernmental transfers. ^1/ These mutual settlement accounts (the natures of which vary according to the changes in budget laws during each year) are usually “settled” at the end of each quarter, and final settlements are made at the end of the fiscal year, after all other accounts have been closed. ^2/ Such transactions are not considered to indicate a true complementary period in that they are not transactions between government and the rest of the economy. If, however, transactions with the rest of the economy are made during the period that accounts are kept open, such complementary period transactions should be removed from GFS data on expenditure or revenue.

It appears that the Russian Federation does have a genuine complementary period-or at least did have in 1992 and classified under Category The Russian Federation, like many FSU countries, operates a system of cash management under which the amount of money spent must not exceed the level of revenue received in the government bank accounts. In 1992, there were lengthy delays between the dates on which receipts were apparently deposited in the banking system and the dates on which the money was credited to the relevant government accounts. The Ministry of Finance obtained a short-term, interest-free revolving loan from the Central Bank of Russia to cover this “float.” At the end of the fiscal year, the loan expired and the spending ministries were unable to finance their scheduled expenditure because the revenue collected by the banking system had yet to be credited to their accounts. A decision was therefore made to hold the 1992 accounts open until February 20, 1993 to record as expenditure transactions that had been (1) made during January 1993 and the first three weeks of February 1993 and (2) financed by revenue received by the banking system in 1992, but not credited to the government accounts until 1993. Because these expenditure transactions took place between the government and the rest of the economy after the end of the fiscal year, they constitute complementary period transactions that should be removed for compilation of GFS data. It appears that the situation affecting the 1992 data in the Russian Federation was an ad hoc arrangement, and it is not known whether it recurred in 1993.

Finally, although much of the data in the fiscal reports of FSU countries is recorded on a gross basis, certain data are recorded in the fiscal reports on a net basis. For example, data on government borrowing from central banks appear to be presented on a net basis only; no data are provided on gross drawings and repayments. Similarly, data on central government lending to local governments are recorded on a net basis. As a result, monthly and quarterly reports for these items can show fluctuations in lending and borrowing levels, which are seemingly inconsistent with the cumulative nature of the data.

6. Consolidation

The method of consolidation used in the source documents also affects the compilation of GFS data. ^1/

In FSU countries operations of consolidated general government are recorded in the accounts of the state budgets. STA technical assistance missions that have prepared data for central, local and general governments of selected FSU countries have verified that the process of consolidation used to prepare state budget data is consistent with GFS methodology and that all intergovernmental transfers between local and central governments appear to have been eliminated. The exception is Turkmenistan, where local government revenue shown in the state budget appears to include intergovernmental transfers from the central government.

Intergovernmental transfers between the various units of local government do not, however, appear to be eliminated from the source documents. ^2/ As a result, the levels of local government revenue and expenditure in FSU countries may often be overstated. For example, in Azerbaijan the revenue data in the source document were overstated in 1991 by 17 percent and the expenditure figure was overstated by 18 percent because intergovernmental transactions between the various units were not eliminated from the local government data. ^1/

7. Integrity of the source data

The integrity of the source document data also affects the compilation of GFS data in FSU countries.

As stated earlier, the data in the source documents are not formally audited. Under the former USSR, government units reporting quarterly and annual data to the Ministry of Finance were required to obtain confirmation from the banks that the data reported were consistent with the bank account balances. Some FSU countries, such as Estonia and Kazakhstan, have continued the system of informal quarterly auditing, but, in Kazakhstan, there is pressure from the central bank to discontinue the practice. As a result, a recent STA technical assistance mission to Kazakhstan observed that increasing amounts were being placed in the “Other Expenditure” categories to account for discrepancies between amounts reported by spending ministries and the central bank account balances. ^2/ It is not known whether this problem exists with data in the annual reports.

The presence of inconsistencies in the data is another problem observed by STA technical assistance missions in many FSU countries. The component figures often do not add to the total. Lines for “statistical discrepancies” have therefore had to be added by the missions to ensure that the GFS data balance. Some of the inconsistencies arise when the annual reports are typed, with transposed figures or typing errors. Even documents that are copies of computer printouts may have errors as they are usually not prepared using a spreadsheet program and, as recently proved to be the case in Kazakhstan, there may be no system of numerical consistency.

1. Exchange taxes arising from foreign currency surrenders

A number of FSU countries operate a system of compulsory surrenders to hard currency funds of varying portions of the foreign currency received by exporters, with surrenders being exchanged for domestic currency at a rate that is less than the market rate. The issue has been raised as to whether these receipts from compulsory surrender of foreign currency can be treated as revenue in the GFS measurement of the deficit or surplus.

GFS methodology separates these transactions into two parts, only one of which is treated as revenue. That portion of receipts from compulsory surrenders of foreign exchange that can be attributed to the difference between the market rate of exchange and the compulsory surrender rate of exchange would be considered to be an exchange tax in GFS methodology (that is, a tax levied on the sale of foreign currency that constitutes a levy on the payment for exports, rather than a tax on the export itself). These receipts would therefore be treated as revenue for the purpose of measuring the deficit of the country and classified under Category 6.5 of GFS Table A, Exchange taxes.^1/

The remaining portion of receipts from compulsory surrenders of foreign exchange would not be included in GFS data, but would be deemed to have been exactly offset by expenditure for the purchase of foreign exchange. In other words, neither the amount of domestic currency spent by hard currency funds to purchase foreign currency, nor the amount of foreign currency received by the hard currency funds in return for domestic currency are considered to be expenditure or revenue in GFS methodology. Instead, they would be considered to be merely the exchange of one financial asset for another, which is an activity of the financial sector, and therefore outside the definition of government. ^2/ However, if the hard currency funds subsequently use the foreign currency to purchase goods and services abroad on behalf of government, then the domestic currency equivalent of the funds actually spent should be treated as expenditure of government, in exactly the same way as if government had purchased foreign currency from the central bank to pay for its foreign currency obligations.

In Turkmenistan in 1992 and in Uzbekistan in May 1993, compulsory surrender systems have been replaced with explicit taxes on payments for exports, payable in foreign currency. ^{3/ 4/} These taxes would also be classified as an exchange tax under Category 6.5 of GFS Table A, Exchange taxes, and included in the measurement of the deficit or surplus.

2. Valuation of transactions of the hard currency funds

Quite frequently, records of the operations of the hard currency funds in FSU countries are maintained in foreign currencies only, raising the issue of how to value in domestic currency those transactions that represent expenditure or revenue of government.

In GFS methodology, the recommended treatment is that transactions should be valued at the market exchange rate prevailing at the time of each transaction. Often, however, this is not practicable. Other options that have been used by STA technical assistance missions and area departments are to use the average quarterly or the average annual exchange rates to calculate the domestic currency equivalent of the foreign currency. If an average exchange rate is to be used, the recommended method would be to use the average for the shortest period possible-in other words, it is preferable to use the average for the relevant month, rather than the average for the relevant year.

Regardless of the method used, it is very important to bear in mind that the use of different valuation methods can lead to differences in the data, especially in times of volatile movements in the exchange rate as experienced in many FSU countries recently. ^1/

3. Treatment of centralized export regimes for strategic goods

In several FSU countries, such as Turkmenistan and Uzbekistan, the centralized export regimes inherent in the former USSR state order system continue to play a significant role in the economy. For example, in Turkmenistan, 60 percent of GDP is estimated to be derived from production of natural gas, and of that, 90 percent is exported through the government owned enterprise, Turkmengas. ^2/ In Uzbekistan, where production of cotton for export plays a major role in the economy, 80 percent of cotton production must be sold to the state. ^3/

The buying and selling of market goods is a commercial activity and any expenditure or revenue of centralized export regimes should not be included in GFS data. The only transactions that would normally be included in the GFS measurement of the deficit or surplus would be: (1) subsidies or transfers from government to the export monopoly to cover operating losses (which would be treated as expenditure and classified under Category 3.1.1 of GFS Table C, Subsidies and other current transfers to nonfinancial public enterprises); (2) lending by government to the export monopoly (which would be treated as lending minus repayments and classified under Category 8.2 of GFS Table C, Domestic lending minus repayments to nonfinancial public enterprises); and (3) transfers to government of the profit of the export monopoly (which would be treated as tax revenue and classified under Category 6.3 of GFS Table A, Profits of export and import monopolies). In addition, transfers from government to recapitalize the export monopoly, or to purchase capital equipment, or for amortization payments for debt of the export monopoly would be treated as capital expenditure and classified under Category 7.1.2 of GFS Table C, Capital transfers to nonfinancial public enterprises.

4. Indirect subsidies arising from counterpart funds for foreign currency allocations

In a number of FSU countries, including the Russian Federation, governments allocate foreign currency obtained through various means (including compulsory surrenders or foreign borrowing) to selected enterprises in return for specified amounts of domestic currency. The domestic currency is normally required to be deposited by the enterprises into special counterpart accounts usually held by governments outside the budget. ^1/ Often the amount of domestic currency to be deposited into the counterpart account is calculated at an exchange rate favorable to the enterprise, implying a subsidy to the enterprise for their imports. The issue has been raised as to how these transactions should be recorded in GFS data.

The transactions involve the effective sale of foreign currency by government to the enterprises and can therefore be viewed as the opposite side of purchases of foreign currency under the compulsory surrender schemes discussed earlier. The same treatment as recommended for compulsory surrenders of foreign currency would be applied to foreign currency allocation transactions-that is, only the difference between the domestic currency equivalent of the foreign currency valued at the market rate of exchange and the amount of the domestic currency received by government from the “sale” of the foreign currency to the enterprise would be included in GFS data.

To illustrate, we can assume that the amount of foreign currency allocated to the enterprise was the equivalent of 100 billion rubles when valued at the market exchange rate, and that the amount of domestic currency paid into the counterpart account by the enterprise in exchange for the foreign currency was 50 billion rubles. The subsidy to the enterprise would be 50 billion rubles and would be classified in GFS expenditure data under Category 3.1.1 of GFS Table C, Subsidies to nonfinancial enterprises, (if a current transfer) or under Category 7.1.2 of Table C, Capital transfers to nonfinancial public enterprises (if a capital transfer).

5. Treatment in GFS of other indirect subsidies arising from loans-in-kind and commodity loans

In many FSU countries there are other indirect subsidies that arise from transactions involving loans-in-kind or commodity loans. Countries such as Armenia, Belarus, and Georgia have received substantial amounts of such loans from abroad in recent years. For example: Armenia received wheat from the European Union (EU); Belarus received grain loans from the United States; Georgia received foreign loans in fuel and wheat. ^1/ To the extent that these commodities have been sold to bread manufacturers and other enterprises at prices that would result in proceeds less than the debt liabilities incurred by governments, indirect subsidies from the governments to the enterprises have been involved. The issues are how these indirect subsidies should be treated in GFS and how they can be separately identified for Fund operational purposes.

There are three separate and explicit transactions involved, each requiring separate recording in GFS data. The first transaction is borrowing by the government under the loan-in-kind. This liability should be recorded in GFS financing and debt data as being borrowing by the government at the time the debt liability is deemed to have been incurred-which may be the time the goods are shipped, or the time at which government receives the goods, or at some other point in time, depending on the terms of the contract with the lender.

The second transaction is expenditure by government of that “borrowing,” through the transfer to the enterprise of the goods received under the loan-in-kind. This transaction should be recorded in GFS data at the same time as the debt liability is recorded, regardless of when the enterprise may receive the goods. The transaction should be classified as a current transfer to the enterprise under Category 3.1.1 of GFS Table C, Subsidies and other current transfers to nonfinancial public enterprises, and should be valued as the domestic currency equivalent of the debt liability of the government, calculated using the exchange rate prevailing at the time the debt liability was incurred. In other words, the amount recorded as expenditure in GFS data must match exactly the amount recorded in the GFS financing and debt data.

The third transaction is the subsequent receipt by government of money paid by the enterprise for the commodities received. These monies are typically received by government some time after the enterprise has received the goods. In the case of bread manufacturers, for example, the government would usually receive payment from manufacturers only after sale of the bread made with the imported wheat or grain. These receipts from enterprises would be treated as nontax revenue of the government and classified under Category 8.2 of GFS Table A, Entrepreneurial and property income from nonfinancial public enterprises and public financial institutions.

The indirect subsidy to the enterprise would be measured as being the difference between the amount of the government transfer to the enterprise (shown under Category 3.1.1 of GFS Table C) and the amount of money from the sale of the goods subsequently paid by the enterprise to government (shown under Category 8.2 of GFS Table A). To facilitate the calculation of the size of these indirect subsidies for Fund operational purposes, it is recommended that the transactions be separately identified under “of which” lines in Category 3.1.1 of GFS Table C and Category 8.2 of GFS Table A.

This treatment would also apply to those circumstances when government has borrowed abroad and used the foreign currency to pay for purchases of goods such as wheat and grain, which the government subsequently allocates to various manufacturers at prices resulting in proceeds less that the domestic currency equivalent of the debt liability of the government. Although these loans are not loans-in-kind, the nature of the indirect subsidy to the enterprises is the same. ^1/

6. Proceeds from the noncommercial sale of goods received through loans-in-kind and grants-in-kind

In many FSU countries a number of government ministries, for example, the health ministries, have apparently received loans-in-kind of items such as medicine, etc. As stated earlier, these loans-in-kind should be recorded in GFS data because they represent a liability of the government. However, there is some uncertainty about how to treat in GFS the receipts from any noncommercial sale of goods received under loans-in-kind, or from any sales of goods received under grants-in-kind.

In the case of loans-in-kind, the domestic currency equivalent of the loan amount should be entered in the GFS data as borrowing (positive financing) at the time the debt liability is incurred and simultaneously an equal amount should be entered as expenditure. ^1/ (In the case of a loan-in-kind of medicine received by the Ministry of Health, for example, the expenditure would be classified under Category 5 of GFS Table B, Health, and under Category 1.3 of GFS Table C, Purchases of other goods and services.)

The recommended treatment of receipts from any subsequent sale of the goods assumes that the sales are noncommercial in nature. The GFS Manual states that “government departments engaged in the usual social or community activities of government…, should not be considered public industries unless the fees set for these services are clearly designed to approximate the full costs of production.” ^2/ If it is assumed that the activity is a usual social or community activity of government, such as the purchase of medicines for distribution to the public in the case of the health ministries, and that any sales of medicines are incidental to the main activity of the government ministry, then “the entire gross proceeds of their sale are included in revenue.” ^3/ The receipts would therefore be treated as nontax revenue and classified under Category 9 of GFS Table A, Administrative fees and charges and nonindustrial and noncommercial sales.

The receipts from sales of goods received under grants-in-kind should be treated in the same way as those received through the noncommercial sale of goods received under loans-in-kind, although they would be classified under Category 12 of GFS Table A, Other nontax revenue, rather than under Category 9. However, as the grant-in-kind does not involve any future liability of the government, no other entries are made in the GFS data, except to include a memorandum item to the tables stating the value of these grants-in-kind, and to record any subsequent expenditure financed by these revenues.

7. Receipts from the sale of gold

The issue of receipts from sales of gold affects several of the FSU countries, notably the Russian Federation and Uzbekistan. Both these countries are significant gold producers and in both countries the governments are involved in buying and selling gold on a monopoly basis through the State Precious Metals Committee (Goskomdragmet). The authorities in the Russian Federation have in the past presented a case to the Fund arguing that any revenue from sales of the gold reserves of the State Precious Metals Committee should be treated as revenue of the government in measuring the deficit or surplus.

In the Russian Federation the situation is complicated by the fact that the State Precious Metals Committee transactions in gold encompass commercial operations of trading in gold, as well as the management of the gold reserves. ^1/ These two separate, although interrelated, functions are treated in different ways in the GFS data.

Commercial operations from sales of gold from current production would be treated in the same way as any other commercial activity, with only the net profit or loss transferred to the government being included in the GFS data; the profit from sales of gold production would be included as revenue, and the loss would be included as expenditure. ^2/

Management of a country’s gold reserves is a function of the monetary authorities and therefore outside the government sector. In compiling GFS data, any transactions of the monetary authorities’ functions should be eliminated from the statistics for government and replaced with a single imputed transaction representing the net flow of funds to or from government from these activities. (Any net flow to government should be treated as positive financing from the monetary authorities subsector, and any net flow from government should be shown as negative financing from the monetary authorities.) This treatment, however, does not apply to profits flowing to government from these monetary authority operations, as they are deemed to be similar to profits from the central bank and therefore should be treated as revenue in GFS data and included in the measurement of the deficit or surplus.

It is important to ensure that profits transferred to government do not include unrealized profits on the revaluation of the gold reserves. Any such amounts equivalent to unrealized profits or losses that are transferred to government should be included in GFS data as positive financing (unrealized profits) or as negative financing (unrealized losses). Only realized profits transferred by the monetary authorities to government would be classified as revenue, under Category 8.2 of GFS Table A, Entrepreneurial and property income from nonfinancial public enterprises and public financial institutions.

Stocks of gold held by the State Precious Metals Committee in the Russian Federation should be regarded as forming part of the international reserves of that country, the management of which is a function of the monetary authorities. These stocks would therefore be treated in GFS data in the manner described above. Appendix XI provides more detail on the proposed methodology for calculating the profits from gold transactions in the Russian Federation.

8. Treatment of profits of the central bank

In a number of FSU countries, for example, Armenia, Kazakhstan, Moldova, and the Russian Federation, the central banks recently started to transfer profits to their governments on a regular basis. In some instances the initial transfers of profits were sizeable. In Kazakhstan, for example, 25 billion rubles were transferred to government in the first quarter of 1993 and a further 30 billion in the second quarter of 1993. ^1/ The size of these initial transfers raised concerns about whether the receipts represented “real” profits of the central banks or whether they were merely credits from the central banks to the governments under another name, which should be treated as financing.

The explanation given to the recent STA technical assistance mission in Kazakhstan for the large size of the transfers was that the central bank had not transferred its profits to the government for some time, and that therefore the initial transfers represented the accumulated profit of the past several years. Time did not permit further investigation of this explanation by the STA mission, but it may be useful to reiterate the GFS methodology concerning the transfer of the central bank profits to the government.

Calculation of the profit of central banks should cover their entire operations, without any attempt being made to distinguish between profits on sales of gold reserves or foreign exchange reserves and profits on the rest of the activities of the organizations. The profits should be compiled in accordance with accounting procedures that exclude unrealized profits from the revaluation of holdings of foreign exchange or gold reserves. The profits of the central banks actually transferred to governments would be treated as revenue and classified under Category 8.2 of GFS Table A, Entrepreneurial and property income from nonfinancial public enterprises and public financial institutions.

9. Government loans to foreign countries

It has been observed that, in at least one instance, Fund staff reports have treated loans made by the government of one FSU country to another FSU country as negative financing on the part of the lender. ^2/ This treatment is not consistent with GFS methodology.

The lending minus repayments aggregate in the GFS data includes all purchases and sales of financial assets by government made for public policy purposes, and covers both lending to domestic enterprises and lending abroad. The loan by the government of Azerbaijan to Georgia to enable the latter to purchase fuel would appear to have been made for public policy purposes rather than for liquidity management reasons. In terms of the international methodology, therefore, it should be classified either under Category 9.1 of GFS Table C, Lending minus repayments abroad, to governments and international organizations (assuming that the loan was made to the government of Georgia) or under Category 9.3 of GFS Table C, Other lending minus repayments abroad, if the loan was made to a nonfinancial public enterprise in Georgia.

As lending, GFS methodology would include the transaction in the measurement of the deficit or surplus of the lending country. In the GFS data of the borrowing country, however, the loan would be treated as being positive financing, which would not affect the measurement of the deficit. To quote the GFS Manual: “The two sides of a single intergovernmental lending transaction would be treated asymmetrically: as lending, determining the deficit, in the statistics of the lending government, and as borrowing, financing the deficit, in the statistics of the borrowing government.” ^1/

10. Proceeds from privatization

In almost all FSU countries privatization programs are underway, albeit on a relatively small scale in many instances. Often these programs are being administered by separate privatization funds, many of which are extrabudgetary, although in some countries (such as Kazakhstan) these funds are now incorporated into the budget. Treatment of proceeds from the privatization programs differs across staff reports of the Fund. In some instances, proceeds are treated as negative lending. In other cases, they are treated as capital revenue and in yet other instances they are treated as financing.

GFS methodology is clear on how these proceeds should be treated, and divides the receipts into three separate categories. First, “privatization” receipts from sales of government buildings and housing should be treated as capital revenue under Category 13 of GFS Table A, Sales of fixed capital assets. Second, receipts that arise from privatization of land should also be treated as capital revenue and classified under Category 15 of GFS Table A, Sales of land and intangible assets. Finally, receipts that are proceeds from the privatization of enterprises should be treated in GFS data as negative lending, and classified under the appropriate functional category of GFS Table B1, Lending minus Repayments by Function, and under Category 8.2 of GFS Table C, Domestic lending minus repayments to nonfinancial public enterprises. Privatization proceeds from sales of enterprises are viewed as sales of the government’s holdings of equity in the enterprises that had been “purchased” through earlier transfers to, and capitalizations of, the enterprises. In all three cases, therefore, privatization proceeds are included in the measurement of the deficit or surplus, as defined in GFS methodology.

Privatization proceeds from sales of enterprises are not treated as financing in GFS methodology. This reflects the fact that GFS, unlike the System of National Accounts (SNA), distinguishes between acquisitions of financial assets (including equity participations) for public policy purposes and those financial assets that are acquired for liquidity management purposes. The former are treated in GFS as lending and the latter as negative financing.

It is recognized, however, that the “one-off” nature of privatization proceeds warrants separate identification of these receipts for analytical and Fund operational purposes. It is therefore suggested that both the gross lending and the repayments be treated as separate lines in the fiscal data, rather than combined and shown as net lending, which is the standard GFS presentation for the lending minus repayments aggregate. Separate “of which” lines could be added to the repayments data to identify any receipts from privatization of enterprises, and the data from these “of which” lines could then be used as building blocks to construct an alternative measurement of the deficit/surplus.

11. Government assumption of enterprise and central bank debt

At the end of 1992, many of the central banks of FSU countries had significant negative balances in their correspondent accounts with the Central Bank of Russia. During 1993, agreement was reached to convert these negative balances from liabilities of the various central banks to debt liabilities of the relevant governments. In addition, some FSU countries that had arrears under the state trading agreements with other FSU countries agreed to convert the unpaid balances into government debt. Furthermore, in some instances the government assumed the domestic debt of enterprises.

There are numerous examples. In 1993, the government of Armenia assumed the Central Bank of Armenia debt to the Central Bank of Russia, with the negative ruble balance of the correspondent account being converted into medium-term loans from the government of Russia totalling US$45 million. Also in 1993, the government of Georgia signed agreements with the governments of Armenia, Azerbaijan and Russia converting into interest-bearing state loans the unpaid balances of the state trade agreements as at the end of 1992. At the end of 1992, 166 billion rubles of outstanding short-term debt owed by the public enterprises to the banking system was converted to government debt by the government of Turkmenistan. As well, in June 1993, the government of Turkmenistan assumed the 136 million rubles debt of the Central Bank of Turkmenistan to the Central Bank of Russia. In May 1993, Ukraine reached agreement with Russia to convert into government debt the US$1 billion of 1992 interenterprise arrears and the US$1.5 billion outstanding correspondent account balances of the Central Bank of the Ukraine with the Central Bank of Russia. In addition, in July 1993, the Ukraine government agreed to convert into government debt US$ 28.4 million of arrears under a state trade agreement with Georgia. In Uzbekistan, the deficits at the end of 1992 in the correspondent accounts with the Central Bank of Russia were converted into government debt of 57.3 million rubles. Finally, In 1993 the government of one of the FSU countries assumed a significant amount of defaulting enterprise debt that had been lent under government guarantee. ^1/

There has been some debate about the appropriate treatment of these assumptions of debt on the part of governments of FSU countries and, specifically, whether the increase in debt should be shown in financing, with an offsetting entry in expenditure.

GFS is a cash-based system and therefore does not treat noncash assumptions of debt by government as financing and expenditure at the time of assumption. ^2/ However, as debt generated without a cash transaction can nevertheless give rise to an obligation to repay in cash, such debt should be included in the debt statistics of government. ^3/ Subsequent cash payments made for interest on the debt would be treated in GFS data as expenditure, while any subsequent cash payments for amortization of the debt would be included as negative financing. The amortization payments would also be recorded in GFS debt statistics as a reduction in the outstanding level of debt. In GFS methodology, therefore, only the interest payments for the assumed debt affect the measurement of the deficit or surplus of the government that assumed the debt.

GFS methodology does, however, recognize the necessity of being able to identify the noncash debt issuances of governments, and the system makes provision for the recording of this information under memorandum item 12 of the financing table, GFS Table D, Financing by Type of Debt Holder, which covers the noncash issuance of debt repayable in cash. ^4/ In addition, the debt table, GFS Table G, Outstanding Debt by Type of Debt Instrument, provides for a reconciliation between net borrowing during the period and the level of outstanding debt at the end of the period. One of the many reasons for differences between the two figures is memorandum item 14.1, noncash issuance of debt.^5/

12. Treatment of grants in GFS methodology

It has also been noted that grants are on occasion included in foreign borrowings in Fund staff reports. ^6/ GFS places grants into a specific aggregate, separate from revenue or financing transactions. This separation reflects one of the key distinctions that underlie the entire GFS analytical framework, namely, the distinction between repayable and nonrepayable transactions. Grants are defined as being “unrequited, nonrepayable, noncompulsory government receipts from other governments or international institutions.” ^{1/ 2/} Borrowing, on the other hand, is a repayable transaction.

Grants are included in the GFS measurement of the deficit or surplus. However, GFS also recognizes that grants are essentially different from revenue in that the recipient government has little control over the amount or the continuity of grants; in the case of revenue the government has the sovereign power to raise taxes. The GFS framework therefore uses the “building block” approach, and separates out grants from other nonrepayable receipts to facilitate the construction of alternative definitions of the deficit or surplus, which might not include grants in the measurement of the deficit. Even if grants are placed “below the line” under an alternative definition of the deficit, however, they should always be separately identified, because their nonrepayable nature makes them quite distinct from borrowing transactions.

13. Government on-lending and quasi-fiscal operations of the monetary authorities and banks

As discussed in Section IV on the coverage of lending activities of government, there have been significant sums on-lent to enterprises in recent years in a number of FSU countries. In most cases, these on-lending activities have been financed by domestic or foreign borrowing and have been transacted at preferential rates for public policy purposes through a variety of government agencies, as well as through the central bank or the commercial banks acting under the directive of parliament. The issue has arisen as to whether or not these transactions should be included in the measurement of the deficit or surplus in GFS methodology.

In GFS the criterion for making this decision is not which organization transacts the loan, but, rather, which organization is liable to repay the debt incurred to finance the loans. If the debt (whether incurred through foreign borrowing or through domestic credits from the central bank) is the liability of government, then the lending should be included in the GFS data and, hence, in the measurement of the deficit or surplus. If, however, government merely guarantees to repay the debt in event of default by the borrowing enterprise, then the lending is not included in GFS data as a transaction of government. ^3/ This distinction is critical to ensure symmetry between borrowing and expenditure transactions in GFS data.

It is acknowledged that this distinction has been the subject of some concern when comparing the deficit of one FSU country, such as Estonia, which appears to transact all public policy lending through government agencies, and those, such as Georgia or Belarus, that transact significant amounts of such lending through the central bank or commercial banks. ^1/ It is also acknowledged that the distinction can create some difficulties in monitoring the level of certain activities of government, especially when the mechanism used to make the loans varies from year to year. To illustrate using the case of Ukraine, in 1992 and again in the first quarter of 1994, lending to the enterprises was undertaken through the government accounts. In 1993, however, the lending to the enterprises was transacted through the National Bank of the Ukraine, without being charged to the government accounts. ^2/

In recognition of these difficulties, it is recommended that the authorities in FSU countries be encouraged to record the level of quasi-fiscal lending undertaken by the financial sector. This information can then be used by Fund staff as a “building block” to add to GFS data if required, as was done in the recent Fund staff report on Ukraine. It would be essential, however, to ensure that the data are separately identified in any measurement of the deficit or surplus, so that they can be removed by analysts wishing to measure the deficit or surplus as defined in the international methodology.

When the lending is considered to be part of the activities of government, the issue has also been raised about the treatment in GFS of the “indirect subsidy” arising from preferential interest rates on the loans. These “subsidies” are effectively included in GFS data, but are not separately identified. They are included in the measurement of the deficit or surplus as the difference between interest payments made by government to the central bank or the foreign lender (which would be included in Category 2 of GFS Table C, Interest), and the lower amount of interest received by government from the enterprise. (These latter receipts would be included under Category 8.3 of GFS Table A, Other property income.) Separate “of which” lines could be established for these two GFS categories if it were felt to be important for Fund operational purposes that data be identifiable on the indirect subsidies to the enterprises arising from the preferential

14. Measurement of arrears of government units

The budgetary “cash management” system used in many FSU countries limits monthly expenditure to amounts equal to the level of revenue received during the relevant period. This practice has led to a build up in arrears of unpaid obligations of government units. The Fund has a need to monitor the level of these arrears in order to be able to assess the true level of fiscal performance and the effect on demands for credit from the monetary system.^1/

Although GFS is a cash-based system, it acknowledges that payment statistics alone cannot meet all the needs for economic and financial analysis of government operations, and therefore recommends the compilation of supplementary noncash data to augment the GFS data. ^2/ The GFS tables make provision for the recording on data on what is known as floating debt in GFS terminology, which is defined as payment orders written for which checks have not yet been issued or paid. ^3/ Both the changes in the level of unpaid obligations and the outstanding level of unpaid obligations are recorded in GFS data as memorandum items to the expenditure tables, the financing tables and the debt tables. (As Memorandum Item 17 of GFS Table C, and as Memorandum Item 14 in GFS Table D, both entitled Change in floating debt of unpaid obligations, and as Memorandum Item 13 in GFS Table F, Floating debt of unpaid obligations.)^4/ These data can be used as supplementary information to construct a measurement of the deficit or surplus closer to an accrual basis than the strictly cash-based GFS concept.

The concept of floating debt, however, differs somewhat from the concept of arrears. Arrears are defined as the unpaid overdue obligations of government. Not all floating debt, therefore, is necessarily arrears, in that if floating debt is measured as the difference between payment orders Issued and checks paid it will include an element that is not an overdue obligation, but merely the normal float of checks in transit. Furthermore, the data on floating debt will not include overdue obligations resulting “from delays in the processing of incoming bills prior to the issuance of payment orders”. ^5/ Arrears of this nature can be significant in some countries, but are not usually reflected in the fiscal accounts and can generally be measured only by obtaining information from the individual ministries. The GFS system recommends that changes in the levels of such arrears should also be shown as a memorandum item. ^6/

No FSU country appears to have adequate records on the level of arrears. In large measure this reflects the fact that no centralized records are kept of the stages in the expenditure cycle between allocations made by the finance ministries to the spending ministries and payments made through the government bank accounts. In other words, there are no centralized records kept of commitments or deliveries. Although it is expected that the establishment of treasury units will eventually overcome this problem, it is recommended that, as an interim solution, systems be introduced requiring spending ministries to regularly report timely data to the finance ministries on the outstanding level and change in the level of arrears.

15. Treatment in GFS of specific receipt items

Descriptions given in the classification code books used in FSU countries can be obscure or misleading. On the basis of knowledge gained by successive STA technical assistance missions, a bridge table that classifies the USSR codes into appropriate GFS codes was prepared in late 1992, and now serves as the basis for classifying revenue and expenditure in the individual FSU countries. (See Appendix IX.) The rationale for the treatment in GFS of the more unusual receipt items will be explained in this section.

Razdel 5 (Rent Payments) has been classified as a tax under Category 5.6 in GFS Table A, Other taxes on goods and services. Notwithstanding the title of the razdel, the item does not cover receipts from rental of property or land, but, rather, is a compulsory payment for the right to undertake an activity. Razdel 10 (Timber Revenue) has also been classified under Category 5.6, as it was decided that these were compulsory levies similar to a production tax. (Royalties on production of oil, which have recently been introduced in a number of FSU countries, would also be treated as a production tax under Category 5.6 in the GFS classification.)

Razdel 9 (Reimbursements of Budget Expenditures) is not negative expenditure as might be assumed from the title, but repayments of loans made to NFPE’s and other domestic institutions, such as non-profit organizations. It has therefore been classified in the GFS data as negative lending, rather than as revenue.

The title of Razdel 12.41 (Patent Fees) might imply that these receipts should be classified under Category 8 of CFS Table A, Entrepreneurial and property income. However, these fees are not connected with patents issued for inventions, etc., but are a form of license issued to lawyers and other self-employed persons authorizing them to conduct a business. They have therefore been classified as tax revenue under Category 5.5.1 of GFS Table A, Business and professional licenses.

Other razdels with obscure titles are Razdel 12.42 (Revenue from the Inspection of Correctional Institutions) and Razdel 19 (Receipts from Revaluation of Balances of Goods and Inventories). Razdel 12.42 covers receipts from fines paid from the wages and salaries of convicted criminals in lieu of serving prison sentences, and has therefore been classified as nontax revenue under Category 10 in GFS Table A, Fines and forfeits. Razdel 19 has been classified as a tax under Category 1.2, Corporate taxes on income, profits and capital gains, as it was decided that these receipts were a tax on the “windfall” gains resulting from revaluation of stocks of price-controlled goods when price increases are decreed.

Razdel 12.46 (Compensation for the Cost of Treatment Given to Victims of Criminal Offenses) was treated as negative expenditure in the GFS tables, rather than as a fine, because the receipts were directly related to the cost of medical treatment of the victim of the crime, rather than a fine imposed for the criminal act itself.

Another unusual receipt item is Razdel 37.5 (Compensation for Losses of Agricultural Production due to Withdrawal of Agricultural Lands). This is money paid by state-owned farms and other organizations as compensation for withdrawal of land from agricultural use to permit, for example, the construction of a workshop. The compensation is not for sale of the land, but for the loss of agricultural production from the land. It was decided that it was neither a tax nor receipts from rent or from the sale of land. Furthermore, it was decided that the receipt item was not a fine or forfeit in the strict sense. It was therefore classified as nontax revenue under Category 12 of GFS Table A, Other nontax revenue.

Finally, Azerbaijan has apparently recently introduced license fees to replace the system of income tax assessments for the self-employed that was felt to be too administratively cumbersome. In GFS these “license fees” would be treated as a tax under Category 1, Taxes on income, profits and capital gains, rather than as an administrative fee or charge.

16. Reconciliation of fiscal and monetary data

In GFS a standard procedure used to check the validity of data prepared from the fiscal reports of a country is to compare data on domestic financing of the deficit or surplus and data on the level of outstanding domestic debt with data on the net government liabilities or assets held by the monetary authorities and the domestic money banks. Although the figures will rarely be able to be reconciled exactly, it would normally be expected that the two sets of data would be reasonably comparable.

Although considerable effort is devoted to trying to ensure that data on government operations in the two macroeconomic data systems correspond as closely as possible, there are a number of reasons why these aggregates of the fiscal and monetary statistics may differ. Some reasons relate to legitimate, methodologically sound differences in the treatment of the same operations in each of the statistical systems. They include the treatment of noncash operations (for example, the noncash debt issuance or debt assumption by the government) which are not reflected in the GFS financing data but are fully reflected in the monetary statistics on claims on government. Similarly, book entries related to revaluations will result in a change in the banking sector claims on government, but will not be included in the cash-based GFS financing data. Other reasons for discrepancies between the fiscal and monetary statistics relate to imperfections, deficiencies and errors in compiling the actual data. They include discrepancies in coverage, misclassifications (for example, in GFS between bank and nonbank financing, and in the monetary statistics in the sectoral classification of deposits and claims), and differences in timing. ^1/

In many FSU countries it has not proved possible to reconcile domestic financing information in the GFS data with the comparable data in the money and banking statistics. Partly this reflects the differences in coverage, valuation, timing, etc. Between the two macroeconomic data systems. However, in FSU countries there is the added complication that the government accounts held at the banks cannot usually be readily identified. For example, two recent STA technical assistance missions that attempted to reconcile fiscal and monetary data found that part of the problem was the confusing state of the chart of accounts of the central banks. In Azerbaijan, the mission recommended that the Ministry of Finance and the National Bank of Azerbaijan liaise to ensure consistency in recording of government liabilities and deposits in the central bank’s accounting system.^2/ In Estonia, the mission found that, while the Bank of Estonia was collaborating with the Ministry of Finance in an attempt to improve the identification of central government units and their corresponding budget and non-budget accounts, “the identification of the local government accounts will be a slower process.” ^3/

The problems with identifying the government accounts in the monetary data should be borne in mind when comparing fiscal and monetary data of FSU countries, especially in those countries where the technical memoranda stipulate that the monetary data be used for monitoring of Fund programs.

1. Provision of high-level support for development of macroeconomic data

The first priority is seen as ensuring that work on developing adequate macroeconomic statistics in FSU countries is supported at very high levels of government. Without this support for change, progress will be slow.

STA multitopic technical assistance missions have recommended that the authorities in FSU countries establish statistical councils consisting of representatives of the Office of the President, the Ministry of Finance, the State Committee on Statistics (or Ministry of Statistics or Office of Statistics), the central bank and other relevant government bodies. It is envisaged that these councils would be chaired by a senior member of government and would meet at least once a month to discuss the latest available macroeconomic data and to consider the ongoing development of new statistical systems. It is also envisaged that such a council would be responsible for ensuring that legislative and regulatory structures are in place to facilitate both the complete and accurate reporting of macroeconomic statistics and the publication of statements of economic indicators on a timely and regular basis; statements that would summarize, for example, the key aggregates of fiscal data, balance of payments, exchange rates, domestic interest rates, consumer price index and a monetary survey.

It is hoped that the recent appointment of STA statistical advisors in the Baltic countries, Kazakhstan, the Russian Federation, and the Ukraine will result in progress in this area.

2. Establishment of a GFS data compilation unit

Extensive training in GFS methodology has already been provided to FSU countries. ^1/ While this training will continue, the next essential step is considered to be establishment of GFS compilation units in each FSU country. These units should be adequately staffed with trained personnel, who should also be given the authority to liaise with staff of counterpart units at the central banks, as may be necessary. Although establishment of such units is considered to be of key importance, few FSU countries have taken this step as yet.

Location of such a unit is an issue to be decided by the authorities within each country. However, the experience of STA has been that there are advantages in locating the units within the finance ministries, perhaps within the budget monitoring section, where staff of the unit would have ready access to timely and detailed fiscal data. These data would include not only data from the fiscal reporting system, but also data provided by sections within the finance ministries responsible for recording debt transactions and information on the activities of extrabudgetary units and social security schemes.

3. Extension of coverage of the fiscal reporting system to include all units of government

As has been emphasized throughout this paper, one of the major shortcomings of the fiscal data currently compiled in FSU countries is the limited coverage of the reporting system. A very high priority is the establishment as soon as practicable of a formalized system for monthly reporting by central and local government extrabudgetary units and social security schemes of data on their revenue, grants, expenditure, lending, repayments and financing transactions. ^1/ The reporting system should also be extended to those activities of government that are not transacted through formal extrabudgetary funds, but rather through informal arrangements such as counterpart funds or other government accounts not included in the budgetary data. Furthermore, the obligation to report such data should be a legal requirement, and specific provision for this should be made in the statistical legislation of FSU countries.

4. Revision of existing classification codes

A major impediment to development of adequate fiscal data in FSU countries continues to be the inadequacies of the existing classification codes. Revision of the classification codes is therefore considered to be a high priority.

In addition, the work currently being undertaken under the auspices of the Fiscal Affairs Department of the Fund to develop treasury units in FSU countries provides an opportunity to revise the accounting codes that underlie the fiscal data. An ideal accounting system would be one that was designed to facilitate compilation of GFS data, and it is recommended that any revisions of the accounting and fiscal classification codes take into account the international classification categories set out in the GFS tables in Appendix IV.

If a major review of the classification codes is not undertaken immediately, it is essential that, as an interim measure, the existing razdel codes be amended as soon as possible to separate amortization payments from interest payments, to distinguish between expenditure and lending activities of the government, and to reduce the unacceptably high level of expenditure that cannot be classified by either function or economic type. The statia codes used to classify expenditure by economic type should also be amended in line with the categories in GFS Table C, using the classification bridge tables in Appendix X.

5. Establishment of registers for government debt and contingent liabilities

The next priority area of development is seen to be the establishment of centralized registers to record details of the level of government debt, both domestic and foreign, and the transactions affecting that debt. These registers should cover all outstanding debt of government, including debt incurred through loans-in-kind. Coverage of the data in the debt register should match the coverage of government in each FSU country, including debt data on extrabudgetary funds and social security schemes.

Data on the domestic and foreign financing transactions of government should be prepared monthly by the units responsible for maintaining these registers and should be fully reconciled with the data in the fiscal reports. These debt data should include sufficient detail to permit the compilation of GFS financing tables-for example, details of type of debt instrument, type of debt holder, amount of money borrowed or securities issued, and amount of debt repaid or securities redeemed. The unit should also prepare information on the amount of interest paid and amount of bank charges and fees paid and reconcile such information with data in the fiscal reports.

In addition to a debt register, a central register should be established to record details of contingent liabilities of government arising from government guarantees of debt of enterprises or other organizations.

6. Amendment of consolidation methods for intergovernmental transactions within the same level of government

The method used to compile fiscal data for local government should be amended to ensure that all intergovernmental transactions between the various units of local government are removed and that the level of activity of local government is not overstated. This would require the elimination from the data of transfers from higher levels of local government, such as regions (rayons), to lower levels of local government, such as towns (gorods). It would also require elimination from the data of transfers in the opposite direction, such as taxes collected by lower levels of local government that are transferred to higher levels of local government.

7. Liaison with central banks to identify government accounts

As indicated earlier, an important method of checking the integrity of GFS data is to compare financing and debt data with data in the monetary survey on the banking system’s claims on government and deposits of government. Such comparisons are hindered in FSU countries by the inability to identify all relevant accounts of central and local government in the banking system. An important priority in development of GFS reporting systems in FSU countries is therefore liaison with the central banks to identify the government accounts. This work should be undertaken by staff of the units established to compile GFS data, working in conjunction with their colleagues at the central banks and other relevant banks.

8. Establishment of a system to record arrears

In addition to the cash-based GFS data, supplementary non-cash information is needed on the level and changes in the unpaid overdue obligations of government, known as arrears. The absence of treasury units in many FSU countries means that data on arrears are not readily available to policy makers and analysts. It is considered important that, as an interim measure, systems be established for regular and timely reporting by spending units to the finance ministries of outstanding levels and changes in arrears.

9. Establishment of a system to record levels of quasi-fiscal lending by the financial sector

As discussed earlier in this paper, quasi-fiscal lending is frequently undertaken by the financial sector in FSU countries, creating problems of data comparability in those countries where lending activities may move back and forth between government agencies and the financial sector in different fiscal periods. An important priority is therefore seen as being the establishment of a system to compile data on levels of quasi-fiscal lending undertaken by the financial sector. It is envisaged that this data would be used as a “building block”, supplementing data on the fiscal operations of government in construction of an alternative definition of the deficit or surplus.