VI. Fiscal and Public Sector
  • 1 0000000404811396 Monetary Fund
  • | 2 0000000404811396 Monetary Fund


The government budget plays a key role in a country’s economic policies, both directly (through its spending and revenue programs) and indirectly (through the effect of government budgetary actions—such as borrowing from the banking sector—on other macroeconomic sectors). These effects make budgetary policy an important policy instrument in any financial program. Therefore, it is necessary to prepare accurate projections for the revenue, expenditure, and financing components that comprise the budget.

1. Introduction

The government budget plays a key role in a country’s economic policies, both directly (through its spending and revenue programs) and indirectly (through the effect of government budgetary actions—such as borrowing from the banking sector—on other macroeconomic sectors). These effects make budgetary policy an important policy instrument in any financial program. Therefore, it is necessary to prepare accurate projections for the revenue, expenditure, and financing components that comprise the budget.

a. Defining government

The government sector is generally defined to include those functions carried out by public authorities in the exercise of the compulsory powers granted through the political process. In practice, this definition encompasses a wide range of nonmarket services, including income transactions financed by other economic sectors and carried out by a number of government entities.1 These entities include central, state (or provincial), and local governments, as well as a number of quasi-governmental agencies such as publicly owned financial and nonfinancial institutions, social security schemes, and supranational authorities. Public enterprises that operate according to commercial principles and that do not figure actively in public policy are typically excluded. In analyses of fiscal policy and its role in the economy, the scope of government is normally restricted to those agencies the authorities use to implement public policy. However, owing to a lack of adequate data, fiscal analyses are often restricted to central government operations in many countries. The financial sector is defined separately from the public sector in order to highlight government dealings with and access to financial institutions.

b. The objectives of the forecasting exercise

The primary objective of this workshop is to project the overall budget deficit and the financing requirements that feed through the other sectors and thus influence (among other things) primary target variables such as growth and inflation. A structured approach that estimates government revenues and expenditures directly and the deficit as a residual is generally the preferred procedure, particularly because it provides the flexibility to consider the potential impacts of various policy alternatives. This approach merits careful consideration.

It is generally agreed that conceptually the best modeling approach adopts a comprehensive or “general equilibrium” system to capture the feedback among variables. Such an approach can be used to study the effect of different tax rates on the tax base over time (and thus to determine revenue estimates) and the effect of government subsidies on resource allocation and the supply of goods. In practice, however, the general equilibrium approach yields few dividends and is overly complex. Resource reallocation on a scale sufficiently large to have a noticeable impact typically takes time, and the “noise” from other unforeseeable developments during that time brings the validity of the approach into question. In the short term (one year), that is the focus here, the use of the partial equilibrium approach is usually as reliable and computationally more tractable, since forecasts are based on simple projections of primary economic indicators (the independent variables in regression equations). It should also be recognized that special problems arise in forecasting for the many countries whose underlying economic structures are changing substantially. Economic forecasting techniques, irrespective of whether they follow a general or partial equilibrium approach, cannot be directly applied without reference to structural changes that might affect established behavioral relations.

There are two basic approaches to forecasting the budgetary position. One is to prepare the baseline scenario, which involves the use of underlying trends to project the likely budgetary outturn in the absence of any exceptional developments or new initiatives. In this case, each major component (revenue and expenditure) can be projected on the basis of expected changes in the most influential proxy variables (see section 2). However, in the context of the financial programming exercise, it may be determined that on the basis of certain overall indicators of budgetary performance—such as the overall and current account balances—fiscal policy should be adjusted to support other macroeconomic targets. The overall deficit may be too large, for example, contributing to excessive growth in the money supply or pushing external debt beyond prudent limits, or the current account balance may not generate enough savings to finance the investment effort. In this case, fiscal policy is constrained, and attaining the summary targets may require adjustments in individual revenue and expenditure components through discretionary budget measures. Even in the second case, however, a good initial estimate of the likely budget position in the absence of any new fiscal initiatives is essential to determining appropriate economic policies.

The basic approach to the estimation procedure has three steps. The first involves projecting the most important revenue and expenditure categories using quantitative relationships derived from detailed estimates of the relevant bases (such as taxable sales and imports) for these variables. In cases where estimates of directly linked proxy variables are not available, the variables to be projected should be related to pertinent and available macroeconomic variables (GDP and imports, for example). Other minor revenue and expenditure categories are then projected collectively on the basis of trends or relationship to GDP. The results are then totaled to provide the estimated budgetary deficit, and the various financing sources used to derive the estimated amount of domestic bank financing required. It should be noted that in the initial baseline projection, this amount is estimated as a residual.

2. Estimation Concepts

a. Revenues2

The main revenue distinctions are between tax and nontax items and, with respect to taxes, between direct and indirect. Taxes are classified as direct if the individual characteristics of the unit being taxed are considered in assessing the tax liability and indirect if they are not. While these distinctions are of considerable interest in appraising tax policy and figure prominently in the standard revenue data presentations, they have virtually no role in the forecasting process. More relevant to forecasting indirect taxes is the distinction between specific and ad valorem taxes. Specific taxes are unrelated to the prices of the taxed items themselves and thus do not increase with inflation, whereas ad valorem taxes are imposed as a percentage of the sale price and so rise in line with inflation. In the absence of discretionary measures, receipts from ad valorem taxes rise much faster than receipts from specific taxes, particularly as inflation accelerates.3 The response of revenue collection to changes in output is obviously an important relationship and has been formalized in the concept of elasticity.

The elasticity of a revenue source is a measure of responsiveness to changes in income. More precisely, elasticity is defined as the percentage change in revenue collections from a given tax structure relative to the percentage change in the income base. In mathematical terms, this relationship is written:


where e is elasticity, T^ and Y, respectively, are tax receipts from an unchanged tax system and national income, and ∆T^ and ∆Y are the corresponding changes in revenue and income during the period. Revenue elasticity is greater than one if revenue grows faster than income during the period and less than one if incomes grow faster than revenues. Ad valorem taxes are normally more elastic than specific taxes.

The definition of tax elasticity closely parallels the concept of price elasticity that is part of general economic analysis and hence shares many of the same insights into market relationships. The more inelastic the demand for a taxed good, the more responsive total tax collections are to increases in the tax rate, since the inelasticity of demand indicates that consumers are willing to accept much higher prices for a good rather than to forego its use. Clearly, tax policy is not (and should not be) based simply on the ease of taxing goods with inelastic demand, although inelasticity is one factor that must be considered in determining tax policy. Indeed, it has been used in arguments supporting high sumptuary tax rates for alcohol and tobacco, two goods that exhibit inelasticity of consumption.

A closely related concept is that of buoyancy. The buoyancy of a tax is the percentage increase in revenue collections relative to the percentage increase in income. A quick comparison shows that the definitions of elasticity and buoyancy are identical, except that there is no reference to “a given tax structure” in the latter. The mathematical representation of buoyancy is


and is exactly the same as that for elasticity; the only difference lies in the interpretation of the revenue variables.

For buoyancy, the percentage change in tax revenues (∆T/T) is measured on the basis of actual tax collections; for elasticity, the actual collections used in equation 6.2 are adjusted for the effects of any discretionary adjustments to the tax code undertaken during the sample period. If the adjustments enhance revenues, buoyancy will exceed elasticity, as collections will exceed the receipts that would have been generated in the absence of discretionary measures. In the opposite case, when discretionary measures decrease revenues, elasticity will exceed buoyancy. Since elasticity depicts tax responsiveness to changes in income in the absence of adjustments to the revenue system, it is the correct concept to use for projecting future receipts before undertaking additional policy measures. Boxes 6.1 and 6.2 discuss two techniques that can be used to make this adjustment.4

One final implication of the elasticity concept is of interest here. While most taxes are linked in some form or another to the overall level of income, in some cases the tax base itself can be separated from general economic developments. Trade taxes, for example, are fundamentally linked to the level of imports. In these cases, revenue growth may be caused primarily by changes in the tax base rather than by the elasticity of the tax itself, and it is important to understand exactly what the source of the revenue growth is. One way to approach this problem is to rewrite the basic relationship for elasticity (6.1) as two separate elasticities, in the following form:

Regression Technique

If no information is available on the revenue effects of discretionary tax changes, but the number of years with discretionary tax changes is small relative to the sample with revenue data, tax elasticity can be estimated from actual tax collection data using an econometric estimation with dummy variables that isolate the discretionary effect. The dummy variables “capture” the shift in the relationship created by the discretionary changes, as in:



DUMi = 0 during years with no discretionary changes; and

DUMi = 1 during years when discretionary changes are introduced

The coefficient b measures tax elasticity, excluding the effect of the discretionary change in revenue collections in year i and (b + c) measures the buoyancy, including the discretionary effect.


Thus, the elasticity of a tax equals its elasticity relative to its tax base (B) times the elasticity of the tax base relative to income. This approach may provide insight into the high elasticities of taxes that are generally considered inelastic but that have a rapidly expanding tax base. The approach is only useful, however, if separate estimates of the tax base or its relationship to income growth are available—for instance, the import estimates prepared in conjunction with the workshop on the external sector.

b. Expenditures

This section reviews two factors—inflation and exchange rate changes—that can complicate the process of estimating expenditure aggregates. While much has been written on techniques of revenue forecasting, the literature on estimating government expenditure is relatively small. Several factors may account for this imbalance. First, in some respects the best short-term projections are simple mechanical extrapolations of basic underlying accounting relationships rather than sophisticated econometric models. The wage bill, for example, is directly linked to the number of employees (as augmented by new hiring) and the general wage rate (as affected by the annual wage increment). Second, in many cases the expenditure side of the budget and particularly maintenance costs bear the brunt of any fiscal adjustment efforts, making these items notoriously subject to official intervention and difficult to project. Third, programs not amenable to discretionary changes (such as welfare entitlements) usually constitute a major share of government expenditure and are essentially autonomous.

Proportional Adjustment Technique

If the number of discretionary tax changes is large relative to the sample period but revenue estimates are available for discretionary changes D, the actual revenue collections may be adjusted to exclude these discretionary changes using the Proportional Adjustment Technique. The elasticity can then be estimated from the adjusted data using the following econometric equation:


This technique assumes that the proportional effect of discretionary changes on actual collections is unchanged from year to year. The adjusted revenue series (AT) can be estimated from actual revenue collections (T) and the estimates of discretionary changes (D) by the following procedure.

Since there is no revenue adjustment in the base year t:


The proportional effect of the discretionary change in the preceding year is:


Substituting (1) into (2):


For any year, the t–n equation (3) can be generalized to:


To assert that many government expenditure items are not immediately controllable is not to deny that they may follow discernible trends, however, or that underlying trends cannot be used to project future government expenditures. In countries that experience sudden changes in the general inflation rate, projections of nominal price changes are important to accurate forecasting of budgetary outlays, since large movements in prices can overshadow a reallocation of real resources that reflects changing public priorities. For this reason, budgetary expenditure projections must take account of inflationary trends.

Typically, changes in the consumer price index (CPI), rather than the GDP deflator, are considered the most appropriate variable in forecasting government procurement costs, as these changes also incorporate the changing prices of imported goods and services. The two most common time frames for measuring inflation rates are point-to-point and period averages. Point-to-point statistics identify the rate of inflation (increase in the CPI) between two distinct points in time (usually a year apart), such as December 31, 1989 and December 31, 1990. Period averages report the increase in the mean of the price index during a period, again usually a year. As point-to-point statistics provide a more accurate measure of inflation at discrete points in time and better identify responses (such as monetary growth) to specific policy initiatives, these statistics are normally given more emphasis in programming exercises. However, it takes longer to lower average inflation rates than point-to-point rates once inflation has become endemic. For the purposes of budgetary projections, then, inflation as measured by period averages (and for the budget year in particular) is more meaningful, as it more accurately incorporates the increase in government procurement costs over the projected year compared with those of the previous year.

The exchange rate is a second parameter that can have profound implications for the budgetary accounts. The overall effect on the various revenue and expenditure categories depends, of course, on the relative importance of items denominated in foreign currency in each category. On the revenue side, grants are generally denominated in foreign currency, and import taxes are implicitly based on the foreign currency costs of imported goods. On the expenditure side, the procurement costs of imported goods and services, such as petroleum products, are clearly related to the exchange rate, as are the servicing of foreign debt and many investment outlays. Because these investment outlays are frequently financed by external loans, which are also usually denominated in foreign currency, cost increases for capital projects related to the exchange rate often have minimal effects on domestic financing needs but can raise the overall budgetary deficit. For these reasons, significant changes in the exchange rate need to be carefully factored into the budgetary projections.

c. Financing

Decisions on deficit financing have far-reaching implications for the economy, and the potential effects should be carefully weighed. In analyzing deficit financing, governments frequently devote most of their attention to the term structure of new debt issued, with a view to minimizing future interest charges. Although the cost of budgetary financing is an important issue and should be a factor in the choice of public debt instruments, it is not the only consideration in financing decisions. Public borrowing has a special role in the economy: it is the benchmark market for financial instruments in industrial countries and the primary determinant of monetary growth in many developing economies.

One important consideration is the expansionary impact different forms of government borrowing can have on the money supply. To the extent that foreign financing is used for external purchases and thus adds to the supply of goods available to meet domestic demand, it does not augment the money supply. However, the long-term implications for debt servicing and the sustainability of the budgetary position need to be considered separately. Of the three types of domestic financing, nonbank borrowing is viewed as the least expansionary. Because financial assets are transferred from the private to the public sector, aggregate demand is merely being shifted rather than increased. However, efficiency considerations regarding the relative roles of the public and private sectors may limit the extent to which such transfers should occur. Commercial bank funds are considered potentially less expansionary than central bank borrowing, since limits on commercial banks’ lending capacity mean that government demand for credit may start to crowd out private sector demand. Finally, central bank borrowing is considered the most expansionary source of deficit financing, as it represents the creation of high-powered money. The large budgetary deficits several countries experienced during the 1980s raised justifiable concerns regarding the inflationary consequences of their methods of financing their budgets. Governments can sometimes finance large deficits with minimal recourse to bank borrowing by directing the country’s main nonbank financial institutions (such as pension funds) to invest almost entirely in government securities.

3. Structure of the Sri Lanka Budget

An activist government policy spanning several decades has imposed significant burdens on Sri Lanka’s public resources, both directly (in the level of government expenditures) and indirectly (in the form of production and consumption distortions that have impeded growth and revenue development). These pressures have been manifested in poor fiscal performance, with the budget showing large deficits for many years. The adequacy of the tax system in particular has been called into question. The current system is inelastic, and revenues would have fallen significantly as a share of GDP if the government had not undertaken a series of ad hoc discretionary measures to bolster them.5 While structural issues are key to assessing the revenue system, it is also important to recognize that the revenue effort has been sustained at a comparatively high level by international standards and that a more fundamental issue is the continued strain on public resources imposed by the high level of government outlays. In large part, these programs are the result of the long-standing practice of using public resources to achieve social objectives. To the extent that some of these programs have been funded outside the budget, the analysis would benefit from an expansion of its coverage to include public enterprise operations, but these data are not available.

In many respects, the year 1977 represented a turning point in Sri Lanka’s economic policies away from an inward-oriented economic strategy to an outward-looking one intended to encourage growth. After the first two years of adjustment, however, fiscal policy became less ambitious, and the budget stance showed less flexibility. With inflation averaging more than 10 percent per annum, most of the growth in revenue and expenditures was related to price effects, and real trends in revenues and expenditures broadly followed growth rates averaging some 4 percent a year (Table 6.1). Less emphasis was placed on government economic intervention after 1977, and expenditure levels measured as a share of GDP fell sharply (from about two-fifths to about one-third at the start of the 1980s) and then stabilized. Revenues moved in tandem, declining from about one-fourth of GDP before 1980 to slightly more than one-fifth thereafter, leaving the overall budget deficit equivalent to more than one-tenth of GDP over the period. External financing tapered off at the end of the decade, so that the government was forced to borrow heavily from the domestic financial system. The government was generally able to finance much of its domestic borrowing from nonbank sources, but on occasion bank lending approached 5 percent of GDP.

Table 6.1.

Sri Lanka: Summary of Central Government Operations, 1986–91

article image
Source: IMF Institute database.

Revenue trends have been relatively stable over the past decade (Table 6.2). Tax revenues declined from well over 90 percent of total receipts before 1980 to about 85 percent afterwards; major factors in this decline included the phasing out of export taxes under the economic reforms, low elasticity of other taxes, and rapid growth in nontax revenue sources (first in the form of higher profit remittances from the Central Bank, and later in the decade, as interest rates were liberalized, higher income from investments). After 1984, the tax-to-GDP ratio remained at about 18 percent, and the nontax revenue-to-GDP ratio at about 3 percent.

Table 6.2.

Sri Lanka: Summary of Central Government Revenue by Component, 1986–91

article image
Source: IMF Institute database.

Collection of tax arrears.

In principle, the tax system uses import duties to protect domestic industries, the turnover tax to raise the bulk of budgetary revenues, and a broadened excise duty to supplement budgetary resources and promote social objectives. In practice, one-half of all revenues are collected from two taxes (the general sales and turnover tax and import duties); each accounts for about one-fourth of total collections. The income tax and excises are each responsible for about one-tenth of revenues. This heavy reliance on trade and domestic consumption for government revenues is common among developing countries but limits the redistributive role of budgetary operations. The low elasticity of Sri Lanka’s income tax structure (as evidenced by the stable income tax-to-GDP ratio) is atypical, however, and is in part due to the small share of the relatively more progressive personal income tax in total taxes on income (the less progressive corporate income tax accounts for the remainder).6 The turnover tax is a general levy on imports, manufactured goods, and selected services that is imposed at the distribution stage; a credit system for manufacturing inputs reduces cascading effects. The regressive nature of this tax has led to the evolution of a differentiated and complicated rate structure that imposes higher rates on luxury items and lower rates on basic commodities. Collections from the tax expanded rapidly in the early 1980s when imports were added to the base, and the trade-related component now accounts for nearly half of all turnover tax revenues. Collections from domestic sources have remained nearly constant in nominal value since 1985.

Even with the rapid growth in turnover tax revenues, taxes on international trade remain the most important source of budgetary revenue, approaching 7 percent of GDP despite the decline in most export taxes. During the 1980s, export tax receipts fell from nearly 6 percent of GDP to less than 1 percent of GDP; duties on tea and rubber still provide the bulk of export tax receipts. The import tax structure is characterized by a multiplicity of generally ad valorem rates ranging from 5 percent to 250 percent. The share of import duties to GDP has nearly doubled over the decade, in large part because of the depreciation of the real effective exchange rate in the mid-1980s.

The other major categories of taxes have been stable sources of revenue. Excise taxes—mainly on alcoholic and nonalcoholic beverages, tobacco products, motor fuel, vehicles, cosmetics, and jewelry—are generally levied at specific rates, and frequent tariff adjustments are required to maintain revenue shares. The individual income tax generates only about one-third of total income tax collections, with the corporate tax accounting for the rest. The personal income tax has a progressive structure with four marginal rates ranging from 10 percent to 40 percent. Bracket ranges have not been adjusted frequently to offset the effects of inflation, so that the highest marginal rate applies at a relatively low level of income, eroding its progressivity somewhat. Nevertheless, in 1989, individuals in the top bracket (6 percent of all taxpayers) paid 65 percent of total collections. A number of factors, including the wide range of special deductions, exemptions, and reliefs have also eroded the tax base, making the personal income tax somewhat less progressive.

After exceeding 10 percent of total revenues in the early 1980s, the share of corporate income tax collections has declined by about half because of lower yields from state corporations and the provision of tax holidays to a wide spectrum of business activities. Corporate tax contributions are even more skewed than personal income tax receipts, with fewer than 300 companies accounting for over 90 percent of the total. The corporate income tax structure is fairly complex, ranging from 33.5 percent for most small companies to 50 percent for large ones, with a surtax of 15 percent.

After a period of exceptionally large development outlays in the first half of the 1980s related to several major government initiatives, capital expenditure receded as a share of budgetary spending by the end of the decade, falling from about one-half of all expenditures to about one-third (Table 6.3). Most of the increase in capital outlays had been in the form of transfers to public corporations, reflecting a continuing reliance on government initiatives to lead the growth effort, particularly in expanding irrigation and access to electricity and water. The reduction in net lending preceded the fall in capital outlays, declining from over 5 percent of GDP in the late 1970s to about 1 percent of GDP in recent years.

Table 6.3.

Sri Lanka: Economic Classification of Expenditure, 1986–91

article image
Source: IMF Institute database.

Growth in real current outlays roughly matched the expansion in the economy during the 1980s, and the current expenditure-to-GDP ratio remained essentially unchanged at the end of the decade at slightly over 20 percent. An important factor holding down current expenditures through the mid-1980s was the compression in outlays for wages and for goods and services. However, while wage costs remained low due to a significant erosion in government real wages that was only partly reversed by a review of the salary structure in 1988, purchases of goods and services doubled during the 1980s. Defense outlays rose sharply as the security situation deteriorated, quintupling as a share of GDP to 5 percent in the five years ending in 1988. The security situation clearly weakened the government’s ability to respond to emerging economic problems and contributed to a slowing of investment. One category of expenditures that reflected the move to market incentives was outlays for transfers and subsidies, which declined by about 2 percentage points of GDP (to about 6 percent) after 1980. Cuts in outlays for food stamps and subsidies were responsible for the entire reduction, but the cuts were partly retracted in 1989 owing to heightened government concerns about the adequacy of the social safety net. This concern also led to the initiation of a more broad-based poverty alleviation program. As budget deficits remained large over the decade, public debt as a share of GDP rose from 75 percent in 1978 to over 100 percent a decade later. Although the government had access to concessional external lending, interest costs as a share of GDP nearly doubled over the same interval, to 6 percent.

The 1991 budget was framed in the context of the ongoing SAF program with the IMF and envisioned continued progress in reducing the budget deficit (excluding grants) to less than 10 percent of GDP. This reduction was to be achieved despite a significantly lower ratio of revenue to GDP, reflecting the initial stages of a comprehensive tax and tariff reform aimed at improving the allocative efficiency of the tax system, in part by reducing effective rates of protection. Current expenditure was to bear the main burden of adjustment, with outlays budgeted to fall by 3 percentage points of GDP despite the costs associated with introducing an expanded welfare program and privatizing two large companies in 1990. Capital expenditure was to rise moderately to 9 percent of GDP. Budgetary financing plans provided for the repayment of nearly SL Rs 5 billion to the Central Bank, equivalent to more than 1 percent of GDP.

4. Estimating Sri Lanka’s Budget

It is particularly difficult to make budgetary projections in developing countries such as Sri Lanka, where the economic structure is evolving rapidly. The wise budget analyst adopts an eclectic approach, drawing on available information and employing a variety of techniques as circumstances suggest. The discussion that follows focuses on regression equations developed on the basis of historical data, but when structural reforms disrupt the continuity of existing relationships, analysts will have to use their own judgment in order to derive reasonable results. It is particularly important to account for the effects of new programs or exogenous developments in 1991. Simple comparisons such as trends relative to GDP and independent growth rates are convenient “reality checks” that supplement more quantitative approaches.

a. Revenues

The specific econometric estimates (based on 1980–89 data) for a number of the most important taxes are presented in Box 6.3. Data from earlier years have been omitted as obsolete, since they were based on an economic structure that disappeared after the economic reforms were initiated in 1977. The tax elasticities were estimated from revenue series adjusted using the proportional adjustment technique (Box 6.2 and Exercise 1.b) to eliminate the effects of the discretionary tax measures adopted during the estimation period. The improvements in forecasting from this process are directly related to the accuracy of the original revenue estimates. These estimates are based on the discretionary measures and are often of variable quality. Revenues and expenditures not explicitly discussed here should be estimated either separately or, where common trends exist, collectively on the basis of these trends or past shares of GDP. Throughout the estimation procedures, the analyst should remain pragmatic and impose reality checks on any projections, particularly with respect to tax revenues as a share of GDP.

(1) Income taxes

The personal and corporate income taxes should be estimated separately to reflect their different tax bases: personal income and the operating surplus of enterprises, respectively. In practice, however, enterprise data are not estimated separately, and GDP is the only available proxy variable on which to base projections. A complicating factor is the widespread practice of granting business tax holidays, which has eroded corporate tax recent years. The impact of these exemptions is particularly difficult to estimate and may affect the estimated income tax elasticity.

(2) Taxes on trade

The restructuring of export taxes in 1985 effectively terminated them as a major revenue source, and thereafter import duties made up the bulk of trade-related tax collections. The sharply different trends in the two trade taxes suggest that each tax should be treated separately. In view of the relatively small contribution of export duties, potential revenues can be estimated either on the basis of past trends with respect to GDP or the value of exports of the relevant products. Note that the equations below provide two alternative approaches to projecting import tax collections: as a function either of nominal GNP or of nominal imports.

Sri Lanka: Tax Equations

Three equations which may be used to estimate income tax collections are:

(3) General sales and turnover tax

The wide range of goods subject to general sales and turnover taxes make receipts in these categories exceptionally difficult to project. The plateauing of revenues on domestically produced goods after 1985, and the continuing robust growth in taxes on imports since their introduction in 1982, suggest that these taxes should also be treated separately. Unfortunately, information about discretionary changes in both categories is available only on an aggregative basis, making it impossible to prepare separate adjusted revenue series.

(4) Excise taxes

Excise taxes on alcoholic beverages and tobacco products are levied on a specific rather than an ad valorem basis and account for virtually all receipts in this category. To maintain their revenue share, these taxes have been adjusted on a regular basis, accounting for the wide discrepancy between actual collections and the adjusted revenue series (which excludes the effect of discretionary changes). Because these taxes are specific, real consumption is conceptually the most relevant independent variable tying the tax base to revenues. Unfortunately, the regressions yield no meaningful relationship between these variables. The best alternative approach assumes an annual increase in these taxes and applies the buoyancy estimate provided in question (c) at the end of this chapter.

b. Expenditures

The following discussion examines the key expenditure items that must be projected to determine the structural relationships that can be utilized to estimate future expenditure levels.

(1) Public sector employee compensation

The public sector wage bill should be the most identifiable component of government expenditure. Statistics on the number of employees are normally available (showing type and grade) in a census available from the public service commission, and the wage structure is a matter of public record. Typically, future recruitment requirements are explicitly identified during the budget process and matched with retirement plans; promotion policies and general wage increments, which are elements of public policy, are determined as part of budget preparations. In general, increases in the wage bill from promotions built into the salary structure are minimal, because new employees replace workers who retire. As statistics on the number of employees and average salary rates are not available here, the projected increase in each can be used to calculate the projected rise in the wage bill itself. That is, the projected growth in the number of employees (in percent) can be multiplied by the programmed percentage increase in their salaries (keeping in mind the wage policies identified above) to estimate growth in the wage bill.

(2) Current expenditures on other goods and services

These outlays constitute the government’s main operating expenses and cover mostly overhead and maintenance. They can be compressed somewhat with little immediate impact on government functions and hence are often the first items affected by expenditure controls. Over the long term, however, the goods and services these outlays provide are essential to efficient government operations, and cutting back too severely can curtail the delivery of essential services, undermine their effectiveness, and retard development objectives. School buildings are of little value without teachers, for instance, and teachers need resource materials. Similarly, agricultural extension agents need to make field inspections to provide effective technical advice, and it is more economical to repair roads on a timely basis than to delay repairs until the entire road needs to be rebuilt.

Unless special constraints are imposed on these expenditures, they can be projected to increase in line with nominal GDP. However, there are a number of special factors that should also be considered. First, it is important to recognize that some of these goods and services are imported (petroleum, for example) and that exchange rate movements can have a major impact on their cost. Moreover, if the capital expenditure program expands significantly, the new projects will require additional recurrent resources, which should be explicitly accounted for in projecting outlays for goods and services.

(3) Interest payments

The rapid expansion of public sector debt has catapulted debt service into one of the largest expenditure items, if not the largest item, in most national budgets. In Sri Lanka, public sector debt service has increased nearly threefold in five years.

Conceptually, projecting interest charges is straightforward: the domestic and foreign components are estimated separately and the outstanding stock of each charged the anticipated interest rate. Data on external and domestic debt and servicing costs are provided in Table 6.4. In terms of domestic debt, interest rates may be an element of public policy, and the interest rate used should reflect that decision. For external debt, interest rates should be projected on the basis of assumed deviations from the implicit rate calculated on the basis of 1989 foreign interest payments and these interest charges (which are in foreign currency) converted to domestic outlays through the exchange rate. Interest on debt incurred during the year as a result of any budgetary deficit is similarly estimated and combined with interest costs on outstanding debt to provide a comprehensive figure.

Table 6.4.

Sri Lanka: Central Government Debt, 1986–90

(In millions of Sri Lanka rupees, end of period)

article image
IMF Institute database.

(4) Subsidies and transfers

It has been argued that the macroeconomic implications of fiscal policies are overemphasized, to the detriment of the microeconomic implications. The budget position is important not only because the deficit is an important component of aggregate demand, but also because the budget shows clearly where the government has interfered with market forces. In that spirit, outlays for subsidies and transfers represent the costs of programs established to achieve public initiatives. Such outlays are difficult to eliminate once they have been introduced. In the absence of specific initiatives to the contrary, then, such outlays will continue to reflect previous trends.7

(5) Capital expenditures

In theory, capital expenditures are more likely to be adjusted than other types of expenditures, and some public finance scholars have proposed that capital outlays, rather than taxes, should be the primary means of making adjustments to provide the “appropriate” fiscal stance. In practice, it is not easy to turn capital projects off and on as perceptions about the appropriate fiscal stimulus change, particularly in developing countries, where external financing (often concessional) is a major source of funding. Such adjustments have, however, been a major element of Sri Lanka’s fiscal policy, regardless of the need to encourage supply-side expansion in order not to jeopardize the economy’s growth potential.

Capital expenditure for each year is set within the context of a rolling five-year investment program.8 Although the program is a compilation of Sri Lanka’s investment requirements over a prolonged period and does not take account of specific resource constraints, it is amended each year in light of both the available foreign financing and other budgetary priorities. In the last five years, Sri Lanka’s capital expenditure level has hardly varied in nominal terms; in fact, it has shown a sharp decline relative to GDP. A major element has been the reduction in capital transfers to public enterprises, largely owing to changes that made enterprises responsible for financing their capital expenditures directly and to an increase in net lending.

5. Exercises and Issues for Discussion

This section presents issues for discussion and exercises related to the material just presented.

a. Exercises

(1) Assuming that there are no exceptional measures, prepare baseline estimates for the 1991 budget. In preparing the forecast, draw on the projections of nominal GDP, inflation, and foreign trade from other workshops. The estimates for external and domestic financing should be consistent with capital flows in the balance of payments and private sector credit requirements, with bank borrowing a residual.

(2) Calculate an adjusted data series for central government tax collections in 1985-89. Use the estimates for the effects of discretionary tax measures provided in Box 6.4 and the proportional adjustment method presented in Box 6.2. Then calculate estimates for tax elasticity and buoyancy. Which is larger? What policy implications does this difference give rise to? Would a regression equation help in this case?

Effects of Discretionary Tax Measures

article image

(3) Compare the estimates for tax elasticities provided in Box 6.3 with the following buoyancy estimates obtained from regression equations that use actual tax receipts as a dependent variable.



PIT = Personal income taxes

CIT = Corporate income taxes

MT = Import taxes

S and T = Sales and turnover taxes

ECT = Excise taxes

b. Issues for discussion

(1) Discuss the assumptions and possible limitations of the proportional data adjustment method.

(2) Discuss the reasons for choosing the appropriate independent variable in a given relationship to forecast import taxes.

(3) Compare the estimation results of the different specifications for individual tax categories.

(4) Compare the estimates for central government tax revenues prepared by estimating individual revenue sources separately with the estimate generated by the regression equation (using adjusted data):


(5) Would regression equations help in estimating expenditures?

(6) Discuss the reasons the authorities have been so successful in securing domestic nonbank financing for the budget deficit.

Table 6.5.

Sri Lanka: Problems and Policies With Respect to Macroeconomic and Structural Adjustment (Fiscal and Public Sector)

article image

Appendix - Sri Lanka: Summary of the Tax System, May 1989

Sri Lanka: Summary of the Tax System, May 1989

article image
article image
article image
article image
article image
article image
article image
article image
article image
Source: Central Bank of Sri Lanka.

Companies with issued capital not exceeding SL Rs 500,000.

Thirty-three and one-third (331/3) percent if the aggregate amount of remittances is less than one-third of taxable income; if the aggregate amount of remittances is not less than one-third of taxable income, the tax rate is 11 1/9 percent.


See IMF, Manual on Government Finance Statistics (Washington, D.C.: 1986), pp. 7-29. The services provided often include some market transactions, which are usually incidental to the main government operations (except in centrally planned economies, where the government sector assumes far greater direct responsibility for economic activities).


For a discussion of budget issues related to economies in transition, sec Vito Tanzi, “Fiscal Policy and the Economic Restructuring of Economies in Transition,” Working Paper 93/22 (Washington, D.C: IMF, 1993).


However, in economies with high rates of inflation, collection lags can also become an important source, in real terms, of lost government revenues. See Vito Tanzi, “Inflation, Lags in Collection, and the Real Value of Tax Revenue,” IMF Staff Papers, Vol. 24, No. 1 (March 1977), pp. 154-67.


For details, consult Sheetal K. Chand, “Some Procedures for Forecasting Tax Revenue in Developing Countries,” Departmental Memoranda 75/91 (Washington, D.C: IMF 1975).


A comprehensive review of the tax structure in Sri Lanka is provided in the Appendix Table.


Typically, under a progressive tax structure with higher marginal tax rates for higher incomes, tax receipts increase faster than incomes as the economy grows.


In some cases, it may be possible to identify the costs of specific important subsidies by analyzing the per unit cost and then appending this information to summary estimates prepared from other programs that have been projected separately.


This program is used as a key document for the periodic (usually annual) meetings of Sri Lanka’s Aid Group.