TheWorld Bank J a n u a r y PREMnotes 2 0 0 4 n u m b e r 8 3 Public Sector Salary bonuses in revenue departments: do they work? Many revenue departments offer bonuses in hopes of improving worker performance. But such mechanisms pose enormous risks--and so require careful design and cautious implementation. Compensation provides a major incentive for important are attracting, motivating, and Bonuses do not workers to perform well. Accordingly, bonus retaining skilled staff--efforts that are strongly payments are often used to enhance staff influenced by compensation packages, includ- automatically effectiveness and efficiency--and have ing potential bonuses. There is also reason become increasingly common in revenue to believe that bonuses can help prevent cor- increase the departments around the world. A recent ruption, especially if it results from unrea- World Bank study analyzed the bonus sys- sonably low pay for tax officials. However, effectiveness tems applied by revenue authorities in seven there is no conclusive evidence on the link countries, and was complemented by ques- between compensation and corruption. of revenue tionnaires completed by tax and customs administrations in seven others. Though the A system's design should departments sample was small, the study generated inter- reflect its goals esting findings and points to areas requir- To be effective, a revenue department's bonus ing further analysis. system must be generous enough to make a The countries analyzed indicate that bonus difference in employees' take-home pay. In systems are fraught with danger and so must addition, criteria for allocating bonuses should be designed with the utmost care. There is reflect the system's goals. If the main goal is no evidence that bonuses automatically in- to increase revenue, the bonus system should crease the effectiveness of revenue depart- reward revenue-generating efforts by, for ments. And because staff performance is example, measuring revenue in excess of a influenced by factors other than salaries and given target. bonuses, it is not easy to determine the effects Such a limited focus is risky, however, of such systems. This note reviews reasons for because it can undermine other aspects of introducing bonus systems and discusses a performance. For instance, Denmark's bonus number of crucial design features, particu- scheme initially focused exclusively on enhanc- larly the importance of effective and trans- ing revenue, leading to taxpayer harassment. parent performance appraisal systems based Similarly, Latvian taxpayers felt gouged by tax on clear performance measures. collectors eager to qualify for bonuses. As a result both systems had to be modified. Why offer bonuses? If the aim is to make the revenue depart- A revenue department is supposed to collect ment more efficient in specific operational government revenue at low cost and low areas, performance criteria should reflect inconvenience to taxpayers. Many factors expected results--such as the number and influence the effectiveness and efficiency results of audits conducted in a given period, of revenue authorities. But among the most the reduction in taxpayer complaints, and from the development economics vice presidency and povert y reduction and economic management network the increased satisfaction of importers with for revenue administration staff. To deal the customs regime. with this resentment, efforts may be needed A bonus system's target group also needs to explain the crucial function of mobiliz- to be based on expected results. Should the ing budget revenue to finance public services, system be limited to managers, as it initially as well as the high risk of corruption in rev- was in Denmark? Should it cover all staff enue departments. equally, as in Ghana and Latvia? Or should In Ghana the bonus system for the Inter- it provide equal bonuses to all staff plus nal Revenue Service (IRS), introduced in supplemental bonuses based on individual 1986, substantially raised take-home pay for performance, as in Brazil? In Romania indi- IRS staff. Although this situation was initially vidual performance determines 70 percent accepted by other civil servants, support Performance of the bonus, with the rest based on col- eroded over time. As a result the compen- lective performance. In South Africa the sation of IRS staff became the reference evaluations must ratio is half and half, while in Iran collective point for salary demands in the rest of the performance accounts for 90 percent of the civil service. In response the government be fair and based reward. In Morocco evaluations are based froze IRS staff compensation until the rest on individual performance, but all staff of the civil service reached its level, under- on robust, relevant receive some bonus. mining the system's initial objective. indicators Award criteria should have Proper performance indicators widespread legitimacy are crucial Without internal and external legitimacy a Performance evaluations must be fair and bonus system will not be sustainable, because based on robust, relevant indicators. These stakeholders in the revenue administration indicators should be clear to staff and to will work against the system. Excessive dis- managers preparing evaluations, and mea- cretion in allocating bonuses undermines sured performance should be clearly attrib- internal legitimacy. Albania addressed this utable to the staff member receiving a issue by installing a three-tier responsibil- bonus. Indicators must provide a sound basis ity system so that no single manager has the for measuring and comparing individuals, final say in awarding bonuses: staff are teams, and departments, including changes allowed to see their evaluations, there is an over time. This requires agreed definitions, appeals mechanism, and the names of bonus standardized concepts, and efficient infor- recipients are published. mation collection. Good indicators mini- Staff rotation, common in most customs mize the risk of data manipulation and do administrations, helps prevent collusion not neglect performance areas that are not between managers and staff and can signal measured. Targets used as performance indi- that evaluations are unbiased. A bonus sys- cators should be SMART: specific, measur- tem that by design excludes a large portion able, achievable, relevant, and timed. of staff--such as the Philippine system, Most of the countries surveyed rely more which only benefits staff involved in detect- on effectiveness than efficiency indicators ing customs fraud--risks undermining staff and generally use just three or four. Evalu- morale and ultimately its own effective oper- ations and negotiations are especially impor- ation. The more discretionary factors that tant in systems with unclear indicators, and are built into a system, the greater are the ensuring their legitimacy requires special efforts required to build confidence in it. attention. In the absence of a universal set External legitimacy is also important, of indicators that suits all countries, each because the budget resources required to revenue administration needs to develop its fund the bonus system need to be appro- own--taking into account intended goals, priated on a regular basis. Government institutional arrangements, the civil service approval is difficult to obtain and sustain if culture, and the relationships between the the rest of the civil service objects to bonuses pay system for the system's beneficiaries and PREMnote 83 January 2004 that for the rest of the civil service. The base pay is lower, bonuses are far more sub- designs of the bonus systems in Denmark stantial. For instance, in Morocco's customs and Brazil are instructive in this context. administration performance bonuses have replaced all others and can reach 100 per- Denmark's system is complex but effective cent of base salaries--aiding recruitment Denmark's Customs and Taxation Depart- efforts. In Romania bonuses can reach ment has experimented with bonus systems three times base salary. since the early 1990s, and settled on a new one in 1999. Initially limited to regional Brazil's scheme has had mixed results managers, the scheme was extended to all Brazil's revenue department has also exper- managers in 2000 and to all staff in 2001. imented with bonus systems over the past Regional managers can obtain a 5­15 per- decade. Early systems distributed about two- To enhance cent salary supplement, depending on the thirds of the fines collected from taxpayer size of the office they manage, plus a per- audits. Bonuses were initially based on indi- effectiveness formance bonus capped at 15 percent of vidual performance but evolved into equal base salary. The best-performing man- salary supplements for all staff. and efficiency, agers--those that exceed by 100 percent Bonuses still have a collective component, the targets stated in their performance con- with 40 percent determined by whether a a bonus system tracts--receive an additional bonus, but the unit achieves its revenue collection target. total performance bonus cannot exceed 25 (The full bonus is awarded if 97 percent must be legitimate, percent of base salary. of the target is reached--and forfeited if Regional managers' performance bonuses less than 90 percent is achieved.) But in 1999 objective, are based on several components: the revenue department reintroduced · 40 percent is determined by hard bonuses for individual performance, and transparent, and numeric results. From about a hundred these account for the other 60 percent. goals, the department chairman and Supervisors conduct performance evalua- easy to administer regional director each select four or five tions, which staff can review and appeal. All that reflect the circumstances of regional staff are eligible for bonuses of up to 50 per- offices and offer verifiable improvements cent of base salary. in effectiveness and efficiency. Since the bonus system was introduced, · 20 percent depends on achieving a per- auditors have increasingly focused on large formance target set by the minister of taxpayers with high revenue potential--sug- finance. gesting that collective bonuses provide their · 20 percent depends on achieving targets intended incentives. But bonuses for indi- for human resource management (such vidual performance are poorly managed, as results from staff surveys or reductions with nearly all staff members receiving the in staff sick leave). maximum score. Greater reliance on man- · 20 percent depends on contributions agement reviews of individual performance toward achieving the department's strate- evaluations could prevent this outcome. gic objectives. The process starts with a self-assessment by Conclusion the regional manager, backed by internal sta- In some cases bonuses may enhance a rev- tistics, and is followed by sometimes arduous enue department's effectiveness and effi- negotiations with the department chairman. ciency. To do so, the bonus system must be One criticism of the Danish bonus sys- legitimate, objective, transparent, and easy tem is that it risks not being attractive to administer. But these requirements pose enough to influence staff performance. substantial challenges for design and imple- The same has been said of bonus systems mentation. Only well-designed schemes will for revenue departments in some other achieve the desired results. industrial countries, such as Canada. In Although bonus schemes are increasingly developing countries, where civil service popular in revenue administrations in devel- PREMnote 83 January 2004 oping countries, current information on their Perry, James L. 1991. "Linking Pay to Perfor- operations and effects does not confirm their mance: The Controversy Continues." In universal success and usefulness. More infor- Carolyn Ban and Norma M. Riccucci, eds., mation is needed on these schemes. More- Public Personnel Management: Current Con- over, it might be possible to achieve their cerns, Future Challenges. New York: Longman. intended results by placing staff in revenue Teel, Kenneth S. 1986. "Compensation: Are administrations on a higher pay scale than Merit Raises Really Based on Merit?" Per- the rest of the civil service. Such a move would sonnel Journal 65 (March): 88­95. be justified by the importance of their func- tion and could help contain corruption-- This note was written by Luc De Wulf (Consul- and would avoid the pitfalls of the schemes tant, PRMTR) and is based on a 2001 study by Only well-designed described here. PLS Ramboll that was largely financed by the Dan- ish Governance Trust Fund. The full study is schemes will Further reading available on the Website of the Tax Policy and Mookerjee, Dilip. 1998. "Incentive Reforms Administration Thematic Group (www.world- achieve the in Developing Country Bureaucracies: bank.org/publicsector/taxation). Lessons from Tax Administration." In Boris If you are interested in similar topics, con- desired results Pleskovic and Joseph E. Stiglitz, eds., sider joining the Tax Policy and Administration Annual World Bank Conference on Develop- Thematic Group. Contact Michael Engelschalk ment Economics 1997. Washington, D.C.: (x87764) or click on Thematic Groups on World Bank. PREMnet. This note series is intended to summarize good practices and key policy findings on PREM-related topics. The views expressed in the notes are those of the authors and do not necessarily reflect those of the World Bank. PREMnotes are widely dis- tributed to Bank staff and are also available on the PREM Website (http://prem). If you are interested in writing a PREMnote, email your idea to Madjiguene Seck. For additional copies of this PREMnote please contact the PREM Advisory Service at x87736. PREMnotes are edited by Paul Holtz and laid out by Suzanne Luft. Prepared for World Bank staff