Journal Issue

Efficient Public Sector Downsizing

International Monetary Fund. External Relations Dept.
Published Date:
January 1997
  • ShareShare
Show Summary Details

Public sector downsizing is an important part of many developing countries’ economic reform efforts. Carrying out downsizing operations successfully is not easy, however, and requires careful analysis, planning, and implementation.

PUBLIC sector downsizing is becoming an increasingly important ingredient of economic reform in developing countries. Among the unfortunate legacies of state-led development, bloated bureaucracies and overstaffed public enterprises are especially problematic. Labor redundancies are particularly severe in the transition economies of Central Europe and the former Soviet Union, where the shift from central planning to market orientation requires millions of workers to leave the public sector. But retrenchment is needed in other regions, too. In Latin America and South Asia, decades of protective policies have led to the proliferation of white elephants and “sick” industries, some of which cannot be salvaged. All over the world, technological progress is making natural monopolies disappear, exposing formerly somnolent public utilities to harsh competition. As traditional and authoritarian ways are replaced by more modern and democratic ones, countries’ governments and business leaders are showing a greater willingness to correct the employment excesses that have resulted from past patronage and cronyism.

In many developing countries, the extent of labor redundancies is so vast that any serious downsizing may actually be politically infeasible, at least if it is to rely on involuntary dismissals. Consequently, a voluntary approach to reducing public sector employment has been popular with developing country governments and the multilateral organizations and donor countries that assist them. More specifically, severance pay is offered to encourage redundant workers to quit and thus overcome their resistance to downsizing, restructuring, and privatization.

Until recently, one of the main obstacles to implementing the voluntary approach to public sector downsizing was its extremely high cost. In a typical national public sector downsizing operation, hundreds of thousands of workers need to be relocated to the private sector, with the average compensation and retraining package amounting to several thousand dollars per worker. A single operation may therefore cost hundreds of millions of dollars. Unfortunately, the countries where such operations are most needed are usually cash strapped. Recent changes in the attitude of some multilateral organizations toward mass retrenchment, however, have significantly eased these constraints. For instance, in February 1996, the World Bank’s operational rules were modified to allow lending to fund severance pay, provided that it is being offered as part of a plan to restructure the public sector. And other multilateral organizations have also demonstrated that they favor the voluntary approach to public sector downsizing. As a result, many developing countries are either planning or already undertaking public sector downsizing operations.

While the efficiency gains from public sector downsizing operations are potentially large, there is also considerable risk of mishandling them. It is painfully clear that the public sectors of many developing countries are plagued with large numbers of workers who contribute little or nothing to output or welfare. The issue is whether the use of severance pay packages will encourage low-productivity workers to depart or rather induce their more productive peers to do so.

Returns to downsizing

Downsizing operations, or projects, are not different in nature from more standard investment projects, and they should therefore be subject to the same kind of evaluation. Standard investment projects are undertaken only when their benefits outweigh their costs. When product and factor markets are competitive, investment projects’ benefits and costs assessed at market prices provide an accurate measure of their benefits and costs to society. The connection between the financial and economic returns to be derived from downsizing projects, however, is much more tenuous. As part of a broader effort to revive the use of project analysis in connection with the formulation of development policies (Devarajan and others, 1995), it may be useful to clarify this connection.

Financial returns to downsizing result from the reduction in the wage bill of the agency or enterprise being restructured. When the present value of this reduction is higher than the up-front costs of the project—for severance pay, retraining programs, and the like—downsizing has positive financial returns. Economic returns, in contrast, result from a better allocation of labor across sectors. When displaced workers add more to output or welfare when they are outside the public sector than when they were in it, downsizing has positive economic returns. In the simplest case, where labor productivity in the public sector is close to zero, all downsizing projects will have positive economic returns. More commonly, however, financial returns may be a poor indicator of economic returns. High financial returns to downsizing projects, however, should come as no surprise. If, as is often done in practice, the equivalent of a couple of years of salary is offered to whoever is willing to leave the public sector, the up-front costs can frequently be recovered, owing to reductions in the sector’s wage bill, in only two years. But this calculation does not address the issue of whether workers who take this offer will prove to have been redundant.

There is an indirect measure of the economic return to downsizing—namely, the percentage of displaced workers who are subsequently rehired by the same agency or enterprise. Extensive rehiring reveals a poorly handled downsizing process. In the best case, it implies that workers who were essential to the operation were mistakenly considered redundant. In the worst case, it suggests that workers who had no intention of permanently leaving the public sector were able to cash in by accepting “golden handshakes” (attractive buyout offers). The absence of any rehiring, however, is not necessarily an indication of a downsizing project’s success. Indeed, another type of labor misallocation consists of retaining public sector workers who have low productivity The sample considered, which was drawn from a cross-country survey of downsizing operations (Haltiwanger and Singh, 1997), is small (Rama, 1997), but the fact that rehiring was observed in 40 percent of the operations, and was very substantial in 20 percent of them, suggests that economic returns to downsizing have been low.

Assessing redundancies

Even if the individual productivity of all public sector workers were perfectly observable, identifying the redundant ones would not be a straightforward task. This is because overstaffing is only one among several distortions and imperfections. Typically, government agencies and state-owned enterprises are also characterized by “wrong” pay scales, with those at the bottom of the hierarchy earning more, and those at the top earning less, than their private sector counterparts. Moreover, they usually benefit from subsidies and transfers from the budget, and these, in turn, are financed by distortionary taxes. Other imperfections lie outside, rather than within, the agencies or enterprises that are restructured. For instance, it may take months or years for displaced workers to land new jobs. And mass displacement of public sector employees may also affect the jobs and earnings of other workers, as is often seen following substantial downsizings in one-company towns.

Tackling only one distortion thus cannot guarantee an improvement in economic efficiency. The best solution is, of course, to remove all distortions and imperfections simultaneously. In state-owned enterprises, privatization would go a long way toward accomplishing this. Private owners would have neither any interest in keeping “wrong” pay scales nor easy access to government subsidies and transfers. The other imperfections could be addressed by formulating appropriate regulations and creating adequate incentives. Consequently, downsizing should not be undertaken before privatization when the latter is feasible. If it is, chances are that the government’s decisions about the numbers and kinds of workers to be separated from the enterprise will not mesh well with the plans of the new owners. And even if all redundant workers were correctly identified by the government, its separation costs per worker would probably be too high.

This first-best solution to the problems of an overmanned and inefficient government enterprise or agency may, however, be unattainable. Privatization is not an option for many government agencies, and it may not be advisable for a few state-owned enterprises. Also, privatization may not be feasible on political grounds, at least until the government shows that it can overcome labor resistance and trim employment. If privatization is ruled out, however, some of the distortions discussed above may not be entirely avoidable. For instance, changing the pay scale or achieving full cost recovery may be beyond the reach of many government agencies or state-owned enterprises. A second-best approach is therefore needed. This involves using a “shadow” wage, in much the same way that shadow prices are used in the economic appraisal of investment projects. More specifically, workers whose productivity is lower than the shadow wage should be considered redundant, even if their productivity is higher than their actual wage. Conversely, workers whose productivity is higher than their shadow wage should be retained, even if their productivity is lower than their actual wage. The appropriate formulas for the shadow wage can be found in Rama (1997).

Adverse selection

When individual productivity is difficult to measure, as is often the case in government administration, the shadow price approach described above cannot be implemented. Other strategies are therefore needed to carry out the downsizing. In practice, the typical strategy involves two steps. First, an estimate of the percentage of redundant workers is produced by comparing the actual employment level to some supposedly optimal employment benchmark in terms of economic efficiency Second, severance pay is offered to those willing to resign their public sector job voluntarily. An obvious problem with this strategy is that the percentage of workers accepting the buyout offer may differ from the estimated percentage of redundant workers. That percentage will depend on the level at which severance pay is set. But a far more serious problem—the adverse selection problem—concerns the composition of stayers and leavers.

If severance pay is used to induce voluntary separation from government agencies, the hard-working employees would be the first ones to leave. This is because they would experience a smaller welfare loss than the lazy ones if they left the public sector.

Hence the need to design the appropriate mechanisms to identify the hard-working employees. The main difference between severance pay offers and these other mechanisms is that the latter should create incentives for such workers to stay in the public sector. Because lazier employees stand to lose more if they are displaced, they are more likely to value job security than harder-working ones. As a result, an offer made to a government agency’s employees that would permit those who relinquished tenure to switch to better-paid fixed-term contracts could appeal to hardworking employees but look unattractive to lazy ones. In practice, a menu of options, possibly including a standard severance pay offer, may be needed. The key is not the precise list of items in the menu, but rather the use of self-selection as a substitute for perfect information about the characteristics of public sector workers.

Downsizing in Ecuador’s central bank

In the early 1990s, the Central Bank of Ecuador (Banco Central de Ecuador, or BCE) had 5,800 relatively well-paid employees on its payroll. Political patronage and bad management were partly to blame for the overstaffing. Apart from this, other redundancies were necessary owing to financial liberalization, which had reduced the scope for direct credit allocation by the BCE. Whatever the causes, a very crude assessment done by the management of the BCE suggested that half of the 5,800 employees were redundant. The management of the BCE thus came to accept that a reduction in personnel was warranted. This reduction took place in two quite distinct phases.

A first round in 1992 led to the departure of 1,400 employees on a strictly voluntary basis. Severance pay packages amounting to the equivalent of up to 30 years of national per capita income were offered to everyone below the minimum retirement age. Chaos ensued, since most employees thought the compensation package offered was too generous to last and therefore rushed to apply for severance. Work effort dropped. Despite the widespread perception that the offer was generous, the BCE employees’ trade union fought to improve it. Meanwhile, BCE employees above the minimum retirement age lobbied to obtain access to the new package. Not everyone was pleased though. The Congress considered it an abuse of government funds. A parliamentary commission was created to investigate the issue, and the chairman of the BCE faced the possibility of a trial. Although he did not go to jail, he lost his job. In the meantime, the productivity of the BCE declined further, owing to the lack of selectivity in the personnel reduction.

The second phase took place in 1994, in a very different context. It was decided that job separations had to be explicitly based on the role and functions of the BCE, as defined by the 1992 law on “monetary regulation and state banking.” In practice, this entailed the classification of all personnel into three groups, labeled A, B, and C. Employees in Group A were those the BCE could not afford to lose; they were therefore denied access to the new severance pay package. At the other end of the spectrum, employees in Group C were clearly redundant. The problem was that they could not legally be fired. So, on a case-by-case basis, they were informed that new laws in preparation would eliminate job security for government employees, and that the BCE’s next offer of compensation (that is, after the relevant laws were enacted) would be much less generous than its current one. The use of this threat was successful, in that all of the employees in Group C agreed to leave. Finally, all employees in Group B were given the choice of staying in the BCE or taking the severance pay offer. In this instance, the severance pay offer made to Groups B and C did not exceed the equivalent of 10 years of national per capita income. Overall, 1,400 employees left during the second phase. Those who stayed at the BCE have subsequently experienced a thorough overhaul of the wage structure and working conditions. Job descriptions and performance evaluations have been introduced. Nine BCE buildings, out of a total of 14, have been sold in order to facilitate regrouping of the remaining personnel. Reportedly, the morale of those who stayed with the BCE has improved.

This sequence of a downsizing program, with the first phase accounting for almost all of the mistakes and the second phase being considerably more effective, is not specific to the BCE. Many other government agencies around the world have gone through a similar, two-phase downsizing process. The challenge is, of course, to get it right the first time.

This box is based on Rama and Maclsaac, 1997.


The need to compensate displaced public sector workers stems from the welfare losses they are bound to experience as a result of displacement. There are several dimensions to these losses, including possible gaps in earnings, stability, and work effort between public and private sector jobs. In situations where public sector workers cannot be legally dismissed, any compensation that falls short of each worker’s perceived loss will fail to encourage his or her voluntary departure. But even where dismissals are legally feasible, insufficient compensation may derail the restructuring program, since public sector workers are among the most vocal and influential interest groups in many societies. Finally, if policymakers’ focus extends beyond legal and political constraints, they may be able to justify compensation on fairness grounds.

Compensation offers may, however, contradict the broader objectives of economic policy reform in developing countries. Many reform efforts supported by multilateral organizations and donor countries are aimed at reorienting public expenditures toward the neediest. The rationale for these efforts is that the poor in general, and the rural poor in particular, have a disproportionately weak voice in the policymaking process. Some would claim that there is a conflict between supporting such efforts to tilt the budgetary process in favor of the poor and lending generous amounts of money to finance severance pay packages for public sector workers, who usually are not poor. Although the decision to offer severance pay should be made on a case-by-case basis, there is clearly no justification for overcompensating displaced workers.

Applying the rule of thumb most commonly used to compensate displaced public sector workers frequently creates problems. Usually, severance pay is set as a multiple of the last public sector wage. As a result, those at the top of the hierarchy (say professionals) are offered much better deals than those at the bottom (say, janitors). But those at the top have less to lose, in relative terms, if they leave. Because of the egalitarian nature of the public sector, their wages are usually below the corresponding private sector wages, while the opposite is true for those at the bottom of the hierarchy. A severance pay offer based on public sector wages would therefore overcompensate those at the top of the hierarchy but fail to encourage the departure of those at the bottom. Note, however, that redundancies tend to be more prevalent at the bottom.

The rules of thumb used to set severance pay are generally based on one or two of the observable characteristics of public sector workers, including their wage. But several case studies show that a much better prediction of their welfare losses could be made by using information about other observable characteristics. For instance, welfare losses from displacement tend to be larger for those with less education and more seniority in the job. They also tend to be larger for women and for employees who have large families. Although the precise determinants vary from country to country, the studies show that a better tailoring of severance pay offers could reduce the total cost of downsizing by 20 percent or more (Assaad, 1997). Of course, in some countries, it may not be possible to base severance pay on characteristics such as gender. But even with an imperfectly tailored severance, the savings would far exceed the cost of producing and processing the data required for the tailoring.


In this context, a brief comparison between downsizing operations and structural adjustment programs may be warranted. One of the main criticisms of structural adjustment programs has focused on their adverse impact on the effectiveness of government. By compressing government wages, it is argued, these programs have encouraged the most skilled workers to leave the public sector. And there is little doubt that better outcomes could have been achieved by getting rid of the less skilled public sector workers and using the resulting savings to offer better wages and working conditions to the more skilled ones. A mishandled downsizing program, however, would not improve matters. The more skilled public sector workers would still leave, probably in larger numbers and at a much higher cost to taxpayers.

Several practical steps can be taken to ensure that public sector downsizing actually increases economic efficiency. They amount to five simple propositions. First, downsizing should not precede privatization when the latter is advisable economically and feasible politically. Governments generally do not excel at making managerial decisions, and designing downsizing operations is one of these. Second, the activities to be discontinued and the workers to be separated from their jobs should be identified before any severance pay offer is made. The number and composition of leavers should not be the outcome of a severance pay offer made across the board. Third, the mechanisms used to identify redundant workers should be adapted to each case. Engineers may be needed to assess labor productivity in state-owned enterprises, but game theorists may be better at designing appropriate menus of options to be offered to employees of government agencies. Fourth, overstaffing is only one among several distortions and imperfections in or around the public sector that affect the optimal extent of downsizing. Wage gaps between public and private sector jobs, inefficient tax systems, long periods of unemployment, and the possibility of shifting the costs of severance pay or early retirement to other agencies may all blur the true returns to downsizing. And fifth, the overcompensation of displaced workers should be avoided. Public sector employees are usually not poor, so that offering them excessively generous severance pay conflicts with separate efforts to reorient government expenditures toward the neediest.

If full compensation is justified (and this is a big if), labor market data can be used to predict workers’ losses from displacement, thus enabling the government to tailor its offerings of severance pay.

This article draws on a World Bank research project on public sector retrenchment and efficient compensation schemes. See

Martin Ramaa Uruguayan national, is a Senior Economist in the World Bank’s Development Economics Research Group.


    RaguiAssaad1997“An Analysis of Compensation Programs for Redundant Workers in Egyptian Public Enterprise,”unpublished paperUniversity of Minnesotaand forthcoming in the World Bank Economic Review.

    ShantanayanDevarajan and others1995“Reviving Project Appraisal at the World Bank,”Policy Research Working Paper No. 1496 (Washington: World Bank).

    JohnHaltiwanger and ManishaSingh1997“Cross-country Evidence on Public Sector Retrenchment,”unpublished paperUniversity of Maryland.

    MartínRama1997“Efficient Public Sector Downsizing,”World Bank Policy Research Working Paper and forthcoming in the World Bank Economic Review.

    MartínRama and DonnaMaclsaac1997“Earnings and Welfare after Retrenchment: Central Bank Employees in Ecuador”unpublished (Washington: World Bank).

Other Resources Citing This Publication