A Decade of Civil Service Reform in Sub-Saharan Africa
  • 1 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund

Contributor Notes

Author’s E-Mail Address: Ilienert@imf.org

This paper assesses a decade of experience in civil service reform in a sample of 32 sub-Saharan African countries. Many countries have made an important start towards reducing excessive staffing levels and the nominal wage bill, but less progress has been made in decompressing salary differentials in favor of higher-grade staff. In the CFA franc zone countries, real wages fell sharply after the 1994 devaluation, but the wage bill relative to tax revenue is still high in many countries. There is a need to consolidate quantitative first-generation reforms that contribute to macroeconomic stabilization. Equally important is the need to make progress on qualitative second-generation reforms, especially remuneration and promotion policies that reward performance and measures to improve civil service management. Such policies will require strong political commitment by governments.


This paper assesses a decade of experience in civil service reform in a sample of 32 sub-Saharan African countries. Many countries have made an important start towards reducing excessive staffing levels and the nominal wage bill, but less progress has been made in decompressing salary differentials in favor of higher-grade staff. In the CFA franc zone countries, real wages fell sharply after the 1994 devaluation, but the wage bill relative to tax revenue is still high in many countries. There is a need to consolidate quantitative first-generation reforms that contribute to macroeconomic stabilization. Equally important is the need to make progress on qualitative second-generation reforms, especially remuneration and promotion policies that reward performance and measures to improve civil service management. Such policies will require strong political commitment by governments.

I. Introduction

During the 1980s, many sub-Saharan African countries recognized that their civil services were not providing public goods and services in the most cost-effective way. Wage bills were generally too high, mainly because of excess staffing. Although governments had begun to reduce their role in economic development, downsizing the civil service was slow in following suit. Ironically, despite high wage bills, civil servants were often underpaid, especially professionals in the highest grades, making it difficult for governments to retain their most competent employees.

To remedy these deficiencies, governments sought the assistance of international donors and creditors in drawing up reform programs. In particular, the World Bank helped devise comprehensive strategies for civil service reform, the implementation of which was included in structural adjustment lending programs.2 Also, since civil service wage bills were absorbing a large share (sometimes more than 50 percent) of budgetary revenues, it was recognized that a reduction in the civil service wage bill would often need to play a crucial role in programs that called for a reduction in unsustainable fiscal deficits. Especially under its Enhanced Structural Adjustment Facility (ESAF), IMF-supported programs sought to ensure that “first-generation” civil service reforms—those which contribute primarily to macroeconomic stabilization—were an integral part of adjustment strategies. Increasingly, IMF-supported programs included structural benchmarks for the total nominal wage bill, and occasionally these were supplemented by limits on civil service employment or by specific measures such as the completion of a census of government employees by a certain date.

This study examines a decade of reform experience in 32 countries, with the focus being on documenting overall trends. The paper does not attempt to model systematically why particular outcomes occurred, nor how the IMF and World Bank contributed to developments.3 The sample includes all 29 countries that embarked on IMF programs supported by ESAF arrangements before 1995, together with three middle-income countries. Appendix 1 lists the 32 countries and the caveats concerning the data. After examining the pre-1985 conditions prevalent in sub-Saharan African countries (Section II), the paper discusses the objectives of civil service reform programs (Section III). The key instruments for restructuring employment and remuneration are assessed in Section IV, before their impact on the wage bill, employment and real wages is discussed in Section V. The paper concludes with some policy proposals for future civil service reforms (Section VI).

II. Key Features of Civil Service in the Prereform Period

Prior to systematic reform of the civil service, most countries experienced increases in the size of their wage bills. During the 10-15 years before 1985, nearly half of the countries for which data are available saw their wage bills rise by more than 1 percentage point of GDP, around one-third saw little change, and a few countries reduced their wage bills relative to GDP.4 In general, increasing wage bills reflected a combination of two main elements: a rapid rise in employment, offset to some extent by a compression of wages, especially of civil servants in the upper echelons of the salary structure.

A. Rapid Increase in Employment

In the decade prior to 1985, many countries experienced annual growth rates of employment in excess of 5 percent (including Ghana, Kenya, Mali, Mauritania, Niger, Tanzania, and Togo), with some as high as 10 percent per annum. Zambia is one of the few countries whose growth (about 1 percent per annum) was constrained to below the population growth rate. The rapid employment increases reflected: (1) the large degree of government intervention in economic development, including most investment and many commercial activities, which required a large supporting civil service; (2) unclear mandates of the civil service, other governmental bodies, and state-owned enterprises (in some countries, parastatal enterprise staff were included in civil service payrolls); (3) the State was often an employer of “first resort,” with guarantees of civil service jobs for graduates; (4) the need for educating and providing health care to burgeoning populations; and (5) political patronage, whereby supporters of those holding power were rewarded with civil service positions. Combined with strong institutional barriers to retrenching civil service staff,5 excess employment inevitably resulted.

B. Salary Compression

There are three aspects to the wage compression experienced prior to 1985. First, across-the-board falls in real wages were often experienced. Outside the CFA franc countries, this was usually the result of high inflation, coupled with only modest increases in nominal wages. Few countries had formal indexation mechanisms, so that when inflation accelerated in the 1970s (mainly for oil importing countries, or those whose exchange rates were devalued to more realistic levels), there was rapid erosion of purchasing power. For example, in 1985 the average civil service salary in real terms in Tanzania was only one-quarter of its level of a decade earlier.6

Second, when wage increases were accorded, either higher percentage increases were granted to lower-level employees or the same nominal increments were provided to all staff. These policies resulted in a particularly strong erosion of management-level salaries. For example, in Zambia, in 1971 an assistant director’s salary was seventeen times the lowest-paid salaried employee, but by 1986, this ratio had dropped to less than four.7

Third, many countries had deliberately allowed the monetary value of fringe benefits to cushion the erosion of real wages due to incomplete indexation of nominal wages to inflation. For instance, the ratio of nonwage benefits in total compensation rose in The Gambia from 12 percent in 1982 to 22 percent in 1988 and in Senegal from 25 percent in 1980-85 to 43 percent in 1989.

C. Consequences of Overstaffing and Compressed Wages

The compression of both the salary level and its structure enabled governments to accommodate the rapid expansion of their labor forces. However, this had adverse consequences on the civil service. In addition to excess employment, these included:

  • A demotivated staff, a decline in work effort, absenteeism, moonlighting, and daylighting.8 Demotivation also showed up in poor civil service management and a loss of sense of responsibility.

  • Difficulties in recruiting and retaining technical and professional staff, due especially to large salary differentials with the private, parastatal, or donor sectors.9

  • Increases in nontransparent forms of remuneration, namely a multiplicity of nonwage benefits that were either paid in cash or in kind.

  • Strong incentives to accept illicit forms of incomes (bribes, using or stealing government property, sale of government services, and so forth).

  • These features of the civil service also had economy-wide consequences, including:

  • The wage bill became too large relative to government revenues. Thus, the wage bill became an increasingly important contributor to growing fiscal deficits.

  • The civil service wage bill became too large relative to nonwage expenditures. World Bank Public Expenditure Reviews revealed that there was severe compression of outlays on recurrent goods and services, resulting in teachers and health workers without materials, unmaintained roads and infrastructure, and so forth.

III. Objectives of Civil Service Reform

Stated most simply, the objective of civil service reform is to maximize the value of every dollar spent on each government employee. This aim must necessarily be set in the broader context of maximizing the productivity of all government expenditures, and achieving the most appropriate balance between wage and nonwage outlays. Like other expenditures, civil service wage outlays are necessarily subject to the constraint of limited revenues and financing. And in the medium term, civil service reform needs to be set in the context of achieving a sustainable budgetary situation, which in turn is necessary (but insufficient) to attain the outcomes of macroeconomic adjustment programs, namely high-quality economic growth,10 a low inflation rate, and a sustainable balance of payments situation.

In seeking greater efficiencies in wage outlays, governments have generally followed three specific objectives, these being to (1) downsize the civil service so that it is both affordable and in line with the role government assigns itself; (2) provide the incentives, skills, and motivation to civil service employees to enable them to provide essential public goods and services; and (3) enhance civil service management and accountability at all levels.

In pursuing these broad objectives, several intermediate targets can be identified. “First-generation” reforms—those contributing primarily to macroeconomic stabilization—focus particularly on quantitative adjustments of the wage bill, which may be necessitated by changes in macroeconomic circumstances, both external and domestic. Such civil service reforms aim to:

  • Reassess the role government wishes to play in “core” and “non-core” activities.

  • Decrease employment or redeploy staff to priority sectors, following the identification of nonessential functions and duplication within the civil service.

  • Verify that the number of civil servants is consistent with those on the payroll, and that budgeted outlays are adequate (i.e., no wage arrears for staff physically present).

  • Explore alternative options for efficient delivery of public services, by subcontracting the provision of public goods or services to the private sector.

  • Ensure that changes in real salaries are consistent with macroeconomic stability.

  • Restructure total remuneration, particularly to make noncash benefits more transparent.

These components are aimed primarily at objective (1) above. Although cost cutting by squeezing real wages has often been an element of civil service reform, beyond a certain point, it becomes counterproductive, since a continuing erosion of wages vis-a-vis private sector salaries leads to an outflow of skilled staff. The efficiency of those remaining in the civil service is lowered, demotivation increases, and there are strong incentives for uncontrolled absenteeism, moonlighting, and corruption. Thus, to attain objectives (2) and (3) above, “second generation” reforms—those that focus particularly on improving the quality of remaining civil servants—need to be pursued. Such reforms do not exclude the necessity for further quantitative adjustments, but their main objective is to improve efficiency and boost economic growth. Specifically, second generation reforms aim to:

  • Shift the focus from labor inputs to labor outputs of expenditure programs, by emphasizing the performance of each civil servant in achieving time-bound objectives.

  • Restructure remuneration in order to minimize the departure of skilled staff. This involves a narrowing of private/public sector salary differentials.

  • Change promotion and personnel management policies, which, along with the restructured salary scales, ensure that merit and initiative are rewarded.

  • Provide training to upgrade skills of existing or new employees.

  • Reassess the mix of wage and nonwage spending on productive activities. This may require new budgetary practices to allow accountable civil service managers to choose the most appropriate blend of labor and nonlabor inputs.

From the above discussion, it is clear that there are many intermediate objectives, all of which are designed to attain the above-mentioned final objectives of civil service reforms. Without accompanying this multiplicity of objectives by a clear prioritization of the most urgent reforms, it is difficult to assess the “success” of civil service reforms. Potential conflicts of objectives are particularly obvious when it comes to restructuring salaries: on the one hand, to achieve fiscal adjustment by “first-generation” reforms, real wages may have to decline if there is resistance to downsize employment; on the other hand, without a sharp increase in the real salaries of civil servants that are most prone to be “lost” to nongovernment sectors, the quality of remaining civil servants declines.

Such a conflict is also seen regarding employment. “First-generation” reforms emphasize reductions in the size of the civil service, with minimal attention paid to improving the skills of remaining civil servants. As a result, some of the most skilled staff quit the civil service prematurely. Recognizing that “first-generation” reforms may have been overemphasized, countries are now attempting to attain a more appropriate balance between desirable downward quantitative adjustments and upward qualitative adjustments.

IV. Instruments of Civil Service Reform

Over the past decade, the low-income countries of sub-Saharan Africa focused on reducing employment and real wages. This section describes and assesses the main instruments used for achieving the objectives of such reforms.

A. Employment-reducing Measures

Censuses and removal of “ghost” workers

Prior to 1985, very few countries had reliable employment data. Centralized payrolls included “ghosts” that usually were not included on the establishment register maintained by the Civil Service Ministry. In some cases, civil servants who retired continued to receive wages in addition to a pension, since the central payroll was not notified, or possibly, there was collusion between the wage-disbursing agency and the retiree. In other cases, a ghost was a civil servant drawing two salaries: one for himself and the other for a ghost worker. Ghosts were particularly prevalent among daily paid employees. In one case, an entirely fictitious government unit was drawing salaries for staff allegedly “employed” by it.

A first task of reform was to discern the true number of civil servants. To this end, some countries performed head counts of employees. Although this led to the detection of some ghosts by discovering falsified or erroneous files, these searches were usually not systematic. Moreover, since they relied on ministerial returns of employees and not on the counting of staff physically present, head counts often only partly resolved the problem. To fully rectify abnormalities in payrolls, many countries carried out a census of employees (see Table 1, column 1).

Table 1.

Instruments for Reducing or Rationalizing Employment, 1985-96

(Table shows years in which measures were taken; 89=1989, and so forth)

article image
Sources: Internal IMF documents; various World Bank reports.

Several countries had already removed employment guarantees prior to 1985.

Censuses proved to be useful diagnostic devices, since they identified ghosts and paved the way for the politically easy step of removing them. However, the implementation of censuses proved to be complex. To be effective, the number of people identified by the Civil Service Ministry needed to be cross-checked with those on the Treasury’s payroll, and durable computerized links established between civil service and payroll data. Another lesson is that censuses need to be comprehensive, covering all ministries and agencies, as well as daily paid workers (in several cases, censuses were incomplete).11

Following censuses, ghost removal occurred in more than half of the countries shown in Table 1 (column 2), and over varying lengths of time. In some countries (e.g., Benin, Comoros, and Côte d’Ivoire), these exercises took one to three years, while in others (e.g., Guinea, Tanzania, Togo, and Uganda) they took five to ten years to complete. The number of ghosts identified was usually substantial: for example, in Uganda and Ghana the equivalent of 8 percent and 4 percent respectively of the total civil service were identified.12

Restraint or freeze on recruitment

A second measure implemented with little political resistance was the application of stringent limits on recruitment. A first step in restraining recruitment is to abolish vacant posts. Several countries took such action (e.g., The Gambia, Guinea, Kenya, Tanzania, and Zimbabwe). However, the abolition of vacant posts neither reduces employment nor cuts the wage bill in the short term, although ultimately savings are made. A second relatively painless way of restraining recruitment is to remove the government guarantee of employment to new graduates (see Table 1, column 4). Many countries took such steps in the early 1980s, prior to the beginning of more comprehensive civil service reforms. In some French-speaking countries, although the guarantee was initially maintained for graduates, governments restrained student entry into tertiary education institutions. Third, as a recruitment condition, some countries introduced the requirement to pass competitive entrance examinations (e.g., Mauritania, Côte d’Ivoire, and Niger).

The preceding measures usually only result in restricting the rate of expansion of total employment. To freeze employment levels, new recruitment must be restricted to the number of vacancies arising from exits, that is, recruitment can only take place when a person retires or resigns. A “one-to-one” recruitment policy was adopted in several countries, including Burundi, Côte d’Ivoire, and Senegal. Some countries went further: new recruitment was restricted to less than the number of annual departees (e.g., Benin, where one recruitment was permitted for every three departures). In the Central African Republic, rather than applying the “one-for-three” rule in terms of staff numbers, it was applied in terms of the wage bill: for every one franc spent on new recruitment, three francs had to be saved on the wage bill in other ways, including reducing the work week from 40 to 34 hours. This allowed greater flexibility in recruiting staff and, at the same time, restrained the wage bill.

A freeze on hiring is a blunt instrument that goes further: if fully implemented, the level of employment declines. Many countries implemented freezes on recruitment (see Table 1, column 5). In Madagascar for example, the suspension of recruitment covered all sectors, and departing workers were not replaced during 1992-93, resulting in a 3 percent fall in employment. However, in most countries, priority sectors, especially education and health, were exempt. Despite the large size of these two sectors (they usually account for more than half of total civil service employment), partial recruitment freezes proved to be successful in reducing employment in some countries. However, in a few countries, unrecorded hiring thwarted downsizing efforts.

Enforcing retirement or lowering the retirement age

A number of countries enforced retirement on employees once they reached the mandatory retirement age (see Table 1, column 6). In a few instances, the quantitative impact on reducing employment was quite large. For instance in Guinea, enforced retirement at age 55 affected 4,090 employees during 1985-89 (6 percent of the 1985 ministry staff).13 Some countries experimented with reducing the statutory retirement age. For instance, Togo reduced the retirement age to 50 (from 55) or to a maximum of 30 years of service. Such measures may reduce the number of civil servants, but since state-financed pensions (possibly accompanied by departure indemnities) are usually paid to retiring staff in place of regular salaries, there may only be a marginal impact on overall public finances. This is because savings in the wage bill are partly offset by a deterioration in the finances of the public pension scheme (due to both a higher number of pensioners and lower employee contributions).

Voluntary early retirement

Early retirement schemes were extensively used to encourage employees approaching the end of their careers to quit the civil service. The following are examples of single-year departures under voluntary retirement schemes: Côte d’Ivoire, 3,600 voluntary departures in 1994 (3½ percent of total civil service); Senegal, over 1,000 in 1990/91 (1½ percent reduction). When cumulated over several years, such departures can be quite significant, for example, in Guinea-Bissau, 1,033 people or 7 percent of the total establishment left during 1992-94.14 Unfortunately, many skilled employees were induced to leave: such adverse incentive effects inevitably reduced civil service efficiency.


The politically most difficult decision is to retrench existing civil service employees who are not close to retirement age. Ideally, lay-offs should be aimed at those who have ceased to be productive and are reluctant to be retrained to acquire new skills. However, effective mechanisms to induce the departure of nonperforming civil servants have been missing from most reform efforts. Osman (1994) notes that clientism, reward for political patronage, and lack of accountability inherited from colonial times has not encouraged the laying-off of staff in Africa. It is also difficult to retrench staff in societies where the private sector’s capacity to absorb excess labor is weak, as the risk of social upheaval is exacerbated.

In countries where substantial retrenchment did occur, this was often accomplished by reducing the number of daily paid or contractual employees. Laying off such employees is facilitated by the fact that they do not have accumulated retirement and other service-related benefits. As examples, over one-half of daily-paid workers or temporary workers were laid off in the early stages of reform in The Gambia, Lesotho, Mozambique, and Uganda (see Table 1, column 8).

With the objective of retrenching permanent staff, several countries introduced voluntary departure packages. For example, in Uganda, about 4,500 people of the 160,000 reduction during 1990-96 were recipients of voluntary departure packages.15 As with early retirement schemes, it was often the most skilled employees that took advantage of these packages, which were costly to implement (see below).

Several countries carried out functional reviews of ministries and their staffing needs. These reviews focussed on desirable positions within a given ministry; by themselves, such reviews do not lead to a reduction of permanent staff. Nonetheless they provide the basis for retrenchment.

Assistance to departing workers

Voluntary retirement was induced by offering attractive severance packages, which provided a variety of financial and in-kind benefits. The main component of a retiree’s departure package was usually conditional on length of service; the age of the volunteer was also a factor. The service-related amount received varied considerably between countries: from 1.2 months for each year of service (The Gambia) to 4 months (Tanzania, capped at 40 months total severance payment). In some countries, the severance payment was a flat amount for those eligible (e.g., 12 months in Guinea-Bissau). In many countries, the financial component alone accounted for 3-5 years’ salary (see Table 2), which is rather generous. In addition, other benefits such as loans for establishing small businesses, employment counseling, training, food-for-work programs, and so forth, were available to departing workers. However, these programs were relatively small, and on occasion, they were not implemented (e.g., the nonpayment of reinsertion loans in Mali accentuated social unrest among departees).

Table 2.

Size of Severance Payments, Selected Countries

(Applicable to a tenured employee with 20 years’ service)

article image
Sources: Table 7.4 of Nunberg (1994); de Merode and Thomas (1994); IMF documents; Tanzania (1996); Uganda (1994)

The main component in 1993 was a flat K Sh 60,000 “golden handshake”; for staff in G grade this provided 15-25 months’ salary; for those in E grade, it was 25-38 months’ salary. In lieu of notice, 3 months’ salary was also paid. In addition, gratuities and pensions were paid to pensionable staff.

Maximum of 60 months is for lowest-grade staff below 48 years; for staff aged 48-55 years, maximum is 48 months’ salary.

The up–front lump sum payment for a pensionable Chief Personnel Officer who opts to take two-thirds of his/her pension immediately upon abolition of office (the remaining one-third would be taken as a pension when eligible age is reached). Other benefits include: three months’ free rent, three months’ “notice,” plus allowances for transportation, training, and severance.

During 1985–90, governments generally bore the cost of severance packages from their own budgets. However, in the 1990s, severance payment costs have typically been paid by external donors,16 who consider that a one-time downsizing of the civil service is a worthwhile investment in improving civil service efficiency. The intervention by donors generally allowed severance payments to be paid on time (some countries had experienced payment arrears). To the extent that donors are willing to finance expensive departure packages, governments have little incentive to reduce their cost. However, in some countries, the costs of severance packages became excessive (e.g., Zambia) and donors were not willing to finance departure packages until costs were trimmed.

Eligibility for severance packages was not conditional on performance. Thus, several countries’ civil services lost highly-skilled workers. A challenge for countries is to reduce the cost of severance packages and change their design so that only poor performers are provided with an incentive to depart. High-quality staff should be provided with performance-related remuneration, as an incentive to remain in the civil service.

Redeployment within the civil service, decentralization, and contracting out

With the decision to switch from a policy of state intervention to one where private sector ownership is promoted, there was an urgent need for a fundamental review of the functions of the civil service. About half of the countries conducted a comprehensive review of job positions, known variously as “efficiency audits” (e.g., Ghana, 1994-97), “management audits” (e.g., Malawi, 1995-97), and “organizational audits” (e.g., Senegal in 1994) (see Table 1, column 10). Following audits, some countries established a new organizational structure for the entire civil service (e.g., Guinea in 1987-90), whereas others reorganized specific ministries on a piecemeal basis.

Following reclassification, governments adapted recruitment and departure policies in an attempt to ensure that the number of positions was equated with the number of staff. This involved staff redeployment and layoffs. However, following functional audits, the process of eliminating overlapping or obsolete functions was slow. For example, in only a few countries was action taken to reduce the number of ministries (e.g., Sierra Leone, from 22 ministries to 15; Uganda, from 34 to 21), by merging those with overlapping functions. Also, only in a few cases were some embassies closed (e.g., Senegal in 1991/92).

In some countries (e.g., Ethiopia, Uganda), decentralization of government activities to local government has been a feature of recent reforms, the aim being to bring government policies closer to the people. Some donors regard this as a feature of good governance, and are financing decentralization of operations, especially in the social sectors. Such policies do not necessarily reduce the total government wage bill (i.e., central plus local governments): if administrative structures at the center are not reduced, there is a risk of duplicating overhead at central and local government levels. But if only central government wages and employment are monitored, it may appear that downsizing is occurring. In contrast, the removal of state enterprise employees from the civil service payroll is a downsizing of total general government in a statistical sense; however, such a change has little impact on overall public finances unless subsidies to public enterprises are removed simultaneously. In the case of Guinea, civil service and public enterprise reform proceeded in tandem: the impact of downsizing was not offset by higher subsidies to public enterprises.

The contracting out of certain government activities has reduced the wage bill in several countries. For example, Côte d’Ivoire began hiring private-sector services to replace daily paid workers, and Zimbabwe contracted out cleaning and catering services (difficulties were experienced in attempting to contract out the maintenance of government vehicles). Such subcontracting results in efficiency gains provided it is delivered competitively, that is, the bidding for government contracts is truly competitive, and there is neither collusion nor corruption in the transfer process. In like fashion, the selling of “government department” enterprises (e.g., the government printing press in Uganda), whose employees were previously civil servants, results in permanent budgetary savings to the extent that greater competition reduces the per unit price of the same goods and services.

Military demobilization

Although this study focuses on civil service reform, in some countries, the total wage bill was reduced significantly because of army demobilization (e.g., Mozambique, Sierra Leone, Togo, Uganda, and Zimbabwe). To the extent that laid-off soldiers are not recruited into the civil service, army demobilization does not thwart civil service reform efforts. This study has not been able to identify the extent to which laid-off military personnel have been reintegrated into the Ministry of Defense or other ministries.

B. Restraining Wages and Rationalizing the Pay Structure

When reconsidering the size of the wage bill, governments may place emphasis on either: (1) restraining the growth of (or enforcing declines in) real salaries per employee; or (2) deliberately increasing real wages of selected groups, especially those in the upper echelons, so that the revised salary structure rewards effort and competence, and minimizes the gap with comparable wages in the private and parastatal sectors. These two areas need not be mutually exclusive: real wage increases for higher-grade staff could, in principle, be offset by real wage declines for lower-grade employees. However, this does not appear to have been the case in any of the countries under review. Alternatively, increases for higher-grade staff could be “paid for” by declines in civil service employment. This was the case in the few countries where salary decompression featured prominently in reform programs. In changing the level and structure of remuneration, the following instruments were used:

Wage freezes or partial indexation

To reduce the size of the wage bill, regular general wage adjustments can either be suspended temporarily or reduced to below the inflation rate, so that real wages fall. Most countries took such actions, and some low-inflation countries reduced nominal salaries across the board (see Table 3, columns 1-2). The policy of not totally offsetting price increases was the main instrument used to reduce the size of the real wage bill in the CFA franc countries following the devaluation in early 1994.

Table 3.

Instruments for Reducing or Rationalizing Remuneration, 1985-96

(Table shows years in which measures were taken; 89 = 1989, and so forth)

article image
Sources: IMF, World Bank Reports.

Compared with retrenching civil servants, the erosion of real wages is politically less costly. Although successful in meeting the short-term cost-reduction objective, a major danger with this policy is that a wage explosion may occur in later years, especially in countries where civil service trade unions exercise considerable power (e.g., Malawi, Zambia, and Zimbabwe). Moreover, such wage restraint can accentuate the compression of salaries, especially if there are lower percentage increases for higher-income earners. Although the longer-term costs of such short-term expedients should therefore be carefully considered, prolonged wage freezes or declines in real wages were successful in permanently reducing the wage bill in countries where wage drift was significant (see next subsection), or where a large external shock demanded a sharp reduction in the wage bill.

Limit automatic increments

In many countries, especially in francophone Africa, automatic “within-grade” increments were granted to all civil servants on the anniversary date of their appointment. Such increments were often more important than those due to promotions (see Robinson, 1990), which also tended to be automatic. Automatic increments—whether they are “within-grade” or promotion related—lower the morale of well-performing civil servants and encourage a slack work ethic. Such increments may also neutralize the impact of other measures such as limits on recruitment. In several countries they were used for circumventing general wage freezes. For example, in the CFA franc countries, the continuation of real wage increases was virtually entirely due to wage drift (e.g., Senegal, where wage drift was about 4 percent per annum in the late 1980s).

In recognition of these shortcomings, countries such as Mali and Senegal departed from such practices early in their reform programs (see Table 3, columns 3-4); other countries (predominately in West Africa) followed suit in the early 1990s. The suspension of automatic increments not only renders it possible to contain the growth of the wage bill, but it also leaves room for granting higher meritorious increments to well-performing civil servants. The suspension of automatic increases based solely on length of service is therefore an integral element of rationalizing the salary structure.

Consolidating nonwage benefits into salaries

The provision of nonwage benefits to permanent civil servants was a widespread practice in sub-Saharan Africa. Many of these benefits were in-kind, including subsidized housing, transportation benefits (free car, driver, and gasoline), education allowances, medical allowances, servants, and uniforms. Prior to reforms, they often formed a significant component of total compensation. For example, the cash value of such benefits amounted to 41 percent of average total compensation in Central African Republic (1987), while for top-grade employees in Tanzania, the monetized equivalent of all fringe benefits were more than ten times basic monthly pay (Nyirabu and Mudzonga, 1993). In Eritrea, top government officials received food rations and basic housing, but no salary (Wescott (1994), pp. 223-24). In addition, some top civil servants received honoraria for being placed on Boards of Directors of public enterprises, or were provided with generous allowances when on official travel.

With such substantial nonwage benefits, the value of total compensation is no longer transparent,17 since their monetary value is not known with accuracy. Another non-transparency was the multiplicity of benefits. For instance, in Senegal, 175 of the 250 “Items of Pay” were supplements to base pay; in Zambia, housing benefits were provided either in kind (government-owned housing) or in cash (for employee-owned or rented accommodation). Moreover, nonwage benefits are inequitable on two counts. First, the bulk of nonwage benefits were received by senior civil servants based at ministry headquarters. Although less senior staff were entitled to nonwage benefits, they often did not receive them. Second, nonwage benefits are a source of untaxed income for higher-income civil servants: the average tax rate on total remuneration is less than that of lower-grade employees not receiving such benefits. Another problem was that when they were monetized during reform programs, insufficient budgetary resources were available for their payment. Finally, nonwage benefits are subject to abuse: for example, government-provided cars can be used for personal excursions; in-kind food can be sold for profit on markets; sick leave can be claimed when urgent moonlighting needs arise.

For these reasons, a desirable first step in salary rationalization is the monetization of in-kind benefits and the integration of most of them into a simplified compensation structure. Given their high monetary value, most countries experienced difficulties in integrating them with monetized salaries. For example, in 1986 Sierra Leone abolished the provision of subsidized rice to all civil servants, but maintained it for security personnel; Kenya abolished the special accommodation and medical allowances, but maintained other allowances. In some cases, during the early stages of reform, nonwage benefits were increased even further to offset the previous erosion of cash wages. For example, the rent and transport allowances were raised in Tanzania in 1988/89, the residency and transport allowances were increased in Niger during 1988-92, and Benin introduced noncash wage payments in 1989. Another problem was the valuation and transparent selling of government assets (houses and cars) which were providing in-kind income.

Since the beginning of the 1990s, countries have acted more aggressively to reduce fringe benefits (see Table 3, column 6). For example, Burkina Faso cut the housing allowance by half (1991-93), Côte d’Ivoire compressed special benefits (1991-93), and Uganda monetized all in-kind benefits. A few countries, including Ghana, Guinea-Bissau, Sierra Leone, and Uganda, have been successful in consolidating nearly all nonwage benefits into total remuneration (a “clean wage” policy).

Wage decompression

Decompressing salary differentials is a necessary step to ensure that highly skilled staff do not quit the service in favor of higher private sector wages. It is also necessary (but not in itself sufficient) to reduce bribe taking.18 In principle, a new decompressed salary structure should be introduced following a functional review of the civil service that redefines job specifications. However, in most countries completing such reviews, there were relatively long delays before rationalized salary scales were introduced. Indeed, in many countries, including nearly all CFA franc countries, where real salary compression has been severe since 1994, there has been very little attempt to widen salary differentials. In contrast, a few countries explicitly set the objective of decompressing salary differentials in favor of the higher echelons. For example, Ghana targeted an increase in the ratio of the highest to the lowest salary. After raising it to about 6:1 in 1986, the Government increased it much further, to 11:1 by 1992.

There are three key factors in determining how wide the salary structure should be: (1) the extent to which private sector total remuneration exceeds that for comparable civil service jobs; (2) the maintenance of adequate differentials between starting wages and top salaries, so as to reward effort and performance of career civil servants; and (3) the affordability of decompression. In countries with large fiscal imbalances, a prudent policy is to decompress those salaries where there are obvious shortages of skilled manpower (e.g., Lesotho and Uganda introduced special skills bonuses for employees in short supply), consistent with an improving overall fiscal position.

Performance-related promotions and pay

Since the mid-1990s, many countries have set as an objective the introduction of performance-related pay. For example, in Mozambique, one of the aims of the 1995-97 program included “the introduction of career streams with productivity incentives, merit promotions, and performance bonuses.” In Uganda, the 1995-98 program aims at developing “a performance recognition system to identify and reward outstanding performers.”

Certain preconditions for implementing performance-related promotion and pay have to be fulfilled, namely that it must be based on a system that: (1) includes a method of measuring and quantifying performance; (2) is applied equitably and consistently to all staff; (3) includes checks and balances in the form of independent “judges” involved in assessing individuals; (4) enforces discipline (e.g., being on duty for the required number of hours); and (5) is transparent and simple to administer.

Progress in introducing performance-related pay and promotion has been limited. This is unsurprising, since the final preconditions—transparency and simplicity—are not easy to implement: there are considerable administrative requirements, and staff need to be well trained in performance management. Nonetheless, efforts in this direction are being made. For example, besides the above-mentioned “scarcity” allowances, “retention” allowances have been paid in The Gambia, Tanzania, Zambia, and Zimbabwe; and Uganda introduced A-B-C ratings for staff during the downsizing exercise. Few countries have introduced performance indicators based on tangible indicators of output (e.g., for the health sector, the number of vaccinations by week).

C. Improving Civil Service Management

Civil service reform is not confined to cost containment and rationalization of pay and employment: institutions need to be developed to support ongoing reviews of civil service efficiency. In particular, there is a need to train managers, upgrade the skills of existing staff, enhance accountability by improving record keeping and establishment control, institute staff audits, revise the legal framework governing the civil service, and apply modern technology to public sector management.19

Progress in these areas has been relatively limited, the main exceptions being the establishment of computerized links between the payroll and establishment registers in several countries, and the completion of the first rounds of functional reviews.

V. An Assessment of a decade of reform

Having examined the various instruments that have been used in civil service reforms, this section assesses the impact of wage and employment policy changes on the size of the wage bill, real wages, and employment in 32 sub-Saharan African countries. The 29 low-income ESAF countries have been subdivided into three groups: the 11 countries of the CFA franc zone, and two groups for the 18 non-CFA franc countries: 9 “high wage bill” countries and 9 “low wage bill” countries.20 This was done since countries with very low wage bills could justifiably increase the size of the wage bill if excessive wage compression was a strong impediment to civil service efficiency. On the other hand, countries with excessive wage bills necessarily need to reduce excess employment and, possibly, inappropriately high salaries.

A. Adjustment in the Nominal Wage Bill

For the average of the 32 countries, there has been a modest reduction in the nominal wage bill relative to GDP: over the past decade, the decline has exceeded 1 percentage point of GDP, from nearly 7 percent to just under 6 percent of GDP (Figure 1 and Appendix Table 4). This level is still high compared with other developing countries: in both Asia and Latin America, the ratio is around 5 percent of GDP.21 However, there is a large variance around these averages. At one extreme is Uganda, whose wage bill has risen from 1 percent of GDP in 1987 to 3 percent in 1996, a development which has been facilitated by a buoyant tax performance over the past five years. At the other extreme is Lesotho, whose wage bill was reduced to 15 percent of GDP by 1996 (in 1997, Zimbabwe’s wage bill is estimated to jump to 15 percent of GDP).

There was virtually no change in the average wage bill relative to GDP for the 32 countries until 1993. Nonetheless, prior to the early 1990s, some of the early participants in SAF-supported IMF programs (notably The Gambia, Guinea-Bissau, Madagascar, Mozambique, Sierra Leone, Tanzania, and Uganda) lowered their wage bill by an average of more than 1 percentage point of GDP. During this period, the CFA franc countries witnessed a rapid rise in the size of their wage bills, particularly the three stand-by countries. However, following the devaluation of the CFA franc in early 1994, the wage bill in the 14 CFA franc countries fell on average by nearly 3 percentage points of GDP, with a dramatic decline in the three stand-by countries.22 Similar developments occurred in the share of the wage bill in total current expenditures (see Appendix Table 5). However, despite the downward adjustment, the wage bill in the CFA franc countries with ESAF programs was still absorbing nearly 41 percent of total current expenditures in 1996, in contrast to only 33 percent in the non-CFA franc countries.

The ratio of the wage bill to revenue provides a measure of the affordability of the wage bill: unlike capital expenditures, which are often financed by donors in sub-Saharan Africa, salaries are paid out of domestic resources. In the 32 countries under review, salaries took an increasingly larger share of tax revenue during 1986-93, absorbing over 46 percent of total revenue in 1993 (Appendix Table 6). Since then, sharp declines in the CFA franc countries have been witnessed, bringing their average down from over 60 percent in 1993 to under 40 percent in 1996, with a particularly strong fall in the three stand-by CFA franc countries. Nonetheless seven CFA franc countries were still devoting more than 40 percent of total revenue to salaries in 1996. Interestingly, salaries in the “high wage” non-CFA countries were absorbing much less revenue throughout the entire period than in the CFA franc countries (Figure 2). It may be concluded that the wage policies adopted following the CFA franc devaluation facilitated a much desired reduction in the wage bill, whose size is now somewhat closer to that in the other sub-Saharan African countries. Nonetheless, wage bills are still relatively high compared with domestically-generated resources, notably taxes. Part of this problem lies in the low capacity of CFA franc countries to generate revenue.23

Figure 1. Government Wage Bill as a Percent of GDP 1/
Figure 1. Government Wage Bill as a Percent of GDP 1/
Source: IMF1/ See Appendix 1 for classification of countries
Figure 2. Government Wage Bill as a Percent of Total Revenue Excluding Grants 1/
Figure 2. Government Wage Bill as a Percent of Total Revenue Excluding Grants 1/
Source: IMF1/ See Appendix 1 for classification of countries

B. Adjustment in Real Wages

On the basis of a more limited sample of 23 countries, the average salary per civil servant as a multiple of GDP per capita rose until 1993 (See Figure 3 and Appendix Table 7). Since 1993, the ratio has, on average fallen from 8.3 to 7.6 in 1996. This compares with a ratio of about 3 in Asia and Latin America.24 The downward adjustment of wage levels relative to GDP per capita in Africa is mostly explained by developments in the CFA franc countries. In contrast, since 1990, this ratio has increased in the non-CFA franc countries, mainly reflecting large real wage increases in a few countries (see below).

Figure 3. Government Wage Bill per Employee as a Multiple of GDP per Capita 1/
Figure 3. Government Wage Bill per Employee as a Multiple of GDP per Capita 1/
1/ See Appendix 1 for classification of countries

This ratio provides a first indication of the extent to which average civil service salaries are out of line with those in countries at a similar stage of economic development. However, its size depends not only on the level of civil service salaries but also on the structure of GDP, employment (labor force participation), and population.25 In subsistence economies such as Burkina Faso, Mali, and Chad, which have seen a considerable fall in this ratio, civil servants’ salaries are still 12-14 times GDP per capita. In contrast, in the three stand-by CFA franc countries, all of which have oil sectors (which boosts GDP per capita), this ratio has declined from 5 to 3¼ during 1992-96. Thus, when using this indicator to assess the adequacy of average civil service salaries relative to another country, the comparison should be confined to countries with similar economic structures and per capita incomes.

Real wages per civil servant continued to fall in most countries during the past decade. In the 20 countries excluding Uganda for which continuous data are available since 1990, the fall has averaged 2 percent per annum (Appendix Table 8), with a particularly sharp fall in the CFA franc countries after 1993. However, around this average, there has been a wide diversity of experience: in 6 countries, there were increases, ranging up to 4 percent per annum (Lesotho) during 1990-96, whereas in the 14 other countries, there were declines averaging up to 12 percent per annum (Cameroon). In some countries (e.g., Lesotho and Malawi), real wages rose because governments were unable to resist large real wage increases in particular years.26 However, in a few countries, especially Uganda, a conscious effort was made to increase real wages from low levels.27

In the majority of countries, where real wage declines occurred, this was usually achieved by granting limited across-the-board nominal wage increases in the face of high inflation. In several CFA franc countries, where inflation typically accelerated to 40-50 percent in 1994 following the devaluation, the nominal wage rise was limited to 10 percent; in the case of Cameroon there was no nominal increase (following declines in nominal salaries in late 1993 of 35-54 percent, depending on grade); and in Congo, following a wage freeze during 1994, nominal wages were reduced by 15 percent in 1995.

With average real wages continuing to fall in the majority of countries, it is pertinent to examine whether there was any decompression of upper echelon wages during the period, as this was an objective of reform in many countries. The available evidence suggests that there was little desirable change in the salary structure in the countries that experienced decreases in real wages. On the contrary, in some of the countries that witnessed the strongest declines in real wages, there was a further compression of upper-grade salaries. For example, in Cameroon, the highest civil service salary was reduced to 6 times the lowest salary in 1994, compared with 10 times during 1984-92.

In contrast, in countries where the salary structure was extremely compressed (e.g., Ethiopia and Mozambique), rapid decompression featured prominently in the early stages of reform.28 Other early reformers (e.g., The Gambia, Ghana, Guinea, and Uganda) set targets for decompression of the salary structure, and these were generally attained. In these countries, there were strong real wage increases throughout the decade (see Appendix Table 8); yet by 1996, their wage bills (as a percent of GDP) had either not risen significantly or had risen to levels still well below the African average. This was usually because the salary increases were accompanied by significant falls in employment (see below).

Progress in reducing differentials between civil service and private sector salaries has been limited. In making such comparisons, caution is required, because: (1) civil servants’ total remuneration includes both cash and in-kind benefits, with the latter particularly difficult to quantify; (2) civil service jobs and functions may not have any direct private sector comparators; (3) civil servants enjoy security of employment,29 which provides an additional benefit that is difficult to quantify; and (4) the formal private sector is small in some countries: the more appropriate comparator is either parastatal enterprises or international development agencies. For example, in Lesotho, the salary scale mid-point in the electricity company was 40 percent higher than in the civil service. Similarly, in Malawi, support staff and middle grade employees in the donor sector were found to earn three to six times comparable jobs in the civil service (Price Waterhouse, 1993).

In making comparisons between civil service and nongovernment remuneration, there are two options: (1) to make direct comparisons for selected professions such as accountants, engineers, computer programmers (i.e., the upper/middle echelons of the civil service) as well as for messengers, chauffeurs (lower echelons); or (2) to compare average salaries (however, representative data on private sector salaries are scarce in Africa). The former is clearly preferable to the latter, particularly if both monetized and in-kind benefits are captured. Unfortunately, under the civil service reform programs put in place in the past decade, neither method has been used systematically to monitor differentials; only a few “snapshots” are available.

For seven CFA franc countries, average civil service remuneration at the beginning of the 1990s, inclusive of both monetized and in-kind benefits, was higher than private sector remuneration. This differential ranged from 26 percent in Mali, to around 100 percent in Côte d’Ivoire and Senegal (see Lachaud, 1996).31 However, Lachaud finds that most of the public/private sector salary gap is attributable to the characteristics of individuals, especially their higher level of education and training, and their experience (measured by age). When adjusted for these factors, civil service wages were also found to be on par with those in state enterprises (except in Burkina Faso). However, the devaluation induced a sharp decline in civil service real wages in CFA franc countries and improved the capacity of private companies (especially those exporting) to pay higher salaries; anecdotal evidence suggests that there is now a negative gap between civil service and private (or parastatal) salaries in these countries, particularly for staff in the upper grades.

The available evidence for the non-CFA franc countries is mixed. Schiavo-Campo, et. al. (1997b) provides evidence that civil service salaries are on par with those for similar skills in the private sector. In other continents, government jobs carry a 10-20 percent “discount,” reflecting greater job security. However, some individual country surveys show that average base salaries in the civil service are somewhat higher than those for comparable jobs in the private sector. This is partly due to the fact that, in some countries, remuneration at the lower end of the salary scale is higher than for nongovernment salaries, especially when the impact of housing or other nonwage benefits is included (e.g., Malawi,32 Zambia, and Zimbabwe). In this context, in the process of reform, some countries monetized nonwage benefits and certain cash benefits were integrated into a revised base salary structure. This had the effect of widening the measured differentials between lower- and upper-grade base salaries, since usually only higher-grade staff were eligible for nonwage benefits.

C. Adjustment in Employment

Limited employment data prevent a full analysis for all countries: only 19 countries have continuous time series covering 1990-96, with the quality and comparability requiring these data to be treated with considerable caution.33 With these caveats, employment declines during the past decade occurred in about one-third of the countries, and in six countries (Benin, Central African Republic, Guinea, Madagascar, Mali, and Uganda), the number of civil service servants in 1996 was more than 10 percent lower than a decade earlier (Appendix Table 9). Downsizing became more widespread in the early 1990s in the CFA franc countries, with most of them witnessing marked declines in staff (Burkina Faso and Chad are exceptions). Outside the CFA franc zone, only in Guinea, Sierra Leone, Tanzania, and, especially, Uganda, was there a significant downsizing during 1990-96. In other countries, despite plans to reduce total employment, this has generally not occurred.

Initially much of the downsizing was achieved by removing “ghosts” and temporary workers. More recently, an increasing proportion of the employment declines have stemmed from early retirement or voluntary retrenchment with generous severance payments. Although many countries have discussed strategies for a more limited role for government, only a few have closed a significant number of ministries.

These employment declines must be seen against the continuing rapid population growths of 2-3 percent per annum: there has been a marked decline in the ratio of civil service employment to population. By 1996, only 1 percent of the population were civil servants (Figure 4), which is considerably lower than in other developing countries, where nearly 3 percent of the population is employed by government.34 This reflects, inter alia, constrained budgetary resources and the burgeoning youth population in sub-Saharan

Figure 4. Civil Service Employment as a Percent of Population 1/
Figure 4. Civil Service Employment as a Percent of Population 1/
Source: IMF1/ See Appendix 1 for classification of countries
Africa.35 In education alone, which usually accounts for some 40 percent of the civil service in sub-Saharan Africa, student/teacher ratios are considerably higher than in other developing countries.36 However, this does not imply that educational quality is necessarily lower: research on student performance suggests that the proportion of outlays devoted to teaching materials is a better guide to quality. In sub-Saharan Africa, often 10-20 students share one textbook. This suggests that recruitment of teachers to public schools may not necessarily result in an improvement in the quality of education. Yet, as noted earlier, many countries excluded education from recruitment freezes.

VI. Conclusions and Directions for Future Policies

Most sub-Saharan African low-income countries have made only limited progress towards achieving the objectives of civil service reform discussed in Section III. Although many countries have made important beginnings in restructuring the civil service, so that its size is becoming more consistent with a less-interventionist state, less progress has been made in providing more appropriate incentives to skilled staff through revised remuneration and promotion policies. Only a few countries have persistently and boldly pursued comprehensive reforms that have resulted in marked declines in employment, accompanied by a strong decompression of the eroded salaries of higher-grade civil servants.

Although employment reduction has occurred in many countries, especially when viewed relative to rapidly expanding populations, large differentials of employment/population ratios across countries remain, suggesting that there is still further scope for downsizing. The ministries in which this could occur could be guided by regular functional reviews, as well as sectoral indicators such as pupil/teacher ratios, patient/health worker ratios, taxes collected per tax collector. Although there is no optimal size of the civil service, the role the state can play is effectively constrained, inter alia, by the amount of tax revenue that it can collect. But even in countries like Uganda, where revenue performance has improved markedly from a very low level and where the civil service has already declined very significantly, the authorities are continuing to find scope for further downsizing. Hence, nearly all countries may be able to improve public sector productivity by further retrenchment, especially of staff whose contribution to the delivery of public goods and services is limited. This could occur by:

  • Ensuring that all “ghosts” are eliminated. Countries that have not completed censuses must take this essential first step. In countries where at least a first census has been taken, “ghosts” continue to reappear, even after several years of reform. This suggests that further efforts should be made to achieve full coherence between the numbers registered on the establishment database and those being paid. Only once this consistency is achieved can the monthly payroll be used with confidence for the annual budget projections of the wage bill.

  • Being bolder in restructuring government. Much of the downsizing so far has been either by removing “ghosts” or laying off temporary workers. More recently, donor— financed voluntary retrenchments have also contributed to outflows of staff approaching retirement age, many of them skilled employees. Only a few countries—those which experienced the strongest employment declines—took bold actions such as closing about one-third of their ministries.37 To pursue this avenue, strong political commitment to the reform process is an essential ingredient.

  • Re–examining the policy of excluding education and health from recruitment freezes. In particular, there may be scope for greater reliance on private sector provision in these two important social sectors. In education, in some countries, families prefer to pay a higher price for primary education at private schools, whose quality is often higher than in state schools. In health, one country (Comoros) removed all health workers from the government payroll. By confining the government’s role in health to the reimbursement of private providers of a limited number of essential medical services, countries outside Africa have downsized their civil services substantially.38

  • Retrenching nonperforming civil servants. Countries should focus on designing mechanisms for assessing employee performance objectively, then proceed to institutionalize arrangements for laying off nonperforming staff.

  • Reducing the costs of severance packages. The provision of up to 5 years’ salary plus other benefits is a strong incentive for capable staff to leave the civil service. Less generous departure packages should be aimed at poor performers, and should exclude senior staff with management or technical skills.39 In some countries, civil service pension laws need to be reformulated so as to reduce the incentive for skilled workers to retire prematurely.

Total remuneration in many countries still needs to be restructured comprehensively so as to enhance transparency and improve ability to recruit and retain skilled staff. Since salary decompression has not been widespread, there is an urgent need to introduce differentiated real wage changes, with increases in favor of those in the higher grades, especially those in professions who quit because private sector wages are higher. In this respect, there is a need to:

  • Direct the largest proportion of general salary increases to those in the higher grades. In low-inflation countries, this may imply that the totality of any nominal increase is allocated to upper-grade staff.

  • Reduce lower-grade remuneration in countries where it exceeds private sector equivalents.

  • Systematically monitor total remuneration at various levels of the civil service relative to nongovernment jobs that require similar skills and experience.

  • Replace automatic “in-grade” salary increases and promotions that are based solely on seniority by transparent policies that reward the most competent staff on the basis of performance (whatever their age) and penalize poor performers40

  • Complete the monetization of in-kind nonwage benefits and integrate most cash allowances into the base salary structure. The payment of special bonuses to attract (or retain) skilled staff are appropriate interim steps towards a rationalized structure.

A more general issue is the trade-off between “first-generation” reforms—cost-containment measures directed at reducing employment and restraining real wage growth—and “second generation” reforms that improve the incentive structure of the civil service. To date, most progress has been in quantitative adjustment and this has often been to the detriment of the quality of the civil service. One guideline for deciding whether emphasis should continue to be placed on further quantitative adjustment is to examine the extent to which the fiscal constraint is binding. If, for example, a country’s wage bill still exceeds, say 40 percent of budgetary revenue, the focus on cost containment should continue, although when reviewing wage policies, greater attention needs to be placed on improving the structure of total remuneration. On the other hand, in a country that has a relatively low wage bill/revenue ratio, and wage outlays relative to nonwage expenditures are considered appropriate, the focus should principally be on increasing salaries for skilled professionals, especially when the gap with private sector salaries is large.

Although continuing reform efforts need to be differentiated according to the priority attached to the three reform objectives discussed in Section III, all countries need to devote increasing attention to improving the incentives that increase the efficiency of the service, and ultimately economic growth. The most urgent needs are probably to increase higher-grade remuneration and reward good performance, thereby boosting staff morale. Since there is strong evidence that low salaries are highly correlated with civil service corruption,41 reform programs also need to address factors impacting on corruption, such as the lack of explicit performance standards, highly subjective recruitment and promotion procedures, and lack of administrative sanctions (e.g., for failing to respect civil service codes of ethics, or not being in the workplace for the required number of statutory hours).

More generally, reforms are needed to improve civil service management. There are considerable merits to decentralizing management in order to increase managers’ autonomy, including the choosing of labor and nonlabor inputs, and even the setting of wages. However, there are dangers of doing this prior to establishing firm control of recruitment and the payroll at the central level.42 Assistance from the World Bank and other donors will continue to be essential in designing and financing comprehensive reform strategies, including more appropriate civil service regulations; new policies to ensure that personnel selection, training, and compensation are based on merit, not on connections; enhanced accountability of civil servants, especially for those working in sectors where the temptation to accept bribes is more prevalent (e.g., customs and tax administration); the impact of reforms on civil service pension systems;43 and budgetary reforms to increase the transparency of government spending (e.g., the awarding of government contracts).44 Finally, the ultimate success of civil service reforms will depend on political willingness to implement such broad-ranging policies.


I. Country Classification

article image

II. Sources of Data and Caveats on their Use

The data are derived from IMF sources, which in turn are received from national authorities. Care is required in using them, since both the government wage bill and civil service employment data are prone to consistency and coverage problems.

The raw data underlying Tables 4-7 of this appendix is the wage bill in national currency terms, divided by different aggregates, also in national currency terms. The wage bill generally covers gross cash remuneration, inclusive of income taxes as well as both employer (government) and employee social security (pension) contributions. Although it would be preferable to exclude the salaries of the military, this generally has not been possible (Ethiopia is one exception). Stevens (1994b) shows clearly the coverage problems of the “civil” service; to the extent possible, the data include the police and other internal security, all education sector employees (university staff are sometimes excluded), and daily and other temporary workers. Ideally, the data should include all levels of general government, but in nearly all cases, regional or local governments are excluded. Since the wage bill is available for fiscal years, the data are shown under the calendar year if the fiscal year begins before June 30, otherwise it is shown in the following calendar year.45

In contrast to the wage bill, employment data (see Table 9) generally exclude the military, and in some cases, internal security and diplomats as well (e.g., Benin). Thus, average wages per employee (and hence the data shown in Tables 7-8) are often biased upwards, perhaps by 20 percent or so, depending on the size of the military wage bill. They may also be biased upwards if the salaries of nongovernment contractual personnel for public education and health (not included in employment data) are classified under “government salaries” (instead of under “transfers”). On the other hand, the wage bill includes only cash payments: sizeable in-kind payments for housing, transportation, and food would offset these upward biases to some extent. Employment data can also vary according to methodological differences: “employment” is exaggerated if the source is establishment data inclusive of unfilled vacancies. Employment can also be biased upwards if it is based on payroll data inclusive of a large number of ghosts. As with the government wage bill, employment generally refers to central government civil servants only (Mali is an exception: regional employees are included).

In Table 4, the most recent estimates of nominal GDP have been taken. Estimates of GDP are often revised substantially by national statisticians. For example, the national accounts in Lesotho were revised in March 1997, resulting in a substantial downward revision of nominal GDP, especially for the years prior to 1990. In Table 8, the consumer price index has been used for calculating real wages per civil servant.

The authors were reluctant to adjust raw data. However, the employment data for Benin prior to 1992 were adjusted on the basis of the known decline of 2,610 in 1992. Finally, for the countries for which there are missing employment data in some years (e.g., Equatorial Guinea, Ghana, Niger, and Tanzania), the earlier series may not be consistent in coverage with the later series. However, no adjustments have been applied to these data.

Table 4.

Government Wage Bill as a Percent of GDP, 1986-96

article image
Source: IMF.

Unweighted average of 32 countries.

Table 5.

Wage Bill as a Percent of Current Expenditures, 1986-96

article image
Source: IMF.

Unweighted average of 32 countries.

Table 6.

Wage Bill as a Percent of Total Revenue Excluding Grants, 1986-96

article image
Source: IMF.

Unweighted average of 32 countries.

Table 7.

Wage Bill per Employee as a Multiple of GDP per Capita, 1986-96

article image
Source: IMF.

Unweighted average of 24 countries for which continuous data are available beginning 1991.